IBS Special Issue by Wiley-Blackwell

4th

Sp eci al issue: Int ern at ional Bi ogeography Society Meet ing

Ecography 33: 209 211, 2010 doi: 10.1111/j.1600-0587.2010.06677.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Carsten Rahbek. Accepted 17 May 2010

Celebrating the diversity of biogeographical research Special issue: International Biogeography Society, 4th biennial meeting David Nogue´s-Bravo and Carsten Rahbek D. Nogue´s-Bravo (dnogues@bio.ku.dk) and C. Rahbek, Center for Macroecology, Evolution and Climate, Dept of Biology, Univ. of Copenhagen, Universitetsparken 15, DK-2100, Copenhagen, Denmark.

Biogeography aims to understand the temporal and spatial distribution of life on Earth. Biogeographical research is aimed not only at describing where organisms live, at what densities, with whom, and how it all relates to the environmental and geographical setting but also why this is so. The International Biogeography Society, IBS, is a young and vibrant international and interdisciplinary society contributing to the advancement of all studies of the geography of nature, including spatial ecology (<www.biogeography.org>). In January 2009, the 4th International Conference of the International Biogeography Society took place in Merida on the Yucatan Peninsula, Mexico. Ecography provided financial support, acting as the sponsor of the Symposium of Extinction Biogeography and contributing to student travel awards. In addition, Ecography was the officially designated journal for publishing some of the many exciting talks and posters presented at the conference. All of the papers in this special issue of Ecography arose from the IBS conference. They have all been subject to external peer review, subsequent revision, and final editorial decisions of acceptance/rejection.

209

ISSUE

Understanding past and current extinctions and their spatio-temporal dynamics is of tremendous direct interest, but insight from such studies is also of importance in understanding the impacts of contemporary and future global changes in land use and climate on species. Southern European peninsulas, for example, were traditionally recognized as glacial refugia where many species survived during the ice ages. In a study using species distribution models in a phylogeographical research framework, Vega et al. (2010) challenge this view by showing that it is plausible that the pygmy shrew had northern refugia during the Last Glacial Maximum. However, climate change was not the only factor affecting global or local extinctions during the Late Quaternary. Humans were also a well-known factor causing the extinction of species, mainly on islands, where humans have disrupted key ecosystem functions. To minimize the unwarranted effect of disrupted ecosystem functioning, Hansen et al. (2010) propose that humans should actively replace extinct taxa by introducing analogue taxa with presumed similar ecological functions as the extinct species. They illustrate this approach with some taxon substitution projects on islands using large tortoises as examples. Also on islands, the dramatic extinction debt revealed by Triantis et al. (2010) calls for better management, including the restoration and expansion of native forests. Species living at the top of mountains are like oceanic

IBS SPECIAL

The special issue starts with an article by the local organizers (Vazquez-Dominguez and Arita 2010) that provides an introduction and overview of the biogeographic history of the Yucatan Peninsula, the setting of the conference before it delves into a series of 22 papers that represents the diversity of what constitutes biogeographical research in the 21st century. The first series of papers focuses on speciation, extinction and migration as the three key principal forces that drive the distribution of biological diversity. Understanding when, how and where new species arise is of fundamental importance to our basic understanding of biodiversity on Earth. Reconstructing the evolutionary history of the family Oriolidae by generating a molecular phylogeny based on both nuclear and mitochondrial DNA sequence data, Jønsson et al. (2010) shed new light on how species in this clade dispersed first from their Australian area of origin to Asia and then onwards to Africa before back-colonising Asia and the Indonesian archipelago. The hypothesis that diversification rates are higher in active than in passive tectonic settings is explored in the paper by Badgley (2010), and Casner and Pyrcz (2010) show that speciation of butterflies in tropical mountain regions occurs primarily within elevational bands. Using a global database on the world’s amphibians, Hof et al. (2010) find an indication of historical signals in the realized climatic niches of species.

ISSUE

IBS SPECIAL

islands in that they are also expected to be highly exposed to extinctions because there appears to be nowhere to migrate upwards when lower altitudes warm up as a consequence of global warming. However, upward migrations of species tracking climate change may not be the only possible scenario; some species could go against the flow. Lenoir et al. (2010), focusing on the latter, discuss potential mechanisms for unexpected downward range shifts of mountain plant species under climate change. The impact of climate change on species distribution has traditionally been attempted by using species distribution modeling, but the usefulness of this model approach may well be affected by the uncertainty embedded in the climate models used to forecast future climatic conditions (Real et al. 2010). Migration of species through evolutionary and ecological time has profoundly shaped biogeographical patterns at different scales, from populations to whole continental biotas. Using 240 datasets, Jenkins et al. (2010) show that in the era of landscape genetics, ‘‘isolation by distance’’ still matters in modern population genetics. Migrations across biogeographic boundaries such us the Great American Biotic Interchange, have profoundly shaped current patterns of biological diversity in the New World. In a Special Feature within this special issue, introduced by Riddle and Hafner (2010), a small series of papers focus on understanding the timing and the biological consequences of the Great American Biotic Interchange (Cody et al. 2010, Smith and Klicka 2010), on the vicariance processes in Middle America (Daza et al. 2010), and on the biogeographic patterns across the Mexican Transition Zone (Morrone 2010). These contributions provide novel results and illustrate fresh research venues to revisit traditional biogeographical questions that are rooted in the research legacy of classical biogeographers such as Alfred Russel Wallace (Riddle and Hafner 2010). Speciation, extinction and dispersal in interaction with the dynamics of abiotic and ecological processes are traditionally viewed as what determine current biogeographical patterns, including life history traits. Different approaches to study body size patterns and their drivers in Pacific island birds are explored by Olalla-Ta´rraga et al. (2010) and Boyer and Jetz (2010), respectively. These studies are followed by a study assessing factors thought to cause patterns in the geographical distribution of African palm species (BlachOvergaard et al. 2010), and a study assessing the interspecific range size variability of butterflies in relation to life history traits and geographic features of species distributions (Garcia-Barros and Romo Benito 2010). Not only do the distribution of species and patterns of diversity vary in time and space, so do the derived and underlying distributions of geographical ranges sizes of species assemblages. Krabbe Borregaard and Rahbek (2010) highlight the potential of using range-diversity plots for generating and testing hypotheses about how general ecological processes shape the location and size of species ranges and species richness. The study illustrates that much is still to be learned concerning the causes of large-scale patterns of species richness, a theme which is also the focus of Kreft et al.’s (2010) study on the global species richness pattern of ferns and seed plants. They suggested that taxon-specific ecological and life-history traits play

210

an important role in defining global richness gradients. Another classic research area of biogeography is the relationship between richness and area. In the last paper of the special issue, Guilhaumon et al. (2010) has contributed as a Software note, an R-package that allows users to easily implement model selection and parameter estimation to assess uncertainties in species area-relationship models. This special issue illustrates the current convergence of different academic fields such as evolutionary biology, ecology, phylogeography, and global change biology within a biogeographic framework to explain large scale patterns of biological diversity. The holistic nature of biogeography constitutes both a challenge but also an exciting opportunity for inter-disciplinary research. In light of the ongoing species extinction crisis caused primarily by habitat alteration and global changes in land-use with the recent added focus on the impact of global changes in climate on biological systems, a diverse research program is as important and relevant as ever in the history of biogeographical research. We hope that this special issue, presenting and promoting presentations at the International Biogeography Society’s conference in 2009 as peer-reviewed scientific journal papers, will contribute to a more thorough understanding of life on Earth.

References Badgley, C. 2010. Tectonics, topography, and mammalian diversity. Ecography 33: 220 231. Blach-Overgaard, A. et al. 2010. Determinants of palm species distributions across Africa: the relative roles of climate, non-climatic environmental factors, and spatial constraints. Ecography 33: 380 391. Boyer, A. G. and Jetz, W. 2010. Cross-species and assemblagebased approaches to Bergmann’s rule and the biogeography of body size in Pacific island birds. Ecography 33: 369 379. Casner, K. L. and Pyrcz, T. W. 2010. Patterns and timing of diversification in a tropical montane butterfly genus Lymanopoda (Nymphalidae, Satyrinae). Ecography 33: 251 259. Cody, S. et al. 2010. The Great American Biotic Interchange revisited. Ecography 33: 326 332. Daza, J. M. et al. 2010. Using regional comparative phylogeographic data from snake lineages to infer historical processes in Middle America. Ecography 33: 343 354. Garcia-Barros, E. and Romo Benito, H. 2010. The relationship between geographic range size and life history traits: is biogeographic history uncovered? A test using the Iberian butterflies. Ecography 33: 392 401. Guilhaumon, F. et al. 2010. mmSAR: an R-package for multimodel species area relationship inference. Ecography 33: 420 424. Hansen, D. M. et al. 2010. Ecological history and latent conservation potential: large and giant tortoises as a model system for taxon substitutions. Ecography 33: 272 284. Hof, C. et al. 2010. Phylogenetic signals in the climatic niches of the world’s amphibians. Ecography 33: 242 250. Jenkins, D. G. et al. 2010. A meta-analysis of isolation by distance: relic or reference standard for landscape genetics? Ecography 33: 315 320. Jønsson, K. A. et al. 2010. Phylogeny and biogeography of Oriolidae (Aves: Passeriformes). Ecography 33: 232 241.

Krabbe Borregaard, M. and Rahbek, C. 2010. Dispersion fields, diversity fields and null models: uniting range sizes and species richness. Ecography 33: 402 407. Kreft, H. et al. 2010. Contrasting environmental and regional effects on global pteridophyte and seed plant diversity. Ecography 33: 408 419. Lenoir, J. et al. 2010. Going against the flow: potential mechanisms for unexpected downslope range shifts in a warming climate. Ecography 33: 295 303. Morrone, J. J. 2010. Fundamental biogeographic patterns across the Mexican Transition Zone: an evolutionary approach. Ecography 33: 355 361. Olalla-Ta´rraga, M. A´. et al. 2010. Cross-species and assemblagebased approaches to Bergmann’s rule and the biogeography of body size in Plethodon salamanders of eastern North America. Ecography 33: 362 368. Real, R. et al. 2010. Species distribution models in climate change scenarios are still not useful for informing policy planning: an

uncertainty assessment using fuzzy logic. Ecography 33: 304 314. Riddle, B. R. and Hafner, D. J. 2010. Integrating pattern with process at biogeographic boundaries: the legacy of Wallace. Ecography 33: 321 325. Smith, B. T. and Klicka, J. 2010. The profound influence of the Late Pliocene Panamanian uplift on the exchange, diversification, and distribution of New World birds. Ecography 33: 333 342. Triantis, K. A. et al. 2010. Extinction debt on oceanic islands. Ecography 33: 285 294. Va´zquez-Domı´nguez, E. and Arita, H. T. 2010. The Yucatan peninsula: biogeographical history 65 million years in the making. Ecography 33: 212 219. Vega, R. et al. 2010. Northern glacial refugia for the pygmy shrew Sorex minutus in Europe revealed by phylogeographic analyses and species distribution modelling. Ecography 33: 260 271.

IBS SPECIAL ISSUE

211

Ecography 33: 212 219, 2010 doi: 10.1111/j.1600-0587.2009.06293.x # 2010 The Authors. Journal compilation # 2010 Ecography

The Yucatan peninsula: biogeographical history 65 million years in the making Ella Va´zquez-Domı´nguez and He´ctor T. Arita

IBS SPECIAL

ISSUE

E. Va´zquez-Domıńguez (evazquez@ecologia.unam.mx), Inst. de Ecologıá, Universidad Nacional Autońoma de Me´xico, Ap. Postal 70-275, Ciudad Universitaria, Me´xico DF, 04510, Me´xico. H. T. Arita, Centro de Investigaciones en Ecosistemas, Universidad Nacional Autońoma de Me´xico, Antigua Carretera a Pa´tzcuaro No. 8701, Col. Ex-Hacienda de San Jose´ de La Huerta, 58190, Morelia, Michoacań, Me´xico.

The fourth biennial meeting of the International Biogeography Society (IBS) in Merida, Yucatan in January 2009 represented a double opportunity for Mexican biologists. First, it fostered the integration of the large community of Mexican biogeographers with the activities of the IBS. Second, the meeting allowed us to welcome a large number of delegates from distant parts of the world who were able to visit what has been considered an obligate destination for nature lovers and cultural tourists alike: the Yucatan peninsula. As Edward O. Wilson pointed out, besides economic power every country has two additional and important types of wealth: cultural and natural. Cultural richness is a naturally embedded component of the Mexican way of life. It is manifested in the rich legacy of ancient Mesoamerican civilizations, in the remarkable diversity of human groups, indigenous languages and dialects, local customs and food, and also in the seamless integration of modernity with tradition that can be seen in every major city. This cultural wealth is paralleled by an amazing natural richness, best illustrated by the country’s extremely high biological diversity. Mexico is the only nation in the world containing the totality of a continental border between two biogeographic realms, the Nearctic and the Neotropical. The mixing of elements of these two regions across a highly heterogeneous landscape is the perfect recipe for a ‘‘megadiverse’’ country like Mexico. Mexico’s cultural and natural richness becomes rapidly evident to any traveller to the Yucatan peninsula. Consider the caves of the southern part of the state of Yucatan as an example. In many of these caves, a casual visitor will notice a multitude of fossil seashells embedded in the walls and ceiling. Looking down, she could find small pieces of Maya ceramics interspersed with the sediment, and perhaps even a piece of the tooth of a Pleistocene horse Equus conversidens. These three interesting elements are in fact separated by orders of magnitude of time (Fig. 1): the limestone with the shells is of Oligocene origin, ca 25 million yr old, the horse became extinct some 10 000 yr ago, and the piece of ceramic is around 800 yr old. Furthermore, a much larger 212

and older piece of evidence of a past event might be in front of the visitor: many of the large sinkholes that punctuate the landscape of the Yucatan are located along the rim of the Chicxulub crater, a 180-km wide scar created by the impact of an enormous asteroid 65 million yr ago that is believed to have caused the mass extinction event of the end of the Cretaceous. Standing in front of this diverse mixing of elements of various origins, one cannot help being amazed by the particularities of the geologic, evolutionary and cultural history of the Yucatan that have produced the present-day diversity of this unique part of Mexico. In this introduction to the special section of papers presented at the IBS meeting, we offer a brief overview of the biological and cultural features that make the Yucatan peninsula such a special place. When choosing a Mexican venue for the IBS meeting, our first option was Merida, the ‘‘White City’’, the peaceful and charming capital of the state of Yucatan. What better place could it be for a biogeography meeting than atop a 180-km wide, 65 million yr old crater that testifies one of the most spectacular events in the history of life on Earth?

65 million years of history Chicxulub: the dinosaur connection It can be safely stated that the biological history of the Yucatan started, or at least was reset, 65 million yr ago (Fig. 1). The Cretaceous-Tertiary (KT) episode that happened then is one of the so-called ‘‘big-five’’ extinction events in the history of life on Earth (Raup and Sepkoski 1982, Alroy et al. 2008). The KT episode wiped out 75% of all animal species, including entire clades such as non-avian dinosaurs, ammonites, rudists and inoceramid bivalves (Marshall and Ward 1996). Current knowledge strongly suggests that the KT event was triggered by the collision of a 10-km asteroid with what is now the northern Yucatan peninsula, producing a 100 million megaton explosion that in an instant obliterated the geological

Figure 1. Time line of major events in the biogeographical history of the Yucatan peninsula. Periods of the Pre-Hispanic era are: Paleoindian, Archaic, Preclassic, Classic, and Postclassic. Note the logarithmic scale.

profile of the region and extinguished all life in hundreds of kilometres around. When first proposed (Alvarez et al. 1980), the idea of an extraterrestrial object hitting the Earth 65 million yr ago was received with considerable scepticism. The hypothesis was supported by the strong empirical evidence of a spike in iridium concentration in sediments 65 million yr old, exactly at the K-T transition. Because iridium is an extremely rare element on Earth but occurs in measurable concentrations in extraterrestrial objects, the most plausible explanation for such a spike was a space object colliding with the Earth, disintegrating in an explosion that dispersed iridium-rich sediment all over the world. The Alvarez team calculated that the hypothetical asteroid or comet should have measured ca 10 km, and should have produced a crater 200 km in diameter. One of the problems with the theory was that no crater of the right age and size was known at the time. Shortly after the Alvarez et al. (1980) paper was published, Allan Hildebrand and Stein Boynton developed a theoretical model for an asteroid impact as predicted by Alvarez and collaborators, and called for a search for the

missing crater. Unknown to Hildebrand and Boynton, evidence for a candidate crater fitting the theoretical description had been found in the 1960s and 1970s during oil exploration drillings financed by the Mexican oil company PEMEX. In 1981, Glen Penfield used the PEMEX data to describe an underwater crater north of Yucatan, but his report went unnoticed. Many years later, as Penfield and Hildebrand joined forces, the evidence was finally published in a scientific paper proposing an underground circular feature 180 km in diameter centred in the coastal town of Chicxulub as a formal candidate for the missing crater of the KT impactor (Hildebrand et al. 1991). Today, most scientists have accepted the idea that an extraterrestrial impact caused the KT mass extinction, and that the Chicxulub crater is indeed the scar of that episode (Fig. 1, 2) (Schulte et al. 2010). Statistical analyses of the fossil record show that besides background extinction throughout most of the Cretaceous, there was a clear mass extinction of ammonites coinciding with the KT boundary (Marshall and Ward 1996). There is also physical evidence of the short- and long-term effects of the collision in areas adjacent to the peninsula and farther away: the

IBS SPECIAL ISSUE

Figure 2. Yucatan peninsula (states of Yucatan, Campeche and Quintana Roo), with location of the Chicxulub crater (ring of cenotes) and Sierrita de Ticul-Loltun cave, and the limits of the Yucatan Peninsula Biotic Province.

213

100-million megaton explosion, mega-tsunamis with 300m waves, wildfires hundreds of kilometres away from ground zero, massive forest destruction, global climate change with a subsequent reduction of 80% in the photosynthesis rate, and of course mass extinction of plant and animal species (Kring 2007). Recent analyses of material extracted from the site have dated the Chicxulub crater at 0.3 million yr (Myr) before the KT horizon, casting doubt on the idea that the Chicxulub object was the sole detonator of the KT extinction (Keller et al. 2004). These findings, which suggest a more complex series of events, including even multiple impacts, have ignited a new round of controversy regarding the Chicxulub site. A recent study has added a new and surprising twist to the story (Bottke et al. 2007). Simulation models of the dynamics of the group of asteroids called the Baptistina family show that they could have originated with a fragmentation of a large asteroid 160 Myr BP, perhaps due to a collision with another object. According to the model, the largest piece resulting from the fragmentation is the present-day asteroid 298-Baptistina, which still resides in the asteroid belt. Smaller pieces were scattered and many of them entered the inner Solar System. Bottke et al. (2007) speculate that the spectacular Tycho crater in the Moon is the result of the collision of one of this Baptistina objects 109 Myr BP. The Chicxulub crater could have been produced by another of the Baptistina objects that ended its 95 Myr pilgrimage with an explosive encounter with Earth 65 Myr BP. If this is correct, Tycho and Chicxulub could be ‘‘sister craters’’ produced by an amazing sequence of improbable events. Under the sea: Cenozoic biogeography

IBS SPECIAL

ISSUE

For more than 100 Myr, from the Cretaceous until the Pleistocene, numerous marine transgressions submerged the area of what is now the Yucatan peninsula under warm

tropical waters. During this time limestone strata were formed with the remains of ancient coral reefs and seashells, including the uppermost Miocene-Pliocene (24 2 Myr BP) Carrillo Puerto Formation, a 15-m thick deposit of almost pure calcium carbonate that surrounds the shallow portions of present-day karst systems. Thus, the whole peninsula is basically a large limestone slab, submerged for millions of years, that is slowly emerging from south to north and where older deposits are located near the base. When the Baptistina object collided with Earth at the end of the Cretaceous, what is now the Yucatan peninsula was a shallow coastal shelf at the southern extreme of North America. For millions of years, south of this tip there was a wide ocean separating North and South America, producing the independent evolution of early New World mammals in ‘‘splendid isolation’’ (Simpson 1980) until the Panamanian land bridge connected the two land masses ca 3.1 2.8 Myr BP (Fig. 3), triggering the great American biotic interchange (GABI), a major mixing of biotas from South and North America that shaped the high-level taxonomic composition of modern floras and faunas of the New World (Marshall et al. 1982, MacFadden 2006, Webb 2006). Because the northern portion of the peninsula did not emerge until a few million years ago, the role of the Yucatan peninsula in the evolution of the terrestrial faunas of the Caribbean region during the Cenozoic was probably minor. Dispersal and vicariant theories have been proposed for the colonization of the Antilles and posterior in situ evolution (Da´valos 2004, Hedges 2006, Ricklefs and Bermingham 2008). Both types of hypotheses, however, call for a South American origin for the major vertebrate clades in the Antilles, with arrival and isolation times varying from theory to theory but pointing to around 35 Myr ago (Iturralde-Vinent and MacPhee 1999). An alternative hypothesis for the origin of some vertebrate groups in the Antilles is the existence of a connection between the Yucatan peninsula and the islands at the beginning of the

Figure 3. Emerged landmasses during the Middle Eocene (49 37 Myr BP), when North and South America were not connected (dotted line), and during the Pliocene (3.1 2.8 Myr BP) after the closure of the Isthmus of Panama (solid line), based on the model of Heinicke and collaborators (Fig. 4; Heinicke et al. 2007). Abbrevations: NA: North America, MA: Middle America, SA: South America, PA: Proto Antilles, Cu: Cuba, BB: Bahama Bank, Ja: Jamaica, Hi: Hispanola, PR: Puerto Rico, LA: Lesser Antilles.

214

Cenozoic, with posterior isolation and vicariant evolution, a model that fits data for cichlid fish (Chakrabarty 2006). In any of the scenarios, biotic interchange between Yucatan and the Caribbean islands is thought to have been minimal throughout most of the Cenozoic, accounting for the present-day low similarity between the vertebrate faunas of these two areas (Va´zquez-Miranda et al. 2007). Nevertheless, for certain groups the Yucatan peninsula served as a bridge for dispersal between Central America and the Caribbean islands. For example, the present-day distribution of eleutherodactyline frogs is best explained by a model that includes several events of dispersal over water from and to South America and from Cuba to Yucatan at different times of the Cenozoic (Heinicke et al. 2007). Similarly, the evolution of mormoopid bats probably involved dispersal over water from the northern Neotropics to Central America and from there to the Antilles, most likely through the Yucatan peninsula (Da´valos 2006). Interesting examples of evolution in the Yucatan fauna come from invertebrates inhabiting the fresh-water or anchihaline underground bodies of water. Shrimps of the genus Typhlatya are represented in the peninsula by three species occurring in fresh-water habitats. However, the divergence of this clade predates the origin of its present-day habitat, according to molecular data (Hunter et al. 2008). This result implies that the three species must have originated in marine habitats (the original medium of the genus) before the end of the Pliocene, when the freshwater habitats started to form. Subsequently, the three fully-formed species could have invaded the new habitat. The great American biotic interchange and the configuration of modern biotas

ISSUE

215

IBS SPECIAL

The emergence of the Panama isthmus and the subsequent great American biotic interchange, peaking at approximately 2.8 Myr BP, marked the start of the processes that have configured the modern floras and faunas of the Yucatan peninsula and other Middle American regions (Fig. 1; MacFadden 2006). Before the closure of the Panama isthmus, all Mexican mammal faunas were completely North American in composition (Ferrusquia-Villafranca 2003, Webb 2006). Today, faunas of Middle America (Mexico plus Central America) are a rich and complex mixture of North and South American components, a clear evidence of the processes associated with the GABI. North American mammalian faunas north of the Tropic of Cancer still consist mostly of elements of native families, with only a few South American components, such as opossums (Didelphidae), armadillo (Dasypodidae), a handful of phyllostomid and mormoopid bats. In contrast, faunas of tropical Mexico are rich assemblages that include both North and South American elements, some of which have evolved into idiosyncratic Mesoamerican endemics. Some families of North American origin (e.g. Tapiridae, Felidae, Sciuridae) are represented in the peninsula by species typically considered tropical (e.g. Baird’s tapir Tapirus bairdii, jaguar Panthera onca and Deppe’s squirrel Sciurus deppei). In addition, clades of South American origin

are represented by primates (black howler monkey Alouatta pigra and Geoffroy’s spider monkey Ateles geoffroyi), marsupials, bats, cingulata and pilosa (armadillo Dasypus novemcinctus and northern tamandua Tamandua mexicana) and hystricognath rodents (Central American agouti Dasyprocta punctata, and spotted paca Cuniculus paca). Because of the geological history of the Yucatan peninsula, present-day faunas of the northern part of the peninsula are of recent origin, B2.8 Myr. With very few exceptions, vertebrate faunas of northern Yucatan are subsets of the fauna of the base of the peninsula (the Peten and adjacent areas). Bats of the state of Yucatan, for example, represent a subset that cannot be distinguished from random samples of the fauna of the base of the peninsula, except that species with high dispersal capability are overrepresented in the northern fauna (Arita 1997). This suggests that faunas of the northern part of the peninsula originated simply by dispersal of species from the south. The big-eared climbing rat Ototylomys phyllotis for example, diverged and dispersed from South America toward Middle America coinciding in time with the GABI, and its present range includes the Yucatan peninsula (Gutie´rrez-Garcıá and Va´zquez-Domıńguez unpubl.). An example involving vicariant and dispersal events is the evolution of cantil pitvipers of the genus Agkistrodon (Parkinson et al. 2000). Phylogeographical studies showed that the genus Agkistrodon originally occupied relatively temperate habitats and evolved toward more tropical ones; the species Agkistrodon bilineatus, present now in the Yucatan peninsula, shows a historical initial divergence between populations from the eastern and western coasts in Mexico, with a posterior dispersal of one population to the Yucatan peninsula through subhumid corridors along northern Central America that diverged into a different subspecies, Agkistrodon bilineatus russeolus. These examples show that despite its apparent simplicity, the process of conformation of the Yucatan fauna can have many variations that depend on the idiosyncratic features of the different clades that are involved (Arita and Va´zquezDomıńguez 2003). As pointed out by Webb (2006), a major problem faced when studying the GABI is the lack of fossils of the right age at the right place. We have fossils either from the middle Miocene (well before the GABI) or from the Pleistocene (after the important processes had happened). In the Yucatan peninsula, fossils come from cave deposits of recent origin such as those from the Loltun cave in the southern part of the state. Sixty-eight animal species in ten orders, 25 families and 52 genera have been recorded as fossils in the cave, ranging in time between 30 000 and 500 yr (Arroyo-Cabrales and A´lvarez 2003 and references therein). Among the mammals found in the Loltun deposits are seven extinct species, including Pleistocene horses Equus conversidens, saber-toothed cats Smilodon fatalis, wolfs Canis dirus, mastodonts Cuvieronius sp. and camels Hemiuauchenia sp., together with the bat Desmodus draculae and the marsupial Marmosa lorenzoi (Arroyo-Cabrales and A´lvarez 2003). These extinctions probably were caused by changes in climate or by the arrival of humans, although there is no direct evidence for either of these two factors.

IBS SPECIAL

ISSUE

The Holocene and Anthropocene Evidence of human occupation of the Yucatan peninsula in the Paleoindian period has been found in the Loltun cave and in Belize, where distinctive Clovis points have been recovered from deposits in Ladyville, that have been dated at 9000 to 7500 BC (Kelly 1993). Since then, the region has been populated by humans without interruption (Fig. 1). During the Classic period (250 900 AD), a large Maya city such as Tikal harboured populations of at least 50 000 people, or perhaps even 100 000 according to some estimates, and when the Spaniards arrived in 1519, the human population in the peninsula was probably as high as it is today (Fig. 1). Many people, thinking always of the great cities of the Classic period, visualize the Maya as a vanished human group, without realizing that they still number in the millions all over southern and eastern Mexico and in Central America. Just as in other parts of the world, the cycles of development and decline of human groups in the Yucatan peninsula have been closely tied to climate conditions and the use of natural resources (Diamond 2005). There is mounting evidence of significant climate shifts in the peninsula associated with global conditions. For example, important sea level changes in the Yucatan peninsula 121 kyr ago, during the last interglacial period have been documented by analyzing fossil reefs of the peninsula (Blanchon et al. 2009). The study showed changes in sea level of up to 3 m in ecological time, perhaps as fast as in a few decades. Paleobotanical studies have documented dramatic changes in the vegetation cover of the peninsula. Only four thousand years ago, the Peten region was warmer and much dryer than it is today, and extensive savannahs existed in what is now covered with tropical rainforest. Forests began to dominate the landscape only about 2500 yr ago. In eastern Middle America, including sites in the Yucatan peninsula, the period with the densest tropical forests and deepest lakes coincides with the socalled Little Ice Age, 1350 1850 AD (Lozano-Garcı´a et al. 2007). In recent years, several studies have shown a strong correlation between changes in climate and the demise of the Classic Maya city-states. These studies are made possible not only by modern techniques that allow the reconstruction of past climates, but also by the precise calendar (the ‘‘long count’’) that the Classic Mayas used to record important events (Sharer and Traxler 2005). In every important city, steles were erected every 19 yr and 265 d to mark the start of a new k’atun (period of 7200 days). In the year 790, at least 45 such monuments were built, but 100 yr later only a dozen were produced, and on 15 January, 909, a sole stele was carved in the city of Tonina, in the highlands of Chiapas. This decline in the elaboration of monuments testifies the fall of each of the major Maya city-states of the Classic period: Palenque and Yaxchilan, in the Usumacinta river basin were abandoned first, at the beginning of the 9th century. Then, cities of present-day Belize and Guatemala followed suit and disintegrated by 860. Finally, the mega-metropolis of the Peten, such as Tikal and Calakmul were deserted before 910. This sequence is important because it points to the fact

216

that there was not a single collapse of the Classic Maya, but a series of events that took almost 100 yr to develop. In 1995, a study of stable oxygen isotopes (d18O) from sediments in a lake in Quintana Roo showed the existence of important dry spells coinciding with the end of the Classic period (Hodell et al. 1995), suggesting the idea that a ‘‘megadrought’’ could have triggered the fall of the cities. Years later, a study of the sediments of the Cariaco basin in South America, which allow the estimation of year by year rainfall patterns, demonstrated the existence not of a single event, but a series of extremely dry periods corresponding with the end of the Classic (Haug et al. 2003). Even more, the driest years (dated at 760, 810, 860 and 910 AD) coincided with the sequence of abandonment of the main Maya sites. This result points to severe drought as one of many possible causes of the collapse of Classic Maya citystates. New data on d18O from the northern peninsula shows that the 15th century abandonment of some Postclassic sites, such as Mayapan, also coincides with a particularly harsh dry spell (Hodell et al. 2007), and recent evidence from several camps corroborates the megadrought theory (Pringle 2009). Of course, the so-called collapse of the Classic Mayas was a very long and complex process that involved other environmental, social, political and religious factors as well. On 24 March, 1519, a new type of biogeographical process took place in the coast of Tabasco. Sixteen horses that arrived with the army of Herna´n Corte´s became the first animals introduced by Europeans into continental North America (Fig. 1). It is ironic that horses, which had evolved in North America only to become extinct there at the end of the Pleistocene (MacFadden 2006), gave a small band of a few hundred Spaniards the leverage to vanquish the powerful Aztec empire. After the Pleistocene extinctions, the native Middle American fauna lacked large mammals suitable for human use, so big domesticated animals were totally unknown to Mesoamerican Indians. Horses caused a tremendous impression on natives, becoming one of the most powerful weapons of the conquistadors. They were also the first in a long list of plants and animals introduced by Europeans that changed the structure and functioning of many ecosystems. In the other direction, many native crops of the Yucatan peninsula were exported to the rest of the world. Two plants in particular played important roles in the configuration of the modern landscape of the Yucatan. The tapped sap of the sapodilla tree Manilkara zapota was used since pre-Hispanic times to produce a gum (the chicle) that could be chewed. From the 1870s, when the chewing gum was introduced to the United States until the mid 1940s, the increasing demand for natural sapodilla gum was so big that it fostered the exploration of the forests of southern Yucatan in search of more trees to be exploited. These explorations contributed to the finding of many Maya ruins (Sharer and Traxler 2005), and catapulted the economic development of the whole area. After the invention of artificial substitutes for chicle in the 1940s, the demand for the natural product plummeted. Today, chicle is harvested only for specialized markets, mostly in Asia, that still prefer chewing gum based on natural products.

The other plant that drove Yucatan’s economy for a long time was henequen Agave fourcroydes. During the 19th century, demand for henequen or sisal fibers soared. At one point, up to 85% of the fiber used worldwide came from Yucatan, making the state one of the more affluent regions of Mexico by the 1880s. It was at this time that the huge haciendas in Yucatan’s countryside flourished and the luxurious houses in Merida were built. Unfortunately for the environment, extensive plantations of the agave plant ruined hundreds of thousands of hectares, in an area where poor soil makes very difficult the cultivation of other crops. After World War I, the development of artificial fibres and the competition from countries that had smuggled henequen plants out of Yucatan and were producing their own fibre marked the end of the ‘‘green gold’’ boom in the peninsula. The chicle tree and the henequen plant are only two examples of the complicated processes involving the use of natural resources in the Yucatan. Today’s Yucatan environments are the result of millions of years of evolution, but also of the direct interaction with humans within the past 10 000 yr. Even within Biosphere reserves, such as in the Calakmul area in the southern part of the peninsula, the landscape is a complex matrix of natural and humanmodified environments whose intermingling determines the dynamics of the rich biological diversity of the region (Vester et al. 2007).

Present-day Yucatan The geographic setting

Diversity patterns and conservation

217

ISSUE

The definition of biogeographic provinces is commonly based on the homogeneous distribution of the biota of a region, compared to that of adjacent areas. Early surveys of the Yucatan demonstrated a Neotropical affinity for its flora and fauna, evidenced by their composition. In particular, the Yucatan fauna is similar to that of other tropical dry zones, but it differs due to the presence of elements from the more humid areas of the south. This distinctiveness has prompted most scholars to consider the Yucatan peninsula a biotic province on its own, the Yucatan Peninsula Biotic Province (Fig. 2; Goldman and Moore 1945) or a region with two provinces, the Peten province in the south and the Yucatan province in the north. The former scheme has received much more support from recent analyses of geological and physiographic features, and of the distribution of plants, birds and mammals (Fa and Morales 1993, Morrone 2005). Vascular plants in the Yucatan peninsula are very diverse, reaching 2600 3000 species. The six most common families represent ca 41% of the total flora of the region, including Fabaceae, Poaceae and Orchidaceae. However, species richness in the Yucatan is lower than that of comparable Neotropical regions of similar size. This fact results basically from the fairly recent origin of the peninsula, its relatively dry flat terrain, and its lack of superficial water, all of which preclude the presence of the many different microclimates and local heterogeneity that are typical of other Neotropical zones (Carnevali et al. 2003). Approximately 7% of the flora is endemic to the Yucatan peninsula, with some very distinctive, conspicuous and even dominant species, such as Acacia gaumeri (Fabaceae) and Myrmecophila christinae (Orchidaceae). Four genera are restricted to the province: Golmanella, Harleya, Plagiolophus and Asemnantha. Endemic species follow a particular distribution pattern, geographically divided in three parts: a northern ‘‘belt’’, with species such as Ipomea sororia (Convolvulaceae) and Mammilaria

IBS SPECIAL

The ecological features and biogeography of the present-day Yucatan peninsula show the indelible mark of its 65 million yr history, as well as the evident effect of modern human activity. Politically, the peninsula comprises the entire territory of the states of Campeche, Yucatan and Quintana Roo. From a geomorphological point of view it also includes Belize, the Peten area of Guatemala and small portions of the Mexican states of Chiapas and Tabasco (Fig. 2). The limestone bedrock determines a terrain that is typical karst, dominated by a low and relatively flat plain of porous limestone with little soil. The highest point in the north is only 250 m (750 ft) in the Sierrita de Ticul. Surface water, in the form of small lakes and rivers, is confined to the southern part of the peninsula. In the north, all water reservoirs are underground, where there is a complex freshwater saltwater interface (Escolero et al. 2007). The karst is also characterized by a large number of caves and cenotes (water-filled sinkholes) such as those at the rim of the Chicxulub crater (Perry et al. 1995, Schulte et al. 2010), that provide unique habitats for plants and animals (Arita 1996, MacSwiney et al. 2009, Va´zquez-Domı´nguez et al. 2009). Most of the region is warm and subhumid, but climate follows a pattern from dry in the north-northwest of the peninsula to very humid in the south-southeast. Temperature and rainfall vary from high mean annual temperatures (268C) and low annual rainfall (500 mm) in the northwest to lower temperatures and more abundant rainfall in the southeast (1400 2000 mm; Orellana et al. 2003).

Throughout most of the peninsula there is a very well defined rainy season from June to October, although winter rains are not uncommon in the south. Proximity to the Tropic of Cancer and the influence on the region of the Atlantic Bermuda-Azores anticyclone create both a high atmospheric activity and a strong north to south gradient of atmospheric pressure. This, together with the effect of trade winds and the influence of tropical perturbations allow the formation of hurricanes, a defining climatic feature of the whole Caribbean (Orellana et al. 2003). Because of its position, the Yucatan is hit harder and with higher frequency by hurricanes on the east coast, contributing to the east-to-west gradient of humidity that determines the physiognomy, phenology and structure of the vegetation of the peninsula. Hence, vegetation also follows a SE-NW gradient, from tropical rainforests in the Peten to tropical scrubland in the extreme NW portion of the peninsula. Extensive areas between these two extremes were originally covered with deciduous or semideciduous tropical forests (Carnevali et al. 2003).

ISSUE

IBS SPECIAL

heyderi spp. gaumeri (Cactaceae); a southern zone with dominant species like Golmanella sarmentosa (Celastraceae), Epidendrum martinezii (Malvaceae) and Maytenus schippii (Orchidaceae); and a widespread component including species like Hampea trilobata (Malvaceae) or Acacia gaumeri (Carnevali et al. 2003). Vertebrate assemblages in the Yucatan peninsula are rich in species, but less diverse than comparable regions in central and western Mexico (Lee 1980, Arita 1997, Zambrano et al. 2006). Nonetheless, high phylogenetic and taxonomic diversity characterise the region when compared with some areas of Central and tropical South America and Africa (Schipper et al. 2008). Few endemics, low richness of restricted species and higher representation of wide ranging species are also noticeable patterns (Arita and Rodriguez 2002, Arita and Va´zquez-Domıńguez 2003, Schipper et al. 2008). Finally, another defining feature of the peninsula is its low beta diversity when compared with other regions in Mexico (Arita and Rodriguez 2002). This is a consequence of the ‘‘everyone is everywhere’’ distribution pattern, shown most clearly by mammals, which in turn is the result of the peninsula’s simple topography, lack of geographical barriers and low habitat heterogeneity. Simpson’s peninsula effect a decrease in species diversity from the base to the tip of peninsulas is clearly observed in the Yucatan peninsula (Simpson 1964). This is more evident for mammals, which vary in number from around 130 species in the base to 90 in the tip; for frogs, with 22 species in the base and nine in the north (Lee 1980), and for bats, with 85 species present in the south and only 31 in the north (Arita 1997). Likewise, the flora follows a conspicuous diversity pattern along the SE-NW rain gradient; the humid communities in the south having more species than their northern counterparts. An exception is seen for snakes and lizards, which are less diverse at the centre of the peninsula, and increase their richnesss towards the tip (Lee 1980). Endemics include ca 20 reptiles, seven birds and 10 mammals. Richness of endemic amphibian and reptilian species follows an inverse ‘‘peninsula’’ pattern in which more endemic species occur in the north than in the south. In contrast, there is no distinctive gradient of richness of endemic mammals and birds, most of which are widely distributed within the peninsula (Arita and Va´zquezDomıńguez 2003). Many endemic mammals, for instance, are distributed all over the peninsula and sometimes marginally to the piedmont of the highlands of Chiapas and Guatemala. Species showing this pattern include the Yucatan yellow bat Rhogeessa aeneus, the Yucatan squirrel Sciurus yucatanensis, Hatt’s vesper rat Otonyctomys hatti, and the Yucatan black howler monkey Alouatta pigra. All these particular biogeographic traits make the Yucatan an important place for conservation strategies despite the moderate absolute species richness of the region. For example, parts of the Yucatan have been identified as priority areas for the conservation of trees of the tropical deciduous forest (Cue-Bar et al. 2006) and carnivores (Valenzuela-Galvań and Va´zquez 2008). Likewise, the Yucatan peninsula is a hotspot for endemic helminth parasites of freshwater fishes (Aguilar-Aguilar et al. 2008).

218

Conclusion Biogeography is by necessity a historical science, in the sense that present-day patterns of diversity and distribution of species cannot be understood without considering the geological and evolutionary history of the region. In today’s Yucatan, patterns are the result of a wide variety of processes that have shaped the environments of the peninsula at different time scales, from 65 million yr to a few decades. Moreover, a full understanding of those processes is necessary to face the present and future conservation challenges posed by the complex intermixing of natural and social elements that characterize the peninsula. As Danish philosopher Soren Kierkegaard once wrote, life can only be understood backwards but it must be lived forwards.

References Aguilar-Aguilar, R. et al. 2008. Richness and endemism of helminth parasites of freshwater fishes in Mexico. Biol. J. Linn. Soc. 94: 435 444. Alroy, J. et al. 2008. Phanerozoic trends in the global diversity of marine invertebrates. Science 321: 97 100. Alvarez, L. W. et al. 1980. Extraterrestrial cause for the Cretaceous-Tertiary extinction. Experimental results and theoretical interpretation. Science 208: 1095 1108. Arita, H. T. 1996. The conservation of cave-roosting bats in Yucatan, Mexico. Biol. Conserv. 76: 177 185. Arita, H. T. 1997. Species composition and morphological structure of the bat fauna of Yucatan, Mexico. J. Anim. Ecol. 66: 83 97. Arita, H. T. and Rodriguez, P. 2002. Geographic range, turnover rate and the scaling of species diversity. Ecography 25: 541 550. Arita, H. T. and Va´zquez-Domıńguez, E. 2003. Fauna y la conformacioń de la provincia bio´tica yucateca: Biogeografıá y macroecologıá. In: Colunga-Garcıá Marıń, P. and Larque´Saavedra, A. (eds), Naturaleza y sociedad en el a´rea maya, pasado, presente y futuro. Academia Mexicana de Ciencias, CICY, pp. 60 80. Arroyo-Cabrales, J. and .A´lvarez, T. 2003. A preliminary report of the late Quaternary mammal fauna from Lotuń Cave, Yucatań, Mexico. In: Schubert, B. W. et al. (eds), Ice Age cave faunas of North America. Indiana Univ. Press and Denver Museum of Nature and Science, pp. 262 272. Blanchon, P. et al. 2009. Rapid sea-level rise and reef backstepping at the close of the last interglacial highstand. Nature 458: 881 885. Bottke, W. F. et al. 2007. An asteroid breakup 160 Myr ago as the probable source of the K/T impactor. Nature 449: 48 53. Carnevali, G. et al. 2003. Flora y vegetacioń de la penıńsula de Yucatań. In: Colunga-Garcıá Marıń, P. and Larque´Saavedra, A. (eds), Naturaleza y sociedad en el a´rea maya, pasado, presente y futuro. Academia Mexicana de Ciencias, CICY, pp. 53 68. Chakrabarty, P. 2006. Systematics and historical biogeography of Greater Antillean Cichlidae. Mol. Phylogenet. Evol. 39: 619 627. Cue-Bar, E. M. et al. 2006. Identifying priority areas for conservation in Mexican tropical deciduous forest based on tree species. Interciencia 31: 712 719. Da´valos, L. M. 2004. Phylogeny and biogeography of Caribbean mammals. Biol. J. Linn. Soc. 81: 373 394.

219

ISSUE

MacSwiney, M. C. et al. 2009. Insectivorous bat activity at cenotes in the Yucatan Peninsula, Mexico. Acta Chiropt. 11: 139 147. Marshall, C. R. and Ward, P. D. 1996. Sudden and gradual molluscan extinctions in the latest Cretaceous of western European Tethys. Science 274: 1360 1363. Marshall, L. G. et al. 1982. Mammalian evolution and the Great American Interchange. Science 215: 1351 1357. Morrone, J. J. 2005. Hacia una sıńtesis biogeogra´fica de Me´xico. Rev. Mex. Biodivers. 76: 207 252. Orellana, R. et al. 2003. Presente, pasado y futuro de los climas de la penıńsula de Yucatań. In: Colunga-Garcıá Marıń, P. and Larque´-Saavedra, A. (eds), Naturaleza y sociedad en el a´rea maya, pasado, presente y futuro. Academia Mexicana de Ciencias, CICY, pp. 37 52. Parkinson, C. L. et al. 2000. Phylogeography of the pitviper clade Agkistrodon: historical ecology, species status, and conservation of cantils. Mol. Ecol. 9: 411 420. Perry, E. et al. 1995. Ring of cenotes (sinkholes), northwest Yucatan, Mexico its hydrogeologic characteristics and possible association with the Chicxulub impact crater. Geology 23: 17 20. Pringle, H. 2009. A new look at the Mayas’ end. Science 324: 454 456. Raup, D. M. and Sepkoski, J. J. 1982. Mass extinctions in the marine fossil record. Science 215: 1501 1503. Ricklefs, R. E. and Bermingham, E. 2008. The West Indies as a laboratory of biogeography and evolution. Phil. Trans. R. Soc. B 363: 2393 2413. Schipper, J. et al. 2008. The status of the World’s land and marine mammals: diversity, threat, and knowledge. Science 322: 225 230. Schulte, P. et al. 2010. The Chicxulub asteroid impact and mass extinction at the Cretaceous Paleogene Boundary. Science 327: 1214 1218. Sharer, R. and Traxler, L. 2005. The ancient Maya, 6th ed. Stanford Univ. Press. Simpson, G. G. 1964. Species density of North American recent mammals. Syst. Zool. 12: 57 73. Simpson, G. G. 1980. Splendid isolation: the curious history of South American mammals. Yale Univ. Press. Valenzuela-Galvań, D. and Va´zquez, L. B. 2008. Prioritizing areas for conservation of Mexican carnivores considering natural protected areas and human population density. Anim. Conserv. 11: 215 223. Va´zquez-Domıńguez, E. et al. 2009. Contrasting genetic structure in two codistributed freshwater fish species inhabiting highly seasonal systems. Rev. Mex. Biodivers. 80: 181 192. Va´zquez-Miranda, H. et al. 2007. Biogeographical patterns of the avifaunas of the Caribbean Basin Islands: a parsimony perspective. Cladistics 23: 180 200. Vester, H. F. M. et al. 2007. Land change in the southern Yucatan and Calakmul Biosphere Reserve: effects on habitat and biodiversity. Ecol. Appl. 17: 989 1003. Webb, S. D. 2006. The great American biotic interchange: patterns and processes. Ann. Mo. Bot. Gard 93: 245 257. Zambrano, L. et al. 2006. Fish community structure in freshwater karstic wetlands of the Yucatan peninsula, Mexico. Ichthyol. Explor. Freshwaters 17: 193 206.

IBS SPECIAL

Da´valos, L. M. 2006. The geography and diversification in the mormoopids (Chiroptera: Mormoopidae). Biol. J. Linn. Soc. 88: 101 118. Diamond, J. M. 2005. Collapse: how societies choose to fail or succeed. Viking. Escolero, O. et al. 2007. Dynamic of the freshwater saltwater interface in the karstic aquifer under extraordinary recharge action: the Merida Yucatan case study. Environ. Geol. 51: 719 723. Fa, J. E. and Morales, L. E. 1993. Patterns of mammalian diversity in Mexico. In: Ramamoorthy, R. et al. (eds), Biological diversity of Mexico: origins and distribution. Oxford Univ. Press, pp. 319 361. Ferrusquia-Villafranca, I. 2003. Mexico’s middle Miocene mammalian assemblages: an overview. In: Flynn, L. (ed.), Vertebrate fossils and their context: contributions in honor of Richard H. Tedford. American Museum of Natural History, pp. 321 347. Goldman, E. A. and Moore, R. T. 1945. The biotic provinces of Mexico. J. Mammal. 26: 347 360. Haug, G. H. et al. 2003. Climate and the collapse of Maya civilization. Science 299: 1731 1735. Hedges, S. B. 2006. Paleogeography of the Antilles and origin of West Indian terrestrial vertebrates. Ann. Mo. Bot. Gard. 93: 231 244. Heinicke, M. P. et al. 2007. Major Caribbean and Central American frog faunas originated by ancient oceanic dispersal. Proc. Nat. Acad. Sci. USA 104: 10092 10097. Hildebrand, A. R. et al. 1991. Chicxulub crater: a possible Cretaceous/Tertiary boundary impact crater on the Yucatań Peninsula, Mexico. Geology 19: 867 871. Hodell, D. A. et al. 1995. Possible role of climate in the collapse of Classic Maya civilization. Nature 375: 391 394. Hodell, D. A. et al. 2007. Climate and cultural history of the northeastern Yucatan Peninsula, Quintana Roo, Mexico. Clim. Change 83: 215 240. Hunter, R. L. et al. 2008. Phylogeny and historical biogeography of the cave-adapted shrimp genus Typhlatya (Atyidae) in the Caribbean Sea and western Atlantic. J. Biogeogr. 35: 65 75. Iturralde-Vinent, M. A. and MacPhee, R. D. E. 1999. Paleogeography of the Caribbean region: implications for Cenozoic biogeography. Bull. Am. Mus. Nat. Hist. 238: 1 95. Keller, G. et al. 2004. Chicxulub impact predates the K-T boundary mass extinction. Proc. Nat. Acad. Sci. USA 101: 3753 3758. Kelly, T. C. 1993. Preceramic projectile-point typology in Belize. Ancient Mesoamerica 4: 205 227. Kring, D. A. 2007. The Chicxulub impact event and its environmental consequences at the Cretaceous-Tertiary boundary. Paleogeogr. Paleoclimatol. Paleoecol. 255: 4 21. Lee, J. C. 1980. An ecogeographic analysis of the herpetofauna of the Yucatań Peninsula. Univ. Kansas Misc. Publ. 67: 1 75. Lozano-Garcıá, M. S. et al. 2007. Tracing the effects of the Little Ice Age in the tropical lowlands of eastern Mesoamerica. Proc. Nat. Acad. Sci. USA 104: 16200 16203. MacFadden, B. J. 2006. Extinct mammalian biodiversity of the ancient New World tropics. Trends Ecol. Evol. 21: 157 165.

Ecography 33: 220 231, 2010 doi: 10.1111/j.1600-0587.2010.06282.x # 2010 The Author. Journal compilation # 2010 Ecography Subject Editor: Douglas A. Kelt. Accepted 29 April 2010

Tectonics, topography, and mammalian diversity Catherine Badgley C. Badgley (cbadgley@umich.edu), Dept of Ecology and Evolutionary Biology and Museum of Paleontology, Univ. of Michigan, Ann Arbor, MI 48109, USA.

IBS SPECIAL

ISSUE

Terrestrial vertebrates show striking changes in species richness across topographic gradients. For mammals, nearly twice as many species per unit area occur in topographically complex regions as in adjacent lowlands. The geological context of this pervasive biogeographic pattern suggests that tectonic processes have a first-order impact on regional diversity. I evaluate ecological, evolutionary, and historical influences of tectonics and topography on the regional diversity of terrestrial mammals, focusing on the hypothesis that diversification rates are higher in active versus passive tectonic settings. Ten predictions follow from this hypothesis. 1) The timing of peaks in speciation should be congruent with the timescale for tectonic episodes. 2) The rates of speciation and genetic differentiation of populations should be greater for species inhabiting topographically complex regions than spatially continuous landscapes. 3) If topographic complexity per se promotes diversification, then a cluster of young divergences should occur for montane species compared to lowland relatives. 4) Endemism in tectonically active regions should reflect origination within the region rather than range reduction from larger areas. 5) Extinction rates should differ for lineages in tectonically active regions compared to adjacent lowlands. 6) The relationship between local and regional species richness should differ between topographic settings because of higher beta diversity in topographically complex regions. 7) Species originating in topographically complex regions should colonize adjacent lowlands more often than the reverse pattern. 8) North-south mountain ranges should have higher regional species richness than east-west mountain ranges. 9) Areas with multiple mountain ranges should have higher regional species richness than comparable areas with single mountain ranges. 10) Global climate changes should affect diversification in tectonically active regions. Research addressing these topics places elevational diversity gradients into a geohistorical context and integrates data from modern biotas and the fossil record.

One of the most striking patterns in biogeography is the high species richness of terrestrial vertebrates in topographically complex regions compared to adjacent regions of low relief. The pattern involves both the accumulation and spatial turnover of species along steep elevational and environmental gradients, resulting in high regional species richness. The association between complex topography and high species richness has been documented for many groups of terrestrial vertebrates (mammals, birds, amphibians) and vascular plants, as well as for different continental regions (Qian and Ricklefs 2008). The geological context of this diversity pattern suggests that tectonic and associated erosional processes, which create gradients in topographic complexity, have a first-order impact on regional diversity. In this paper, I evaluate the potential causes of elevated richness of terrestrial mammals in areas of high versus low topographic complexity resulting from different tectonic histories, with data from modern and fossil mammalian faunas and lineages. I review ecological, evolutionary, and historical processes that could determine the major features of the general pattern. In particular, I focus on the hypothesis that evolutionary processes affecting speciation, extinction, dispersal, and adaptation differ in active versus 220

passive tectonic settings. Several predictions that follow from this hypothesis are evaluated in light of current data and suggest directions for new research.

Background Mountainous regions are well known to harbor greater species richness than adjacent lowland areas for many groups and regions, resulting in spectacular diversity hotspots for terrestrial vertebrates (Humboldt 1805, Simpson 1964, Rahbek and Graves 2001, Sechrest et al. 2002, Grenyer et al. 2006, Wiens et al. 2007, Thomas et al. 2008). Increasing topographic complexity creates new habitat, enlarges environmental gradients, establishes barriers to dispersal, and isolates populations, potentially contributing to adaptation to new environmental conditions and speciation in excess of extinction for terrestrial organisms. For freshwater fishes, topographic complexity reduces habitat area and connectedness and results in elevated extinction rates (Smith et al. pers. comm.). Although this idea has old roots (Simpson 1964, Cracraft 1985, Moritz et al. 2000, Brown 2001), it has received far

less attention than latitude and the associated environmental and historical correlates in relation to taxonomic and ecological diversity; and predictions associated with the influence of topography remain to be adequately tested. For mammals, the association of high species richness with regions of complex topography has been documented on several continents. In the first continent-wide analysis of mammalian biogeography, Simpson (1964) highlighted latitudinal and longitudinal gradients in species density of extant North American mammals. The longitudinal gradient follows topographic complexity (Fig. 1). Simpson (1964, p. 69) noted that ‘‘where there are latitudinal gradients, these are additive with topographic gradients, the two accounting for most of the pattern’’. Kerr and Packer (1997) documented the predictive power of topography, interpreted as a surrogate for habitat heterogeneity, (A)

West

East

2500

Elevation (m)

2000 1500 1000 500 0 130

120

110

100

W Longitude

(B)

120

Species richness

110 100 90 80 70 60 50 40 130

120

110

100 W Longitude

120

Species richness

110 100 90 80 70 60 50

r = 0.85

40 500 1000 1500 2000 2500 Elevation (m)

Figure 1. A transect across North America at 408N demonstrates the strong correlation between elevation and mammalian species density. Data from equal-area quadrats in Badgley and Fox (2000). (A) Average elevation of quadrat in meters above sea level. (B) Species density (including bats) per quadrat. (C) Scatterplot and Pearson correlation coefficient between elevation and species density.

221

ISSUE

IBS SPECIAL

(C)

on species density of North American mammals. Badgley and Fox (2000) showed that climatic and topographic variables predict most of the variation in both species density and ecological structure of North American mammals. South American mammals also show a strong longitudinal gradient: species density across the Andes is twice as great as in the Amazon Basin at the same latitude (Patterson et al. 2005, InfoNatura 2007). In Europe, spatial clustering of mammalian species density and ecological structure reflects climate and physiography (Heikinheimo et al. 2007). In equatorial Africa, mammalian faunas (excluding bats) of the East African Rift system harbor 106 122 species, whereas faunas from the Congo Basin contain 56 78 species (Badgley unpubl.). Most Australian mammal species inhabit the coastal mountains and dissected plateaus of eastern and northeastern Australia compared to the vast lowland desert interior (data from Strahan 1995). Although the documentation of Asian mammals is uneven, regional compilations show exceptionally high species richness and extinction risk in the eastern Himalayas and mountainous peninsulas of southeastern Asia compared to adjacent lowlands or high plateaus (Sechrest et al. 2002, Schipper et al. 2008). Where the mammalian fossil record permits comparison of coeval assemblages from nearby lowlands and uplands, species richness is greater in the upland assemblage. In an early Cenozoic example, Gunnell and Bartels (2001) compared well sampled, Middle Eocene mammalian faunas from basin-center and basin-margin areas of the Green River Basin in Wyoming, USA. The basin-margin site at the edge of actively rising mountains contained substantially greater species richness of mammals and other vertebrate groups than did coeval assemblages from the basin center. Presumably, the higher richness reflected the ecotone between lowland and upland habitats. Several groups of mammals showed evidence of speciation in situ (ancestordescendant pairs or sister species with an ancestral species in an older interval). A late Cenozoic example contrasts fossil assemblages from the Miocene of Pakistan and southwestern China. In floodplain sediments from the Himalayan foreland basin of Pakistan, the greatest species richness of mammalian assemblages over a 12-million yr period was 70 species documented at 10 Ma (million years ago, unpubl., Barry et al. 2002). In contrast, a late Miocene fossil locality at the same latitude from a montane valley in Yunnan Province (southwestern China) preserved ca 120 species (Badgley et al. 1988, Z. Qiu pers. comm.). Both localities have mammals indicative of mesic forests and are documented by 1000 fossil specimens. Even more compelling are increases in species richness following tectonic episodes that altered topographic gradients or geographic barriers. Barnosky and Carrasco (2002) and Kohn and Fremd (2008) documented increases in regional species or generic richness of mammals from the montane western United States in step with Middle Miocene tectonic extension, volcanism, and rifting. These modern and historical examples document high mammal richness in tectonically active regions compared to nearby lowland environments and motivate this inquiry into the different processes that have shaped this widespread pattern.

IBS SPECIAL

ISSUE

Evolutionary consequences of different tectonic regimes The focus of this paper is the hypothesis that tectonically active areas, such as active mountain belts and rift valleys, are engines of mammalian diversity by increasing diversification rates relative to those in tectonically quiescent regions. I am not concerned with species richness or turnover within elevational gradients of individual mountains or mountain ranges per se, but with processes that generated differences in species richness, taxonomic composition, and ecological diversity between major geologic provinces. The spatial scale of mammalian geographic ranges, ecoregions, and life zones is the scale over which the evolutionary component of diversity gradients is most evident (Ricklefs and Schluter 1993). Tectonic activity converts landscapes of low elevation, low relief, and spatially continuous climate and vegetation into landscapes of high relief with steep climatic gradients and fragmented vegetation zones. These transformations occur over thousands to millions of years and are driven by plate subduction, rifting, and mantle hotspots (Windley 1995). Complex topography may result from deformation at convergent (e.g. Andes, Himalayas) or divergent (e.g. East African Rift, Rio Grande Rift) plate boundaries. Changes in elevation and relief alter climatic and atmospheric conditions, habitat area, and connectivity among habitats. For terrestrial mammals, the operational properties are continuity and area of habitats, proximity and connectivity among areas of similar habitat, strength of environmental and resource gradients, and topographic stability of habitats (Coblentz and Riitters 2004). Topography interacts with climate at many spatial scales, from the microhabitats of hillsides to entire mountain ranges and massive plateaus. Changes in temperature, atmospheric pressure, and humidity with elevation mimic latitudinal climatic changes, such that alpine tundra and high-elevation deserts have climatic, edaphic, and floral parallels with arctic tundra and polar deserts (Barbour and Billings 1999). The altitudinal temperature gradient changes with latitude and global temperature. Today, for example, the snowline occurs at 4500 m at the equator, rises toward the Tropics of Cancer and Capricorn, then falls with increasing latitude, reaching sea level near the poles. Snowlines fell during the last glaciation, by up to 1000 m in the tropics (Porter 2001), rose during the current interglacial, and are rising further from modern global warming (IPCC 2007). Both climate and physiography influence the effectiveness of high mountains or plateaus as barriers to dispersal. Janzenâ&#x20AC;&#x2122;s (1967) assertion that mountain passes are higher in the tropics refers to the higher extension of habitable life zones in tropical versus temperate regions, lower seasonal temperature variations throughout the tropics, and different thermal tolerances of tropical versus temperate and boreal organisms. The predictions about thermal tolerances have been borne out better for ectotherms than for endotherms (Ghalambor et al. 2006). Nonetheless, for birds (Rahbek and Graves 2001) and mammals (McCain 2005), climatic conditions effectively sort species along elevational gradients. Thus, topography cannot be fully separated from climate and climatic history in its ecological or evolutionary influences on diversity. 222

The processes that increase regional (and local) species richness are intensified in montane regions. The environmental gradients have the potential to accommodate many kinds of species in superjacent life zones (Merriam 1894, Lomolino et al. 2006). Immigration occurs from adjacent life zones as well as within zones. During Quaternary glacial cycles, areas of montane vegetation expanded during cool intervals, creating low-elevation connections for dispersal (Brown 1978, Grayson 1993, Thompson and Anderson 2000). Geographic isolation of habitats on individual mountains or mountain ranges increases opportunities for genetic differentiation and speciation, as well as extinction (Cracraft 1985, Brown 2001). Steep environmental gradients present strong selection gradients, creating circumstances for disruptive selection to act on contiguous populations (Endler 1977, Moritz et al. 2000). These conditions should affect small mammals more than large ones, because of size-related differences in home-range area and resource requirements in relation to habitat area and barriers. Surveys show that small mammals indeed drive the major gradients in species richness (Patterson et al. 1998, Badgley and Fox 2000, Lomolino 2001, McCain 2005). In contrast, tectonically quiescent areas, such as the ancient shields and passive margins of continents, slowly lose habitats through erosion. Such areas exhibit spatially continuous habitats and low climatic heterogeneity that promote geographically extensive populations with high gene flow. Although such areas can potentially accommodate many species, immigrants from different habitats are far away. The low frequency of strong barriers facilitates high gene flow over large areas, with less opportunity for lineage differentiation or speciation in mammals. Selection gradients related to environmental conditions are weak. During Quaternary glacial cycles, bioclimatic zones were latitudinally compressed during glacial advances and expanded during interglacials (Wright et al. 1993, Williams et al. 2004). Compression of climatic zones during glacial periods would have further enhanced gene flow within and among habitats. This contrast in physiography and Quaternary history should lead to differences in the phylogeographic structure of sister taxa occupying regions of high versus low topographic complexity.

The elevational gradient in North America A North American example highlights the contrast in taxonomic and ecological diversity from different tectonic contexts. North America (including Central America) consists of tectonically active western and southern regions and a tectonically passive eastern region. The mountain ranges, basins, and plateaus of the tectonically active region have been created over the last 100 myr (million years), with current tectonic activity concentrated along the western margin of North America and under mantle hotspots, such as the Yellowstone Hotspot (Burchfiel et al. 1992, Pierce and Morgan 1992). For a given latitude, species richness per unit area is twice as great in most areas of the tectonically active west as in the tectonically quiescent east (Fig. 1A B), and richness is strongly correlated with elevation (Fig. 1C). (Active and passive regions occupy comparable areas.) Likewise, certain trophic groups, such as

herbivores and granivores, show high montane species richness, whereas others, notably carnivores and omnivores, do not (Badgley and Fox 2000). Taxonomically, rodents and bats dominate the latitudinal richness gradient, whereas rodents dominate the longitudinal gradient. Rodents constitute just over half of extant North American mammal species (Hall 1981, Wilson and Reeder 2005). Sixty-two percent of rodent geographic ranges lie within the active region compared to 12% within the passive region, while 26% have ranges overlapping both regions (Table 1). Four families cricetids (muroids including voles, deermice, and packrats), geomyids (pocket gophers), heteromyids (pocket and kangaroo mice), and sciurids (squirrels) dominate North American rodent diversity. All four families have more than twice as many species occurring only in the active region compared to the passive region, with about one-fourth or fewer species occurring in both regions. This consistent geographic pattern suggests a difference in macroevolutionary processes between the active and passive regions. At a finer scale, the mammals of Colorado illustrate species richness and turnover along the boundary between active and passive regions (Fig. 2). A strong elevational gradient spans this boundary. Based on detailed documentation of species ranges throughout the state (Fitzgerald et al. 1994), I recorded the eastern and western range limits of all species within 1-degree bands of longitude (ca 100 km wide) for mammals inhabiting the northern half of Colorado (Fig. 2A). Since the Rocky Mountains run northsouth, the longitudinal range boundary is also an elevational boundary. Twenty-six species have ranges principally in the Great Plains (eastern Colorado), with some ranges extending into the Rocky Mountain front (central Colorado); 69 species have ranges within the Rocky Mountains and the plateaus of western Colorado, with some ranges extending further west; 30 species occur throughout Colorado (Fig. 2A). Species richness rises from the Great Plains, peaks at the Rocky Mountain front, and declines slightly further west (Fig. 2B C). The number of range boundaries, a measure of spatial turnover, shows a unimodal pattern, with the highest value at the Rocky Mountain front (Fig. 2D). Notably, the number of range boundaries west of the mountain front is not significantly greater than the number

Active region Passive region Both regions 241 1 117 1 1 1 21 39 60

47 21 1 1 10 5 9

99 1 1 53 1 2 2 2 6 12 18

A number of predictions pertaining to evolutionary and ecological processes follow from the hypothesis that tectonic history is a driver of mammalian diversification. Ten of these are given below with representative data from the neontological and paleontological literature, when available. When neither data nor literature are available, I provide suggestions for appropriate tests. Timing of speciation The timescale for peaks in mammalian speciation should be consistent with the timescale for tectonic changes in topography. Peaks in origination should correspond to peaks of tectonic activity if the increase in topographic complexity forms barriers, fragments populations, and strengthens selection gradients. Alternatively, if montane regions are simply accommodating species that have immigrated from lowland as well as montane regions, then no correspondence between tectonic history and origination rates should occur, thereby falsifying the prediction. Data required to test this prediction include well resolved phylogenies with robust estimates of divergence times for the lineages under consideration, fossils of the focal lineage to document the timing of origination, and geochronological data about the timing of tectonism. Since small mammals drive the major features of the diversity trends, the ideal analyses would involve fossil rodents or insectivores from paired tectonically active and passive regions (Finarelli and Badgley 2010). (Fossil bats are too scarce for such comparisons.) Two studies of mid-Cenozoic mammals demonstrate the potential for comparing origination rates in different tectonic settings (Barnosky and Carrasco 2002, Kohn and Fremd 2008), although the latter analyzed changes in generic richness. Speciation and genetic differentiation The speciation rate should be greater for mammals in topographically complex regions than for sister taxa in spatially continuous landscapes. Speciation rates that are either systematically higher or statistically similar in spatially continuous landscapes relative to topographically complex regions would falsify the prediction. Fragmentation and isolation of habitats by physiographic barriers and strong resource gradients should promote speciation in 223

ISSUE

All rodents (n 387) Aplodontidae (n 1) Castoridae (n 1) Caviidae (n 1) Cricetidae (n 191) Dasyproctidae (n 2) Dipodidae (n 4) Echimyidae (n 3) Erethizontidae (n 3) Geomyidae (n 37) Heteromyidae (n 56) Sciuridae (n 87)

Predictions and preliminary tests

IBS SPECIAL

Table 1. Number of extant North American rodent species with geographic range entirely within the tectonically active region, the passive region, or overlapping both regions. Geographic-range data from Hall (1981), Patterson et al. (2005), Wilson and Reeder (2005), and InfoNatura (2007).

east of the mountain front, signifying that the sustained increase in species richness across western Colorado results from the accumulation of species (mainly from the west) more than from spatial turnover. These patterns as well as the high beta diversity of mammals (also, birds and amphibians) in mountainous areas of North and South America (McKnight et al. 2007, Melo et al. 2009, Qian et al. 2009) provide strong evidence for a general macroecological pattern. But it is necessary to evaluate evolutionary processes in the context of tectonic and environmental history for a more fundamental understanding of the origins of this gradient.

108

109

(A)

107

106

105

104

103

102 41°N

5 9 3 11 1 15

2 2 2 7

40°N 30’

40°N

9 1 2 1 2

39°N 30’

2 1

13 1 4

39°N

(B) Metres

3000

2000

IBS SPECIAL

ISSUE

(D)

No. range boundaries

(C)

Species richness

1000 108.5

107.5

106.5

105.5

104.5

103.5

102.5

108.5

107.5

106.5 105.5 104.5 Degrees longitude

103.5

102.5

100 90 80 70 60 50 40

40 30 20 10 0

Figure 2. Transect across the northern half of Colorado (USA), spanning the boundary between tectonically active and passive regions of North America. (A) The longitudinal extent of mammal species ranges is indicated by horizontal lines across a topographic base map (from GeoMapApp, <www.geomapapp.org/>). Numbers next to lines signify the number of species with that longitudinal range distribution. Dark lines indicate geographic ranges that are entirely within the mountainous region or extend further west; light lines indicate ranges on the high plains or extending to the east. (B) Topographic profile across northern Colorado at ca 408N (from GeoMapApp). (C) Species richness of mammals based on presence of geographic ranges across 1-degree bands of longitude (ca 100 km wide). (D) Number of range boundaries in each 1-degree band of longitude as a measure of spatial turnover. Geographic-range data from Fitzgerald et al. (1994).

topographically complex regions, especially for species with small home range sizes or stenotopic habitat preferences. Sister species in continuous landscapes should experience higher rates of dispersal and gene flow as well as weak resource gradients, thereby lowering the likelihood of reproductive isolation and allopatric speciation. Data relevant for testing this prediction include phylogenetic and phylogeographic studies of mammals with sister taxa in both montane and lowland landscapes. 224

Two examples involving North American rodents support this prediction. Demastes et al. (2002) conducted a phylogenetic analysis of pocket gophers (Geomyidae) from the Mexican Plateau and the Trans-Mexican Volcanic Belt (TMVB). The authors sampled mtDNA from 38 localities across the geographic ranges of Cratogeomys and Pappogeomys. Their maximum-likelihood tree based on cytochrome b showed a different pattern of cladogenesis on the broad, stable Mexican Plateau compared to the young mountains

and deep valleys of the TMVB. From divergence estimated at 3.2 Ma into two lineages of Cratogeomys, one branch gave rise to two species now occupying the northern part of the Mexican Plateau and one occupying the southeastern part of the Mexican Plateau. The other branch split into five clades (that do not correspond neatly to currently recognized species) that live above 2000 m in the TMVB; the distribution of these clades corresponds to geographic subdivisions of the TMVB by volcanic plateaus and rivers. Estimated divergence times imply that since 2.6 Ma, two clades arose in the northern Mexican Plateau and five clades arose in the TMVB, during an interval of volcanic activity in the TMVB and glacial-interglacial climatic changes. Based on extant species, cladogenesis was about twice as great in the volcanically active TMVB as in the stable Mexican Plateau. The second example involves Tamias amoenus (yellowpine chipmunk), which occupies montane conifer forests in northwestern North America, and T. striatus (eastern chipmunk) from eastern North America. Phylogeographic analysis based on sequence variation in cytochrome b revealed 12 clades in T. amoenus (Demboski and Sullivan 2003). The distribution of haplotypes corresponds closely with different mountain ranges in several distinct geological provinces. The degree of sequence divergence (4.5 7.4%) implies differentiation before the mid-Quaternary. In contrast, the phylogeographic structure of Tamias striatus has no concordance with landscape features but indicates population expansion since the last glacial maximum from multiple refugia in the eastern United States (Rowe et al. 2006). Thus, the geographic structure and temporal depth of mtDNA variation differ in congeners that occupy topographically complex versus simple landscapes.

Endemism

Species ages

Extinction rates

Two scenarios need evaluation. First, if montane regions continuously promote speciation, then higher diversification rates should result in a cluster of young divergences compared to lowland relatives. Alternatively, if tectonic activity per se stimulates speciation, then peaks in origination should coincide with the timing of tectonism, irrespective of age. These alternatives could be evaluated with a high-quality fossil record that spans intervals of tectonism. Geographic differences in the timing of divergence of extant species could also test these scenarios. Moritz et al. (2000) document younger species-level divergence ages for Andean small mammals compared to their Amazonian relatives. For example, in murid rodents, the mean genetic distance between sister taxa in the Amazon is twice as large as the mean genetic distance between sister taxa in the Andes, implying that Andean rodent faunas contain younger species. This example supports the first scenario in which montane regions systematically promote speciation, since divergences occurred over millions of years in both regions. If younger species consistently occur in tectonically quiescent versus active areas, or if speciation rate does not change in response to an episode of tectonic activity, then the general hypothesis is falsified.

How mammalian extinction rates compare between topographically complex regions and lowlands is an open question, since topographically complex regions have opposing influences on extinction rates. Three scenarios are possible. First, extinction rates could be higher in topographically complex regions than in lowlands. Diversification could be still higher in the former if per-lineage speciation rates exceed extinction rates. This scenario implies that the average persistence time of mammalian lineages would be shorter and faunal turnover higher in montane regions than in lowlands. The smaller, often fragmented geographic ranges and large distances between areas of suitable habitat in topographically complex regions should elevate extinction risk for non-volant mammal populations (Brown 2001). Isolation of low- to midelevation vegetation zones by high elevation or barren stretches in extensive, linear mountain ranges or of alpine regions on smaller, separated mountain ranges (such as in the Great Basin) should reduce dispersal and increase extinction risk, as is presently occurring for Ochotona princeps (pikas) of the western United States (Grayson 2005). Second, mammalian extinction rates could be similar between topographically complex regions and their adjacent lowlands. Under this scenario, montane

Endemism in tectonically active regions should reflect cladogenesis within the montane region rather than contraction of geographic range(s) from a much larger region. Endemism can result from alternate historical trajectories origination in a region and persistence in that geographic region alone, origination in one region and relocation to an entirely different area, or reduction of geographic range to produce a relict distribution (Lomolino et al. 2006). Endemism resulting from the second or third processes would not support the hypothesis. The geographic isolation of the areas under consideration has a substantial impact on geographic-range expansion, immigration, and endemism. In their comparison of vertebrate faunas from two montane ecosystems of comparable area, Yellowstone National Park in the northern Rocky Mountains and two national parks in northern Patagonia, Barnosky et al. (2001) noted lower mammalian species richness and higher levels of endemism due to origination in situ in Patagonia compared to Yellowstone. They attributed both properties to the isolation of Patagonia from other temperate sources. Well resolved phylogenies can demonstrate whether geographically clustered endemics are also closest relatives, supporting the scenario of endemism reflecting origination and persistence. In addition, data from the fossil record documenting mammalian faunal composition from both montane and adjacent lowland areas over time, as well as indicators of paleoenvironmental history, are needed to evaluate the patterns of endemism in tectonically active regions. Areas with a fossil record of montane and adjacent lowland species include the western U.S. (Carrasco et al. 2005, Janis et al. 2008) and Ethiopia (Yalden and Largen 1992), each with many endemic mammals today.

IBS SPECIAL ISSUE

225

diversification would be driven by speciation and immigration. Third, extinction rates could be lower in topographically complex regions than in lowlands, despite differences in the size and fragmentation of geographic ranges. Under this scenario, species persistence times would be greater and faunal turnover lower in topographically complex regions. Areas of complex topography feature steep environmental gradients with many life zones in close geographic proximity, which may facilitate dispersal among suitable microhabitats. In their analysis of Quaternary vegetation change on the Andean flank, Bush et al. (2004) noted that the rate of change in forest composition was similar during intervals of substantial climatic change and climatically stable periods. They postulated (p. 828) ‘‘that the ease with which species can migrate and avoid extinction in response to climate change may be a major factor promoting diversity in these [montane] systems’’, suggesting that extinction risk is reduced in topographically complex regions. Extinction rates in topographically complex regions should also differ between cooling and warming periods. During intervals of global cooling, extinction rates should decline as ranges expand, isolation is reduced, and dispersal increases (Brown and Kodric-Brown 1977, Brown 1978). During periods of warming, extinction rates should increase as high-elevation habitats shrink and vanish. The North American mammal record has numerous fossil localities across the tectonically active montane regions and tectonically quiescent plains over much of the Cenozoic (Janis et al. 1998, 2008, Carrasco et al. 2005) with high taxonomic and temporal resolution suitable for testing these scenarios.

IBS SPECIAL

ISSUE

Local versus regional species richness If tectonics drives diversification, then regional mammalian richness in tectonically active regions should comprise a set of distinctive local faunas. Beta diversity should be high across topographically complex regions because of vicariance and strong environmental gradients. In contrast, lowlands should feature low beta diversity as a consequence of low speciation rates and large geographic ranges. Local and regional richness should have a strong positive correlation, and local richness and species composition (per unit area) should vary little across the landscape. These principles also apply to mammals of different body size and dispersal ability inhabiting each region. For large-bodied species with large home ranges and high mobility, local and regional species richness should be similar. For small-bodied species with small home ranges and short dispersal distances, local species richness should represent a small fraction of regional species richness. This relationship remains to be tested empirically. On the other hand, the close proximity of many life zones and moderate dispersal capabilities of most mammals lead to higher local species density in topographically complex regions than in topographically continuous ones (Lomolino 2001). Dispersal into adjacent life zones should also enrich lowland faunas close to montane source areas, an effect evident for mammals of Colorado (Fig. 2A) from the montane front into the Great Plains. 226

Export of diversity Topographically complex regions should export species to adjacent lowlands. This pattern could arise when a species originates in the tectonically active region and then expands or shifts its geographic range to the passive region. Higher species richness in topographically complex regions could reduce the likelihood of immigrations from adjacent regions with higher richness serving as a barrier to immigrants (Shea and Chesson 2002). Likewise, species from regions of high diversity could be more successful immigrants to other regions because they experience more biotic interactions in the area of origin, as proposed for reef biodiversity in topographically complex marine habitats (Kiessling et al. 2010). For mammals of northern Colorado (Fig. 2), I assessed the proportion of species whose geographic ranges lie predominantly in the topographically complex region but extend onto the high plains, and vice versa (Fitzgerald et al. 1994). For the montane region (largely west of 1058W), 18 species have extended their geographic range eastward to lower elevations. For the plains region (largely east of 1058W), 18 species have extended their range westward, with three species occurring only at lower elevations (below 2000 m) and 15 species expanding into higher elevations. This example does not support the prediction that higher species richness is a barrier to immigration. The prediction could also be addressed with high-resolution fossil records to assess expansion of geographic ranges over geologic time. The evolutionary counterpart to this ecological pattern involves species from montane regions expanding into lowland regions and the lowland populations then evolving into new species. The question is whether montane regions contribute to the diversity of lowland regions through speciation (even if origination rates occur at a higher frequency within the montane members of the clade) to a greater degree than lowland regions contribute to the diversity of montane regions. Well-resolved phylogenies of clades with broad geographic coverage permit comparison of branching patterns in relation to tectonic context. For pocket gophers (Geomyidae, Spradling et al. 2004), pocket mice in the genus Liomys (Heteromyidae, Rogers and Vance 2005), the major heteromyid lineages (Hafner et al. 2007), and deer mice in the genus Peromyscus (Cricetidae, Bradley et al. 2007), there was considerable concordance between branching patterns and tectonic context, based on data from Table 1. For Liomys, older lineages inhabit the tectonically active area more often than both areas, and species inhabiting both areas occur on young branches (Fig. 3). This pattern suggests that colonization proceeded from the active to the passive region, if modern distributions provide accurate information about area of origination. The opposite pattern occurs for geomyines: older lineages include a higher frequency of species occupying passive regions or both regions today, and species on younger branches occur more frequently in tectonically active areas (Fig. 4), suggesting colonization of the active region from the passive region. These patterns are merely suggestive. Since many species origins are estimated to date to the Miocene or Pliocene, and geographic ranges may shift substantially over time, high-resolution fossil records in

Orientation of mountain ranges Regions with north-south mountain ranges should feature higher regional species richness than areas with east-west

ranges. A north-south orientation facilitates latitudinal range shifts of species in response to global temperature changes, as species move along north-south corridors, and lowers extinction risk (Coblentz and Riitters 2004). In contrast, an east-west orientation of mountain ranges inhibits movement of species north and south during global cooling or warming episodes. This prediction can be evaluated by comparing regional species richness of 227

ISSUE

conjunction with well resolved phylogenies are needed to assess these geographic histories.

IBS SPECIAL

Figure 3. Phylogenetic tree of heteromyid species in the genus Liomys from Rogers and Vance (2005). Tree results from a Bayesian analysis of mitochondrial cytochrome b sequence data. Nodes in bold are supported by Bayesian posterior probabilities 95%; values for other nodes are 50%. Shading indicates whether each population occurs in the tectonically active region or both active and passive regions of Mexico. Gray shading active region, white box both active and passive regions. Most lineages occur in the tectonically active region, and those in both active and passive regions are on younger branches, suggesting that colonization proceeded from the active to passive region. Modified from Fig. 2 in Rogers and Vance (2005).

100 100 100 87 100 99 56 87 75 100 100

2.1 mya

C. planiceps clade E C. fumosus clade D

C. goldmani P. bulleri

99 76

100 100

4.2 mya 98 81

100 100 1.6 mya 2.7 mya 100 100

O. cherriei

100 100

O. underwoodi O. hispidus O. grandis West O. grandis East

100 100

4.9 mya

C. fumosus clade A

C. castanops

2.9 mya

set to 5.0 mya

C. fumosus clade C

C. merriami 100 99

C. fumosus clade B

Z. trichopus G. breviceps G. texensis T. bottae

100 97

1.8 mya 2.6 mya 100 96 100 100

T. umbrinuss T. bulbivorus T. talpoides T. mazama T. monticola

0.05 substitutions per site

IBS SPECIAL

ISSUE

Figure 4. Bayesian phylogeny of geomyids from Spradling et al. (2004), based on analysis of mitochondrial and nuclear genes. Estimated divergence dates are shown for some nodes. Shading indicates whether the species or phylogroup occurs in the tectonically active region, the passive region, or both regions. Gray shading active region, white lettering on black passive region, white box both regions. Species in the outgroup (Thomomys, or ‘‘T ’’ species) and older branches occur more often in the passive region or both regions, and younger branches occur mostly in the active region, suggesting that colonizations proceeded from the passive to active regions. Modified from Fig. 4 in Spradling et al. (2004).

European mammals north and south of the Alps with North American mammals at similar latitudes, and also the regional species pools of the east-west trending Uinta Mountains with the contiguous north-south trending Wasatch Mountains in Wyoming and Utah, USA. Brown and Maurer (1989) made a related point in demonstrating that the geographic orientation of small ranges runs northsouth, whereas large ranges are oriented east-west for North American mammals, and predicted that European mammals would show few ranges oriented north-south but many east-west orientations. The comparison remains to be done. A finding of similar or lower regional richness in northsouth compared to east-west mountain ranges would falsify this prediction. Multiple versus single mountain ranges Areas with multiple mountain ranges should have higher regional species richness than areas with single mountain ranges. The expectation is that areas with multiple mountain ranges would generate and exchange species, resulting in higher regional species richness than areas with single mountain ranges (standardized for latitude and area). This prediction can be addressed by comparing species richness from regions (of equal area) with several small ranges in the Great Basin to single large ranges of the 228

eastern Rocky Mountains (e.g. Wind River Range, Wyoming) or to outlying ranges (e.g. Guadalupe Mountains of western Texas). The Ethiopian highlands, a single mountainous complex, compared to the extensive eastern and western African Rift valleys and highlands offer another example. Global climate change and diversification Global warming or cooling should cause geographic ranges to shift in elevation, but do these changes affect diversification? Global warming causes species ranges to shift to higher elevations, thereby increasing the regional species richness within topographically complex regions. Warming also reduces the area and increases the isolation of habitat zones at higher elevations. Fragmentation and isolation of such habitats could promote speciation in mammals if sustained over geologic time. However, it could also promote extinction on mountaintops, as has occurred since the last glacial maximum (Brown 1978) and continues today (Grayson 2005). Global cooling causes species ranges to shift to lower elevations, reducing regional species richness within topographically complex regions and increasing richness in adjacent lowlands. This process may explain why mammalian richness was more similar in the eastern and western US following the last glacial maximum

than it is today (Cannon 2004). If global warming influences diversification in topographically complex regions, then fossil records from such regions should show a pulse of originations during episodes of global warming, with only a modest increase in species richness (from poleward range shifts) in adjacent lowlands. This process should also cause a clustering of divergences corresponding to the warming interval in the phylogeny of clades from tectonically active regions. Absence of diversification during intervals of global warming would falsify this prediction.

Conclusion

References Badgley, C. and Fox, D. L. 2000. Ecological biogeography of North American mammals: species density and ecological structure in relation to environmental gradients. J. Biogeogr. 27: 1437 1467.

229

ISSUE

Acknowledgements I thank G. R. Smith for insights about the concepts in this paper. Participants at the International Biogeography Society 2009 meeting gave useful feedback and stimulating discussion when I presented these ideas. D. A. Kelt and an anonymous reviewer provided useful comments on the manuscript. B. Miljour assisted with figures.

IBS SPECIAL

I have presented a general hypothesis that tectonic activities on the continents create topographic and climatic heterogeneity that directly affects the magnitude of resource gradients and the connectivity versus isolation among habitats. In turn, these environmental properties promote diversification in tectonically active regions and suppress diversification in tectonically quiescent regions. This hypothesis pertains to terrestrial clades (mammals, plants, birds), whereas aquatic clades (freshwater fishes) experience high extinction rates in tectonically active regions and low extinction rates in passive regions (Smith et al. pers. comm.). The evidence presented here, while preliminary, largely supports the hypothesis of tectonics as a driver of mammalian evolution while highlighting the need for more investigation. Three research questions stand out for refinement and testing of this macroevolutionary hypothesis. First, does topographic complexity per se or active tectonism (increase in topographic complexity and elevational gradients) stimulate diversification? Second, do extinction rates differ significantly between tectonically active and passive regions, or is the contrast in species richness driven primarily by speciation? Third, how does global climate change interact with tectonic history to influence diversification? Well-resolved phylogenies and fossil records are needed to address these questions rigorously. This hypothesis and its recommended tests have several implications for the assessment of diversity gradients in extant mammals (and other groups). Most recent analyses have focused on environmental properties as predictors of alpha diversity, ecological structure, beta diversity, and for evaluation of null models of species distribution at local to regional scales (Currie 1991, Badgley and Fox 2000, McCain 2005, McKnight et al. 2007, Qian et al. 2009). Such analyses quantify the sorting of extant species along environmental gradients but consistently reveal diversity anomalies that are not merely a function of the scale of analysis. Topographic variables (elevation, relief, topographic heterogeneity) are consistently strong predictors of species richness; their significance is interpreted as a surrogate for habitat heterogeneity (Kerr and Packer 1997, Coblentz and Riitters 2004, McKnight et al. 2007). This interpretation highlights ecological mechanisms: greater habitat heterogeneity generates increased species packing. The analysis of diversity gradients also needs to address the geographic and temporal origins of modern diversity. The major hypotheses for explaining the

evolutionary dimensions of diversity gradients (Mittelbach et al. 2007) involve area, time, and climatic stability in relation to diversification rates. The significance of these factors changes when placed into a geologic context at the continental scale. Biome or realm area has a different effect on diversification rate if the biome is fragmented or continuous. Duration of a biome or realm as a site of species accumulation depends on regional rates of speciation and extinction, which should differ according to physiographic properties. Climatic stability was rare at all latitudes over the Cenozoic, which witnessed long-term cooling, short-term warming intervals, and Milankovitch cycles throughout (Zachos et al. 2001, Lyle et al. 2008). This idea also has implications for the relevance of the fossil record, which can make fundamental contributions to the assessment of macroevolutionary differences between different tectonic settings. Most of the continental fossil record prior to the Cenozoic is from depositional environments at low elevations in topographically homogeneous landscapes, such as river basins and coastal plains. Cenozoic fossil sites, however, occur in a range of intermontane and low-elevation settings. Comparison of origination rates, extinction rates, and taxonomic richness in adjacent tectonically active and quiescent settings will provide critical tests of this hypothesis (Finarelli and Badgley 2010). Also, this hypothesis offers a plausible explanation for why the earliest representatives of higher taxa of mammals are so often missing from the fossil record. If these early representatives arose at higher elevation in topographically complex areas in the early Cenozoic or late Cretaceous, they would be unlikely to leave a fossil record. Finally, the long durations of lineages and faunas characteristic of some continental vertebrate records (e.g. Miocene of Pakistan, Permian of Texas, USA) may reflect the tectonic setting rather than characterizing mammalian evolution more generally. In conclusion, this hypothesis links a macroecological pattern in modern mammalian faunas to macroevolutionary processes that responded to the tectonic history of the continents. The predictions suggest a research program focusing on biogeographic processes that determine the regional diversity of mammals across different tectonic settings. This research will place the modern biodiversity hotspots for mammals into an evolutionary and geohistorical context which should increase our appreciation of their uniqueness and elevate the imperative for their conservation.

ISSUE

IBS SPECIAL

Badgley, C. et al. 1988. Paleoecology of a Miocene, tropical, upland fauna: Lufeng, China. Nat. Geogr. Res. 4: 178 195. Barbour, M. G. and Billings, W. D. 1999. North American terrestrial vegetation. Cambridge Univ. Press. Barnosky, A. D. and Carrasco, M. A. 2002. Effects of OligoMiocene global climate changes on mammalian species richness in the northwestern quarter of the USA. Evol. Ecol. Res. 4: 811 841. Barnosky, A. D. et al. 2001. Temperate terrestrial vertebrate faunas in North and South America: interplay of ecology, evolution, and geography with biodiversity. Conserv. Biol. 15: 658 674. Barry, J. C. et al. 2002. Faunal and environmental change in the Late Miocene Siwaliks of northern Pakistan. Paleobiology (Suppl.) 28: 1 71. Bradley, R. D. et al. 2007. Toward a molecular phylogeny for Peromyscus: evidence from mitochondrial cytochrome-b sequences. J. Mammal. 88: 1146 1159. Brown, J. H. 1978. The theory of insular biogeography and the distribution of boreal birds and mammals. In: Harper, K. T. and Reveal, J. L. (eds), Intermountain biogeography: a symposium. Great Basin Naturalist Memoirs, pp. 209 227. Brown, J. H. 2001. Mammals on mountainsides: elevational patterns of diversity. Global Ecol. Biogeogr. 10: 101 109. Brown, J. H. and Kodric-Brown, A. 1977. Turnover rates in insular biogeography: effect of immigration on extinction. Ecology 58: 445 449. Brown, J. H. and Maurer, B. A. 1989. Macroecology: the division of food and space among species on continents. Science 243: 1145 1150. Burchfiel, B. C. et al. (eds) 1992. The Cordilleran orogen: conterminous US. Vol. G-3, Geology of North America, Geological Society of America. Bush, M. B. et al. 2004. 48,000 years of climate and forest change in a biodiversity hot spot. Science 303: 827 829. Cannon, M. D. 2004. Geographic variability in North American mammal community richness during the terminal Pleistocene. Quat. Sci. Rev. 23: 1099 1123. Carrasco, M. A. et al. 2005. Miocene Mammal Mapping Project (MIOMAP). Univ. of California Museum of Paleontology, <www.ucmp.berkeley.edu/miomap/>. Coblentz, D. D. and Riitters, K. H. 2004. Topographic controls on the regional-scale biodiversity of the south-western USA. J. Biogeogr. 31: 1125 1138. Cracraft, J. 1985. Biological diversification and its causes. Ann. Missouri Bot. Gard. 72: 794 822. Currie, D. J. 1991. Energy and large-scale patterns of animal- and plant-species richness. Am. Nat. 137: 27 49. Demastes, J. W. et al. 2002. Systematics and phylogeography of pocket gophers in the genera Cratogeomys and Pappogeomys. Mol. Phylogenet. Evol. 22: 144 154. Demboski, J. R. and Sullivan, J. 2003. Extensive mtDNA variation within the yellow-pine chipmunk, Tamias amoenus (Rodentia: Sciuridae), and phylogeographic inferences for northwest North America. Mol. Phylogenet. Evol. 26: 389 408. Endler, J. A. 1977. Geographic variation, speciation, and clines. Princeton Univ. Press. Finarelli, J. A. and Badgley, C. 2010. Diversity dynamics of Miocene mammals in relation to the history of tectonism and climate. Proc. R. Soc. B, doi: 10.1098.rspb.2010.0348. Fitzgerald, J. P. et al. 1994. Mammals of Colorado. Denver Museum of Natural History. Ghalambor, C. K. et al. 2006. Are mountain passes higher in the tropics? Janzen’s hypothesis revisited. Integr. Comp. Biol. 46: 5 17. Grayson, D. K. 1993. The desert’s past. Smithsonian Inst. Press.

230

Grayson, D. K. 2005. A brief history of Great Basin pikas. J. Biogeogr. 32: 2103 2111. Grenyer, R. et al. 2006. Global distribution and conservation of rare and threatened vertebrates. Nature 444: 93 96. Gunnell, G. F. and Bartels, W. S. 2001. Basin margins, biodiversity, evolutionary innovation, and the origin of new taxa. In: Gunnell, G. F. (ed.), Eocene biodiversity: unusual occurrences and rarely sampled habitats. Kluwer Academic/ Plenum Publ., pp. 403 432. Hafner, J. C. et al. 2007. Basal clades and molecular systematics of heteromyid rodents. J. Mammal. 88: 1129 1145. Hall, E. R. 1981. The mammals of North America. Wiley. Heikinheimo, H. et al. 2007. Biogeography of European land mammals shows environmentally distinct and spatially coherent clusters. J. Biogeogr. 34: 1053 1064. Humboldt, A. v. 1805. Essai sur la geographie des plantes accompagne´ d’un tableau physique des regions equinoxiales, fonde´ sure les mesures execute´es, depuis le dixie`me degre´ de latitude boreale jusqu’au dixie`me degre´ de latitude australe, pendant les anne´es 1799, 1800, 1801, 1802, et 1803. Levrault Schoell. InfoNatura 2007. Animals and ecosystems of Latin America. NatureServe, <www.natureserve.org/infonatura/>. IPCC 2007. Climate change 2007: the physical science basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge Univ. Press. Janis, C. M. et al. 1998. Evolution of Tertiary mammals of North America, Vol. 1: terrestrial carnivores, ungulates, and ungulate-like mammals. Cambridge Univ. Press. Janis, C. M. et al. 2008. Evolution of Tertiary mammals of North America, Vol. 2: small mammals, xenarthrans, and marine mammals. Cambridge Univ. Press. Janzen, D. H. 1967. Why mountain passes are higher in the tropics. Am. Nat. 101: 233 249. Kerr, J. T. and Packer, L. 1997. Habitat heterogeneity as a determinant of mammal species richness in high-energy regions. Nature 385: 252 254. Kiessling, W. et al. 2010. Reefs as cradles of evolution and sources of biodiversity in the Phanerozoic. Science 327: 196 198. Kohn, M. J. and Fremd, T. J. 2008. Miocene tectonics and climate forcing of biodiversity, western United States. Geology 36: 783 786. Lomolino, M. V. 2001. Elevation gradients of species density: historical and prospective views. Global Ecol. Biogeogr. 10: 3 12. Lomolino, M. V. et al. 2006. Biogeography. Sinauer. Lyle, M. et al. 2008. Pacific Ocean and Cenozoic evolution of climate. Rev. Geophys. 46: 1 47. McCain, C. M. 2005. Elevational gradients in diversity of small mammals. Ecology 86: 366 372. McKnight, M. W. et al. 2007. Putting beta-diversity on the map: broad-scale congruence and coincidence in the extremes. PLoS Biol. 5: 2424 2432. Melo, A. S. et al. 2009. Environmental drivers of beta-diversity patterns in New-World birds and mammals. Ecography 32: 226 236. Merriam, C. H. 1894. Laws of temperature control of the geographic distribution of terrestrial animals and plants. Nat. Geogr. 6: 229 238. Mittelbach, G. G. et al. 2007. Evolution and the latitudinal diversity gradient: speciation, extinction and biogeography. Ecol. Lett. 10: 315 331. Moritz, C. et al. 2000. Diversification of rainforest faunas: an integrated molecular approach. Annu. Rev. Ecol. Syst. 31: 533 563.

Patterson, B. D. et al. 1998. Contrasting patterns of elevational zonation for birds and mammals in the Andes of southeastern Peru. J. Biogeogr. 25: 593 607. Patterson, B. D. et al. 2005. Digital distribution maps of the mammals of the western Hemisphere, version 2.0. NatureServe, <www.natureserve.org/>. Pierce, K. L. and Morgan, L. A. 1992. The track of the Yellowstone hotspot: volcanism, faulting, and uplift. Geol. Soc. Am. Mem. 179: 1 53. Porter, S. C. 2001. Snowline depression in the tropics during the last glaciation. Quat. Sci. Rev. 20: 1067 1091. Qian, H. and Ricklefs, R. E. 2008. Global concordance in diversity patterns of vascular plants and terrestrial vertebrates. Ecol. Lett. 11: 547 553. Qian, H. et al. 2009. The latitudinal gradient of beta diversity in relation to climate and topography for mammals in North America. Global Ecol. Biogeogr. 18: 111 122. Rahbek, C. and Graves, G. R. 2001. Multiscale assessment of patterns of avian species richness. Proc. Nat. Acad. Sci. USA 98: 4534 4539. Ricklefs, R. E. and Schluter, D. 1993. Species diversity: regional and historical influences. In: Ricklefs, R. E. and Schluter, D. (eds), Species diversity in ecological communities. Univ. Chicago Press, pp. 350 363. Rogers, D. S. and Vance, V. L. 2005. Phylogenetics of spiny pocket mice (Genus Liomys): analysis of cytochrome-b based on multiple heuristic approaches. J. Mammal. 86: 1085 1094. Rowe, K. C. et al. 2006. Comparative phylogeography of eastern chipmunks and white-footed mice in relation to the individualistic nature of species. Mol. Ecol. 15: 4003 4020. Schipper, J. et al. 2008. The status of the worldâ&#x20AC;&#x2122;s land and marine mammals: diversity, threat, and knowledge. Science 322: 225 230.

Sechrest, W. et al. 2002. Hotspots and the conservation of evolutionary history. Proc. Nat. Acad. Sci. USA 99: 2067 2071. Shea, K. and Chesson, P. 2002. Community ecology theory as a framework for biological invasions. Trends Ecol. Evol. 17: 170 176. Simpson, G. G. 1964. Species density of North American recent mammals. Syst. Zool. 13: 57 63. Spradling, T. A. et al. 2004. DNA data support a rapid radiation of pocket gopher genera (Rodentia: Geomyidae). J. Mammal. Evol. 11: 105 125. Strahan, R. (ed.) 1995. Mammals of Australia. Smithsonian Inst. Press. Thomas, G. H. et al. 2008. Regional variation in the historical components of global avian species richness. Global Ecol. Biogeogr. 17: 340 351. Thompson, R. S. and Anderson, K. H. 2000. Biomes of western North America at 18,000, 6000, and 0 14C yr BP reconstructed from pollen and packrat middens. J. Biogeogr. 27: 555 584. Wiens, J. J. et al. 2007. Phylogenetic history underlies elevational biodiversity patterns in tropical salamanders. Proc. R. Soc. B 274: 919 928. Williams, J. W. et al. 2004. Late Quaternary vegetation dynamics in North America: scaling from taxa to biomes. Ecol. Monogr. 74: 309 334. Wilson, D. E. and Reeder, D. M. 2005. Mammal species of the World. Johns Hopkins Univ. Press. Windley, B. F. 1995. The evolving continents. Wiley. Wright, H. E. Jr et al. 1993. Global climates since the Last Glacial Maximum. Univ. Minnesota Press. Yalden, D. W. and Largen, M. J. 1992. The endemic mammals of Ethiopia. Mammal Rev. 22: 115 150. Zachos, J. et al. 2001. Trends, rhythms, and aberrations in global climate 65 Ma to present. Science 292: 686 693.

IBS SPECIAL ISSUE

231

Ecography 33: 232 241, 2010 doi: 10.1111/j.1600-0587.2010.06167.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Nathan Sanders. Accepted 15 February 2010

Phylogeny and biogeography of Oriolidae (Aves: Passeriformes) Knud A. Jønsson, Rauri C. K. Bowie, Robert G. Moyle, Martin Irestedt, Les Christidis, Janette A. Norman and Jon Fjeldsa˚ K. A. Jønsson (kajonsson@snm.ku.dk), Vertebrate Dept, Zoological Museum, Univ. of Copenhagen, Universitetsparken 15, DK-2100 Copenhagen Ø, Denmark, and Museum of Vertebrate Zoology and Dept of Integrative Biology, 3101 Valley Life Science Building, Univ. of California, Berkeley, CA 94720-3160, USA. R. C. K. Bowie, Museum of Vertebrate Zoology and Dept of Integrative Biology, 3101 Valley Life Science Building, Univ. of California, Berkeley, CA 94720-3160, USA. R. G. Moyle, Natural History Museum and Biodiversity Research Center, Univ. of Kansas, KS 66045-7561, USA. M. Irestedt, Molecular Systematic Laboratory, Swedish Museum of Natural History, P.O. Box 50007, SE-10405 Stockholm, Sweden. L. Christidis, Div. of Research and Collections, Australian Museum, 6 College St, Sydney, New South Wales 2010, Australia, and Dept of Genetics, Univ. of Melbourne, Parkville, Victoria 3052, Australia. J. A. Norman, Sciences Dept, Museum Victoria, GPO Box 666, Melbourne, Victoria 3001, Australia, and Dept of Genetics, Univ. of Melbourne, Parkville, Victoria 3052, Australia. J. Fjeldsa˚, Vertebrate Dept, Zoological Museun, Univ. of Copenhagen, Universitetsparken 15, DK-2100 Copenhagen Ø, Denmark.

IBS SPECIAL

ISSUE

Understanding oscine passerine dispersal patterns out of their Australian area of origin is hampered by a paucity of robust phylogenies. We constructed a molecular phylogeny of the oscine family, Oriolidae, which is distributed from Australia through to the Old World. We used the phylogeny to assess direction and timing of dispersal and whether dispersal can be linked with the well-documented movements of geological terranes in the Indonesian Archipelago. We sampled 29 of 33 species of Oriolidae from fresh tissue and from toe pads from museum specimens, and examined two nuclear introns and two mitochondrial genes. Model-based phylogenetic analyses yielded strong support for clades that generally mirrored classical systematics. Biogeographical analyses and divergence time estimates demonstrated that the family originated in the Australo-Papuan region from where it dispersed first to Asia and then onwards to Africa and the Philippines before back-colonising Asia and the Indonesian archipelago. Thus, contrary to several other avian families in the region, Oriolidae represents a sequential dispersal pattern from Australia to Africa via Asia. However, it is noteworthy that the Pacific islands and archipelagos remain uncolonised and that members inhabiting Wallacea are recent colonisers suggesting that Oriolidae are poorly adapted to island life.

The faunal transition between Asia and Australia has received much attention since Wallace (1860, 1863) noted the remarkable species turnover between the western and eastern Indonesian islands (either side of Wallace’s line). Most noticeable was the replacement of placental mammals to the west (except for bats and rodents) with marsupials to the east. Several avian examples are also known, e.g. woodpeckers to the west and cockatoos to the east (MacKinnon and Phillipps 1993, Coates and Bishop 1997). Today the geological history of the region is wellestablished. Wallacea is of mixed origin consisting of Australo-Papuan and Asian plate fragments as well as new volcanic islands (Hall 1998, 2002), and these land masses consequently harbour biota of different origins. Originally Australo-Papua was part of Gondwana, from which it was separated in the Late Cretaceous at around 80 Mya (Metcalve 1998). About 40 Mya the Australo-Papuan plate started to drift rapidly towards the north, and 10 20 Mya plate fragments began to intermingle in the seas between the two continental areas (Hall 1998, 2002). Although no land connection is yet established, we would expect that volant 232

organisms such as birds, bats and insects could take advantage of theses stepping-stone islands and disperse across the gap. Biogeographical patterns differ substantially among passerine bird groups, reflecting different times of origin and radiation and probably also different ecological adaptations and life-history strategies. However, it is becoming increasingly apparent that understanding geological history is an essential prerequisite for understanding patterns of present species distributions. For example, several studies have examined vertebrate speciation and biogeographical patterns in Indo-Pacific archipelagos (e.g. mammals in the Philippine archipelago, Steppan et al. 2003, Heaney 2005, Jansa et al. 2006, and passerine birds in Pacific archipelagos, Filardi and Moyle 2005, Cibois et al. 2007). However, few studies have yet encompassed the whole region on both sides of Wallace’s line (exceptions exist for amphibians Evans et al. 2003 and passerine birds Jønsson et al. 2008a, Moyle et al. 2009). As robust phylogenetic hypotheses become available, it is now a great challenge to interpret evolutionary relationships in light of

the detailed knowledge of plate tectonics that is available for the region. Evidence supports the origin of passerine birds (Passeriformes) within the Gondwanan supercontinent around the time of the K/T boundary (Barker et al. 2002, 2004, Ericson et al. 2002). Two major lineages within Passeriformes are recognized: the suboscines (Tyranni), which are mainly South American, and the oscines (Passeri), with an Australian origin. The basal lineages within oscines are more or less restricted to the Australo-Papuan region, which thus is assumed to be the area of origin for this diverse radiation (Christidis 1991, Barker et al. 2002, Edwards and Boles 2002, Ericson et al. 2002). Within oscines, the core Corvoidea comprises a group of corvoid birds, which includes such diverse families as cuckoo-shrikes (Campephagidae), African bush-shrikes (Malaconotidae), Old World orioles (Oriolidae), whistlers (Pachycephalidae), vireos (Vireonidae), fantails (Rhipudiridae), birds-of-paradise (Paradiseae), shrikes (Lanidae) and crows (Corvoidea). Core Corvoidea has recently been established to be Australian of origin (Barker et al. 2004, Jønsson and Fjeldsa˚ 2006) and some of the families have dispersed to all other continents (except the Antarctica) and to remote oceanic islands and, in the process, evolved high species diversity (ca 750 spp., sensu Monroe and Sibley 1993). Other families such as birds-of-paradise have restricted ranges in Australo-Papua and adjacent islands, possibly because of constraints linked to their unique reproductive strategy (Irestedt et al. 2009). Within the core Corvoidea, systematics of the Oriolidae (Old World orioles) has received little attention. Orioles are broadly distributed in Australia, Asia and Africa. Several species occur in the Indonesian and Philippine archipelagos on both sides of Wallace’s line, and they therefore form an ideal group for investigating biogeographical history and dispersal patterns out of Australia. In this study we present the first molecular phylogeny of the family Oriolidae, based on both nuclear and mitochondrial DNA sequence data. We use the phylogeny to examine the mode, tempo and timing of biogeographical dispersal patterns out of Australia.

Taxon sampling and laboratory procedures

Alignment and phylogenetic analyses Alignment was performed using MegAlign with some minor manual adjustments. The concatenated alignment consisted of 2365 bp and the lengths of the individual alignments were GAPDH: 317 bp, ODC intron-6 and 7: 612 bp, NADH dehydrogenase subunit 2: 1041 (for some species we only obtained 525 bp) and NADH dehydrogenase subunit 3: 397 bp. Coding genes (ND2 and ND3) 233

ISSUE

Orioles are medium-sized, mostly fruit-eating birds confined to mature forest and forest edges. Members of this fairly homogenous bird family, are traditionally defined to comprise the genera Oriolus and Sphecotheres. However, a recent study by Jønsson et al. (2008b) revealed that two species of Pitohui (P. dicrous and P. kirhocephalus) are closely associated with Oriolidae. Thus in the present study we included all species within Oriolidae except O. crassirostris of Sao Tome, which is morphologically very similar to O. brachyrhynchus, and O. tenuirostris from southeast Asia which is closely related to O. chinensis diffusus. Within the genus Sphecotheres we lack the two species endemic to Wetar and and Timor, which are closely related to the AustraloPapuan Sphecotheres vieilloti (Monroe and Sibley 1993). We included four subspecies of the Oriolus chinensis complex.

IBS SPECIAL

Material and methods

This species is widespread in Asia from India to Indonesia and has many distinct subspecies throughout its range. We also included Oriolus kundoo from southern Asia, which was recently separated from O. oriolus (Walter and Jones 2008). In total 29 species were included in the ingroup (including O. kundoo). DNA from fresh tissue (blood, liver, muscle) was extracted using the Quiagen Dneasy Extraction kit (Qiagen, Valencia, CA), following the manufacturers’ protocol. Two nuclear gene regions, ornithine decarboxylase (ODC) introns 6 to 7 (chromosome 3), and glyceraldehyde-3phosphodehydrogenase (GAPDH) intron-11 (chromosome 1), and two mitochondrial markers NADH dehydrogenase subunit 2 (ND2) and subunit 3 (ND3) were sequenced and used to estimate phylogenetic relationships. Primer pairs used for amplification were: ND2: Lmet (Hackett 1996)/ H6312 (Cicero and Johnson 2001); ND3: ND3-L10755/ ND3-H11151 (Chesser 1999); ODC: OD6/OD8 (Allen and Omland 2003), G3P13/G3P14b (Fjeldsa˚ et al. 2003). The thermocycling conditions included a hotstart at 958C for 5 min, followed by 32 cycles at 958C for 40 s, 54 638C for 40 s, and 728C for 60 s, and was completed by a final extension at 728C for 8 min. One microliter of the amplification products was electrophoresed on a 1.5% agarose gel and revealed under UV light with ethidium bromide to check for correct fragment size and to control for the specificity of the amplifications. PCR products were purified using ExoSap enzymes (Exonuclease and Shrimp Alkaline Phosphatase). Purified PCR products were cyclesequenced using the Big Dye terminator chemistry (ABI, Applied Biosystems) in both directions with the same primers used for PCR amplifications, except for G3P13, which was replaced by G3PintL1 (Fjeldsa˚ et al. 2003), and run on an automated ABI 3100 DNA sequencer. Corresponding laboratory procedures for study skins are detailed in Irestedt et al. (2006). Additional internal primers for study skins are specified in Jønsson et al. (2008a) for GAPDH and in Irestedt et al. (2006) for ODC in addition to two new internal primers for ND2 specifically designed for this study, ND2per330F: ATTCCACTTYTGATTCCCAGAAGT; ND2per340R: CCTTGTAGTACTTCTGGGAATCA; ND2ori500F: AGCYTTAGGRGGATGATTRGGRCT; ND2ori530R: GARGAGAARGCYATRATYTTTCG; ND2ori790F: CAGGCTTCCTCCCAAAATGACT; ND2ori773R: AGTCATTTTGGGAGGAAGCCTG. Sequences were assembled with SeqMan II (DNASTAR). Positions where the nucleotide could not be determined with certainty were coded with the appropriate IUPAC code. GenBank accession numbers are provided in Table 1.

IBS SPECIAL

ISSUE

Table 1. List of taxa included in the study. Acronyms are: AMNH, American Museum of Natural History, USA; ANWC, Australian National Wildlife Collection, Canberra, Australia; FMNH, Field Museum of Natural History, Chicago, USA; KU, Univ. of Kansas, Lawrence, USA; LSU, Louisiana State Univ., USA; MCSNC, Museo Civico di Storia Naturale di Carmagnola, Italy; MNHN, MuseÂ´um National d?histoire Naturelle, Paris, France; MV, Museum Victoria, Australia; MVZ, Museum of Vertebrat Zoology, UC Berkeley, USA; NRM, Naturhistoriska Riksmuseet, Stockholm, Sweden; RMNH, Rijksmuseum van Natuurlijke Histoire, Leiden, Netherlands; UMMZ, Univ. of Michigan Museum of Zoology; UWBM, Univ. of Washington, Burke Museum, Seattle, USA; ZMUC, Zoological Museum, Univ. of Copenhagen, Denmark. Species

Voucher

Origin

GAPDH

ODC

ND2

ND3

Colluricincla megarhynca Colluricincla ferrugineus Coracina caesia Coracina caesia Corvus corone Cyclarhis gujanensis Cyclarhis gujanensis Daphoenositta chrysoptera Dicrurus bracteatus Hylophilus ochraceiceps Hylophilus ochraceiceps Lanius collaris Oriolus albiloris Oriolus auratus Oriolus bouroensis Oriolus brachyrhynchus Oriolus chinensis chinensis Oriolus chinensis diffusus Oriolus chinensis diffusus Oriolus chinensis diffusus Oriolus chinensis maculatus Oriolus chinensis melanisticus Oriolus chlorocephalus Oriolus cruentus Oriolus flavocinctus Oriolus forsteni Oriolus hosii Oriolus isabellae Oriolus kundoo Oriolus larvatus Oriolus melanotis Oriolus mellianus Oriolus monacha persistens Oriolus nigripennis Oriolus oriolus Oriolus oriolus Oriolus percivali Oriolus phaeochromus Oriolus sagittatus Oriolus steerei Oriolus steerei Oriolus szalayi Oriolus trailli Oriolus xanthonotus Oriolus xanthornus Ornorectes cristatus Pachycephala simplex Pitohui dichrous Pitohui kirhocephalus Sphecotheres vielloti Vireo olivaceus Vireo olivaceus Outgroup Menura novaehollandiae Menura novaehollandiae

ANWC39343 MV E506 ZMUC 123521 ZMUC134772 MNHN 13-16 ZMUC128105 LSUMZ103262 MV1311 UWBM68045 ZMUC127900 LSUMZ125496 MNHN 2-26 RMNH.AVES.80981 NRM552082 AMNH111097 LSU B-45144 KU10945 AMNH366779 NRM569620 KU10450 NRM569617 ZMUC123918 NRM569622 LSU B-52617 MV1603 RMNH.AVES.14761 AMNH671235 AMNH768148 NRM 570086 MVZ uncat. JF527 AMNH346175 MNHN 1931-1 NRM569619 LSU B-45335 MCSNC1415 ZMUC138401 NRM569618 NRM553510 MV1225 ZMUC100057 AMNH782012 ANWC27056 MNHN JF484 LSU B-57419 MNHN 4-10D ANWC26733 MV1183 MV E545 FMNH 280697 MV2915 ZMUC124543 UMMZ T978

Australia New Guinea Tanzania Tanzania France Ecuador Bolivia Australia New Guinea Ecuador Bolivia Cameroon Philippines Sierra Leone Buru Ghana Philippines Korea Vietnam China Sumatra Talaud, Indonesia Tanzania Borneo Australia Ceram Borneo Luzon Uzbekistan South Africa Timor China Ethiopia Ghana Italy Denmark Kenya Indonesia Aust Mindanao Negros New Guinea Laos Borneo Thailand New Guinea Australia New Guinea New Guinea Australia Panama USA

EU273377 EU273391 EF052797

EU273357 EU273372

GQ494092 GQ494089 EF052773

GQ494126 GQ494123

MV F722 not vouchered

Australia Australia

were checked for the presence of stop codons or insertion/ deletion events that would have disrupted the reading frame. We used Bayesian inference (Holder and Lewis 2003, Huelsenbeck and Ronquist 2003), as implemented in MrBayes 3.1.2 (Huelsenbeck et al. 2001, Ronquist and Huelsenbeck 2003) to estimate phylogenetic relationships. The most appropriate substitution models were determined 234

DQ406663 EU380473

GQ901708 EU272116 EU380435

GQ901732 AY529949 AY030129

EU380474 EF052813 EU272087

EU272113 EU272109

DQ406662 GQ901805 GQ901799

EU272112 GQ901723 GQ901717

GQ901791 GQ901798 GQ901806 GQ901804 GQ901797 GQ901801 EU273382 GQ901800 GQ901792 EF441221

GQ901709 GQ901716 GQ901724 GQ901722 GQ901715 GQ901719 EU273362 GQ901718 GQ901710 EF441243

GQ901807 GQ901809 GQ901796

GQ901725 GQ901727 GQ901714

GQ901810 GQ901802 GQ901793 EF052755

GQ901720 GQ901711 EU273363

GQ901803

GQ901721

GQ901788

GQ901705

GQ901808 GQ901789 GQ901795 GQ901794 DQ406645 EU273389 EU599245 EU273390 EU273392 GQ901790 EU273394

GQ901726 GQ901706 GQ901713 GQ901712 EU272111 EU273370 EU599259 EU273371 GQ901707 EU273374

EF052784 AY030133 AY529960 GQ901778 GQ901771 GQ901780 GQ901762 GQ901769 GQ901782 GQ901777 GQ901768 GQ901773 GQ901772 GQ901763 GQ901758 GQ901779 GQ901783 GQ901784 GQ901787 GQ901767 GQ901781 GQ901786 GQ901774 GQ901764 EF052693 GQ901775 GQ901776 GQ901759 GQ901770 GQ901785 GQ901760 GQ901766 GQ901765 AY529964 GQ494087 EU600814 GQ494088 GQ494100 GQ901761

GQ901749 GQ901742 GQ901751 GQ901733 GQ901740 GQ901753 GQ901748 GQ901739 GQ901744 GQ901757 GQ901743 GQ901734 GQ901728 GQ901750 GQ901754 GQ901755 GQ901738 GQ901752 GQ901745 GQ901735 GQ494146 GQ901746 GQ901747 GQ901729 GQ901741 GQ901756 GQ901730 GQ901737 GQ901736 GQ494121 EU600797 GQ494122 GQ494134 GQ901731

AY136614 EF441220

EF441242 NC_007883

NC_007883

with MrModeltest 2.0 (Nylander 2004), using the Akaike information criterion (Akaike 1973, Posada and Buckley 2004). Bayesian analyses for the concatenated data set were performed allowing the different parameters (base frequencies, rate matrix or transition/transversion ratio, shape parameter, proportion of invariable sites) to vary between the six partitions (GAPDH, ODC, 1st, 2nd, 3rd codon

positions for mtDNA and tRNA), i.e. mixed-models analyses (Ronquist and Huelsenbeck 2003, Nylander 2004). In all MrBayes analyses, the Markov Chain Monte Carlo (MCMC) were run using Metropolis-coupling, with one cold and three heated chains, for 10 (individual analyses) to 20 million (combined analysis) iterations with trees sampled every 100 iterations. The number of iterations discarded before the posterior probabilities were calculated (i.e. the length of the ‘‘burn-in’’ period) was graphically estimated using AWTY (Wilgenbusch et al. 2004, Nylander et al. 2008) by monitoring the change in cumulative split frequencies. Two independent runs initiated from random starting trees were performed for each data set, and the loglikelihood values and posterior probabilities for splits and model parameters were checked to ascertain that the chains had reached apparent stationarity. We used GARLI 0.95 (Zwickl 2006) to perform maximum likelihood analyses. Five independent analyses (20 million generations for the combined analysis, 10 million generations for individual partitions) were performed. Nodal support was evaluated with 100 nonparametric bootstrap pseudoreplications. Distributions and identification of ancestral areas

Dating analyses We used Beast V1.4.6 (Drummond et al. 2002, 2006, Drummond and Rambaut 2007), to estimate divergence dates within Oriolidae. We assigned the best fitting model, as estimated by MrModeltest2 to each of the partitions. We used ML corrected pairwise distances of ND2 for five wellsupported nodes (PP ]0.99 and ML bootstrap ]95) and a recently published rate extrapolation (2.8% Myr 1) of evolution in ND2 for another family of passerine birds (Norman et al. 2007) to calibrate the tree. Needless to say that this sort of extrapolation carries with it a significant margin of error and thus we emphasize the importance of thinking of the time estimates only as a rough attempt to place diversification events within Oriolidae in a historical context. The following calibration points were used: 1) the split within clade I between Oriolus flavocinctus/melanotis and O. szalayi at 2.43 My90.5 stdv (95% CI 1.608 3.252 My); 2) the most basal split within clade III at 5.43 My90.5 stdv (95% CI 4.608 6.252 My); 3) the split in clade IV between Oriolus chinensis diffusus and Oriolus kundoo/O. oriolus/O. chinensis melanisticus/O. c. chinensis at 2.57 My90.5 stdv (95% CI 1.748 3.392 My); 4) the split in clade VI between Oriolus albiloris/isabellae and O. steerei (Negros) at 2.81 My90.5 stdv (95% CI 1.988 3.632 My) and 5) the split in clade VII between Pitohui dichrous and P. kirhocephalus at 5.04 My90.5 stdv (95% CI 4.218 5.862 My). We assumed a Yule Speciation Process for the tree prior and an uncorrelated lognormal distribution for the molecular clock model (Drummond et al. 2006, Ho 2007). We used default prior distributions for all other parameters and ran MCMC chains for 50 million generations. The analysis was repeated twice. We used the program Tracer (Rambaut and Drummond 2007) to assess convergence diagnostics.

Results Phylogenetic analyses

235

ISSUE

Analyses performed on the concatenated data set (six partitions: GAPDH, ODC, 1st, 2nd, 3rd mtDNA codon positions and tRNA; ML: -ln 22590.74, BI harmonic mean -ln 21450.33) and on the individual partitions (GAPDH: AIC: GTR G, ML: -ln 1820.18, BI harmonic mean -ln 1946.46; ODC: AIC: GTR G, ML: -ln 2705.19, BI harmonic mean -ln 2801.14; ND3: AIC: GTR I G, ML: -ln 3835.97, BI harmonic mean -ln 3744.02; ND2: AIC: GTR I G, ML: -ln 12917.55, BI harmonic mean -ln 12540.15) yielded 50% majority-rule consensus trees that were topologically congruent for well-supported nodes (posterior probability 0.95 and bootstrap values 70%) for ODC, ND2 and ND3. GAPDH, however, does show

IBS SPECIAL

Distributional data have been digitised as part of a global effort to map all avian distributions. The distributional data consist of range maps based on conservative interpolation between documented records and comprehensive literature review, entered in a grid corresponding to 1 1 geographical degrees, using the WorldMap software (Williams 1996). The distributional data can be linked, through clade codes, to the topologies of phylogenetic trees. For a simple visual presentation, species richness maps can be produced for groups of species representing branch-length quartiles, based on number of nodes from the root of the phylogeny. Thus, the 1st quartile represents the 25% of species closest to the root (the least number of nodes from the root to the taxon), and the 4th quartile represents the 25% of most terminal taxa (the highest number of nodes from the root). Where several species are separated from the root by an equal number of nodes, the most recently diverged species according to the chronogram will be placed in the higher branch-length quartile. Ancestral areas for Oriolidae were estimated using DIVA (Dispersal-Vicariance Analysis) ver. 1.1 (Ronquist 1996, 1997). Five geographical regions were recognized: A: Australia/New Guinea; B: Wallacea; C: Eurasia; D: Philippines; E: Africa. Maxarea values were set to two. This is equivalent to assuming that the ancestors of the group in question have the same ability to disperse as their extant descendants and therefore ancestral ranges were similar in size to extant ranges (Sanmartı´n 2003, Nylander et al. 2008). Because DIVA can handle only fully bifurcating trees we were forced to deal with polytomies within clade I and a polytomy at the base of clade II, III/IV and Oriolus xanthornus. Taxa belonging to clade I occur in Wallacea and the Australo-Papuan region and relationships among many of the taxa were unresolved. We ran the analysis twice, once assuming an Australo-Papuan origin (A) of the clade and once assuming a more widespread Australo-Papuan/

Wallacean origin of the clade. For the other polytomy we ran three analyses reflecting the three possible relationships. The analyses were carried out several times exploring the effect of changing the cost settings (codivergence 0 5, duplication 0 5, sorting 0 5, switching 0 5). None of these changes altered the outcome of the analysis suggesting a robust result.

Scores of the best likelihood trees were within 0.5 likelihood units of the best tree recovered in each of the other four GARLI runs, suggesting that the five runs had converged. The ML tree topology was completely congruent with the BI topology for well-supported nodes (posterior probability ]0.95 and bootstrap values ]70). Geographical distributions of branch-length quartiles A visual illustration of how the diversification of orioles changed in time and space (Fig. 2) presents geographical patterns of species richness for four groups defined from different phylogenetic branch-lengths. This approach is naÄąÂ¨ve in the sense that the timing of speciation events does not directly follow from the number of nodes on a branch, and therefore is not directly comparable across the phylogeny. Nevertheless, this approach provides a rough

IBS SPECIAL

ISSUE

some differences in the basal part of the tree; placement of the Pitohui species is in conflict with the other gene trees. Relationships of the Pitohui species in the combined analysis, however, are also supported by Myoglobin sequence data in a study by JĂ¸nsson et al. (2008b). Thus we feel confident that Pitohui dicrous and P. kirhocephalus are in fact sister to Sphecotheres vielloti and that they in turn are part of the family Oriolidae. The nuclear gene trees (GAPDH and ODC) (not shown) generally provide few well-supported clades. This is not unexpected and reflects that the genes evolve too slowly to resolve closely related young species within Oriolidae. The nuclear data, however, provide evidence for the partition of some more basal clades. The ND2 and ND3 gene trees (not shown) provide better resolution in the distal part of the tree. The combined analysis (Fig. 1) of both mitochondrial and nuclear genes generates a robust, densely sampled phylogeny for the entire family Oriolidae.

Figure 1. The 50% majority rule consensus tree of Oriolidae obtained from the Bayesian analysis of the combined dataset (GAPDH, ODC, ND2 and ND3). Support values are indicated to the left of the nodes. Above the branch is the posterior probability (only values above 0.95 are shown, asterisks indicate 1.00 posterior probabilities). Below the branch is the maximum likelihood bootstrap value (only values above 70 are shown) from 100 pseudoreplicates. Ancestral areas of origin according to the DIVA analysis are indicated to the left of nodes and present distributions of terminal taxa are indicated after the taxon names (A Australo-Papua, B Wallacea, C Eurasia including, Sumatra, Java and Borneo, D Philippines and E Africa). Clades discussed in the text are indicated by roman numerals I VII.

236

Figure 2. Richness patterns of Oriolidae, according to branch-length groups (number of nodes from the base of the phylogeny). (A) 1st quartile illustrates the distributions of the nine most basal species. (B C) 2nd and 3rd quartiles illustrate the distribution of the intermediate species in the phylogeny, (D) 4th quartile illustrates the distribution of the nine most terminal species. Light blue colour represents one species, and the warmer colours represent the higher numbers of overlapping species. Highest number of species in one grid is: A: 4, B: 4, C: 2, D: 3.

illustration of an origin in the humid parts of the AustraloPapuan area and a rapid dispersal over to the Eurasian mainland, with diversification in the Greater Sundas (B), and further diversification in Africa (C) and the Orient and a recent back colonisation to Wallacea (D). Dispersal-Vicariance analysis

Dating analyses

237

ISSUE

The results of the BEAST dating analysis (Fig. 3) indicate the origin of Oriolidae to be in the Miocene with a rather deep split 20 Mya between the Sphecotheres/Pitohui clade (VII) and all other orioles (clades I VI). The origin of Oriolus (clades I VI) is ca 13 Mya. The diversification within the group of Australian and Wallacean brown orioles (clade I) is determined to be rather young at 5 Mya and the origin of clades II VI and clades II V when orioles dispersed to Asia is determined to be 12 Mya and 11 Mya,

IBS SPECIAL

To run an ideal DIVA analysis one would want to include several taxa outside the focal group, such that the basal part of the ingroup is no longer the root of the tree. This is because the reliability of ancestral reconstructions becomes increasingly unreliable towards the root of the tree, which can cause the ancestral distributions at the root to include all areas analysed (Ronquist 1996). Basal relationships among corvoid bird families, however, have proven hard to resolve (JĂ¸nsson et al. 2008a) and this somewhat confounds the DIVA analysis in this study because we are unable to determine the closest sister groups of Oriolidae. Thus the DIVA analysis presented is conducted exclusively on the ingroup. The DIVA analysis (Fig. 1 and 3) suggests an AustraloPapuan or a more widespread Australo-Papuan/Eurasian origin of the basal nodes. The widespread Australo-Papuan/ Eurasian result is most likely an artefact for the above mentioned reasons. Several other core Corvoidean families have been demonstrated to be of Australo-Papuan origin

(JĂ¸nsson et al. 2008a) and with several basal species within Oriolidae occurring in Australo-Papua it seems reasonable to assume that this is also the case for Oriolidae. Because the origin of clade I is hard to determine we present the results from the DIVA analysis where we assumed the origin of clade I to be Australian but we have indicated in the figure that the origin could be both Australo-Papuan and Wallacean. From the Australo-Papuan region orioles colonised mainland Asia, as indicated by a Eurasian origin of nodes leading to clades II VI and to clades II V. Further up the tree we find a colonisation of Africa indicated by an African origin of the nodes leading to clades II, III and III/IV, and within clade IV there is a back-colonisation of Asia from Africa.

IBS SPECIAL

ISSUE

Figure 3. Chronogram based on the BEAST analysis of Oriolidae. ML corrected pairwise distances of ND2 and a recently published rate (2.8% Myr 1) of evolution in ND2 for passerine birds were used to calibrate the tree. Bars represent 95% HPD intervals. Ancestral areas of origin according to the DIVA analysis are indicated to the left of nodes (A Australo-Papua, B Wallacea, C Eurasia including, Sumatra, Java and Borneo, D Philippines and E Africa). Clades discussed in the text are indicated to the right by roman numerals I VII.

respectively, whereas dispersal to Africa from Asia (root of clades II IV and clades III IV, are found to be between 8 and 10 Mya. Back-colonisation of Asia from Africa (the split between O. auratus and O. chinensis/oriolus/kundoo is at 5 Mya but dispersal into the Indonesian and Philippine archipelagos by O. chinensis melanisticus and O. c. maculatus did not happen until 3 Mya. Dispersal to the Philippines from Asia (clade VI) took place at 5 Mya.

Discussion Systematics The study establishes that Oriolidae (with the inclusion of Sphecotheres and two Pitohui species) is monophyletic with seven well-supported clades (Fig. 1). Figbirds (Sphecotheres) and Pitohui dichrous and P. kirhocephalus form a clade (clade VII), and this clade is sister to the genus Oriolus (all other orioles). It should be mentioned that recent studies 238

have demonstrated that the genus Pitohui is in fact polyphyletic and that several members have been assigned to other genera. Thus the only species left in the genus Pitohui are the two species within the family Oriolidae (Jønsson et al. 2008b, 2010). A basal group within Oriolus (clade I) includes all the Wallacean brown oriole species and the three Australo-Papuan orioles. Within this assemblage, however, resolution is poor, which may reflect recent rapid radiation and colonisation of Wallacean islands from the Australo-Papuan region. The ‘‘African black-headed orioles’’ are reconstructed in two distinct clades (clades II and III). The data are equivocal about the relationship between these two clades and they occur in a polytomy that also includes a largely Asian clade (IV) and O. xanthornus. One of the African clades (II) contains O. brachyrhynchus and the green-headed O. chlorocephalus. The rest of the ‘‘African black-headed orioles’’ fall in another clade (III). The golden orioles (including both African and Asian species) are found in the

same clade (IV) and we find clear evidence that Oriolus chinensis is polyphyletic. We did not sample all O. chinensis subspecies and therefore we can only state that O. chinensis populations in mainland Asia (O. c. diffusus), the Philippines (O. c. chinensis and O. c. melanisticus) and the Sunda Islands (O. c. maculatus), which are represented in this study, are rather divergent according to DNA data and that species status for at least these three taxa must be considered. A clade of Philippine orioles (clade VI) has O. xanthonotus of Borneo at the base. Here we note that there is only a minor molecular differentiation between O. albiloris and O. isabellae, perhaps indicating that they should be treated as a single taxonomic unit. Finally, there is a clade (V) of Asian red and black orioles, which includes O. cruentus, O. hosii, O. trailli and O. mellianus. All relationships recovered in the present analyses are in concordance with previously recognised superspecies, the only notable differences being possible polyphyly of the African black-headed assemblages, which fall out as two distinct clades. This, however, may simply be due to poor resolution between clades II, III and IV. Furthermore, Oriolus xanthornus has historically been considered part of the red and black Asian clade (V) (Walter and Jones 2008) although it resembles African black-headed orioles by plumage colours. Poor resolution in this part of the phylogeny makes us unable to determine with confidence where it belongs but it does seem to have a closer affinity with the African black-headed species. Evolutionary lability in plumage colours and patterns is seen within many other families of birds such as New World Orioles (Allen and Omland 2003), Minivets (JĂ¸nsson et al. 2010) and bushshrikes (Nguembock et al. 2008). However, this is not the case for Old World orioles, where distinct plumages characterise different lineages. Biogeography

ISSUE

239

IBS SPECIAL

Few studies have addressed biogeographical patterns of vertebrates across Wallacea (Hisheh et al. 1998: fruit bats; Evans et al. 2003: frogs), and the focus has mainly been on organisms that have colonised Wallacea from Asia and not vice versa. Michaux (1998) published a paper on birds, which is mostly an expanded list of birds occurring in certain defined subregions within the Indo-Pacific but that study did not include any analyses of colonisation patterns. More detailed studies of passerine birds of Australo-Papuan origin dispersing across and within Wallacea are now appearing (JĂ¸nsson et al. 2008c, 2010, Norman et al. 2009), and these studies have demonstrated a marked variation in dispersal patterns between families, reflecting differences in ecology and life-history strategies, and possibly also reflecting their respective times of dispersal. Oriolidae is an example of a bird family that has colonised Asia from the Australo-Papuan region. Although the results of the DIVA analysis are somewhat ambiguous at the root of the Oriolidae phylogeny, it is most parsimonious to assume a rather restricted Australo-Papuan origin of the group based on the fact that several basal members are at present distributed in Australo-Papua (Fig. 1, 2). The distribution of the brown oriole species (clade I) within Wallacea comprises O. forsteri and O. bouroensis in the

Ceram-Buru area, part of a microplate that was once connected with the Vogelkop Peninsula of the western Papuan landmass, and O. melanotis of Timor/Wetar, which represents a microplate that detached from mainland Australia in the mid-Miocene and rose above sea-level in the late Pliocene (Hall 1998, 2002). Furthermore, O. phaeochromus inhabits the North Moluccas, and dates back to Late Pliocene/Pleistocene. These islands are of oceanic origin, but were located very close to the Vogelkop Peninsula at this time (Hall 1998, 2002). These ages and distributions suggest that, although several taxa occur on isolated islands within Wallacea, they may only have dispersed a short distance from New Guinea and then drifted to their current locations. No basal members of Oriolidae occur in Asia (Fig. 2A). Poor resolution within clade I makes it difficult to evaluate if multiple colonisations of the Moluccas took place in the Plio/Pleistocene or if a historically widespread brown taxon diversified within the archipelago. It was suggested by Diamond (1982) that the brown orioles mimic the larger and rather aggressive friarbirds to avoid competition during feeding. If this were in fact so, convergent brown plumage evolution within the area where friarbirds exist would indeed seem an advantageous adaptation. These brown orioles are the only orioles that have been successful in maintaining populations on Wallacean islands apart from some subspecies of Oriolus chinensis, which occur in Talaud, Sulawesi and the Lesser Sunda Islands. The origin of clades V and VI is Asian and thus the initial colonisation of Asia, did not leave any trace in Wallacea, suggesting long-distance dispersal or extinction of all intervening populations. The lineage diversity maps suggests an initial proliferation in Sundaland (Fig. 2B), and this was followed by large range expansions in Asia (Fig. 2C, D). The red and black orioles in clade V radiated within southeast Asia from around 10 Mya, and orioles in clade VI colonised the Philippines around 4 Mya (possibly via Palawan). From Asia there is evidence of colonisation of Africa in the Late Miocene by the two African black/greenheaded clades (II and III), at a time when evergreen humid forest probably extended across northern and central Africa (FjeldsaË&#x161; and Bowie 2008) (Fig. 2C). The colonisation pattern of Oriolidae is interesting in several ways. At first sight it seems a straight forward dispersal from Australia to Asia and onwards to Africa, which is the general pattern proposed for oscine passerine birds proposed by Barker et al. (2002) and Ericson et al. (2002). However, the initial leap across Wallacea without colonisation of the archipelago is puzzling. Orioles apparently have a great dispersal capacity over land, exemplified by their rapid colonisation of Asia and Africa but are absent from the Melanesian (and other oceanic) archipelagos, which were successfully colonised by several other corvoid families (e.g. Pachycephalidae, Campephagidae, Monarchidae). Colonisation of both the Philippine and Indonesian archipelagos took place in rather recent time and it is tempting to speculate that orioles are poorly adapted to archipelago life and that the recent archipelago colonisations from both Australia and Asia represent a source to sink spill-over from the mainland. It seems probable that when orioles colonised Asia in the Miocene it must have involved a series of island-hopping events within Wallacea

(Indonesian island chain), maybe followed by extinction of populations inhabiting small volcanic islands. Apart from the recent colonisations of the northern Moluccas and the Philippines, orioles are absent from ophiolitic islands (derived from oceanic crust that was uplifted above sea level mainly through volcanism). Perhaps orioles require rather large and stable land areas with mature forests for the fruits they require, or a vegetation different from the pioneer communities of newly formed islands. They are therefore mostly restricted to the microplates that are close to mainlands and larger landbridge islands. A better understanding of the ecological requirements of orioles may be essential for interpreting the general rarity of orioles within the Wallacean area. Likewise there is no obvious reason why orioles have not colonised the Melanesian archipelagos. Orioles have radiated and dispersed to Asia and Africa already in the Miocene, whereas colonisation of the Philippine and the Indonesian archipelagos did not take place until the Pliocene when Palawan was in place as a landbridge between Borneo and Luzon. Competition with other species of frugivorous birds (the fruit pigeons of Ptilinopus and Ducula are widespred in the Pacific and may be competitors), may have made colonisation of the Philippine and the Indonesian archipelago difficult and also kept orioles completely out of the western Pacific archipelagos.

IBS SPECIAL

ISSUE

Acknowledgements We are grateful to the following institutions for granting access to toe-pad, blood and tissue samples: AMNH, American Museum of Natural History, USA; ANWC, Australian National Wildlife Collection, Canberra, Australia; FMNH, Field Museum of Natural History, Chicago, USA; KU, Univ. of Kansas, Lawrence, USA; LSU, Louisiana State Univ., USA; MCSNC, Museo Civico di Storia Naturale di Carmagnola, Italy; MNHN, Muse´um National d?histoire Naturelle, Paris, France; MV, Museum Victoria, Australia; NRM, Naturhistoriska Riksmuseet, Stockholm, Sweden; RMNH, Rijksmuseum van Natuurlijke Histoire, Leiden, Netherlands; UMMZ, Univ. of Michigan Museum of Zoology; UWBM, Univ. of Washington, Burke Museum, Seattle, USA; ZMUC, Zoological Museum, Univ. of Copenhagen, Denmark. KAJ would like to acknowledge the support from the Australian Museum Postgraduate Awards 2006/07.

References Akaike, H. 1973. Information theory as an extension of the maximum likelihood principle. In: Petrov, B. N. and Csaki, F. (eds), Second International Symposium on Information Theory. Akademiai Kiado, Budapest, pp. 276 281. Allen, E. S. and Omland, K. E. 2003. Novel intron phylogeny (ODC) supports plumage convergence in orioles (Icterus). Auk 120: 961 969. Barker, F. K. et al. 2002. A phylogenetic hypothesis for passerine birds: taxonomic and biogeographic implications of an analysis of nuclear DNA sequence data. Proc. R. Soc. B 269: 295 308. Barker, F. K. et al. 2004. Phylogeny and diversification of the largest avian radiation. Proc. Nat. Acad. Sci. USA 101: 11040 11045. Chesser, R. T. 1999. Molecular systematics of the rhinocryptid genus Pteroptochos. Condor 101: 439 446. Christidis, L. 1991. Molecular and biochemical evidence for the origins and evolutionary radiations of the Australasian avifauna. In: Bell, B. D. and Cossee, R. O. (eds), Proceedings of

240

the XXth International Ornithological Congress. Ornithological Congress Trust Board, Wellington, pp. 392 397. Cibois, A. et al. 2007. Uniform phenotype conceals double colonization by reed-warblers of a remote Pacific archipelago. J. Biogeogr. 34: 1150 1166. Cicero, C. and Johnson, N. K. 2001. Higher level phylogeny of vireos (Aves: Vireonidae) based on sequences of multiple mtDNA genes. Mol. Phylogenet. Evol. 20: 27 40. Coates, B. J. and Bishop, K. D. 1997. A guide to the birds of Wallacea, Sulawesi, the Moluccas and Lesser Sunda Islands, Indonesia. Dove Publ., Australia. Diamond, J. M. 1982. Mimicry of friarbirds by orioles. Auk 99: 187 196. Drummond, A. J. and Rambaut, A. 2007 BEAST v1.4.7. <http://beast.bio.ed.ac.uk/>. Drummond, A. J. et al. 2002. Estimating mutation parameters, population history and genealogy simultaneously from temporally spaced sequence data. Genetics 161: 1307 1320. Drummond, A. J. et al. 2006. Relaxed phylogenetics and dating with confidence. PLoS Biol. 4: 88. Edwards, S. V. and Boles, W. E. 2002. Out of Gondwana: the origin of passerine birds. Trends Ecol. Evol. 17: 347 349. Ericson, P. G. P. et al. 2002. A Gondwanan origin of passerine birds supported by DNA sequences of the endemic New Zealand wrens. Proc. R. Soc. B 269: 235 241. Evans, B. J. et al. 2003. Phylogenetics of fanged frogs: testing biogeographical hypotheses at the interface of the Asian and Australian faunal zones. Syst. Biol. 52: 794 819. Filardi, C. E. and Moyle, R. G. 2005. Single origin of a panPacific bird group and upstream colonization of Australasia. Nature 438: 216 219. Fjeldsa˚, J. and Bowie, R. C. K. 2008. New perspectives on Africa’s ancient forest avifauna. Afr. J. Ecol. 46: 235 247. Fjeldsa˚, J. et al. 2003. Sapayoa aenigma: a New World representative of Old World suboscines. Proc. R. Soc. B (Suppl.) 270: 238 241. Hackett, S. J. 1996. Molecular phylogenetics and biogeography of tanagers in the genus Ramphocelus (Aves). Mol. Phylogenet. Evol. 5: 368 382. Hall, R. 1998. The plate tectonics of the Cenozoic SE Asia and the distribution of land and sea. In: Hall, R. and Holloway, J. D. (eds), Biogeography and geological evolution of SE Asia. Backhuys Publ., pp. 133 163. Hall, R. 2002. Cenozoic geological and plate tectonic evolution of SE Asia and the SW Pacific: computer-based reconstructions, model and animations. J. Asian Earth Sci. 20: 353 431. Heaney, L. R. et al. 2005. The roles of geological history and colonization abilities in genetic differentiation between mammalian populations in the Philippine archipelago. J. Biogeogr. 32: 229 247. Hisheh, S. et al. 1998. Biogeography of the Indonesian archipelago. Mitochondrial DNA variation in the fruit bat, Eonycteris spelaea. Biol. J. Linn. Soc. 65: 329 345. Ho, S. Y. W. 2007. Calibrating molecular estimates of substitution rates and divergence times in birds. J. Avian Biol. 38: 409 414. Holder, M. T. and Lewis, P. O. 2003. Phylogeny estimation: traditional and Bayesian approaches. Nat. Rev. Genet. 4: 275 284. Huelsenbeck, J. P. and Ronquist, F. 2003. MrBayes: a program for the Bayesian inference of phylogeny. Version 3.1.2. <http:// mrbayes.scs.fsu.edu/index.php>. Huelsenbeck, J. P. et al. 2001. MrBayes: Bayesian inference of phylogeny. Bioinformatics 17: 754 755. Irestedt, M. et al. 2006. Nuclear DNA from old collections of avian study skins reveals the evolutionary history of the Old World suboscines (Aves, Passeriformes). Zool. Script. 35: 567 580.

Irestedt, M. et al. 2009. An unexpectedly long history of sexual selection in birds-of-paradise. BMC Biol. 9: 235. Jansa, S. A. et al. 2006. The pattern and timing of diversification of Philippine endemic rodents: evidence from mitochondrial and nuclear gene sequences. Syst. Biol. 55: 73 88. Jønsson, K. A. and Fjeldsa˚, J. 2006. Determining biogeographic patterns of dispersal and diversification in oscine passerine birds in Australia, southeast Asia and Africa. J. Biogeogr. 33: 1155 1165. Jønsson, K. A. et al. 2008a. Explosive avian radiations and multidirectional dispersal across Wallacea: evidence from the Campephagidae and other crown Corvida (Aves). Mol. Phylogenet. Evol. 47: 221 236. Jønsson, K. A. et al. 2008b. Polyphyletic origin of toxic pitohui birds suggests widespread occurrence of toxicity in corvoid birds. Biol. Lett. 4: 71 74. Jønsson, K. A. et al. 2008c. Molecular phylogenetics and diversification within one of the most geographically variable bird species complexes (Pachycephala pectoralis/melanura). J. Avian Biol. 39: 473 478. Jønsson, K. A. et al. 2010. Historical biogeography of an IndoPacific passerine bird family: Pachycephalidae: different colonization patterns in the Indonesian and Melanesian archipelagos. J. Biogeogr. 37: 245 257. MacKinnon, J. and Phillipps, K. 1993. A field guide to the birds of Borneo, Sumatra, Java and Bali. Oxford Univ. Press. Metcalve, I. 1998. Palaeozoic and Mesozoic geological evolution of the SE Asian region: multidisciplinary constramits and implications for biogeography. In: Hall, R. and Holloway, J. D. (eds), Biogeography and geological evolution of SE Asia. Backhuys Publ., pp. 25 41. Michaux, B. 1998. Terrestrial birds in the Indo-Pacific. In: Hall, R. and Holloway, J. D. (eds), Biogeography and geological evolution of SE Asia. Backhuys Publ., pp. 361 391. Monroe, B. L. and Sibley, C. G. 1993. A world checklist of birds. Yale Univ. Press. Moyle, R. G. et al. 2009. Explosive Pleistocene diversification and hemispheric expansion of a ‘‘great speciator’’. Proc. Nat. Acad. Sci. USA 106: 1863 1868. Nguembock, B. et al. 2008. Phylogeny of Laniarius: molecular data reveal L. liberatus synonymous with L. erlangeri and ‘‘plumage coloration’’ as unreliable morphological characters for defining species and species groups. Mol. Phylogenet. Evol. 48: 396 407. Norman, J. A. et al. 2007. Speciation dynamics in the AustraloPapuan Meliphaga honeyeaters. Mol. Phylogenet. Evol. 42: 80 91.

Norman, J. A. et al. 2009. A multigene phylogeny reveals novel relationships for aberrant genera of Australo-Papuan core Corvoidea (Aves: Passeriformes) and polyphyly of the Pachycephalidae and Psophodidae. Mol. Phylogenet. Evol. 52: 488 497. Nylander, J. A. A. 2004. MrModeltest2. <http://www.abc.se/ nylander/>. Nylander, J. A. A. et al. 2008. AWTY (are we there yet): a system for graphical exploration of MCMC convergence in Bayesian phylogenetics. Bioinformatics 24: 581 583. Posada, D. and Buckley, T. R. 2004. Model selection and model averaging in phylogenetics: advantages of Akaike information criterion and Bayesian approaches over likelihood ratio tests. Syst. Biol. 53: 793 808. Rambaut, A. and Drummond, A. J. 2007. Tracer v1.4. <http:// beast.bio.ed.ac.uk/Tracer>. Ronquist, F. 1996. DIVA version 1.1. Computer program and manual. <www.ebc.uu.se/systzoo/research/diva/diva.html>. Ronquist, F. 1997. Dispersal-vicariance analysis: a new approach to the quantification of historical biogeography. Syst. Biol. 46: 195 203. Ronquist, F. and Huelsenbeck, J. P. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572 1574. Sanmartı´n, I. 2003. Dispersal vs. vicariance in the Mediterranean: historical biogeography of the Palearctic Pachydeminae (Coleoptera, Scarabaeoidea). J. Biogeogr. 30: 1883 1897. Steppan, S. J. et al. 2003. Molecular phylogeny of the endemic rodent Apomys (Muridae) and the dynamics of diversification in an oceanic archipelago. Biol. J. Linn. Soc. 80: 699 715. Wallace, A. R. 1860. On the zoological geography of the Malay Archipelago. J. Linn. Soc. Lond. IV: 172 184. Wallace, A. R. 1863. On the physical geography of the Malay Archipelago. J. Roy. Geogr. Soc. 217 234. Walter, B. A. and Jones, P. J. 2008. Family oriolidae (orioles). In: in Del Hoyo, J. et al. (eds), Handbook of the birds of the world Vol. 13. Penduline-tits to shrikes. Lynx Edicions, pp. 692 731. Wilgenbusch, J. C. et al. 2004. AWTY: a system for graphical exploration of MCMC convergence in Bayesian phylogenetic inference. <http://ceb.csit.fsu.edu/awty>. Williams, P. H. 1996. WORLDMAP 4 WINDOWS: software and help document 4.1. Privately distributed, London, UK. Zwickl, D. J. 2006. Genetic algorithm approaches for the phylogenetic analysis of large biological sequence datasets under the maximum likelihood criterion. Ph.D. thesis, The Univ. of Texas at Austin, USA.

IBS SPECIAL ISSUE

241

Ecography 33: 242 250, 2010 doi: 10.1111/j.1600-0587.2010.06309.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Kenneth H. Kozak. Accepted 8 February 2010

Phylogenetic signals in the climatic niches of the world’s amphibians Christian Hof, Carsten Rahbek and Miguel B. Arau´jo C. Hof (chof@bio.ku.dk), Center for Macroecology, Evolution & Climate, Dept of Biology, Univ. of Copenhagen, Universitetsparken 15, DK-2100 Copenhagen Ø, Denmark and Biodiversity and Global Change Lab, Dept of Biodiversity and Evolutionary Biology, National Museum of Natural Sciences (CSIC), C/Jose´ Gutie´rrez Abascal, 2, ES-28006 Madrid, Spain. C. Rahbek, Center for Macroecology, Evolution & Climate, Dept of Biology, Univ. of Copenhagen, Universitetsparken 15, DK-2100 Copenhagen Ø, Denmark. M. B. Arau´jo, Biodiversity and Global Change Lab, Dept of Biodiversity and Evolutionary Biology, National Museum of Natural Sciences (CSIC), C/Jose´ Gutie´rrez Abascal, 2, ES-28006 Madrid, Spain and Rui Nabeiro Biodiversity Chair, CIBIO, Univ. of E´vora, Largo dos Colegiais, PT-7000 E´vora, Portugal.

IBS SPECIAL

ISSUE

The question of whether closely related species share similar ecological requirements has attracted increasing attention, because of its importance for understanding global diversity gradients and the impacts of climate change on species distributions. In fact, the assumption that related species are also ecologically similar has often been made, although the prevalence of such a phylogenetic signal in ecological niches remains heavily debated. Here, we provide a global analysis of phylogenetic niche relatedness for the world’s amphibians. In particular, we assess which proportion of the variance in the realised climatic niches is explained on higher taxonomic levels, and whether the climatic niches of species within a given taxonomic group are more similar than between taxonomic groups. We found evidence for phylogenetic signals in realised climatic niches although the strength of the signal varied among amphibian orders and across biogeographical regions. To our knowledge, this is the first study providing a comprehensive analysis of the phylogenetic signal in species climatic niches for an entire clade across the world. Even though our results do not provide a strong test of the niche conservatism hypothesis, they question the alternative hypothesis that niches evolve independently of phylogenetic influences.

The question of whether related species are also ecologically similar is as old as modern biology (Darwin 1859). Recently, the question has gained increased interest (Losos 2008a, b, Wiens 2008, Vieites et al. 2009a, Dormann et al. 2010), partly because of its implications for understanding global biodiversity gradients (Wiens and Donoghue 2004), and partly because it helps in comprehending how species might adapt to ongoing climate changes (Botkin et al. 2007). Understanding the extent to which there is a phylogenetic signal in ecological niches (the tendency for related species to resemble each other’s ecological characteristics more than species randomly drawn from a phylogeny; Blomberg and Garland 2002, Losos 2008a) helps to formulate hypothesis about niche evolution. This is particularly true if one adopts the view that estimation of the signal strength in climatic niches may serve as a surrogate measure for the rate of climatic niche evolution (Garland 1992, Blomberg et al. 2003, Rheindt et al. 2004, but see Revell et al. 2008, Ackerly 2009). It needs to be added, though, that establishing such a phylogenetic signal does not demonstrate the existence of phylogenetic niche conservatism, which is the tendency of related species’ niches to be even more similar than expected given their phylogeny (Losos 2008a). However, the existence of strong signals in climatic niches do challenge the alternative 242

hypothesis that niches evolve quickly (Broennimann et al. 2007) and independently of phylogeny (Dormann et al. 2010). Despite the relevance of the climatic niche concept to contemporary ecology (Arau´jo and Guisan 2006, Sobero´n 2007), quantitative analyses on the strength of the phylogenetic signal in climatic niche similarities are scarce (but see, e.g. Prinzing et al. 2001). As pointed out by Losos (2008a), most studies investigating phylogenetic signals in ecological niches only include few species at rather small geographic extents. Thus, the need for taxonomically and geographically comprehensive analyses on phylogenetic signals in climatic niches is timely. Here, we provide the first of such analyses and test for the existence and strength of phylogenetic signals in climatic niches for an entire class of organisms, the amphibians, on a global scale. Ideally, one would test hypotheses about niche evolution using measures of the fundamental niche (sensu Hutchinson 1957), since the fundamental niche is the product of the genetics, morphology and physiology of the species, thus being the ‘‘feature’’ which evolves. In a climatic context, the fundamental niche would be the range of combinations of climatic variables in which the species could potentially exist (Austin et al. 1990, Sobero´n 2007). Unfortunately, estimates of the fundamental climatic niches for large

numbers of species are difficult to obtain. Therefore we have to rely on surrogates estimated with the climate envelope of species, i.e. the combination of climatic variables (e.g. means and extremes of precipitation and temperature) that best describes a species’ geographical range. This characterisation can, however, at best represent the realised climatic niche of a species, and will never entirely portray the fundamental climatic niche (see discussion in Arau´jo and Guisan 2006, Colwell and Rangel 2009, Sobero´n and Nakamura 2009). Here, we first used a family-level phylogeny of the world’s amphibians to test for the existence of phylogenetic signals in species climate niches. Then we tested for the existence of phylogenetic signals and measured their strength separately for the three orders of amphibians and for each one of seven biogeographical regions.

Material and methods

Climatic niches were characterised using an ordination approach termed ‘‘outlying mean index’’ (OMI; Dole´dec et al. 2000). In contrast to other ordination techniques, OMI does not make assumptions about the shape of the species’ response curves to the environment and gives equal weight to sites independent of their species richness. OMI gives the species average position (‘‘niche position’’) within environmental space, which represents a measure of the

243

ISSUE

Quantifying climatic niches

IBS SPECIAL

We used distributions for 5527 amphibian species from all three amphibian orders (Anura, Caudata, Gymnophiona, see Supplementary material Table S1 for an overview of the numbers of species included in the dataset). Distribution data were compiled from the ‘‘Global Amphibian Assessment’’ (IUCN 2004). This dataset comprises distribution maps (extent of occurrence polygons) for each species based on documented records and expert knowledge. Although it is the most comprehensive global dataset available for amphibian distributions, many species are listed as ‘‘data deficient’’, due to a lack of knowledge on their real distributions. Climatic data (originally 19 bioclimatic variables at 10’ resolution) were compiled from the WorldClim database (Hijmans et al. 2005). Distribution and climate data were resampled to a 2 2 degree latitudelongitude grid including 5017 terrestrial cells. A taxonomic topology for genus, subfamily, family and higher taxonomic levels was compiled from the ‘‘Amphibian Tree of Life’’ (Frost et al. 2006) and the online database ‘‘Amphibian Species of the World’’ (ASW; Frost 2007). The taxonomic nomenclature of this database is based on a phylogenetic super tree considering the most recent studies of amphibian phylogeny and is thus building upon direct inferences of the evolutionary history of the species. Despite criticism on several aspects of the original ‘‘Amphibian Tree of Life’’ phylogeny (Wiens 2007), the ASW taxonomy is the most comprehensive taxonomic database for amphibians to date and is being used frequently in conservation and evolutionary studies (Blackburn 2008, Corey and Waite 2008, Santos et al. 2009).

distance between the environmental conditions used by the species and the mean environmental conditions of the study area. It also quantifies the variability of environmental conditions used by each species (‘‘niche breadth’’), given by the standard deviation along the respective OMI axes (for more details, see Dole´dec et al. 2000, as well as Thuiller et al. 2004 for a case study using OMI). Here, environmental conditions were measured as a function of eight climatic variables: mean diurnal range of temperature, minimum temperature of the coldest month, annual range of temperature, mean temperature of the warmest quarter, annual precipitation, precipitation seasonality, precipitation of the driest quarter, and precipitation of the warmest quarter (for a detailed description of the derivation of these variables, see Hijmans et al. 2005). These variables include a range of climatic factors (temperature extremes, amount and seasonality of precipitation) which are known to impose constraints on the occurrence and survival of amphibians (Carey and Alexander 2003, Wells 2007), and are often used to model the geographical distributions of individual species (Arau´jo et al. 2006) and species richness (Arau´jo et al. 2008). In the OMI analysis, we used the first and second axes of the ordination since they explained 82 to 96% of the total inertia (Supplementary material Table S2). OMI analyses were performed using the ade4 package in R (Chessel et al. 2004, R Development Core Team 2008). A randomisation test was performed to examine if niche positions along climate gradients could have arisen by chance (Dole´dec et al. 2000); one thousand permutations were obtained for testing niche positions of each species occurring in each one of the biogeographical regions (see below). From the OMI analysis, we also obtained measures of niche breadth along the first and second OMI axes (for more details, see Dole´dec et al. 2000, Thuiller et al. 2004). Species may share ecological traits because of their shared evolutionary history, but also because they occur in similar places (see Freckleton and Jetz 2009, and references therein). For practical reasons, to account for possible confounding effects arising from spatial autocorrelation in niche characteristics and to explore the potential geographic variation in phylogenetic signal strength, all analyses except the one for the family-level phylogeny (see below) were performed separately for each amphibian order and biogeographical realm. Biogeographical realms were classified following the divisions of Sclater (1858) and Wallace (1876), later renamed by Olson et al. (2001): Afrotropics, Australasia, Indo-Malay, Nearctic, Neotropics, Palaearctic, Antarctica, and Oceania (referred to here as ‘‘regions’’; see Supplementary material Table S1 for an overview of the numbers of species for each species set). Because there are no amphibians in Antarctica and only a few across the scattered islands of Oceania, these regions were removed from the analyses. Madagascar harbours a rich amphibian fauna that is quite distinct from the Afrotropical fauna (Duellman 1999, Vieites et al. 2009b); therefore, we added Madagascar as a seventh region. Nevertheless, we are aware that the spatial extent of the regions is still too large to completely rule out any confounding spatial influence on niche similarity. However, the geographic and phylogenetic resolution of our data

does not allow for more sophisticated approaches (as recently proposed by Freckleton and Jetz 2009).

IBS SPECIAL

ISSUE

Testing for phylogenetic signals in climatic niche similarity To test for phylogenetic signals in climatic niche similarity, we used Blomberg’s randomisation test and K statistic, variance component analyses (VCA), analysis of similarity (ANOSIM) and Wilcoxon rank sum tests. Blomberg’s randomisation test for phylogenetic signal assesses whether a given phylogenetic tree (including topology and branch lengths) better fits a set of data assigned to the tree tips (climatic niche positions in our case) as compared with the fit obtained when the data have been randomly permuted across the tree tips (Blomberg et al. 2003). The K statistic indicates the strength of phylogenetic signal, as compared with an expectation based on the tree structure and assuming Brownian motion character evolution. K values equal to 1 indicate a phylogenetic signal resembling the Brownian motion evolution model, values of K 1 or B1 indicate a stronger or weaker signal than the one expected by the Brownian motion model of character evolution (Blomberg et al. 2003). Since no complete phylogeny is yet available for the world’s amphibians, we used the global family-level phylogeny from Roelants et al. (2007). Blomberg’s randomisation and K analyses were performed using the picante package within R, with 1000 randomisations to assess significance (Kembel et al. 2009). With VCA we quantified how much of the niche variance on the species level (among-species variance) can be explained at different taxonomic levels (Venables and Ripley 1999, Prinzing et al. 2001). As taxonomic levels we used the genus, subfamily and family grouping as well as the higher taxonomic categories above the family level as given by Frost et al. (2006). A large proportion of the amongspecies variance in niche position explained at higher taxonomic levels would indicate a phylogenetic signal in climatic niche similarity. On the other hand, all the variance localised among the species would indicate the absence of a phylogenetic signal. We applied VCA with a restricted maximum likelihood approach, using the functions lme and varcomp in the ape package within R (Paradis et al. 2004). We also performed null models to assess if the results of the VCA could be produced by chance alone. The null models simulate the case of no phylogenetic signal running VCA based on a randomised phylogeny. To generate the null models, we randomised the taxonomic assignments of the species and calculated the variance components as the mean of one thousand randomisations. Again, we ran this analysis separately for the three amphibian orders within each region. With ANOSIM a non-parametric test analogous to ANOVA we tested if niche similarities within groups were larger than between groups (Clarke 1993). The procedure started with a calculation of within- and between-group niche dissimilarities, as follows. Euclidean distances between niche positions were calculated for pairwise combinations of all possible pairs of species. The Euclidian distances reflecting niche dissimilarity between pairs of species were then compared within and between 244

taxonomic groups aggregated at the genus and family levels. When the mean within-group niche dissimilarity is smaller than between-group niche dissimilarity, this is interpreted as indicating the presence of a phylogenetic signal in climatic niche similarity; when the mean is larger, it means the phylogenetic signal is lost. Based on 999 permutations, we tested whether within- and between-group niche dissimilarities were more different than expected by chance. ANOSIM was run with the vegan package of R (Oksanen et al. 2009), again separately for each one of the biogeographical regions (see also Supplementary material Fig. S1 for an illustration of the procedure, and Fig. S2 for examples of two species sets). We also calculated the amount of niche overlap along the first and second OMI axis within and between groups (families and genera). To do so, according to the protocol of the ANOSIM analysis, we calculated the pairwise niche overlap for all possible species pairs, again separately for each order and biogeographical region. Species occurring in only one grid cell have by definition a niche breadth of zero and are therefore excluded from the overlap analyses. We then grouped the pairwise niche overlap values into a within-taxon and between-taxon group (the taxon being the family or genus). For each dataset (amphibian order per region), the within- and between-group separation was done 1) for the entire species pool and 2) separately for each taxon (see also Table 2 for details). Wilcoxon rank sum tests (Hollander and Wolfe 1999) were used to test if withingroup overlap was larger than between-group overlap, which would indicate a phylogenetic signal. By applying different methods to test for phylogenetic signal we try to decrease the risk that the outcomes are biased by the uncertainties or problems of a certain method. Results indicating the same tendency for different methods (although not quantitatively comparable) would strengthen the general value of results and support stronger inference. To ensure that the results were not systematically biased by species with niche characterisations that could have arisen by chance, VCA and ANOSIM analyses were performed 1) including all species and 2) including only species with climatic niches significantly better characterised by OMI than expected by chance.

Results In the global analyses on the family level, we found a phylogenetic signal in climatic niches for the first and second OMI axes (p 0.001 and p 0.026, respectively). Signal strength differed considerably among the two axes, the first axis showing a signal stronger than expected from a Brownian motion evolution model (K 1.45), the second axis showing a signal lower than that (K 0.44). The analysis conducted with VCA showed that a high proportion of among-species variance in climatic niche position is explained at higher taxonomic levels (Fig. 1). Results were consistent independently of whether the whole set of species or the sub-set with significant OMI values was considered. In most cases, the analyses of the species for which climatic influences were significant showed an even stronger phylogenetic signal (Supplementary material Fig. S3); this indicates that there were no biases arising from

ISSUE

245

IBS SPECIAL

Figure 1. Results of the variance component analyses (VCA). Variance components are calculated as the proportion of among-species variance in climatic niche positions that is explained at different taxonomic levels (species, genus, family, above-family; see key). The bars are organised from lower (species) to higher (above-family) taxonomic levels. A completely black bar indicates that all variance lies at the species-level, and none is explained at higher taxonomic levels. The analyses were performed separately for the three orders and each of the biogeographical regions (AFR, Afrotropics; AUS, Australasia; IND, Indo-Malay; MAD, Madagascar; NEA, Nearctic; NEO, Neotropics; PAL, Palaearctic). Within one species set (represented by a box), the first and third bars give the observed (obs) values (for the first or second OMI axis, respectively), and the second and fourth bars give the values for the according null model (exp). Null models were conducted by randomising the phylogenetic assignment for the species pool, thus representing the null expectation of no phylogenetic signal in climatic niche similarity (see text for further details).

Table 1. Climatic niche distances for amphibians on the family and genus levels within different biogeographical regions. n

ISSUE

Within

Between

rANOSIM

Families Anura AFR AUS IND MAD NEA NEO PAL

16 9 12 3 8 19 11

2.04 2.03 2.42 1.65 2.31 2.08 2.37

1.18 1.64 1.42 0.94 1.41 1.41 1.47

2.37 3.18 2.50 1.83 2.35 2.37 2.73

1.35 1.87 1.39 1.03 1.35 1.54 1.57

0.14 0.35 0.034 0.10 0.022 0.11 0.14

B0.001 B0.001 B0.001 0.003 0.143 B0.001 B0.001

Caudata IND NEA NEO PAL

2 7 3 4

1.53 2.46 2.13 2.92

0.89 1.73 1.48 1.88

1.37 2.40 2.87 3.61

0.79 1.56 1.77 2.02

0.10 0.0045 0.25 0.21

0.785 0.522 B0.001 B0.001

2 2

2.58 2.14

1.55 1.26

2.56 2.24

1.51 1.50

0.0040 0.0012

0.44 0.462

Genera Anura AFR AUS IND MAD NEA NEO PAL

50 37 67 16 17 126 41

1.96 2.69 2.04 1.62 2.37 1.98 2.08

1.12 1.84 1.43 0.94 1.51 1.51 1.50

2.35 2.92 2.50 1.74 2.34 2.33 2.69

1.34 1.88 1.39 0.99 1.35 1.52 1.56

0.16 0.074 0.19 0.069 0.0042 0.16 0.25

B0.001 0.002 B0.001 0.02 0.454 B0.001 B0.001

Caudata IND NEA NEO PAL

6 20 9 17

1.43 1.37 1.69 2.31

0.78 1.22 1.31 1.78

1.47 2.57 2.38 3.42

0.87 1.66 1.57 1.99

0.0089 0.44 0.28 0.35

0.445 B0.001 B0.001 B0.001

Gymnophiona AFR IND NEO

7 4 10

1.12 2.66 2.12

1.17 1.61 1.32

2.90 2.51 2.17

1.67 1.49 1.31

Gymnophiona AFR IND NEO

IBS SPECIAL

ANOSIM$

Mean niche distance*

0.62 0.052 0.029

B0.001 0.725 0.321

*Mean distances were calculated by averaging all Euclidean distances in niche positions in climatic space between species pairs within a family or genus (Within) or between species pairs that do not share the same family or genus (Between). Furthermore, we give the respective standard deviations (SD) and the number of families or genera (n) within each region used in the analysis. For further details on how the distance values were calculated, see Supplementary material Fig. S1. $ The ANOSIM statistic (rANOSIM) and the associated p values give estimations on the likelihood that the observed differences were significantly different from 0. (For more details, see text.) Values are given separately for the three amphibian orders and for each biogeographical region. ANOSIM values showing significantly larger niche distances for between-group than for within-group species pairs are indicated in bold. Note that the analyses could not be conducted on the family level for Afrotropical Gymnophiona, as all species occurring there belong to the same family. AFR, Afrotropics; AUS, Australasia; IND, Indo-Malay; MAD, Madagascar; NEA, Nearctic; NEO, Neotropics; PAL, Palaearctic.

potentially unreliable niche characterisations. Therefore, results for the full analyses are presented. The observed proportions of explained variance at higher taxonomic levels were consistently larger than those yielded by the null models, which simulated the case of no phylogenetic signal (the only exception were Indo-Malayan Caudata). However, when comparing different regions and orders, we found considerable variation (Fig. 1): for Anura, variance explained above the species level ranged from 7% (Madagascar) up to 76% (Australasia), with most values exceeding the mean value of 49% (averaged across all regions and both OMI ordination axes). For Caudata, values ranged from 0% (Indo-Malay) to 87% (Palaearctic), and the mean was 50%. For Gymnophiona, extreme values for niche variance explained above the species level were 246

10% (Indo-Malay and Neotropics) and 70% (Afrotropics), with a mean of 34%. Tests of niche differences with ANOSIM revealed that within-group niche distances were significantly smaller than between-group distances in the vast majority of cases (Table 1). This outcome matches the findings of the VCA, also indicating the presence of a phylogenetic signal in climatic niches at both the genus and the family levels for most regions and taxa. Again, running the analyses with all species or using only those species for which OMI performed significantly well rendered highly consistent results (Supplementary material Table S3). Despite the consistent trend of within-group niche distances being smaller than between-group distances, we found a small number of cases deviating from the overall pattern. At the

Table 2. Climatic niche overlap analyses for amphibians on the family and genus levels, along the first and second OMI axes (OMI1 and OMI2), within different biogeographical regions. n*

OMI1 Pooled$

Families Anura AFR AUS IND MAD NEA NEO PAL

15 9 12 3 8 19 11

(2) (0) (1) (2) (3) (1) (3)

W B*** W B*** W B*** WBB (n.s.) W B** W B*** W B (n.s.)

Caudata IND NEA NEO PAL

2 7 2 4

(1) (1) (1) (1)

Gymnophiona IND NEO

1 (1) 2 (0)

W B

% WBB

OMI2 Pooled

W B

WBB

10 7 7 0 1 8 3

(0) (1) (1) (0) (2) (3) (4)

0 0 2 0 0 3 1

(5) (1) (2) (3) (5) (5) (3)

W B*** W B*** W B*** W B*** W B (n.s.) W B (n.s.) W B**

6 6 7 1 0 9 4

(4) (2) (1) (1) (3) (1) (2)

1 0 1 1 0 3 3

(4) (1) (3) (0) (5) (6) (2)

W B* WBB*** W B*** W B*

1 2 2 1

(1) (3) (0) (1)

0 1 0 0

(0) (1) (0) (2)

W B (n.s.) W BB*** WBB (n.s.) W B***

0 3 0 2

(1) (2) (1) (0)

0 1 0 0

(1) (1) (1) (2)

WBB (n.s.) W B (n.s.)

0 (0) 0 (1)

0 (1) 0 (1)

WBB (n.s.) W B***

0 (0) 1 (0)

0 (1) 0 (1)

Genera Anura AFR AUS IND MAD NEA NEO PAL

47 34 60 16 15 112 38

(34) (17) (32) (8) (15) (52) (18)

W B*** W B*** W B*** W B (n.s.) W B* W B*** W B***

16 17 22 3 2 39 5

(18) (7) (21) (2) (6) (27) (19)

0 1 0 2 0 4 0

(13) (9) (17) (9) (7) (42) (14)

W B*** W B*** W B*** W B*** W B (n.s.) W B*** W B***

12 9 25 3 4 32 8

(10) (11) (17) (3) (2) (34) (13)

2 1 4 1 0 5 1

(23) (13) (14) (9) (9) (41) (16)

Caudata IND NEA NEO PAL

5 18 7 16

(2) (7) (6) (13)

W B (n.s.) W B*** W B*** W B***

0 4 5 4

(3) (6) (0) (2)

0 1 0 0

(2) (7) (2) (10)

W B (n.s.) WBB* W B* W B***

0 6 0 4

(3) (3) (4) (1)

0 1 0 0

(2) (8) (3) (11)

W B (n.s.) WBB (n.s.) W B***

0 (2) 1 (0) 1 (1)

WBB (n.s.) W B (n.s.) W B***

0 (0) 1 (1) 2 (2)

Gymnophiona AFR IND NEO

4 (7) 2 (3) 9 (7)

0 (2) 0 (1) 0 (7)

0 (4) 0 (0) 0 (5)

*Number of groups (families or genera, respectively) included in the overlap analyses. Values in brackets give the number of groups for which tests could not be performed (e.g. groups that included a single species only or that only consisted of species occurring in a single grid cell). $ For the pooled comparisons, all within- (W) and all between-group (B) values of niche overlap (families or genera, respectively) were pooled and then compared using Wilcoxon rank sum tests. ‘‘W B’’ indicates that within-group overlap was larger than between-group overlap (which would indicate a phylogenetic signal). Asterisks indicate significance levels, ***p B0.001, **pB0.01, *pB0.05, n.s., not significant. % Numbers of groups (families or genera) showing within-group niche overlap being larger or smaller than between-group overlap (numbers in brackets indicate the number of within- or between-group comparisons that were not significant). AFR, Afrotropics; AUS, Australasia; IND, Indo-Malay; MAD, Madagascar; NEA, Nearctic; NEO, Neotropics; PAL, Palaearctic.

showed the overall pattern. As for the other analyses, the results varied considerably among regions, taxa, and the two OMI axes.

Discussion

247

ISSUE

Our analyses provide evidence in support of the idea that phylogenetically related species have similar realised climatic niches, even though the strength of the phylogenetic signal varied considerably across amphibian orders and biogeographical regions. To our knowledge, this is the first study investigating phylogenetic niche signals across an entire class of organisms on a global scale, nevertheless accounting for regional variation. Thus it provides a starting point to address questions related to evolutionary niche dynamics of amphibians.

IBS SPECIAL

family level, 3 out of 13 analyses showed larger withingroup distances than between-group distances. At the genus level, within-group distances were larger than betweengroup distances only for 2 out of 14 data sets (see Table 2 for details). Niche overlap analyses showed that in the majority of cases within-group overlap was larger than between-group overlap (Table 2). In the comparison of pooled within- and between groups, within-family overlap was significantly larger than between-family overlap in 8 out of 13 datasets along the first OMI axis and in 7 out of 13 datasets along the second OMI axis. On the genus level, within-group overlap was significantly larger than between-group overlap in 10 out of 14 datasets along the first OMI axis and in 9 out of 14 datasets along the second OMI axis. Comparing within- and between-group overlap separately for each family or genus per region, still the majority of datasets

ISSUE

IBS SPECIAL

Overall, we found a phylogenetic signal in amphibians’ realised climatic niches, as was first shown at the family level by Blomberg’s randomisation test and K statistic. However, the strength of the signal differed considerably for the two niche axes. Both the VCA and the niche overlap analyses, which were done separately for the different regions and orders, supported the existence of a phylogenetic niche signal among amphibians in the majority of the datasets. Applying a different methodology (ANOSIM) again supported the general finding of a phylogenetic signal. Admittedly, the values for the ANOSIM statistic (rANOSIM, Table 1) are relatively low in many cases, even though the p values indicated statistical significance. These low values may often be attributable to the high species numbers in some regions (e.g. Neotropical or Indo-Malayan Anura), resulting in high significance levels even though the differences might be weak. However, the general tendency confirmed by four different methods and across the majority of the species sets analysed supports the conclusion that the trend is robust. Only a few studies have measured phylogenetic niche signals of clades at large geographical scales. For European plants, Prinzing et al. (2001) found that 28 75% of amongspecies niche variance (niche positions along environmental gradients) was explained at higher taxonomic levels. This result is roughly concordant with our findings. For central European spiders, 20 40% of the variance in niche position in shading and moisture was explained at higher taxonomic levels (Entling et al. 2007). However, the spiders’ phylogenetic signal in ecological traits was consistently lower than in morphological traits ( 70% of morphological variance explained above the species level). For dietary niches of European birds, Bra¨ndle et al. (2002) found that about 70% of the variance was explained at higher taxonomic levels. Even though there are a limited number of studies to compare our results with, our findings are consistent with results previously reported for phylogenetic signals in climatic niches, and also with those in morphological traits or dietary niches. Despite an overall and robust trend of detection of phylogenetic signals in climatic niches, we found considerable variation in the strength of the signal among biogeographical regions and the three amphibian orders. Further analyses are needed to examine such variation in detail. In the context of this study we can only discuss some of the limitations of the analysis and some of the most striking findings. As mentioned before our analyses are based on characterizations of species realised climatic niches. Such niches are incomplete representations of species’ true limits of tolerance to climate variables and so cannot entirely portray fundamental climatic niches (Sobero´n 2007). Obviously the possibility of existence of strong mismatches between the observed realised and the fundamental niches decrease the likelihood of detecting a phylogenetic (i.e. evolutionary) signal and it is impossible to rule out that such mismatches may have caused weak phylogenetic signal in some of our data sets. Nevertheless, given 1) this conceptual mismatch between realised and fundamental climatic niches, and 2) that except for the global family-level analysis we use a

248

taxonomy (albeit based on recent phylogenies) that introduced further uncertainties (see below), one could expect that any phylogenetic signal would be obscured. Given this potential for ambiguity, identifying a consistent pattern across most of the regions even with the data and methods used rather strengthens the conclusion that a phylogenetic signal exists in amphibian climatic niches. However, the coarse spatial resolution of the data may on the other hand weaken the information content of the results. Many of the grid cells (which cover areas of almost 50 000 km2 at the equator) contain strong climatic gradients and exceed the range of many species. Assigning closely related species within one grid cell to the same climatic niche although they actually have very different climatic preferences could inflate the phylogenetic signal. In fact, within areas of rather small extent, closely related species may show strong tendencies of niche divergence (Graham et al. 2004, Knouft et al. 2006, Kozak and Wiens 2007). However, an inflation of the phylogenetic signal should not occur if such species with different niches within the same grid cell belong to different genera or families assigning them to the same niche position would in this case rather weaken the phylogenetic signal in our analyses. In any case, we cannot fully discard the potential inflation of the phylogenetic signal’s strength here, but we emphasize that with our analyses we do not and cannot unravel complex evolutionary mechanisms such as speciation (Kozak and Wiens 2006) or the phylogenetic structuring of local or regional communities (Webb et al. 2002), all of which require data at a much finer spatial and phylogenetic resolution. Besides methodological factors, geographic, taxonomic and climatic idiosyncrasies contribute to the observed variation in the strength of the phylogenetic signal. For Anura, e.g. only Nearctic genera showed a result contradictory to the overall pattern of the ANOSIM analysis, the within-group similarity being slightly lower than the between-group similarity. This result was driven by the low niche similarity within the genus Lithobates, which is the largest genus in the Nearctic Anura (30 species). An examination of the different species reveals that some are widely distributed (e.g. L. sylvatica, L. catesbeiana), but others (e.g. L. dunni, L. onca, L. sevosus) have small ranges located in very different regions within the Nearctic and thus have very different climatic niches. This combination of high species richness and a high within-genus variety of climatic niches may have contributed to the low phylogenetic signal in the Nearctic Anuran genera. Furthermore, taxonomic misclassifications may also influence the failure of detection of a phylogenetic signal (Blomberg et al. 2003). This is a general issue for our analyses, of course, but may be particularly important for Nearctic Anura, as classification of Lithobates as a genus remains controversial (Hillis and Wilcox 2005, Frost et al. 2006, Che et al. 2007). A rather weak phylogenetic signal was also detected at the family level of Nearctic Caudata as indicated by the VCA and the niche overlap analysis (Fig. 1, Table 2). Here, the family Plethodontidae comprises more than three times as many species as the other families combined (143 vs 43 species). The highly diverse Plethodontid salamanders

References Ackerly, D. 2009. Conservatism and diversification of plant functional traits: evolutionary rates versus phylogenetic signal. Proc. Nat. Acad. Sci. USA 106: 19699 19706. Arau´jo, M. B. and Guisan, A. 2006. Five (or so) challenges for species distribution modelling. J. Biogeogr. 33: 1677 1688. Arau´jo, M. B. et al. 2006. Climate warming and the decline of amphibians and reptiles in Europe. J. Biogeogr. 33: 1712 1728. Arau´jo, M. B. et al. 2008. Quaternary climate changes explain diversity among reptiles and amphibians. Ecography 31: 8 15. Austin, M. P. et al. 1990. Measurement of the realized qualitative niche environmental niches of 5 Eucalyptus species. Ecol. Monogr. 60: 161 177. Blackburn, D. C. 2008. Biogeography and evolution of body size and life history of African frogs: phylogeny of squeakers (Arthroleptis) and long-fingered frogs (Cardioglossa) estimated from mitochondrial data. Mol. Phylogenet. Evol. 49: 806 826. Blomberg, S. P. and Garland, T. 2002. Tempo and mode in evolution: phylogenetic inertia, adaptation and comparative methods. J. Evol. Biol. 15: 899 910. Blomberg, S. P. et al. 2003. Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution 57: 717 745. Botkin, D. B. et al. 2007. Forecasting the effects of global warming on biodiversity. Bioscience 57: 227 236. Bra¨ndle, M. et al. 2002. Dietary niche breadth for central European birds: correlations with species-specific traits. Evol. Ecol. Res. 4: 643 657. Broennimann, O. et al. 2007. Evidence of climatic niche shift during biological invasion. Ecol. Lett. 10: 701 709. Carey, C. and Alexander, M. A. 2003. Climate change and amphibian declines: is there a link? Divers. Distrib. 9: 111 121. Che, J. et al. 2007. Phylogeny of Raninae (Anura: Ranidae) inferred from mitochondrial and nuclear sequences. Mol. Phylogenet. Evol. 43: 1 13. Chessel, D. et al. 2004. The ade4 package I one-table methods. R News 4: 5 10. Clarke, K. R. 1993. Non-parametric multivariate analyses of changes in community structure. Aust. J. Ecol. 18: 117 143. Colwell, R. K. and Rangel, T. F. 2009. Hutchinson’s duality: the once and future niche. Proc. Nat. Acad. Sci. USA 106: 19651 19658. Corey, S. J. and Waite, T. A. 2008. Phylogenetic autocorrelation of extinction threat in globally imperilled amphibians. Divers. Distrib. 14: 614 629. Darwin, C. 1859. On the origin of species by means of natural selection, or the preservation of favoured races in the struggle for life. John Murray. Dole´dec, S. et al. 2000. Niche separation in community analysis: a new method. Ecology 81: 2914 2927. Dormann, C. F. et al. 2010. Evolution of climate niches in European mammals? Biol. Lett. 6: 229 232.

249

ISSUE

Acknowledgements We are grateful to Andre´s Baselga, Jochen Bihn, Martin Bra¨ndle, David Nogue´s-Bravo, Ken Kozak, three anonymous reviewers, and, in particular, to Irina Levinsky for

comments on earlier versions of this manuscript. Special thanks to Susanne Fritz for helpful comments and continuous R support. We also thank Michael Krabbe Borregaard, Roland Brandl and Walter Jetz for fruitful discussions. CH and CR acknowledge the Danish National Research Foundation for support to the Center for Macroecology, Evolution and Climate. MBA is funded by the EC FP6 ECOCHANGE project (Contract No 036866-GOCE).

IBS SPECIAL

occupy a great variety of niches (Vieites et al. 2007), whereas the niches of species within each Plethodontid genus are very similar (e.g. for Hydromantes, Desmognathus, or Batrachoseps). Indo-Malayan Caudata showed no clear pattern, possibly because of the low species richness of Caudata in this region (25 species) and because for many species the distributions used represent only a small part of their entire distribution. Thus, this species set is rather negligible. However, we stress that in the vast majority of cases, the total ranges of species are contained within one biogeographical region. Some authors question whether establishing the existence of a phylogenetic signal is a useful pursuit (Wiens and Graham 2005, Wiens 2008). In parallel, calls for ‘‘further research into the extent and occurrence of PNC [phylogenetic niche conservatism], and phylogenetic signal more generally’’ (Losos 2008a, p. 1001) are also common. Testing for the existence of a phylogenetic signal is important as the assumption underlies several types of studies, such as the investigation of diversity gradients and the building of species distribution models for climate change prediction, and because its generality is still under debate. Although the aim of our study was to test for a phylogenetic signal in climatic niches, our analyses provide a baseline for further investigations on climatic niche conservatism in amphibians. Phylogenetic niche conservatism can be defined as the tendency of closely related species to be more similar than expected based on phylogenetic relationships (Losos 2008a); put more broadly, it is the temporal constancy of the ecological niche (Pearman et al. 2008, Nogue´s-Bravo 2009). Niche conservatism is a topic of recent growing interest (Peterson et al. 1999, Prinzing et al. 2001, Wiens 2004, Wiens and Graham 2005, Kozak and Wiens 2006, Freckleton and Jetz 2009, Vieites et al. 2009a, Dormann et al. 2010). Overall, its generality or even existence remains a matter of controversial debate (Losos 2008a, Pearman et al. 2008). Based on our findings, we can draw two conclusions with regard to phylogenetic niche conservatism and temporal niche constancy in amphibians. First, as recently pointed out by Losos (2008a, p. 997), a ‘‘lack of phylogenetic signal is sufficient to indicate that PNC does not occur.’’ Thus, based on our detection of a phylogenetic signal in climatic niches, the niche conservatism hypothesis cannot be rejected. Second, for several regions, we found high values of among-species niche variance explained above the family level (Afrotropical and Palaearctic Anura: 30%, Australasian Anura: 60%). This result lends support to the suggestion of the existence of considerable constancy in climatic niches for a period of time that reaches back to the late Cretaceous or even earlier ( 65 Mya ago), when many of the above-family splits took place (Roelants et al. 2007). However, further studies are needed using finely resolved phylogenetic and climatic data to test for the occurrence and strength of phylogenetic conservatism in amphibian climatic niches.

ISSUE

IBS SPECIAL

Duellman, W. E. 1999. Patterns of distribution of amphibians a global perspective. John Hopkins Univ. Press. Entling, W. et al. 2007. Niche properties of central European spiders: shading, moisture and the evolution of the habitat niche. Global Ecol. Biogeogr. 16: 440 448. Freckleton, R. P. and Jetz, W. 2009. Space versus phylogeny: disentangling phylogenetic and spatial signals in comparative data. Proc. R. Soc. B 276: 21 30. Frost, D. R. 2007. Amphibian species of the world: an online reference. Version 5.1. <http://research.amnh.org/ herpetology/amphibia/index.php>. Frost, D. R. et al. 2006. The amphibian tree of life. Bull. Am. Mus. Nat. Hist., American Museum of Natural History. Garland, T. 1992. Rate tests for phenotypic evolution using phylogenetically independent contrasts. Am. Nat. 140: 509 519. Graham, C. H. et al. 2004. Integrating phylogenetics and environmental niche models to explore speciation mechanisms in dendrobatid frogs. Evolution 58: 1781 1793. Hijmans, R. J. et al. 2005. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25: 1965 1978. Hillis, D. M. and Wilcox, T. P. 2005. Phylogeny of the New World true frogs (Rana). Mol. Phylogenet. Evol. 34: 299 314. Hollander, M. and Wolfe, D. A. 1999. Nonparametric statistical methods. Wiley. Hutchinson, G. E. 1957. Population studies animal ecology and demography concluding remarks. Cold Spring Harbor Symp. Quant. Biol. 22: 415 427. IUCN 2004. Global amphibian assessment. <www. globalamphibians.org/>. Kembel, S. W. et al. 2009. picante: R tools for integrating phylogenies and ecology. R package ver. 0.7-0. Knouft, J. H. et al. 2006. Phylogenetic analysis of the evolution of the niche in lizards of the Anolis sagrei group. Ecology 87: S29 S38. Kozak, K. H. and Wiens, J. J. 2006. Does niche conservatism promote speciation? A case study in North American salamanders. Evolution 60: 2604 2621. Kozak, K. H. and Wiens, J. J. 2007. Climatic zonation drives latitudinal variation in speciation mechanisms. Proc. R. Soc. B 274: 2995 3003. Losos, J. B. 2008a. Phylogenetic niche conservatism, phylogenetic signal and the relationship between phylogenetic relatedness and ecological similarity among species. Ecol. Lett. 11: 995 1003. Losos, J. B. 2008b. Rejoinder to Wiens (2008): phylogenetic niche conservatism, its occurrence and importance. Ecol. Lett. 11: 1005 1007. Nogue´s-Bravo, D. 2009. Predicting the past distribution of species climatic niches. Global Ecol. Biogeogr. 18: 521 531. Oksanen, J. et al. 2009. vegan: community ecology package. R package ver. 1.15-2. Olson, D. M. et al. 2001. Terrestrial ecoregions of the worlds: a new map of life on Earth. Bioscience 51: 933 938. Paradis, E. et al. 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20: 289 290. Pearman, P. B. et al. 2008. Niche dynamics in space and time. Trends Ecol. Evol. 23: 149 158.

Download the Supplementary material as file E6309 from <www.oikos.ekol.lu.se/appendix>.

250

Peterson, A. T. et al. 1999. Conservatism of ecological niches in evolutionary time. Science 285: 1265 1267. Prinzing, A. et al. 2001. The niche of higher plants: evidence for phylogenetic conservatism. Proc. R. Soc. B 268: 2383 2389. R Development Core Team 2008. R: a language and environment for statistical computing. R Foundation for Statistical Computing. Revell, L. J. et al. 2008. Phylogenetic signal, evolutionary process, and rate. Syst. Biol. 57: 591 601. Rheindt, F. E. et al. 2004. Rapidly evolving traits and the comparative method: how important is testing for phylogenetic signal? Evol. Ecol. Res. 6: 377 396. Roelants, K. et al. 2007. Global patterns of diversification in the history of modern amphibians. Proc. Nat. Acad. Sci. USA 104: 887 892. Santos, J. C. et al. 2009. Amazonian amphibian diversity is primarily derived from Late Miocene Andean lineages. PloS Biol. 7: 448 461. Sclater, P. L. 1858. On the general geographical distribution of the members of the class aves. J. Proc. Linn. Soc. (Zool.) 2: 130 145. Sobero´n, J. 2007. Grinnellian and Eltonian niches and geographic distributions of species. Ecol. Lett. 10: 1115 1123. Sobero´n, J. and Nakamura, M. 2009. Niches and distributional areas: concepts, methods, and assumptions. Proc. Nat. Acad. Sci. USA 106: 19644 19650. Thuiller, W. et al. 2004. Relating plant traits and species distributions along bioclimatic gradients for 88 Leucadendron taxa. Ecology 85: 1688 1699. Venables, W. N. and Ripley, B. D. 1999. Modern applied statistics with S-Plus. Springer. Vieites, D. R. et al. 2007. Rapid diversification and dispersal during periods of global warming by plethodontid salamanders. Proc. Nat. Acad. Sci. USA 104: 19903 19907. Vieites, D. R. et al. 2009a. Reconstruction of the climate envelopes of salamanders and their evolution through time. Proc. Nat. Acad. Sci. USA 106: 19715 19722. Vieites, D. R. et al. 2009b. Vast underestimation of Madagascar’s biodiversity evidenced by an integrative amphibian inventory. Proc. Nat. Acad. Sci. USA 106: 8267 8272. Wallace, A. R. 1876. The geographical distribution of animals. Macmillan. Webb, C. O. et al. 2002. Phylogenies and community ecology. Annu. Rev. Ecol. Syst. 33: 475 505. Wells, K. W. 2007. The ecology and behavior of Amphibians. Univ. of Chicago Press. Wiens, J. J. 2004. Speciation and ecology revisited: phylogenetic niche conservatism and the origin of species. Evolution 58: 193 197. Wiens, J. J. 2007. The amphibian tree of life (book review). Q. Rev. Biol. 82: 55 56. Wiens, J. J. 2008. Commentary on Losos (2008): niche conservatism deja vu. Ecol. Lett. 11: 1004 1005. Wiens, J. J. and Donoghue, M. J. 2004. Historical biogeography, ecology and species richness. Trends Ecol. Evol. 19: 639 644. Wiens, J. J. and Graham, C. H. 2005. Niche conservatism: integrating evolution, ecology, and conservation biology. Annu. Rev. Ecol. Evol. Syst. 36: 519 539.

Ecography 33: 251 259, 2010 doi: 10.1111/j.1600-0587.2010.06306.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Catherine Graham. Accepted 15 May 2010

Patterns and timing of diversification in a tropical montane butterfly genus, Lymanopoda (Nymphalidae, Satyrinae) Kayce L. Casner and Tomasz W. Pyrcz K. L. Casner (kaycelu@gmail.com), Dept of Evolution and Ecology and Center for Population Biology, Univ. of California, Davis, CA 95616, USA. T. W. Pyrcz, Zoological Museum, Jagiellonian Univ., 30 060 Krakow, Poland.

Species distributions are a product of contemporary and historical forces. Using phylogenetic and geographic data, we explore the timing of and barriers to the diversification of the Andean butterfly genus Lymanopoda (Nymphalidae, Satyrinae). Clade and species level diversification is coincident with Andean orogeny and Pleistocene glaciation cycles. Lymanopoda has primarily diversified within elevational bands, radiating horizontally throughout the Andes with occasional speciation across elevational boundaries, often associated with ecotones. Narrow elevational ranges and infrequent speciation into adjacent elevational strata suggest that expansion across elevational gradients is relatively difficult. These results are similar to those found in studies of other Andean taxa.

251

ISSUE

Mountain ranges are ideal environments for studying speciation and species distributions. They provide high peaks and low valleys, which provide insurmountable topographic barriers that prevent gene flow and facilitate geographic isolation. Intricate topography and clines in environmental variables such as temperature, aridity, and solar insolence provide opportunity for ecological speciation (Chapman 1917, Elias et al. 2009). The South American Andes in particular are considered a ‘‘species pump’’ (Fjeldsa and Lovett 1997) in the Neotropics, but the mechanistic role that mountains might play in contributing to species richness and diversity patterns remains unclear. Butterflies of the subtribe Pronophilina (Nymphalidae, Satyrinae) are an excellent group to study patterns of diversification. It is the most speciose group of butterflies in the Andean cloud forest with 400 described and 200 undescribed species (Adams 1985, 1986, Pyrcz and Wojtusiak 2002). This species richness is especially impressive in light of the young age of the Andes, which did not reach their present height until 10 6 Ma in the central section (Garzione et al. 2006) and ca 2.7 Ma in the north (Gregory-Wodzicki 2000). Sets of pronophilines are allopatric along the Andes where deep valleys and high passes act as dispersal barriers. Within each part of the Andes, lower-elevation species with broad geographic ranges are replaced by higher-elevation, narrowly distributed congeners. The result is a stair-step distribution up mountain sides with unique species composition in adjacent regions of an extended mountain chain or in nearby cordilleras (Adams 1985, Pyrcz et al. 1999).

IBS SPECIAL

As anthropogenic forces act to alter species’ natural distributions, it is more important than ever to understand ‘‘why species are where they are’’. MacArthur (1972) wrote in the first lines of his commanding work, Geographical ecology, ‘‘To do science is to search for repeated patterns, not simply accumulate facts and to do the science of geographical ecology is to search for patterns of plant and animal life that can be put on a map’’. Contemporary patterns in plant and animal distributions are the product of existing abiotic and biotic forces limiting ranges, and a history of adaptation and speciation in response to past forces. Previous studies have documented and tested existing biotic and abiotic pressures associated with species boundaries (Heller 1971, Terborgh 1971, Heller and Poulson 1972, Brown et al. 1996, Holt and Keitt 2005). In this paper, we explore the historical responses of species to physical and ecological boundaries through patterns of speciation in a montane environment. Environmental gradients are effective backgrounds for studying species distributions and speciation. Abiotic factors impeding range expansion, such as temperature, aridity and salinity, are apparent and easy to measure along gradients. Additionally, distributional patterns are relatively obvious with respect to the environment (Terborgh 1971). Often, changes in abiotic factors along gradients are salient and conspicuously co-vary with biotic richness and composition, making biotic and abiotic barriers more detectable. Over evolutionary time, species’ responses to environmental barriers become apparent through range maintenance, range expansion (Kirkpatrick and Barton 1997), or speciation (Schluter 1996).

IBS SPECIAL

ISSUE

This study focuses on the pronophiline genus Lymanopoda, which displays the aforementioned distributional traits of the subtribe. The genus is apparently monophyletic as it shows several synapomorphies in the adult morphology: 1) the ocelli in forewing cells Cu1 and Cu2 are displaced basally, 2) the hindwing ventral surface median band is broken and displaced in the discal cell, 3) male genitalia possess a superuncus, 4) the female genitalia possess a sclerotised lamella on the distal part of the posterior apophysis (Pyrcz et al. 1999, Pyrcz 2004). Currently, 66 species are recognized (including 5 undescribed) (Lamas 2004). The species of Lymanopoda are exclusively montane and found between 1000 1200 and 3800 4000 m. Two species occur in Central America, and the other 64 species are found in the Andes and its peripheral cordilleras (Sierra Nevada de Santa Marta, Sierra de Turimiquire). The highest species richness occurs at a latitude of 108S on the eastern slopes of the Andes in central Peru (Pasco) where 14 species occur along an elevation gradient. Species richness decreases gradually north- and southwards reaching its distributional limit at latitude 178N and 278S (Fig. 1). The larvae are oligophagous on montane bamboo, primarily of the widespread Andean genus Chusquea in the cloud forests (Adams and Bernard 1981) and Swallenchloa in the paramo (Pyrcz et al. 1999). Adults exhibit low vagility, usually staying in close proximity to their bamboo host plants. Here we explore the evolutionary patterns for 40 species of Lymanopoda along elevational gradients and a north south transect in the South American Andes. Specifically, we test two hypotheses revised from Moritz et al. (2000). 1) Diversification was vertical. This hypothesis is similar to the gradient model whereby speciation is based on divergent selection along an elevational gradient. By this scenario, species along an elevational gradient are more closely related

to one another than to species at similar elevations in neighboring regions. 2) Diversification was horizontal. This is similar to the ‘‘refuge’’ hypothesis proposed by Moritz et al. (2000) whereby speciation is based on allopatry. In this case, closely related species occupy similar elevational bands in adjacent regions. Each hypothesis produces a unique phylogenetic pattern with respect to the geographic ranges of closely related taxa, which we test using a multilocus data set.

Materials and methods Sample collection Most of the 228 specimens were collected by the first author between 2006 2008 and second author between 2004 2008. Other specimens were received as donations from museum and personal collections. We collected specimens from ca 75 locations throughout the northern and central Andes (Bolivia, Peru, Ecuador, Colombia, Venezuela) and Costa Rica. Where possible, multiple samples were obtained, although some species, particularly higher elevation species, are rare and only a single individual was collected. Samples of the sister genus Ianussiusa (Pen˜a et al. 2006) were collected from its range in Colombia and Ecuador and used as the outgroup. We determined elevational and latitudinal range limits through widespread sampling between 1991 and 2009. Elevational ranges were established by sampling along altitudinal gradients at 1200 4200 m. When field work at a single locality lasted 5 d, Van Someren-Rydon baited traps (DeVries 1987) were placed every 50 m in elevation and baited with excrement of carnivorous animals. We checked traps and added fresh bait daily (Pyrcz 2004). When B5 d were available to sample a locality, we used standard entomological nets to collect specimens. To confirm taxonomy of samples, the second author carried out morphological studies of adults at the Museum of Zoology of the Jagiellonian Univ. Samples were also compared to type specimens housed in other major European collections. Systematic arrangement of the genus Lymanopoda follows Lamas (2004). Species were defined using the biological species concept in which species recognition is based on morphological traits that facilitate reproductive isolation, primarily genitalic features and differences in wing pattern. Adult Satyrinae communicate visually, and therefore wing color and markings are considered important to intraspecific recognition. Molecular methods

Figure 1. Andean distribution of genus Lymanopoda.

252

QIagen’s DNEasy extraction kit was used for in-house extractions. DNA was extracted from abdomens for most samples, although in a few cases only legs were available. Purified DNA was resuspended in elution buffer and stored at 208C. We amplified and sequenced 1458 bp of Cytochrome Oxidase subunit I (COI) of the mitochondrion and 4 nuclear genes 403 bp of wingless (wg), 606 bp of ribosomal protein S5 (RpS5), 691 bp of glyceraldehydes3-phosphate dehydrogenase (GAPDH), and 751 bp of elongation factor-1 alpha (EF-1 alpha) following the

PCR protocols, with minor variations, laid out by PenË&#x153;a et al. (2006) and Wahlberg and Wheat (2008). Primer sequences and PCR cycling protocols, are also available at <http://nymphalidae.utu.fi/Nymphalidae/Molecular.htm>. All PCR products were sequenced in both directions, and in most cases there was complete overlap of fragments. All sequencing was done on the Univ. of California at Davis campus in the College of Agriculture and Environmental Sciences Genomics Facility. For a table of specimens, geographic data and accession numbers, see Supplementary material Table S1. Phylogenetic analyses

Dating the tree

253

ISSUE

To estimate divergence times for Lymanopoda, we used a fossil of butterfly Lethe carbieri (Nymphalidae, Satyrinae) from the Late Oligocene (ca 25 Ma) (Nel et al. 1993). Butterfly fossils are exceedingly rare, and the specimen of Lethe carbieri is the closest known fossil record to the genus Lymanopoda, both of which are members of the Satyrini tribe (PenË&#x153;a and Wahlberg 2008, Wahlberg et al. 2009). To estimate the timescale of diversification for Lymanopoda we used the program BEAST 1.5.2 (Drummond and Rambaut 2007) run through the Computational Biology Suite for High Performance Computing (BioHPC) portal housed at Cornell Univ. MODELTEST v3.7 (Posada and Crandall 1998) was used to identify the optimal model of substitution based on the hierarchical likelihood ratio tests (hLRTs) for each gene (EF1 alpha: TrNef G I; GAPDH: TrN G I; RpS5: TrNef G; wg: HKY G I; COI: GTR G I). The BEAST analysis was partitioned by gene and included the model of evolution, unique gamma distribution shape parameter (G), and proportion of invariable sites (I). We used a normal prior for the age of the fossil calibration, which was set to the age of divergence between Lethe and Neope (m 25 Ma and s 1.0). The topology and branch lengths from the Bayesian analysis including Lethe and Neope were used as a starting tree, and a Yule model (Yule 1924, Aldous 2001) was assigned to the tree prior. To account

IBS SPECIAL

When developing biogeographic hypotheses based on phylogenies, it is critical that species relationships be robust. No single analytical tool available today is ideal, therefore we used a multi-step process relying on several analytical techniques to determine and evaluate the best species tree. First, we analyzed each gene independently and in combination using Bayesian analyses with the program MrBayes 3.1.2 (Huelsenbeck and Ronquist 2001, Ronquist and Huelsenbeck 2003). One combination included all genes, mitochondrial and nuclear, and a second analysis included only concatenated nuclear genes. Because concatenation methods have been shown to produce species trees inconsistent with the â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;trueâ&#x20AC;&#x2122;â&#x20AC;&#x2122; phylogeny under certain circumstances (Degnan and Rosenberg 2006, Pollard et al. 2006, Kubatko and Degnan 2007), we used the program Bayesian Untangling of Concordance Knots (BUCKy) (Ane et al. 2007) to identify clades with the greatest concordance among gene trees and assess the validity of concatenated analyses. Sequences were aligned using Sequencher 4.8 (Gene Codes Corporation, Ann Arbor, MI) and codon positions were defined using MacClade 4.06 (Maddison and Maddison 2000). We used the CIPRES portal (Miller et al. 2009) to perform Bayesian analyses on individual sequences using MrBayes. Each gene was partitioned by codon position, and the GTR (generalized time reversible) model with gamma-distributed rate variation across sites and a proportion of invariable sites for sequence evolution was specified. We allowed for partition-specific rates by setting the rate parameter to variable and unlinked the model parameters for gene partitions. We performed 2 replicates with 4 chains for 2 107 generations. The temperature was set to 0.10 to enhance mixing, and chains were sampled every 1000 generations. Combined datasets were partitioned by gene and codon position. Again, two replicates and 4 chains were run for 1 107 generations at a temperature of 0.10, and chains were sampled every 1000 generations. To assess convergence of parameters we checked the standard deviation of split frequencies across the independent runs. Using the online software Are We There Yet (AWTY) (Wilgenbusch et al. 2004, Nylander et al. 2007), we assessed convergence of topologies after a 25% burn-in. We visually inspected the 4 diagnostics AWTY provides: posterior probabilities of clades for non-overlapping samples of trees, pair wise split frequencies for independent MCMC runs, cumulative frequencies for selected splits, and the symmetric tree-difference score between and within runs.

Clades from concatenated trees were assessed using BUCKy. BUCKy estimates concordance among sets of genes based on the assumption that the number of distinct topologies among a survey of genes is small compared to all possible topologies (Ane et al. 2007). BUCKy employs a 2-stage Markov chain Monte Carlo (MCMC) method in which the posterior probability distributions of independent gene trees, derived from the Bayesian analyses above, are input to a second MCMC procedure that estimates a posterior distribution of gene-to-tree maps. A summary of the posterior probabilities of the gene-to-tree maps provides revised posterior probability distributions for each gene, accounting for concordance, and an estimate of the proportion of sampled genes for which given clades are true. Additionally, a primary concordance tree is created from clades with the greatest proportion of support from the revised posterior distributions of individual gene trees. BUCKy analyses only accommodate 31 species so we created 13 pruned sub-trees from the original, full-taxon gene trees, with various combinations of species and gene compositions, using the ape package 2.3-2 (Paradis et al. 2004) for R 2.9.1. We used sequenced data from two closely related genera and two more distantly related genera, to confirm monophyly of Lymanopoda. According to Wahlberg et al. (2009), Lymanopoda is sister to genus Ianussiusa or part of a polytomy with Ianussiusa and Idioneurula. Sequences for Idioneurula eremita, Lethe minerva, and Neope bremeri were acquired from GenBank (accession numbers are in Supplementary material Table S1) and included in a Bayesian analysis with similar parameters to the Lymanopoda only analysis run above.

for lineage-specific rate heterogeneity, branch lengths were allowed to vary under a relaxed clock model with an uncorrelated lognormal distribution. We ran five independent chains of ten million generations each, sampling every 1000 generations and combined results with LogCombiner v 1.5.3 (Drummond and Rambaut 2007). Species range overlap The degree of overlap in latitudinal and elevational ranges was calculated at each node using methods described in Fitzpatrick and Turrelli (2006). The purpose of this exercise was not to reconstruct ancestral ranges, but to estimate the average overlap between species ranges after a certain time since speciation. Because ranges closely follow the north south transect of the Andes and there is very little longitudinal variation, we used the latitudinal coordinates as range boundaries. For terminal species pairs, range overlap was calculated by dividing the area of overlap by the area of the smaller species’ range. For internal nodes, we found the nested averages of pairwise overlaps between species’ ranges in each clade. We also calculated the relative difference in the upper elevational limit for subtending clades at each node. For terminal species, the difference in the maximum elevational range was calculated and divided by the full elevational range of the higher species. Elevational range differences for internal nodes were calculated using nested averages as above. We only calculated range overlap and difference for pairwise relationships supported by the majority of gene trees and BUCKy analyses. Range overlap between species varies between 0 (for no overlap) and 1 (for complete overlap). Range difference values may vary between 0 and and exceed 1 if the difference in elevational ranges is greater than the species’ or clade’s full elevational range.

IBS SPECIAL

ISSUE

Statistical test of clade distributions Additionally, we compared elevational and latitudinal distributions among clades. We demarcated the clades as above based on deep divisions in the tree, which were supported by differences in morphology, gene trees and BUCKy analyses. The median of each species’ elevational and latitudinal range was determined from the distribution data, and each clade was assigned a numerical dummy variable. Using Student’s t-test we compared mean median distributions for each clade pair. We used the FlignerKilleen test and Fisher’s F-test to confirm homogeneity of variances. Phylogenetic signal in ranges We tested phylogenetic signal in latitudinal and elevational ranges using Blomberg et al.’s (2003) K in the R package picante (Kembel et al. 2009). We used the same clade demarcations as above to look for phylogenetic signal in latitudinal (northern limit, southern limit, median) and elevational (upper limit, lower limit, median) ranges. 254

Results Lymanopoda is monophyletic relative to sister genus Ianussiusa and the more distantly-related Idioneurula (Supplementary material Fig. S1). Membership of welldefined clades agreed based on individual analysis of gene trees with the exception of a few taxa, primarily L. eubagioides and L. inde (see below) (see Supplementary material Fig. S2 S6 for gene trees). Because clades produced in individual gene trees were broadly concordant, we used results based on all evidence (i.e. concatenation of all genes). BUCKy’s primary concordance trees, using pruned gene trees, supported clades rendered in the all-evidence tree. However, relationships among major clades varied by gene and analysis, preventing inference of deeper relationships. Placement of L. eubagioides and L. inde differed among gene trees and analyses. Nuclear genes consistently placed L. inde in the L. caracara/L.melia clade, and the placement of L. eubagioides varied by gene. The mtDNA segment COI, and all primary concordance trees from BUCKy placed L. eubagioides and L. inde as sister taxa sister to the L. vivienteni/L. rana clade. Lymanopoda eubagioides and L. inde are sister taxa and sister to the L. caracara/L. melia clade in the concatenated analysis (Fig. 2). By our estimation, the genus Lymanopoda diverged from sister genus Ianussiusa ca 27 Ma, which is 5 million years before Wahlberg et al.’s (2009) estimate. In either case, divergence with Ianussiusa was very early in Andean orogeny, which we discuss more below. Formation of most major clades has occurred within the last 8 10 million years, and most of the species-level diversification has occurred within the last 6 million years. More than half of the sampled species are of Pleistocene or post-Pleistocene origin (Fig. 2). Calculations of range overlap are presented in Table 1. Because relationships among the major clades were variable in the above analyses, we only calculated range overlap and differences for nodes within major clades. Of the 32 nodes where overlap was calculated, 17 nodes demonstrate a high degree of elevational overlap, sharing more than 75% of their elevational range with a sister species/clade, while only ten nodes join clades with a similar degree of overlap in their latitudinal range. Likewise, fifteen nodes join species sharing 25% and less of their latitudinal ranges (indicated by nodes with a triangle (') in Fig. 2), and only 4 nodes show a correspondingly low level of overlap in elevational range. Only 5 nodes join sister clades with an elevational difference of 75% or greater (indicated with a circle ( ) in Fig. 2). The differences in proportions of species overlapping horizontally versus vertically demonstrate a stronger signal of speciation associated with latitudinal rather than elevational shifts. A few cases exist in which sister species segregate one below the other along the forest-paramo ecotone, but because of variation in the elevation of the ecotone due to latitude and aspect, this elevational segregation is not apparent in the data. These cases are based on decades of observation, and have been marked with a square (j) in Fig. 2. There were significant differences in the mean elevational and latitudinal distributions of clades (Fig. 3). Members of the ‘‘L. caracara’’ clade have significantly higher elevational ranges than members of all other clades. The ‘‘L. araneola’’ clade also has a significantly higher

Figure 2. A species tree for genus Lymanopoda with associated latitudinal and elevational distributions for each species. Node numbers, left of nodes and above branches, correspond to Table 1. Posterior probabilities are given on the left of nodes below branches. Circles ( ) indicate elevational range shift of at least 75% relative to a sister clade. Triangles (') indicate nodes that join sister taxa that overlap by no more than 25% in their latitudinal ranges. *Lymanopoda euopis is one of two species (the other is L. cinna of Guatemala) of Lymanopoda that occur outside of the Andes.

If we assume that there is a relationship between the mode of speciation and the current distribution of species’ geographic ranges, we may infer the geography of speciation and look for inherent patterns. The relatively young age of most species of Lymanopoda and distinctive allo- and parapatric species ranges indicate minimum levels of dispersal since speciation that might mask speciation

255

ISSUE

Discussion

patterns (Barraclough and Vogler 2000). The phylogeny and species range data suggests that Lymanopoda has diversified both along elevational gradients and horizontally throughout the Andes. This is supported by a strong phylogenetic signal for both elevational and latitudinal range limits. Although there is evidence for both directions of diversification, a prevalence of clades composed of species within the same elevational strata and significant differences in mean elevation among clades suggests that there has been greater diversification within elevation and across latitude than the converse. Approximately half of all nodes in the phylogeny in which we calculated overlap join sister taxa that have a low level of latitudinal overlap, suggesting speciation associated with a horizontal shift into an adjacent mountain range. A latitudinal shift likely indicates allopatric speciation through either dispersal or vicariance. Both hypotheses are plausible. The incidence of dispersal is difficult to test (Voelker 1999), although it has been shown to play an important role in diversification of other Andean taxa (Remsen 1984, Schulte II et al. 2000). Adams (1985) suggested that the stair-step pattern of pronophiline butterflies is the consequence of vicariance facilitated through Pleistocene glaciation cycles (2.588 million 12 000 BP), wherein warm periods caused species

IBS SPECIAL

distribution than the ‘‘L. caeruleata’’ and ‘‘L. affineola’’ clades, and the ‘‘L. caeruleata’’ and ‘‘L. affineola’’ clades occupy significantly lower elevations than all other clades. Several clades were also distinct in their latitudinal range. Members of the ‘‘L. excisa’’ clade occupy ranges significantly further north than members of the ‘‘L. affineola’’ and ‘‘L. araneola’’ clades, and members of the ‘‘L. altis’’ clade are significantly north of members of the ‘‘L. araneola’’ clade. Significant phylogenetic signal was detected in all six latitudinal and elevational range variables tested (lower elevation limit K 0.32, p 0.015; upper elevation limit K 0.29, p 0.001; mid elevation K 0.32, p 0.001; northern limit K 0.31, p 0.01; southern limit K 0.26, p 0.028; mid latitude K 0.48, p 0.001).

IBS SPECIAL

ISSUE

Table 1. Latitudinal and elevational range overlap and elevational difference in the upper range limits for clades joined at a shared node. Node numbers correspond to the numbers above branches in Fig. 2. Node

Latitudinal overlap

Elevational overlap

Diff. in upper range limits

1 2 3 4 5 6 7 8 9 10 11 12 13fh 13fu 13hu 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

1.00 0.70 0.70 0.54 1.00 0.00 0.50 0.25 0.13 0.13 0.03 0.00 1.00 1.00 1.00 0.33 0.67 0.71 0.00 0.00 0.25 1.00 1.00 0.45 0.69 0.80 1.00 0.00 0.05 0.00 0.00 1.00 0.00 0.00

1.00 0.83 0.42 0.69 1.00 1.00 1.00 1.00 1.00 0.44 0.29 1.00 0.00 0.00 1.00 0.58 0.15 0.31 1.00 0.40 0.23 0.33 1.00 0.75 0.83 0.67 1.00 0.33 0.83 0.73 1.00 1.00 0.50 0.59

0.60 0.30 0.15 0.59 0.33 0.00 0.15 0.13 0.06 0.35 0.57 0.00 1.25 1.67 0.00 0.73 0.71 1.09 0.00 0.10 0.86 0.67 0.00 0.33 0.17 0.67 0.00 0.67 0.25 0.10 0.00 0.00 0.86 0.19

to move up-slope, become isolated and diversify, and glacial periods pushed faunas to lower elevations where reinforcement and dispersal into adjacent mountain ranges ensued. He argued that repetitions of this cycle might produce vertical stacking of species’ ranges through completely allopatric speciation processes. While it appears that the Pleistocene certainly contributed to the species-level diversity of Lymanopoda, dating suggests a much earlier origin for the elevational stratification. The geologic and climatic events occurring between the formation of the early Andes to today are complex and under debate (Sempere et al. 2008). Andean orogeny was neither uniform nor simultaneous throughout the range, varying in timing and rate of uplift from east to west and north to south (Gregory-Wodzicki 2000). Most studies suggest that Andean uplift began in the Eocene or Oligocene but then halted until the Late Miocene (Sempere et al. 2008). The western Cordillera of the central Andes was at no more than half of its current elevation 25 Ma, while the eastern Cordillera only reached half of its modern elevation ca 10 Ma. High altitudes probably emerged first in the central Andes, 138S 288S, and progressed northward (Picard et al. 2008). Surface uplift on the order of 2000 3500 m has occurred in the eastern Cordillera and

256

Altiplano in the last 10 million years. Uplift of the northern Andes was very recent, estimated to have been at no more than 40% of its current elevation 4 Ma in some regions (Gregory-Wodzicki 2000). Timing of the split with Ianussiusa between 20 to 30 Ma corresponds with very early Andean orogeny, and subsequent diversification likely occurred when uplift resumed 10 15 million years later. Dramatic uplift during the Miocene and Pliocene coincides with formation of Lymanopoda’s major clades, and continued uplift in the north and intervals of global cooling during the Pleistocene coincide with much of the specieslevel diversification. Speciation across elevational clines occurred multiple times in the early stages of diversification of Lymanopoda and again more recently, particularly within a few clades. Some shifts in elevation are dramatic and apparent in the data, while other shifts in elevation are associated more with habitat type than strict elevation, and therefore, are not obvious in the range data. Ecotones can occur at different elevations depending on latitude and aspect, and some species’ elevational ranges co-vary with specific habitat. Therefore a few species pairs appear to be partially sympatry according to the data, but they are locally parapatric. For example, L. caracara is a cloud forest paramo ecotone specialist, and L. hazelana occurs in high montane cloud forest directly below L. caracara. The forest paramo ecotone occurs between 3600 m in the north of Ecuador and 3150 m in southern Ecuador, and ranges of L. caracara and L. hazelana closely parallel this gradient. A similar pattern is true for L. excisa and L. nivea, which occur in forest just below the edge. In cases where forests have been logged and the ecotone is lower, ranges similarly extend to lower elevations. Habitat segregation, as demonstrated by these species, could be the result of allopatric speciation followed by secondary contact and subsequent niche partitioning or ecological speciation. In these cases, the ecotone is obvious and the butterfly’s association with the ecotone is apparent, but this is the exception. In most cases, elevational ranges are relatively consistent and the biotic or abiotic factors limiting elevational distributions are unknown, which is why we have used elevation rather than microhabitat as a proxy for range boundaries. A similar case of sympatric/parapatric speciation occurred in antbirds (Thamnophilidae, Percnostola) of northern South America (Braun et al. 2005). The roraiman antbird Percostola saturata, endemic to tepuis of southeastern Venezuela and northern Brazil, is elevationally parapatric to its more widespread lowland sister, the spotted-winged antbird Percnostola leucostigma. Species ranges are structured on environmental variables associated with elevation, such as fast-moving water common in highlands and standing pools common in the lowland, and the borders of the ranges interdigitate according to local habitat. Recent theoretical and empirical studies have shown that parapatric speciation along a gradient and with gene flow, might be more realizable and widespread than previously thought (Schneider et al. 1999, Doebeli and Dieckmann 2003, Doebeli et al. 2005). Results of previous studies and speciation patterns in Lymanopoda suggest that ecotones associated with a change

Figure 3. Box and whisker plot of clades versus mean median elevations. Dark horizontal lines show median values. The top and bottom of boxes represent the 25th and 75th percentiles, respectively. Vertical dashed lines show maximum and minimum values or 2 standard deviations, whichever is smaller. Open circles represent outliers.

257

ISSUE

species. A study of 16 500 montane species spanning 48 degrees of latitude recently confirmed his prediction (McCain 2009). Other studies have directly tested the effects of changing environmental conditions on organisms at various elevations. Montane carabid beetles studied along an elevational transect in Wales showed significantly lower optimal body temperatures than widespread, low elevation species (Buse et al. 2001). Similarly, distributions of two congeneric species of jumping plant lice in Norway were restricted elevationally by heat budgets acting on development rates. Craspedolepta nebulosa developed more efficiently at low temperatures and was able to inhabit higher elevations than its congener, C. subpunctata (Bird and Hodkinson 2005). In addition to direct physiological effects, increases in elevation influence forest structure and composition. Forests become progressively more stunted and open with elevation, which may have indirect effects on fauna. The bamboo host plant of Lymanopoda is abundant up to the forest paramo edge. At lower elevations it grows tall and is often shaded by an extensive canopy cover, while at higher elevations the forest is stunted. Therefore, at timberline Chusquea and Lymanopoda are more exposed to variable weather conditions strong winds, sudden changes in temperature and humidity. Differences in the nutritional value of Chusquea at low and high elevations have not been directly tested, but there is a large body of evidence that

IBS SPECIAL

in elevation may facilitate sympatric/parapatric speciation and further expansion into higher elevations. Janzen (1967) outlined a possible mechanistic explanation for the relatively common observation of elevationally stacked species ranges in the tropics in his seminal work, ‘‘Why mountain passes are higher in the tropics’’. He argued that tropical organisms are exposed to minimal seasonal variation, and therefore are adapted to a narrower range of climatic conditions. Because elevation co-varies with temperature, and the temperature difference rather than absolute height likely determines the efficacy of a barrier, tropical organisms are more likely than their temperate counterparts to encounter temperatures beyond their physiological tolerance, and therefore, dispersal is obstructed by elevational changes. The temperature lapse rate in tropical mountains is 0.5 0.68C for each 100 m of elevation gain (Grubb 1977). Lymanopoda span more than 2000 m of elevation, with low and high elevation species experiencing a temperature difference of ca 10 128C. Differences in day and nighttime temperature in cloud forest average 8 148C, which is greater than seasonal differences. This means that a low-elevation species in the highlands or a highland species at lower elevation would certainly experience temperatures outside its normal climate regime. Janzen went on to predict that this should lead to high fidelity to a specific set of abiotic conditions and lead to narrow elevational distributions in tropical-montane

ISSUE

IBS SPECIAL

suggests that plant nutrient composition changes with elevation (Morecroft and Woodward 1996, Erelli et al. 1998, Cordell et al. 1999). We obtained samples for approximately two thirds of the species of Lymanopoda. Many species lacking from this analysis are endemic to Colombia, which has been the subject of less collecting than other parts of the Andes. Most of these species are localized, rare and live at high elevations. Based on morphology we can place the missing species into existing clades. Five of the missing species L. altaselva, L. lactea, L. labineta, L. schmidti and L. paisa belong to the L. ionius clade and resemble each other and L. ionius in morphology and genitalia. Lymanopoda paisa flies at a similar elevation to L. ionius while the other four fly in adjacent regions at elevations just above L. ionius. A similar situation exists for the low elevation species L. obsoleta and L. ferruginosa, which are broadly distributed and replaced by closely related congeners at higher elevation. Lymanopoda lebbaea˜ flies in the eastern Cordillera of Colombia and is most closely aligned with L. labda, which occurs in the Colombian western and central Cordilleras and south into Ecuador. Lymanopoda maletera is probably most closely related to L. dietzi and flies at a similar elevation in Colombia. Similarly, L. melendeza appears to be most closely aligned with L. marianna and occurs at a similar elevation in Colombia. Without molecular analysis we cannot be certain of sister species relationships, and therefore of the mode of speciation. Morphology, however, suggests that many of the missing species are most closely related to allopatric replacements occurring at similar elevations in adjacent regions. Whether covariates of elevation limit Lymanopoda directly or indirectly is unclear, but relatively narrow ranges and less frequent speciation across elevational boundaries suggest that changing environmental factors along a gradient have played a principal role in the pattern of diversification. Gaining insight into the mechanisms limiting the distribution of species with restricted ranges is of great evolutionary interest and is essential to conservation. Changes in climate, invasive species and habitat destruction have the potential to drastically alter habitats and shift species’ geographic ranges. As we take measures to moderate wide-spread diversity loss, it is important to be equipped with knowledge of why species ‘‘are where they are’’. An evolutionary perspective can inform us of types and strengths of barriers that might prevent range expansion or dispersal. Diversification of Lymanopoda butterflies and other tropical montane faunas appear to be constrained elevationally, and thus, geographically by high mountains and deep valleys. Highly defined habitat requirements combined with difficult dispersal make tropical montane faunas especially vulnerable to anthropogenic forces and priorities for protection. Acknowledgements KLC would like to thank Arthur Shapiro for turning her on to Andean butterflies and PBR, and the National Science Foundation for making the extended South American field trips possible through the aid of a Graduate Research Fellowship. She would also like to thank H. B. Shaffer for use of his molecular lab and Keith Willmott and Rupert Griffiths for their assistance with field work. T. Pyrcz field work was financed by KBN6P04F06711 and DS-MZ-IZ/UJ-2003-2009 grants. T. Pyrcz would like to thank Janusz Wojtusiak (Jagiellonin Univ., Krako´w,

258

Poland), Pierre Boyer (Le Puy Sainte Reparade, France) and Mauro Costa (Caracas, Venezuela). Part of this work was carried out by using the resources of the Computational Biology Service Unit from Cornell Univ., which is partially funded by Microsoft Corporation.

References Adams, M. J. 1985. Speciation of the pronophiline butterflies (Satyridae) of the northern Andes. J. Res. Lepid. Suppl. 1: 33 49. Adams, M. J. 1986. Pronophiline butterflies (Satyridae) of the 3 Andean cordilleras of Colombia. Zool. J. Linn. Soc. 87: 235 320. Adams, M. J. and Bernard, G. I. 1981. Pronophiline butterflies (Satyridae) of the Cordillera De Merida, Venezuela. Zool. J. Linn. Soc. 71: 343 372. Aldous, D. J. 2001. Stochastic models and descriptive statistics for phylogenetic trees, from Yule to today. Stat. Sci. 16: 23 34. Ane, C. et al. 2007. Bayesian estimation of concordance among gene trees. Mol. Biol. Evol. 24: 412 426. Barraclough, T. G. and Vogler, A. P. 2000. Detecting the geographical pattern of speciation from species-level phylogenies. Am. Nat. 155: 419 434. Bird, J. M. and Hodkinson, I. D. 2005. What limits the altitudinal distribution of Craspedolepta species (Sternorrhyncha: Psylloidea) on fireweed? Ecol. Entomol. 30: 510 520. Blomberg, S. P. et al. 2003. Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution 57: 717 745. Braun, M. J. et al. 2005. Avian speciation in the Pantepui: the case of the roraiman antbird (Percnostola [Schistocichla] ‘‘Leucostigma’’ saturata). Condor 107: 327 341. Brown, J. H. et al. 1996. The geographic range: size, shape, boundaries, and internal structure. Annu. Rev. Ecol. Syst. 27: 597 623. Buse, A. et al. 2001. Arthropod distribution on an alpine elevational gradient: the relationship with preferred temperature and cold tolerance. Eur. J. Entomol. 98: 301 309. Chapman, F. M. 1917. The distributin of bird life in Colombia. Bull. Am. Mus. Nat. Hist. 31: 1 169. Cordell, S. et al. 1999. Allocation of nitrogen and carbon in leaves of Metrosideros polymorpha regulates carboxylation capacity and delta C-13 along an altitudinal gradient. Funct. Ecol. 13: 811 818. Degnan, J. H. and Rosenberg, N. A. 2006. Discordance of species trees with their most likely gene trees. PLoS Genet. 5: e68. DeVries, P. J. 1987. The butterflies of Costa Rica and their natural history. Volume 1: Papilionidae, Pieridae, Nymphalidae. Princeton Univ. Press. Doebeli, M. and Dieckmann, U. 2003. Speciation along environmental gradients. Nature 421: 259 264. Doebeli, M. et al. 2005. What we have also learned: adaptive speciation is theoretically plausible. Evolution 59: 691 695. Drummond, A. J. and Rambaut, A. 2007. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evol. Biol. 7: 214. Elias, M. et al. 2009. Out of the Andes: patterns of diversification in clearwing butterflies. Mol. Ecol. 18: 1716 1729. Erelli, M. C. et al. 1998. Altitudinal patterns in host suitability for forest insects. Oecologia 117: 133 142. Fitzpatrick, B. M. and Turrelli, M. 2006. The geography of mammalian speciation: mixed signals from phylogenies and range maps. Evolution 60: 601 615.

Download the Supplementary material as file E6306 from <www.oikos.ekol.lu.se/appendix>.

259

ISSUE

Pen˜a, C. et al. 2006. Higher level phylogeny of Satyrinae butterflies (Lepidoptera: Nymphalidae) based on DNA sequence data. Mol. Phylogenet. Evol. 40: 29 49. Picard, D. et al. 2008. Direction and timing of uplift propagation in the Peruvian Andes deduced from molecular phylogenetics of highland biotaxa. Earth Planetary Sci. Lett. 271: 326 336. Pollard, D. A. et al. 2006. Widespread discordance of gene trees with species tree in Drosophila: evidence for incomplete lineage sorting. PLoS Genet. 10: 173. Posada, D. and Crandall, K. A. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817 818. Pyrcz, T. W. 2004. Pronophiline butterflies of the highlands of Chachapoyas in northern Peru: faunal survey, diversity and distribution patterns (Lepidoptera, Nymphalidae, Satyrinae). Genus 15: 455 622. Pyrcz, T. W. and Wojtusiak, J. 2002. The vertical distribution of pronophiline butterflies (Nymphalidae, Satyrinae) along an elevational transect in Monte Zerpa (Cordillera de Merida, Venezuela) with remarks on their diversity and parapatric distribution. Global Ecol. Biogeogr. 11: 211 221. Pyrcz, T. W. et al. 1999. Contribution to the knowledge of Ecuadorian Pronophilini. Part III. Three new species and five new subspecies of Lymanopoda (Lepidoptera: Nymphalidae: Satyrinae). Genus 10: 497 522. Remsen, J. V. J. 1984. High incidence of ‘‘Leapfrog’’ pattern of geographic variation in Andean birds: implications for the speciation process. Science 224: 171 173. Ronquist, F. and Huelsenbeck, J. P. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572 1574. Schluter, D. 1996. Ecological causes of adaptive radiation. Am. Nat. 148: S40 S64. Schneider, C. J. et al. 1999. A test of alternative models of diversification in tropical rainforests: ecological gradients vs. rainforest refugia. Proc. Nat. Acad. Sci. USA 96: 13869 13873. Schulte II, J. A. et al. 2000. Phylogenetic relationships in the iguanid lizard genus Liolaemus : multiple origins of viviparous reproduction and evidence for recurring Andean vicariance and dispersal. Biol. J. Linn. Soc. 69: 75 102. Sempere, T. et al. 2008. New insights into Andean evolution: an introduction to contributions from the 6th ISAG symposium (Barcelona, 2005). Tectonophysics 459: 1 13. Terborgh, J. 1971. Distribution on environmental gradients: theory and a preliminary interpretation of distributional patterns in the avifauna of the Cordillera Vilcabamba, Peru. Ecology 52: 23 40. Voelker, G. 1999. Dispersal, vicariance, and clocks: historical biogeography and speciation in a cosmopolitan passerine genus (Anthus : Motacillidae). Evolution 53: 1536 1552. Wahlberg, N. and Wheat, C. 2008. Genomic outposts serve the phylogenomic pioneers: designing novel nuclear markers for genomic DNA extractions of Lepidoptera. Syst. Biol. 57: 231 242. Wahlberg, N. et al. 2009. Nymphalid butterflies diversify following near demise at the cretaceous/terriary boundary. Proc. R. Soc. B 276: 4295 4302. Wilgenbusch, J. C. et al. 2004. AWTY: a system for graphical exploration of MCMC convergence in Bayesian phylogenetic inference. <http://king2.csit.fsu.edu/CEBProjects/ awty/awty-start.php>. Yule, G. U. 1924. A mathematical theory of evolution, based on the conclusions of Dr. J. C. Willis. Phil. Trans. R. Soc. B 213: 21 87.

IBS SPECIAL

Fjeldsa, J. and Lovett, J. C. 1997. Geographical patterns of old and young species in African forest biota: the significance of specific montane areas as evolutionary centres. Biodivers. Conserv. 6: 325 346. Garzione, C. N. et al. 2006. Rapid late Miocene rise of the Bolivian Altiplano: evidence for removal of mantle lithosphere. Earth Planetary Sci. Lett. 241: 543 556. Gregory-Wodzicki, K. M. 2000. Uplift history of the central and northern Andes: a review. Geol. Soc. Am. Bull. 112: 1091 1105. Grubb, P. J. 1977. Control of forest growth and distribution on wet tropical mountains: with special reference to mineral nutrition. Annu. Rev. Ecol. Syst. 8: 83 107. Heller, H. C. 1971. Altitudinal zonation of chipmunks (Eutamias): interspecific aggression. Ecology 52: 312 319. Heller, H. C. and Poulson, T. 1972. Altitudinal zonation of chipmunks (Eutamias): adaptations to aridity and high temperature. Am. Midl. Nat. 87: 296 313. Holt, R. D. and Keitt, T. H. 2005. Species’ borders: a unifying theme in ecology. Oikos 108: 3 6. Huelsenbeck, J. P. and Ronquist, F. 2001. MrBayes: Bayesian inference of phylogeny. Bioinformatics 17: 754 755. Janzen, D. H. 1967. Why mountain passes are higher in the tropics. Am. Nat. 101: 233 249. Kembel, S. W. et al. 2009. R tools for integrating phylgenies and ecology. R package ver. 1.0-0, <http:/picante.r-forge. r-project.org>. Kirkpatrick, M. and Barton, N. H. 1997. Evolution of a species’ range. Am. Nat. 150: 1 23. Kubatko, L. and Degnan, J. 2007. Inconsistency of phylogenetic estimates from concatenated data under coalescence. Syst. Biol. 56: 17 24. Lamas, G. (ed.) 2004. Checklist: part 4A. Hesperioidea Papilionoidea. Atlas of Neotropical Lepidoptera. Association for Tropical Lepidoptera/Scientific Publ. MacArthur, R. 1972. Geographical ecology: patterns in the distributions of species. Harper and Rowe. Maddison, D. R. and Maddison, W. P. 2000. MacClade version 4: analysis of phylogeny and character evolution. <www. mesquiteproject.org>. McCain, C. 2009. Vertebrate range sizes indicate that mountains may be ‘higher’ in the tropics. Ecol. Lett. 12: 550 560. Miller, M. et al. 2009. The CIPRES portals. CIPRES, <www.phylo.org/sub_sections/portal>. Morecroft, M. D. and Woodward, F. I. 1996. Experiments on the cause of altitudinal differences in the leaf nutrient contents, size and delta C-13 of Alchemill alpina. New Phytol. 132: 471 479. Moritz, C. et al. 2000. Diversification of rainforest faunas: an integrated molecular approach. Annu. Rev. Ecol. Syst. 31: 533 563. Nel, A. et al. 1993. Un nouveau Lepidoptere Satyrinae fossile de l’Oligocene ud sud-est de la France (Insecta, Lepidoptera, Nymphalidae). Linn. Belg. 14: 20 36. Nylander, J. A. A. et al. 2007. AWTY (are we there yet?): a system for graphical exploration of MCMC convergence in Bayesian phylogenetics. Bioinformatics 24: 581 583. Paradis, E. et al. 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20: 289 290. Pen˜a, C. and Wahlberg, N. 2008. Prehistorical climate change increased diversification of a group of butterflies. Biol. Lett. 4: 274 278.

Ecography 33: 260 271, 2010 doi: 10.1111/j.1600-0587.2010.06287.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: David Nogue´s-Bravo. Accepted 20 January 2010

Northern glacial refugia for the pygmy shrew Sorex minutus in Europe revealed by phylogeographic analyses and species distribution modelling Rodrigo Vega, Camilla Fløjgaard, Andre´s Lira-Noriega, Yoshinori Nakazawa, Jens-Christian Svenning and Jeremy B. Searle R. Vega (rrv9@cornell.edu) and J. B. Searle, Dept of Biology, Univ. of York, PO Box 373, York YO10 5YW, UK. (Present address of R. V.: Dept of Entomology, Comstock Hall 5123, Cornell Univ., Ithaca, NY 14853, USA.) C. Fløjgaard, Ecoinformatics and Biodiversity Group, Dept of Biological Sciences, Aarhus Univ., Ny Munkegade 114, DK-8000 Aarhus C., Denmark and Dept of Wildlife Ecology and Biodiversity, National Environmental Research Inst., Aarhus Univ., Grenaavej 14, DK-8410 Rønde, Denmark. A. Lira-Noriega and Y. Nakazawa, Natural History Museum and Biodiversity Research Center, The Univ. of Kansas, Lawrence, KS 66045, USA. J.-C. Svenning, Ecoinformatics and Biodiversity Group, Dept of Biological Sciences, Aarhus Univ., Ny Munkegade 114, DK-8000 Aarhus C., Denmark.

IBS SPECIAL

ISSUE

The southern European peninsulas (Iberian, Italian and Balkan) are traditionally recognized as glacial refugia from where many species colonized central and northern Europe after the Last Glacial Maximum (LGM). However, evidence that some species had more northerly refugia is accumulating from phylogeographic, palaeontological and palynological studies, and more recently from species distribution modelling (SDM), but further studies are needed to test the idea of northern refugia in Europe. Here, we take a rarely implemented multidisciplinary approach to assess if the pygmy shrew Sorex minutus, a widespread Eurasian mammal species, had northern refugia during the LGM, and if these influenced its postglacial geographic distribution. First, we evaluated the phylogeographic and population expansion patterns using mtDNA sequence data from 123 pygmy shrews. Then, we used SDM to predict present and past (LGM) potential distributions using two different training data sets, two different algorithms (Maxent and GARP) and climate reconstructions for the LGM with two different general circulation models. An LGM distribution in the southern peninsulas was predicted by the SDM approaches, in line with the occurrence of lineages of S. minutus in these areas. The phylogeographic analyses also indicated a widespread and strictly northern-central European lineage, not derived from southern peninsulas, and with a postglacial population expansion signature. This was consistent with the SDM predictions of suitable LGM conditions for S. minutus occurring across central and eastern Europe, from unglaciated parts of the British Isles to much of the eastern European Plain. Hence, S. minutus likely persisted in parts of central and eastern Europe during the LGM, from where it colonized other northern areas during the late-glacial and postglacial periods. Our results provide new insights into the glacial and postglacial colonization history of the European mammal fauna, notably supporting glacial refugia further north than traditionally recognized.

During the Quaternary ice ages substantial areas of northern Europe were covered by ice sheets while permafrost existed in large areas of central Europe, which restricted the distribution of many temperate and warmadapted species to the three southern European peninsulas of Iberia, Italy and the Balkans at the Last Glacial Maximum (LGM; Hewitt 2000). These species are interpreted to have recolonized central and northern Europe from these traditionally recognized southern glacial refugia in response to the late-glacial and postglacial warming (Taberlet et al. 1998, Hewitt 2000). Therefore, southern glacial refugia and the northward postglacial recolonization of central and northern Europe from these areas has become an established biogeographical paradigm (Hewitt 2000).

260

Other studies have, however, provided palaeontological, palynological and phylogeographic evidence that glacial refugia for some temperate and boreal species existed further north than the traditionally recognized southern European refugia, implying a more complex pattern of glacial survival and postglacial recolonization: fossils of temperate mammal species dated to the LGM (albeit rarely small mammals) have been described for a number of sites in central Europe, sometimes in co-occurrence with cold-adapted Pleistocene faunal elements (Sommer and Nadachowski 2006). Macrofossil charcoal (organic plant material]2 mm in diameter) of coniferous and broadleaved trees dating to the Upper Palaeolithic has been found in several sites in Austria (42 23 Kya), Czech Republic (29 24.5 Kya), Croatia (27.8 10.8 Kya) and Hungary

makes it difficult to determine the importance of these regions for the LGM distribution of the species, its postglacial colonization history and its present-day genetic structure. Moreover, the inference of glacial refugia based solely on phylogeographic analyses can be obscured by the extinction of genetic variants, incomplete sampling and large-scale range shifts of the species (Waltari et al. 2007). Hence at this point, although the previous phylogeographic studies suggested the existence of northern glacial refugia for S. minutus, the size and geographic spread of these refugia as well as their role in the postglacial range dynamics of the species remain unclear. The purpose of this study is to assess the distribution of S. minutus during the LGM based on a multidisciplinary approach using more detailed mtDNA-based phylogeographic analyses than conducted hitherto and including SDM-based hindcasting. Only a few studies have tried to estimate potential northern refugial areas in this way, despite the stronger inference allowed by these independent and highly complementary approaches (Waltari et al. 2007). We assessed the following specific study questions: would a more detailed phylogeographic analysis also detect a distinctive ‘‘northern-central European and Siberian’’ lineage as has been previously found? Would this widespread lineage present a genetic signature of population expansion? Would different SDM-based hindcasting approaches predict suitable LGM conditions for S. minutus not only in the southern European peninsulas, but also further north, consistent with northern refugia? Would the combined phylogeographic and SDM approach allow us to estimate more precisely the geographic locations of northern refugia for S. minutus, as well as determine their potential role for its postglacial range dynamics? From the population expansion characteristics, how did the refugial populations colonize their current ranges? Finally, are the rather scant fossil data for S. minutus consistent with our phylogeographic and distributional findings? This study sheds light on the spatial variation of the genetic diversity within the widespread distribution of S. minutus, its postglacial population expansion and colonization of Europe from northern refugia, and contributes towards an emerging new synthesis of the fullglacial distributions of the European biota. The nature of northern refugia also has important implications for the understanding of their biogeographic roles as sources of genetic diversity, areas of speciation, identification of conservation units and preservation of species, particularly in response to future climate change (Kotlı´k et al. 2006, Provan and Bennett 2008).

Materials and methods

IBS SPECIAL ISSUE

(31.5 16.5 Kya), suggesting that these regions were also refugial areas for temperate deciduous species (Willis and van Andel 2004, Magri et al. 2006). Palynological records have shown European beech Fagus sylvatica pollen in several sites in central Europe between the late glacial and postglacial (15 10 Kya), and have shown that none of the three traditional refugial areas was the source for northerncentral European beech populations (Magri et al. 2006). Phylogeographic studies on several small mammals have shown little similarity between Mediterranean and northern populations, and have described genetic clades linking together haplotypes sampled throughout northern-central Europe (Bilton et al. 1998, Kotlı´k et al. 2006). Furthermore, species distribution modelling (SDM) has shown that suitable climatic conditions existed for temperate and boreal species in northern latitudes supporting more northerly refugial areas in Europe (Svenning et al. 2008, Fløjgaard et al. 2009). However, a more comprehensive understanding of the relative importance of southern versus northern refugia in terms of LGM species’ ranges as well as for postglacial recolonization is needed. Here, we use the pygmy shrew Sorex minutus (Mammalia, Soricomorpha), as a model for studying the persistence of populations in northern European refugia during the LGM. Sorex minutus is widely distributed in the Palaearctic, throughout Europe to Lake Baikal (Siberia), including the three southern European peninsulas (Hutterer et al. 2008). The species occurs at low density in a wide range of terrestrial habitats with adequate ground cover (Churchfield and Searle 2008). In southern Europe the distribution becomes patchy and limited to higher altitudes where it occurs with some degree of geographical isolation and differentiation, while in central and northern parts of Europe and in Siberia it is more abundant and populations are more connected and widespread. Previous phylogeographic studies on S. minutus revealed a very widespread and genetically homogeneous ‘‘northerncentral European and Siberian’’ lineage, extending from Britain through central and northern Europe to Siberia (ca 7000 km), but genetically distinct from the southern lineages in Iberia, Italy and the Balkans (Bilton et al. 1998, Mascheretti et al. 2003, McDevitt et al. 2010). These studies suggested that the northern-central European lineage persisted and expanded from one or more central or eastern European refugia located further north than the traditionally recognized southern European refugia. However, the size and locations of the possible northern refugia for S. minutus could not be assessed precisely. Species distribution models combine information about species occurrences with environmental (usually climatic) data found across the study region to estimate the presentday geographical distribution of suitable environmental conditions for the species (Guisan and Zimmermann 2000). Then, the set of environmental conditions can be projected to past conditions to identify areas where there were suitable environmental conditions for the species (hindcasting) (Nogue´s-Bravo 2009), in this case at the LGM. Such SDM-based hindcasting has not been integrated into the previous phylogeographic studies on S. minutus, and the genetic data for central and eastern regions of Europe and in Siberia have been rather incomplete. This

Phylogeographic analyses Samples and laboratory procedures

In total, 123 individuals of S. minutus from Europe and Siberia were used for the phylogeographic analysis of the mitochondrial cytochrome b (cyt b) gene. Sixty-six S. minutus cyt b sequences were obtained from Genbank

261

(AB175132: Ohdachi et al. 2006; AJ535393 AJ535457: Mascheretti et al. 2003). Fifty-seven out of the 123 samples of S. minutus were obtained from northern-central Europe during fieldwork and from museum collections (see Acknowledgements) to increase the molecular data and to provide a more detailed analysis of this region. A sequence of S. volnuchini was used as outgroup (AJ535458: Mascheretti et al. 2003). Genomic DNA was extracted using a commercial kit (Qiagen). Partial cyt b sequences were obtained by PCR using two primer pairs that amplified ca 700 bp of overlapping fragments. PCR amplification was performed in a 50 ml final volume: 1X Buffer, 1 mM each primer, 1 mM dNTP’s, 3 mM MgCl2 and 0.5 U Platinum Taq Polymerase (Invitrogen), with cycling conditions: 948C for 4 min, 40 cycles at 948C for 30 s, 558C for 30 s and 728C for 45 s, and a final elongation step at 728C for 7 min. Purification of PCR products was done with a commercial kit (Qiagen) and sequenced (Macrogen and Cornell Univ. Core Laboratories Center).

IBS SPECIAL

ISSUE

Sequence and phylogenetic analyses

Cyt b sequences were edited in BioEdit 7.0 (Hall 1999) and aligned by eye. For the construction of phylogenetic trees, the model of evolution that best fitted the molecular data was searched using MrModeltest 2.3 (Nylander 2004) using the minimum Akaike information criteria value. The substitution model supported was the General Time Reversible with specified substitution types (A C 0.3663, A G 17.4110, A T 1.0216, C G 2.1621, C T 13.0604, G T 1.0), proportion of invariable sites (0.5332), gamma shape parameter (0.9799) and nucleotide frequencies (A 0.2750, C 0.2996, G 0.1382, T 0.2872). The phylogenetic relationships within S. minutus were inferred by Neighbour-Joining (NJ), Maximum Likelihood (ML) and Bayesian analysis using PAUP* 4.0b10 (Swofford 2000), PhyML 3.0 (Guindon and Gascuel 2003) and MrBayes 3.1 (Huelsenbeck and Ronquist 2001), respectively. Confidence for the phylogenetic relationships in NJ and ML was assessed by bootstrap replicates (10 000 and 500 replicates, respectively). For the Bayesian analysis, two independent runs were performed with 10 million generations and 5 chains each, a sampling frequency every 1000 generations, a temperature of 0.1 for the heated chain and checking for convergence. Trees were summarized after a burn-in value of 2500 to obtain the posterior probabilities of each phylogenetic branch. Phylogenetic networks provide an explicit graphic representation of evolutionary history between sequences in which taxa are represented as nodes and their evolutionary relationships are represented by edges. Most internal nodes represent ancestral states from which more recent and peripheral nodes derive (Avise 2000). A parsimony phylogenetic network of cyt b haplotypes was constructed using the software Network 4.5 (FluxusEngineering) with a median-joining algorithm and a greedy FHP genetic distance calculation method. The median joining algorithm identifies groups of haplotypes and introduces hypothetical (non-observed) haplotypes to construct the parsimony network. 262

Genetic and statistical analyses

Standard sequence polymorphism indices (number of haplotypes, polymorphic sites and parsimony informative sites) and genetic diversity values (p, nucleotide diversity9 SD; h, haplotype diversity) were estimated using Arlequin 3.11 (Excoffier et al. 2005). Population expansion was examined for both the full dataset (Eurasia) and for the ‘‘northern-central European and Siberian’’ lineage using DnaSP 5.0 (Librado and Rozas 2009). In each case a mismatch distribution (distribution of the number of differences between pairs of haplotypes) was estimated to compare the demography of the populations with the expectations of a sudden population expansion model (Rogers and Harpending 1992). The raggedness index (rg), which measures the smoothness of the observed distribution, was computed and the statistical validity of the estimated expansion model was tested by a parametric bootstrap approach as a sum of square deviations (SSD) between the observed and the expected mismatch (Schneider and Excoffier 1999) using Arlequin (10 000 replicates). Three other tests for population expansion were performed in DnaSP using coalescent simulations to test for statistical significance (10 000 replicates): R2 test of neutrality, based on the difference of the number of singleton mutations and the average number of nucleotide differences (Ramos-Onsins and Rozas 2002); Fu’s Fs, a statistic based on the infinite-site model without recombination that shows large negative Fs values when there has been a demographic population expansion (Fu 1997); Tajima’s D, a test for selective neutrality based on the infinite-site model without recombination where significant values appear from selective effects but also from factors such as population expansion, bottleneck or heterogeneous mutation rates (Tajima 1989). Species distribution modelling Important discrepancies in the prediction of the potential distribution of a particular species arise from differences in data sample size (Stockwell and Peterson 2002, Wisz et al. 2008), environmental and/or climatic data (Peterson and Nakazawa 2008), and algorithms (Peterson et al. 2007, but see Phillips 2008). Also, if the occurrence records used to model the distribution do not adequately sample the environmental requirements of the species, the prediction will not truly reflect its potential geographic distribution (Pearson et al. 2007). Therefore, to ensure the robustness of our findings, we modelled the potential distribution of S. minutus in the present and at the LGM using two independent training data sets, two algorithms, namely the maximum entropy algorithm (Maxent; Phillips et al. 2006) and the Genetic Algorithm for Rule-set Prediction (GARP; Stockwell and Noble 1992, Stockwell 1999), and using climate reconstructions for the LGM based on two general circulation models (GCMs). All GIS operations were performed using ArcGIS 9.3 (ESRI, Redlands, CA, USA). Species occurrence data

For the first data set, hereafter termed ‘‘data set 1’’, we used the species records from fieldwork, from two online sources (Global Biodiversity Information Facility, GBIF,

For the present-day SDM we initially considered the 19 bioclimatic variables from the WorldClim dataset at a spatial resolution of 2.5 minutes <www.worldclim.org/>. These climate layers are based on spatially interpolated values of temperature and precipitation gathered from weather stations around the world from 1950 2000 (Hijmans et al. 2005). For the LGM (21 Kya) we used

Modelling algorithms

To assess the variation in the outcome of model predictions due to differences in modelling algorithms, we used Maxent and GARP. Maxent has been shown to perform very well in comparative studies of species distribution modelling compared to GARP (Elith et al. 2006, Phillips and Dudı´k 2008, Elith and Graham 2009, but see also Peterson et al. 2008), while GARP has been shown to perform better than Maxent in transferability studies (Peterson et al. 2007, but see also Phillips 2008). Ultimately, the performance of each algorithm may be properly compared using the corresponding thresholding during model evaluation, since their predictions are not given in the same scale (Peterson et al. 2008). To evaluate the accuracy of our models, the empirical AUC values were compared against the AUC values of 1000 random models, as implemented in Peterson et al. (2008), using the data from the test region. AUC ROC values are expressed as the ratio of the area under the observed curve (i.e. the overall area for which each algorithm predicts as present) to the area under the line that defines a random expectation; consequently, the AUC values are expected to 263

ISSUE

Climate data

the climate reconstructions of the same 19 bioclimatic variables based on the CCSM3 (Collins et al. 2006) and MIROC3.2 (Hasumi and Emori 2004) GCMs <http:// biogeo.berkeley.edu> at a spatial resolution of 2.5 minutes. We used the Jackknife procedure implemented in Maxent with the 19 bioclimatic variables on the two data sets to find the best set of predictor variables. We assessed the performance of the models based on the Area Under the Curve (AUC) values of the Receiver Operating Characteristic (ROC) in the independent test region of Siberia. The worst predictor of the whole set of variables was eliminated, a new model was produced using the remaining variables and the process was repeated until all variables were exhausted. We chose the final set of predictors based on parsimony (i.e. with the fewest number of climatic variables) and with the highest AUC value in the independent test region of Siberia. The final set of predictors comprised the variables Annual Mean Temperature (AMT) and Precipitation of the Warmest Quarter (PWQ); thus, AMT and PWQ were used for estimating the present and LGM distribution of S. minutus. These two variables were not highly correlated (r 0.3550) and models that included only these yielded higher or almost equal AUC values than models that included only one or more variables in combination with AMT and PWQ. In addition, these variables are biologically meaningful for S. minutus considering its broad distribution in northern-central Europe and Siberia and habitat preference for damp and temperate areas (Churchfield and Searle 2008, Hutterer et al. 2008). The modelling was performed with data sets 1 and 2 as inputs in Maxent and GARP, and all models were evaluated on the geographically independent (extrinsic) test data from Siberia. For data set 1 we made models with all 25 subsets. Finally, all models were projected onto the two LGM climate reconstructions to identify the potential distribution of S. minutus.

IBS SPECIAL

and Mammal Networked Information System, MaNIS) and from museum specimens obtained for our study (see Acknowledgements). Most of the data were derived from the following sources: the Atlas of Mammals in Britain (Arnold 1993), the European Environment Agency, the UK National Biodiversity Network, the Highland Biological Recording Group HBRG Mammals data set and the Ministerio de Medio Ambiente y Medio Rural y Marino (Spain). Low precision occurrences, such as presence data taken from the centroids of atlas grids and falsely georeferenced occurrences (i.e. offshore and out-of-range locations), were eliminated from this data set. In total, we collected 536 high-precision unique latitude-longitude localities, but this data set was geographically biased towards western Europe and Britain due to differences in sampling effort across the species’ distribution range (i.e. there are few species records from Siberia and southern Europe). In order to correct for sampling bias, we created 25 random subsets from the original data set to limit the number of unique occurrences to 55 in squares of 5 5 degrees distributed across the extent of the geographical analysis (Wisz et al. 2008). This procedure yielded a total of 146 unique localities for each subset which were more evenly distributed. For the second data set, hereafter termed ‘‘data set 2’’, we used the records from the Atlas of European Mammals (AEM; Mitchell-Jones et al. 1999) which present less geographic bias within Europe, but had a much coarser resolution than data set 1. The AEM uses an approximate equal area grid of 50 50 km based on the Universal Transverse Mercator (UTM) projection and the Military Grid Reference System (MGRS). Records of ‘‘species presence’’ as well as ‘‘presence assumed’’ (i.e. presence was observed before 1970 and no evidence of later extinction) were included in the study and a total of 1178 data points were used. To ensure transferability of our models, we used a geographically independent test data set. We digitized the Eurasian range map for S. minutus (Hutterer et al. 2008) and recorded the species as present in all 50 50 km MGRS grid cells within the outline of the range map. Then, we used the part of the range located east of the European study area (for simplicity referred to hereafter as Siberia) only as a test data set (n 3122 data points). This allowed us to evaluate the performance of the models with both data sets and assess which climatic variables provided the strongest predictive ability in a geographically independent region with relatively LGM-like conditions (Fløjgaard et al. 2009). We used the digitised range map data only for testing, given its much coarser resolution and uncertain quality compared to the occurrence data from data sets 1 and 2.

ISSUE

IBS SPECIAL

be larger than one as the model departs from the random expectation (Peterson et al. 2008). Maxent is a machine-learning technique based on the principle of maximum entropy that fits a probability distribution to the environmental conditions at the locations where a species has been observed (Phillips et al. 2004, 2006). When implemented with ecologically meaningful sets of predictor variables, Maxent produced similar estimates for the locations of glacial refugia as Bioclim, another commonly used, but simpler, modelling technique (Svenning et al. 2008, Fløjgaard et al. 2009). We used the default settings in Maxent 3.2.1 <www.cs.princeton. edu/ schapire/maxent/> with background data limited to Eurasia as described in the species occurrence data section. We converted the continuous logistic output from Maxent into a binary map of predicted suitable environmental conditions for S. minutus using the maximum test sensitivity and specificity threshold because it optimized the correct discrimination of presences and pseudoabsences in the test data. GARP is a genetic algorithm that produces a set of rules that describe the non-random association between environmental variables and occurrence data (Stockwell and Noble 1992, Stockwell 1999). First, the algorithm creates a set of rules based on four basic types (bioclimatic, atomic, negated and logistic regression rules), their individual predictive accuracy is calculated and only those rules with the highest predictive accuracy are retained in the model. The overall performance of the model is evaluated using a subset of presence points. Then, a second generation of rules is produced via the random modification of the previous generation rules, their predictive accuracy is calculated and only those with the highest accuracy are included in the model. Finally, the overall performance of the model is re-evaluated and the process of creation, evaluation and inclusion of rules is repeated until a maximum number of iterations is reached (1000 in this case), or until performance values no longer change appreciably from one iteration to the next (convergence parameter of 1%). We used the version of DesktopGarp as implemented in openModeller ver. 1.0.9 <http://openmodeller.sf.net> using the default parameters (Anderson et al. 2003). We converted the continuous output into a binary map of predicted occurrence of the suitable conditions for S. minutus by assigning a value of 1 for the model values that corresponded to 10% or more of the testing points.

Results Phylogeographic analysis Sequence analysis and phylogenetic reconstructions

A partial sequence of 1110 bp from the S. minutus cyt b was analysed. One hundred and twelve haplotypes were obtained, from which 46 were newly described and deposited in GenBank (accession numbers: GQ494305 GQ494350). There were 894 invariable and 216 variable positions, from which 137 were parsimony informative. All the phylogenetic analyses revealed five distinct lineages (Fig. 1). Samples from the Mediterranean peninsulas clustered in three lineages, namely the ‘‘Iberian’’, ‘‘Italian’’ and ‘‘Balkan’’ groups, corresponding to their 264

geographical origin. There was also a well supported ‘‘Pyrenean’’ lineage with samples from Andorra and Ireland. Samples from northern-central Europe and Siberia clustered together forming a geographically widespread lineage that did not include any individuals from the southern peninsulas, hereafter named as the ‘‘northern-central European’’ lineage. This lineage was composed of 105 sequences (94 haplotypes) with 940 invariable and 170 variable positions, from which 92 were parsimony informative. The phylogenetic network of the northern-central European lineage presented a star-like pattern with three most central haplotypes, named A, B and C, separated by only one mutational step from each other and from which all other sequences derived (Fig. 2). The other phylogroups from the southern peninsulas were much more distantly related and separated by several mutations (data not shown). The central haplotypes A and B were entirely composed of samples from the Netherlands (three and two individuals, respectively), while the third central haplotype (C) belonged to a central Ukrainian specimen from the locality Tishki (5086.27?N, 3386.39?E). There was an apparent geographical subdivision of the samples that were connected to these three central haplotypes (Fig. 2). Only haplotypes from Great Britain and the Netherlands were directly connected to A. Several haplotypes from different countries of northern and central Europe were connected to B, also including some haplotypes from Great Britain and the Netherlands, but there were no haplotypes from eastern Europe or Siberia (except for one sample from Ukraine ambiguously connected to B and C). Haplotypes from northern, central and eastern Europe and Siberia were all directly connected to C, but there were no samples from countries further west than Germany. However, the support for these subdivisions was not strong: equally parsimonious explanations (loops) appeared in the central part of the network between B and C, and there was no supported sub-structure within the northern-central European lineage in the phylogenetic trees. Genetic and statistical analyses

The whole Eurasian sample presented a nucleotide diversity p 0.010990.0055, and a haplotype diversity h 0.9983. The northern-central European lineage had a nucleotide diversity p 0.006790.0035, and a haplotype diversity h 0.9980. Genetic diversity values were not calculated for the southern European lineages because of small sample size. The mismatch distribution of the whole dataset (Eurasia) was bimodal, consistent with pairwise differences between sequences belonging to the same and different lineages (Fig. 3a). The mismatch distribution of the northern-central European lineage showed a unimodal distribution that, visually, fitted almost perfectly over the expected values for a population expansion model (Fig. 3b). There was an observed mean of 7.382 pairwise differences with a variance of 8.152. The goodness of fit test showed no significant differences between the observed and expected values under a sudden expansion model for the northerncentral European lineage (SSD 0.0004, pSSD 0.05; rg 0.0082, p 0.05). Negative and significant Tajima’s D (D 2.5721, p B0.001) and Fu’s Fs (Fs 24.8437,

sudden population expansion (R2 0.0180, p B0.001). The rest of the sequences and lineages that belonged to the more distantly related southern European lineages (Iberian, Italian and Balkan peninsulas) and the Pyrenees were not analysed because of small sample size. Species distribution modelling Predicted present distribution

Species distribution models from Maxent matched the reported distribution of the species (Fig. 4a, c). The models also predicted suitable climatic conditions outside the reported distribution of the species especially in two regions, the Asia Minor-Caucasus region and in the Far East (Fig. 4a, c). The predicted present distribution of S. minutus with GARP was very similar to that of Maxent, it also matched the reported distribution and the predicted suitable climatic conditions in the Asia Minor-Caucasus region and in the Far East (Fig. 4b, d). All Maxent and GARP models were accurate in the test region, with AUC values for both data sets higher than null expectations (p B0.001; mean AUCMAXENT 1.249 0.021 and mean AUCGARP 1.04990.007 for data set 1, and mean AUCMAXENT 1.24990.011 and mean AUCGARP 1.03290.005 for data set 2). Predicted LGM distribution

pB0.001) showed departures from neutrality also consistent with a sudden population expansion. Moreover, the R2 test of neutrality also showed that the northern-central European lineage gave a genetic signature consistent with a

265

ISSUE

Figure 1. Continued.

Figure 1. Bayesian inference tree showing the phylogenetic relationships among Sorex minutus samples (S. volnuchini, outgroup). Five lineages were found (I Pyrenean-Irish, D Italian, j Iberian, ' Balkan, and k northern-central European). The northern-central European lineage is geographically widespread but has not been found within the southern European peninsulas. Values on branches correspond to Bayesian posterior probabilities. Haplotypes are represented with two-letter country codes followed by an identification number (x2, haplotype frequency 2 etc.): AD Andorra, AT Austria, BY Belarus, CH Switzerland, CZ Czech Republic, DE Germany, DK Denmark, ES Spain, FI Finland, FR France, GB Great Britain, IE Ireland, IT Italy, LT Lithuania, MK Macedonia, NL the Netherlands, PL Poland, RU Russia, SE Sweden, SK Slovakia, TR Turkey, UA Ukraine.

IBS SPECIAL

With the two data sets and GCMs, Maxent and GARP predicted suitable LGM climatic conditions in the southern European peninsulas (Fig. 4e l), concordant with southern refugia. In general, suitable LGM conditions with the two data sets, GCMs and algorithms were also predicted north of the southern refugia, particularly throughout central Europe, most of eastern Europe, southern Poland, eastern and southern Ukraine, the Crimea peninsula and the Caucasus. With Maxent, the LGM predictions differed little between data sets or between GCMs, and there were predicted suitable conditions in central and eastern Europe close to the ice sheet (Fig. 4e, g, i, k). With GARP, predictions differed between GCMs: more restricted suitable conditions in central and eastern Europe were predicted with CCSM3 (Fig. 4f, h) than with MIROC3.2 (Fig. 4j, l), but predictions did not differ much between data sets. The most restricted predictions (using GARP with CCSM3) still showed suitable climatic conditions in southern Ireland, central and southern France, western parts

Figure 2. Parsimony median joining haplotype network for the northern-central European lineage of Sorex minutus. Observed haplotypes are shown as grey circles (proportional to frequency) and hypothetical haplotypes are shown as black circles. There is a star-like phylogeny with three central (ancestral) haplotypes. A and B are two central haplotypes from the Netherlands, and C is from central Ukraine. The dotted black line encircles haplotypes directly linked to A, black lines encircle haplotypes directly linked to B and the dashed line encircles haplotypes directly linked to C (the country of origin for haplotypes is shown next to clusters; twoletters country codes as in Fig. 1). For simplicity, haplotypes from the more diverged southern European lineages are not shown, but relate to central-European haplotypes by the addition of several hypothetical haplotypes and 10 mutational steps. The scale bar represents one mutational step.

of Switzerland, a few regions north of the Balkans, the Crimea peninsula and the Caucasus.

Discussion

IBS SPECIAL

ISSUE

Northern glacial refugia revealed by a combined approach Sorex minutus is considered a temperate species, but it is also latitudinally distributed above 608N (i.e. near the Arctic Circle) and altitudinally above 2000 m in regions with permafrost and harsh winters (Mitchell-Jones et al. 1999, Hutterer et al. 2008). Northern non-arctic species like S. minutus could have persisted in high latitude refugia in Europe during the LGM, north of the traditionally recognized Mediterranean refugial areas (Stewart and Lister 2001). This could have been a result of their ecological traits (notably cold tolerance) and biogeographical characteristics that may have determined their response to the glaciations (Bhagwat and Willis 2008). Sorex minutus is, therefore, a suitable model organism for exploring the controversial hypothesis of ‘‘northern’’ glacial refugia. The general concordance of the phylogeographic analyses with the predicted LGM distributions based on species distribution modelling and the concordance between models suggest that we have obtained robust results concerning the LGM distribution of S. minutus. Our phylogeographic analyses provided evidence for a distinct lineage in northern-central Europe, with additional lineages in the Iberian, Italian and Balkan peninsulas in southern Europe. First, the absence of southern haplotypes in northern-central Europe supports the hypothesis that the 266

Figure 3. Mismatch distribution for observed (continuous line) and expected (dashed line) pairwise comparisons under a sudden population expansion model among Sorex minutus cyt b sequences. (a) Mismatch distribution among Eurasian sequences with a bimodal observed distribution where the first peak corresponds to pairwise comparisons among closely related individuals within lineages, while the second peak corresponds to pairwise comparisons among distantly related individuals from different lineages. (b) Mismatch distribution among sequences from the northerncentral European lineage showing a unimodal distribution, a genetic signature which corresponds to the expected distribution for sudden population expansion.

southern peninsulas were areas of endemism and differentiation for S. minutus, but not for northward colonization (Bilton et al. 1998), i.e. the current populations in northern-central Europe were not derived from LGM populations in the traditional southern European refugia. Second, the northern-central European lineage showed a strong signature of population expansion supported by the mismatch distribution and population expansion tests. Finally, ancestral haplotypes in a phylogenetic network can be identified by their central or internal position from where the peripheral, more recent, haplotypes are derived, by the number of haplotypes that arise from them and by their abundance (Avise 2000). The phylogenetic network of the northern-central European lineage showed a starlike pattern with three ancestral haplotypes from distant regions in central and eastern Europe (the Netherlands and Ukraine). This pattern was also consistent with a widespread LGM distribution and congruent with the hypothesis of persistence and postglacial expansion from northern glacial refugia.

267

Figure 4. Continued.

IBS SPECIAL

ISSUE

IBS SPECIAL

The phylogeographic pattern that we observe here did not arise from the low sample size in southern Europe: the few samples from southern peninsulas belonged to lineages differentiated by a large number of mutation steps from the northern-central European lineage; if northern-central Europe had been colonized from southern Europe we would have found northern-central European samples clustering within southern lineages, not forming a separate lineage. Moreover, a phylogeographic study on S. minutus using the mitochondrial Control Region and Y-chromosome introns with more samples from southern peninsulas showed a similar pattern (McDevitt et al. 2010). Nevertheless, further sampling in southern regions and the use of other molecular markers is desirable to investigate the genetic variation and population expansion events within Mediterranean peninsulas, and for the determination of contact zones among lineages. We did not use the mismatch distributions to date the population expansion for the northern-central European lineage because of the lack of a suitable mutation rate for cyt b in S. minutus. Previous studies on Sorex have used mismatch distributions for molecular dating (e.g. Ratkiewicz et al. 2002), but with mutation rates that may not be suitable over short time frames (Ho et al. 2005). The modelling approaches predicted successfully the wide present-day distribution of S. minutus in Eurasia. Therefore, we consider our SDM approaches as giving realistic estimates of the area with suitable climatic conditions for our species and of its potential LGM distribution. A third model using Bioclim with SDM data sets 1 and 2 also resulted in very similar present-day and LGM distributions for S. minutus (data not shown). The potential LGM distributions predicted by our SDM approaches not only included the traditionally recognized southern refugia, but also a wide area across central and eastern Europe, from the unglaciated parts of southern Ireland and Britain to most of the central and southeast European (or Russian) Plain. In particular, the predicted LGM distribution throughout central and eastern Europe encompasses suggested northern refugial areas based on palaeontological and palynological data for other temperate and boreal species (Willis et al. 2000, Willis and van Andel 2004, Magri et al. 2006, Sommer and Nadachowski 2006). Thus, the northern-central European lineage could have persisted in various parts of this wide area during the LGM according to the phylogeographic and the SDM approaches. We note that the central and eastern European LGM distribution was similar with both data sets, particularly when using Maxent (with both GCMs) and when using GARP with MIROC3.2, even though we used very different species records. However, the LGM distributions

when using GARP were more widespread to the north with MIROC3.2 than with CCSM3 GCMs, which could represent variations due to modelling algorithms and GCMs. Also, the predicted present-day suitable climatic conditions outside the reported distribution of S. minutus in the Asia Minor-Caucasus region and in the Far East probably reflect competitive or speciation processes rather than an inaccurate estimation of the suitable climatic conditions. In Asia Minor-Caucasus, S. minutus is replaced by the closely related sister species S. volnuchini, while in the Far East many other Sorex species occur including similarsized species such as S. gracillimus. The predicted LGM distribution of S. minutus appears to be continuous throughout Europe; however, lineage diversification is still plausible: First, the present distribution of S. minutus also appears to be continuous but it is affected by landscape features, not evident at the geographic resolution given, which could have subdivided the species range. Therefore, it could be expected that landscape features at the LGM also affected the distribution of S. minutus. Second, the estimation of the extent of ice sheets in mountainous areas is not precise, so it may be expected that the Iberian and Italian populations remained isolated from the rest of Europe by ice sheets covering the Pyrenees and the Alps, respectively, while the heterogeneous landscape in the Balkans could have been responsible for the limited distribution of the genetic lineage there. Also, different genetic variants could have arisen within regions and could have been maintained there selectively reducing further spread into contiguous regions. Another explanation could be that interspecific competition and/or other non-climatic conditions subdivided the potentially continuous LGM distribution. Insights into postglacial colonization The predicted distribution for S. minutus in the Iberian, Italian and Balkan peninsulas presumably corresponds to the refugial areas where the southern genetic lineages persisted during the LGM. The Pyrenean lineage, here represented by a limited number of Andorran and Irish samples, could have persisted during the LGM in central and south-western France and even in unglaciated areas in southern Ireland, as shown by our SDM models. However, genetic studies support a more recent origin of the Irish pygmy shrew, transported there by humans during the Holocene (Mascheretti et al. 2003, McDevitt et al. 2009, A. D. McDevitt, V. R. Rambau, R. Vega and J. B. Searle pers. comm.). Further molecular sampling in southern Europe is desirable to determine the extent of the

Figure 4. Species distribution modelling of Sorex minutus in the present and at the Last Glacial Maximum (LGM) using different approaches. Two independent data sets, two algorithms, Maximum entropy (Maxent) and Genetic Algorithm for Rule-set prediction (GARP), and climate reconstructions for the LGM based on two general circulation models (CCSM3 and MIROC3.2) were used. Climatic variables were obtained from WorldClim and two were selected as best predictors with a Jackknife procedure: annual mean temperature and precipitation of the warmest quarter. (a d) Maxent and GARP modelled present distributions with data sets 1 and 2. (e l) Maxent and GARP modelled LGM distributions with data sets 1 and 2 using CCSM3 and MIROC3.2. The thick lines (a d) represent the outline of present-day distribution range of the species, the dark shading corresponds to present-day and LGM suitable climatic conditions, and the light gray polygon represents the ice extent at the LGM, about 21 Kya (redrawn from Svendsen et al. 2004). Location of samples used for the phylogeographic analysis is shown (lineages as in Fig. 1: I Pyrenean-Irish, D Italian, j Iberian, ' Balkan, and k northern-central European).

268

geographic distribution of the lineages found there and the contact zones between them. Considering the phylogenetic network for the northerncentral European lineage, the three central (ancestral) haplotypes were located in or near regions where the SDM approaches predicted a potential LGM distribution for S. minutus. These results imply that S. minutus was not dependent on amelioration of the climate at the end of the last glaciation to colonize northern-central Europe from southern refugia; instead, it was already present. As the ice sheets retreated and the climate improved, the range of S. minutus expanded from northern refugia colonizing the rest of northern-central Europe. For example, Scandinavian and the Baltic regions were most likely colonized by pygmy shrews from eastern Europe, not from the west or from southern peninsulas. Thus, the phylogenetic network shows that sequences from Norway, Finland and Lithuania group closely with the Ukrainian central haplotype, which according to the SDM modelling could have survived the LGM in situ on the east European Plain. Likewise, the genetic similarity of samples from the Netherlands and Britain, in comparison to those elsewhere, suggests that the British pygmy shrew originated from populations in the vicinity of the Netherlands, reaching Britain over the landbridge with continental Europe. An alternative explanation is that S. minutus persisted in the unglaciated regions of southern Britain (as predicted by several of our SDM approaches) which were geographically connected and genetically similar to populations in continental Europe during the LGM. Whatever the explanation, as ice sheets retreated, S. minutus belonging to the northern-central European lineage was able to colonize the northern parts of mainland Britain. Further support from fossils and phylogeographic analyses

269

ISSUE

Acknowledgements Specimens and species records of Sorex minutus were made available by several museums and we acknowledge the help of the curators from the following institutions: Dept of Ecology and Evolution (Univ. de Lausanne, Switzerland), Univ. na Primorskem and Research Centre of Koper (Slovenia), Natuurhistorisch Museum (Rotterdam, the Netherlands), Dipartimento di Ecologia (Univ. della Calabria, Italy), Museo di Anatomia Comparata and Museo di Zoologia ‘‘La Sapienza’’ (Univ. di Roma, Italy) and Natuurmuseum Brabant (Tilburg, the Netherlands). We are very grateful for the tissue samples provided by Boris Krysˇtufec, Allan McDevitt, Glenn Yannic, Jacques Hausser, Jan Zima, Frı´ða Jo´hannesdo´ttir, Holger Bruns, Peter Borkenhagen and Petr Kotlı´k. We thank David Nogue´s-Bravo and two anonymous referees for their valuable comments. Bayesian analyses were run at the Computational Biology Service Unit from Cornell Univ. which is partially funded by Microsoft Corporation. We gratefully acknowledge financial support to R. Vega (181844) and A. Lira-Noriega (189216) from CONACyT (Me´xico), and to J.-C. Svenning from the Danish Natural Science Research Council (grant 272-07-0242).This work represents the fruits of the discussion of our work presented at the 4th Biennial

IBS SPECIAL

Northern refugia in central Europe and further east, north of the traditional Mediterranean refugia, have been hypothesized in phylogeographic analyses for a number of small mammal species other than the pygmy shrew, including the field vole Microtus agrestis (Jaarola and Searle 2002), bank vole Clethrionomys glareolus (Deffontaine et al. 2005, Kotlı´k et al. 2006), root vole Microtus oeconomus (Brunhoff et al. 2003), common vole Microtus arvalis (Heckel et al. 2005) and the common shrew Sorex araneus (Bilton et al. 1998, Yannic et al. 2008). For bank voles, root voles, field voles and common voles, predictions of their potential LGM distribution based on SDM were also consistent with northern refugia (Fløjgaard et al. 2009). Most of the phylogeographic studies point to the Carpathians as a likely northern refugial area, but a refugium in this area could have included broader regions of Hungary, Slovakia, Czech Republic, Moldova and Poland, supported by the occurrence of temperate mammal fossil records in the area (Sommer and Nadachowski 2006) and by our results. Also, the region of the Dordogne in south-western France was situated outside the LGM permafrost area and has temperate mammal fossil records dated to the end of the LGM. Therefore, it has been suggested as another likely refugium north of

the traditionally recognized southern refugia (Sommer and Nadachowski 2006), further supported by our findings. In addition, there are a few but important fossil records of S. minutus from several localities north of the southern refugia, radiocarbon dated close to the LGM or earlier (S3P Faunal Database <www.esc.cam.ac.uk/research/researchgroups/oistage3/stage-three-project-database-downloads>). These fossil remains have been found in sites in France (26 Kya), Belgium (38 40 Kya), Germany (23 29 Kya) and Hungary (20 22 Kya). In conclusion, a wide northern LGM distribution for S. minutus is supported by the combined use of a phylogeographic and species distribution modelling approach. The SDM methodologies provide evidence for a central and eastern European LGM distribution of S. minutus, where the northern-central European lineage could have been distributed. Additionally, the SDM approaches reveal potential LGM distributions for S. minutus in southern refugia, consistent with the lineages present in those regions. The phylogeographic analyses, however, indicate that the southern refugia were not the postglacial source of the current and widespread northern-central European populations. The other phylogeographic and SDM studies on small mammals, mammal and plant fossil records, and S. minutus fossil remains presented here provide additional evidence consistent with or directly supportive of our findings. Our results contribute to the understanding of persistence and colonization from glacial refugia further north than traditionally recognized. They also provide new insights into the location and importance of refugial areas for the persistence of populations and genetic lineages during climate change. The use of S. minutus as a model exemplifies how the combined use of phylogeography and species distribution modelling can be applied to understand present-day biodiversity patterns, and can predict and test the past distribution of species to gain insight into the colonization patterns, differentiation and biogeography of species.

Conference of the International Biogeography Society (8 12 January 2009, Me´rida, Yucata´n, Me´xico).

IBS SPECIAL

ISSUE

References Anderson, R. P. et al. 2003. Evaluating predictive models of species’ distributions: criteria for selecting optimal models. Ecol. Model. 162: 211 232. Arnold, H. R. 1993. Atlas of mammals of Britain. Her Majesty’s Stationary Office, London. Avise, J. C. 2000. Phylogeography: the history and formation of species. Harvard Univ. Press. Bhagwat, S. A. and Willis, K. J. 2008. Species persistence in northerly glacial refugia of Europe: a matter of chance or biogeographical traits? J. Biogeogr. 35: 464 482. Bilton, D. T. et al. 1998. Mediterranean Europe as an area of endemism for small mammals rather than a source for northwards postglacial colonization. Proc. R. Soc. B 265: 1219 1226. Brunhoff, C. et al. 2003. Holarctic phylogeography of the root vole (Microtus oeconomus): implications for late Quaternary biogeography of high latitudes. Mol. Ecol. 12: 957 968. Churchfield, S. and Searle, J. B. 2008. Pygmy shrew Sorex minutus. In: Harris, S. and Yalden, D. W. (eds), Mammals of the British Isles: handbook. Mammal Society, pp. 267 271. Collins, W. D. et al. 2006. The Community Climate System Model: CCSM3. J. Climate 19: 2122 2143. Deffontaine, V. et al. 2005. Beyond the Mediterranean peninsulas: evidence of central European glacial refugia for a temperate forest mammal species, the bank vole (Clethrionomys glareolus). Mol. Ecol. 14: 1727 1739. Elith, J. and Graham, C. H. 2009. Do they? How do they? WHY do they differ? On finding reasons for differing performances of species distribution models. Ecography 32: 66 77. Elith, J. et al. 2006. Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29: 129 151. Excoffier, L. et al. 2005. Arlequin ver. 3.0: an integrated software package for population genetics data analysis. Evol. Bioinform. Online 1: 47 50. Fløjgaard, C. et al. 2009. Ice age distributions of European small mammals: insights from species distribution modelling. J. Biogeogr. 36: 1152 1163. Fu, Y. X. 1997. Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection. Genetics 147: 915 925. Guindon, S. and Gascuel, O. 2003. A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst. Biol. 52: 696 704. Guisan, A. and Zimmermann, N. E. 2000. Predictive habitat distribution models in ecology. Ecol. Model. 135: 147 186. Hall, T. A. 1999. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/ NT/XP. Nucleic Acids Symp. Ser. 41: 95 98. Hasumi, H. and Emori, S. 2004. K-1 coupled GCM (MIROC) description. Center for Climate System Research, Univ. of Tokyo. Heckel, G. et al. 2005. Genetic structure and colonization processes in European populations of the common vole, Microtus arvalis. Evolution 59: 2231 2242. Hewitt, G. M. 2000. The genetic legacy of the Quaternary ice ages. Nature 405: 907 913. Hijmans, R. J. et al. 2005. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25: 1965 1978.

270

Ho, S. Y. W. et al. 2005. Time dependency on molecular rate estimates and systematic overestimation of recent divergence times. Mol. Biol. Evol. 22: 1561 1568. Huelsenbeck, J. P. and Ronquist, F. 2001. MrBayes: Bayesian inference of phylogeny. Bioinformatics 17: 754 755. Hutterer, R. et al. 2008. Sorex minutus. IUCN 2009, IUCN Red List of Threatened Species, ver. 2009.2 <www.iucnredlist.org>, accessed 12 May 2009. Jaarola, M. and Searle, J. B. 2002. Phylogeography of field voles (Microtus agrestis) in Eurasia inferred from mitochondrial DNA sequences. Mol. Ecol. 11: 2613 2621. Kotlı´k, P. et al. 2006. A northern glacial refugium for bank voles (Clethrionomys glareolus). Proc. Nat. Acad. Sci. USA 103: 14860 14864. Librado, P. and Rozas, J. 2009. DnaSP ver. 5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25: 1451 1452. Magri, D. et al. 2006. A new scenario for the Quaternary history of European beech populations: palaeobotanical evidence and genetic consequences. New Phytol. 171: 199 221. Mascheretti, S. et al. 2003. How did pygmy shrews colonize Ireland? Clues from a phylogenetic analysis of mitochondrial cytochrome b sequences. Proc. R. Soc. B 270: 1593 1599. McDevitt, A. D. et al. 2009. Genetic variation in Irish pygmy shrews Sorex minutus (Soricomorpha: Soricidae): implications for colonization history. Biol. J. Linn. Soc. 97: 918 927. McDevitt, A. D. et al. 2010. Postglacial re-colonization of continental Europe by the pygmy shrew (Sorex minutus) inferred from mitochondrial and Y chromosomal DNA sequences. In: Habel, J. C. and Assman, T. (eds), Relict species phylogeography and conservation biology. Springer, in press. Mitchell-Jones, A. J. et al. 1999. The atlas of European mammals. Poyser Natural History. Nogue´s-Bravo, D. 2009. Predicting the past distribution of species climatic niches. Global Ecol. Biogeogr. 18: 521 531. Nylander, J. A. A. 2004. MrModeltest ver. 2. Program distributed by the author. Evolutionary Biology Centre, Uppsala Univ. Ohdachi, S. D. et al. 2006. Molecular phylogenetics of soricid shrews (Mammalia) based on mitochondrial cytochrome b gene sequences: with special reference to the Soricinae. J. Zool. 270: 177 191. Pearson, R. G. et al. 2007. Predicting species distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. J. Biogeogr. 34: 102 117. Peterson, A. T. and Nakazawa, Y. 2008. Environmental data sets matter in ecological niche modeling: an example with Solenopsis invicta and Solenopsis richterii. Global Ecol. Biogeogr. 17: 135 144. Peterson, A. T. et al. 2007. Transferability and model evaluation in ecological niche modeling: a comparison of GARP and Maxent. Ecography 30: 550 560. Peterson, A. T. et al. 2008. Rethinking receiver operating characteristic analysis applications in ecological niche modeling. Ecol. Model. 213: 63 72. Phillips, S. J. 2008. Transferability, sample selection bias and background data in presence only modelling: a response to Peterson et al. 2007. Ecography 31: 272 278. Phillips, S. J. and Dudı´k, M. 2008. Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecography 31: 161 175. Phillips, S. J. et al. 2004. A maximum entropy approach to species distribution modeling. In: Greiner, R. and Schuurmans, D. (eds), Proc. 21st Int. Conf. on Machine Learning. ACM Press, pp. 655 662.

Phillips, S. J. et al. 2006. Maximum entropy modeling of species geographic distributions. Ecol. Model. 190: 231 259. Provan, J. and Bennett, K. D. 2008. Phylogeographic insights into cryptic glacial refugia. Trends Ecol. Evol. 23: 564 571. Ramos-Onsins, S. E. and Rozas, J. 2002. Statistical properties of new neutrality tests against population growth. Mol. Biol. Evol. 19: 2092 2100. Ratkiewicz, M. et al. 2002. The evolutionary history of the two karyotypic groups of the common shrew, Sorex araneus, in Poland. Heredity 88: 235 242. Rogers, A. R. and Harpending, H. 1992. Population growth makes waves in the distribution of pairwise genetic differences. Mol. Biol. Evol. 9: 552 569. Schneider, S. and Excoffier, L. 1999. Estimation of demographic parameters from the distribution of pairwise differences when the mutation rates vary among sites: application to human mitochondrial DNA. Genetics 152: 1079 1089. Sommer, R. S. and Nadachowski, A. 2006. Glacial refugia of mammals in Europe: evidence from fossil records. Mammal Rev. 36: 251 265. Stewart, J. R. and Lister, A. M. 2001. Cryptic northern refugia and the origins of the modern biota. Trends Ecol. Evol. 16: 608 613. Stockwell, D. R. B. 1999. Genetic algorithms II. In: Fielding, A. H. (ed.), Machine learning methods for ecological applications. Kluwer, pp. 123 144. Stockwell, D. R. B. and Noble, I. R. 1992. Introduction of sets of rules from animal distribution data: a robust and informative method of data analysis. Math. Comput. Simulat. 33: 385 390.

Stockwell, D. R. B. and Peterson, A. T. 2002. Effects of sample size on accuracy of species distribution models. Ecol. Model. 148: 1 13. Svendsen, J. I. et al. 2004. Late Quaternary ice sheet history of northern Eurasia. Quat. Sci. Rev. 23: 1229 1271. Svenning, J.-C. et al. 2008. Glacial refugia of temperate trees in Europe: insights from species distribution modeling. J. Ecol. 96: 1117 1127. Swofford, D. L. 2000. PAUP*. Phylogenetic analysis using parsimony (*and other methods), ver. 4. Sinauer. Taberlet, P. et al. 1998. Comparative phylogeography and postglacial colonization routes in Europe. Mol. Ecol. 7: 453 464. Tajima, F. 1989. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123: 585 595. Waltari, E. et al. 2007. Locating Pleistocene refugia: comparing phylogeogaphic and ecological niche model predictions. PLoS One 7: e563. Willis, K. J. and van Andel, T. H. 2004. Trees or no trees? The environments of central and eastern Europe during the Last Glaciation. Quat. Sci. Rev. 23: 2369 2387. Willis, K. J. et al. 2000. The full-glacial forests of central and south-eastern Europe. Quat. Res. 53: 203 213. Wisz, M. S. et al. 2008. Effects of sample size on the performance of species distribution models. Divers. Distrib. 14: 763 773. Yannic, G. et al. 2008. A new perspective on the evolutionary history of western European Sorex araneus group revealed by paternal and maternal molecular markers. Mol. Phylogenet. Evol. 47: 237 250.

IBS SPECIAL ISSUE

271

Ecography 33: 272 284, 2010 doi: 10.1111/j.1600-0587.2010.06305.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editors: David Nogue´s-Bravo and Carsten Rahbek. Accepted 7 March 2010

Ecological history and latent conservation potential: large and giant tortoises as a model for taxon substitutions Dennis M. Hansen, C. Josh Donlan, Christine J. Griffiths and Karl J. Campbell D. M. Hansen (dmhansen@stanford.edu), Dept of Biology, Stanford Univ., 371 Serra Mall, CA 94305, USA. C. J. Donlan, Advanced Conservation Strategies, P.O. Box 1201, Midway, UT 84049, USA, and Copeland Fellow in Global Sustainability, Amherst College, Amherst, MA 01002, USA, and Dept of Ecology and Evolutionary Biology, Cornell Univ., Ithaca, NY 14853, USA. C. J. Griffiths, School of Biological Sciences, Univ. of Bristol, Woodland Road, Bristol, BS8 1UG, UK, and Inst. of Environmental Sciences, Univ. of Zurich, 190 Winterthurerstrasse, CH-8057 Zurich, Switzerland. K. J. Campbell, Island Conservation, LML, 100 Shaffer Road, Santa Cruz, CA 95060, USA, and School of Integrative Systems, Univ. of Queensland, Gatton, Queensland 4343, Australia.

IBS SPECIAL

ISSUE

Starting in the late 1970s, ecologists began unraveling the role of recently extinct large vertebrates in evolutionary ecology and ecosystem dynamics. Three decades later, practitioners are now considering the role of ecological history in conservation practice, and some have called for restoring missing ecological functions and evolutionary potential using taxon substitutes extant, functionally similar taxa to replace extinct species. This pro-active approach to biodiversity conservation has proved controversial. Yet, rewilding with taxon substitutes, or ecological analogues, is now being integrated into conservation and restoration programmes around the world. Empirical evidence is emerging that illustrates how taxon substitutions can restore missing ecological functions and evolutionary potential. However, a major roadblock to a broader evaluation and application of taxon substitution is the lack of practical guidelines within which they should be conducted. While the International Union for Conservation of Nature’s reintroduction guidelines are an obvious choice, they are unsuitable in their current form. We recommend necessary amendments to these guidelines to explicitly address taxon substitutions. A second impediment to empirical evaluations of rewilding with taxon substitutions is the sheer scale of some proposed projects; the majority involves large mammals over large areas. We present and discuss evidence that large and giant tortoises (family Testudinidae) are a useful model to rapidly provide empirical assessments of the use of taxon substitutes on a comparatively smaller scale. Worldwide, at least 36 species of large and giant tortoises went extinct since the late Pleistocene, leaving 32 extant species. We examine the latent conservation potential, benefits, and risks of using tortoise taxon substitutes as a strategy for restoring dysfunctional ecosystems. We highlight how, especially on islands, conservation practitioners are starting to employ extant large tortoises in ecosystems to replace extinct tortoises that once played keystone roles.

Starting in the late 1970s, ecologists began unraveling the role of recently extinct large vertebrates in evolutionary ecology and ecosystem dynamics. For example, for the first time, the ecology of large-seeded fruits in the Americas and divaricating plants in New Zealand were viewed as anachronistic, due to the missing large vertebrates that once influenced their evolutionary ecology (Greenwood and Atkinson 1977, Janzen and Martin 1982). Such views based on ecological history came at a time when evidence was mounting that humans played a significant, if not the major, role in the extinctions of the late Pleistocene (Martin and Klein 1984). Some three decades later, practitioners are now considering the role of ecological and evolutionary history in conservation practice. Some researchers have highlighted the underappreciated importance of evolutionary processes in effective biodiversity conservation planning (Erwin 1991, Atkinson 1998, Crandall et al. 2000, Ashley et al. 2003). 272

Others have gone further and called for restoring missing ecological functions and evolutionary potential with the introduction of related or sometimes unrelated taxa as analogues or substitutes for extinct species, often referred to as rewilding. While ‘‘rewilding’’ was originally coined by Soule´ and Noss (1998), the term’s meaning has been recently expanded in the scientific literature and media to include proposed reintroductions that incorporate ecological history back to the Pleistocene epoch (Atkinson 2001, Jones 2002, Steadman and Martin 2003, Galetti 2004, Donlan et al. 2005, 2006, Zimov 2005). We define taxon substitution as the replacement of extinct taxa by the introduction of analogue taxa related or ecologically similar to replace the ecological functions of the extinct species. Recently, empirical research has begun to illustrate how interactions of extinct species can be restored by using related or functionally similar taxa as taxon substitutes (Bond et al. 2004, Hansen et al. 2008, Griffiths et al.

2010). While the concept of rewilding remains a controversial means of restoring ecosystem processes (Caro 2007), taxon substitution projects are gaining acceptance within the public sector and a number of ambitious projects, firmly based on recent ecological history, are already underway (Zimov 2005, Curry 2008, Marris 2009). However, while some of these projects focused on taxon substitutions are based on sound science and justification, others may be misguided by bad historical information or dubious justifications. It is thus imperative that projects be judged on a case-by-case basis. In this paper we first briefly discuss how existing guidelines and definitions fall short in providing an overall framework to help guide and inform taxon substitutions. The most suitable framework in which taxon substitutions should be addressed is the reintroduction guidelines of the International Union for the Conservation of Nature (IUCN 1998). Currently, those guidelines are limited to sub-species level substitutions (Soorae 2008). We propose that the IUCN reintroduction guidelines be revised to explicitly encompass taxon substitutions and promote a more holistic and dynamic approach to restoration. Revised guidelines are not only needed to provide a framework for how to implement well-thought-out taxon substitution projects, but are also particularly needed in order to discourage moving forward on projects when they are not justified scientifically, socio-politically, or pragmatically. In addition to guidelines, a second challenge to taxon substitutions is the sheer scale and accompanying controversy of many of the proposed projects (Galetti 2004, Donlan et al. 2005, Zimov 2005, Caro 2007). In response, we propose rewilding with large and giant tortoises (family Testudinidae) as a model to rapidly advance our understanding of taxon substitutions and provide much-needed empirical assessments of rewilding as a restoration tool.

Taxon substitutions and the IUCN reintroduction guidelines

Large and giant tortoises: models for taxon substitutions

273

ISSUE

Tortoises of the family Testudinidae occur on most continents (except Antarctica and Australia) and on many isolated islands as a result of oceanic dispersal (Bonin et al. 2006, Crumly 2009). Many species, however, have gone extinct since the late Pleistocene. Within the last few millennia, the majority of tortoise extinctions occurred on islands. These recent tortoise extinctions present an opportunity to vet, implement, and evaluate the conservation potential of taxon substitutions. In that spirit, we provide an overview of extant and recently extinct large and giant tortoises, highlight the important roles of extant and extinct tortoises in some ecosystems, and argue that tortoises are a low-risk taxon for substitutions. Finally, we present and discuss several case stories that illustrate how extant tortoises can be suitable analogues for their recently extinct counterparts.

IBS SPECIAL

Under the current IUCN guidelines for reintroductions, the aims of taxon substitutions fall within ‘‘conservation or benign introductions’’, defined as attempts to ‘‘establish a species, for the purpose of conservation, outside its recorded distribution but within an appropriate habitat and ecogeographical area’’ (IUCN 1998). However, the aims behind taxon substitutions and conservation introductions can differ. Conservation introductions deal exclusively with species that have ‘‘become globally or locally extinct, or extirpated, in the wild’’ (IUCN 1998), while taxon substitutions may involve species that may not be threatened within their native range. The IUCN guidelines would benefit from incorporating this distinction. Reintroductions and translocations have historically been viewed in isolation from other conservation or restoration efforts, with a strong focus on avoiding extinction (Armstrong and Seddon 2008). The IUCN guidelines state that ‘‘a conservation/benign introduction should be undertaken only as a last resort when no opportunities for reintroduction into the original site or range exist and only when a significant contribution to the conservation of the species will result’’ (IUCN 1998). This inherently promotes a

single-species approach, ignoring the potential for restoring lost or currently dysfunctional species interactions by using taxon substitutions. A recent review identified that the main goal of a majority of reintroduction projects was to increase the number of individuals or populations of the target species; in contrast, only two of 62 projects specifically listed restoration of species interactions as a goal (Soorae 2008). Reintroduction biology would benefit by incorporating the recent emphasis across conservation biology that focuses on ecological effectiveness and species interactions (Soule´ et al. 2003, 2005, Ripple and Beschta 2007, Wright et al. 2007, Papanastasis 2009, Kaiser-Bunbury et al. 2010). With respect to restoring functional relationships within ecosystems, we propose that: 1) taxon substitutions, as defined above, are explicitly included as a stand-alone justification for a species introduction, and 2) reintroduction guidelines should include the use of taxa above the sub-species level under taxon substitutions, when it can be empirically demonstrated that the proposed substitute fulfills some ecological function(s) of the extinct taxon. A vital role of conservation scientists is to inform policy and planning, while practitioners strive to implement action based on sound science. If empirical evidence exists that demonstrates how dysfunctional or lost species interactions and ecosystem processes can benefit from taxon substitutions without negatively impacting human society, native biodiversity or ecosystem functions it should be a clear goal to support such work. In order to contribute to biodiversity conservation, taxon substitutions must therefore be viewed and executed in a cost-benefit framework. Our proposed revision of the IUCN reintroduction guidelines would facilitate a much-needed transparent debate on the role of taxon substitutions in biodiversity conservation, and provide a framework to advance the science and application of taxon substitutions in restoration projects. Alternatively, a more all-inclusive set of introduction guidelines could be advanced by returning to the original definitions of translocations (IUCN 1987), along the lines recently suggested by Armstrong and Seddon (2008). In this case, taxon substitutions simply become a specialised case of introductions.

Global distribution and status of large tortoises In general, we follow the taxonomy of Fritz and Havasˇ (2007), supplemented by recent findings in molecular studies (Austin et al. 2003, Le et al. 2006, Fritz and Bininda-Emonds 2007, Poulakakis et al. 2008). We include only tortoises with reported straight carapace or plastron lengths of 30 cm, which was chosen as our cut-off point because tortoises above this length are typically referred to as ‘‘large’’ and almost all known recently extinct tortoises are 30 cm. At least 36 species of large and giant tortoises have gone extinct since the Pleistocene, with the majority occurring on islands and vanishing in the late Pleistocene (Table 1). At least 32 species are still extant, with the majority of higherorder taxa found on continents (Table 2). The only remaining species of giant tortoises growing to more than one meter carapace length are found on the isolated islands of Gala´pagos in the Pacific Ocean and the Aldabra Atoll in the Indian Ocean; Geochelone (Centrochelys) sulcata in the African Sahel belt comes close, with lengths of up to 83 cm.

Additional extinct tortoise species continue to come to light: sub-fossil specimens have been recently discovered both in the Mediterranean and Caribbean regions (Caloi et al. 1986, Meylan and Sterrer 2000, Chesi et al. 2007, Steadman et al. 2007). Since the late Pleistocene, human predation and anthropogenic impacts have been major causes of tortoise extinction and endangerment. This is particularly welldocumented for some of the recent extinctions on islands, including Madagascar, the Mascarenes, and the Gala´pagos (Van Denburgh 1914, Cheke and Hume 2008, Pedrono 2008). There is also ample evidence of early human tortoise-hunting in mainland habitats from the Paleolithic and onwards, including the Mediterranean Rim and southern Africa (Stiner et al. 1999, Klein and Cruz-Uribe 2000, Blasco 2008). Some tortoise extinctions, however, occurred prior to human contact. For example, the Caribbean tortoise Hesperotestudo bermudae could have been lost due to partial submergence of its low-rise island home during recent interglacials (Meylan and Sterrer 2000, Olson et al. 2006).

IBS SPECIAL

ISSUE

Table 1. Extinct large and giant tortoises from the Pleistocene to Holocene. Taxon

Distribution

Island/mainland

Last record

Maximum carapace length (cm)

Aldabrachelys abrupta Aldabrachelys grandidieri Cheirogaster gymnesica Cheirogaster sp. Chelonoidis cubensis Chelonoidis elata Chelonoidis elephantopus Chelonoidis phantastica Chelonoidis? sellowi Chelonoidis sombrerensis Chelonoidis wallacei Chelonoidis sp. Chelonoidis sp. Chelonoidis sp. Chelonoidis? sp. Cylindraspis indica Cylindraspis inepta Cylindraspis peltates Cylindraspis triserrata Cylindraspis vosmaeri Geochelone burchardi Geochelone robusta Geochelone sp. Geochelone sp. Geochelone sp. Gopherus donlaloi Hesperotestudo crassiscutata

Madagascar Madagascar Minorca, Balearics Pituysic Islands, Balearics Cuba, Brazil Cuba Floreana, Gala´pagos Fernandina, Gala´pagos Uruguay Sombrero Island Rabida, Gala´pagos Santa Fe, Gala´pagos Great Abaco, Bahamas Dominican Republic Curac¸ao Reunion, Mascarenes Mauritius, Mascarenes Rodrigues, Mascarenes Mauritius, Mascarenes Rodrigues, Mascarenes Canary Islands Malta Bahamas Navassa Island Barbados Mexico Southern USA, Central America Kansas, USA Florida, USA Texas, USA Southern USA Celebes, Indonesia Ryukyu Islands, Japan Java, India India Mona Island

Island Island Island Island Mainland and island Island Island Island Mainland Island Island Island Island Island Island Island Island Island Island Island Island Island Island Island Island Mainland Mainland

Holocene Holocene Pleistocene

115 125

Hesperotestudo equicomes Hesperotestudo incisa Hesperotestudo johnstoni Hesperotestudo wilsoni Manouria margae Manouria oyamai Megalochelys atlas Megalochelys cautleyi Monachelys monensis

Mainland Mainland Mainland Mainland Island Island Mainland and island Mainland Island

Pleistocene Pleistocene Holocene Holocene Pleistocene Late Pleistocene Holocene Holocene Holocene Holocene Pleistocene Holocene Holocene Holocene Holocene Holocene Pleistocene Pleistocene Late Pleistocene Pleistocene Late Pleistocene Pleistocene Pleistocene Pleistocene Holocene Pleistocene Late Pleistocene Pleistocene Pleistocene Pleistocene

References

1, 2, 3 1, 2, 3, 4 1, 5 5 1, 6 1 7 86 7, 8 1 100 1, 6, 9 82 7, 10 7 46 11 60 12, 13 80 14 60 2, 15 ‘‘Large’’ 2, 15 42 2 ‘‘Giant’’ 2, 15 110 15, 16 1 120 1, 17, 18 60 6 40 6 60 19 54 (plastron) 20 150 1, 13, 21

120 150 ‘‘Giant’’ 180 50

1 1 1 1, 22 1, 23 24 1, 2 1 1, 6

1: Auffenberg 1974, 2: Arnold 1979, 3: Pedrono 2008, 4: Bour 1984, 5: Sondaar and van der Geer 2005, 6: Auffenberg 1967, 7: MacFarland et al. 1974a, 8: Ernst and Barbour 1989, 9: Lazell 1993, 10: Steadman et al. 1991, 11: Steadman et al. 2007, 12: Franz and Woods 1983, 13: Meylan and Sterrer 2000, 14: Hoijer 1963, 15: Arnold 1980, 16: Stoddart and Peake 1979, 17: Caloi et al. 1986, 18: Hunt and Schembri 1999, 19: Ray 1964, 20: Reynoso and Montellano-Ballesteros 2004, 21: Cisneros 2005, 22: Moodie and Devender 1979, 23: Hoijer 1951, 24: Takahashi et al. 2003.

274

Table 2. Extant large and giant tortoises. Species

Distribution

Island/mainland

Maximum carapace length (cm)

References

Aldabrachelys gigantea Astrochelys radiata Astrochelys yniphora Chelonoidis carbonaria

Aldabra, Seychelles Southern Madagascar Northwest Madagascar Northern South and Central America, introduced to Islands of Caribean Southern South America Northern South America and Trinidad Pinta, Gala´pagos Wolf volcano, Isabela, Gala´pagos San Cristobal, Gala´pagos Santiago, Gala´pagos Pinzon, Gala´pagos Sierra Negra, Isabela, Gala´pagos Espanola, Gala´pagos Darwin volcano, Isabela, Gala´pagos Santa Cruz, Gala´pagos Alcedo volcano, Isabela, Gala´pagos Cerro Azul, Isabela, Gala´pagos South Africa, southern Namibia India, Pakistan, Sri Lanka Burma Central and North Africa, Sahel-belt North-central Mexico

Island Island Island Mainland

105 40 45 70

1 1 1, 2 1, 3

Mainland Mainland and island Island Island Island Island Island Island Island Island Island Island Island Mainland Mainland and island Mainland Mainland Mainland

1, 1 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 5 1 1, 1, 3

Mainland Mainland Mainland

43 82 98 104 90 102 84 102 75 103 105 125 110 30 38 30 83 40 (fossils up to 100) 40 38 33

Mainland Mainland Mainland and island

30 40 60

3, 6 3, 5 3, 6

Mainland Mainland Mainland Mainland

33 70 34 40

3, 6 1, 3, 5 3 3

Chelonoidis chilensis Chelonoidis denticulata Chelonoidis abingdoni Chelonoidis becki Chelonoidis chatamensis Chelonoidis darwini Chelonoidis ephyppium Chelonoidis guntheri Chelonoidis hoodensis Chelonoidis microphyes Chelonoidis porteri Chelonoidis vandenburghi Chelonoidis vicina Chersina angulata Geochelone elegans Geochelone platynota Geochelone (Centrochelys) sulcata Gopherus flavomarginatus Gopherus agassizii Gopherus polyphemus Indotestudo elongata Indotestudo travancorica Kinixys erosa Manouria emys Manouria impressa Stigmochelys pardalis Testudo boettgeri Testudo marginata

South-western USA, Mexico South-eastern USA Asia (Nepal, India, China, Burma, Malaysia, Thailand, Cambodia, Vietnam) Western India Central West Africa Burma, Thailand, Malay Peninsula, Sumatra, Borneo Burma, Thailand, Malay Peninsula, Vietnam Eastern to southern Africa South-eastern Europe Greece, southern Balkan

3 4 4 4 4 4 4 4 4 4 4 4 3 5

3 3 3, 6

1: Ernst and Babour 1989, 2: Pedrono 2008, 3: Bonin et al. 2006, 4: MacFarland et al. 1974a, 5: Branch 2008, 6: Auffenberg 1974.

Tortoises as ecological and evolutionary keystone species

275

ISSUE

True land tortoises (family Testudinidae) arose around 55 million years ago, and are part of the oldest surviving reptile lineage (Auffenberg 1974, Bonin et al. 2006). The slow metabolism of tortoises and their ability to withstand long periods without food or water have enabled them to colonise almost all continents and many islands, with most species found in subtropical and tropical regions. Tortoises are important components of many ecosystems, and often attain high densities and biomass (Iverson 1982). For example, Astrochelys radiata density estimates in Madagascar vary from 1250 to 5400 tortoises km 2 (Leuteritz et al. 2005). On Aldabra, biomass of A. gigantea has been estimated to be between 3.5 and 58 tonnes per square kilometer more than the combined biomass of various species of large mammalian herbivores in any African wildlife area (Coe et al. 1979). In some African game parks, tortoise biomass outweighs that of several species of large mammalian herbivores (Iverson 1982, Branch 2008). Most extant tortoise species are highly generalised herbivores, frugivores or omnivores (Grubb 1971, Milton 1992, Bonin et al. 2006,

IBS SPECIAL

Similarly, there is evidence that Aldabra was re-colonised by giant tortoises from Madagascar at least three times, following sea-level changes that caused temporal submergence (Taylor et al. 1979). Projected anthropogenic increases in sea level may thus threaten the world’s largest remaining population of giant tortoises, Aldabrachelys gigantea, on Aldabra Atoll. Therefore, in some cases taxon substitution cannot be justified based on redressing past anthropogenic extinctions, but could be debated if the introduction of a generalised herbivore is deemed to be facilitating the desired trajectory of an ecosystem restoration project in such places. Indeed, this approach could be a good example of ‘‘restoring for the future’’ (Choi 2007, Macdonald 2009), e.g. maximising future ecosystem resilience. Few extant tortoises have been studied in sufficient detail to assign an updated IUCN Red List Category. Researchers have argued, however, that almost all extant tortoises are declining, and that many species should be considered endangered (Bonin et al. 2006, Branch 2008). Current threats to tortoises include collection by humans, introduced predators, and climate change (Erasmus et al. 2002, Bonin et al. 2006).

ISSUE

IBS SPECIAL

Branch 2008). Tortoises do not masticate their food, have a relatively simple digestive system, and many species have flexible digestive responses that are determined by diet (Guard 1980, Bjorndal 1989, Barboza 1995, Hailey 1997, McMaster and Downs 2008). It is likely that extinct tortoises had similarly broad diets. In many ecosystems, tortoises are thus likely to be or have been keystone species; not in the classical sense as it pertains to ecosystem importance in relation to biomass (Paine 1969), but rather in relation to the topological position and importance of tortoises in interaction- and food webs (Jorda´n 2009). A good example is the gopher tortoise Gopherus polyphemus, which influences a number of key processes in North American long-leaf pine grasslands and forest ecosystems, including herbivory, seed dispersal, nutrient cycling, and creating and maintaining habitat heterogeneity via trampling or digging of burrows (Kaczor and Hartnett 1990, Carlson et al. 2003, Birkhead et al. 2005, van Lear et al. 2005, Means 2006). Oceanic island ecosystems also offer many examples; given tortoises’ propensity for long-distance oceanic dispersal, they were likely often among the first large, non-volant vertebrates to colonise oceanic islands thus shaping these isolated ecosystems from early on in their history (Hnatiuk 1978, Arnold 1979, Meylan and Sterrer 2000, Gerlach et al. 2006). The resulting long, shared ecological and evolutionary histories of island tortoises and their plant communities has shaped many plant-tortoise interactions, many of which have since been lost as a result of tortoise decline or extinction (Iverson 1987, Strasberg 1996, Eskildsen et al. 2004, Gibbs et al. 2008, Hansen et al. 2008, Hansen and Galetti 2009, Griffiths et al. 2010). For example, ‘‘tortoise turf ’’, a plant community of endemic grass, herb and sedge species and engineered by continuous tortoise grazing and trampling, is thought to have been common on islands throughout the Indian Ocean before tortoises went extinct; it is now restricted to Aldabra (Merton et al. 1976, Cheke and Hume 2008). Furthermore, evidence is mounting that tortoises are or were important seed dispersers on continents and islands in ecosystems ranging from coastal shrub and dry deserts to rainforests (Rick and Bowman 1961, Hnatiuk 1978, Milton 1992, Varela and Bucher 2002, Strong and Fragoso 2006, Hansen et al. 2008, Jerozolimski et al. 2009). Tortoises can eat large amounts of fruits and swallow relatively large fruits and seeds. For example, yellow-footed tortoises Chelonoidis denticulata in Brazil with average carapace lengths of only 25 30 cm defecated seeds up to 4.0 1.7 cm in size (Jerozolimski et al. 2009). Variable gut passage times have been reported for tortoises, with average values ranging from a few days to three weeks, allowing for mean dispersal distances of several hundred metres (Rick and Bowman 1961, Hansen et al. 2008, Jerozolimski et al. 2009). Tortoises represent low-risk, high-impact taxon substitutions On many islands, tortoise extinction has resulted in dysfunctional ecosystems with respect to seed dispersal and herbivory (Gibbs et al. 2008, Hansen et al. 2008, Hansen and Galetti 2009, Griffiths et al. 2010). On 276

continents, the greater array of extant native herbivores and frugivores has likely helped buffer the ecological losses of tortoises (Hansen and Galetti 2009). Thus, tortoise taxon substitutions are arguably more imperative and appropriate on islands. Indeed, the impact and conservation value of tortoise taxon substitutions on islands is likely to be greater than suggested for mainland scenarios, due to the simpler ecosystems that have only recently been subjected to anthropogenic impacts (Kaiser-Bunbury et al. 2010). Tortoises can be regarded as low-risk taxon substitutes (Griffiths et al. 2010). Due to their highly generalised diets and relatively minimal reintroduction requirements, it is likely that tortoises introduced as taxon substitutions would be able to reestablish some ecosystem functions of the extinct tortoises and become integral parts of their new ecosystems. We highlight five reasons for large tortoises being particularly well suited for taxon substitutions. 1) Populations of large tortoises have high intrinsic growth rates and are easy to breed or rear in captivity. If juveniles are headstarted in captivity, they have high survival rates even in the presence of introduced predators (MacFarland et al. 1974a). 2) Tortoises are easy and cheap to fence in. This is especially important for their use in the relatively small conservation management areas found on many oceanic islands. Moreover, within fenced areas, it is easy to up- and down-regulate tortoise numbers and size of individuals, even in large areas or on a seasonal basis. Excess individuals can be kept in holding pens elsewhere, or cordoned-off sections of the restoration area, and require comparatively little husbandry. Similar techniques are used for livestock de facto taxon substitutes for extinct large mammalian herbivores in large-scale continental grassland restoration projects (Papanastasis 2009). 3) Their versatility enables them to be introduced into a wide range of habitats of varying qualities including highly degraded areas, making tortoises an attractive option for early-stage restoration efforts. There is some evidence that native plant species and communities evolved to withstand tortoise herbivory on islands (Merton et al. 1976, Eskildsen et al. 2004). This can lead to tortoise taxon substitutes actively preferring introduced and invasive plant species, leading to competitive release for the native species and thus further facilitating habitat recovery (Griffiths et al. 2010). 4) The risk of negatively impacting disease dynamics of the native fauna is small. Reptile diseases and parasites are typically species-specific, with little risk of transfer to other reptiles or other vertebrates (Cooper and Jackson 1981). However, several tortoise species and populations are increasingly affected by within-species diseases (Flanagan 2000). Thus, disease screening and quarantine measures are essential before tortoise taxon substitutions, especially if sourcing individuals from several populations. 5) While there are naturalised populations of mediumand large-sized tortoises in several places around the world (e.g. Balearic Islands, Caribbean Islands, Lever 2003), the risk of tortoises becoming invasive pests is remote, given their life history traits. More importantly, the nature of tortoises facilitates management; the removal of a recently introduced population is feasible if deemed necessary. There are important considerations and risks that will need mitigating before moving forward on any tortoise

taxon substitution program. Because of their highly generalised diet, precautions must be taken to avoid tortoises assisting in the spread of invasive plant species via defecated seeds. Giant Aldabra tortoises released on Curieuse Island in the Seychelles have been observed feeding on fruits and seeds of several invasive plant species, and may be a contributing factor to their spread there (Hambler 1994). Similarly, tortoise taxon substitutions in the Gala´pagos could lead to an increased rate of invasion of some plant species, such as bramble Rubus niveus which tortoises consume (R. Atkinson pers. comm.). Proper quarantine measures that determine passage times for 100% of ingested seeds are critical. When A. gigantea tortoises were quarantined before translocation to Round Island, Mauritius, some seeds took as long as three months to pass through the tortoises’ guts (Griffiths unpubl.). Another point to consider is the length of time required for tortoises to reach full size; while breeding and rearing large tortoises to use in rewilding projects may be straightforward, the time required may be a disadvantage in relation to projects that need a here-and-now capacity for restoring ecosystem function. In Gala´pagos, for example, there are plenty of small juvenile tortoises in the breeding centre that could be used immediately for taxon substitution projects. Yet with respect to potentially controlling biomass of invasive plants, the impact of one adult tortoise would be much greater than that of dozens of small juveniles. To swiftly reach specific restoration goals it may therefore be preferable to also use translocated adults. Selection of the taxon to be used for substitution must be strongly supported by ecological history, and balanced between phylogeny and natural history (Martin 1969, Donlan et al. 2006). For taxon substitutions whose goal centers on restoring species interactions or ecosystem function, choosing the genetically closest extant tortoise as a substitute may in some cases not be an appropriate selection criterion (contrary to what the IUCN reintroduction guidelines currently advise; IUCN 1998). This could be the case in an ecosystem where the closest relative of an extinct desert tortoise species is found in a rainforest, or in ecosystems where morphological divergence between species has led to more or less separate diets or feeding behaviours (see the Gala´pagos case story below for an illustrative example).

Giant tortoises are the flagship species of the Gala´pagos. Fifteen tortoise species are generally recognised, of which four are extinct and one is extinct in the wild (Table 1 and Table 2; for a recent discussion of taxonomic status of the Gala´pagos giant tortoises, see Russello et al. 2010). Species generally occur(ed) singly on separate islands, except for the largest island, Isabela, which has five species, each more or less separated by volcanic features. Two general carapace types exist, the rounded ‘‘domed’’ and the ‘‘saddleback’’ which rises sharply in the front (Van Denburgh 1914). The saddleback is considered an adaptation for browsing on elevated vegetation in dry habitats (Fig. 1A), while the domed is primarily found in wetter habitats where grazing is more common (Fig. 1B; Fritts 1984). Human over-exploitation in the 1800s was the primary cause of the tortoise extinctions (Van Denburgh 1914), with invasive mammals impeding the conservation of many of the remaining species (MacFarland et al. 1974b, Fritts et al. 2000). Poaching by humans is still a significant threat to several tortoise populations (e.g. on southern portions of Isabela, Fritts et al. 2000). Populations of several endangered species have been supplemented with captive-bred or -reared tortoises (MacFarland et al. 1974a, Fritts 1984). After a series of invasive mammal eradication campaigns (Campbell et al. 2004, Cruz et al. 2005, 2009), Gala´pagos restoration plans are considering taxon substitutions on two islands, Pinta and Floreana, where tortoise extinctions have occurred (Charles Darwin Foundation 2009, Gala´pagos Conservancy 2009). For Pinta Island tortoises C. abingdoni, the Espanola tortoise C. hoodensis is the prime candidate for taxon substitutions based on molecular data, along with sharing the same saddleback morphology (Poulakakis et al. 2008; Fig. 1C). Pinta Island restoration has been complicated by the fact that female tortoises from Isabela Island that are housed with Lonesome George (the sole remaining Pinta tortoise) recently laid eggs for the second time (although potentially infertile like the first batch from 2008). These events have led to speculation that George’s genetics may be salvageable (Russello et al. 2007, Tran 2009). For the extinct Floreana tortoise C. elephantopus the closest relatives of the extinct tortoise are four of the species on Isabela (Poulakakis et al. 2008), but recent molecular studies have also uncovered several tortoises with a recent Floreana ancestry (Russello et al. 2010). These findings could lead to a conservation dilemma: repatriate Floreana with available extant tortoises to restore lost dynamics as soon as possible, or initiate a long-term breeding programme to create a lineage of tortoises with a genetical make-up very similar to the extinct species? 277

ISSUE

There have been several tortoise translocation projects worldwide, usually as part of single-species conservation projects rather than establishing new populations (Pedrono and Sarovy 2000, Atkinson 2001, Tuberville et al. 2005, 2008). These case studies, along with the abundant information on the ecology and conservation of North American tortoises (Bury and Germano 1994), are informative with respect to vetting potential tortoise taxon substitution programs. Bolson tortoises Gopherus flavomarginatus were recently reintroduced to New Mexico from Mexico after being absent in the United States for thousands of years (Truett and Phillips 2009). Whether one views this as an introduction or a reintroduction depends on which restoration benchmark is applied

Gala´pagos Islands

IBS SPECIAL

Lessons from tortoise reintroductions, translocations, and taxon substitutions

(Donlan and Martin 2004). Within the conventional post-Columbian view, Bolson tortoises could be viewed as non-native species which is indeed the status assigned to it by the US National Park Service (Houston and Schreiner 1995). In contrast, it can be viewed as a reintroduction from a prehistoric view that stretches back to the late Pleistocene, where this tortoise still roamed much of the southern USA (Morafka 1988, Truett and Phillips 2009).

ISSUE

IBS SPECIAL

Figure 1. Large and giant tortoises and examples of potential and ongoing taxon substitution projects. In the Gala´pagos Islands there are two types of tortoises, reflecting adaptations to two main herbivory regimes: saddleback shells for browsing (A), and domed shells for grazing (B). Ideally, taxon substitutions should take such ecological information into consideration. For example, captive-bred juveniles of the saddleback Chelonoidis hoodensis from Espanola (C) could be used as taxon substitutes for the extinct saddleback C. elephantopus from Floreana (but see Gala´pagos case story in main text). In Mauritius, on the two small islands Ile aux Aigrettes and Round Island, ongoing tortoise taxon substitution projects aim to replace the recently extinct endemic species Cylindraspis triserrata and C. inepta. On Ile aux Aigrettes, giant Aldabra tortoises Aldabrachelys gigantea were introduced in 2000 and act as important seed dispersers and herbivores (D). Several tortoise nests have been found on the island and the eggs successfully reared in state-of-the-art incubators (E). On Round Island, A. gigantea and the smaller Madagascan radiated tortoise Astrochelys radiata were introduced in June 2007 (F H). In a reserve created in 2007 in Rodrigues, practitioners wait for 100 000 native plants to grow large enough to allow several hundred A. gigantea and A. radiata to graze and browse freely (I). Photo credits: (A) by F. J. Sulloway, (B) by CJD, (C) by R. J. Hobbs, (D, E) by DMH, (F, G, H) by CJG, (I) by Matjazˇ Kuntner.

278

While recent genetic information may be seen to complicate the management decision of the actual selection of the appropriate tortoise species for the potential repatriation onto Floreana and Pinta Island, such information is nonetheless informative. When combined with ecological knowledge, managers are positioned to proceed with a taxon substitution program based on sound science if deemed appropriate. Given that the invasive mammal populations the main driver of biodiversity loss and ecosystem degradation have been eliminated or are in the process of being eliminated, ecosystem restoration via taxon substitution is a logical next step in the conservation of the GalaÂ´pagos Islands. Mascarene Islands

Seychelles Before human arrival, many islands in the Seychelles housed giant tortoises. After human settlement in the mid 1700s, over-exploitation combined with depredation by introduced predators lead to the extinction of most populations of Seychelles tortoises by the early 1800s (Arnold 1979, Stoddart and Peake 1979). Molecular evidence from living tortoises and museum specimens strongly suggests that all Seychelles Aldabrachelys tortoises form one species, A. gigantea (Austin et al. 2003, Palkovacs et al. 2003). Today, A. gigantea is only found in the wild on Aldabra. However, there are claims of several extant Aldabrachelys species (Gerlach 2004, but see Frazier 2006). Whatever the eventual outcome of these taxonomical deliberations, the Seychelles have provided some valuable lessons in giant tortoise translocation be they taxon substitutions or reintroductions and offer much potential for future rewilding projects involving tortoises. For example, there was a large-scale translocation of a total of 250 tortoises, from Aldabra to Curieuse between 1978 and 1982 (Stoddart et al. 1982, Hambler 1994). Even though tortoises have been stolen from the island, or have died, this project is partly a success from a tourism point of view, but the effects on the ecosystem have been little studied (Stoddart et al. 1982, Samour et al. 1987, Hambler 1994), and worryingly include the dispersal by tortoises of invasive plants (Hambler 1994). On Cousine, introduced tortoises have been credited with restoring large-herbivore grazing, seed dispersal, and creating or maintaining habitat for endangered invertebrates (Samways et al. 2010). Several other islands harbour (re)introduced A. gigantea tortoises, including Bird, Denis, Silhouette, and Moyenne (Gerlach 2004, Hansen unpubl.). Many of these islands are privately owned tourist destinations, and the giant tortoises are often portrayed as a major attraction, highlighting the potential for economical as well as ecological justification for tortoise taxon substitutions on islands. Madagascar

279

ISSUE

Madagascar was home to two species of giant tortoises, Aldabrachelys grandidieri and A. abrupta. They went extinct in pre-European times, but likely as recently as 1250 and 750 yr ago, respectively (Burleigh and Arnold 1986). They occurred over large parts of the island, from coastal habitats to the central highlands, and often in sympatry (Arnold 1979, Pedrono 2008). Even though both species had domeshaped shells, isotope analyses of subfossil remains suggest some niche-differentiation in diet, with the larger A. grandidieri perhaps mostly a grazer confined to open areas,

IBS SPECIAL

The premier examples of the use of taxon substitutes to replace extinct tortoises in rewilding projects come from two of the Mascarene Islands in the Indian Ocean: Mauritius and Rodrigues. These islands were each home to two of the five species of giant tortoises from the endemic Mascarene genus Cylindraspis that went extinct between the early 1700s and mid 1800s (Table 1, Arnold 1979, Austin and Arnold 2001). Mauritius and its offshore islets were home to the two endemic species Cylindraspis inepta and C. triserrata. Remnants of coastal tortoise habitat are today restricted to offshore islets, most notably Ile aux Aigrettes and Round Island. Both islands are regarded as showcases for restoration as their degraded states are being reversed with a suite of well-planned restoration projects, including intensive systematic weeding programs and the eradication of introduced predators and herbivores prior to the introduction of endangered endemic plants and animals (Jones 2008). Both islands once harboured Cylindraspis tortoises (Cheke and Hume 2008). To advance the restoration of these islands, several individuals of Aldabrachelys gigantea from captive herds in Mauritius were introduced to Ile aux Aigrettes in 2000 and Round Island in 2007 (Jones 2002; Fig. 1D, F and G). Several Madagascan radiated tortoises Astrochelys radiata from captive-bred stocks were also introduced to Round Island, to examine which extant tortoise species is a more suitable substitute for the extinct Cylindraspis tortoises (Griffiths et al. 2010) (Fig. 1F and H). A major goal of these taxon substitutions is to restore lost grazing and seed dispersal functions. Preliminary results are encouraging on both islands: tortoises are dispersing seeds of several native plants and are selectively grazing exotic plant species, such as the highly invasive Leucena leucocephala (Fabaceae) on Ile aux Aigrettes (Fig. 1D). With proper management, tortoise grazing and browsing is likely to replace ongoing intensive manual weeding. The A. gigantea tortoises on Ile aux Aigrettes are already breeding, with some eggs hatching in situ and others collected for rearing in an incubator, providing the next generation of taxon substitutes for further restoration projects (Fig. 1E). Rodrigues was once home to the two giant tortoise species C. peltastes and C. vosmaeri (Arnold 1979). Rodrigues has suffered the extinction of most of its terrestrial vertebrates and was considered one of the most

degraded island ecosystems worldwide (Gade 1985). Several integrated restoration projects have been initiated since the late 1990s. In 2007, a nature reserve was created, which aims to recreate a large tract of a Rodrigues ecosystem as it is thought to have been 400 yr ago. Around 100 000 native and endemic shrubs and trees have been planted (Fig. 1I), and introduced A. gigantea and A. radiata tortoises are already grazing and browsing in parts of the reserve (Weaver and Griffiths 2008).

and the slightly smaller A. abrupta preferring to browse in more shrubby or forested habitats (Burleigh and Arnold 1986, Pedrono 2008). Many Madagascan plants possess seemingly anachronistic anti-herbivory traits (Grubb 2003), which may well be a result of selective pressures once exerted not only by extinct elephant birds (Bond and Silander 2007), but also by the extinct giant tortoises. The paleoecology of Madagascar was evoked by Burney (2003), who lamented the recent loss of all the large Madagascan vertebrates and then looked into a feasible future, where some of the functional ghosts could be resurrected with taxon substitutions. Madagascar may, in fact, provide large-scale restoration experiments in the near future. Several scientists have suggested using A. gigantea from Aldabra or from captive or rewilded stock elsewhere as taxon substitutes (D. A. Burney, C. J. Raxworthy and O. L. Griffiths pers. comm.). Captive A. gigantea in Madagascar seem to do well, even under severe neglect, so chances for successful introductions with minimum population management are quite high. Current efforts to model the likely distribution of A. abrupta, based on plentiful subfossil remains, could serve as a template for where in Madagascar A. gigantea could be used as a taxon substitute (C. J. Raxworthy pers. comm.).

IBS SPECIAL

ISSUE

The Caribbean Given the number of islands, their geological history, and their proximity to continental regions with large tortoises, there are relatively few known extinct tortoises from this region (ten species; Table 1). However, some fossil remains have only been discovered recently (Meylan and Sterrer 2000, Steadman et al. 2007), and it is likely that future work will increase the tally. The South American red-footed tortoise Chelonoidis carbonaria (and possibly C. denticulata) has been introduced to many islands in the Caribbean (Lever 2003). Chelonoidis carbonaria has been present in a naturalised state on some of the islands for decades, and perhaps several hundred years (Lever 2003). Future studies of Caribbean C. carbonaria populations could be compared to our knowledge about the species in its native continental South America (Bjorndal 1989, Moskovits and Bjorndal 1990, Jerozolimski et al. 2009), and provide valuable information for potential tortoise rewilding projects in the Caribbean. Interestingly, the naturalised C. carbonaria tortoises reported on Barbados are already de facto taxon substitutes for an extinct giant tortoise species (‘‘Geochelone’’; genus indet.) that occurred on the island (Table 1, Ray 1964, Lazell 1993). Conservation problems in the endangered Caribbean dry forests include invasive plants and a lack of seed dispersal services, and restoration here relies heavily on human intervention (Ray and Brown 1995). Building on the preliminary successful tortoise taxon substitutions in Mauritius (see above), we suggest that tortoise rewilding may well be a cost-effective way to facilitate even large-scale dry forest restoration in the Caribbean, with the tortoises acting as seed dispersers and herbivores. 280

Conclusion Despite global potential for resurrecting lost species interactions and restore degraded ecosystem functions, taxon substitutions remain controversial. We suggest that a healthy debate on the applicability of taxon substitutions could be facilitated by including guidelines for them within an expanded IUCN species translocation framework. This would have the added benefit of promoting species interactions and functional integrity of ecosystems as integral parts of all translocation projects. Furthermore, conducting taxon substitutions and reintroductions within a proper experimental framework will facilitate the interpretation of ecosystem responses, providing direction and insight for future management actions, as well as providing ideal templates for studies in community ecology (Armstrong and Seddon 2008). Due to their controversial nature, taxon substitution projects will likely be attracting closer scientific scrutiny than comparable taxon reintroductions. This extra scrutiny is justified due to the potential for unwanted consequences brought on by novel species interactions (Ricciardi and Simberloff 2009). Taxon substitutions are often advocated on grounds of reviving lost or dysfunctional ecosystem dynamics (Hamann 1993, Galetti 2004, Donlan et al. 2006, Gibbs et al. 2008, Hansen et al. 2008, Griffiths et al. 2010). These process-oriented hypotheses lend themselves toward an experimental a priori approach that is often lacking in re-introduction projects with a strict species conservation focus (Armstrong and Seddon 2008). Additionally, projects that use non-threatened species as taxon substitutions are useful to experimentally explore factors affecting translocation success, a luxury that translocation projects with endangered species can ill afford. Hypothesisdriven and explicit guidelines for taxon substitutions would also help discourage and prevent programs that are not justified on scientific, historical, or socio-political grounds. We have highlighted how the extinction biogeography of tortoises offers a model to provide much-needed empirical evaluation of taxon substitutions and rewilding efforts. For endangered tortoise species, we believe in situ conservation should take priority over their use in taxon substitution projects. But even in these cases, translocation or captive breeding could provide animals for taxon substitution projects elsewhere, affording the species one or several additional refuges from possible extinction. Considering how few extant species of large and particularly giant tortoises remain globally and how many of these have rapidly dwindling populations, heeding Aldo Leopold’s advice from an ecological, evolutionary, and historical perspective, is likely wise and doing so in the wild via taxon substitutions is perhaps ironic to some abiding by the precautionary principle: ‘‘the first rule of intelligent tinkering is to save all the pieces’’. Acknowledgements We thank Jack Williams and others with the International Biogeography Society for the invitation to present our views expressed here. We thank Rachel Atkinson, Massimo Delfino, Jack Frazier, Charles Crumly, Mauro Galetti, Richard Hobbs, Maria Norup, Gary Roemer, Frank Sulloway and Alfredo Valido for providing information or photos, hard-to-get publications, translations and/or comments on earlier drafts of the paper. DMH was

funded by the Velux Foundation, CJG by the Dulverton Trust and the Research Fund of the Univ. of Zu¨rich, and the Copeland Fellows Program at Amherst College funded CJD.

References

ISSUE

281

IBS SPECIAL

Armstrong, D. P. and Seddon, P. J. 2008. Directions in reintroduction biology. Trends Ecol. Evol. 23: 20 25. Arnold, E. N. 1979. Indian Ocean giant tortoises: their systematics and island adaptations. Phil. Trans. R. Soc. B 286: 127 145. Arnold, E. N. 1980. Recently extinct reptile populations from Mauritius and Re´union, Indian Ocean. J. Zool. 191: 33 47. Ashley, M. V. et al. 2003. Evolutionarily enlightened management. Biol. Conserv. 111: 115 123. Atkinson, I. A. E. 1998. Conserving plants as evolutionary entities: successes and unanswered questions from New Zealand and elsewhere. Aliso 16: 103 112. Atkinson, I. A. E. 2001. Introduced mammals and models for restoration. Biol. Conserv. 99: 81 96. Auffenberg, W. 1967. Notes on West Indian tortoises. Herpetologica 23: 34 44. Auffenberg, W. 1974. Checklist of fossil tortoises (Testudinidae). Bull. Florida State Mus. 18: 121 251. Austin, J. J. and Arnold, E. N. 2001. Ancient mitochondrial DNA and morphology elucidate an extinct island radiation of Indian Ocean giant tortoises (Cylindraspis). Proc. R. Soc. B 268: 2515 2523. Austin, J. J. et al. 2003. Was there a second adaptive radiation of giant tortoises in the Indian Ocean? Using mitochondrial DNA to investigate speciation and biogeography of Aldabrachelys (Reptilia, Testudinidae). Mol. Ecol. 12: 1415 1424. Barboza, P. S. 1995. Digesta passage and functional anatomy of the digestive tract in the desert tortoise (Xerobates agassizii). J. Comp. Phys. B 165: 193 202. Birkhead, R. D. et al. 2005. Patterns of folivory and seed ingestion by gopher tortoises (Gopherus polyphemus) in a southeastern pine savanna. Am. Midl. Nat. 154: 143 151. Bjorndal, K. A. 1989. Flexibility of digestive responses in two generalist herbivores, the tortoises Gechelone carbonaria and Geochelone denticulata. Oecologia 78: 317 321. Blasco, R. 2008. Human consumption of tortoises at Level IV of Bolomor Cave (Valencia, Spain). J. Archaeol. Sci. 35: 2839 2848. Bond, W. J. and Silander, J. A. 2007. Springs and wire plants: anachronistic defences against Madagascar’s extinct elephant birds. Proc. R. Soc. B 274: 1985 1992. Bond, W. J. et al. 2004. Plant structural defences against browsing birds: a legacy of New Zealand’s extinct moas. Oikos 104: 500 508. Bonin, F. et al. 2006. Turtles of the World. Johns Hopkins Univ. Press. Bour, R. 1984. Taxonomy, history and geography of Seychelles land tortoises and fresh-water turtles. In: Stoddart, D. R. (ed.), Biogeography and ecology of the Seychelles Islands. Dr W. Junk, pp. 281 307. Branch, B. 2008. Tortoises, terrapins & turtles of Africa. Struik Publ., South Africa. Burleigh, R. and Arnold, E. N. 1986. Age and dietary differences of recently extinct Indian Ocean tortoises (Geochelone s. lat.) revealed by carbon isotype analysis. Proc. R. Soc. B 227: 137 144. Burney, D. A. 2003. Madagascar’s prehistoric ecosystems. In: Goodman, S. M. and Benstead, J. (eds), The natural history of Madagascar. Univ. of Chigaco Press, pp. 47 51. Bury, R. B. and Germano, D. J. (eds) 1994. Biology of North American tortoises. Fish and Wildlife Research 13. United

States Dept of the Interior, National Biological Survey, Washington, DC, USA. Caloi, L. et al. 1986. La fauna a vertebrati terrestri del Pleistocene delle isole del Mediterraneo. Geol. Rom. 25: 235 256. Campbell, K. et al. 2004. Eradication of feral goats Capra hircus from Pinta Island, Gala´pagos. Oryx 38: 328 333. Carlson, J. E. et al. 2003. Seed dispersal by Gopherus polyphemus at Archbold Biological Station, Florida. Florida Sci. 66: 147 154. Caro, T. 2007. The Pleistocene re-wilding gambit. Trends Ecol. Evol. 22: 281 283. Charles Darwin Foundation 2009. <www.darwinfoundation. org/>, accessed 27 August 2009. Cheke, A. S. and Hume, J. P. 2008. Lost land of the Dodo. Christopher Helm. Chesi, F. et al. 2007. Middle Pleistocene giant tortoises from Sicily. In: Tintori, A. and Boccaletti, M. (eds), VII Giornate di Paleontologia della Societa` Paleontologica Italiana, p. 19. Choi, Y. D. 2007. Restoration ecology to the future: a call for new paradigm. Restor. Ecol. 15: 351 353. Cisneros, J. C. 2005. New Pleistocene vertebrate fauna from El Salvador. Rev. Brasileira Paleontol. 8: 239 255. Coe, M. J. et al. 1979. The biomass, production and carrying capacity of giant tortoises on Aldabra. Phil. Trans. R. Soc. B 286: 163 176. Cooper, J. E. and Jackson, O. F. 1981. Diseases of Reptilia. Academic Press. Crandall, K. A. et al. 2000. Considering evolutionary processes in conservation biology. Trends Ecol. Evol. 15: 290 295. Crumly, C. R. 2009. Totoises. In: Gillespie, R. G. and Clague, D. A. (eds), Encyclopedia of islands. Univ. of California Press, pp. 921 926. Cruz, F. et al. 2005. Conservation action in the Gala´pagos: feral pig (Sus scrofa) eradication from Santiago Island. Biol. Conserv. 121: 473 478. Cruz, F. et al. 2009. Bio-economics of large-scale eradication of feral goats from Santiago Island, Gala´pagos. J. Wildl. Manage. 73: 191 200. Curry, A. 2008. Pleistocene park: where the auroxen roam. Wired Magazine September 22. Donlan, C. J. and Martin, P. S. 2004. Role of ecological history in invasive species management and conservation. Conserv. Biol. 18: 267 269. Donlan, C. J. et al. 2006. Pleistocene rewilding: an optimistic agenda for twenty-first century conservation. Am. Nat. 168: 660 681. Donlan, J. et al. 2005. Re-wilding North America. Nature 436: 913 914. Erasmus, B. F. N. et al. 2002. Vulnerability of South African animal taxa to climate change. Global Change Biol. 8: 679 693. Ernst, C. H. and Barbour, R. W. 1989. Turtles of the world. Smithsonian Inst. Press. Erwin, T. L. 1991. An evolutionary basis for conservation strategies. Science 253: 750 752. Eskildsen, L. I. et al. 2004. Feeding response of the Aldabra giant tortoise (Geochelone gigantea) to island plants showing heterophylly. J. Biogeogr. 31: 1785 1790. Flanagan, J. 2000. Disease and health considerations. In: Klemens, M. W. (ed.), Turtle conservation. Smithsonian Inst. Press, pp. 85 95. Franz, R. and Woods, C. A. 1983. A fossil tortoise from Hispaniola. J. Herpetol. 17: 79 81. Frazier, J. 2006. Giant tortoises of the Indian Ocean. Herpetol. Rev. 37: 368 373. Fritts, T. H. 1984. Evolutionary divergence of giant tortoises in Gala´pagos. Biol. J. Linn. Soc. 21: 165 176. Fritts, T. H. et al. 2000. Progress and priorities in research for the conservation of reptiles. In: Sitwell, N. et al. (eds), Science

ISSUE

IBS SPECIAL

for conservation in Gala´pagos. Bulletin de’l Inst. Royal des Siences Naturalles de Belgique, pp. 39 45. Fritz, U. and Bininda-Emonds, O. R. P. 2007. When genes meet nomenclature: tortoise phylogeny and the shifting generic concepts of Testudo and Geochelone. Zoology 110: 298 307. Fritz, U. and Havasˇ, P. 2007. Checklist of chelonians of the world. Vertebr. Zool. 57: 149 368. Gade, D. W. 1985. Man and nature on Rodrigues: tragedy of an island common. Environ. Conserv. 12: 207 215. Gala´pagos Conservancy 2009. <www.Gala´pagos.org>, accessed 27 August 2009. Galetti, M. 2004. Parks of the Pleistocene: recreating the Cerrado and the Pantanal with megafauna. Natureza and Conservaçaõ 2: 93 100. Gerlach, J. 2004. Giant tortoises of the Indian Ocean. Chimaira, Frankfurt. Gerlach, J. et al. 2006. The first substantiated case of trans-oceanic tortoise dispersal. J. Nat. Hist. 40: 2403 2408. Gibbs, J. P. et al. 2008. The role of endangered species reintroduction in ecosystem restoration: tortoise-cactus interactions on Espanola island, Gala´pagos. Rest. Ecol. 16: 88 93. Greenwood, R. M. and Atkinson, I. A. E. 1977. Evolution of divaricating plants in New Zealand in relation to moa browsing. Proc. N. Z. Ecol. Soc. 24: 21 33. Griffiths, C. J. et al. 2010. The use of extant non-indigenous tortoises to replace extinct ecosystem engineers: a restoration tool. Rest. Ecol. 18: 1 7. Grubb, P. 1971. The growth, ecology and population structure of giant tortoises on Aldabra. Phil. Trans. R. Soc. B 260: 327 372. Grubb, P. J. 2003. Interpreting some outstanding features of the flora and vegetation of Madagascar. Perspect. Plant Ecol. Evol. Syst. 6: 125 146. Guard, C. L. 1980. The reptilian digestive system: general characteristics. In: Schmidt-Nielsen, K. et al. (eds), Comparative physiology: primitive mammals. Cambridge Univ. Press, pp. 43 51. Hailey, A. 1997. Digestive efficiency and gut morphology of omnivorous and herbivorous African tortoises. Can. J. Zool. 75: 787 794. Hamann, O. 1993. On vegetation recovery, goats and giant tortoises on Pinta Island, Gala´pagos, Ecuador. Biodivers. Conserv. 2: 138 152. Hambler, C. 1994. Giant tortoise Geochelone gigantea translocation to Curieuse Island (Seychelles): success or failure? Biol. Conserv. 69: 293 299. Hansen, D. M. and Galetti, M. 2009. The forgotten megafauna. Science 324: 42 43. Hansen, D. M. et al. 2008. Seed dispersal and establishment of endangered plants on oceanic Islands: the Janzen-Connell model, and the use of ecological analogues. PLoS One 3: e2111, doi: 10.1371/journal.pone.0002111. Hnatiuk, S. H. 1978. Plant dispersal by the Aldabran giant tortoise, Geochelone gigantea (Schweigger). Oecologia 36: 345 350. Hoijer, D. A. 1951. Pygmy elephant and giant tortoise. Sci. Monthly 72: 3 8. Hoijer, D. A. 1963. Geochelone from the Pleistocene of Curaçao, Netherlands Antilles. Copeia 1963: 579 580. Houston, D. B. and Schreiner, E. G. 1995. Alien species in national parks: drawing lines in space and time. Conserv. Biol. 9: 204 209. Hunt, C. O. and Schembri, P. J. 1999. Quaternary environments and biogeography of the Maltese Islands. In: Mifsud, A. and

282

Savona Ventura, C. (eds), Facets of Maltese prehistory. The Prehistoric Society of Malta, p. 243. IUCN 1987. IUCN position statement on the translocation of living organisms: introductions, re-introductions, and restocking. Prepared by the Species Survival Commission in collaboration with the Commission on Ecology and the Commission on Environmental Policy, Law and Administration. IUCN 1998. Guidelines for re-introductions. IUCN/SSC Reintroduction Specialist Group, IUCN, Gland, Switzerland and Cambridge, UK. Iverson, J. B. 1982. Biomass in turtle populations: a neglected subject. Oecologia 55: 69 76. Iverson, J. B. 1987. Tortoises, not dodos, and the Tambalacoque tree. J. Herpetol. 21: 229 230. Janzen, D. H. and Martin, P. S. 1982. Neoptropical anachronisms: the fruits the Gomphoteres ate. Science 215: 19 27. Jerozolimski, A. et al. 2009. Are tortoises important seed dispersers in Amazonian forests? Oecologia 161: 517 528. Jones, C. G. 2002. Reptiles and amphibians. In: Perrow, M. R. and Davy, A. J. (eds), Handbook of ecological restoration. Cambridge Univ. Press, pp. 355 375. Jones, C. G. 2008. Practical conservation on Mauritius and Rodrigues: steps towards the restoration of devastated ecosystems. In: Cheke, A. S. and Hume, J. P. (eds), Lost land of the Dodo. Christopher Helm, pp. 226 259. Jorda´n, F. 2009. Keystone species and food webs. Phil. Trans. R. Soc. B 364: 1733 1741. Kaczor, S. and Hartnett, D. 1990. Gopher tortoise (Gopherus polyphemus) effects on soils and vegetation in a Florida USA sandhill community. Am. Midl. Nat. 123: 100 111. Kaiser-Bunbury, C. N. et al. 2010. Conservation and restoration of plant animal mutualisms on oceanic islands. Perspect. Plant Ecol. Evol. Syst. 12: 131 143. Klein, R. D. and Cruz-Uribe, K. 2000. Stone age population numbers and the average tortoise size at Byneskranskop Cave 1 and Die Kelders Cave 1, Southern Cape Province, South Africa. S. Afr. Archaeol. Bull. 38: 26 30. Lazell, J. D. J. 1993. Tortoise, cf. Geochelone carbonaria, from the Pleistocene of Anguilla, northern Lesser Antilles. J. Herpetol. 27: 485 486. Le, M. et al. 2006. A molecular phylogeny of tortoises (Testudines: Testudinidae) based on mitochondrial and nuclear genes. Mol. Phylogenet. Evol. 40: 517 531. Leuteritz, T. E. J. et al. 2005. Distribution, status, and conservation of radiated tortoises (Geochelone radiata) in Madagascar. Biol. Conserv. 124: 451 461. Lever, C. 2003. Naturalized reptiles and amphibians of the world. Oxford Univ. Press. Macdonald, D. W. 2009. Lessons learnt and plans laid: seven awkward questions for the future of reintroductions. In: Hayward, M. W. and Somers, M. J. (eds), Reintroduction of top-order predators. Wiley-Blackwell, pp. 411 448. MacFarland, C. et al. 1974a. The Gala´pagos giant tortoises (Geochelone elephantopus) part II: conservation methods. Biol. Conserv. 6: 198 212. MacFarland, C. G. et al. 1974b. The Gala´pagos giant tortoises (Geochelone elephantopus) part I: status of the surviving populations. Biol. Conserv. 6: 118 133. Marris, E. 2009. Reflecting the past. Nature 462: 30 32. Martin, P. S. 1969. Wanted: a suitable herbivore. Nat. Hist. 78: 35 39. Martin, P. S. and Klein, R. G. 1984. Quaternary extinctions: a prehistoric revolution. Univ. of Arizona Press.

283

ISSUE

Russello, M. A. et al. 2010. DNA from the past informs ex situ conservation for the future: an ‘‘extinct’’ species of Galapagos tortoise identified in captivity. PLoS One 5: e8683, doi: 8610.1371/journal.pone.0008683. Samour, H. J. et al. 1987. A survey of the Aldabra giant tortoise population introduced on Curieuse Island, Seychelles. Biol. Conserv. 41: 147 158. Samways, M. J. et al. 2010. Restoration of a tropical island: Cousine Island, Seychelles. Biodivers. Conserv. 19: 425 434. Sondaar, P. Y. and van der Geer, A. A. E. 2005. Evolution and extinction of Plio-Pleistocene island ungulates. Quaternaire 2: 241 256. Soorae, P. S. 2008. Global re-introduction perspectives: reintroduction case-studies from around the globe. IUCN/ SSC Re-introduction Specialist Group, Abu Dhabi, UAE. Soule´, M. and Noss, R. 1998. Rewilding and biodiversity: complementary goals for continental conservation. Wild Earth 8: 18 28. Soule´, M. E. et al. 2003. Ecological effectiveness: conservation goals for interactive species. Conserv. Biol. 17: 1238 1250. Soule´, M. E. et al. 2005. Strongly interacting species: conservation policy, management, and ethics. Bioscience 55: 168 176. Steadman, D. W. and Martin, P. S. 2003. The late Quaternary extinction and future resurrection of birds on Pacific islands. Earth-Sci. Rev. 61: 133 147. Steadman, D. W. et al. 1991. Chronology of Holocene vertebrate extinction in the Gala´pagos Islands. Quat. Res. 36: 126 133. Steadman, D. W. et al. 2007. Exceptionally well preserved late Quaternary plant and vertebrate fossils from a blue hole on Abaco, The Bahamas. Proc. Nat. Acad. Sci. USA 104: 19897 19902. Stiner, M. C. et al. 1999. Paleolithic population growth pulses evidenced by small animal exploitation. Science 283: 190 194. Stoddart, D. R. and Peake, J. F. 1979. Historical records of Indian Ocean giant tortoise populations. Phil. Trans. R. Soc. B 286: 147 158. Stoddart, D. R. et al. 1982. Tortoises and tourists in the western Indian Ocean: the Curieuse experiment. Biol. Conserv. 24: 67 80. Strasberg, D. 1996. Diversity, size composition and spatial aggregation among trees on a 1-ha rain forest plot at La Reunion. Biodivers. Conserv. 5: 825 840. Strong, J. N. and Fragoso, J. M. V. 2006. Seed dispersal by Geochelone carbonaria and Geochelone denticulata in northwestern Brazil. Biotropica 38: 683 686. Takahashi, A. et al. 2003. A new species of the genus Manouria (Testudines: Testudinidae) from the Upper Pleistocene of the Ryukyu Islands, Japan. Paleontol. Res. 7: 195 217. Taylor, J. D. et al. 1979. Terrestrial faunas and habitats of Aldabra during the late Pleistocene. Phil. Trans. R. Soc. B 286: 47 66. Tran, M. 2009. Lonesome George, the last Gala´pagos giant tortoise, may become a dad. <http://guardian.co.uk>, accessed 28 August 2009. Truett, J. and Phillips, M. 2009. Beyond historic baselines: restoring bolson tortoises to Pleistocene range. Ecol. Restor. 27: 145 151. Tuberville, T. D. et al. 2005. Translocation as a conservation tool: site fidelity and movement of repatriated gopher tortoises (Gopherus polyphemus). Anim. Conserv. 8: 349 358. Tuberville, T. D. et al. 2008. Long-term apparent survival of translocated gopher tortoises: a comparison of newly released and previously established animals. Biol. Conserv. 141: 2690 2697.

IBS SPECIAL

McMaster, M. K. and Downs, C. T. 2008. Digestive parameters and water turnover of the leopard tortoise. Comp. Biochem. Physiol. A 151: 114 125. Means, D. B. 2006. Vertebrate faunal diversity of longleaf pine ecosystems. In: Jose, S. et al. (eds), The longleaf pine ecosystem: ecology, silviculture, and restoration. Springer, pp. 157 213. Merton, L. F. H. et al. 1976. Giant tortoise and vegetation interactions on Aldabra Atoll part 1: inland. Biol. Conserv. 9: 293 304. Meylan, P. A. and Sterrer, W. 2000. Hesperotestudo (Testudines: Testudinidae) from the Pleistocene of Bermuda, with comments on the phylogenetic position of the genus. Zool. J. Linn. Soc. 128: 51 76. Milton, S. J. 1992. Plants eaten and dispersed by adult leopard tortoises Geochelone pardalis (Reptilia, Testudinae) in the southern Karoo. S. Afr. J. Zool. 27: 45 49. Moodie, K. B. and Devender, T. R. V. 1979. Extinction and extirpation in the herpetofauna of the Southern High Plains with emphasis on Geochelone wilsonii (Testudinidae). Herpetologica 35: 198 206. Morafka, D. J. 1988. Historical biogeography of the bolson tortoise. Ann. Carnegie Mus. 57: 47 72. Moskovits, D. K. and Bjorndal, K. A. 1990. Diet and food preferences of the tortoises Geochelone carbonaria and Geochelone denticulata in northwestern Brazil. Herpetologica 46: 207 218. Olson, S. L. et al. 2006. Geological constraints on evolution and survival in endemic reptiles on Bermuda. J. Herpetol. 40: 394 398. Paine, R. T. 1969. A note on trophic complexity and community stability. Am. Nat. 103: 91 93. Palkovacs, E. P. et al. 2003. Are the native giant tortoises from the Seychelles really extinct? A genetic perspective based on mtDNA and microsatellite data. Mol. Ecol. 12: 1403 1413. Papanastasis, V. P. 2009. Restoration of degraded grazing lands through grazing management: can it work? Restor. Ecol. 17: 441 445. Pedrono, M. 2008. The tortoises and turtles of Madagascar. Natural History Publications (Borneo). Pedrono, M. and Sarovy, A. 2000. Trial release of the world’s rarest tortoise Geochelone yniphora in Madagascar. Biol. Conserv. 95: 333 342. Poulakakis, N. et al. 2008. Historical DNA analysis reveals living descendants of an extinct species of Gala´pagos tortoise. Proc. Nat. Acad. Sci. USA 105: 15464 15469. Ray, C. E. 1964. A small assemblage of vertebrate fossils from Spring Bay, Barbados. J. Barbados Mus. Hist. Soc. 31: 11 22. Ray, G. J. and Brown, B. J. 1995. Restoring Caribbean dry forests: evaluation of tree propagation techniques. Restor. Ecol. 3: 86 94. Reynoso, V.-H. and Montellano-Ballesteros, M. 2004. A new giant turtle of the genus Gopherus (Chelonia: Testudinidae) from the Pleistocene of Tamaulipas, Mexico, and a review of the phylogeny and biogeography of gopher tortoises. J. Vertebr. Paleontol. 24: 822 837. Ricciardi, A. and Simberloff, D. 2009. Assisted colonization is not a viable conservation strategy. Trends Ecol. Evol. 24: 248 253. Rick, C. M. and Bowman, R. I. 1961. Gala´pagos tomatoes and tortoises. Evolution 15: 407 417. Ripple, W. J. and Beschta, R. L. 2007. Restoring Yellowstone’s aspen with wolves. Biol. Conserv. 138: 514 519. Russello, M. A. et al. 2007. Lonesome George is not alone among Gala´pagos tortoises. Curr. Biol. 17: 317 318.

IBS SPECIAL

ISSUE

Van Denburgh, J. 1914. Expedition of the California Academy of Sciences to the Gala´pagos Islands 1905 1906. Proc. Cal. Acad. Sci. 2: 203 374. Van Lear, D. H. et al. 2005. History and restoration of the longleaf pine-grassland ecosystem: implications for species at risk. For. Ecol. Manage. 211: 150. Varela, R. O. and Bucher, E. H. 2002. Seed dispersal by Chelonoidis chilensis in the Chaco dry woodland of Argentina. J. Herpetol. 36: 137 140.

284

Weaver, E. A. and Griffiths, O. L. 2008. A guide to La Vanille Re´serve des Mascareignes. Bioculture Press, Rivie`re des Anguilles, Mauritius. Wright, S. J. et al. 2007. The plight of large animals in tropical forests and the consequences for plant regeneration. Biotropica 39: 289 291. Zimov, S. A. 2005. Pleistocene park: return of the mammoth’s ecosystem. Science 308: 796 798.

Ecography 33: 285 294, 2010 doi: 10.1111/j.1600-0587.2010.06203.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Helmut Hillebrand. Accepted 12 March 2010

Extinction debt on oceanic islands Kostas A. Triantis, Paulo A. V. Borges, Richard J. Ladle, Joaquıń Hortal, Pedro Cardoso, Clara Gaspar, Francisco Dinis, Ene´sima Mendonça, Lućia M. A. Silveira, Rosalina Gabriel, Catarina Melo, Ana M. C. Santos, Isabel R. Amorim, Se´rvio P. Ribeiro, Artur R. M. Serrano, Jose´ A. Quartau and Robert J. Whittaker K. A. Triantis (konstantinos.triantis@ouce.ox.ac.uk), Biodiversity Research Group, Oxford Univ. Centre for the Environment, South Parks Road, Oxford, OX1 3QY, UK, and Dept de Cieˆncias Agra´rias, Univ. dos Açores, CITAA (Azorean Biodiversity Group), Terra-Cha˜, PT-9700851, Angra do Heroı´smo, Terceira, Açores, Portugal. P. A. V. Borges, P. Cardoso, C. Gaspar, F. Dinis, E. Mendonça, L. M. A. Silveira, R. Gabriel, C. Melo and I. R. Amorim, Univ. dos Açores, Dept de Cieˆncias Agra´rias, CITAA (Azorean Biodiversity Group), Terra-Cha˜, PT9700-851, Angra do Heroı´smo, Terceira, Açores, Portugal. R. J. Ladle, Biodiversity Research Group, Oxford Univ., Centre for the Environment, South Parks Road, Oxford, 0X1 3QY, UK. J. Hortal, NERC Centre for Population Biology, Imperial College at Silwood Park, Ascot, SL5 7PY, UK. A. M. C. Santos, Div. of Biology, Imperial College at Silwood Park, Ascot, SL5 7PY, UK and Dept de Cieˆncias Agra´rias, Univ. dos Açores, CITAA (Azorean Biodiversity Group), Terra-Cha˜, PT-9700-851, Angra do Heroı´smo, Terceira, Açores, Portugal. S. Ribeiro, Univ. Federal de Ouro Preto, DEBIO/Inst. de Cieˆncias Exatas e Biologicas, Lab. Evolutionary Ecology of Canopy Insects, 35400-000, Ouro Preto, MG, Brazil. A. R. M. Serrano and J. A. Quartau, Centro de Biologia Ambiental/Dept de Biologia Animal, Faculdade de Ciencias da Univ. de Lisboa, R. Ernesto de Vasconcelos, C2, PT-1749-016 Lisboa, Portugal. R. J. Whittaker, Biodiversity Research Group, Oxford Univ. Centre for the Environment, South Parks Road, Oxford, OX1 3QY, UK, and Centre for Macroecology, Evolution and Climate, Dept of Biology, Univ. of Copenhagen, DK-2100 Copenhagen, Denmark.

such as the introduction of non-native species (Paulay 1994, May et al. 1995, Blackburn et al. 2004, Steadman 2006, Hanski et al. 2007, Whittaker and Ferna´ndez-Palacios 2007). Habitat destruction is rarely absolute and typically results in many species being reduced to a few small, isolated populations, each susceptible to a variety of stochastic factors such as random fluctuations in demography, changes of the local environment and the erosion of genetic variability (Lande 1993). Hence, it can take several generations for the full impact of habitat destruction and fragmentation to be visible in the number of extinctions (Tilman et al. 1994, Helm et al. 2006, Vellend et al. 2006). 285

ISSUE

In their natural state, oceanic islands typically support a substantial proportion of endemic species, many of which have been lost as a direct consequence of recent human habitation (Steadman 2006, Whittaker and Ferna´ndez-Palacios 2007). The biodiversity ‘‘crisis’’ is thus nowhere more apparent and in need of urgent action than on remote islands (Paulay 1994). The majority of the documented extinctions since ca AD 1600 are of species endemic to oceanic islands. Although the specific causes of these extinctions are often difficult to attribute (Whittaker and Ferna´ndez-Palacios 2007), the primary drivers are the habitat destruction and fragmentation universally associated with human colonization, in combination with other factors

IBS SPECIAL

Habitat destruction is the leading cause of species extinctions. However, there is typically a time-lag between the reduction in habitat area and the eventual disappearance of the remnant populations. These ‘‘surviving but ultimately doomed’’ species represent an extinction debt. Calculating the magnitude of such future extinction events has been hampered by potentially inaccurate assumptions about the slope of species area relationships, which are habitat- and taxon-specific. We overcome this challenge by applying a method that uses the historical sequence of deforestation in the Azorean Islands, to calculate realistic and ecologically-adjusted species area relationships. The results reveal dramatic and hitherto unrecognized levels of extinction debt, as a result of the extensive destruction of the native forest: 95%, in B600 yr. Our estimations suggest that more than half of the extant forest arthropod species, which have evolved in and are dependent on the native forest, might eventually be driven to extinction. Data on species abundances from Graciosa Island, where only a very small patch of secondary native vegetation still exists, as well as the number of species that have not been found in the last 45 yr, despite the extensive sampling effort, offer support to the predictions made. We argue that immediate action to restore and expand native forest habitat is required to avert the loss of numerous endemic species in the near future.

ISSUE

IBS SPECIAL

This time-lag represents an ‘‘extinction debt’’ (Tilman et al. 1994) a future ecological cost of habitat destruction that may not be initially apparent in studies made shortly after habitat loss has occurred. For this reason it is probable that the true ecological costs of the historically recent spate of habitat destruction, disturbance and fragmentation on many oceanic islands are yet to be realised (Diamond 1989), i.e. there exist many extant but seriously imperilled species. Developing methods to quantify the magnitude and taxonomic distribution of the extinction debt is clearly important for effective conservation planning and prioritization. However, accurate assessment of extinction rates and their extrapolation into the future requires robust longterm data on species occurrences data which are rarely available, especially for less conspicuous taxa such as invertebrates. The lack of appropriate knowledge has led to an inevitable reliance on indirect measures and theoretical projections of extinctions (McDonald and Brown 1992, Heywood et al. 1994, May et al. 1995, Pimm et al. 1995, Brooks et al. 1997, Rosenzweig 2001, Brook et al. 2003, Whittaker et al. 2005, Kuussaari et al. 2009, Ladle 2009). One of the most commonly used methods for estimating future extinctions is to extrapolate from the characteristic form of the classic island species area relationship [S cAz, where S is the number of species, A is (island) area, and c and z are constants] derived from island biogeography theory (Preston 1962, MacArthur and Wilson 1967). The consequences of habitat loss under this framework can be predicted following the ‘‘rule of thumb’’ calculation that a 10-fold decrease in area results in a twofold decrease in species (Darlington 1957), or alternatively, when an area of habitat is reduced by 90%, the number of species eventually drops to one half. This approach has been applied at varying sometimes very coarse scales to forecast species losses as a function of habitat loss due to factors such as deforestation (Brooks et al. 2002) or future climate change (Thomas et al. 2004). Even though the accuracy of this approach critically rests upon accurate estimation of the slope (z) of the relationship (Rosenzweig 2001, Whittaker et al. 2005, Lewis 2006, Whittaker and Ferna´ndez-Palacios 2007), it has been commonplace to assume z 0.25 across a range of different taxonomic groups, scales and ecogeographical systems (May et al. 1995, Brooks et al. 2002, Thomas et al. 2004). Although arthropods represent the bulk of all known living species, the level of threat imposed by global environmental changes to arthropod diversity remains poorly documented (Brooks et al. 2006, Fonseca 2010). Dunn (2005) has estimated that roughly 44 000 insect extinctions have occurred in the last 600 yr, but the number of extinctions documented during this period is 61 species (IUCN 2009; the respective number for arachnids is zero). Here, we apply a method that uses the historical information on deforestation on the Azores (a remote Atlantic Ocean archipelago) to generate more accurate estimates of local extinctions or extirpations (hereafter extinctions) for the endemic forest-dependent species of three well-studied groups of arthropods from the Azores, namely the spiders (Araneae), the true bugs (Hemiptera) and the beetles (Coleoptera). This approach has been used in a few mainland systems (Pimm and Askins 1995, Helm et al. 286

2006, see also Kuussaari et al. 2009 for a recent review) but we are not aware of any similar study on islands, despite the widely accepted notion that islands and especially oceanic islands have suffered and will probably suffer increased extinctions following habitat loss. The Azores constitute an ideal model system for assessing extinction debt because: 1) they have lost 95% of their original native forest during the six centuries of human occupation; 2) being one of the most isolated archipelagos on Earth they support a significant number of single island endemic species (SIE; i.e. endemic species restricted to one island) (Borges et al. 2005b, Borges and Hortal 2009, Cardoso et al. 2010); 3) the history of human settlement and deforestation is well known (Frutuoso 1963, Silveira 2007), and; 4) extensive distributional data exist for a range of taxa (Borges et al. 2005b).

Methods Study area The first human settlements were established in the Azores (Supplementary material Fig. S1) around AD 1440. More than 550 yr of human presence has taken its toll on the local fauna and flora, 420 species of which (out of the 4467 total terrestrial taxa known from the Azores) are endemic to the archipelago (Borges et al. 2005b). Today, ca 70% of the vascular plant species and 58% of the arthropod species found in the Azores are exotic, many of them invasive (Borges et al. 2005b, 2006). The native ‘‘laurisilva’’, a humid evergreen broadleaf laurel forest, was the predominant vegetation form in the Azores before human colonization in the 15th century (ca AD 1440). Here, we consider as ‘‘native forest’’ both the humid evergreen broadleaf laurel forest and other native forest types such as the Juniperus brevifolia- and Erica azorica-dominated forests. The Azorean laurisilva differs from that found on Madeira and on the Canary Islands as it includes just a single species of Lauraceae (Laurus azorica), although also featuring several species of sclerophyllous and microphyllous trees and shrubs (e.g. J. brevifolia and E. azorica), and luxuriant bryophyte communities, covering all available substrata (Gabriel and Bates 2005). The destruction of the native forest in the Azores has followed a clear temporal sequence. At the time of human colonization the archipelago was almost entirely covered by forest (ca AD 1440) (Martins 1993, Silveira 2007). By 300 yr ago (ca AD 1700) human activities had restricted the native forest in most islands to areas above 300 m a.s.l. and by ca AD 1850, areas with native forest were mainly present above 500 m a.s.l. (Silveira 2007). The development of an economy dependent on milk production during the last decades of the 20th century drove a further reduction of native forest area, with the clearing of large fragments at mid- and high-altitude for pasture, further decreasing the native forest to its current extent of 2.5% of the total area of the archipelago (B58 km2 in total). Thus, in B600 yr 95% of the original native forest has been destroyed (Gaspar 2007, Gaspar et al. 2008, Table 1).

Data

To explore the impact of native forest destruction on current levels of endemic arthropod species richness, we 287

ISSUE

Calculation of extinction debt

IBS SPECIAL

As a result of the exhaustiveness of taxonomic work, the relative poorness of the Azorean fauna, and the intensive sampling during the last ten years (see Supplementary material for an analytical description of the sampling method), the Borges et al. (2005b) checklist (updated also with recent unpublished data) includes virtually all arthropod species native to the Azores, reported and described from 1859 (DroueÂ¨t 1859) up to today, as well as an accurate account of their presence or absence in all the islands of the archipelago. The data for the Araneae, Hemiptera and Coleoptera are particularly comprehensive (Borges et al. 2005b, Borges and Wunderlich 2008, Cardoso et al. 2010). In this context, even if more species remain to be discovered from the islands in the future (e.g. Borges and Wunderlich 2008), we can reasonably regard each island as being currently proportionally equally well-sampled. In 1998, 60 native species (excluding Crustacea, Acari, Collembola, Hymenoptera and Diptera) were known to be SIE. During 1999 and 2000, 64 transects were set up, covering all remnants of native forest in the Azorean islands (BALA project) (Borges et al. 2005a, Ribeiro et al. 2005, Table 1). Eight species out of the original 60 SIE were found in other islands, but also 13 new species were described, nine of them being SIE (Borges and Wunderlich 2008). During 2003 and 2004, 38 new transects were set up in the same forest remnants (Gaspar 2007, Gaspar et al. 2008). After this intensive additional round of surveys, only one further species previously thought to be a SIE was found in another island, demonstrating the high reliability of the current checklist at the island level. Based on previous work (Borges and Brown 1999, Borges et al. 2005a, 2006, 2008, Ribeiro et al. 2005, Gaspar 2007, Borges and Wunderlich 2008, Gaspar et al. 2008) the endemic arthropods were classified as native forest dependent and non-forest dependent species (e.g. cave-adapted species, native grassland specialists, species also surviving in exotic forests or other man-made habitats). A species was considered forest-dependent (i.e. forest specialist) when 85% or more of its individuals have been collected in native vegetation (see Forest dependent endemic species in Supplementary material Table S1). Only the forest-dependent species endemic to the archipelago (59 species in total) were considered for further analyses; these species represent 56% of all the endemic species of the taxa considered. Despite the intensive survey effort recently carried out in anthropogenic habitats on some of the islands (Terceira, Pico, Graciosa and Santa Maria; Borges and Brown 1999, Borges et al. 2005a, 2006, 2008, Borges and Wunderlich 2008; see also Supplementary material), none of the species considered as a native forest endemic here has been found to have large populations in any other type of land use (B15% of their total numbers of individuals, after standardising for sampling effort; see details in Supplementary material Table S1). The completeness and comparability of these surveys was verified using a number of sampling effort algorithms (see Sampling effort analysis in the Supplementary material). The respective species lists of endemic forest specialists for the above three taxa were extracted for the areas of native

forest corresponding to four points in time (below). This step was undertaken using SQL-based queries on the ATLANTIS-Azores database by means of the Atlantis Tierra 2.0 software (Zurita and Arechavaleta 2003, Borges et al. 2005b, Table 2). The ATLANTIS-Azores database includes an exhaustive checklist created by many taxonomists, who have recently performed a detailed revision of the taxonomic status of many species, identified many synonyms and improved the list of Azorean arthropods (Borges et al. 2005b). This database includes the spatial distribution of all recorded species specimens in a 500 500 m grid, based on both literature and unpublished field data, hence allowing us to obtain the list of species for any region within any of the islands. Here we extracted four different species lists for each taxon, each one of them chosen to correspond to the extent of native forest at four known points in time before and since human colonization (Table 1; Fig. 2 with the island of Terceira as an illustration). They were as follows: a) for the total area of each island, i.e. all known forest specialist species reported from the island. This reflects the near 100% forest cover of the islands before the arrival of humans; AD 1440, herein T1. b) For areas above 300 m, including only those species reported above this elevational limit and corresponding to the extent of the native forest ca AD 1700, T2. c) For areas above 500 m, the extent of the native forest at ca AD 1850, T3. d) for the present area occupied by native forest, including only those species currently reported from native forest remnants within each island, AD 2000, T4. The slight differences in the number of species denoted for (a), (b) and (c) are due to the fact that some species have been recorded only from the lowland areas which have been sequentially lost over time. As Raheem et al. (2009) have recently shown, the influence of pre-fragmentation patterns of species turnover can persist despite habitat loss and fragmentation, with the spatial pattern in species distribution before disturbance persisting to the present. Thus, we avoided considering each island as a priori biogeographically homogeneous before habitat destruction, in terms of species distribution in the different elevational zones considered. The differences between the species number for the total island area (a) and for the current extent of the native forest (d) (Table 2) are due to the inclusion in (a) of historical records of species presences in low and mid altitudes where the native forest is now absent. This means that if a species has been reported in the past from a lowland area where the native forest is now absent and this species is not found in any of the areas currently covered by native forest, the species was included in list (a) but not in list (d). Thus, for this latter category we are not following the simple elevational criterion used for (b) and (c) but we are instead using the actual distribution of the native forest patches. The current area of native forests for all the islands (Table 1) was estimated based on digital aerial photography of the islands and field work (Gaspar 2007, Gaspar et al. 2008).

IBS SPECIAL

ISSUE

288 Table 1. Basic characteristics of the islands of the Azores (main source: Borges and Hortal 2009; see also Methods). Latitude and longitude refer to the centre of the island, and are given in decimal degrees. Total area of the island approximates the forest cover before the arrival of humans; AD 1440, T1; area above 300 m corresponds to the extent of the native forest ca AD 1700, T2; area above 500 m, the extent of the native forest ca AD 1850, T3; and the present area of forest remnants is for AD 2000, T4. : absence of native forest; *currently there is no primary native forest on Graciosa and Corvo Islands. On Graciosa only a very small patch of secondary native vegetation occurs; this patch is dominated by small-sized Erica azorica, an early successional endemic shrub. Island Graciosa Corvo Santa Maria Faial Sa˜o Jorge Sa˜o Miguel Pico Flores Terceira Total

Latitude o N

Longitude o W

Altitude (m)

Total area of island (km2), T1

Area above 300 m (km2), T2

Area above 500 m (km2), T3

Present area of forest remnants (km2), T4

Maximum age (Ma)

39.0 39.4 36.9 38.6 38.7 37.7 38.5 39.4 38.7

27.6 31.0 25.1 28.5 27.9 25.5 28.2 30.9 27.2

398 718 587 1043 1053 1103 2351 915 1023

62 17 97 172 246 757 433 142 402 2328

3.48 9.33 13.19 80.45 170.56 352.39 261.66 95.18 177.60 1163.84

5.44 0.21 36.59 90.35 186.02 188.30 52.58 70.09 629.58

* * 0.09 2.26 2.93 3.31 9.52 15.71 23.45 57.27

2.50 0.71 8.12 0.73 0.55 4.01 0.25 2.90 3.52

Table 2. The number of forest-dependent endemic arthropod species in the four different habitat areas, corresponding to the extent of native forest at four known points in time, before and following human colonization (Supplementary material Table S2 and Methods for details). Island

Coleoptera Total area, T1

Graciosa Corvo Flores Faial Pico Sa˜o Jorge Terceira Sa˜o Miguel Santa Maria

2 1 8 4 14 4 11 17 14

Araneae

Area 300 m, Area 500 m, Present area, T2 T3 T4 2 1 7 3 13 4 10 17 13

1 6 3 13 4 9 11 12

6 3 13 4 9 11 12

Total area, T1 3 0 11 8 10 11 11 11 7

Hemiptera

Area 300 m, Area 500 m, T2 T3 2 0 11 8 10 11 11 10 7

0 11 7 10 11 11 9 6

Present area, T4

Total area, T1

10 7 10 11 10 9 6

3 2 5 5 4 6 8 6 3

Area 300 m, Area 500 m, Present area, T2 T3 T4 1 2 5 5 4 6 7 5 3

2 4 5 4 6 7 5 3

3 3 4 4 5 5 3

richness of the endemic taxa. We also tested the predictive accuracy of the two species area age models (for the total area and the area above 300 m) by testing the correlation between the observed and the predicted number of species. Finally, in order to evaluate our predictions, we compare the average species abundance per transect (i.e. average number of individuals of archipelagic endemic forestdependent species per transect) of Graciosa Island with the rest of the islands of the archipelago. Currently there is no primary native forest on Graciosa; only a very small patch of secondary native vegetation occurs, dominated by small-sized Erica azorica, an early successional endemic shrub. Hence we predict that the surviving forest-dependent species that are present in several islands will show smaller abundances within transects on Graciosa, indicative of a progressive reduction of their populations towards extinction. All analyses were carried out using STATISTICA 6.1 (StatSoft 2003).

Results For the total island area and the area above 300 m, the species area age model applied was significant (p B0.05) for each of the arthropod taxa considered (Table 3), with most of the explained variance attributable to area. However, for the area above 500 m and the present area covered by native forest, neither the species area age relationships nor the respective species area relationships were statistically significant for any of the three taxa considered (Supplementary material Table S2). We thus used the first two benchmark relationships, for total area ( AD 1440, T1) and area above 300 m ( AD 1700, T2) (Fig. 1 and 2B), to represent the baseline conditions for estimation of current extinction debt. Hence, we used the parameters estimated for the total area of the islands (Pred. 1; Table 4), and that of the area above 300 m (Pred. 2; Table 4) to estimate the number of endemic forest arthropods that â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;shouldâ&#x20AC;&#x2122;â&#x20AC;&#x2122; be present and, by direct comparison with the number of extant species, derive the number of species to go extinct (i.e. the extinction debt) for each taxon (Table 4 and Supplementary material S3). For all three arthropod taxa considered, our results clearly indicate that the majority of the endemic forestdependent species are expected to go extinct in time, especially on those islands on which the native forest has been restricted to small areas, namely Santa Maria, SaË&#x153;o Miguel, SaË&#x153;o Jorge and Faial, or on which it has been totally removed, namely Graciosa and Corvo (Table 1 and 4). Terceira, the island with the largest remnants of native

Taxon/island area Coleoptera (total area) Coleoptera ( 300 m) Araneae (total area) Araneae ( 300 m) Hemiptera (total area) Hemiptera ( 300 m)

Equation LogS 0.915 0.678 LogA 0.076 G LogS 0.383 0.471 LogA 0.116 G LogS 0.979 0.780 LogA 0.026 G LogS 0.318 0.531 LogA 0.067 G LogS 0.060 0.321 LogA 0.007 G LogS 0.088 0.347 LogA 0.016 G

SE intercept 0.288 0.198 0.189 0.238 0.184 0.146

SE bA

SE bG

0.126 0.092 0.170 0.153 0.080 0.067

0.025 0.026 0.03 0.04 0.016 0.019

2.6 2.6 2.6 2.6 2.6 2.6

0.87 0.86 0.79 0.68 0.73 0.82

F-value p-value 20.14 18.78 11.06 6.33 7.96 13.27

B0.01 B0.01 0.01 0.03 0.02 B0.01

289

ISSUE

Table 3. The species area age equations used for predicting extinctions. S: number of forest-dependent archipelagic endemic species; A: area; G: geological age; b: standard error for non-standardized regression coefficients (see Methods for details). The degrees of freedom (DF), F and p-values are also presented. For all the models tested see Supplementary material Table S2.

IBS SPECIAL

assumed a multiple linear relationship between species number (S), area (A) and the geological age of each island (G), ) i.e. Log S b1 b2 Log A b3 G, for the endemic forest-dependent species of Araneae, Hemiptera and Coleoptera. For number of species and area we used the conventional logarithmic transformations (log10) to estimate the equation parameters (Borges and Brown 1999, Borges and Hortal 2009, cf. Rosenzweig 2001). For the particular case of the single island, where the number of Araneae species was zero we used the conventional practice of raising the values for all islands by 0.5. Inclusion of island age (Supplementary material) follows previous theoretical and empirical work showing that age can influence the evolutionary dynamics of oceanic islands, as reflected in levels of endemism (Whittaker et al. 2008, Borges and Hortal 2009). Including island age means that we do not assume that the islands were in a pure â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;ecologicalâ&#x20AC;&#x2122;â&#x20AC;&#x2122; immigration extinction equilibrium prior to human colonization. Instead, the number of endemic forest species prior to human colonization is assumed to be a longer-term outcome of immigration, speciation and extinction dynamics. We calculated our species area age relationships using four different â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;habitat areasâ&#x20AC;&#x2122;â&#x20AC;&#x2122; corresponding to the extent of native forest at four known points in time: AD 1440 (total area), AD 1700 (area above 300 m), AD 1850 (above 500 m) and AD 2000 (current extent) (see above). If â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;relaxationâ&#x20AC;&#x2122;â&#x20AC;&#x2122; of species numbers has not yet taken place or is incomplete (i.e. an extinction debt remains) then the best fitting species area age model will correspond to the remaining area of forest at some past time. However, which â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;past timeâ&#x20AC;&#x2122;â&#x20AC;&#x2122; may not be the same for each taxon due to differences in their ecology and life history. Additionally, we tested the effectiveness of the applied model against a number of different models, e.g. including measures of island elevation, log-transformed age values, and considering quadratic models of geological age, i.e. G G2 (Whittaker et al. 2008). An alternative explanation for the lack of relationship between the current extent of native forest and the number of forest dependent species is that larger islands originally had more species as a consequence of their larger area. Thus, due to their larger species pool, more species would be expected to be found in fragments within larger islands. To test this mechanism we evaluated the relationship between the number of the archipelagic endemic species of the three taxa considered here and the total area of each island and compared its explanatory power with the respective species area age relationship. If larger islands have more species, then the species area model will be the best for the species

IBS SPECIAL

ISSUE

Figure 1. Species area relationships for the endemic forest arthropods of the three groups studied (Coleoptera, Araneae, Hemiptera), for the areas of native forest corresponding to four known points in time (see text). In order to exclude the effect of island age on species richness, for purposes of visual representation we present the relationship between the residuals of the log (species) age relationship, (i.e. geological age-independent richness) against log (area; km2). While the relationships for the total area (AD 1440, T1) and the area above 300 m (AD 1700, T2) were statistically significant for all taxa, for the area above 500 m (AD 1850, T3) and the present area of the native forest (AD 2000, T4) they are not statistically significant for any taxon (see Supplementary material Table S2 for details). Solid lines are regression trend-lines, and dashed lines are 95% confidence intervals. Non-significant relationships are shown here for purposes of comparison.

forest, has the smallest number of predicted future extinctions. The estimated proportion of extinctions per island varies from 50 to 99% for Coleoptera, 60 to 99.5% for Araneae and 49 to 85% for Hemiptera. Amongst the three taxa, Hemiptera are at the lowest overall risk of extinction. The mean predicted percentage of extinctions for all the islands is: Coleoptera, 91.56% (95.68%; Pred. 1) and 74% (915.82%; Pred. 2), Araneae, 94.81% (94.41%; Pred. 1) and 80.81% (910.73%; Pred. 2), and Hemiptera, 68.56% (912.42%; Pred. 1) and 67% (913.06%; Pred. 2). These projections are in accordance with the distribution of the taxa across the island group since the percentage of endemic forest-dependent species present in three or fewer islands is 72% for Coleoptera, 47% for Araneae and 36% for Hemiptera. In the multiple regression models applied, the age parameter was statistically significant only in the case of Coleoptera; hence, when it was excluded from the models applied for spiders and Hemiptera, the predictions remained the same (without any statistically significant difference for the values presented). However, we applied the species area age model in all cases for purposes of comparison (Table 3). Note that this does not affect the statistical significance of the relationships used, i.e. the relationships estimated based on the area above 500 m and the current area of the native forest remain statistically 290

non-significant even when only area is considered (Supplementary material Table S2), and the calculated parameters remain statistically indistinguishable for the cases where age has no significant contribution (Supplementary material Table S2). Additionally, the models we report were always better, based on the adjusted R2 values and the Akaikeâ&#x20AC;&#x2122;s information criterion values (AIC), than were models considering elevation or quadratic age (results not shown). The species area model for the archipelagic endemic species was the best model (i.e. lower value of AIC) only for Araneae (see Alternative mechanism in Supplementary material and Table S4), indicating that at least for Coleoptera and Hemiptera, the hypothesis that larger islands have more species, independent of the current area of the native forests, can be ruled out. The general pattern arising from the cross-checking of the predictive accuracy of the two species area age models used (Supplementary material Table S5) demonstrates that using the parameter estimations from the species area age model of the areas 300 m over-predicts the number of species that are present when applied to the total area of the islands, while the use of the parameters arising from the species area age model for the total area leads to an underestimation of the species present in areas above 300 m (Supplementary material Table S5 and further discussion in the Supplementary material). In all cases

72 70 57 58 58 54 77 55 80 54 49 79 76 85 84 0.90 0.84 2.30 1.13 1.79 1.21 2.96 1.43 0.48 0.83 0.86 2.11 1.13 1.80 1.23 2.40 1.28 0.46 3 2 5 5 4 6 8 6 3 96 76 91 74 97 83 94 83.5 98 92 86 60 97 85 99.5 93 0.71 0.54 2.90 0.83 1.65 0.93 4.42 1.69 0.47 0.12 0.11 1.02 0.21 0.62 0.25 1.52 0.34 0.03 3 0 11 8 10 11 11 11 7 90.5 59.5 86 50 86 67 94 81 96 91 93 80 82.5 57.5 97 88 99 92 0.81 0.50 2.69 0.75 1.29 0.81 4.68 2.11 1.17 0.19 0.14 1.15 0.25 0.59 0.28 1.92 0.56 0.10 2 1 8 4 14 4 11 17 14 Graciosa Corvo Flores Faial Pico Sa˜o Jorge Terceira Sa˜o Miguel Santa Maria

Species loss (%) Pred. 2 ( 300) Pred. 1 (ALL) Hemiptera Species loss (%) Pred. 2 ( 300) Pred. 1 (ALL) Araneae Species loss (%) Pred. 2 ( 300) Pred. 1 (ALL) Coleoptera

ISSUE

Brook et al. (2003), studying a wide range of terrestrial and freshwater taxa from Singapore, inferred that 34 87% of species identified as forest specialists had gone extinct following deforestation in Singapore. They referred to these as catastrophic extinctions and warned that 13 42% of regional populations in south east Asia will be lost over the next century due to habitat loss, in the absence of remedial action. Our estimates for the magnitude of the extinction debt among forest-dependent endemic arthropods in the Azores are even higher than these startling figures and suggest that more than half of the extant species might eventually be driven to extinction due to habitat loss; a habitat loss which is almost complete ( 95% of the original extent of the native forest) and has occurred in B600 yr. The severity of the deforestation, both in terms of the spatial extent and the temporal scale, has clearly reduced

IBS SPECIAL

Discussion

Island

the Durbin-Watson test, applied to detect the presence of autocorrelation, indicates that the residuals are not positively autocorrelated, except for the Araneae 300 m dataset, for which the test is not conclusive (Supplementary material Table S6) and the coefficient of determination (R2) of the relationship between observed and predicted number of species (log-transformed values) was higher than 0.65. The results of the comparison of species average abundance on Graciosa Island with the rest of the islands, where native forest still exists, clearly indicate that for the clear majority of the eight species for which available data exists, there is a clear pattern of lower abundances in Graciosa Island (Supplementary material Table S7).

Table 4. Predicted extinctions. Number of forest-dependent archipelagic endemic arthropods of Coleoptera, Araneae and Hemiptera for the nine Azorean Islands and the respective predicted number of species that should be found based on the species area age models calculated using the total area of each island (Pred. 1) and the area of each island above 300 m (i.e. area occupied by native forest ca 300 yr ago; Pred. 2). Currently there is no native forest on Graciosa and Corvo Islands. The lower and upper bound of 95% confidence limits for both predicted responses are presented in Supplementary materal Table S3.

Figure 2. The sequential reduction of the native forest and the respective species area relationships. (A) The elevational distribution of native forest in historical times for the island of Terceira (Azores; using Atlantis Tierra 2.0 software and Silveira 2007). Red (total area, T1): before human occupation, (almost complete coverage of island’s area); orange (area 300 m, T2): ca 300 yr ago (300 500 m); yellow (area 500, T3): ca 160 yr ago (above 500 m); green (present area, T4): current distribution. (B) A schematic representation of the effects of the sequential reduction of the native forest on the species area relationships of endemic forest arthropods. The dashed line in T4 represents the future species area relationships, extrapolated from T1 and T2 (see text). The magnitude of the extinction debt is represented by the difference between current species richness (solid green line) and the future predictions (dashed lines).

291

ISSUE

IBS SPECIAL

the opportunities for forest-dependent species to cope with the changes in their environment. At face value, these figures constitute a powerful warning to island conservationists that the worst of the extinction crisis is by no means over. Furthermore, in spite of the fact that some archipelagic endemic species may benefit from a degree of population reinforcement between habitat fragments or islands (see also Borges et al. 2008), the parallel reduction of the native forest across all islands in the last 600 yr has greatly diminished the probability of such source-sink dynamics rescuing species from global extinction. Hence, we would also anticipate a correspondingly large number of archipelagic-scale species extinctions for Azorean endemic arthropods in the future as the extinction debt is settled. Amongst the three studied taxa, our analyses suggest that Araneae and Coleoptera are at greater risk of extinction per island, compared to Hemiptera. This may be partially related to the ecological characteristics and requirements of the species in each group, with Hemiptera typically exhibiting higher dispersal abilities and having a smaller proportion of species endemic to a single island (SIE: 6%). In contrast, both Araneae and Coleoptera have high proportions of SIEs, 19.4 and 18.9% respectively. Additionally, spiders, the most important arthropod predators in the Azores, are expected to be relatively intolerant to the destruction and disturbance of natural forests on these islands (Cardoso et al. 2007, 2010) as shown for other high trophic level taxa (Whittaker and Fernańdez-Palacios 2007). We recognise that other processes may be involved in the extinctions to come apart from habitat loss, but at the same time these area-based models can offer an effective descriptor of the combined effects of other causes (see also Hanski et al. 2007, Yaacobi et al. 2007). One such additional factor is undoubtedly the significant pressure exerted by exotic species (Blackburn et al. 2004, Whittaker and Fernańdez-Palacios 2007), which already comprise 58% of the total Azorean arthropod fauna (68% of Araneae, 60% of Coleoptera and 47% of Hemiptera, Borges et al. 2005b, 2006). The figures that we report here are likely to be more accurate than previous predictions because we have focused our attention on endemic forest species that have evolved in and are only found in association with the native forest. Endemic forest dependent species are unlikely to show a range expansion to anthropogenic habitats under land-use changes. Hence, we avoid additional ‘‘noise’’ caused by generalist species that may well be able to survive in other (i.e. anthropogenic) habitats. For example, there is no evidence that the endemic forest arthropods on Terceira can establish viable populations within other forest or vegetation types on the island (Borges and Wunderlich 2008, Borges et al. 2008, see also Methods). Furthermore, we base our predictions on two baseline curves, and not on a single one as usually applied, an approach providing fairly conservative estimates of the present extinction debt, taking into account the crude but reasonably well-founded habitat distributional data available. However, it should also be recognised that the projected extinctions arising from the use of the species area models involve several uncertainties (May et al. 1995, Lewis 2006, Vellend et al. 2006, Whittaker and Fernańdez-Palacios 2007, Kuussaari et al. 292

2009, Ladle 2009) and can never completely replace species-level assessments for the identification of extinction threat (Kotiaho et al. 2005, Whittaker et al. 2005, Kuussaari et al. 2009). Nevertheless, for many species of conservation concern the collection of appropriately detailed information is an unrealistic target. It is therefore important that we develop more realistic indirect measures and theoretical projections of extinctions, based on as pragmatic a set of assumptions as possible (Heywood et al. 1994, May et al. 1995, Whittaker et al. 2005). Here, by using taxon-specific z-values derived from species area relationships of the same taxon in the same island system, we would argue that our extinction estimates are likely to prove more realistic and robust than previous analyses (see Yaacobi et al. 2007 for a similar example on habitat islands). It is highly probable that since the original settlement of humans on the Azores a number of arthropods and other poorly known taxa have already become extinct due to deforestation (cf. Brook et al. 2003, Hanski et al. 2007, Cardoso et al. 2010). Thus, given that a large fraction of the island’s forest had already been cleared before the first reliable standardized sampling (Borges et al. 2005a, 2006, 2008, Ribeiro et al. 2005, Gaspar 2007, Gaspar et al. 2008, Borges and Wunderlich 2008), the extinction of species most sensitive to disturbance probably went unrecorded (Cardoso et al. 2010). In point of fact, at least five SIE beetle species (Bradycellus chavesi, Calathus extensicollis, Calathus vicenteorum, Nesotes azorica, Ocydromus derelictus), recorded early in the 20th century, have not been recorded since 1965 and might therefore be considered extinct (Borges et al. 2000). Moreover, many other SIEs are extremely rare and under threat (Borges et al. 2006), and are particularly scarce in standardized samples (Supplementary material Table S1 for Terceira Island). While seven individuals of Calathus lundbladi, an endemic species of Sa˜o Miguel, were found in four traps during 1989, just one individual was collected in 120 traps in the 1999 2000 survey (Borges et al. 2005a). The case of Graciosa Island is in accord with the above (Supplementary material Table S7); although species abundance responses to forest loss and fragmentation can be strikingly idiosyncratic (Fahrig 2001), and phenomena like density compensation as a result of the extinction of competitors and/or predators cannot be excluded (Whittaker and Ferna´ndez-Palacios 2007; Supplementary material Table S7), the very small fragment of secondary native vegetation in Graciosa, which is highly disturbed, can be considered as the ‘‘last refuge’’ for the endemic forest-dependent species on that island. These species are already on an ecological trajectory towards extinction. Although, it is possible that some forest specialist species might be able to find a refuge in exotic forests (Supplementary material Table S1), the durability and viability of these populations are probably limited (Borges, unpubl.). Conclusively proving the extinction of a small arthropod species will be practically impossible within such a large area as the Azorean archipelago (2328 km2), but we concur with others (Hanski et al. 2009, Ladle 2009), that given the great importance of understanding the processes and rates of species extinctions, analyses based on indirect evidence can be informative.

archipelagos that have experienced anthropogenic habitat loss (Mueller-Dombois and Fosberg 1998, Rolett and Diamond 2004, Steadman 2006) and where the temporal sequence of habitat loss can be at least crudely estimated. Acknowledgements KAT, PAVB, RG and RJW designed the research, PAVB, CG, FD, LMAS, RG, CM, AMCS, IRA, PC, SPR, JH, ARMS, JAQ gathered the data, KAT, PAVB, EM, RJW and PC analysed the data, KAT, RJL, JH, PC, PAVB and RJW wrote the paper. All authors discussed the results and commented on the manuscript. We thank G. Mace, V. Brown, J. Sadler, S. Bhagwat, J. Lobo, A. Jime´nez-Valverde, A. Parmakelis, S. Sfenthourakis, S. Meiri, attendees of the 2009 International Biogeography Society meeting in Merida, and especially Albert Phillimore and Andy Purvis for discussions and comments on previous drafts. We also thank Helmut Hillebrand, Robert Dunn and two anonymous referees for valuable comments on the manuscript. KAT was supported in this work by a Marie Curie Intra-European Fellowship Program (project ‘‘SPAR’’, 041095) held in the OUCE, by a FCT Fellowship (SFRH/BPD/44306/ 2008) and from the Academic Visitors Program of the NERC Centre for Population Biology. PAVB and RG worked on this project under the DRCT project M2.1.2/I/017/2007 and the EU projects INTERREGIII B ‘‘ATLAˆNTICO’’ (2004 2006) and BIONATURA (2006 2008).

References

293

ISSUE

Blackburn, T. M. et al. 2004. Avian extinction and mammalian introductions on oceanic islands. Science 305: 1955 1958. Borges, P. A. V. and Brown, V. K. 1999. Effect of island geological age on the arthropod species richness of Azorean pastures. Biol. J. Linn. Soc. 66: 373 410. Borges, P. A. V. and Wunderlich, J. 2008. Spider biodiversity patterns and their conservation in the Azorean archipelago, with descriptions of new species. Syst. Biodivers. 6: 249 282. Borges, P. A. V. and Hortal, J. 2009. Time, area and isolation: factors driving the diversification of Azorean arthropods. J. Biogeogr. 36: 178 191. Borges, P. A. V. et al. 2000. Ranking the Azorean natural forest reserves for conservation using their endemic arthropods. J. Insect Conserv. 4: 129 147. Borges, P. A. V. et al. 2005a. Ranking protected areas in the Azores using standardised sampling of soil epigean arthropods. Biodivers. Conserv. 14: 2029 2060. Borges, P. A. V. et al. 2005b. A list of the terrestrial fauna (Mollusca and Arthropoda) and flora (Bryophyta, Pteridophyta and Spermatophyta) from the Azores. Direcc¸a˜o Regional de Ambiente and Univ. Azores. Borges, P. A. V. et al. 2006. Invasibility and species richness of island endemic arthropods: a general model of endemic vs. exotic species. J. Biogeogr. 33: 169 187. Borges, P. A. V. et al. 2008. Insect and spider rarity in an oceanic island (Terceira, Azores): true rare and pseudo-rare species. In: Fattorini, S. (ed.), Insect ecology and conservation. Research Signpost, Kerala, India, pp. 47 69. Brook, B. W. et al. 2003. Catastrophic extinctions follow deforestation in Singapore. Nature 424: 420 423. Brooks, T. M. et al. 1997. Deforestation predicts the number of threatened birds in insular southeast Asia. Conserv. Biol. 11: 382 394. Brooks, T. M. et al. 2002. Habitat loss and extinction in the hotspots of biodiversity. Conserv. Biol. 16: 909 923. Brooks, T. M. et al. 2006. Global biodiversity conservation priorities. Science 313: 58 61.

IBS SPECIAL

Precise estimation of the time to extinction of each species under threat remains an unrealistic aim, for it will vary from island to island and from species to species. The scarce available information suggests that delayed extinctions are more likely to occur in species with longer generation times, e.g. mammals as opposed to insects, (see review in Kuussaari et al. 2009), but recent studies on invertebrates (Raheem et al. 2009, Sodhi et al. 2009) have shown a resilience of some invertebrate species to the effects of forest loss; with many species requiring only very small areas to persist for extended periods (see also discussion in Samways 2006). These results suggest a need for caution in generalizing about relaxation and species loss based on data for ecologically different taxa, such as vertebrates and especially birds. Despite the extensive destruction of the Azorean native forest, the remaining network of patches within some of the islands and the overall remaining area in the archipelago might be sufficient for delaying relaxation for long periods of time or even sustain viable populations for some species. Hence, the time lag may be considerable, even for invertebrates of short life cycles. We conclude that large-scale conservation efforts need to be implemented if the high extinction debt we have identified is to be deferred or avoided. Human-induced fragmentation, land-use changes and invasive species have already been identified as important threats to Azorean biodiversity (Martins 1993, Borges et al. 2000, 2006, Borges and Wunderlich 2008). Our analyses strongly reinforce this message: the conservation of the Azorean natural heritage, and that of many other oceanic islands, will largely depend on establishing an integrated large-scale strategy to manage both indigenous and non-indigenous species while simultaneously protecting the remnants of native habitat (i.e. forest in the Azorean context) and, ideally, increasing their extent. This point is corroborated by the case of the Azorean bullfinch Pyrrhula murina, an endemic passerine bird species confined to eastern Sa˜o Miguel and living almost exclusively in the laurel forest. The species, locally abundant in the second half of 19th and early 20th century, has suffered through widespread loss of native forest and invasion by exotic vegetation, which has largely overrun the remaining patches of natural vegetation within the bullfinch’s breeding range. This led to a dramatic decline, to B100 individuals, in the late 1970s. Following the implementation in 2003 of a five-year LIFE-Nature project, a central objective of which was to increase the habitat of the Azores bullfinch, mainly through promoting the regeneration of the laurel forest and the control of the exotic flora (Ramos 1996, 2005, Guimara˜es and Olmeda 2008), the population had increased to an estimated 400 pairs by the year 2006 (Guimara˜es and Olmeda 2008). In the absence of focused and well-resourced interventions, the legacy of past and current deforestation on oceanic islands will be an inexorable process of biodiversity loss stretching well into the future. Many extant species may already have passed crucial thresholds of population size and/or genetic diversity that typically precede extinction, meaning that the species are becoming highly sensitive to demographic and environmental stochasticity (Schoener et al. 2003). The approach to estimating extinction debt outlined in this work may be suitable for application to many other analogous systems, including numerous oceanic

ISSUE

IBS SPECIAL

Cardoso, P. et al. 2007. Biotic integrity of the arthropod communities in the natural forests of Azores. Biodivers. Conserv. 16: 2883 2901. Cardoso, P. et al. 2010. Drivers of diversity in Macaronesian spiders and the role of species extinctions. J. Biogeogr. 37, doi: 10.1111/j.1365-2699.2009.02264.x. Darlington, P. J. 1957. Zoogeography: the geographical distribution of animals. Wiley. Diamond, J. M. 1989. The present, past and future of humancaused extinctions. Phil. Trans. R. Soc. B 325: 469 477. Droue¨t, H. 1859. Coleópte`res Açoreéns. Rev. Mag. Zool. 11: 243 259. Dunn, R. R. 2005. Modern insect extinctions, the neglected majority. Conserv. Biol. 19: 1030 1036. Fahrig, L. 2001. How much habitat is enough? Biol. Conserv. 100: 65 74. Fonseca, R. C. 2010. The silent mass extinction of insect herbivores in biodiversity hotspots. Conserv. Biol. 23: 1507 1515. Frutuoso, G. 1963. The sixth book on longing for the land. Inst. Cultura de Ponta Delgada. Gabriel, R. and Bates, J. W. 2005. Bryophyte community composition and habitat specificity in the natural forests of Terceira, Azores. Plant Ecol. 177: 125 144. Gaspar, C. S. 2007. Arthropod diversity and conservation planning in native forests of the Azores archipelago. Ph.D. thesis, Univ. Sheffield. Gaspar, C. et al. 2008. Diversity and distribution of arthropods in native forests of the Azores archipelago. Arquipe´lago-Life Mar. Sci. 25: 1 30. Guimara˜es, A. and Olmeda, C. 2008. Management of Natura 2000 habitat. 9360 *Macaronesian laurel forests (Laurus, Ocotea). European Commission. Hanski, I. et al. 2007. Deforestation and apparent extinctions of endemic forest beetles in Madagascar. Biol. Lett. 3: 344 347. Hanski, I. et al. 2009. Deforestation and tropical insect extinctions. Biol. Lett. 5: 653 655. Helm, A. et al. 2006. Slow response of plant species richness to habitat loss and fragmentation. Ecol. Lett. 9: 72 77. Heywood, V. H. et al. 1994. Uncertainties in extinction rates. Nature 368: 105. IUCN 2009. IUCN Red List of Threatened Species. Version 2009.2. <www.iucnredlist.org>. Kotiaho, J. S. et al. 2005. Predicting the risk of extinction from shared ecological characteristics. Proc. Nat. Acad. Sci. USA 102: 1963 1967. Kuussaari, M. et al. 2009. Extinction debt: a challenge for biodiversity conservation. Trends Ecol. Evol. 24: 564 571. Ladle, R. J. 2009. Forecasting extinctions: uncertainties and limitations. Diversity 1: 133 150. Lande, R. 1993. Risks of population extinction from demographic and environmental stochasticity and random catastrophes. Am. Nat. 142: 911 927. Lewis, O. T. 2006. Climate change, species area curves and the extinction crisis. Phil. Trans. R. Soc. B 361: 163 171. MacArthur, R. H. and Wilson, E. O. 1967. The theory of island biogeography. Princeton Univ. Press. Martins, A. M. F. 1993. The Azores westernmost Europe: where evolution can be caught red-handed. Bol. Mus. Municipal Funchal S2: 181 198. May, R. M. et al. 1995. Assessing extinction rates. In: Lawton, J. H. and May, R. M. (eds), Extinction rates. Oxford Univ. Press, pp. 1 24.

Download the Supplementary material as file E6203 from <www.oikos.ekol.lu.se/appendix>.

294

McDonald, K. A. and Brown, J. H. 1992. Using montane mammals to model extinctions due to global change. Conserv. Biol. 6: 409 415. Mueller-Dombois, D. F. and Fosberg, F. R. 1998. Vegetation of the tropical Pacific islands. Springer. Paulay, G. 1994. Biodiversity on oceanic islands: its origin and extinction. Am. Zool. 34: 134 144. Pimm, S. L. and Askins, R. A. 1995. Forest losses predict bird extinctions in eastern North America. Proc. Nat. Acad. Sci. USA 92: 9343 9347. Pimm, S. L. et al. 1995. The future of biodiversity. Science 269: 347 350. Preston, F. W. 1962. The canonical distribution of commonness and rarity: part I. Ecology 43: 185 215. Raheem, D. C. et al. 2009. Fragmentation and pre-existing species turnover determines land-snail assemblages of tropical rain forest. J. Biogeogr. 36: 1923 1938. Ramos, J. A. 1996. Introduction of exotic tree species as a threat to the Azores bullfinch population. J. Appl. Ecol. 33: 710 722. Ramos, J. A. 2005. The Priolo and the natural altitude forest, 2nd ed. Caˆmara Municipal de Nordeste. Ribeiro, S. P. et al. 2005. Canopy insect herbivores in the Azorean laurisilva forests: key host plant species in a highly generalist insect community. Ecography 28: 315 330. Rolett, B. and Diamond, J. 2004. Environmental predictors of pre-European deforestation on Pacific islands. Nature 431: 443 446. Rosenzweig, M. L. 2001. Loss of speciation rate will impoverish future diversity. Proc. Nat. Acad. Sci. USA 98: 5404 5410. Samways, M. J. 2006. Insect extinctions and insect survival. Conserv. Biol. 20: 245 246. Schoener, T. W. et al. 2003. Life-history models of extinction: a test with island spiders. Am. Nat. 162: 558 573. Silveira, L. M. A. 2007. Learning with history: interaction with nature during the human colonization in Terceira Island. M.Sc. thesis, Univ. Azores. Sodhi, N. S. et al. 2009. Insect extinctions on a small denuded Bornean island. Biodivers. Conserv. 19: 485 490. StatSoft 2003. STATISTICA (data analysis software system), version 6.1. StatSoft, Tulsa, OK. Steadman, D. W. 2006. Extinction and biogeography of tropical Pacific birds. Univ. Chicago Press. Thomas, C. D. et al. 2004. Extinction risk from climate change. Nature 427: 145 148. Tilman, D. et al. 1994. Habitat destruction and the extinction debt. Nature 371: 65 66. Vellend, M. et al. 2006. Extinction debt of forest plants persists for more than a century following habitat fragmentation. Ecology 87: 542 548. Whittaker, R. J. and Ferna´ndez-Palacios, J. M. 2007. Island biogeography: ecology, evolution, and conservation, 2nd ed. Oxford Univ. Press. Whittaker, R. J. et al. 2005. Conservation biogeography: assessment and prospect. Divers. Distrib. 11: 3 23. Whittaker, R. J. et al. 2008. A general dynamic theory of oceanic island biogeography. J. Biogeogr. 35: 977 994. Yaacobi, G. et al. 2007. Habitat fragmentation may not matter to species diversity. Proc. R. Soc. B 274: 2409 2412. Zurita, N. A. and Arechavaleta, M. 2003. Database of the biodiversity of the Canary Islands (Banco de datos de Biodiversidad de Canarias). Bol. Soc. Entomol. Aragonesa 32: 293 294.

Ecography 33: 295 303, 2010 doi: 10.1111/j.1600-0587.2010.06279.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Robert K. Colwell. Accepted 18 January 2010

Going against the flow: potential mechanisms for unexpected downslope range shifts in a warming climate Jonathan Lenoir, Jean-Claude Ge´gout, Antoine Guisan, Pascal Vittoz, Thomas Wohlgemuth, Niklaus E. Zimmermann, Stefan Dullinger, Harald Pauli, Wolfgang Willner and Jens-Christian Svenning J. Lenoir (lenoir.john@gmail.com), The Ecoinformatics & Biodiversity Group, Dept of Biological Sciences, Aarhus Univ., Ny Munkegade 114, DK-8000 Aarhus C, Denmark and AgroParisTech, UMR1092 AgroParisTech-INRA, Laboratoire d’Etude des Ressources Foreˆt-Bois (LERFoB), 14 rue Girardet, FR-54000 Nancy, France. J.-C. Ge´gout, AgroParisTech, UMR1092 AgroParisTech-INRA, Laboratoire d’Etude des Ressources Foreˆt-Bois (LERFoB), 14 rue Girardet, FR-54000 Nancy, France, and Center for Advanced Studies in Ecology and Biodiversity (CASEB), Depto de Ecologia, Pontificia Univ., Cato´lica de Chile, Alameda 340 C.P. 6513677, Santiago, Chile. A. Guisan, Fac. of Biology and Medicine, Dept of Ecology & Evolution, Univ. of Lausanne, Baˆtiment Biophore, CH-1015 Lausanne, Switzerland. P. Vittoz, Fac. of Biology and Medicine, Dept of Ecology & Evolution, Univ. of Lausanne, Baˆtiment Biophore, CH-1015 Lausanne, Switzerland and Fac. of Geosciences and Environment, Univ. of Lausanne, Baˆtiment Biophore, CH-1015 Lausanne, Switzerland. T. Wohlgemuth and N. E. Zimmermann, Swiss Federal Research Inst. for Forest Snow and Landscape Research WSL, Zu¨rcherstrasse 111, CH-8903 Birmensdorf, Switzerland. S. Dullinger, Vienna Inst. for Nature Conservation and Analyses, Giessergasse 6/7, AT-1090 Vienna, Austria, and Fac. Centre for Biodiversity, Dept of Conservation Biology, Vegetation and Landscape Ecology, Univ. of Vienna, Rennweg 14, AT-1030 Vienna, Austria. H. Pauli, Inst. of Mountain Research: Man and the Environment (IGF) of the Austrian Academy of Sciences, c/o Fac. Centre for Biodiversity, Dept of Conservation Biology, Vegetation and Landscape Ecology, Univ. of Vienna, Rennweg 14, AT-1030 Vienna, Austria. W. Willner, Vienna Inst. for Nature Conservation and Analyses, Giessergasse 6/7, AT-1090 Vienna, Austria. J.-C. Svenning, The Ecoinformatics & Biodiversity Group, Dept of Biological Sciences, Aarhus Univ., Ny Munkegade 114, DK-8000 Aarhus C, Denmark.

295

ISSUE

However, most of the studies that reported expected range shifts towards higher elevations have detected species moving towards lower elevations as well. Table 1 provides an illustrative, non-comprehensive survey of such studies from the recent years: in summary they demonstrate that ca 65% of the species have shifted their mid-range positions upslope, 10% have not changed their mid-range positions, and 25% have shifted their mid-range positions downslope (Table 1). In addition, according to a global review of the literature published until the beginning of the 21st century (Parmesan and Yohe 2003) ca 20% of the species have adjusted their ranges towards lower elevations and/or southern latitudes. Hence, a considerable fraction of the investigated species has shown range shifts that are inconsistent with the forecasted effects of climate warming. These downslope movements seem very unlikely to occur as a direct consequence of rising temperatures, but the potential mechanisms involved have received little attention. Stochastic fluctuations in the positions of individuals, or populations, together with measurement errors, represent one such potential ‘‘mechanism’’. However, many, though not all of the studies reporting downslope shifts have explicitly tested the observed changes in single species’ ranges for significant deviation from random fluctuations. For plants, significant downslope shifts have been reported for 5 of 46 species displaying significant mid-range shifts between the periods 1905 1985 and 1986 2005 (Table S2

IBS SPECIAL

In as much as the elevation gradient in species composition is often thought to be driven by the corresponding temperature gradient, species ranges are both expected and predicted to shift upward in response to climate warming. Indeed, there are numerous reports of species moving towards higher elevations in response to the rising temperatures for both animals (Konvicka et al. 2003, Tryjanowski et al. 2005, Wilson et al. 2005, Franco et al. 2006, Hickling et al. 2006, Moritz et al. 2008, Raxworthy et al. 2008, Chen et al. 2009) and plants (Klanderud and Birks 2003, Walther et al. 2005, Pauli et al. 2007, Kelly and Goulden 2008, Lenoir et al. 2008, Parolo and Rossi 2008, Vittoz et al. 2008, Lenoir et al. 2009), and the evidence for significant upslope migrations now seems overwhelming regardless of the position along latitudinal (Klanderud and Birks 2003, Konvicka et al. 2003, Wilson et al. 2005, Raxworthy et al. 2008, Chen et al. 2009) or elevational (Walther et al. 2005, Pauli et al. 2007, Kelly and Goulden 2008, Lenoir et al. 2008, Vittoz et al. 2008, Lenoir et al. 2009) gradients. Due to this empirical evidence and, perhaps, the intuitive expectation of rising elevational ranges as a consequence of a warming climate, ecologists have primarily focused on elaborating on the mechanisms and consequences of such upslope shifts including (Colwell et al. 2008): 1) biotic attrition in the lowland tropics, 2) gaps between current and projected elevational ranges (range-shift gaps), and 3) mountaintop extinctions in the long-term.

Raxworthy et al. (2008) Franco et al. (2006) Hickling et al. (2006) Wilson et al. (2005) Archaux (2004) Konvicka et al. (2003) 65 m per decade 52 m between periods 10 m per decade 119 m between periods 2 m between periods 60 m between periods 20% 33% 31% 26% 15% 22% 7% 0% 0% 0% 76% 0% 73% 67% 69% 74% 10% 76% 1993 2003 1970 1990 and 2004 2005 1960 2000 1967 1973 and 2004 1970s and 2000s 1951 1980 and 1995 2001 Animal Butterfly Animal Butterfly Bird Butterfly

30 3 329 19 41 119

65 m between periods 31% 0% 69%

The Vosges, Jura, Massif Central Pyrenees, Alps and Corsican mountains, France Tsaratanana mountains, Madagascar England England Sierrade Guadarrama, central Spain Alps, France Czech Republic 1905 1985 and 1986 2005 Plant

171

12% 0% 59% 90% Mount Kinabalu, Borneo Santa Rosa mountains, south California Moth Plant

1965 and 2007 1977 1978 and 2006 2007

102 10

Downward shifts No shifts Upward shifts

29% 10%

68 m between surveys 65 m between periods

Chen et al. (2009) Kelly and Goulden (2008) Lenoir et al. (2008)

Author references Mean shift of the central position Percentage of species Number of species Geographic locations Sampling periods Species groups

Table 1. Studies that have documented changes of the mid-range positions (i.e. mean, median or optimum) of speciesâ&#x20AC;&#x2122; elevational distributions in relation to current climate change. Only studies comparing at least two inventories are listed here. This list is informal and not intended to be exhaustive.

ISSUE

IBS SPECIAL 296

in Lenoir et al. (2008)). For reptiles and amphibians, Raxworthy et al. (2008) have found significant downslope shifts for 2 of 7 species displaying significant mid-range shifts between 1993 and 2003. For birds, Archaux (2004) has reported significant downslope shifts for 5 of 8 species displaying significant mid-range shifts between the 1970s and the 2000s. However, it could be argued that starting from the nearconsensual report of a general and strong upward shift signal in species mid-range positions along the elevation gradient, the relevant random distribution of species mid-range shifts for testing the occurrence of significant downslope movements should be centred around some expected upslope shift, according to the observed climatic trends, rather than around zero (i.e. a null expectation of constant distributions). Indeed, the relevant question is: which proportion of species would we expect to shift their mid-range position downslope by chance alone (i.e. due to random population fluctuations, data idiosyncrasies, and observer errors) despite an average upslope trend of z m? This proportion is highly dependent on the variation in mid-range shifts caused by such â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;stochastic processesâ&#x20AC;&#x2122;â&#x20AC;&#x2122; alone (Fig. 1): the narrower this variability, the lower the proportion of species likely to show such stochastic downslope shifts. However, if random fluctuations and observer errors do not fully explain the observed downslope shifts, what mechanisms might then drive these unexpected range changes? One potential mechanism is land-use-related habitat modification, which has already been shown to cause downslope range shifts in particular settings (HaÂ¨ttenschwiler and KoÂ¨rner 1995, Archaux 2004). In addition, however, such shifts may also result from changes in species interactions (facilitation, competition, and predation) in response to climate warming (Hughes 2000). Here, we propose a conceptual and testable model that explains the observed downslope shifts of species as resulting from the effects that both climate and land use change, separately or in concert, might have on species interactions. We start from the proposition that species are often limited by physical stresses at one margin, but by biotic interactions at the other, more favourable, margin of their distribution along environmental gradient (McArthur 1972, Connell 1978, Brown et al. 1996, Brown and Lomolino 1998, Leathwick and Austin 2001, Normand et al. 2009). We then argue that both climate warming and land-use-related habitat modification may increase levels of disturbance in these ecosystems leading to: 1) a transient reduction of the importance of competition as a limiting factor on species distributions; and 2) an associated potential range expansion towards lower elevations for species whose lower elevation margin was previously strongly limited by competition.

Climate warming may cause temporary downslope range shifts due to transient competitive release at the lower margin of species distribution Climate warming may not only affect species distributions via altering abiotic conditions but also by changing the importance or intensity of species interactions (Hughes

Figure 1. Null expectation of the proportion of downslope midrange shifts despite an average upslope trend of z m for (a) narrower range of random variation and (b) wider range of random variation. Broken curves represent the observed distribution of shifts in species mid-range elevation from a colder to a warmer period. Unbroken curves represent the random distribution in species mid-range elevation within a single statistical population. The gray filling illustrates the expected proportion of random downslope shifts in mid-range positions in the absence of climate change (50%). The black filling illustrates the expected proportion of random downslope shifts in mid-range positions despite a warming-driven upslope trend of z m (B50%). The vertical broken line displays the average upslope trend of z m.

ISSUE

297

IBS SPECIAL

2000, Brooker 2006). Indeed, it has been demonstrated that changes in species interactions may override autecological responses to a changing climate and even reverse community trajectories (Suttle et al. 2007). Before explaining our model which is based on such changes in species interactions, we would like to pinpoint the two most important conceptual cornerstones it rests upon. First of all, the lower-margin-competition versus highermargin-stress limitation concept (McArthur 1972, Connell 1978, Brown and Lomolino 1998) suggests competitive effects as the major mechanism for setting the lower limit of speciesâ&#x20AC;&#x2122; elevational ranges. More generally, the importance of competition is thought to increase as environmental severity decreases, a theory known as the stress-gradient hypothesis in plant ecology (Bertness and Callaway 1994, Callaway and Walker 1997, Brooker et al. 2005, Maestre et al. 2009). This concept might be less applicable towards the lowest elevations of the lowest latitudes in subtropical to tropical areas, where temperature and water stress vary in

opposite directions, and where lower range margins may be relatively often set by drought-induced stress (Normand et al. 2009). However, the stress-gradient hypothesis appears generally valid in elevational ranges where the macroclimate does not present drought-induced stress gradients that complicate the effect of the elevational gradient (Callaway 1998, Callaway et al. 2002). Thus, this concept will apply to many mountainous regions, notably in moist temperate and tropical areas. As a corollary, many species in such systems will likely be characterized by realised climatic niches being smaller than fundamental ones (Vetaas 2002) in particular towards lower elevations (due to biotic interactions). For example, Vetaas (2002) has suggested that competition plays an important role in constraining the climatic ranges of four Himalayan Rhododendron species in their native habitat as compared to their climatic ranges in ornamental gardens and arboreta, three of which being able to grow there under a far wider range of climatic conditions (generally warmer). Therefore, any mechanism that alleviates competitive exclusion is likely to induce changes in species realised distributions. Secondly, the importance of competition in structuring communities is likely reduced by increased levels of disturbance (Dayton 1971, Connell 1978, Huston 1979, McAuliffe 1984, Brooker and Callaghan 1998, Brooker 2006). Thus, if climate warming increases disturbance levels within a specific ecosystem, it is reasonable to expect that it will, to a certain degree, relax the role of competition as a selective filter for community assembly. Indeed, climate warming affects ecosystems stochastically through an increasing frequency and intensity of events such as drought-induced insect outbreaks (Allen et al. in press), heat-induced wildfires (Schumacher and Bugmann 2006), windthrows (Usbeck et al. in press), and permafrost degradations (Cannone et al. 2007). All these climaticextremes-induced effects can be viewed as disturbances under definitions such as destruction of plant biomass (Grime 1979), cause of mortality (Huston 1979), and disruption of ecosystem, community or population structure (Pickett and White 1985). Indeed, the increasing frequency of climatic extremes, indicated by changes in the interannual variability around mean values of climate parameters, have already been reported to influence species range margins (Zimmermann et al. 2009). The interplay between the idea of disturbance-related release from competition and the above-mentioned stress-gradient hypothesis gives the key to understanding potential downslope range shifts despite climate warming. We start outlining our conceptual model by restricting the temperature-elevation-stress relationship to elevational ranges where heat and drought have hardly any impacts, e.g. from temperate montane to alpine ecosystems (Fig. 2a). Following the stress-gradient hypothesis, the importance of competition is therefore likely to decrease upwards along the elevation gradient (Fig. 2b), from favourable abiotic conditions at low elevations (warmer conditions) to harsh abiotic conditions at high elevations (cold damage during winter, reduced energy during the growing season). To simplify matters in our conceptual model, we distinguish between two illustrative species with exactly the same realised distribution along the elevation gradient (as defined

ISSUE

IBS SPECIAL

Figure 2. Conceptual model of changes in species’ elevational distributions in response to climate warming. Panels depict: (a) the general decreasing trend in mean annual temperature along the elevation gradient, and an overall warming between two periods; (b) the general decrease in the importance of competition along a gradient of environmental severity represented by the elevation gradient, and the potential impact of warmer conditions on the relationship between elevation and the importance of competition; (c) the realised and the potential distributions along the elevation gradient for a species strongly limited by competitors from below, and the potential upslope and/or downslope range shifts of the realised distribution due to both climate warming and competitive release; (d) the realised and the potential distributions along the elevation gradient of a species less limited by competitors from below, and the resulting upslope range shift of the realised distribution due to climate warming (competitive release does not allow shifts towards lower elevations that are no longer climatically suitable); and (e) all the resulting combinations of changes in species elevational distributions involving contraction, expansions, or both simultaneously, and the observed proportions of species displaying upward, stable, and downward shifts in elevation in response to recent climate warming. Blue colours represent initial conditions before climate warming, whereas red colours represent changed conditions after an increase in temperatures. Red broken lines represent the impact of a reduction in the importance of competition after climate warming. The elevation gradient considered in our conceptual model ranges from temperate montane to alpine ecosystems where heat and drought have hardly any impact.

from their realised niche), but differing in the way they fill their potential distribution area (as defined from their fundamental ecophysiological niche) (Fig. 2c, d). Hereafter, 298

we will consistently refer to the ‘‘potential distribution’’ as the range corresponding to the species’ fundamental ecophysiological niche, and to the ‘‘realised distribution’’

Habitat modification as an alternative mechanism causing downslope range shifts

299

ISSUE

Species may also shift downslope as a direct consequence of habitat modification, with or without involving competitive release, either following natural disturbances (windthrows, fires, and avalanches), human-induced disturbances or permanent habitat changes (recreational activities, land use changes, and management practices), or due to other local changes in habitat suitability. For example, in the Swiss Central Alps, Ha¨ttenschwiler and Ko¨rner (1995) have suggested that the cessation of forest cattle grazing and the high level of nitrogen deposition may have led to denser and more exuberant ground vegetation, thereby enhancing the replacement of Pinus sylvestris seedling populations by those of P. cembra below the present lower margin of P. cembra adult trees. Similarly, Archaux (2004) has suggested that the increase of conifer areas at the expense of broad-leaved trees due to changes in forest management might cause both coniferous- and deciduous-forest bird species to shift their mean elevation downwards. We note that habitat modification in conjunction with climate warming may explain upslope range shifts as well. As an illustration, in the Swiss Alps, Gehrig-Fasel et al. (2007) have reported that 90% of upslope shifts in the local tree line are due to ingrowth and the filling of gaps indicating that land use is the primary driver over climate warming in many instances, although the two drivers may also act in combination. In a conceptual model involving climate change and herbivory pressure, Cairns and Moen (2004) have highlighted a potential pathway for the interaction of both climate change and herbivore pressure on tree line fluctuations leading to upslope migration, a stationary state, or retrogression of tree lines. Hence, habitat modification has often been claimed to be an important driver of elevational range shifts, acting in concert with climate warming or even outweighing it (Ha¨ttenschwiler and Ko¨rner 1995, Archaux 2004, Cairns and Moen 2004, Gehrig-Fasel et al. 2007, Vittoz et al. 2009). Nevertheless, habitat modification may well involve increased disturbance levels and thereby cause release from competition independently of climate warming, but with similar effects on species’ elevational distributions. Humaninduced disturbances are likely to be more frequent in lowland areas, however, given a generally increasing degree of anthropogenic habitat modification towards lower elevations of most mountainous settings (Nogue´s-Bravo et al. 2008). Consequently, the reduction in the importance of competition due to disturbances should be more important there, allowing range expansions of the realised distribution of species towards lower elevations as long as climatically suitable sites are available down below their lower margins. Therefore, species that fill only part of their potential distribution areas along the elevation gradient (Fig. 2c) are more likely to shift downwards in response to habitat modification alone, especially if their potential distribution areas do not shift upwards. For example, it has been suggested that unplanned vegetation destruction (burning and grazing), removing the pressure from competitive dominants from below, has enabled alpine and subalpine species in New Zealand to increase their elevational distributions downwards (Halloy and Mark 2003).

IBS SPECIAL

as the range corresponding to the species’ realised niche. The first illustrative species is strongly limited at the lower margin of its elevational distribution by competitors from below (low ‘‘realisation’’ of the potential distribution), whereas the second one is much less limited by biotic interactions (high ‘‘realisation’’ of the potential distribution). It is difficult to infer the physiological climatic requirements of these two species from their actual range since their elevational distributions are identical. Nevertheless, they might respond differently to climate warming because of this distinct ‘‘realisation’’ of their ecophysiological range. Let us now introduce the effect of climate warming in our conceptual model (red colours in Fig. 2). As argued above, warming-induced disturbances are likely to transiently reduce the importance of competition along the elevation gradient (Fig. 2b). Simultaneously, climate warming is likely to shift species’ potential distribution along the elevation gradient towards higher elevations (Fig. 2c, d). However, the transient reduction in the importance of competition at the lower margin of a species’ elevational distribution is likely to particularly benefit species that currently have a greater part of their potential distribution unfilled because of competition. For such species, climate warming allows range expansion towards lower elevations from which they had hitherto been competitively excluded (Fig. 2c). In contrast, species which are currently little limited by competition and largely fill their potential distribution areas along the elevation gradient will not be able to move downwards due to the upward shift of habitats climatically suitable to them (Fig. 2d). Figure 2e gives a full account of the climatically driven range shifts along the elevation gradient conceivable under this model. To sum up, species with a low ‘‘realisation’’ of their potential distribution areas along the elevation gradient are especially good candidates for downslope shifts, if they additionally have good dispersal abilities and a wide fundamental climatic niche. For example, Vetaas (2002) has suggested that Rhododendron species with their lower margins hitherto set by competition can grow under warmer climatic conditions, i.e. expand towards lower elevations. Additionally, species mid-range shifts might result not only from expansions at range margins alone, but also from changes in species local abundance (Fig. 2e). For example, species with low competitive abilities, but a wide potential distribution along the elevation gradient associated with a high degree of plasticity (Fig. 2c) might be already present downwards as remnant populations living at environmental extreme sites where the competitive species are excluded by abiotic factors (Eriksson 2000). In this latter case, downslope mid-range shifts might simply result from an increasing abundance of the species with these outlier populations acting as ‘‘expansion foci’’. Finally, we note that downslope range shifts under this model would be likely to be temporary, as the importance of competition may become reasserted if climate change slows down or come to a halt, while species will eventually be forced upwards in elevation if climate change continues and conditions at lower elevations shift beyond the fundamental climatic niche of the species.

IBS SPECIAL

ISSUE

Towards a unified view of climate warming and habitat modification effects on downslope range shifts Let us now consider a more unified point of view integrating the effects of climate warming and habitat modification. On the one hand, habitat modification may act in concert with climate warming to cause a reduction in the importance of competition at the lower margin of a species’ elevational distribution and to allow a potential downslope shift in its mid-range position due to potential expansions at the competitively-released margin (Fig. 2e). This is especially true for species that fill only part of their potential distribution areas along the elevation gradient (Fig. 2c). On the other hand, habitat modification may restrict upslope range expansions due to climate warming or even cause regression patterns at the higher margin of a species’ elevational distribution (Fig. 2e). For example, species may partially fail to colonise new climatically suitable areas at the higher margins of their elevational distributions if they are constrained by habitat fragmentation (Honnay et al. 2002). Additionally, other non-climatic factors such as restricted space towards mountaintops (Colwell et al. 2008), limited dispersal ability (Bossuyt et al. 1999, Hermy et al. 1999, Svenning and Skov 2006), and/or edaphic constraints may restrict, or at least delay warming-induced upslope shifts. Finally, natural habitat modification may further delay warming-induced upslope shifts through unexpected patterns of regression at the higher margins of species’ elevational distributions (Fig. 2e). For example, Cannone et al. (2007) have suggested that warming-induced permafrost degradations at high elevations may trigger habitat disturbances, in the form of debris flow and landslides, causing unexpected patterns of regression in vegetation coverage above 2500 m. This constitutes another migration barrier that restricts upslope migrations to disturbance-adapted species (Cannone et al. 2007). Such limitations to upslope migrations are coherent with observations in different species groups: extinctions at the lower margins of species distributions have been reported to be more common than colonisations at the higher margins (Wilson et al. 2005, 2007, Kelly and Goulden 2008, Moritz et al. 2008, Lenoir et al. 2009). This may sometimes result in no upslope migration, but rather in local changes in species abundance over time (Wilson and Nilsson 2009). In such cases, mortality-induced shifts may take place more rapidly than do recolonisation-induced shifts associated with both migration and establishment processes (Davis 1989). The resulting pattern is a transient ‘‘lean’’ upslope (Breshears et al. 2008). The few establishments of a given species towards higher elevations may fail to compensate for the losses at lower elevations leading to transient declines in species richness or biotic attrition not only at lower elevations (Colwell et al. 2008), but across the whole elevation gradient (Fig. 2 in Wilson et al. (2007)). This configuration is transient and, again, is likely to open a ‘‘window of opportunity’’ for highly vagile and plastic species that might shift either upslope or downslope to fill the gaps initiated either by climate warming or habitat modification. Such a process leaves biological communities with reduced numbers of species, and dominated by more 300

mobile and opportunistic species (Warren et al. 2001). Increased frequency of windthrows across central European forests during the last few decades (Usbeck et al. in press) is one example of disturbances that is likely to produce ‘‘windows of opportunities’’ for vagile species with a high degree of plasticity. Thus, habitat modifications strongly interact with climate warming and contribute to bias competitive release even further towards lower elevations, making downslope range shifts of some species more likely than with climate warming alone.

Other plausible causes of downslope range shifts Of course, downslope range shifts could be driven by changes in other aspects of climate than mean temperature, e.g. precipitation regime, snow cover duration, water balance, or seasonality in climate parameters. These complex aspects of climate variability may heavily influence species range margins (Zimmermann et al. 2009), and thus more complex environment-competition interactions are likely to cause unexpected range shifts in response to climate warming. At high elevations, for example, warmer temperatures may decrease the winter snow cover duration (Beniston 2005), and thus may cause frost damage at the higher margin of a species’ elevational distribution, which in turn may alleviate the competitive effect of this species on other ones potentially migrating towards lower elevations (Fig. 2c). Consequently, the competitive control that this species exerts on the distributions of the species above will likely become less tight. Additionally, the influx towards areas vacated by upslope shifting competitive species is likely to occur both from above and below. This should result in some species shifting upslope and others shifting downslope, and others even expanding towards both sides without changing in their mid-range position (Fig. 2e). Therefore, complex environment-competition interactions may also cause downslope range shifts, but temporarily before other stronger competitors invade.

A case study from French mountain forests: proportion of random downslope movements According to our null expectation, the proportion of random downslope mid-range shifts despite an average upslope trend of z m is dependent on the range of random variation among shifts (Fig. 1). We used data from a previous study focusing on the shifts in the elevational position of plant species’ maximum probability of presence (optimum) (Lenoir et al. 2008) to estimate this range of random variation among shifts and then assess the proportion of random downslope mid-range shifts despite an average upslope trend of z m. That study found an average upslope trend of 65 m, among 171 plant species, between a 1905 1985 dataset and a carefully matched 1986 2005 dataset for French mountain forests. Each dataset comprised 3991 surveys. To estimate the range of random variation among shifts, we constructed two random

datasets, each built by randomly drawing (with replacement) 3991 forest surveys from the 1905 1985 dataset described by Lenoir et al. (2008). Because the increasing trend in mean annual temperature and the warmest records have mostly occurred since the late 1980s (Jones et al. 2001), we chose to draw our two random datasets from the first period to avoid potential strong differences in climatic conditions between the two randomly-drawn datasets. Indeed, it was important that the two bootstrap samples represented an identical range of environmental conditions. We then used the same analytical method as described in Lenoir et al. (2008) to compute species elevation optimum for each of the 171 studied species in each of the two bootstrap samples. The estimated elevational optimum position was bounded between the lowest and the highest elevations in the 1905 1985 dataset. However, instead of computing the difference in each species optimum elevation between two different periods, we here computed the difference in each species optimum elevation between two bootstrap samples from the same period to get the distribution of differences in optima expected from stochasticity alone. We repeated this procedure 1000 times using the ‘‘boot’’ library (Canty and Ripley 2008) in R (R Development Core Team 2009). Averaging cross the 1000 iterations, we found a mean difference in species optimum elevation of 0.4 m (standard deviation: 10.5 m) and a confidence interval for mean at 95% ranging from 14.2 m (standard deviation: 10.7 m) to 13.3 m (standard deviation: 10.6 m). Finally, to assess the proportion of downslope mid-range shifts expected at random despite an average upslope trend of 65 m in French mountain forest plants, we simply repositioned, for each of the 1000 iterations, the distribution of random differences in optima 65 m upwards, i.e. to the right, by adding 65 m to the

location of each shift (Fig. 3a). We then estimated the expected proportion of random downslope shifts, despite an average upslope trend of 65 m, as the proportion of shifts in the left tail of the distribution, truncated at 0 m shift (see solid dark gray bars in Fig. 3a). Across the 1000 iterations (Fig. 3b), we found a proportion of 16% (standard deviation: 4%) of species expected to have optima below the 1905 1985 elevational mean, by chance alone, which is about half the proportion of downslope shifts we originally found (30%) between 1905 1985 and 1986 2005 (Lenoir et al. 2008). Thus, the number of species found to move downwards along the elevation gradient is approximately twice as high as expected by chance under the observed general upward trend. We note that this estimated proportion of species ‘‘going against the flow’’ (14%) is higher than the number of species with significant downslope shifts (5/171) found when individual species optimum elevation were tested for significant difference from a constant species optimum elevation expectation (Table S2 in Lenoir et al. (2008)). However, this probability reflects the conservative nature of the test we used for testing differences in individual species optima along the elevation gradient (Lenoir et al. 2008). The five forest plant species displaying significant individual downslope shifts despite an average upslope trend of 65 m are Clinopodium vugare, Dryopteris dilatata, Quercus pubescens, Rubus fructicosus and Saxifraga cuneifolia. Two of these species have seeds dispersed by birds (Quercus pubescens and Rubus fructicosus), the spores of Dryopteris dilatata are tiny and hence easily going with the wind, while Clinopodium vulgare is dispersed by epizoochory (Rameau et al. 1993). Hence, four of these five species have efficient dispersal mechanisms, most likely an important trait for allowing downslope range shifts despite climate warming.

IBS SPECIAL 301

ISSUE

Figure 3. Distribution of (a) observed and random differences in optima along the elevation gradient for 171 plant species using published data from French mountain forests (Lenoir et al. 2008), the distribution of random differences in optima representing a single illustrative case of the 1000 bootstrap iterations, and (b) distribution of the proportion of random downslope mid-range shifts despite an average upslope trend of 65 m for the 1000 bootstrap iterations. Solid light gray bars represent the observed shifts in species optimum elevation between 1905 1985 and 1986 2005. Unfilled bars represent the distribution of random shifts in species optimum, computed by comparing two bootstrap samples drawn from the 1905 1985 dataset, after repositioning this distribution 65 m upslope, i.e. to the right (see text for details). Solid dark gray bars represent the proportion of downslope mid-range shifts expected at random, despite an average upslope trend of 65 m. The vertical dark broken line displays the average upslope trend of 65 m. The vertical white broken line displays the proportion of random downslope mid-range shifts despite an average upslope trend of 65 m for the single illustrative case of the 1000 bootstrap iterations.

Although Saxifraga cuneifolia has low dispersal abilities (barochory) and a distribution mainly restricted between 1500 and 2000 m, it also occurs sporadically in the lowlands, reaching down to 300 m in French mountain forests (Rameau et al. 1993). Thus, Saxifraga cuneifolia might represent a species strongly limited by competitors from below (Fig. 2c). Additionally, Rubus fructicosus, and to a lesser extent Clinopodium vulgare, and Quercus pubescens seedlings, are highly reactive to canopy opening, i.e. positively influenced by disturbance. While we do not intend to validate our conceptual model with these examples, we note that these five downslope shifting species have a series of traits making them likely to respond to climate change in a way outlined by our model. However, a much more thorough empirical testing of this model is clearly needed.

IBS SPECIAL

ISSUE

Conclusion We suggest that downslope range shifts of species may constitute an indirect biotic response to both climate warming and habitat modification rather than representing just random effects due to stochastic fluctuations in population distributions or observer errors. The concept presented here should become part of a general framework for future studies of changes in species distributions in response to climate warming. In our conceptual model, we assume, on a timescale too short for adaptative change, that downslope shifts primarily occur for species that are strongly limited by competition at their lower elevation range margin, and therefore have a realised distribution that do no fill their potential distribution areas almost completely along the elevation gradient. To test this hypothesis, one could select two sets of species: one set of species that have significantly shifted downslope and another set of species that have significantly shifted upslope. One could then compare their realised distribution in their natural habitats with their potential distribution areas additionally assessed by experiments (common garden or ecotron). In such an experiment, we would expect larger differences between the realised and the potential distributions along the elevation gradient for the set of species that have significantly shifted downslope. Although downslope range shifts, particularly where solely driven by warming-induced competitive release, should be only transient, we underpin the necessity to take the hitherto neglected downslope range shifts of species more explicitly into consideration when making predictions of the effects of future climate change scenarios on species distributions.

Acknowledgements We are thankful to the numerous people who collected data and to people who managed and provided the databases for use, particularly Ingrid Seynave and Patrice de Ruffray. We thank Romain Bertrand and Urs Treier for inspiring discussions. Three anonymous referees and Robert K. Colwell provided many helpful comments and suggestions. We gratefully acknowledge grants from the National Inst. for Agricultural Research (to J. Lenoir), the FP6-ECOCHANGE project of the EU commission (grant GOCE-CT-036866 to S. Dullinger, A.

302

Guisan and N. E. Zimmermann), and the Danish Natural Science Research Council (grant #272-07-0242 to J.-C. Svenning).

References Allen, C. D. et al. in press. A global overview of drought and heatinduced tree mortality reveals emerging climate change risks for forests. For. Ecol. Manage. Archaux, F. 2004. Breeding upwards when climate is becoming warmer: no bird response in the French Alps. Ibis 146: 138 144. Beniston, M. 2005. Mountain climates and climatic change: an overview of processes focusing on the European Alps. Pure Appl. Geophys. 162: 1587 1606. Bertness, M. D. and Callaway, R. 1994. Positive interactions in communities. Trends Ecol. Evol. 9: 191 193. Bossuyt, B. et al. 1999. Migration of herbaceous plant species across ancient-recent forest ecotones in central Belgium. J. Ecol. 87: 628 638. Breshears, D. D. et al. 2008. Vegetation synchronously leans upslope as climate warms. Science 105: 11591 11592. Brooker, R. W. 2006. Plant plant interactions and environmental change. New Phytol. 171: 271 284. Brooker, R. W. and Callaghan, T. V. 1998. The balance between positive and negative plant interactions and its relationship to environmental gradients: a model. Oikos 81: 196 207. Brooker, R. W. et al. 2005. The importance of importance. Oikos 109: 63 70. Brown, J. H. and Lomolino, M. V. 1998. Biogeography. Sinauer. Brown, J. H. et al. 1996. The geographic range: size, shape, boundaries and internal structure. Annu. Rev. Ecol. Syst. 27: 597 623. Cairns, D. M. and Moen, J. 2004. Herbivory influences tree lines. J. Ecol. 92: 1019 1024. Callaway, R. M. 1998. Competition and facilitation on elevation gradients in subalpine forests of the northern Rocky Mountains, USA. Oikos 82: 561 573. Callaway, R. M. and Walker, L. R. 1997. Competition and facilitation: a synthetic approach to interactions in plant communities. Ecology 78: 1958 1965. Callaway, R. M. et al. 2002. Positive interactions among alpine plants increase with stress. Nature 417: 844 848. Cannone, N. et al. 2007. Unexpected impacts of climate change on alpine vegetation. Front. Ecol. Environ. 5: 360 364. Canty, A. and Ripley, B. 2008. Boot: Bootstrap R (S-PLUS) functions. /<www.R-project.org/>. Chen, I. C. et al. 2009. Elevation increases in moth assemblages over 42 years on a tropical mountain. Proc. Nat. Acad. Sci. USA 106: 1479 1483. Colwell, R. K. et al. 2008. Global warming, elevational range shifts and lowland biotic attrition in the wet tropics. Science 322: 258 261. Connell, J. H. 1978. Diversity in tropical rain forests and coral reefs. Science 199: 1302 1310. Davis, M. B. 1989. Lags in vegetation response to greenhouse warming. Clim. Change 15: 75 82. Dayton, P. K. 1971. Competition, disturbance, and community organization: the provision and subsequent utilization of space in a rocky intertidal community. Ecol. Monogr. 41: 351 389. Eriksson, O. 2000. Functional roles of remnant plant populations in communities and ecosystems. Global Ecol. Biogeogr. 9: 443 449. Franco, A. M. A. et al. 2006. Impacts of climate warming and habitat loss on extinctions at speciesâ&#x20AC;&#x2122; low-latitude range boundaries. Global Change Biol. 12: 1545 1553.

303

ISSUE

Parmesan, C. and Yohe, G. 2003. A globally coherent fingerprint of climate change impacts across natural systems. Nature 421: 37 42. Parolo, G. and Rossi, G. 2008. Upward migration of vascular plants following a climate warming trend in the Alps. Basic Appl. Ecol. 9: 100 107. Pauli, H. et al. 2007. Signals of range expansions and contractions of vascular plants in the high Alps: observations (1994 2004) at the GLORIA* master site Schrankogel, Tyrol, Austria. Global Change Biol. 13: 147 156. Pickett, S. T. A. and White, P. S. 1985. The ecology of natural disturbance and patch dynamics. Academic Press. R Development Core Team 2009. R: a language and environment for statistical computing. R Foundation for Statistical Computing, /<www.R-project.org/>. Rameau, J. C. et al. 1993. Flore forestie`re franc¸aise guide e´cologique illustre´ tome 2: montagnes. Inst. pour le De´veloppement Forestier. Raxworthy, C. J. et al. 2008. Extinction vulnerability of tropical montane endemism from warming and upslope displacement: a preliminary appraisal for the highest massif in Madagascar. Global Change Biol. 14: 1703 1720. Schumacher, S. and Bugmann, H. 2006. The relative importance of climatic effects, wildfires and management for future forest landscape dynamics in the Swiss Alps. Global Change Biol. 12: 1435 1450. Suttle, K. B. et al. 2007. Species interactions reverse grassland responses to changing climate. Science 315: 640 642. Svenning, J. C. and Skov, F. 2006. Potential impact of climate change on the northern nemoral forest herb flora of Europe. Biodivers. Conserv. 15: 3341 3356. Tryjanowski, P. et al. 2005. Uphill shifts in the distribution of the white stork Ciconia ciconia in southern Poland: the importance of nest quality. Divers. Distrib. 11: 219 223. Usbeck, T. et al. in press. Wind speed measurements and forest damage in Canton Zurich (central Europe) from 1891 to winter 2007. Int. J. Climatol. Vetaas, O. R. 2002. Realized and potential climate niches: a comparison of four Rhododendron tree species. J. Biogeogr. 29: 545 554. Vittoz, P. et al. 2008. One century of vegetation change on Isla Persa, a nunatak in the Bernina massif in the Swiss Alps. J. Veg. Sci. 6: 671 680. Vittoz, P. et al. 2009. Low impact of climate change on subalpine grasslands in the Swiss northern Alps. Global Change Biol. 15: 209 220. Walther, G. R. et al. 2005. Trends in the upward shift of alpine plants. J. Veg. Sci. 16: 541 548. Warren, M. S. et al. 2001. Rapid responses of British butterflies to opposing forces of climate and habitat change. Nature 414: 65 68. Wilson, R. J. et al. 2005. Changes to the elevational limits and extent of species ranges associated with climate change. Ecol. Lett. 8: 1138 1146. Wilson, R. J. et al. 2007. An elevational shift in butterfly species richness and composition accompanying recent climate change. Global Change Biol. 13: 1873 1887. Wilson, S. D. and Nilsson, C. 2009. Arctic alpine vegetation change over 20 years. Global Change Biol. 15: 1676 1684. Zimmermann, N. E. et al. 2009. Climatic extremes improve predictions of spatial patterns of tree species. Proc. Nat. Acad. Sci. USA 106: 19723 19728.

IBS SPECIAL

Gehrig-Fasel, J. et al. 2007. Tree line shifts in the Swiss Alps: climate change or land abandonment? J. Veg. Sci. 18: 571 582. Grime, J. P. 1979. Plant strategies and vegetation processes. Wiley. Halloy, S. R. P. and Mark, A. F. 2003. Climate-change effects on alpine plant biodiversity: a New Zealand perspective on quantifying the threat. Arct. Antarct. Alp. Res. 35: 248 254. Ha¨ttenschwiler, S. and Ko¨rner, C. 1995. Responses to recent climate warming of Pinus sylvestris and Pinus cembra within their montane transition zone in the Swiss Alps. J. Veg. Sci. 6: 357 368. Hermy, M. et al. 1999. An ecological comparison between ancient and other forest plant species of Europe, and the implications for forest conservation. Biol. Conserv. 91: 9 22. Hickling, R. et al. 2006. The distributions of a wide range of taxonomic groups are expanding polewards. Global Change Biol. 12: 450 455. Honnay, O. et al. 2002. Possible effects of habitat fragmentation and climate change on the range of forest plant species. Ecol. Lett. 5: 525 530. Hughes, L. 2000. Biological consequences of global warming: is the signal already apparent? Trends Ecol. Evol. 15: 56 61. Huston, M. 1979. A general hypothesis of species diversity. Am. Nat. 113: 81 101. Jones, P. D. et al. 2001. The evolution of climate over the last millennium. Science 292: 662 667. Kelly, A. E. and Goulden, M. L. 2008. Rapid shifts in plant distribution with recent climate change. Proc. Nat. Acad. Sci. USA 105: 11823 11826. Klanderud, K. and Birks, H. J. B. 2003. Recent increases in species richness and shifts in altitudinal distributions of Norwegian mountain plants. Holocene 13: 1 6. Konvicka, M. et al. 2003. Uphill shifts in distribution of butterflies in the Czech Republic: effects of changing climate detected on a regional scale. Global Ecol. Biogeogr. 12: 403 410. Leathwick, J. R. and Austin, M. P. 2001. Competitive interactions between tree species in New Zealand’s old-growth indigenous forests. Ecology 82: 2560 2573. Lenoir, J. et al. 2008. A significant upward shift in plant species optimum elevation during the 20th century. Science 320: 1768 1771. Lenoir, J. et al. 2009. Differences between tree species seedling and adult altitudinal distribution in mountain forests during the recent warm period (1986 2006). Ecography 32: 765 777. Maestre, F. T. et al. 2009. Refining the stress-gradient hypothesis for competition and facilitation in plant communities. J. Ecol. 97: 199 205. McArthur, R. H. 1972. Geographycal ecology: patterns in the distribution of species. Harper and Row. McAuliffe, J. R. 1984. Competition for space, disturbance, and the structure of a benthic stream community. Ecology 65: 894 908. Moritz, C. et al. 2008. Impact of a century of climate change on small-mammal communities in Yosemite National Park, USA. Science 322: 261 264. Nogue´s-Bravo, D. et al. 2008. Scale effects and human impact on the elevational species richness gradients. Nature 453: 216 219. Normand, S. et al. 2009. Importance of abiotic stress as a rangelimit determinant for European plants: insights from species responses to climatic gradients. Global Ecol. Biogeogr. 18: 437 449.

Ecography 33: 304 314, 2010 doi: 10.1111/j.1600-0587.2010.06251.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: David Nogue´s-Bravo. Accepted 12 March 2010

Species distribution models in climate change scenarios are still not useful for informing policy planning: an uncertainty assessment using fuzzy logic Raimundo Real, Ana Luz Ma´rquez, Jesu´s Olivero and Alba Estrada R. Real (rrgimenez@uma.es), A. Luz Ma´rquez and J. Olivero, Biogeography, Diversity, and Conservation Research Team, Dept of Animal Biology, Fac. of Sciences, Univ. of Malaga, ES-29071, Malaga, Spain. A. Estrada, Inst. de Investigacio´n en Recursos Cinege´ticos IREC (CSIC-UCLM-JCCM), Ronda de Toledo s/n, ES-13071, Ciudad Real, Spain.

IBS SPECIAL

ISSUE

We compared the effect of general circulation models and greenhouse gas emission scenarios on the uncertainty associated with models predicting changes in areas favourable to animal species. Given that mountain species are particularly at risk due to climate warming, we selected one amphibian (Baetic midwife toad), one reptile (Lataste’s viper), one bird (Bonelli’s eagle), and one mammal (Iberian wild goat) present in Spanish mountains to model their distributional response to climate change during this century. Climate forecasts for the whole century were provided by the Agencia Estatal de Meteorologı´a (AEMET; National Meteorological Agency) of Spain, which adapted the general circulation models CGCM2 and ECHAM4 and produced expected temperature and precipitation values for Spain according to the A2 and B2 emission scenarios. We constructed separate models of the species response to spatial, topographic, human, and climate variables using current values of the corresponding variables. We predicted future areas favourable to the species by replacing the current climate values with those expected according to each climate change scenario, while keeping spatial, topographic and human variables constant. Fuzzy logic was used to compute the coincidence between predictions for different emission scenarios in the same global circulation model, and the consistency between predictions for the same emission scenario applying different general circulation models. In general, coincidences were higher than consistencies and, thus, discrepancies between predictions were more attributable to uncertainty in global circulation models, i.e. our insufficient knowledge concerning the effect of the oceans and atmosphere on climate, than to the putative effect of different emission scenarios on future climates. Our conclusion is that species distribution models in climate warming scenarios are still not useful for informing emission policy planning, although they have great potential as tools once consistencies become higher than coincidences.

Species distribution models are usually employed to assess the potential changes in the distribution of species in response to different factors (Barbosa et al. 2003, Munõz et al. 2005, Munõz and Real 2006, Farfań et al. 2008, Luoto and Heikkinen 2008). Several studies have focused on modelling species distribution shifts in response to climate change (Beaumont et al. 2005, Levinsky et al. 2007, Foody 2008) to monitor the effects of the increase in global average temperature recorded over the last century and predicted for the present century (IPCC 2007). One of the main aims of modelling biogeographical responses to climate change is to inform policy planning by providing a kind of virtual feedback on greenhouse gas emissions. This kind of modelling uses future climates, which are predicted according to the combination of Atmosphere-Ocean General Circulation Models (AOGCMs) and special reports on emission scenarios (SRESs). This procedure incorporates several sources of uncertainty, some of them related to the existence of different 304

AOGCMs (over 20 models available), characterized by different hypotheses on the effect of the oceans and the atmosphere on climate, and SRESs (40 scenarios of future emissions developed by the Intergovernmental Panel on Climate Change), characterized by different storylines of future socio-economic and technological development (Beaumont et al. 2008). The use of different AOGCMs may produce conflicting projected distributions of a species (Xu and Yan 2001). Nevertheless, species distribution models in climate-warming scenarios (which include the effects of AOGCMs and SRESs) will be useful for informing emission policy planning if differences in predicted effects due to differences in SRES are significantly higher than those due to differences in AOGCMs. Fuzzy set theory may be applied to the analysis of species distribution modelling (Robertson et al. 2004, Gevrey et al. 2006, van der Broekhoven et al. 2006, Estrada et al. 2008, Real et al. 2009), and may be used to assess the models’ projections

to the future and the different kinds of uncertainty associated with them. The Mediterranean region is considered to be highly responsive to climate change, because of its geographical situation between temperate central Europe and arid northern Africa (Sa´nchez et al. 2004, Giorgi and Lionello 2008). Mountain ecosystems seem to be particularly sensitive to global warming and are of particular concern to policy planners (Foster 2001, Nogue´s-Bravo et al. 2007, Trivedi et al. 2008), especially in the Mediterranean region (Nogue´s-Bravo et al. 2008). Mountain species ranges might shift more rapidly in response to climate change, as mountains often retain more natural habitats than lowlands do, and because in mountains the microclimate varies with elevation and the species may track climate change over shorter distances (Wilson et al. 2007). Several mountain species have indeed shifted to higher altitudes across European mountains (Grabherr 1994, Klanderud and Birks 2003, Pen˜uelas and Boada 2003, Wilson et al. 2005, Pauli et al. 2007, Wilson et al. 2007). Mountain species are, therefore, especially suited to evaluating the potential effects of future climate change on distributions. We used fuzzy logic to compare the effect of AOGCMs and SRESs on the uncertainty associated with models predicting changes in areas favourable to four mountain species in mainland Spain during the 21st century.

Methods Species and study area

Climatic data

We modelled the distribution of each species in 10 10 km UTM cells with variables related to four explanatory factors taken separately: spatial situation, topography, human activity, and climate (Table 1). For climate we used the values for the period 1961 1990. Real et al. (2008b) showed that while in the north of Europe energy availability is the main factor limiting species distribution, in the south their distribution seems to be more affected by climatic stress due to an excess of environmental energy. We used maximum temperatures because this variable is more representative of this type of climatic stress; in the Iberian Peninsula an increase in maximum temperatures may be potentially more detrimental to species distributions than an increase in minimum temperatures. Non-climatic factors such as topography, human activity, history and population dynamics may have an effect on species distributions (Real et al. 2008a, 2009). As the species may show differential responses to these factors (De Frene et al. 2009), their relative importance should be assessed together with climate before projecting species distribution models to the future. The inclusion of spatial variables in a model can reveal a geographical trend in distribution that could be associated with historical events or with the species population dynamics (Legendre 1993, Real et al. 2003). On the other hand, latitude and longitude also affect the climatic variables (Ma´rquez et al. 2004). Consequently, the true Table 1. Explanatory factors and associated variables used to model the species distributions. Factors

Code

Variables

Spatial situation

La Lo A S SE WE DHi U100

Latitude (8N)(1) Longitude (8E)(1) Mean altitude (m)(2) Slope (8) (calculated from altitude) Southward exposure degree(3) Westward exposure degree(3) Distance to the nearest highway (km)(1) Distance to the nearest urban centre with 100 000 inhabitants (km)(1) Distance to the nearest urban centre with 500 000 inhabitants (km)(1) Human population density in 2000 (number of inhabitants km 2)(4) Annual precipitation (mm)(5) Spring precipitation (mm)(5) Summer precipitation (mm)(5) Autumn precipitation (mm)(5) Winter precipitation (mm)(5) Annual maximum temperature(5) January maximum temperature(5) July maximum temperature(5) Spring maximum temperature(5) Summer maximum temperature(5) Autumn maximum temperature(5) Winter maximum temperature(5)

Topography

Human activity

U500 HPd Climatic

PAn PSp PSu PAu PWi TAn TJa TJu TSp TSu TAu TWi

Sources: (1)I.G.N. (1999); (2)US Geological Survey (1996); (3)Shuttle Radar Topography Mission (SRTM), Farr and Kobrick (2000); (4) ORNL (2001); (5)Agencia Estatal de Meteorologı´a of Spain (AEMET), Ministerio de Medio Ambiente <www.aemet.es/es/elcli ma/cambio_climat/escenarios>.

305

ISSUE

The climatic variables were the result of regionalizing the general circulation models, which combined AOGCMs and SRESs, to Spain. This regionalization was done by the Agencia Estatal de Meteorologı´a (AEMET; National Meteorological Agency <www.aemet.es/es/elclima/cambio_ climat/escenarios>) of Spain (Brunet et al. 2007), which used two AOGCMs: CGCM2 from the Canadian Climate Centre for Modelling and Analysis, and ECHAM4 from the Max Planck Inst. fu¨r Meteorologie; and two SRESs: A2 and B2 (Nakicenovic et al. 2000). These SRESs represent an intermediate position regarding the wide range of

Distribution modelling

IBS SPECIAL

We selected four mountain species in mainland Spain: Baetic midwife toad Alytes dickhilleni, Lataste’s viper Vipera latasti, Bonelli’s eagle Aquila fasciata, and Iberian wild goat Capra pyrenaica to model their distributional responses to climate change during this century. Distribution data were extracted from Martı´ and del Moral (2003), Pleguezuelos et al. (2004), and Palomo et al. (2007) and are taken to represent the species distributions in 1990. The study area was located in the Mediterranean region, a transition zone between the temperate climate of central Europe and the arid climate of northern Africa. As small changes in the processes that control those climates can lead to important changes in the Mediterranean climate, this area is important to analyse the effect of future climate changes on biodiversity (Giorgi and Lionello 2008, Nogue´s-Bravo et al. 2008).

projected shifts in temperature and precipitation (Brunet et al. 2007).

ISSUE

IBS SPECIAL

effect of climate should be assessed in the context of the spatial influences on the species distribution and on climate. As for the human variables, major urban centres and population density, although referring to 1999 and 2000, respectively, do not differ greatly from the situation in 1990, whereas the highways built after 1990 were already major roads in this year. Therefore, all these variables are representative of the general pattern in effect in 1990. For each species and factor (spatial situation, topography, human activity, and climate) we performed logistic regression of presence/absence with each variable related to the factor separately. To control for the increase in type 1 errors due to multiple tests (Benjamini and Hochberg 1995, GarcÄąÂ´a 2003), we controlled the false discovery rate (FDR) using the procedure proposed by Benjamini and Hochberg (1995), only accepting the variables that were significant under an FDR of q B0.05. We then performed forward-backward stepwise logistic regression of presence/ absence data on each subset of significant predictor variables related to the factor. In this way we obtained for each species four multivariate models, one for each factor considered. These factor models show the response of the species to spatial, topographic, human, and climatic variables, separately. We obtained a combined model performing forwardbackward stepwise selection of the variables that were involved in any factor model. We then applied the favourability function (Real et al. 2006), which allows direct comparison of favourability values for species differing in their prevalence. In biological terms, this function has proven to be able to reflect species abundance (Real et al. 2009) and performs correctly when transferring models between different geographical areas (Barbosa et al. 2009). We assessed the discrimination power of these models by calculating Cohenâ&#x20AC;&#x2122;s kappa, sensitivity, specificity, and their Correct Classification Rate (CCR), using the favourability value of F 0.5 as classification threshold, and the Area Under the Curve (AUC) of the Receiver Operating Characteristic, which is independent of any favourability threshold (Hosmer and Lemeshow 2000). Lobo et al. (2008) recommended not using AUC as a comparative measure of accuracy between model results when the occupancy of the species in the territory differ, which is the case here. We used AUC because it provides a measure of the degree to which a species is restricted to a part of the variation range of the modelled predictors, which is valuable information in models intended to predict the response of the species to changes in these predictors. The goodness-offit of the models was assessed using the Hosmer and Lemeshow test (Hosmer and Lemeshow 2000).

Models of this kind are inductively obtained from current distribution data following certain induction rules that guarantee agreement with them and, thus, do not require validation with the training data. Models should be validated according to, and specifically for, the specific purpose they are built for. Although true validation of models built to be transferred to the future is not possible at present (ArauÂ´jo et al. 2005), we determined whether the models were at least spatially transferable within the study area, by assessing if, after recalibrating them using a training dataset, they perform similarly well in a pseudovalidation dataset. We subsequently divided our whole dataset in a randomly distributed recalibration set and a remaining pseudovalidation set. The proportion of the whole dataset used to constitute the recalibration set was determined by the expression: [1 (p 1)1=2 ] 1 , where p is the number of predictor variables in each distribution model (Fielding and Bell 1997). In each model we assessed the agreement between the results of the recalibration and pseudovalidation by comparing their corresponding Cohenâ&#x20AC;&#x2122;s kappa, sensitivity, specificity, CCR, and AUC, taking into account that unsatisfactory coherence between recalibration and pseudovalidation results does not imply that the recalibrated model is wrong (ArauÂ´jo et al. 2005), and even less that the whole model is wrong. We projected to the future the models based on the whole datasets, as they usually perform better than those based on a subset (Fielding and Bell 1997, ArauÂ´jo et al. 2005). Future projection assessment using fuzzy set theory Areas favourable to each species were projected to the future by replacing the current (1961 1990) climatic values in the favourability models with those expected according to each AOGCM and SRES for the following time periods (2011 2040, 2041 2070, 2071 2100). Values of the spatial, topographic and human variables were not modified. The process of environmental modelling can be understood as the identification of the fuzzy set of areas favourable to each species (Estrada et al. 2008). In the favourability models, the favourability values represent the degrees of membership of each area to the fuzzy set of areas favourable to the species. We used various fuzzy logic operations (Kuncheva 2001) to calculate, for each future projection, several features of the predicted impact of climate change on the species favourability, namely the favourability overlap (O), the favourability maintenance (M), the predicted shift in favourability (S), and the increment in favourability (I) with respect to the 1961 1990 period:

Table 2. Variables included in the favourability models for each species and combination of AOGCM and SRES. Variables codes as in Table 1.

A. dickhilleni CGCM2-A2 CGCM2-B2 ECHAM4-A2/B2

306

Lo, La, A, S, PAn, PSu, TSp, Tau Lo, La, A, S, PAn, PSu, TSp, TAu Lo, La, A, S, U500, PAn, PSp

V. latasti

A. fasciata

Lo, La, A, U100, U500, PSp, TJa, TWi A, Dhi, U100, U500, PSp, TJa, TWi Lo, La, A, PAn, PWi

Lo, A, S, PSu, TJu A, S, PAn, TJu Lo, A, S, PSu, TJu, TSu, TAu

C. pyrenaica Lo, A, S, Dhi, U100, PSp, PSu, PAu, Twi Lo, A, S, Dhi, U100, PSp, PSu, PAu, Twi Lo, La, A, S, Dhi, PAn, PAu, TJa

Table 3. Values obtained for different discrimination assessment measures of the favourability models obtained for the period 1961 1990. Cohen’s kappa, sensitivity, specificity and Correct Classification Rate (CCR) have been calculated using the favourability value of F 0.5 as a classification threshold. AUC: Area Under the Curve of the Receiver Operating Characteristic. Goodness-of-fit was assessed with the Hosmer and Lemeshow test (H-L). * p B0.01, n.s. p 0.05. cFp is the cardinality of the fuzzy set of favourable areas modelled for each species and referring to the 1961 1990 period. Pre: prevalence.

cFp

Kappa

Sensitivity

Specificity

CCR

AUC

H-L

A2 B2 A2/B2

480.61 480.45 455.09

0.43 0.429 0.456

0.971 0.971 0.978

0.928 0.927 0.934

0.929 0.928 0.935

0.986 0.986 0.987

1.696 n.s. 1.65 n.s. 0.586 n.s.

A2 B2 A2/B2

2235.89 2234.29 2116.79

0.234 0.211 0.276

0.643 0.618 0.719

0.689 0.682 0.683

0.681 0.67 0.689

0.737 0.731 0.768

12.59 n.s. 33.66* 16.29 n.s.

A2 B2 A2/B2

1930.27 1935.98 1912.35

0.422 0.414 0.406

0.808 0.803 0.799

0.775 0.772 0.768

0.781 0.777 0.773

0.853 0.847 0.861

16.20 n.s. 8.06 n.s. 11.12 n.s.

A2 B2 A2/B2

1505.59 1507.66 1488.95

0.487 0.489 0.498

0.875 0.878 0.889

0.818 0.818 0.82

0.826 0.826 0.83

0.916 0.916 0.92

10.66 n.s. 9.79 n.s. 16.50 n.s.

Pre

A. dickhilleni

CGCM2 0.031 ECHAM4

V. latasti

CGCM2 0.179 ECHAM4

A. fasciata

CGCM2 0.164 ECHAM4

C. pyrenaica

CGCM2 0.138 ECHAM4

c(Ff þFp ) c(Ff Fp )

c(Ff þFp ) c(Fp )

Min[c(Fp ) c(Ff þFp ); c(Ff ) c(Ff þFp )] c(Fp ) c(Ff ) c(Fp ) c(Fp )

Coincidence

c(FA2 þFB2 ) c(FA2 FB2 )

where, FA2 is the predicted future favourability according to the AOGCM and the scenario A2, and FB2 is the predicted future favourability according to the AOGCM and the scenario B2. Consistence is defined here as the agreement between predictions for a given SRES applying different AOGCMs, and is computed as follows: Consistence

c(FC þFE ) c(FC FE )

where, FC is the predicted future favourability according to the circulation model CGCM2, and FE is the predicted

Table 4. Values obtained for Cohen’s kappa (K), Correct Classification Rate (CCR) and Area Under the Curve (AUC) of the Receiver Operating Characteristic for each model on the recalibration set (rc) and on the pseudovalidation set (va). SE: standard error of Kappa.

CGCM2

A. dickhilleni ECHAM4

ECHAM4 CGCM2

A. fasciata ECHAM4 CGCM2

C. pyrenaica ECHAM4

Kva

CCRrc

CCRva

AUCrc

AUCva

A2 B2 A2/B2

0.423 0.422 0.453

0.051 0.051 0.052

0.457 0.454 0.469

0.063 0.062 0.063

0.933 0.933 0.939

0.920 0.919 0.923

0.987 0.987 0.988

0.982 0.982 0.982

A2 B2 A2/B2

0.241 0.227 0.275

0.021 0.021 0.020

0.204 0.196 0.261

0.033 0.032 0.032

0.689 0.681 0.687

0.665 0.655 0.681

0.746 0.739 0.775

0.710 0.706 0.749

A2 B2 A2/B2

0.423 0.467 0.509

0.024 0.024 0.024

0.346 0.365 0.371

0.034 0.034 0.033

0.786 0.808 0.822

0.758 0.763 0.762

0.862 0.853 0.872

0.836 0.835 0.839

A2 B2 A2/B2

0.477 0.476 0.371

0.024 0.024 0.034

0.488 0.490 0.421

0.041 0.035 0.048

0.822 0.821 0.873

0.815 0.816 0.868

0.914 0.914 0.916

0.922 0.921 0.929

307

ISSUE

CGCM2

V. latasti

Krc

IBS SPECIAL

where, c(X ) is the cardinality of the X fuzzy set, that is, the sum of all cells’ degrees of membership in the fuzzy set X. Ff is the fuzzy set of future areas favourable to the species, and the degree of membership of each cell to Ff is defined by the future favourability value for the species in the cell. Fp is the fuzzy set of present areas favourable to the species, and the degree of membership of each cell to Fp is defined by the present favourability value for the species in the cell. Ff þFp is the intersection between future and present favourabilities, and the degree of membership of each cell to Ff þFp is defined by the minimum of the two favourability values for the species in the cell. Ff Fp is the union between future and present favourabilities, and the degree of membership of each cell to Ff Fp is defined by the maximum of the two favourability values for the species in the cell. Positive values of increment (I ) indicate the expansion in favourability for the species (E Max[0, I ]), that is, a gain

in favourable areas, whereas negative values of I mean a net loss of areas favourable to the species (L Min[0, I ]). These features of the predicted impact of climate change on species favourability would be informative for policy planning if the coincidence between predictions for different SRESs using the same AOGCM is lower than the consistency between predictions for the same SRES applying different AOGCMs. Coincidence is here defined as the concurrence between predictions according to two SRESs for a given AOGCM and time period. These are computed as follows:

future favourability according to the circulation model ECHAM4. Mean coincidences and consistencies were compared using ANOVA after controlling for the normality of their distributions using the Kolmogorov-Smirnov test.

Results Scenarios A2 and B2 applied to the circulation model ECHAM4 produced the same values for the period 1961 1990, both for precipitation and temperature, which is why there is only one favourability model for this AOGCM in the initial period (Table 2 and 3).

The recalibrated models’ performances were similar in the recalibration and the pseudovalidation datasets (Table 4). The Cohen’s kappa and AUC values for the favourability models based on the whole dataset were intermediate between the corresponding values obtained on the recalibration and pseudovalidation datasets (Table 3 and 4). For A. dickhilleni and V. latasti, the predicted maintenance rate of the favourability was, in most cases, medium-high ( 0.70) and with slight shifts of the favourable areas (the maximum shift rate being 0.145) (Table 5, Fig. 1 and 2). However, for these species a clear loss of favourable areas was sometimes predicted, reaching net favourability loss values 0.25. For A. dickhilleni a net expansion was predicted for the beginning and middle of

Table 5. Values of the rates of overlap (O), maintenance (M ), shifting (S ), increment (I ), expansion (E ) and net loss (L) of favourability predicted for each future projection with respect to the 1961 1990 period. cFf is the cardinality of the fuzzy set of areas favourable predicted for the respective future period; c (Ff þFp) is the cardinality of the intersection between future and present favourability; and c (Ff Fp) is the cardinality of the union between future and present favourability. Species

A. dickhilleni

CGM-SRES

Period

cFf

CGCM2-A2

2011 2040 2041 2070 2071 2100 2011 2040 2041 2070 2071 2100 2011 2040 2041 2070 2071 2100 2011 2040 2041 2070 2071 2100

0.807 0.655 0.223 0.736 0.777 0.517 0.622 0.53 0.506 0.515 0.636 0.625

0.993 0.706 0.225 0.976 0.904 0.522 0.987 0.899 0.579 0.98 0.719 0.859

0.007 0.079 0.008 0.024 0.096 0.010 0.013 0.101 0.145 0.020 0.131 0.141

0.224 0.215 0.768 0.302 0.067 0.468 0.573 0.597 0.276 0.881 0.15 0.233

0.224 0 0 0.302 0.067 0 0.573 0.597 0 0.881 0 0.233

0 0.215 0.768 0 0 0.468 0 0 0.276 0 0.15 0

588.19 377.10 111.81 625.66 512.71 255.57 715.86 726.60 329.57 856.17 386.75 561.30

2011 2040 2041 2070 2071 2100 2011 2040 2041 2070 2071 2100 2011 2040 2041 2070 2071 2100 2011 2040 2041 2070 2071 2100

0.433 0.451 0.648 0.8 0.583 0.75 0.698 0.661 0.784 0.729 0.856 0.773

0.433 0.451 0.649 0.805 0.583 0.997 0.705 0.674 0.985 0.754 0.938 0.932

0.000 0.099 0.002 0.006 0.000 0.003 0.010 0.020 0.015 0.035 0.062 0.068

0.567 0.451 0.348 0.189 0.417 0.326 0.286 0.306 0.241 0.212 0.033 0.138

0 0 0 0 0 0.326 0 0 0.241 0 0.033 0.138

0.567 0.451 0.348 0.189 0.417 0 0.286 0.306 0 0.212 0 0

968.14 1229.02 1456.61 1811.27 1302.59 2962.84 1513.02 1469.10 2627.95 1669.30 2187.16 2407.90

2011 2040 2041 2070 2071 2100 2011 2040 2041 2070 2071 2100 2011 2040 2041 2070 2071 2100 2011 2040 2041 2070 2071 2100

0.858 0.741 0.689 0.799 0.731 0.684 0.77 0.462 0.678 0.713 0.533 0.586

1 1 1 1 1 1 0.97 1 0.931 1 1 1

0.000 0.000 0.000 0.000 0.000 0.000 0.030 0.000 0.069 0.000 0.000 0.000

0.166 0.35 0.45 0.252 0.368 0.462 0.229 1.166 0.305 0.403 0.878 0.705

0 0 0 0 0 0 0 0 0 0 0 0

2251.11 2605.87 2798.89 2423.85 2648.42 2830.40 2351.00 4142.15 2495.49 2683.03 3591.39 3260.56

2011 2040 2041 2070 2071 2100 2011 2040 2041 2070 2071 2100 2011 2040 2041 2070 2071 2100 2011 2040 2041 2070 2071 2100

0.731 0.588 0.365 0.742 0.652 0.648 0.818 0.705 0.522 0.797 0.643 0.561

0.999 1 1 0.999 1 0.999 0.993 0.996 1 0.988 1 1

0.001 0.000 0.000 0.001 0.000 0.001 0.007 0.004 0.000 0.012 0.000 0.000

0.366 0.701 1.742 0.345 0.533 0.543 0.206 0.408 0.916 0.227 0.555 0.781

0 0 0 0 0 0 0 0 0 0 0 0

2056.25 2561.38 4128.31 2027.69 2311.26 2325.79 1796.22 2096.12 2852.83 1826.95 2315.32 2651.99

CGCM2-B2 ECHAM4-A2 ECHAM4-B2

V. latasti

CGCM2-A2 CGCM2-B2 ECHAM4-A2 ECHAM4-B2

IBS SPECIAL

ISSUE

A. fasciata

CGCM2-A2 CGCM2-B2 ECHAM4-A2 ECHAM4-B2

C. pyrenaica

CGCM2-A2 CGCM2-A2 ECHAM4-A2 ECHAM4-B2

308

Figure 1. Favourability predicted at each 10 10 km UTM square of mainland Spain for A. dickhilleni according to each climatic model and for each period.

Discussion

309

ISSUE

Our results show that discrepancies between predictions were more attributable to the lack of knowledge concerning the effect of oceans and atmosphere on climate (general circulation models) than to the putative effect of different emission scenarios on future climate. Uncertainty is inherent to the climate system and to nature in general and, thus, cannot be avoided (Beaumont et al. 2008, Baer and Risbey 2009). However, it can be, and should be, assessed and taken into account when modelling biodiversity responses to climate change (ArauÂ´jo et al. 2005). Numerous factors contribute to the emergence of uncertainties at each step of the process leading to modelling the species response to climate change (Dormann et al. 2008), and they are transmitted to the following steps (Beaumont et al. 2007). Thus, we are addressing a kind of uncertainty which arises at the final stage of the favourability modelling procedure, but whose source resides in a previous step, namely, in the depth of our knowledge about the effects on global climate of oceans and the atmosphere, on the one hand, and greenhouse gasses, on the other. Both consistence and coincidence are inversely proportional to uncertainty. Consistence provides information about the uncertainty associated with our understanding of the basic principles governing global climate, and uncertainty should be kept low, that is, higher values of consistence are to be preferred. However, it is not the case that every kind of uncertainty is unhelpful, because that associated with the existence of different SRESs is the kind of uncertainty we, as a human

IBS SPECIAL

the century with a net favourability loss during the final three decades, whereas the opposite occurs with V. latasti (Table 5). For A. fasciata and C. pyrenaica, predicted maintenance rates of the favourable areas were very high ( 0.93) and shifting rates were very low for all the AOGCMs and SRESs used (Table 5, Fig. 3 and 4). No net loss of favourable areas was detected for them, but important favourability expansions were frequently predicted. Table 6 shows the coincidences between predictions for different SRESs using the same AOGCM, and the consistencies of results derived from different AOGCMs assuming the same SRES. Coincidence values did not differ significantly when using different AOGCMs, nor did consistence values differ significantly when using different SRESs. However, coincidences (mean 0.768, n 24) were higher than consistencies (mean 0.668, n 24) and these differences were statistically significant (t 2.29, gl 46, p 0.027). This difference could not be attributed to the different sets of climatic variables selected in the different AOGCM-SRES combinations, because after using the same variables in all the models (those obtained in the ECHAM4-A2/B2 combination) differences between coincidences (mean 0.838, n 24) and consistencies (mean 0.679, n 24) were even higher and more significant (t 4.405, gl 46, pB0.001). Coincidences were significantly lower in ectotherms when compared with endotherms (t 2.996, gl 22, p 0.007) and consistence values were also significantly lower in ectotherms (t 5.811, gl 22, p B0.001).

IBS SPECIAL

ISSUE

Figure 2. Favourability predicted at each 10 10 km UTM square of mainland Spain for V. latasti according to each climatic model and for each period.

Figure 3. Favourability predicted at each 10 10 km UTM square of mainland Spain for A. fasciata according to each climatic model and for each period.

310

Figure 4. Favourability predicted at each 10 10 km UTM square of mainland Spain for C. pyrenaica according to each climatic model and for each period.

society, can affect and, thus, is of value to emission policy planning. If the expected responses of species to different SRESs were similar, then uncertainty would be low, but this would imply that the way we control our gas emissions will not affect species distributions much. This may be useful information for policy-making, but not for greenhouse emissions policy planning, as the policy maker would have Table 6. Coincidences between predictions using different SRESs for each circulation model and consistencies between predictions for each emission scenario using different AOGCMs. Period

Coincidences

Consistencies

CGCM2

ECHAM4

2011 2040 2041 2070 2071 2100 Mean

0.873 0.735 0.434 0.680

0.776 0.532 0.587 0.632

0.720 0.507 0.333 0.520

0.638 0.667 0.444 0.583

V. latasti

2011 2040 2041 2070 2071 2100 Mean

0.532 0.824 0.492 0.616

0.872 0.670 0.882 0.808

0.511 0.582 0.496 0.530

0.624 0.532 0.669 0.609

A. fasciata

2011 2040 2041 2070 2071 2100 Mean

0.863 0.881 0.884 0.876

0.839 0.867 0.756 0.821

0.872 0.629 0.801 0.767

0.803 0.729 0.841 0.791

C. pyrenaica

2011 2040 2041 2070 2071 2100 Mean

0.915 0.902 0.563 0.794

0.964 0.871 0.918 0.918

0.791 0.772 0.691 0.751

0.812 0.800 0.775 0.796

Total mean

0.738

0.787

0.635 0.688

311

ISSUE

A. dickhilleni

IBS SPECIAL

Species

no reason to prefer, by including the expected effect of the emissions scenarios on species distribution in the cost/ benefit balance, one SRES to another. Consequently, species distribution models will be more informative in this respect if coincidences are lower. In areas with great spatial variability of climate, such as the Iberian Peninsula, it is particularly important to use reliable AOGCMs to forecast future climate change (SaÂ´nchez et al. 2004). Although our analyses did not evaluate the reliability of AOGCMs, in order for general knowledge on the effects of oceans and atmosphere to be considered reliable it is a necessary, although not sufficient, requirement that the uncertainty associated with the differences between AOGCMs to be low. In our case, competing AOGCMs differ in their simulation of average climate values, thus generating a kind of uncertainty which is implicit in the climate variables used in our models and which is transmitted to our analyses. This adds to the methodological uncertainty associated with the modelling procedures (Thuiller et al. 2008), which the modelling community can deal with through the analysis of different modelling alternatives. Given that we used the same methodology to produce all the models, it is unlikely that our modelling procedure biased the uncertainty associated with AOGCMs and SRESs. In this way, the uncertainty associated with disagreements between different AOGCMs could be assessed and compared to that associated with the existence of different SRESs. The latter should be higher than the former, as policy planners can only affect the emission of gasses producing global warming. If uncertainty

ISSUE

IBS SPECIAL

about the future distribution of areas favourable to a species is not clearly associated with the different SRESs, then choosing between them is not likely to produce the difference desired for the future distribution of the species. The use of fuzzy logic to assess the effect of climate change on species distribution bypassed the loss of information implicit in the use of a threshold for converting model output values into predicted presences and absences to be compared with actual data (Parra and Monahan 2008, Randin et al. 2009). The fact that the favourability function is a membership function was essential to calculate the increment, overlap, maintenance, shifting, expansion, and net loss of the predicted favourability for every species preserving all the information included in the individual favourability values. In this respect, our results are meant to exemplify a useful way to assess the comparative effects of AOGCMs and SRESs on simulations of future species distributions. The expected response to climate change tends to be species-specific (see, for example, Levinsky et al. 2007, Seoane and Carrascal 2008, Virkkala et al. 2008). Our results suggest a negative effect of climate change on the areas favourable to the ectotherm species A. dickhilleni and V. latasti, whereas the other two species, A. fasciata and C. pyreniaca, could benefit from a regional increase in temperature (Table 4, Fig. 1, 2, 3 and 4). Although these results concern too few species to warrant any generalization, they agree with those of Aragoń et al. (2010), who found that the influence of climate on Iberian species distributions is stronger in ectothermic vertebrates. In addition, coincidences were lower in ectotherms when compared with endotherms, which seems to indicate that the distribution of areas favourable to the more sensitive amphibians and reptiles are more likely to be affected by emission policy decisions than those of the endothermic birds and mammals. However, our consistence values were also lower in ectotherms and, thus, the overall informative value was not really higher. In summary, uncertainties related to the AOGCM employed were bigger than those related to the SRES used for every species analysed. This adds to accumulating evidence that agreement between projections using different AOGCMs is currently insufficient. Variation among different AOGCMs was found to be larger than the expected impact of the different SRESs when predicting crop growth (Audsley et al. 2006), regional climatic features (Rowell 2006, De´que´ et al. 2007, Paeth et al. 2008) or river flow regimes (Prudhomme and Davies 2009). An improvement in knowledge on the effect of oceans and atmosphere on climate is needed if really informative models are to be produced. Our conclusion is that species distribution models in climate warming scenarios are still not useful for informing emission policy planning, although they have great potential as tools once consistencies become higher than coincidences. Acknowledgements This work was partially financed by the Consejerıá de Innovacioń, Ciencia y Empresa, Junta de Andalucıá, Spain (project P05-RNM-00935) and the Ministerio de Educacioń y Ciencia of Spain and FEDER (projects CGL2006-09567 and CGL2009-11316). A. Estrada has a postdoctoral contract jointly financed by the European Social Fund and by the Junta de

312

Comunidades de Castilla-La Mancha (Spain), in the framework of the Operational Programme FSE 2007-2013. The Agencia Estatal de Meteorologı´a of Spain provided the climatic data. We are grateful to R. Hidalgo and P. Acevedo for his help with the processing of the climatic variables and the calibration of the models, respectively. We thank D. Nogue´s-Bravo and J. Lobo for their comments on a previous version of the manuscript.

References Aragoń, P. et al. 2010. The contribution of contemporary climate to ectothermic and endothermic vertebrate distributions in a glacial refuge. Global Ecol. Biogeogr. 19: 40 49. Arau´jo, M. B. et al. 2005. Reducing uncertainty in projections of extinction risk from climate change. Global Ecol. Biogeogr. 14: 529 538. Audsley, E. et al. 2006. What can scenario modelling tell us about future European scale agricultural land use, and what not? Environ. Sci. Policy 9: 148 162. Baer, P. and Risbey, J. S. 2009. Uncertainty and assessment of the issues posed by urgent climate change. An editorial comment. Clim. Change 92: 31 36. Barbosa, A. M. et al. 2003. Otter (Lutra lutra) distribution modeling at two resolution scales suited to conservation planning in the Iberian Peninsula. Biol. Conserv. 114: 377 387. Barbosa, A. M. et al. 2009. Transferability of environmental favourability models in geographic space: the case of the Iberian desman (Galemys pyrenaicus) in Portugal and Spain. Ecol. Model. 220: 747 754. Beaumont, L. J. et al. 2005. Predicting species distributions: use of climatic parameters in BIOCLIM and its impact on predictions of species’ current and future distributions. Ecol. Model. 186: 250 269. Beaumont, L. J. et al. 2007. Where will species go? Incorporating new advances in climate modelling into projections of species distributions. Global Change Biol. 13: 1368 1385. Beaumont, L. J. et al. 2008. Why is the choice of future climate scenarios for species distribution modelling important? Ecol. Lett. 11: 1135 1146. Benjamini, Y. and Hochberg, Y. 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. B 57: 289 300. Brunet, M. et al. 2007. Generacioń de escenarios de cambio clima´tico para Espanã. Ministerio de Medio Ambiente, Madrid. De Frene, P. et al. 2009. Unravelling the effects of temperature, latitude and local environment on the reproduction of forest herbs. Global Ecol. Biogeogr. 18: 641 651. De´que´, M. et al. 2007. An intercomparison of regional climate simulations for Europe: assessing uncertainties in model projections. Clim. Change 81: 53 70. Dormann, C. F. et al. 2008. Prediction uncertainty of environmental change effects on temperate European biodiversity. Ecol. Lett. 11: 235 244. Estrada, A. et al. 2008. Using crisp and fuzzy modelling to identify favourability hotspots useful to perform gap analysis. Biodivers. Conserv. 17: 857 871. Farfań, M. A. et al. 2008. Distribution modelling of wild rabbit hunting yields in its original area (S Iberian Peninsula). Ital. J. Zool. 75: 161 172. Farr, T. G. and Kobrick, M. 2000. Shuttle Radar Topography Mission produces a wealth of data. EOS Trans. Am. Geophys. Union 81: 583 585. Fielding, A. H. and Bell, J. F. 1997. A review of methods for the assessment of prediction errors in conservation presence/ absence models. Environ. Conserv. 24: 38 49.

313

ISSUE

Paeth, H. et al. 2008. Uncertainties in climate change prediction: El Ninõ-Southern Oscillation and monsoons. Global Planet Change 60: 265 288. Palomo, L. J. et al. 2007. Atlas y Libro Rojo de los Mamı´feros Terrestres de Espanã. Direccioń General para la Biodiversidad-SECEM-SECEMU, Madrid. Parra, J. L. and Monahan, W. B. 2008. Variability in 20th century climate change reconstructions and its consequences for predicting geographic responses of California mammals. Global Change Biol. 14: 2215 2231. Pauli, H. et al. 2007. Signals of range expansions and contractions of vascular plants in the high Alps: observations (1994 2004) at the GLORIA* master site Schrankogel, Tyrol, Austria. Global Change Biol. 13: 147 156. Pen˜uelas, J. and Boada, M. 2003. A global change-induced biome shift in the Montseny mountains (NE Spain). Global Change Biol. 9: 131 140. Pleguezuelos, J. M. et al. (eds) 2004. Atlas y libro rojo de los anfibios y reptiles de Espanã. Direccioń General de Conservacioń de la Naturaleza-Asociacioń Herpetolo´gica Espanõla, Madrid, Spain. Prudhomme, C. and Davies, H. 2009. Assessing uncertainties in climate change impact analyses on the river flow regimes in the UK. Part 2: future climate. Climatic Change 93: 197 222. Randin, C. et al. 2009. Climate change and plant distribution: local models predict high-elevation persistence. Global Change Biol. 15: 1557 1569. Real, R. et al. 2003. Relative importance of environment, human activity and spatial situation in determining the distribution of terrestrial mammal diversity in Argentina. J. Biogeogr. 30: 939 947. Real, R. et al. 2006. Obtaining environmental favourability functions from logistic regression. Environ Ecol. Stat. 13: 237 245. Real, R. et al. 2008a. Modelling chorotypes of invasive vertebrates in mainland Spain. Divers. Distrib. 14: 364 373. Real, R. et al. 2008b. Using chorotypes to deconstruct biogeographical and biodiversity patterns: the case of breeding waterbirds in Europe. Global Ecol. Biogeogr. 17: 735 746. Real, R. et al. 2009. Conservation biogeography of ecologicallyinteracting species: the case of the Iberian lynx and the European rabbit. Divers. Distrib. 15: 390 400. Robertson, M. P. et al. 2004. A fuzzy classification technique for predicting species’ distributions: applications using invasive alien plants and indigenous insects. Divers. Distrib. 10: 461 474. Rowell, D. P. 2006. A demonstration of the uncertainty in projections of UK climate change resulting from regional model formulation. Clim. Change 79: 243 257. Sańchez, E. et al. 2004. Future climate extreme events in the Mediterranean simulated by a regional climate model: a first approach. Global Planet Change 44: 163 180. Seoane, J. and Carrascal, L. M. 2008. Interspecific differences in population trends of Spanish birds are related to habitat and climatic preferences. Global Ecol. Biogeogr. 17: 111 121. Thuiller, W. et al. 2008. Predicting global change impacts on plant species’ distributions: future challenges. Perspect. Plant Ecol. 9: 137 152. Trivedi, M. R. et al. 2008. Potential effects of climate change on plant communities in three montane nature reserves in Scotland, UK. Biol. Conserv. 141: 1665 1675. US Geological Survey 1996. GTOPO30. Land Processes Distributed Active Archive Center (LP DAAC), EROS Data Center. <http://edcdaac.usgs.gov/gtopo30/gtopo30.asp>, accessed 22 September 1999. van der Broekhoven, E. et al. 2006. Fuzzy rule-based macroinvertebrate habitat suitability models for running waters. Ecol. Model. 198: 71 84.

IBS SPECIAL

Foody, G. M. 2008. Refining predictions of climate change impacts on plant species distribution through the use of local statistics. Ecol. Inform. 3: 228 236. Foster, P. 2001. The potential negative impacts of global climate change on tropical montane cloud forests. Earth-Sci. Rev. 55: 73 106. Garcıá, L. V. 2003. Controlling the false discovery rate in ecological research. Trends Ecol. Evol. 18: 553 554. Gevrey, M. et al. 2006. Estimating risk of events using SOM models: a case study on invasive species establishment. Ecol. Model. 197: 361 372. Giorgi, F. and Lionello, P. 2008. Climate change projections for the Mediterranean region. Global Planet Change 63: 90 104. Grabherr, G. 1994. Climate effects on mountain plants. Nature 369: 448. Hosmer, D. W. and Lemeshow, S. 2000. Applied logistic regression, 2nd ed. Wiley. I.G.N. 1999. Mapa de carreteras. Penıńsula Ibe´rica, Baleares y Canarias. Inst. Geogra´fico Nacional/Ministerio de Fomento, Madrid. IPCC 2007. Summary for policymakers climate change 2007: the physical science basis. In: Solomon, S. et al. (eds), Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge Univ. Press, pp. 1 18. Klanderud, K. and Birks, H. J. B. 2003. Recent increases in species richness and shifts in altitudinal distributions of Norwegian mountain plants. Holocene 13: 1 6. Kuncheva, L. I. 2001. Using measures of similarity and inclusion for multiple classifier fusion by decision templates. Fuzzy Sets Syst. 122: 401 407. Legendre, P. 1993. Spatial autocorrelation: trouble or new paradigm? Ecology 74: 1659 1673. Levinsky, I. et al. 2007. Potential impacts of climate change on the distributions and diversity patterns of European mammals. Biodivers. Conserv. 16: 3803 3816. Lobo, J. et al. 2008. AUC: a misleading measure of the performance of predictive distribution models. Global Ecol. Biogeogr. 17: 145 151. Luoto, M. and Heikkinen, K. 2008. Disregarding topographical heterogeneity biases species turnover assessments based on bioclimatic models. Global Change Biol. 14: 483 494. Ma´rquez, A. L. et al. 2004. Dependence of broad-scale geographical variation in fleshy-fruited plant species richness on disperser bird species richness. Global Ecol. Biogeogr. 13: 295 304. Martı´, M. and del Moral, J. C. (eds) 2003. Atlas de las aves reproductoras de Espanã. Direccioń General de Conservacioń de la Naturaleza-Sociedad Espanõla de Ornitologıá, Madrid, Spain. Munõz, A. R. and Real, R. 2006. Assessing the potential range expansion of the exotic monk parakeet in Spain. Divers. Distrib. 12: 656 665. Munõz, A. R. et al. 2005. Modelling the distribution of Bonelli’s eagle in Spain: implications for conservation planning. Divers. Distrib. 11: 477 486. Nakicenovic, N. et al. 2000. Emission scenarios. A Special Report of Working Group III of the Intergovernmental Panel on Climate Change. Cambridge Univ. Press. Nogue´s-Bravo, D. et al. 2007. Exposure of global mountain systems to climate warming during the 21st century. Global Environ. Change 17: 420 428. Nogue´s-Bravo, D. et al. 2008. Climate change in Mediterranean mountains during the 21st century. Ambio 37: 280 285. ORNL 2001. LandScan 2000 Global Population Database. Oak Ridge National Laboratory (ORNL), Oak Ridge, TN.

IBS SPECIAL

ISSUE

Virkkala, R. et al. 2008. Projected large-scale range reductions of northern-boreal land bird species due to climate change. Biol. Conserv. 141: 1343 1353. Wilson, R. J. et al. 2005. Changes to the elevational limits and extent of species ranges associated with climate change. Ecol. Lett. 8: 1138 1146.

314

Wilson, R. J. et al. 2007. An elevational shift in butterfly species richness and composition accompanying recent climate change. Global Change Biol. 13: 1873 1887. Xu, D. and Yan, H. 2001. A study of the impacts of climate change on the geographic distribution of Pinus koraiensis in China. Environ. Int. 27: 201 205.

Ecography 33: 315 320, 2010 doi: 10.1111/j.1600-0587.2010.06285.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Thorsten Wiegand. Accepted 3 February 2010

A meta-analysis of isolation by distance: relic or reference standard for landscape genetics? David G. Jenkins, Michael Carey, Justyna Czerniewska, Jennifer Fletcher, Tyler Hether, Amanda Jones, Stacy Knight, Joseph Knox, Tonya Long, Mary Mannino, Morgan McGuire, Andrea Riffle, Shannon Segelsky, Logan Shappell, Andrew Sterner, Treanna Strickler and Rosanna Tursi D. G. Jenkins (dgjenkin@mail.ucf.edu), M. Carey, J. Czerniewska, J. Fletcher, T. Hether, A. Jones, S. Knight, J. Knox, T. Long, M. Mannino, M. McGuire, A. Riffle, S. Segelsky, L. Shappell, A. Sterner, T. Strickler and R. Tursi, Dept of Biology, Univ. of Central Florida, 4000 Central Florida Blvd., Orlando, FL 32816-2368, USA.

Isolation by distance (IBD) has been a common measure of genetic structure among populations and is based on Euclidean distances among populations. Whereas IBD does not incorporate geographic complexity (e.g. dispersal barriers, corridors) that may better predict genetic structure, a new approach (landscape genetics) joins landscape ecology with population genetics to better model genetic structure. Should IBD be set aside or should it persist as the most simple model in landscape genetics? We evaluated the status of IBD by collecting and analyzing results of 240 IBD data sets among diverse taxa and study systems. IBD typically represented a low proportion of variance in genetic structure (mean r2 0.22) in part because many studies included relatively few populations (mean 11). The number of populations studied (N) was asymptotically related to IBD significance; a study with 9 populations has only 50% probability of significance, while one with 23 populations will have 90% probability of significance. Surprisingly, ectothermic animals were significantly (p 0.0018) more likely to have significant IBD than endotherms, which suggests a metabolic basis underlying gene flow rates. We also observed marginally significant effects on IBD significance for a) taxa in general and b) dispersal modes within actively-dispersing endotherms. Other factors analyzed (genetic markers, genetic distances, habitats, active or passive dispersal, plant growth form) did not significantly affect IBD, likely related to typical N. For multiple reasons we conclude that IBD should continue as the simplest reference standard against which all other, more complex models should be compared in landscape genetics research.

315

ISSUE

component of IBD estimation continues to be based on straight-line Euclidean distance among populations. With the development of landscape ecology came more advanced approaches to represent landscape complexity (Turner and Gardner 1991). The new discipline of landscape genetics (Manel et al. 2003) fuses landscape ecology with population genetics to incorporate geographic information far more sophisticated than Euclidean distance to explain genetic structure (Guillot et al. 2005, Spear et al. 2005, Holderegger and Wagner 2006, Storfer et al. 2007). Landscape genetics was recently defined as ‘‘research that explicitly quantifies the effects of landscape composition, configuration and matrix quality on gene flow and spatial genetic variation’’ (Storfer et al. 2007) and is likely to become more commonly used to analyze genetic structure. Here we treat this approach as being relevant to metapopulation genetics (Olivieri et al. 1995, Hanski and Gaggiotti 2004), given the original definition of a metapopulation as a ‘‘population of populations’’ (Hanski and Simberloff 1997).

IBS SPECIAL

Gene flow is central to population genetics because the rate of gene flow into a local population contributes to population success or failure, in concert with local selection (Holt and Gomulkiewicz 1997). The converse of gene flow is genetic isolation, and Isolation By Distance (IBD; Wright 1943) analysis has long been the standard approach to express genetic differentiation as a function of distance. Euclidean distance has been used because IBD is based on the island and stepping-stone models of population genetic structure (Wright 1943, Kimura and Weiss 1964), and distance serves as a simple, estimable proxy for the myriad factors that isolate populations. Given this rich history, much has been theorized and advanced through the years regarding IBD (Slatkin 1993, Rousset 1997, Bossart and Prowell 1998, Bohonak 1999). Throughout its history, IBD estimation evolved to include more sophisticated calculations of genetic structure ( Jensen et al. 2005) that kept pace with advances in statistical methods, increasing computer power and rapidly advancing molecular methods. However, the distance

What value then remains for IBD analyses? In one possible view, IBD may be considered as a 20th century paradigm, to be fully replaced by 21st century landscape genetics. In that case, IBD would soon be relegated to historical interest only and omitted from future analyses. Alternatively, IBD may continue to be useful as the simplest baseline method for relative evaluation of more sophisticated and difficult analyses. For example, multi-model inference (Burnham and Anderson 2002) can be used to identify population genetic models that most efficiently capture the most information, relative to other models. In that context, IBD may remain useful as the most simple model for relative comparisons among landscape genetics models. We note that IBD is not a null model, but instead predicts genetic differentiation as a function of distance; a null model invokes no such mechanism and is typically based on randomization (Gotelli and Graves 1996). Interestingly, we found no synoptic evaluation of empirical IBD analyses in the literature that may help clarify its role in an era of landscape genetics. The purpose of this meta-analysis was to evaluate the published evidence for IBD among diverse organisms to answer three questions: 1) are IBD results sensitive to study methods (genetic markers, genetic distance estimators, number of populations)? If so, results must be interpreted appropriately, and IBD analyses may have limited application. If not, IBD may be generally applicable. 2) What patterns emerge in IBD within and among diverse groups of organisms? Some taxonomic groups may be expected to have more significant and clear IBD patterns than other taxa. We do not explore detailed phylogenetic patterns here, but first compare coarse taxonomic groups. We also compared ectothermic vs endothermic animals and organisms grouped by different dispersal modes. 3) Given the above, what role might IBD analyses fill amidst landscape genetic approaches? Questions 1 and 2 were answered by statistical analyses of collected data; question 3 was answered by consideration of answers to questions 1 and 2 and properties of collected IBD studies.

IBS SPECIAL

ISSUE

Methods We collected articles during 2008 from the peer reviewed scientific literature that contained genetic distances among populations and geographic distances or location data for those populations. We make no claims that this sample of the IBD literature adequately represents any given taxon or geographic region. Instead, we consider our collection a substantial sample that should represent general patterns. The population genetics literature is diverse in methods and data reporting customs. No standard approach exists for reporting genetic and geographic distances, so we had to make decisions when processing data. Numerous papers provided only geographic coordinates or maps, which we converted to Euclidean distances among populations using great circle calculations to accurately represent distances on a spherical Earth. Some studies reported statistical outcomes (e.g. Mantel tests) but did not provide genetic or geographic distance data in those cases we analyzed Mantel test outcomes only but could not compute IBD further. 316

Genetic distances were reported in the literature as a variety of markers and statistics, though many report pairwise FST values among populations. For those data sets, we calculated genetic distance as [FST/(1 FST)] (Rousset 1997) to standardize values among studies. Because RST, FST, and uST are analogues of FST (Halliburton 2004), we calculated equivalent formulae for those genetic distance estimators. We also analyzed data sets reported using Nei’s genetic distance (D) data, but without recalculation as above. We calculated IBD using the IBD Web Service (IBDWS; Jensen et al. 2005), as did some other authors. Statistical analyses in IBDWS included Mantel tests of significant correlation between the [FST/(1 FST)] and loge(distance) matrices (Rousset 1997, Jensen et al. 2005), and results of reduced major axis (RMA) regression (slope, and the coefficient of determination, r2). RMA regression is a form of model II regression and most appropriate when the independent variable (geographic distance in our analyses) includes error (Hellberg 1994, Sokal and Rohlf 1995). RMA regression is standard in IBDWS and appropriate here because we estimated some distances from published maps or geographic coordinates, and thus included error. Regressions analyzed the relationship between [FST/(1 FST)] and loge(distance) per Rousset (1997). Slope coefficients are not often reported in IBD analyses, apparently related to concerns with non-independence of points. However, we found that published IBD plots often presented the points and slope graphically, and we considered slope to be an important measure of effect (as in any regression). The correlation coefficient r is sometimes reported in IBD analyses; we were interested in r2 because landscape genetics approaches may be viewed as an effort to increase variance ‘‘explained’’ (r2) relative to Euclidean IBD outcomes. To answer question 1, we tested the hypotheses that IBD outcomes (Mantel p value, slope, r2) varied a) with the number of populations studied (N), b) among genetic markers (e.g. allozymes, microsatellites, etc.), or c) among genetic distance estimators. Analyses of Mantel p values were based on IBDWS analyses and values reported in the literature, whereas analyses of slope and r2 were based on IBDWS analyses alone. We tested for the effect of study scope (N) using regressions, where N was the independent factor and Mantel p value, slope, or r2 were the dependent factors. We selected among several regression models using Akaike information criteria (AIC; Burnham and Anderson 2002). We tested for potential effects of genetic markers and genetic estimators on IBD outcomes in two ways: 1) with Mantel p values, slope, or r2 as continuous data (by factorial ANOVA, with markers, estimators, and marker X estimator interaction as factors), and 2) with Mantel p values represented as binary data (p B0.05 or p 0.05), where markers and estimators were tested by x2. As an additional test that controlled for potential variation in analytical method, we also conducted these same analyses on IBDWS outcomes alone. To answer question 2, we tested the hypotheses that IBD outcomes (Mantel p value, slope, r2) varied among taxa, dispersal modes of all organisms (active or passive), metabolic categories of animals (ectotherm or endotherm),

or habitats (marine, terrestrial, freshwater). We used factorial ANCOVA for these analyses, where the covariate was N studied. Finally, when considering the potential future value of IBD (question 3), we computed regressions of the relationship between Mantel p values and IBD r2 and compared regression models using AIC. We also computed a multiple regression to predict IBD r2 by Mantel p values and N. Throughout, variables were transformed as necessary for homogeneity of variance, and statistical analyses were computed with SPSS v16.

Results We obtained 240 data sets that were analyzed by Mantel tests; of those, 143 data sets were analyzed using IBDWS by us or by authors. Most data sets were recent, based on microsatellite markers and FST (Fig. 1). Though the data may not represent the full history of IBD analyses, they do represent the status of modern IBD analysis around the time that landscape genetics began. Overall, the average Mantel p value was 0.166, the mean number of populations studied (N) was 11.1, the mean IBD r2 value was 0.22, and the mean slope [(FST/(1 FST)):loge(km)] was 0.81. Of the study parameters we analyzed, only N significantly affected Mantel test p values; significance did not depend on the genetic markers, genetic distance estimators, or a marker X distance interaction in the sampled studies (Table 1). The number of populations (N) was inversely related to Mantel test p values, as one may expect; more populations in a study tended to contribute to lower Mantel test p values. The inverse relationship was significant but not highly predictive of IBD significance

ISSUE

317

IBS SPECIAL

Figure 1. (A) Most studies included in the meta-analysis were recently published, reflecting IBD immediately prior to and at the advent of landscape genetics. (B) Microsatellite (ms) data were most frequently recorded, followed by mitochondrial DNA (mt) and allozyme (allo) data. Other markers included AFLs, RAPD, minisatellites, ISSR, cpSSR, and SSCP. (C) FST genetic distance data were most commonly obtained, though FST analogs (RST, FST, and uST), Neiâ&#x20AC;&#x2122;s D, and other distances (Mhat, Nem) were also recorded.

(r2 0.104). However, logistic regression of binary Mantel test significance was significant (p B0.001) and fairly predictive (64% correctly predicted) (Table 1). The logistic prediction indicated that 9 populations were needed to achieve 50% probability of significant IBD, 17 populations were needed to achieve 75% probability of significant IBD, and 24 populations were required to achieve 90% probability of significant IBD (Fig. 2). Despite significant effects on Mantel test results, N did not significantly affect IBD r2 or slope (Table 1). Concordant with results for Mantel tests, IBD slopes and r2 values (from IBDWS analyses) also did not differ significantly among genetic markers or genetic distance estimators. In addition, we found no significant interaction between genetic markers and genetic distance estimators, meaning that IBD significance did not depend on the combination of genetic markers and genetic distance estimators used. Based on the significant effect of N on IBD statistical significance, we included log10(N) as a covariate in subsequent analyses of biological variables. Among the analyzed biological variables, taxa differed marginally for Mantel test p values when assessed as binary data (i.e. pB0.05 or p 0.05) or continuous variables (Table 1). This weak effect was revealed with a factorial ANCOVA that included the interactive effects of habitat on taxa differences, though habitats did not differ for IBD outcomes even after accounting for taxonomic effects and N. The interaction effect was due to reptiles, which occurred in all habitat categories (terrestrial, amphibious, and aquatic), as revealed by re-analysis after removing reptiles from the data (Table 1). Marginal differences among taxa for Mantel test outcomes did not translate to significant differences among taxa for IBD slopes or r2 values (Table 1). Grouped more broadly, animals and plants also did not have significantly different Mantel test outcomes (ANCOVA, p 0.576; not listed in Table 1). Several surprises were found in comparisons. IBD outcomes were not significantly different between active and passive dispersers (Table 1). Most passive-dispersing data sets represented non-animal taxa (especially plants), but no significant differences were observed among plant habits (herb, shrub, tree) for IBD outcomes (Table 1). However, ectothermic animals were significantly more likely to have significant IBD than endothermic animals (x2, p 0.018; Table 1). Within actively-dispersing ectotherms, we compared dispersal modes (walking, flying, swimming) for IBD outcomes but found no significant differences. A marginally significant difference was observed among dispersal modes of actively-dispersing endotherms, but this relatively weak effect did not translate to significant effects on slope or r2. Finally, Mantel test p values were significantly and negatively related to IBD r2 values, as might be expected, and this relationship co-varied with N (Table 1, Fig. 3). Overall, IBD studies conducted with more populations were more likely to obtain a significant correlation between genetic and geographic distances (significant Mantel p value), but were less likely to obtain a predictive IBD trend (i.e. greater r2; Fig. 3).

Table 1. Summary of statistical analyses on IBD data. Significant outcomes (p B0.05) are highlighted in bold; marginally significant outcomes are in italics. Factors tested

IBD variable

Statistical test

Outcome

Question 1: study methods No. of populations

Mantel p

Regressions and AIC

IBD slope IBD r2 Mantel p IBD slope IBD r2 Mantel p Mantel p

Regressions and AIC Regressions and AIC ANOVA ANOVA ANOVA x2 (binary) x2 (binary)

Inverse (p 0.011 (1.256/N); pB0.001; R2 0.104; AIC wi 0.72 Linear; n.s.d. (p 0.370); R2 0.005; AIC wi 0.35 Inverse; n.s.d. (p 0.125); R2 0.011; AIC wi 0.38 All factors are n.s.d. (p 0.70) All factors are n.s.d. (p 0.90) All factors are n.s.d. (p 0.20) n.s.d. (p 0.854) n.s.d. (p 0.224)

Mantel p Mantel p

x2 (binary) ANCOVA(7)

IBD slope IBD r2 Mantel p IBD slope IBD r2 Mantel p IBD slope IBD r2 Mantel p IBD slope IBD r2 Mantel p Mantel p IBD slope IBD r2

ANCOVA ANCOVA ANCOVA ANCOVA ANCOVA ANCOVA ANCOVA ANCOVA ANCOVA ANCOVA ANCOVA x2 (binary) ANCOVA ANCOVA ANCOVA

IBD r2

Regressions and AIC

IBD r2

Multiple regression (backward elimination)

Genetic markers, genetic distance estimators, and Marker Distance Interaction (1) Genetic distance estimators Genetic markers Question 2: biology Taxa Taxa, Habitat, and Taxa Habitat Interaction (2)

Dispersal mode

(3)

Actively-dispersing homeotherms

(4)

Actively-dispersing poikilotherms

(5)

Metabolism (animals)

(6)

Question 3: IBD value Mantel p loge(Mantel p) No. of populations

n.s.d. (p 0.086) Taxa: p 0.046 Habitat: n.s.d. (p 0.483) Interaction: p 0.018 All factors are n.s.d. (p 0.90) All factors are n.s.d. (p 0.90) n.s.d. (p 0.270) n.s.d. (p 0.641) n.s.d. (p 0.377) Dispersal Modes: p 0.044 Dispersal Modes: n.s.d. (p 0.402) Dispersal Modes: n.s.d. (p 0.132) Dispersal Modes: n.s.d. (p 0.130) Dispersal Modes: n.s.d. (p 0.656) Dispersal Modes: n.s.d. (p 0.580) p 0.018 n.s.d. (p 0.412) n.s.d. (p 0.412) n.s.d. (p 0.858) Logarithmic (r2 0.063 0.04*loge(p)); pB0.001; R2 0.259; AIC wi 0.35 r2 0.137 0.052*loge(p) 0.011*N; p B0.001; R2 0.368

(1)

IBS SPECIAL

ISSUE

Genetic markers (N): allozymes (41), mtDNA (42), microsatellites (128), and others (RAPD, AFLP, ISSR, cpSSR, SSCP, minisatellites; 33). Genetic distance estimators (N): FST/(1 FST) (170), FST analogs (RST, FST, uST) (31), Neiâ&#x20AC;&#x2122;s D (21). (2) Taxa (N) amphibians (33), reptiles (28), birds (33), fish (42), mammals (45), arthropods (12). The significant interaction for Mantel p values was due to reptiles; when reptiles were removed from analysis, taxa remained significantly different (p 0.042) but interaction was n.s.d. (p 0.468). (3) Dispersal mode (N) active (191) or passive (53). (4) Active dispersal modes within homeotherms (N) active flying (39), active swimming (18), and active walking (21). (5) Active dispersal modes within poikilotherms (N) active flying (13), active swimming (48), active swimming and walking (8), passive (42). (6) Animal metabolism (N) poikilotherm (112; 41 n.s.d., 71 sig. Mantel p) or homeotherm (78; 42 n.s.d., 36 sig. Mantel p). (7) The covariate (number of populations studied) was significant (pB0.05) in all ANCOVAs.

Discussion Isolation by distance analyses are based on linear distances between populations and are the simplest possible model to characterize metapopulation genetic structure. A simple model may not be expected to respond to interactive details among diverse data sets, and we found that IBD results did not respond to multiple factors. For example, IBD results were insensitive to the choice of genetic marker or genetic distance estimator. Rather than indicating that IBD is robustly diagnostic across study systems, we view this result as indicating that IBD is a fuzzy instrument due to its simplicity and the often-small study scope used. Many IBD studies have not included sufficiently large numbers of populations to be assured of detecting IBD if it exists, as evidenced by 55% probability of significant IBD for the mean number of populations studied (N 11). Studies that included more populations had greater probability to 318

observe significant IBD, but we found relatively few such studies. We found it interesting that predictive capability (r2) for IBD results was inversely correlated (though not strongly so) with N. We hypothesize that IBD analyses fail to capture nonlinear effects of complex landscapes and historical processes (e.g. postglacial dispersal pathways) on genetic structure as more populations across a landscape were included in a study. Given the advent and rapid growth of landscape genetics approaches (Manel et al. 2003, Storfer et al. 2007), this is obviously not a novel hypothesis. However, no synoptic meta-analysis of empirical IBD outcomes has existed to support this conceptual and methodological transition. The contrast between the effects of N on Mantel p and IBD r2 indicates the promise of landscape genetics for better understanding metapopulation genetic structure. Landscape genetics models that incorporate factors beyond Euclidean

Figure 2. Predicted relationship between the number of populations studied (N) and the probability of obtaining a significant IBD analysis (Mantel p value B0.05). Bold curve is the logistic regression predicted relationship. Horizontal and vertical lines relate probabilities of 50, 75 and 90% to the corresponding N required to attain each probability. See Table 1 for additional regression statistics.

distance (e.g. potential dispersal barriers) are likely to increase the variance represented (i.e. attain greater r2) to essentially fill in the empty right-hand wall of Fig. 3 and enable strongly predictive models of population genetic structure among many populations. We predict that a similar, future meta-analysis of landscape genetics outcomes will find less contrast between Mantel p and ‘‘explained’’ variance due to N, as was observed in this study of IBD outcomes. Phylogenetic and physiological traits appeared to underlie IBD patterns, if our crude taxonomic and metabolic categories were any indication. Clearly, a more sophisticated and sensitive approach to phylogenetic signatures (Webb

ISSUE

319

IBS SPECIAL

Figure 3. The predictive value (variance ‘‘explained’’) of IBD analyses (IBD r2) was a function of 1) Mantel p value (i.e. significant correlation between genetic distance and geographic distance matrices) and 2) the number of populations analyzed in a study (N). Most studies represented a combination of few populations and low Mantel p. Some of those studies attained high r2, but many did not. The goal of landscape genetics approaches may be visualized here as placing more points high on the right-hand wall (more populations and significant Mantel p values).

et al. 2002) is needed to more fully explore this possibility, given the availability of robust supertrees among diverse taxa. Whereas endothermic animals were essentially equiprobable for significant or nonsignificant IBD, ectothermic animals were nearly twice as likely to exhibit significant IBD. The difference between ectotherms and endotherms for IBD results suggests a metabolic basis for gene flow and population genetic structure. Ectotherms must modulate activity and/or location based on external temperatures, and are necessarily constrained to disperse within strict temperature limits and temporal windows of opportunity (Janzen 1967, Ghalambor et al. 2006). Endotherms may be less constrained by those same conditions, and thus significant IBD for a endothermic organism may be more heavily weighted to processes other than climatic conditions, such as physical/chemical habitat, food web structure, etc. If so, then population genetic structure of ectothermic animals may be more strongly affected by climate change than endotherms, but endotherms may be able to adjust ranges with climate change more readily (assuming other factors such as habitat fragmentation are not important). Ectotherms and endotherms have been compared for range geometry and evolutionary rates (Martin and Palumbi 1993, Pfrender et al. 1998) but macroecological differences among diverse ectotherm and endotherm lineages for gene flow have not been examined to date. Such a macroecological investigation will need to address body size because it may express potential ectotherm-endotherm differences in combination with standardized temperatures (Gillooly et al. 2001) and because it affects dispersal distances among active dispersers (Jenkins et al. 2007). Analyses of study scale effects (Gaston and Blackburn 1996) and phylogeny (Webb et al. 2002) will also be valuable to understand and predict effects of broad-scale factors (e.g. climate change, fragmentation) on genetic structure of diverse organisms. Different dispersal modes (flying, crawling, swimming) contributed to variation among active dispersing endotherms for Mantel p values, but this effect did not translate to IBD slope or r2. These traits should be influenced by phylogenetic effects, but further analysis with an even greater data set will be needed to parse the effects of taxa (e.g. birds, bats, and insects) separate from dispersal modes (e.g. flying). At this point, only weak evidence exists for differences among very different dispersal modes for metapopulation genetic isolation. We do not expect this statement to persist after landscape genetics approaches have been more fully applied. Overall, our analyses do not present a compelling case for IBD as a stand-alone analytical approach to describe population genetic structure of metapopulations; this 20th century method does not suffice in the 21st century. Indeed, IBD is typically but one of several analyses in most population genetics papers we collected, though these studies did not apply landscape genetics approaches. The relative insensitivity of IBD may be related to the generally low predictive ability of many published IBD outcomes ( 50% of r2 values were B0.20). In other words, it is difficult to detect subtleties when the picture is blurry. Despite its shortcomings, we recommend that IBD analysis continue to be a vital component of landscape genetics studies. The shortcomings we report here are partly an effect of relatively few studied populations. As more

populations are studied in large-scale landscape genetics efforts, IBD patterns may be better resolved. The logistic model presented here for IBD significance as a function of N (Fig. 3) may serve as one guideline for adjusting sampling scale to obtain more predictive IBD results, and may help guide landscape genetics analyses as well. Isolation by distance should serve as the baseline for evaluation of more complex landscape genetics models that may exceed IBD’s ability to represent spatial pattern in genetic structure. The extent that a landscape model exceeds an IBD model is the important point, and thus reveals the value of IBD to landscape genetics. We do not suggest that the basic theory underlying IBD will be overthrown, but rather that the actual dispersal distances among populations will become better estimated, so that the simple proxy (Euclidean distance) can be surpassed by more realistic dispersal pathways. Multimodel inference approaches (Burnham and Anderson 2002) should be used to compare alternative landscape genetic models to IBD, and we suggest that this approach be considered as a requirement for landscape genetics analyses. In this manner, IBD will persist as part of 21st century analyses of metapopulation genetics, but will do so in a role that highlights the added strength of landscape genetics approaches and relates 21st century approaches to those of the 20th century for continuity. Acknowledgements We thank Eric Hoffmann, Doug Bruggeman, and an anonymous reviewer for helpful comments, Andy Bohonak and colleagues for providing IBDWS as a valuable online tool for IBD analyses, and all the authors of papers we meta-analyzed for their hard work in collecting and publishing their results.

IBS SPECIAL

ISSUE

References Bohonak, A. J. 1999. Dispersal, gene flow, and population structure. Q. Rev. Biol. 74: 21 45. Bossart, J. L. and Prowell, D. P. 1998. Genetic estimates of population structure and gene flow: limitations, lessons and new directions. Trends Ecol. Evol. 13: 202 206. Burnham, K. P. and Anderson, D. R. 2002. Model selection and multimodel inference: a practical information-theoretic approach, 2nd ed. Springer. Gaston, K. J. and Blackburn, T. M. 1996. Range size-body size relationships: evidence of scale dependence. Oikos 75: 479 485. Ghalambor, C. K. et al. 2006. Are mountain passes higher in the tropics? Janzen’s hypothesis revisited. Integr. Comp. Biol. 46: 5 17. Gillooly, J. F. et al. 2001. Effects of size and temperature on metabolic rate. Science 293: 2248 2251. Gotelli, N. J. and Graves, G. R. 1996. Null models in ecology. Smithsonian Inst. Press.

320

Guillot, G. et al. 2005. A spatial statistical model for landscape genetics. Genetics 170: 1261 1280. Halliburton, R. 2004. Introduction to population genetics. Benjamin Cummings. Hanski, I. and Gaggiotti, O. E. 2004. Ecology, genetics and evolution of metapopulations. Academic Press. Hanski, I. and Simberloff, D. 1997. The metapopulation approach, its history, conceptual domain, and application to conservation. In: Hanski, I. and Gilpin, M. E. (eds), Metapopulation biology: ecology, genetics, and evolution. Academic Press, pp. 5 26. Hellberg, M. E. 1994. Relationships between inferred levels of gene flow and geographic distance in a philopatric coral, Balanophyllia elegans. Evolution 48: 1829 1854. Holderegger, R. and Wagner, H. H. 2006. A brief guide to landscape genetics. Landscape Ecol. 21: 793 796. Holt, R. D. and Gomulkiewicz, R. 1997. How does immigration influence local adaptation? A reexamination of a familiar paradigm. Am. Nat. 149: 563 572. Janzen, D. H. 1967. Why mountain passes are higher in the tropics. Am. Nat. 101: 233 249. Jenkins, D. G. et al. 2007. Does size matter for dispersal distance? Global Ecol. Biogeogr. 16: 415 425. Jensen, J. L. et al. 2005. Isolation by distance, web service. BMC Genet. 6: 13, v.3.16, <http://ibdws.sdsu.edu/>. Kimura, M. and Weiss, G. H. 1964. The stepping stone model of population structure and the decrease of genetic correlation with distance. Genetics 49: 561 576. Manel, S. et al. 2003. Landscape genetics: combining landscape ecology and population genetics. Trends Ecol. Evol. 18: 189 197. Martin, A. P. and Palumbi, S. R. 1993. Body size, metabolic rate, generation time, and the molecular clock. Proc. Nat. Acad. Sci. USA 90: 4087 4091. Oliveri, I. 1995. Metapopulation genetics and the evolution of dispersal. Am. Nat. 146: 202 228. Pfrender, M. E. et al. 1998. Patterns in the geographical range sizes of ectotherms in North America. Oecologia 115: 439 444. Rousset, F. 1997. Genetic differentiation and estimation of gene flow from f-statistics under isolation by distance. Genetics 145: 1219 1228. Slatkin, M. 1993. Isolation by distance in equilibrium and nonequilibrium populations. Evolution 47: 264 279. Sokal, R. R. and Rohlf, F. J. 1995. Biometry: the principles and practice of statistics in biological research. W H Freeman. Spear, S. F. et al. 2005. Landscape genetics of the blotched tiger salamander (Ambystoma tigrinum melanostictum). Mol. Ecol. 14: 2553 2564. Storfer, A. et al. 2007. Putting the ‘landscape’ in landscape genetics. Heredity 98: 128 142. Turner, M. G. and Gardner, R. H. 1991. Quantitative methods in landscape ecology. Ecological Studies Series, Springer. Webb, C. O. et al. 2002. Phylogenies and community ecology. Annu. Rev. Ecol. Syst. 33: 475 505. Wright, S. 1943. Isolation by distance. Genetics 28: 114 138.

Ecography 33: 321 325, 2010 doi: 10.1111/j.1600-0587.2010.06544.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: David Nogue´s-Bravo. Accepted 21 February 2010

SYMPOSIUM

Integrating pattern with process at biogeographic boundaries: the legacy of Wallace

Society in Me´rida, Me´xico. Here, we develop a background to these papers (Cody et al. 2010, Daza et al. 2010, Morrone 2010, Smith and Klicka 2010) by summarizing several highlights in the historical focus of biogeographers on boundaries.

Wallace’s Line: the iconic biogeographic boundary

321

ISSUE

Wallace was not satisfied with providing only a detailed description of the pattern of biotic transition across Wallace’s Line he went on to speculate on the geological histories that underlay these differences: ‘‘I believe the western part to be a separated portion of continental Asia, the eastern the fragmentary prolongation of a former Pacific continent’’ (Wallace letter to Henry Bates in 1858 as reported by Berry 2002). At the same time that Wallace was delineating this geographically abrupt transition, he also was keenly aware that there were ongoing biogeographic processes that revealed a more complex situation: ‘‘The separation between these two regions [Sundaland and Wallacea] is not so absolute. There is some transition. There are species and genera common to the eastern and western islands’’ (Wallace 1860, p. 175). His grappling with these complexities reached its zenith when contemplating the history of the fauna on the island of Sulawesi (Celebes): ‘‘Its fauna presents the most puzzling relations, showing affinities to Java, to the Philippines, to the Moluccas, to New Guinea, to continental India, and even to Africa; so that it is almost impossible to decide whether to place it in the Oriental or the Australian region’’ (Wallace 1876 v. I, p. 389). Beyond the more purely biogeographic observations of the patterns of biotic transition, Wallace took advantage of the wealth of biological diversity across islands within the Malay Archipelago to generate fundamental insights into the processes of evolutionary diversification and of speciation. In a paper now known as the Sarawak Law paper, he used an array of biogeographic and geological observations

IBS SPECIAL

Biogeography is a vital discipline today because of its extraordinarily integrative nature, drawing from and informing biological and Earth sciences in order to explain the history and future of life on our planet. Yet, even as we continue to build more sophisticated syntheses using molecular genetics, GIS-based distribution modelling, and ever-better analytical and visualizing approaches, we should recall that exploring causal connections between biological and Earth history is not a particularly new endeavor. For example, the biogeographic principles advocated in the late 1800s by Alfred Russel Wallace (see Box 2.1 in Lomolino et al. 2006) were infused with ideas associating distributional and diversification histories of organisms with geology and climate. But even a century before Wallace, in 1761 Compte de Buffon had recognized the differences between mammals in the New World and Old World tropics and proposed a rudimentary evolutionary causation for their divergence and distribution based on separation of formerly united continents. Certainly, de Buffon and Wallace were not alone during their times in describing the non-random distributions of animals and plants, exploring causal explanations (see contributions reproduced in the ‘‘Early Classics’’ section of Lomolino et al. 2004), and recognizing regions of rapid transition between geographically distinct biotas. Wallace stood out, however, in the magnitude and synthetic nature of his focus on a single biogeographic boundary. After spending eight years exploring the mosaic of islands that lay between southeast Asia and New Guinea/Australia, Wallace concluded that ‘‘ . . .we may consider it established that the Strait of Lombock (only 15 miles wide) marks the limits and abruptly separates two of the great zoological regions of the world’’ (Wallace 1860, pp. 173 174). He called this region the Malay Archipelago and it includes the most famous biogeographic transition of all, named Wallace’s Line by T. H. Huxley in 1868. The four papers in this Special Feature were first presented in January 2009 in the ‘‘Patterns and Processes at Biogeographic Boundaries’’ symposium convened at the 4th Biennial Meeting of the International Biogeography

BIOGEOGRAPHIC BOUNDARIES

B. R. Riddle (brett.riddle@unlv.edu), School of Life Sciences, Univ. of Nevada Las Vegas, Las Vegas, NV 89154-4004, USA. D. J. Hafner, Museum of Southwestern Biology, Univ. of New Mexico, Albuquerque, NM 87131, USA.

Brett R. Riddle and David J. Hafner

BIOGEOGRAPHIC BOUNDARIES ON

SYMPOSIUM ISSUE

IBS SPECIAL

to conclude that ‘‘Every species has come into existence coincident both in space and time with a pre-existing closely allied species’’ (Wallace 1855, p. 186). In the Ternate paper (Wallace 1858) he outlined the role of natural selection in the production of those new species. With these and a host of other papers, and with books including The Malay Archipelago in 1869 and Island Life in 1880, he firmly established the empirical importance of this biogeographic transition zone during the nascent years of the Darwinian revolution in biology. Throughout the 20th century, evolutionary biologists and ecologists built upon the foundation that Wallace had established in the Malay Archipelago, with a good deal of concern for establishing the pattern of transition (e.g. the lines of Wallace, modified Wallace, Huxley, modified Huxley, Sclater, Weber [favored by Mayr 1944], Lydekker, etc.). Perhaps of most lasting influence, this biogeographic transition zone provided core evolutionary insights for Ernst Mayr (1942), and also played a major role in the formulation of new and often controversial models in evolutionary and ecological biogeography (Wilson 1959, Diamond 1974, 1975).

Beyond Wallace’s Line: the Nearctic Neotropical transition zone

Wallace established three fundamental attributes of the Malay Archipelago that subsequently focused the attention of biogeographers on other biogeographic boundaries. The region marked an abrupt transition between long-isolated biotas, each of which represented hotspots of diversification and speciation, and complex patterns of interchange resulted as those once-isolated biotas collided. Wallace recognized similar atributes at the Mexican sub-Region of the Neotropical Region as it forms a boundary with the Nearctic Region. He used his vast compilations of both marine and terrestrial biotic affinities to postulate ‘‘it almost certain that the union of North and South America is comparatively a recent occurrence, and that during the Miocene and Pliocene periods, they were separated by a wide arm of the sea . . .[and that] when the evidence of both land and sea animals support each other as they do here, the conclusions arrived at are almost as certain as if we had (as we no doubt some day shall have) geological proof of these successive subsidences’’ (Wallace 1876 v. II, pp. 57 59). Wallace relied on evidence from a relatively good mammalian fossil record to interpret biogeographic pattern and ecological consequences following the invasion of North American by a South American mammalian biota: ‘‘We have here unmistakable evidence of an extensive immigration from South into North America, not very long before the beginning of the Glacial epoch . . . How such large yet defenceless animals as tapirs and great terrestrial sloths, could have made their way into a country abounding in large felines equal in size and destructiveness to the lion and the tiger, with numerous wolves and bears of the largest size, is a great mystery . . . and the fact that no such migration had occurred for countless preceding ages, proves that some great barrier to the entrance of terrestrial mammalia which had previously existed, must for a time have been removed’’ (Wallace 1876 v. I, pp. 131 132). 322

In his reliance on mammals, Wallace initiated what developed throughout the 20th century into a dominate bias in the study of Late Cenozoic biogeographic and evolutionary dynamics between the Nearctic and Neotropical regions. Other notable observations that emphasized the Pliocene interchange of mammals between North and South America included those of Karl A. von Zittel (1891) and Hermann von Ihering (1900). However, the modern stage was set by George Gaylord Simpson (1950), who famously sorted South American mammals into three temporally distinct faunas that he postulated to represent waves of immigration, most likely from North America: earliest Paleocene Ancient Immigrants (Oldtimers); Late Eocene to Miocene Island Hoppers; and late Tertiary to Recent Newcomers, the South American component of what now is known as the Great American Biotic Interchange (GABI). Simpson (1950), like Wallace, was clearly interested in both the ecological and biogeographic attributes of northern and southern faunas that might explain the pattern of Neotropical and Nearctic interchange that he inferred: ‘‘Middle America is a faunal filter . . . Its ecological characteristics . . . determined which stocks were involved in faunal interchanges between North and South America and which are now immobilized to the north and to the south. The filtering action is not sharply localized. It begins well to the north (and west), roughly at the edge of the Lower Sonoran life zone in southwestern United States, and also reaches far to the south and east, more or less to the edge of the Guiana highlands and thence southward and westward’’ (Simpson 1950, p. 387). Simpson’s attempt to delineate the geography and ecology of the transition zone between Nearctic and Neotropical biotas was continued into the latter 20th century primarily by entomologist Gonzalos Halffter: ‘‘I have defined as the Mexican Transition Zone . . . part of the southwestern United States, all of Mexico, and a large part of Central America extending to the Nicaraguan lowlands’’ (Halffter 1987, p. 95). Halffter’s Mexican Transition Zone has subsequently been refined through contributions including Ortega and Arita (1998) and many studies by Juan J. Morrone and his collaborators. It became a hallmark of the GABI that modern mammal biotas in South America have a much higher percentage of taxa with northern ancestry than do biotas in North America with South American ancestry. Simpson surmised that mammals with northern ancestries were so successful in South America for two reasons. First, he considered them poised for a great invasion of South America following development of a land bridge because they had been adapting to tropical ecosystems in Central America, similar to those of northern South America, throughout the Cenozoic. Second, once in South America, the northern invaders were better competitors because of their long evolutionary history of being tested and surviving within the intense milieu of the World Continent: ‘‘When ecological vicars met, one or the other generally became extinct . . . Those [northern invaders] extant in the Plio-Pleistocene were the ones that had been successful in a long series of competitive episodes. They were specialists in invasion and in meeting competitive invaders’’ (Simpson 1950, pp. 382 383).

Integrating pattern with process at biogeographic boundaries

Advances in molecular biogeography and phylogeography

Advances in reconstructing Earth history

Finally, the contribution from Morrone synthesizes a number of studies that collectively have shed light on biogeographic pattern across the Mexican Transition Zone, the 323

ISSUE

Advances in the integrative nature of biogeography

IBS SPECIAL

A geology infused by plate tectonics has brought into focus the dynamic relationship between plate boundaries and biogeographic boundaries. Many of the grandest biogeographic boundaries, including those discussed in detail here as well as the northward movements of India and Africa, represent collision boundaries between Gondwanan and Laurasian continental plates and adjacent oceanic plates, producing complex mosaics of island arcs, accreted terranes, mountain chains, and uplifted plateaus. Wallace surely would have been satisfied in knowing how closely the first part of his speculation about the geological history of the Malay Archipelago region fit with modern reconstructions of Pleistocene cycles of connectivity and isolation of islands on the Sunda Shelf (Sundaland) with each other and with continental Asia (Voris 2000). However, he was not correct in postulating a former Pacific continent, and would have been astonished by the northward migration of the Australian plate during the Cenozoic Era, and the production of island arc volcanoes, accreted terranes, and minor plates along the leading edges of the Australian and Pacific oceanic plates (Wallacea) as they subduct under the Euarasian continental plate (Hall 2001, 2002). One can wonder how his puzzlement about the complex biota of Sulawesi would have changed had he known that this island represents the tectonic suturing of Gondwanan and Laurasian terranes during the Late Miocene (Hall 2001). At the Nearctic Neotropical boundary, Wallace’s prediction that great advances would allow geology to one day catch up to the strong biogeographic signal of recent suturing of North and South American biotas has been realized, although with some differences in details (Coates and Obando 1996, Iturralde-Vinent 2006, Kirby et al. 2008). Of great importance to biogeographers attempting to decipher the complexity of this biotic history are

Modern studies of the biogeography of boundaries have been revolutionized with molecular genetics-based biogeography and phylogeography (Riddle et al. 2008) for several reasons. First, biogeographers can estimate phylogenetic and phylogeographic structure with great accuracy; often revealing a wealth of cryptic diversity in taxa within and across a boundary region. Second, biogeographers now have the capacity to address historical and ecological patterns in those taxa within a co-distributed biota that do not have a good fossil record. Third, many of today’s studies use properties of molecular evolution to estimate independently timeframes of isolation, interchange, and diversification. The Malay Archipelago coincides closely with delineations identified variously as the Indo-Malay Archipelago, Indo-Malay-Philippine Archipelago, Indo-West Pacific, Indo-Australian Archipelago, Malesia, and Coral Triangle (Hoeksema 2007). Biogeographers are using molecular approaches to reveal patterns of diversity and processes of diversification within the Archipelago in both terrestrial (Heaney et al. 2005, van Welzen et al. 2005, Newbound et al. 2008, Uy et al. 2009) and marine (Barber et al. 2000, Briggs 2000, Lourie and Vincent 2004, Carpenter and Springer 2005, Hoeksema 2007, Williams and Duda 2008, Bellwood and Meyer 2009) taxa and biotas, providing critical information for biodiversity protection in one of the world’s great hotspots. Turning again to questions of isolation, interchange, and diversification across the Nearctic Neotropical transition zone, the contributions here by Smith and Klicka and by Cody et al. offer a complementary pair of analyses illustrating how biogeographers can use time-calibrated molecular phylogenies to begin to truly put the ‘‘Biotic’’ into the GABI, addressing the generality of the mammalbased patterns across a suite of non-mammalian taxa. Daza et al. also employ time-calibrated molecular phylogenies within a novel analytical approach to illustrate advances in our potential to address hypotheses of historical vicariance on the complex geographic tapestry of Middle America.

BIOGEOGRAPHIC BOUNDARIES

With this brief background, we are now in a position to ask: what properties of biogeographic boundaries continue to draw the focus of modern biogeographers? We suggest that minimally the following three advances are highly relevant to addressing this question, and are illustrated variously in the collection of studies presented in this Special Feature.

emerging details about the timeframes and processes of landscape evolution, including the uplift of the northern Andes, the complex upland and lowland geological mosaic in Central America, development of the TransMexican Volcanic Belt, uplift of the mountain ranges and plateaus in northern Mexico, and opening of the Gulf of California. The four papers in this volume illustrate how today’s studies of interactions within and across boundaries are being addressed within an interactive framework that can reciprocally inform biogeographic and geological reconstructions.

SYMPOSIUM

With his synthesis of biogeography, evolution, and ecology, Simpson laid the foundation for subsequent decades of investigations into the GABI led by the paleontologists S. David Webb and Larry G. Marshall. More information began to emerge on non-mammalian groups, and the ‘‘state-of-the-art’’ was summarized in a compendium edited by Stehli and Webb (1985), followed by updated data and models during the 1990s for mammals (Vrba 1992, Webb 2006), reptiles (Cadle and Greene 1993), and plants (Burnham and Graham 1999).

BIOGEOGRAPHIC BOUNDARIES ON

SYMPOSIUM ISSUE

IBS SPECIAL

core region of biotic reticulation between Nearctic and Neotropical elements. A fundamental contribution here lies in his synopsis of a step-wise, integrative approach that is an attempt to break through the log-jam of debate over approaches and concepts that has plagued historical biogeography for decades. Refreshingly, Morrone acknowledges that a multitude of legitimate questions reside within biogeography, and that different approaches are optimized to address different questions, but that an ultimate goal might be to assemble each into a series of analytical steps within a fully integrative research program (see also Riddle and Hafner 2006) that seeks to develop a geobiotic scenario for a given biogeographic system. One can, for example, view each of the other three contributions here as important contributions to, but not the entire story of, a fully realized geobiotic scenario for the Nearctic Neotropical transition zone.

Biogeographic boundaries across Earth Beyond the Malay Archipelago and Nearctic Neotropical transition zone, biogeographers are actively exploring patterns and processes at a host of other biogeographic boundaries. For example, molecular-based studies are being combined with ecological, climatic, and fossil information to unravel the geobiotic scenarios of isolation, interchange and diversification between Palaearctic and Nearctic biotas across Beringia (Debruyne et al. 2008, DeChaine 2008, Elias and Crocker 2008, Haile et al. 2009); between the Indian subcontinent and Eurasia (Bossuyt and Milinkovitch 2001, Conti et al. 2002, Van Bocxlaer et al. 2006); and between Africa and Eurasia (Voelker 2002, Kodandaramaiah and Wahlberg 2007). We hope the papers in this Special Feature prove to be a valuable demonstration of several of the ways in which emerging approaches and concepts are allowing biogeographers to fully explore the fascinating patterns and processes at biogeographic boundaries that Alfred Russel Wallace could only begin to tap at the dawn of biogeography some 150 yr ago.

References Barber, P. H. et al. 2000. Biogeography a marine Wallace’s line? Nature 406: 692 693. Bellwood, D. R. and Meyer, C. P. 2009. Searching for heat in a marine biodiversity hotspot. J. Biogeogr. 36: 569 576. Berry, A. 2002. Infinite tropics: an Alfred Russel Wallace anthology. Verso. Bossuyt, F. and Milinkovitch, M. C. 2001. Amphibians as indicators of early tertiary ‘‘out-of-India’’ dispersal of vertebrates. Science 292: 93 95. Briggs, J. C. 2000. Centrifugal speciation and centres of origin. J. Biogeogr. 27: 1183 1188. Burnham, R. J. and Graham, A. 1999. The history of neotropical vegetation: new developments and status. Ann. Missouri Bot. Gard. 86: 546 589. Cadle, J. E. and Greene, H. W. 1993. Phylogenetic patterns, biogeography, and the ecological structure of Neotropical snake assemblages. In: Ricklefs, R. E. and Schluter, D. (eds),

324

Species diversity in ecological communities: historical and geographical perspectives. Univ. of Chicago Press, pp. 281 293. Carpenter, K. E. and Springer, V. G. 2005. The center of the center of marine shore fish biodiversity: the Philippine Islands. Environ. Biol. Fish. 72: 467 480. Coates, A. G. and Obando, J. A. 1996. The geologic evolution of the Central American Isthmus. In: Jackson, J. B. C. et al. (eds), Evolution and environment in tropical America. Univ. of Chicago Press, pp. 21 56. Cody, S. et al. 2010. The great American biotic interchange revisited. Ecography 33: 326 332. Conti, E. et al. 2002. Early tertiary out-of-India dispersal of Crypteroniaceae: evidence from phylogeny and molecular dating. Evolution 56: 1931 1942. Daza, J. M. et al. 2010. Using regional comparative phylogeographic data from snake lineages to infer historical processes in Middle America. Ecography 33: 343 354. Debruyne, R. et al. 2008. Out of America: ancient DNA evidence for a new world origin of late quaternary woolly mammoths. Curr. Biol. 18: 1320 1326. DeChaine, E. G. 2008. A bridge or a barrier? Beringia’s influence on the distribution and diversity of tundra plants. Plant Ecol. Divers. 1: 197 207. Diamond, J. M. 1974. Colonization of exploded volcanic islands by birds: the supertramp strategy. Science 184: 803 806. Diamond, J. M. 1975. Assembly of species communities. In: Cody, M. L. and Diamond, J. M. (eds), Ecology and evolution of communities. Belknap Press, pp. 342 444. Elias, S. A. and Crocker, B. 2008. The Bering Land Bridge: a moisture barrier to the dispersal of steppe-tundra biota? Quat. Sci. Rev. 27: 2473 2483. Haile, J. et al. 2009. Ancient DNA reveals late survival of mammoths and horse in interior Alaska. Proc. Nat. Acad. Sci. USA 106: 22352 22357. Halffter, G. 1987. Biogeography of the montane entomofauna of Mexico and Central America. Annu Rev. Entomol. 32: 95 114. Hall, R. 2001. Cenozoic reconstructions of SE Asia and the SW Pacific: changing patterns of land and sea. In: Metcalfe, I. et al. (eds), Faunal and floral migrations and evolution in SE Asia Australasia. A. A. Balkema (Swets & Zeitlinger Publ.), pp. 35 56. Hall, R. 2002. Cenozoic geological and plate tectonic evolution of SE Asia and the SW Pacific: computer-based reconstructions, model and animations. J. Asian Earth Sci. 20: 353 431. Heaney, L. R. et al. 2005. The roles of geological history and colonization abilities in genetic differentiation between mammalian populations in the Philippine archipelago. J. Biogeogr. 32: 229 247. Hoeksema, B. W. 2007. Delineation of the Indo-Malayan centre of maximum marine biodiversity: the coral triangle. In: Renema, W. (ed.), Biogeography, time, and place: distributions, barriers, and islands. Springer, pp. 117 178. Iturralde-Vinent, M. A. 2006. Meso-Cenozoic Caribbean paleogeography: implications for the historical biogeography of the region. Int. Geol. Rev. 48: 791 827. Kirby, M. X. et al. 2008. Lower Miocene stratigraphy along the Panama Canal and its bearing on the Central American Peninsula. PLoS One 3: e2791. Kodandaramaiah, U. and Wahlberg, N. 2007. Out-of-Africa origin and dispersal-mediated diversification of the butterfly genus Junonia (Nymphalidae: Nymphalinae). J. Evol. Biol. 20: 2181 2191. Lomolino, M. V. et al. (eds) 2004. Foundations of biogeography: classic papers with commentaries. Univ. of Chicago Press. Lomolino, M. V. et al. 2006. Biogeography, 3rd ed. Sinauer.

BIOGEOGRAPHIC BOUNDARIES

Van Bocxlaer, I. et al. 2006. Late Cretaceous vicariance in Gondwanan amphibians. PLoS One 1: e74. van Welzen, P. C. et al. 2005. Plant distribution and plate tectonics in Malesia. Biol. Skrifter 55: 199 217. Voelker, G. 2002. Systematics and historical biogeography of wagtails: dispersal versus vicariance revisited. Condor 104: 725 739. von Ihering, H. 1900. The history of the neotropical region. Science 12: 857 864. von Zittel, K. A. 1891. Handbuch der Palaeontologie, vol. 4, Band Vertebrata (Mammalia). Ouldenberg, Munich. Voris, H. K. 2000. Maps of Pleistocene sea levels in southeast Asia: shorelines, river systems and time durations. J. Biogeogr. 27: 1153 1167. Vrba, E. S. 1992. Mammals as a key to evolutionary-theory. J. Mammal. 73: 1 28. Wallace, A. R. 1855. On the law which has regulated the introduction of new species. Ann. Mag. Nat. Hist. 16: 184 196. Wallace, A. R. 1858. On the tendency of varieties to depart indefinitely from the original type. J. Proc. Linn. Soc. (Zool.) 3: 53 62. Wallace, A. R. 1860. On the zoological geography of the Malay Archipelago. J. Proc. Linn. Soc. (Zool.) 4: 172 184. Wallace, A. R. 1876. The geographical distribution of animals. Macmillan. Webb, S. D. 2006. The Great American Biotic Interchange: patterns and processes. Ann. Missouri Bot. Gard. 93: 245 257. Williams, S. T. and Duda, T. F. 2008. Did tectonic activity stimulate Oligo-Miocene speciation in the Indo-West Pacific? Evolution 62: 1618 1634. Wilson, E. O. 1959. Adaptive shift and dispersal in a tropical ant fauna. Evolution 13: 122 144.

SYMPOSIUM

Lourie, S. A. and Vincent, A. C. J. 2004. A marine fish follows Wallaceâ&#x20AC;&#x2122;s Line: the phylogeography of the three-spot seahorse (Hippocampus trimaculatus, Syngnathidae, Teleostei) in southeast Asia. J. Biogeogr. 31: 1975 1985. Mayr, E. 1942. Systematics and the Origin of Species. Columbia Univ. Press. Mayr, E. 1944. Wallaceâ&#x20AC;&#x2122;s Line in the light of recent zoogeographic studies. Q. Rev. Biol. 19: 1 14. Morrone, J. J. 2010. Fundamental biogeographic patterns across the Mexican Transition Zone: an evolutionary approach. Ecography 33: 355 361. Newbound, C. N. et al. 2008. Markedly discordant mitochondrial DNA and allozyme phylogenies of tube-nosed fruit bats, Nyctimene, at the Australian-oriental biogeographical interface. Biol. J. Linn. Soc. 93: 589 602. Ortega, J. and Arita, H. T. 1998. Neotropical Nearctic limits in Middle America as determined by distributions of bats. J. Mammal. 79: 772 783. Riddle, B. R. and Hafner, D. J. 2006. A step-wise approach to integrating phylogeographic and phylogenetic biogeographic perspectives on the history of a core North American warm deserts biota. J. Arid Environ. 65: 435 461. Riddle, B. R. et al. 2008. The role of molecular genetics in sculpting the future of integrative biogeography. Prog. Phys. Geogr. 32: 173 202. Simpson, G. G. 1950. History of the fauna of Latin America. Am. Sci. 38: 361 389. Smith, B. T. and Klicka, J. 2010. The profound influence of the Late Pliocene Panamanian uplift on the exchange, diversification, and distribution of New World birds. Ecography 33: 333 342. Stehli, F. G. and Webb, S. D. (eds) 1985. The Great American Biotic Interchange. Plenum Press. Uy, J. A. C. et al. 2009. Plumage and song differences mediate species recognition between incipient flycather species of the Solomon Islands. Evolution 63: 153 164.

IBS SPECIAL ISSUE

325

The Great American Biotic Interchange revisited Sarah Cody, James E. Richardson, Valentı´ Rull, Christopher Ellis and R. Toby Pennington S. Cody, The Royal Botanic Garden, 20a Inverleith Row, Edinburgh, EH3 5LR, UK and Dept of Botany and Molecular Evolution, Senckenberg Research Inst., Senckenberganlage 25, DE-60325 Frankfurt, Germany. J. E. Richardson, The Royal Botanic Garden, 20a Inverleith Row, Edinburgh, EH3 5LR, UK and Univ. de Los Andes, Depto de Ciencias Biolo´gicas, Cra 1A No. 18A-10, Edificio J Piso 4, Bogota´, Colombia. V. Rull, Consejo Superior de Investigaciones Cientı´ficas (CSIC), Inst. Bota`nic de Barcelona (IBB), Palynology and Paleoecology Group, Passeig del Migdia s/n, ES-08038 Barcelona, Spain. C. Ellis and R. T. Pennington (t.pennington@rbge.ac.uk), The Royal Botanic Garden, 20a Inverleith Row, Edinburgh, EH3 5LR, UK.

The ‘‘Great American Biotic Interchange’’ (GABI) is regarded as a defining event in the biogeography of the Americas. It is hypothesized to have occurred when the Isthmus of Panama closed ca three million years ago (Ma), ending the isolation of South America and permitting the mixing of its biota with that of North America. This view of the GABI is based largely upon the animal fossil record, but recent molecular biogeographic studies of plants that show repeated instances of long-distance dispersal over major oceanic barriers suggest that perhaps the land bridge provided by the isthmus may have been less necessary for plant migration. Here we show that plants have significantly earlier divergence time estimates than animals for historical migration events across the Isthmus of Panama region. This difference in timing indicates that plants had a greater propensity for dispersal over the isthmus before its closure compared with animals. The GABI was therefore asynchronous for plants and animals, which has fundamental implications for the historical assembly of tropical biomes in the most species-rich forests on the planet.

IBS SPECIAL

ISSUE

SYMPOSIUM

BIOGEOGRAPHIC BOUNDARIES

Ecography 33: 326 332, 2010 doi: 10.1111/j.1600-0587.2010.06327.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: David Nogue´s-Bravo. Accepted 8 February 2010

The Isthmus of Panama exists today as a narrow strip of land that separates the Pacific Ocean and the Caribbean Sea, linking together South and Central America. Coates and Obando (1996) suggested that early in its formation, ca 15 Ma, the nascent isthmus consisted of a series of volcanic islands, arranged in an arc between South and Central America, which were formed as a result of the subduction of the Pacific plate under the Caribbean plate. The deposition of water-borne sediment filled in the gaps between them and complete closure of the isthmus occurred around three Ma. An alternative view suggests that the Isthmus of Panama existed initially as a peninsula of southern Central America, as early as 19 Ma, rather than as an island chain (Kirby et al. 2008). The peninsula was, however, separated from South America by a deep marine channel before connecting with it three Ma. Both models estimate that the completion of the land bridge occurred approximately three Ma, so if organisms used the landbridge to move between South and Central America, we estimate that migration must have occurred after this date. The predominant biogeographic paradigm regarding the Panama Isthmus is that its closure ended the ‘‘splendid’’ isolation of South America (Simpson 1980), causing an exchange of biota regarded as so fundamental that is has become termed the ‘‘Great American Biotic Interchange’’ (GABI; Stehli and Webb 1985). The original view of the GABI was based largely upon the mammalian fossil record 326

(e.g. reviews by Simpson 1980, Stehli and Webb 1985, Webb and Rancy 1996, Burnham and Graham 1999), which indicates few migrations before the late Pliocene and the Pleistocene (Burnham and Graham 1999). The palaeobotanical record, which is poorer than the animal fossil record and based principally on pollen (Burnham and Graham 1999), upholds this view of the GABI. Evidence derived from similarities in palynomorphs indicates that, like the mammalian faunas, North and South American floras were distinct until the late Tertiary, with greater similarities developing in the Plio-Pleistocene and especially in the Pleistocene (Graham 1992, Burnham and Graham 1999). The closure of the Panama Isthmus is therefore considered to be a defining event in the neotropical biogeography of both plants and animals (Gentry 1982, Wendt 1993, Burnham and Graham 1999). Dated molecular phylogenies allow us to determine the times at which sister lineages split from each other. The conventional view of the GABI, implying a special role for a continuous terrestrial corridor for migration of biota, is called into question since numerous recently published dated molecular biogeographic studies have shown that historical long-distance dispersal over major oceanic barriers has to be repeatedly invoked to explain plant distributions (Richardson et al. 2004, Lavin et al. 2005; reviewed by Pennington et al. 2006). Related to this, an intriguing result that has emerged from two biogeographic meta-analyses of

327

ISSUE

Figure 1. Hypothetical phylogeny of plant species endemic to South (black branches) and Central America (dotted branches) showing geographical structure around the Panama Isthmus. Branch lengths are proportional to time. The Central American clade nested within a South American group indicates migration from south to north. ‘‘S’’ and ‘‘C’’ indicate stem and crown nodes of the Central American clade. The movement from South to Central America could have occurred at any time between the stem and crown nodes (along the dashed line) and therefore the date of the stem node is the earliest possible time of south north movement.

IBS SPECIAL

BIOGEOGRAPHIC BOUNDARIES

We compiled information from 40 published and three unpublished dated molecular phylogenetic studies of terrestrial plant and animal lineages (Supplementary material Table S1), mostly drawn from a recent review paper by Rull (2008). Of these 16 were plant studies and the rest were of animals. Parsimony optimization of a two state

Materials and methods

geographic character (Central America vs South America) was used to infer unambiguous migration events across the Isthmus of Panama region. Where optimization was ambiguous, nodes were not considered in the analysis. In determining the timing of migration the distinction between stem and crown nodes needs to be highlighted (Fig. 1) because the date of the stem node is the earliest possible time of migration (i.e. a maximum estimate) and the date of the crown node the latest (i.e. a minimum estimate; see Fig. 1 and caption for more details). Because in some cases the difference in age between the stem and crown node is large, there is considerable uncertainty of the exact timing of the dispersal event. Therefore, where possible, both the minimum and maximum divergence times of both stem and crown nodes that indicate transcontinental migration were recorded along with the genes used and the methods of calibration and dating (Supplementary material Table S1 and associated list of references). Methods commonly used to date phylogenies are reviewed in Renner (2005) and Rutschmann (2006). It is evident from studies that date their phylogenies using more than one approach that different age estimates for the same node on the topology can be obtained (Benton and Ayala 2003, Bromham and Penny 2003). We therefore developed criteria for choosing amongst alternative options. Studies that used the Isthmus of Panama as a calibration to date the phylogeny were discarded since they already assume the effects of the closure rather than testing it. Studies that did not produce a phylogeny but dated lineage divergence through the conversion of genetic distance into ages using a substitution rate calculated from other taxa (a ‘‘borrowed’’ rate) were dismissed because clear identification of ‘‘isthmus nodes’’ was lacking. Explicitly phylogenetic studies that used a borrowed rate were considered acceptable, but dated phylogenies calibrated using external evidence were preferred. In such dated phylogenies, where possible the date chosen was the one that was calibrated using fossils rather than geological events because of the possibility of dispersal (Renner 2005). In dated phylogenies calibrated using external evidence, in the face of different dates resulting from different methods (penalized likelihood (Sanderson 2002), Bayesian (Thorne and Kishino 2002), nonparametric rate smoothing (Sanderson 1997)) we favoured dates calculated by penalized likelihood and Bayesian methods because NPRS has been shown to over-estimate dates (Lavin et al. 2005). Supplementary material Table S1 gives information concerning taxa, mode of calibration, analytical approach, and dates of stem and crown nodes (with confidence intervals). We assessed the influence of the completion of the Isthmus of Panama land bridge on plant and animal migration events by calculating the probability that each transcontinental crossing occurred before the closure of the Isthmus of Panama using two different measures, both of which pose the question: what is the probability of migration from Central to South America, or vice versa, before three Ma when the isthmus closed? The first approach, ‘‘raw probability’’, determines the probability (p 0 1) of migration over the isthmus before its closure (Fig. 2 and 3). The second approach, ‘‘accumulated likelihood’’, is a function of the time elapsed between the

SYMPOSIUM

multiple lineages in temperate and subtropical biomes (Sanmartıń et al. 2001, Donoghue and Smith 2004, Sanmartıń and Ronquist 2004) is a possible difference in the relative frequency of dispersal in the biogeographic history of animals and plants. In a study of dated phylogenies of 18 plant and 54 animal clades from the southern hemisphere, Sanmartıń and Ronquist (2004) discovered that dates for the same geographic divergences were older for animals than for plants. In general, tectonic events better explained the animal patterns, and greater amounts of dispersal had to be invoked to explain the plant patterns. A similar result was reported for dated phylogenies of 66 plant and 39 animal clades containing disjunctions between the temperate forests of north America and eastern Asia (Sanmartıń et al. 2001, Donoghue and Smith 2004), which was interpreted as the outcome of more recent overwater dispersal for plants compared to overland migration by animals. In this paper we carry out a meta-analysis of dated molecular phylogenies of animal and plant clades that are distributed in both South and Central America in order to examine whether migrations coincided with the formation of the Isthmus of Panama. In these phylogenies, we searched for instances in which the geographic structure of sister lineages are such that we can infer an unambiguous migration between South and Central America or vice-versa (Fig. 1). Our principal goal was to investigate whether the timing of migration across the Isthmus region differed in plants and animals, and whether both depended upon the completion of a terrestrial migration corridor as the conventional model of the GABI implies.

Potential range of ‘arrival’ = 3 MY 9

BIOGEOGRAPHIC BOUNDARIES

SYMPOSIUM

Potential range of ‘arrival’ = 2 MY

Probability of arrival before closure = 1

Probability of arrival before closure = (distance in time/age range) = 3

= 4 MY

Land-bridge

Figure 2. Calculation of raw probability values. Taxa with a range of divergence times that are older than three Ma will receive a raw probability value of one.

Figure 4. Calculation of accumulated likelihood values. Accumulated likelihood values take into account how much earlier or later than the closure of the Isthmus of Panama migration occurred.

migration of a species and closure of the isthmus landbridge, relative to phylogenetic uncertainty expressed as the estimated age range between minimum and maximum divergence times (Fig. 4). In addition to these approaches, differences in the estimated divergence times were statistically compared for both stem and crown node data between animal and plant taxa. For each taxon, a mid-point was calculated between the maximum and minimum estimated divergence; midpoints were compared between plants and animals using a nonparametric Mann-Whitney U test.

that are less than three Ma, receive a probability value of zero because these divergence time estimates indicate migration over the isthmus after its closure. Taxa that have a range of node values that intersect the three Ma mark have scores calculated by assuming an equal probability of trans-isthmian crossing throughout this range of dates and then estimating the probability of migration before land bridge closure. These probabilities will lie between zero and one (Fig. 3). For example, taxa with a minimum divergence time of two Ma and a maximum divergence time of seven Ma could have crossed the isthmus before it closed at three Ma, or they could have crossed the isthmus after it closed. The raw probability of crossing before isthmus closure (Pr) is the proportion of the age range that lies before three Ma:

Raw probability This method assigns a probability of one to those taxa that have minimum node values that are greater than three Ma and so indicate migration over the isthmus before its closure (Fig. 2). Conversely, those taxa with maximum node values

ISSUE

8 7

Potential range of ‘arrival’ = 5 MY

IBS SPECIAL

where Mx is the maximum divergence time (earliest potential age of arrival), Mn is the minimum divergence time (latest potential age of arrival) and I is the age of closure (three Ma). In our worked example, Mx 7 and Mn 2 Ma, giving a Pr value of 0.8: which is the raw probability that this taxon crossed the isthmus before three Ma. To compare between plants (with fewer studies) and animals (with a greater range of available study data), values of Pr were plotted along a standardized axis of rank position (standardized between zero and 1). Accumulated likelihood values

Land-bridge Probability of arrival after closure = 0.2

1 0

Figure 3. Calculation of raw probability values. Taxa with a range of divergence time estimates that extend both before and after three Ma will receive a raw probability between zero and one. The precise figure will depend on the proportion of the range that is older than three Ma. In the case illustrated the taxon will receive a raw probability of 0.8.

328

(1)

Probability of arrival before closure = 0.8

4 3

Pr (Mx I )=(Mx Mn)

Raw probabilities are constrained to values between zero and 1, and do not take into consideration how much time before or after the closure of the isthmus migration occurred. For example, taxa exhibiting minimum divergence times that indicate trans-continental crossing 45 Ma are given the same raw probability score (Pr 1) as those taxa exhibiting minimum divergence times that correspond to a crossing 3.5 Ma. Given the potential broad time-frame in the dating process it is quite conceivable that the taxon with a divergence time of 45 Ma is more likely to have crossed the isthmus before it had closed, compared to a

Pc(n) (Mn I )=(Mx Mn)

(3)

Pc(i) f[Mn (Mx Mn)=2] I g=[(Mx Mn)=2]

(4)

Accordingly, we used eq. 2 (Pc(p)) where the minimum node value is greater than three Ma, eq. 3 (Pc(n)) where the maximum node value is less than three Ma, and eq. 4 (Pc(i)) where the maximum node value is greater than three Ma and the minimum node value is less than three Ma. Note that eq. 2 is functionally equivalent to eq. 1, and that Pc scores are in fact a cumulative measure of Pr . For example, using eq. 3, taxa exhibiting divergence time estimates that indicate migration soon after the closure of the isthmus are given lower (i.e. near zero) negative probability values than those taxa that crossed a long time after the closure of the isthmus. The more negative the probability, the less likely that migration occurred before closure of the isthmus. Where maximum and minimum node data are identical (i.e. Mx Mn 0) we used a default value of 0.3 Ma (equivalent to the 10th percentile for all differences between values of Mx and Mn). To compare between plants (with fewer studies) and animals (with a greater range of available study data), values of Pc were

We identified 58 unambiguous migration events across the Isthmus of Panama region in 16 phylogenies for plants and 27 for animals. Figure 5 indicates that plants generally have older stem node ages than animals. Eight of the 25 plant migrations (32%) have stem node divergence dates that are over 20 Ma whereas all animals have stem node divergence dates that are below 20 Ma. The entire range of stem node divergence times for plants spans nearly 50 Ma compared with less than 20 Ma for animals. Based on minimum stem node values, plants have been making cross-continental migrations between South and Central America for the past 50 Ma whereas the majority of animals (31/33; 94%) did not embark on this journey until after ten Ma. Only 19 (33%) plant and animal taxa out of 58 have stem node age ranges that indicate crossing after the closure of the isthmus. Of these, 13 are animals and six are plants. Crown node data in Fig. 6 reveal the same distinction between plant and animal divergence times, albeit somewhat less clearly. Of the 46% of crown node taxa that crossed after closure, six are plants, and 11 are animals. Of those taxa that have divergence times indicating migration across the seaway before ten Ma (Fig. 5 and 6), the majority are plants with the exception of eels (Synbranchidae), rodents (Sigmodontinae) and bats (Balantiopteryx). Of these animals with divergence dates over ten Ma, a bat may potentially fly across and the eel has been shown to tolerate salt water. Our analyses indicate that plants are more likely to have migrated over the isthmus prior to its closure three Ma (Fig. 7 and 8). The left-skewed plant data in Fig. 7 indicates a greater likelihood that plant taxa crossed the isthmus before it closed. Figure 8 shows that accumulated likelihood scores for plants lie above the trend of animals; this is particularly

BIOGEOGRAPHIC BOUNDARIES

(2)

Results

Pc(p) (Mx I )=(Mx Mn)

plotted along a standardized axis of rank position (standardized between zero and 1).

SYMPOSIUM

taxon with a divergence time of 3.5 Ma. For this reason, we also used an â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;accumulated likelihood valueâ&#x20AC;&#x2122;â&#x20AC;&#x2122; (Pc) that assigns higher scores to taxa that indicate migration much before isthmus closure and lower scores to taxa with divergence time estimates that indicate a crossing much closer to the timing of the closure of the isthmus. These scores are calculated by determining the distance between maximum or minimum node values and the three Ma closure date, and then dividing this value by the range between the extreme node values (reflecting uncertainty in the phylogenetic time-frame). This calculation estimates the number of times the node age range fits into the distance between the date of migration and the date of isthmus closure (Fig. 4).

IBS SPECIAL ISSUE

Figure 5. Chronologically arranged stem node divergence times for animals (filled bars) and plants (unfilled bars).

329

apparent towards the right hand side of the graph where divergence dates tend towards the older end of the spectrum. The non-parametric Mann-Whitney U test is consistent with the results of all other analyses. For both stem and

0.8

0.6

0.4

0.2 Animals Plants

0.0 0.0

0.2

0.4

0.6

0.8

1.0

(b) 1.0 0.8

Raw probability

IBS SPECIAL

ISSUE

Rank (proportion of sample)

0.6

0.4

0.2 Animals Plants

0.0 0.0

0.2

0.4

0.6

0.8

1.0

Rank (proportion of sample)

Figure 7. Raw probability scores based on plant and animal stem node (a) and crown node data (b) indicating the probability of migration over the isthmus before its closure.

330

crown node data, the estimated plant divergence times occur significantly earlier than for equivalent animal taxa: U 215.5, p 0.002 (stem node data) and U 101.1, p 0.033 (crown node data).

Discussion

(a) 1.0

Raw probability

BIOGEOGRAPHIC BOUNDARIES ON

SYMPOSIUM

Figure 6. Chronologically arranged crown node divergence time estimates for animals (filled bars) and plants (unfilled bars).

Our results show clearly that lineages of plants had a greater overall likelihood of crossing the Isthmus of Panama region before it closed, implying a new view for the GABI that plants tended to move first. Interestingly, the majority of plant and animal phylogenies indicate trans-continental migration prior to the closure of the isthmus (39 plant and animal taxa with minimum stem node ages that are older than three Ma out of 58 plant and animal taxa in the stem node data set, i.e. 67%). These findings indicate one of two things: either both plants and animals had a greater than expected ability to disperse over water between South and Central America before three Ma; or alternatively, an earlier land connection, such as the one hypothesized by Bermingham and Martin (1998) may be plausible. Bermingham and Martin (1998) used divergence time estimates from freshwater fish that were intolerant of salt water to postulate that a land bridge must have existed between South and Central America three to seven Ma. Because no geological evidence has been found to support the Bermingham-Martin hypothesis, we suggest that our results indicate that both plants and animals commonly crossed the narrowing isthmus region before its final closure. A statistically significant difference in timing of migration between plants and animals is evident from both stem and crown node ages (Fig. 5, 6, 7, 8). That the crown node data display essentially the same pattern is important with regard to migration in relation to the specific date of the closure of the Isthmus of Panama because the crown node dates are minimum estimates and therefore a more conservative test of whether migration occurred before the Isthmus closed.

(a) 80 60

–20 0.0

0.2

0.4

0.6

0.8

1.0

(b) 60 Accumulated likelihood

Animals Plants

40 30 20 10 0 –10 0.0

0.2

0.4

0.6

Rank (proportion of sample)

Figure 8. Accumulated likelihood values based on stem node (a) and crown node data (b). Scores greater than zero indicate a balanced migration probability before the isthmus closed and scores below zero indicate a migration probability after the isthmus closed. The greater the positive score, the earlier the minimum divergence date is before three Ma, and the narrower the estimated range in phylogenetic divergence dates. The greater the negative score, the later the maximum divergence date is after three Ma.

References

331

ISSUE

Benton, M. J. and Ayala, F. J. 2003. Dating the tree of Life. Science 300: 1698 1700. Bermingham, E. and Martin, A. P. 1998. Comparative mtDNA phylogeography of neotropical freshwater fishes: testing shared history to infer the evolutionary landscape of lower Central America. Mol. Ecol. 7: 499 517. Bromham, L. and Penny, D. 2003. The modern molecular clock. Nat. Rev. Genet. 4: 216 224. Burnham, R. J. and Graham, A. 1999. The history of Neotropical vegetation: new developments and status. Ann. Missouri Bot. Gard. 86: 546 589. Coates, A. G. and Obando, J. A. 1996. The geologic evolution of the Central American Isthmus. In: Jackson, J. B. C. et al. (eds), Evolution and environment in tropical America. Chicago Univ. Press, pp. 21 56. Donoghue, M. J. and Smith, S. A. 2004. Patterns in the assembly of temperate forests around the northern Hemisphere. Phil. Trans. R. Soc. B 359: 1633 1644.

IBS SPECIAL

We acknowledge several possible sources of error, including inadequate sampling of taxa, varying dating methodology and different means of calibration when drawing comparisons between phylogenetic dates reported in literature (Renner 2005, Rutschmann 2006). In particular, our animal dataset is is dominated by vertebrates, especially birds, and in future could be improved by the addition of more invertebrates. In this context, it is interesting to note that the animal dataset of Sanmartı´n et al. (2001), which was dominated by studies of insects and lacking in studies of vertebrates, still showed evidence for less historical dispersal than a corresponding plant dataset for the same disjunction in the northern hemisphere (Donoghue and Smith 2004, Pennington et al. 2006). This lends support to the conclusion of the present study, which is clearly consistent with plants having a greater capacity than animals for dispersal, and subsequent establishment of founder populations in new territories. Earlier over-water migration by plants is likely to reflect the ability of their propagules to traverse large distances and to establish founder populations successfully. Seeds are often produced in great numbers and some are particularly well adapted to wind or water dispersal. Seed dormancy means

Acknowledgements We thank our editor David Nogue´s Bravo and two anonymous reviewers for their valuable input. Jane Squirrel is thanked for assistance with statistical analyses. Paulina Hechenleitner, Alexandra Muellner and Josh Clayton are thanked for providing unpublished data. The Univ. of Los Andes, Bogota, Colombia is thanked for hosting JER during a six month sabbatical period (August January 2008) which enabled completion of his contribution to this work.

BIOGEOGRAPHIC BOUNDARIES

1.0

0.8

Rank (proportion of sample)

that they can remain viable even after a long journey and need not germinate until conditions are favorable. Perhaps more crucially, flexibility of reproduction either via vegetative propagation or self-fertilization means that some plants can colonize without needing a sexual partner. The few meta-analyses contrasting the biogeographic patterns of plants and animals (this paper, Sanmartı´n et al. 2001, Donoghue and Smith 2004, Sanmartin and Ronquist 2004) are limited in their taxonomic scope and suffer methodological problems such as phylogenetic dates calculated using a variety of methods (Pennington et al. 2006). Despite these shortcomings, they indicate that in the temperate and subtropical northern (Sanmartı´n et al. 2001, Donoghue and Smith 2004) and southern hemispheres (Sanmartin and Ronquist 2004), and in the tropics (this paper), that plants show a greater tendency for dispersal than animals and have geographic disjunctions that occurred at times not correlated with specific geological events. If this pattern is general it has substantial implications for models of community assembly over evolutionary time (Donoghue and Smith 2004), envisaging resident faunas needing to be resilient to more dynamic floras. In the case of the Isthmus of Panama, the different biogeographies of plants and animals across this region has major implications for our understanding of biome assembly in the Neotropics, the most species-rich region in the world. The better ability of plants to disperse and establish means that forests such as in Amazonia, even when South America was an isolated continental island prior to its land connection with Central America, were subject to invasion by ‘‘foreign’’ plant species and resident animals and plants would constantly have had to evolve in response to this rain of immigrants.

SYMPOSIUM

Accumulated likelihood

Animals Plants

BIOGEOGRAPHIC BOUNDARIES ON

Download the Supplementary material as file E6327 from <www.oikos.ekol.lu.se/appendix>.

IBS SPECIAL

ISSUE

SYMPOSIUM

Gentry, A. H. 1982. Neotropical floristic diversity: phytogeographical connections between Central and South America, Pleistocene climatic fluctuations, or an accident of Andean orogeny? Ann. Mo. Bot. Gard. 69: 557 593. Graham, A. 1992. Utilisation of the Isthmian landbridge during the Cenozoic palaeobotanical evidence for timing, and the selective influence of altitudes and climate. Rev. Palaeobot. Palynol. 72: 119 128. Kirby, M. X. et al. 2008. Lower Miocene stratigraphy along the Panama Canal and its bearing on the Central American Peninsula. PLoS One 3: 2791. Lavin, M. et al. 2005. Evolutionary rates analysis of Leguminosae implicates a rapid diversification of lineages during the Tertiary. Syst. Biol. 54: 575 594. Pennington, R. T. et al. 2006. Insights into the historical construction of species-rich biomes from dated plant phylogenies, neutral ecological theory and phylogenetic community structure. New Phytol. 172: 605 616. Renner, S. S. 2005. Relaxed molecular clocks for dating historical plant dispersal events. Trends Plant Sci. 10: 550 558. Richardson, J. E. et al. 2004. Historical biogeography of two cosmopolitan families of flowering plants: Annonaceae and Rhamnaceae. Phil. Trans. R. Soc. B 359: 1495 1508. Rull, V. 2008. Speciation timing and neotropical biodiversity: the Tertiary-Quaternary debate in the light of molecular phylogenetic evidence. Mol. Ecol. 17: 2722 2729. Rutschmann, F. 2006. Molecular dating of phylogenetic trees: a brief review of current methods that estimate divergence times. Divers. Distrib. 12: 35 48.

332

Sanderson, M. J. 1997. A nonparametric approach to estimating divergence times in the absence of rate constancy. Mol. Biol. Evol. 14: 1218 1231. Sanderson, M. J. 2002. Estimating absolute rates of molecular evolution and divergence times: a penalized likelihood approach. Mol. Biol. Evol. 19: 101 109. Sanmart覺織n, I. and Ronquist, F. 2004. Southern hemisphere biogeography inferred by event-based models: plant versus animal patterns. Syst. Biol. 53: 216 243. Sanmart覺織n, I. et al. 2001. Patterns of animal dispersal, vicariance and diversification in the Holarctic. Biol. J. Linn. Soc. 73: 345 390. Simpson, G. G. 1980. Splendid isolation: the curious history of South American mammals. Yale Univ. Press. Stehli, F. G. and Webb, D. S. 1985. The great American biotic interchange. Plenum Press. Thorne, J. L. and Kishino, H. 2002. Divergence time and evolutionary rate estimation with multilocus data. Syst. Biol. 51: 689 702. Webb, S. D. and Rancy, A. 1996. Late Cenozoic evolution of the neotropical mammal fauna. In: Jackson, J. B. C. et al. (eds), Evolution and environment in tropical America. Univ. Chicago Press, pp. 335 358. Wendt, T. 1993. Composition, floristic affinities and origins of the Mexican Atlantic slope rain forests. In: Ramamoorthy, T. P. et al. (eds), Biological diversity of Mexico, origins and distribution. Oxford Univ. Press, pp. 595 680.

Ecography 33: 333 342, 2010 doi: 10.1111/j.1600-0587.2009.06335.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Carsten Rahbek. Accepted 4 December 2009

Separated throughout most of the Cenozoic era, North and South America were joined during the mid-Pliocene when the uplift of Panama formed a land bridge between these two continents. The fossil record indicates that this connection allowed an unprecedented degree of inter-continental exchange to occur between unique, previously isolated biotic assemblages, a phenomenon now recognized as the ‘‘Great American Biotic Interchange’’. However, a relatively poor avian fossil record has prevented our understanding the role of the land bridge in shaping New World avian communities. To address the question of avian participation in the GABI, we compiled 64 avian phylogenetic studies and applied a relaxed molecular clock to estimate the timing of trans-isthmus diversification events. Here, we show that a significant pulse of avian interchange occurred in concert with the isthmus uplift. The avian exchange was temporally consistent with the well understood mammalian interchange, despite the presumed greater vagility of birds. Birds inhabiting a variety of habitats and elevational zones responded to the newly available corridor. Within the tropics, exchange was equal in both directions although between extratropical and tropical regions it was not. Avian lineages with Nearctic origins have repeatedly invaded the tropics and radiated throughout South America; whereas, lineages with South American tropical origins remain largely restricted to the confines of the Neotropical region. This previously unrecognized pattern of asymmetric niche conservatism may represent an important and underappreciated contributor to the latitude diversity gradient.

333

ISSUE

each continent soon after the formation of the Panamanian Isthmus (Webb 1976); 2) an early wave of xeric-adapted species was followed by a second wave of mesic-adapted species (Webb and Rancy 1996); 3) there was an equal exchange of taxa between continents (Marshall et al. 1982, Webb and Marshall 1982); and 4) most exchange had ceased by the onset of the mid-Pleistocene (Webb 1976). Over time, lineages moving into South America were more ‘‘successful’’ than those moving in the opposite direction. Today, 50% of modern South American genera have North American origins whereas only about 10% of North American mammal genera are derived from southern immigrants (Webb and Marshall 1982). Avian bones are thin and light relative to those of mammals and as a consequence fossilize poorly. The depauperate nature of the avian fossil record precludes a direct fossil-based comparison of the responses of birds and mammals to the land bridge formation (Mayr 1964, Olson 1985, Vuilleumier 1985). The few avian fossils available do suggest some interchange during the late Cenozoic, although the timing is unclear (Vuilleumier 1985). Furthermore, there is the perception that because birds can

IBS SPECIAL

The continents of North and South America have been isolated from one another throughout most of their histories. The fossil record indicates that a transient Beringian connection between North America and eastern Asia existed throughout much of the Cenozoic (Simpson 1947, Tiffney and Manchester 2001). Unlike North America, South America was an island continent during this time, allowing for the evolution of an ancient and largely endemic biota (Patterson and Pascual 1972, Simpson 1980, Vuilleumier 1984, Ricklefs 2002). Species assemblages unique to each continent met abruptly some 3 million years ago (mya, Coates and Obando 1996) upon the final uplift of southern Central America and the formation of a new terrestrial corridor, the Panamanian land bridge. This uplift initiated a process that led to unprecedented ecological and evolutionary consequences for previously isolated biotas (Simpson 1980), an event now referred to as the Great American Biotic Interchange (GABI). The mammalian response to the land bridge formation was written from the rich fossil deposits on both continents. Among the many evolutionary inferences gleaned from the mammalian fossil record are: 1) immigrant taxa appeared on

BIOGEOGRAPHIC BOUNDARIES

B. T. Smith (btsmith@unlv.nevada.edu), School of Life Sciences, Univ. of Nevada, Las Vegas, 4505 S. Maryland Parkway Box 454004, Las Vegas, NV 89154, USA, and Marjorie Barrick Museum of Natural History, Univ. of Nevada, Las Vegas, 4505 S. Maryland Parkway Box 454012, Las Vegas, NV 89154-4012, USA. J. Klicka, Marjorie Barrick Museum of Natural History, Univ. of Nevada, Las Vegas, 4505 S. Maryland Parkway Box 454012, Las Vegas, NV 89154-4012, USA.

Brian Tilston Smith and John Klicka

SYMPOSIUM

The profound influence of the Late Pliocene Panamanian uplift on the exchange, diversification, and distribution of New World birds

BIOGEOGRAPHIC BOUNDARIES ON

SYMPOSIUM

fly, their dispersal abilities differ from non-volant organisms such as terrestrial mammals (Lomolino et al. 2006). The colonization of distant islands by birds provides ample evidence that they can cross water barriers. The overwater dispersal abilities of some groups of birds are well known (‘‘tramps’’ sensu Mayr and Diamond 2001; Turdus thrushes, Voelker et al. 2009). However, other species of birds, such as those found in the Neotropical forest understory, do not readily disperse across water gaps (Hayes and Sewlal 2004, Moore et al. 2008) and the propensity for such dispersal is simply unknown for most birds. Thus, one cannot determine from the evidence at hand whether avian lineages regularly crossed between North and South America prior to the land bridge completion. Intercontinental dispersal may be approximated by identifying trans-isthmus diversification events in well resolved molecular phylogenies and then applying a molecular clock. Although, molecular clocks are not without controversy (see Methods) they provide a previously unavailable perspective on how birds responded to the presence of a terrestrial corridor connecting these previously isolated continents. We approach this question in two ways. First, we used molecular data and a phylogenetic tree-based approach to summarize the timing and patterns of recent avian trans-isthmus exchange. We then compared discernible patterns to those inferred from the fossil record for mammals and we evaluated dispersal direction and the influence of elevation and habitat preference on exchange. Second, we used a taxonomic approach in which we summarized and contrasted the historical and present-day distributions of Nearctic and Neotropical families to understand how the deep history of avian groups has affected present-day diversity patterns of New World birds. When combined, our analyses allowed us to determine the relative role of the Panama land bridge formation on avifaunal exchange at different temporal and taxonomic scales.

IBS SPECIAL

ISSUE

Materials and methods We performed a literature search for molecular phylogenies on New World avifauna. Studies were selected if they had complete or nearly complete taxon sampling and if the constituent taxa included both North and South American lineages. Suitable datasets were downloaded from Genbank using the program Geneious v. 2.5.4 (Drummond et al. 2006). In all, we compiled 64 molecular phylogenetic studies of New World birds, summarizing those events in which sister lineages have diverged after intercontinental dispersal. The birds that we studied are an ecologically and taxonomically diverse group that includes representation from 11 orders, 34 families, and over 100 genera (complete taxon list provided in Supplementary material). Range evolution In order to identify trans-isthmus diversification events we reconstructed phylogenetic trees for each dataset in order to perform range evolution analyses. To select a model of 334

sequence evolution for each dataset we used the program MrModeltest v. 2.3 (Nylander 2004) and the Akaike information criterion (AIC). We constructed Maximum Likelihood phylogenetic trees in the program PAUP* 4.0b 10 (Swofford 2002) using the heuristic tree search, (TBR branch swapping and 10 random addition replicates). Tree topology was verified by comparison to the original published tree. We then performed a likelihood ratio test for clock-like evolution. For trees that were not clock-like, we used the program BEAST v. 1.4.8 (Drummond and Rambaut 2007) and applied a relaxed molecular clock (molecular rates are discussed below) with a Yule process speciation prior. Each analysis was initially run for 10 000 000 generations and sampled every 1000. We used the program Tracer v. 1.4 (Rambaut and Drummond 2007) to assess the mixing of MCMC chains in the analyses and to determine the burn-in. If the runs did not converge or we did not achieve reliable ESS values ( 200) we re-ran the analysis for more generations. All trees generated prior to the point of stationarity were discarded. In cases where the greater topology was unresolved, we reduced the dataset to the clade of interest. To estimate the direction of dispersal we employed the dispersal-extinction-cladogenesis (DEC) model (Ree and Smith 2008) in the program Lagrange 2.0.1 (Ree et al. 2005). We defined six geographically relevant New World regions; North America (from Panama, north of the Darien, to Alaska), South America, the Greater Antilles, the Lesser Antilles, the Hawaiian Archipelago, and the Galapagos Archipelago. For each dataset, we set the basal divergence time to 1.0 because we were only interested in the relative timing of events for this particular analysis. We chose not to place dispersal time constraints on the analyses in order to optimize the data and allow all practical range evolution scenarios to occur. On trees that had an unresolved pre-isthmus ancestral area, we placed an ancestral area constraint to reduce uncertainty concerning the geographic origin of the clade. To justify the use of such a constraint, we performed ancestral state reconstruction using parsimony in Mesquite v. 2.6 (Maddison and Maddison 2009). Constraints were applied only if ancestral state reconstruction inferred that the basal node in a phylogeny was unambiguously assigned to one continent, and that node was older than 4 million years. This procedure eliminated a discontiguous ancestral area in both North and South America prior to the isthmus uplift. Diversification times For each phylogeny, we identified diversification (cladogenetic) events between North and South American taxa that represent a dispersal event across the isthmus. We recognize that a lag time between dispersal and diversification events exists, and that the duration of this time varies across species depending on such factors as ecology and dispersal ability. For this study, we assumed that the lag time between dispersal and lineage diversification was negligible with respect to the evolutionary time scale of the interchange. We reasoned that although that some intercontinental gene flow likely persisted for a

The estimated time period of the final isthmus uplift and the subsequent formation of a terrestrial corridor have been inferred from multiple lines of evidence. Although most researchers agree that the Panamanian land bridge was present by the mid-Pliocene, there is no consensus on precisely when this event occurred. Evidence based on marine deposits suggests that the connection between the Caribbean and Pacific was cut off by ca 3.1 mya (Coates and Obando 1996). Radio-isotopic dating from mammalian fossil beds indicates an approximate isthmus closure at 3 mya (Marshall et al. 1979); however, these same mammalian data have been interpreted as evidence for a more recent isthmus closure at 2.5 mya (Stehli and Webb 1985). The most recent geological estimate indicated that North and South America may have been connected as early as 4 mya (Kirby et al. 2008). To account for the uncertainty in isthmus closure times, we used two different estimates for the beginning of the post-land bridge period, the widely used date of 3.1 mya (Coates and Obando 1996) and the most recently published estimate of 4 mya (Kirby et al. 2008).

BIOGEOGRAPHIC BOUNDARIES

Isthmus of Panama completion time estimates

(Arbogast and Slowinski 1998); nor, did we address time dependent rate variation (Ho et al. 2005). Both issues potentially affect a small proportion of our estimates in the tail of our distribution. Had we addressed these potential errors, a subset of points would be moved toward the present, strengthening our interpretation of the results. We did not account for coalescent variance (Edwards and Beerli 2000). Simulation studies have shown that a single vicariant event can generate a wide range of genetic distances among geminate species pairs (Hickerson et al. 2006). Discrete diversification pulses that were undetected with our analyses may have occurred during the GABI. Finally, any lag time between gene and population divergence (Edwards and Beerli 2000) would cause some of our estimates, particularly those associated with more recent diversification events, to be younger than we estimated.

SYMPOSIUM

short time after dispersal occurred, the relatively linear configuration of southern Central America, the geographic bottleneck at its contact point with South America, and the climatic and geologic turmoil that occurred at the time of the isthmus uplift led to relatively rapid genetic isolation. To improve precision, mitochondrial DNA (mtDNA) sequences of all the descendant lineages arising from each dispersal/diversification event (called diversification events hereafter) were re-run through MrModeltest v. 2.3 (Nylander 2004) to obtain a sequence model that was unaffected by the phylogeny it was nested in. We reran each individual dataset through BEAST following the same methodology as described above. To evaluate the robustness of our results, we additionally performed analyses with alternative clock calibration times, a uniform clock, strict clocks, as well as uncorrected genetic distances. Four different mitochondrial genes were used in the trans-isthmus dataset: cytochrome b (cyt-b), cytochrome oxidase I (COI), NADH dehydrogenase subunit II (ND2) and ATP-synthase 6 and 8 (ATPase6&8). We assessed the variability of rates among mtDNA loci by using pairwise comparisons of average intrageneric genetic distances for each gene used (Fig. 1). These analyses indicated that cyt-b and COI evolve at similar rates and for these we applied the widely used cyt-b rate of 2.0% sequence divergence per million years (0.010 substitutions/site/lineage/million years, [s/s/l/m], Lovette 2004). Our comparisons indicated that ND2 and ATPase6&8 evolve at approximately 1.25 times the rate of cyt-b (slope of regression line, Fig. 1), thus we applied a rate of 2.5% per million years (0.0125 s/s/l/m) for these genes. Rate heterogeneity among taxa and across genes has led some researchers to question the general utility of a molecular clock (Pereira and Baker 2006, Nabholz et al. 2009) however, other studies have shown that when analyzed carefully (as we have done here) divergence time estimates can be robust and informative (Weir and Schluter 2008). We acknowledge other sources of bias in estimates based on a single locus such as mtDNA. For example, we did not correct our data for ancestral polymorphism

Phylogenetic tree-based comparative diversification analyses

12 10 8 6 Y = 1.27x + 0.59 R2= 0.810

2 0 0

2 4 6 8 10 12 Average intrageneric p-distance, Cyt-b

Figure 1. Plot of relative rates of cyt-b and ND2 genes for 23 genera of New World birds. The dotted line indicates equal rates.

335

ISSUE

To test the hypothesis that the land bridge was critical for avian exchange we compared observed diversification time estimates with appropriate null models (Zink et al. 2004). We used two complementary approaches. First we simulated distributions of diversification events using the program Phyl-O-Gen v. 1.2 (Rambaut 2002). These distributions were generated with parameters that would reflect a range of possible diversification scenarios. We used two different constant growth scenarios, moderate speciation (birth rate 0.2, extinction rate 0.2) and high speciation (birth rate 0.8, extinction rate 0.2) We also generated distributions affected by extinction pulses, using a birth rate of 0.2 and episodic extinction events that removed 20% (low extinction pulse) and 80% (high extinction pulse) of the lineages in a simulation. We converted branching events in simulated phylogenies into

IBS SPECIAL

Average intrageneric p-distance, ND2

BIOGEOGRAPHIC BOUNDARIES ON

SYMPOSIUM ISSUE

IBS SPECIAL

time using the program LASER (Rabosky 2006) by setting the basal divergence of each simulation to 12 million years. Diversification events were periodically sampled to generate a simulated distribution. To make the distribution comparable to the distribution of empirically derived diversification events, all points less than 0.04 million years were omitted. We realized that simulated distributions may not capture the complexity of diversification occurring within and among multiple avian lineages. To address this issue, we employed a second approach that used the results from our BEAST analysis to generate a distribution of all diversifications occurring within 28 completely sampled avian genera that are represented in both North and South America. Temporal estimates of all diversifications (n 413) occurring within the 28 selected genera were then compiled into a single distribution that represents an empirically derived estimate of background diversification over time. To allow direct comparison with our distribution of isthmus related diversifications (n 135), we generated a distribution that represents a temporal baseline of cladogenetic events occurring over the last 8 million years. From this distribution we took 100 random draws of 135 diversification events. To compare the simulated and empirical distributions of diversification events with distribution of transisthmus events, we used the nonparametric two-sample Kolmolgorov Smirnov test (K-S test) across the entire temporal distribution (0 8 mya), as well as to pre-land bridge (4 8 and 3.1 8 mya) and post-land bridge (0 4 and 0 3.1 mya) time intervals. Historical and ecological diversification patterns We further explored the timing and pattern of dispersal by performing additional K-S tests on direction of dispersal and the potential role of elevation and habitat preference. Following the designations of Stotz et al. (1996), we separated lineages into two broad elevation categories, highland and lowland. The transitional zone between montane and lowland habitats varies across mountain ranges and regions in the Neotropical region (Stotz et al. 1996). To account for this complexity, we defined lowland birds as those never occurring above 2000 m and highland birds as those never occurring lower than 1000 m (fide Stotz et al. 1996). This analysis was used to test the hypothesis that birds occupying montane habitat ‘‘islands’’ may be predisposed to disperse with greater regularity than lowland forest birds. Lineages were also divided into two broad ecological categories based on predefined habitat and foraging strata categories (Stotz et al. 1996). These included birds that are obligately dependent on the vertical structure occurring in true forest habitats (hereafter: forest birds) and ‘‘non-obligates,’’ birds occurring predominantly in successional, edge, and scrub habitats (hereafter: edge/scrub birds). This approach allowed us to use birds to evaluate the ecogeographic model (e.g. alternating periods of rainforest and drier, more open habitat corridors) that has been proposed to explain the mammalian interchange (Webb 1991, Vrba 1992). 336

Taxonomy-based comparative biogeographic analysis The timing of avian exchange is not only important for discerning the impact of the Panamanian land bridge but it also provides critical insight into how the ancestral origin of avian groups affected the present-day distribution of New World avian diversity. To examine the contribution of historical assemblages to present-day patterns of diversity we identified New World avian families for which an ancestral origin had been inferred using molecular data. For each family, we summarized the number of species now inhabiting either Nearctic or Neotropical regions (source: Sibley and Monroe 1990). To test whether families with northern ancestral origins have a greater number of species in the Neotropical region than families with southern ancestral origins have in the Nearctic (Mayr 1964), we used a generalized linear model with a quasibinomial error term to account for over dispersion.

Results We identified 135 trans-isthmus diversification events that occurred at multiple taxonomic scales including intraspecific, inter-specific, inter-generic, and inter-familial branching events (Supplementary material). Overall, our results indicate that the land bridge was critical in facilitating intercontinental exchange and subsequent diversification; although, a number of geologic, geographic, and climatic factors that are known to have occurred during this time period certainly played a role. We obtained qualitatively similar results irrespective of the clock approach used (results not shown), suggesting that considerable signal is present in these data. Diversification between North and South America occurred from the mid-Miocene through the Late Pleistocene, with 76% of the transisthmus diversification events we identified occurring within the last 4 million years. A histogram of trans-isthmus diversification times (Fig. 2) suggests a pulse from 4 to 2 mya, coincident with the land bridge formation. This pulse was followed by a pronounced decrease in the frequency of diversification events during the Pleistocene. The distribution of trans-isthmus diversifications differed significantly from all simulations (Fig. 3; Table 1) but we could not reject the background distribution (i.e. all diversifications accruing within 28 genera). However, when our data are partitioned into biologically relevant time intervals, more precise patterns of diversification are recovered. All comparisons for the earliest time interval (4 8 or 3.1 8 mya) indicate that early diversification events between North and South America followed the expected null diversification rate (Table 1). Importantly, this trend strongly shifts in the recent interval (regardless of which isthmus closure times estimate is used), a time attributed to the GABI. During this period, the frequency of transisthmus diversifications exceeds those derived from both the simulated and background distributions (Table 1). Within the post-land bridge period, a decline in diversification is evident in the isthmus-related and background distributions, a phenomenon shown to be common in birds (Weir 2006, Phillimore and Price 2008). We suggest that the

16 14 12

SYMPOSIUM

10 8 6 4

Trans-isthmus diversification events

6 7 8 9 10 11 12 13 14 15 16 Time (million years ago)

Figure 2. Histogram of trans-isthmus diversification events. Diversification times were estimated by employing a gene-specific relaxed molecular clock. The shaded area corresponds to the post-land bridge time period. Taxonomic rank is indicated by color: intraspecific (red); interspecific (blue); intergeneric (gold); and interfamilial (grey).

0.8

moderate speciation high speciation low extinction pulse high extinction pulse trans-isthmus “background”

0.7 0.6 0.5 0.4 0.3 0.2

(n 72) and highland (n 21) birds (Fig. 4b) indicates no difference between these groups (K-S test, D 0.2480, p 0.233). We did, however, detect potentially different temporal trends for forest and scrub/edge birds (Fig. 4c). The mean diversification time of scrub/edge birds (n 25; 3.33 mya, 2.55 4.11) was older than the obligate forest birds (n 49; 2.67 mya, 2.07 3.27) although the distributions were not significantly different (K-S test, D 0.248, p 0.224). When family-level biogeography was examined, we detected contrasting evolutionary histories (Table 2). Of the 12 families we identified as having northern origins, three are constrained today to the tropical region while the remaining nine occur widely across both Nearctic and Neotropical regions. Conversely, of the 16 families we identified as having southern (i.e. South American) origins, most invaded tropical Central America but only two were able to push further north to colonize Nearctic regions. The families with northern origins had a significantly greater proportion of species in the Neotropical region (83.4% of species from northern families now occur in the Neotropical region) than families with southern origins in the Nearctic (2.7% of species from southern families occur in the Nearctic; 2 P [t1 9.627] B0.001).

0.1 0.0 0

2 3 4 5 6 Time (million years ago)

Birds and mammals, a shared history Molecular and fossil based evolutionary reconstructions can be affected by forms of bias unique to each approach, making direct comparisons between such studies difficult. Nevertheless, the avian response to the Isthmus of Panama formation that we describe shares a close resemblance to the late Pliocene interchange described for mammals. The temporal consistency between the avian and mammalian histories underscores the dramatic impact that the newly 337

ISSUE

Figure 3. Relative frequency of simulated, empirical and transisthmus diversification events. Moderate speciation (birth rate 0.2, extinction rate 0.2); high speciation (birth rate 0.8, extinction rate 0.2); low extinction pulse (birth rate 0.2) with episodic extinction events that removed 20% of lineages; high extinction pulse (birth rate 0.2) with episodic extinction events that removed 80% of the lineages. ‘‘Background’’ diversification times were estimated for all cladogenetic events occurring within 28 widely distributed (i.e. with elements occurring on both continents) New World avian genera.

Discussion

IBS SPECIAL

Relative frequency of diversification events

notably steeper decline apparent in the trans-isthmus distribution reflects a density dependent effect. Within the tropics, the direction of exchange (Fig. 4a) was symmetrical and reciprocal (north to south, n 52; south to north, n 50; K-S test, D 0.195, p 0.259) but different trends emerge following the onset of the Pleistocene. Dispersals into South America slowed during this time while dispersal into North America reached its highest frequency. The role of elevation on trans-isthmus dispersal appears to be negligible. Our comparison between lowland

BIOGEOGRAPHIC BOUNDARIES

Distribution modeled

0 8 (mya)

4 8 (mya)

3.1 8 (mya)

0 4 (mya)

0 3.1 (mya)

Moderate speciation High speciation Low extinction pulse High extinction pulse ‘‘Background’’

0.6166$ 0.3623$ 0.5824$ 0.3333$ 0.1495

0.2575 0.2256 0.2363 0.2280 0.2336

0.1772 0.1200 0.1596 0.1348 0.2272

0.5840$ 0.3626$ 0.5533$ 0.3132$ 0.2129*

0.5445$ 0.3078$ 0.5078$ 0.2560* 0.2449*

*pB0.05. $ pB0.001.

formed terrestrial corridor had on the two previously isolated biotas. For both birds and mammals, some lineages were able to disperse across the water gap prior to isthmus completion; although, in both groups intercontinental exchange for the vast majority was facilitated by the presence of the land bridge. While birds can fly, our result suggests that most birds are subject to the same dispersal constraints as their mammalian counterparts. The similarities between the avian and mammalian responses to the isthmus completion are also seen in the patterns of interchange. The shared decline in trans-isthmus dispersal over time likely reflects a real pattern that may be driven by the unavailability of ecological space. We suggest that this pattern is the result of a density dependent effect (the priority effect, MacArthur 1972), such that exchange across the isthmus would have decreased over time due to

SYMPOSIUM

BIOGEOGRAPHIC BOUNDARIES

Table 1. Results of comparisons between the distribution of trans-isthmus diversifications and alternative diversification scenarios. To compare the simulated and empirical distributions of diversification events with the trans-isthmus distribution, we used the nonparametric two-sample Kolmolgorov Smirnov test (K-S test). The test statistic shown is the D statistic, a measure of difference between two distributions.

competition with congeners (or diversifying conspecifics) that had already crossed the isthmus. Other previously identified factors that may lead to a genetic signature of density dependent diversification, such as extinction and inadequate taxon sampling (Rabosky and Lovette 2008), are not easily explained by the data. An increase in extinction rates that would cause a decline in diversification is not likely. There is no a priori prediction that would allow extinction to differentially affect recent colonists (last 2 million years) at a much higher rate than that experienced by earlier colonists. Inadequate taxon sampling is also not likely, as 35% of the points in the data are intra-specific genetic breaks identified by widespread geographic sampling. Our data suggest that by the onset of the Pleistocene, the majority of species in geographic proximity to the land

0.30 0.25

south to north north to south

0.20 0.15 0.10 0.05 0.00

IBS SPECIAL

ISSUE

Relative frequency

0.30 0.25

lowland birds

0.20

highland birds

0.15 0.10 0.05 0.00 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00

forest canopy birds edge/scrub birds

3 4 5 6 Time (million years ago)

Figure 4. The frequency of diversification events relative to historical and ecological variables. (a) dispersal direction (K-S statistic: D 0.1946, p 0.259); (b) elevation (D 0.2480, p 0.233); and (c) habitat group (D 0.2482, p 0.224).

338

Table 2. Breeding distributions of New World avian families for which ancestral origins were inferred from molecular data. Ancestral north

25 9 34 38 18 11 66 402 65 76 33 81

0 0 8 13 17 4 9 0 49 20 15 35

Neotropical

Nearctic

150 331 35 2 11 32 17 235 209 31 400 51 4 44 10 49

0 18 0 0 0 0 0 0 0 0 27 0 0 0 0 0

*Old World members of family are excluded from this total. $ Thraupidae is secondarily South American.

Intercontinental avian exchange and the latitudinal diversity gradient

339

ISSUE

Overall, the exchange of lineages in response to the Isthmus of Panama closure played a major role in assembling the present-day Neotropical avifauna. The molecular record shows that within the tropics, the number of birds dispersing between the continents in each direction was similar; however, from a broader geographic perspective this symmetry disappears. For both birds and mammals, the GABI had relatively little affect on diversity in temperate North America. We suggest that this phenomenon is best explained by an asymmetry in niche conservatism and this asymmetry has contributed to the long-recognized latitudinal diversity gradient (Mittelbach et al. 2007). Avian families with northern origins occur widely in both the Nearctic and Neotropical regions, thus northern lineages have successfully diversified throughout South

IBS SPECIAL

bridge had already crossed, even though source pools continued to receive new immigrant species from the interchange. Once these source taxa immigrated, their descendants only rarely dispersed back into their continent of origin. This is evident by the infrequency of backcrosses between continents, with 85% of the points represented by single dispersal events. Nearly half of the lineages that did backcross represented older events at higher taxonomic levels. It is unlikely that our results are confounded by sampling effects such as over-emphasizing highland taxa (at the expense of more continuously distributed lowland species) or by selective study of only those taxa presumed a priori to possess complex population structure. Our data include intraspecific divergences for 34 lowland taxa that are presumed to be continuously distributed across the Isthmus of Panama. In the course of summarizing the relevant available data we encountered only one instance of haplotype sharing across the isthmus, suggesting that most lowland taxa do differ across this biogeographic break and that diversification occurs soon after transisthmus dispersal. The similarity between the highland and lowland distributions is somewhat surprising given that the distributions of many highland birds are fragmented by intervening lowland valleys. Our result may indicate that highland birds are able to readily disperse through lowland corridors during interglacial habitat shifts. Regardless,

BIOGEOGRAPHIC BOUNDARIES

Psittacidae* Trochillidae Ramphastidae Semnornithidae Capitonidae Bucconidae Galbulidae Furnariidae Thamnophilidae Tityridae Tyrannidae Pipridae Melanopareidae Rhinocryptidae Conopophagidae Grallaridae

Nearctic

Ancestral south

Neotropical

SYMPOSIUM

Trogonidae* Momotidae Mimidae Vireonidae Corvidae* Polioptilidae Troglodytidae Thraupidae$ Parulidae Icteridae Cardinalidae Emberizidae*

No. of species

additional data may be required to further clarify the role of elevation on trans-isthmus dispersal. We identified only 21 highland diversification events compared to seventytwo for lowland birds. The disparity in sample size between highland and lowland birds suggests that the land bridge corridor had a lesser impact in facilitating dispersal for highland species because the corridor only directly connected lowland habitats. The mammalian GABI is known for the exchange of savanna species despite the present-day lack of a dry habitat corridor between the Americas. Our dataset contains several non-rainforest species (xeric, scrub, and edge taxa) that would have difficulty dispersing through the extensive rainforest on the isthmus today. The existence of historically more continuous dry habitats within the Neotropical region (Pennington et al. 2000) and a drier isthmus habitat corridor (Webb 1991) coincide with a known Pliocene warming period (Zachos et al. 2001). Thus, it seems reasonable that large mammalian herbivores and birds adapted to more xeric habitats may have dispersed at this time with true forest birds crossing sometime later. Although our analysis lacked statistical significance, differing temporal trends for these two ecological groups are apparent (Fig. 4c). Scrub/edge birds have an older mean diversification time and the frequency of dispersal declines in the Pleistocene which is consistent with the early wave of large mammalian herbivores. In contrast, forest birds show a pattern similar to that seen for a second wave of mesic phase mammals (Webb and Rancy 1996), with both groups reaching their highest frequency in the Pleistocene. As rainforest continued to expand on the land bridge, many potential immigrants (xeric, scrub, and edge species) from the north would have been cut off, likely contributing to the slowdown in birds dispersing southward. Assuming our interpretations are correct, the decline in the rate of diversification events was driven in part by density dependent effects and also by habitat changes that may have prevented edge/scrub taxa from continuing to participate in the GABI. Despite the overall slowdown in avian exchange our results indicate that for tropical South American birds, the GABI does not appear to be over.

BIOGEOGRAPHIC BOUNDARIES ON

SYMPOSIUM ISSUE

IBS SPECIAL

America (Table 2). Conversely, the southern families that participated in the GABI remain almost entirely restricted to the Neotropical region, as they have been for tens of millions of years. As a specific example, the oscine and suboscine passerines are a large and diverse group representing over one half of all extant avian species (Sibley and Monroe 1990). Based on molecular dating, the oscines arrived in the New World between 34 and 14 mya, via multiple independent dispersal events, likely through Beringia (Barker et al. 2004). The suboscines have occupied South America some 65 million years, prior to its drifting from Antarctica and Australia (Barker et al. 2004). Despite having northern ancestral origins, 55% of New World oscine species now breed in South America, many of them in tropical habitats. In contrast, only 2.4% of suboscines have secondarily adapted to North American temperate zone habitats (Table 2). The difficulty tropical organisms have in colonizing seasonal temperate zone environments is widely recognized. Niche conservatism at this scale, is an important component of the Tropical Niche Conservatism hypothesis (TNC), a model invoked to help explain the latitudinal diversity gradient (Wiens and Donoghue 2004). The observed asymmetry is also consistent with another component of the TNC model, the ‘‘time for speciation’’ effect (Stephens and Wiens 2003). Most ancestrally southern bird lineages now occurring at the northern limits of the Neotropical region, have arrived since the time of isthmus uplift and may not have had sufficient time to adapt to temperate conditions. Conversely, the ability of the northern birds to repeatedly invade the tropics may be linked to their long exposure to tropical conditions in Central America well before the uplift of the isthmus. One part of the TNC hypothesis that does not entirely explain the distribution of New World avian biodiversity is the role of shifting tropical habitats. During the early Tertiary, tropical forests were once much more widespread across the globe and have since contracted to equatorial latitudes (Behrensmeyer et al. 1992). This contraction along with the inability of tropical organisms to adapt to extratropical habitats has been argued be to a major driver of the latitudinal diversity gradient in New World birds (Hawkins et al. 2006). Support for this pattern in the Old World is seen in fossil evidence that indicates relatives of the present-day Ethiopian (Afrotropical) avifauna were once distributed in northern Europe and the United States (reviewed in Feduccia 1999, Ksepka and Clarke 2009). But, in the New World proper, neither fossil evidence (Vuilleumier 1985) nor molecular data support a unified contraction of the Neotropical avifauna, because the North and South American avifauna remained largely isolated from one another until the Panamanian land bridge formation (Fig. 2). The early Cenozoic North American tropical avifauna is largely extinct and has contributed very little to the extant Neotropical avifauna. We were able to identify only two Neotropical families (Trogonidae and Motmotidae) which are presumed to have northern tropical origins (Feduccia 1999, Witt 2004, DaCosta and Klicka 2008). The contrasting biotic histories of the Old and New World are reflected in their geology. The Old World continents have a long history of connectivity which is in sharp 340

contrast to the prolonged isolation of South America from North America. Conclusion Our study suggests the formation of the Isthmus of Panama was critical for facilitating avian exchange in a manner similar to that known for mammals. This exchange, contributed to a reconfiguration of the taxonomic composition of New World avian communities. The GABI was the process that united the isolated North and South American tropical avifaunas into a single biogeographic unit, the Neotropical region. However, the GABI had little impact on avian diversity in the Nearctic. This asymmetric exchange between the Nearctic and Neotropical regions provides an additional insight into why species diversity is higher towards the equator. The interchange of birds from ancestrally northern families increased avian diversity in the tropics, whereas, birds from the ancient South America avifauna contributed little to extant diversity in the Nearctic. We suggest that the ability of organisms with Nearctic origins, to colonize and radiate within the tropics may represent an important and underappreciated contributor to the latitudinal diversity gradient. Acknowledgements We thank Jef Jaeger, Robert Zink, Gary Voelker, Garth Spellman, Matt Miller, Brett Riddle, the UNLV Systematics group, and four anonymous reviewers for their comments on this manuscript. We acknowledge Richard Ree, who helped us with our Lagrange analyses, and Jason Weckstein for the use of his unpublished data. Jorge Perez and Daniel Cadena provided critical specimen material and Cheryl Vanier provided assistance with statistical tests. Some of the analyses were performed on the Computational Biology Service Unit at Cornell Univ., an entity funded in part by the Microsoft Corporation. We thank the field collectors, specimen preparators, and curators whose efforts contributed to this study. This work was supported in part by an NSF grant (DEB 0315469, to J.K.) and research funds provided by the UNLV Museum Foundation.

References Arbogast, B. S. and Slowinski, J. B. 1998. Pleistocene speciation and the mitochondrial DNA clock. Science 282: 1955a. Barker, F. K. et al. 2004. Phylogeny and diversification of the largest avian radiation. Proc. Nat. Acad. Sci. USA 101: 11040 11045. Behrensmeyer, A. K. et al. 1992. Terrestrial ecosystems through time. Evolutionary paleoecology of terrestrial plants and animals. Univ. of Chicago Press. Coates, A. and Obando, J. 1996. The geologic evolution of the Central American Isthmus. In: Jackson, J. et al. (eds), Evolution and environment in tropical America. Univ. of Chicago Press, pp. 21 56. DaCosta, J. M. and Klicka, J. 2008. The Great American Interchange in birds: a phylogenetic perspective with the genus Trogon. Mol. Ecol. 17: 1328 1343. Drummond, A. J. and Rambaut, A. 2007. BEAST: Bayesian evolutionary analysis sampling trees. BMC Evol. Biol. 7: 214.

BIOGEOGRAPHIC BOUNDARIES IBS SPECIAL ISSUE

341

Rabosky, D. L. 2006. LASER: a maximum likelihood toolkit for detecting temporal shifts in diversification rates. Evol. Bioinform. 2: 257 260. Rabosky, D. L. and Lovette, I. J. 2008. Density-dependent diversification in North American wood warblers. Proc. R. Soc. B 275: 2363 2371. Rambaut, A. 2002. Phyl-O-Gen: phylogenetic tree simulator package v 1.1. /<http://evolve.zoo.ox.uk/software/ PhyloGen//>. Rambaut, A. and Drummond, A. J. 2007. Tracer v 1.4. /<http://beast.bio.ed.ac.uk/tracer/>. Ree, R. H. and Smith, S. A. 2008. Maximum likelihood inference of geographic range evolution by dispersal, local extinction, and cladogenesis. Syst. Biol. 57: 4 14. Ree, R. H. et al. 2005. A likelihood framework for inferring the evolution of geographic range on phylogenetic trees. Evolution 59: 2299 2311. Ricklefs, R. E. 2002. Splendid isolation: historical ecology of the South American passerine fauna. J. Avian Biol. 33: 207 211. Sibley, C. G. and Monroe, B. L. J. 1990. Distribution and taxonomy of birds of the world. Yale Univ. Press. Simpson, G. G. 1947. Holarctic mammalian faunas and continental relationships during the Cenozoic. Bull. Geol. Soc. Am. 58: 613 688. Simpson, G. G. 1980. Splendid isolation: the curious history of South American mammals. Yale Univ. Press. Stehli, F. and Webb, S. 1985. The Great American Biotic Interchange. Plenum Press. Stephens, P. R. and Wiens, J. J. 2003. Explaining species richness from continents to communities: the time-for-speciation effect in Emydid turtles. Am. Nat. 161: 112 128. Stotz, D. F. et al. 1996. Neotropical birds: ecology and conservation. Univ. of Chicago Press. Swofford, D. L. 2002. PAUP* v. 4.0b10: phylogenetic analysis using parsimony (and other methods). Sinauer. Tiffney, B. H. and Manchester, S. R. 2001. The use of geological and paleontological evidence in evaluating plant phylogeographic hypotheses in the northern hemisphere Tertiary. Int. J. Plant Sci. 162: 3 17. Voelker, G. et al. 2009. Repeated trans-Atlantic dispersal catalysed a global songbird radiation. Global Ecol. Biogeogr. 18: 41 49. Vrba, E. S. 1992. Mammals as a key to evolutionary theory. J. Mammal. 73: 1 28. Vuilleumier, F. 1984. Faunal turnover and development of fossil avifaunas in South America. Evolution 38: 1384 1396. Vuilleumier, F. 1985. Fossil and recent avifaunas and the Interamerican Interchange. In: Stehli, F. and Webb, S. (eds), The Great American Biotic Interchange. Plenum Press, pp. 387 424. Webb, S. D. 1976. Mammalian faunal dynamics of the Great American Interchange. Paleobiology 2: 220 234. Webb, S. D. 1991. Ecogeography and the Great American Interchange. Paleobiology 17: 266 280. Webb, S. D. and Marshall, L. G. 1982. Historical biogeography of recent South American land mammals. In: Mares, M. A. and Genoways, H. H. (eds), Evolution of Neotropical mammals. Pymatuning Laboratory of Ecology, pp. 39 54. Webb, S. D. and Rancy, A. 1996. Late Cenozoic evolution of the Neotropical mammal fauna. In: Jackson, J. et al. (eds), Evolution and environment in tropical America. Univ. of Chicago Press, pp. 335 358. Weir, J. T. 2006. Divergent timing and patterns of species accumulation in lowland and highland Neotropical birds. Evolution 60: 842 855. Weir, J. T. and Schluter, D. 2008. Calibrating the avian molecular clock. Mol. Ecol. 17: 2321 2328.

SYMPOSIUM

Drummond, A. J. et al. 2006. Geneious v.2.5.4. /<www. geneious.com//>. Edwards, S. V. and Beerli, P. 2000. Perspective: gene divergence, population divergence, and the variance in coalescence time in phylogeographic studies. Evolution 54: 1839 1854. Feduccia, A. 1999. The origin and evolution of birds. Yale Univ. Press. Hawkins, B. A. et al. 2006. Post-Eocene climate change, niche conservatism, and the latitudinal diversity gradient of New World birds. J. Biogeogr. 33: 770 780. Hayes, F. E. and Sewlal, J. N. 2004. The Amazon river as a dispersal barrier to passerine birds: effects of river width, habitat and taxonomy. J. Biogeogr. 31: 1809 1818. Hickerson, M. J. et al. 2006. Comparative phylogeographic summary statistics for testing simultaneous vicariance across taxon-pairs. Mol. Ecol. 15: 209 224. Ho, S. et al. 2005. Time dependency of molecular rate estimates and systematic overestimation of recent divergence times. Mol. Biol. Evol. 22: 1561 1568. Kirby, M. X. et al. 2008. Lower Miocene stratigraphy along the Panama Canal and its bearing on the Central American Peninsula. PloS ONE 3: e2791. Ksepka, D. T. and Clarke, J. A. 2009. Affinities of Palaeospiza bella and the phylogeny and biogeography of mousebirds (Coliiformes). Auk 126: 245 259. Lomolino, M. V. et al. 2006. Biogeography. Sinauer. Lovette, I. J. 2004. Mitochondrial dating and mixed-support for the ‘‘2% rule’’ in birds. Auk 121: 1 6. MacArthur, R. H. 1972. Geographical ecology: patterns in the distribution of species. Princeton Univ. Press. Maddison, W. P. and Maddison, D. 2009. Mesquite: a modular system for evolutionary analysis v 2.6. /<http:// mesquiteproject.org/>. Marshall, L. G. et al. 1979. Calibration of the Great American Interchange. Science 204: 272 279. Marshall, L. G. et al. 1982. Mammalian evolution and the Great American Interchange. Science 215: 1351 1357. Mayr, E. 1964. Inferences concerning the Tertiary American bird faunas. Proc. Nat. Acad. Sci. USA 51: 280 288. Mayr, E. and Diamond, J. 2001. The birds of northern Melanesia: speciation, ecology, and biogeography. Oxford Univ. Press. Mittelbach, G. G. et al. 2007. Evolution and the latitudinal diversity gradient: speciation, extinction and biogeography. Ecol. Lett. 10: 315 331. Moore, R. P. et al. 2008. Experimental evidence for extreme dispersal limitation in tropical forest birds. Ecol. Lett. 11: 960 968. Nabholz, B. et al. 2009. The erratic mitochondrial clock: variations of mutation rate, not population size, affect mtDNA diversity across birds and mammals. BMC Evol. Biol. 9: 54. Nylander, J. 2004. MrModeltest v 2.3. Evolutionary Biology Centre, Uppsala Univ. Olson, S. L. 1985. The fossil record of birds. In: Farner, D. S. et al. (eds), Avian biology. Academic Press, pp. 179 238. Patterson, B. and Pascual, R. 1972. The fossil mammal fauna of South America. In: Keast, A. et al. (eds), Evolution, mammals, and southern continents. State Univ. of New York Press. Pereira, S. L. and Baker, A. J. 2006. A mitogenomic timescale for birds detects variable phylogenetic rates of molecular evolution and refutes the standard molecular clock. Mol. Biol. Evol. 23: 1731 1740. Pennington, R. T. et al. 2000. Neotropical seasonally dry forests and Quaternary vegetation changes. J. Biogeogr. 27: 261 273. Phillimore, A. B. and Price, T. D. 2008. Density-dependent cladogenesis in birds. PloS Biol. 6: e71.

Download the Supplementary material as file E6335 from / <www.oikos.ekol.lu.se/appendix/>.

IBS SPECIAL

ISSUE

SYMPOSIUM

BIOGEOGRAPHIC BOUNDARIES

Wiens, J. J. and Donoghue, M. J. 2004. Historical biogeography, ecology and species richness. Trends Ecol. Evol. 19: 639 644. Witt, C. C. 2004. Rates of molecular evolution and their application to Neotropical avian biogeography. Ph.D. thesis, Louisiana State Univ.

342

Zachos, J. et al. 2001. Trends, rhythms, and aberrations in global climate 65 Ma to present. Science 292: 686 693. Zink, R. M. et al. 2004. The tempo of avian diversification during the Quaternary. Proc. R. Soc. B 359: 215 220.

Ecography 33: 343 354, 2010 doi: 10.1111/j.1600-0587.2010.06281.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Ken Kozak. Accepted 6 April 2010 *The first two authors contributed equally.

SYMPOSIUM

Using regional comparative phylogeographic data from snake lineages to infer historical processes in Middle America

Understanding how historical processes have either similarly, or differentially, shaped the evolution of lineages or biotic assemblages is important for a broad spectrum of fields. Gaining such understanding can be particularly challenging, however, especially for regions that have a complex geologic and biological history. In this study we apply a broad comparative approach to distill such regional biogeographic perspectives, by characterizing sets of divergence times for major biogeographic boundaries estimated from multiple codistributed lineages of snakes. We use a large combined (mitochondrial gene sequence) phylogeographic/phylogenetic dataset containing several clades of snakes that range across Middle America the tropical region between Mexico and northwestern South America. This region is known for its complex tectonic history, and poorly understood historical biogeography. Based on our results, we highlight how phylogeographic transition zones between Middle and South America and the Nicaragua Depression appear to have undergone multiple episodes of diversification in different lineages. This is in contrast to other examples we find where apparently a single vicariant period is shared across multiple lineages. We specifically evaluate the distributions of divergence time estimates across multiple lineages and estimate the number of temporal periods of lineage diversification per biogeographic break. Overall, our results highlight a great deal of shared temporal divergence, and provide important hypotheses for yet unstudied lineages. These multi-lineage comparisons across multiple spatial and temporal scales provide excellent predictive power for identifying the roles of geology, climate, ecology and natural history in shaping regional biodiversity.

343

ISSUE

or tectonic information, or where little historical consensus is available (Riddle et al. 2008, Castoe et al. 2009). In historical biogeography, vicariance and dispersal are considered the major forces that determine the divergence and geographic distribution of lineages (Nelson and Platnick 1981, Ree and Smith 2008, Ree and Sanmart覺織n 2009). Neither of these two processes are, however, easily extracted from any single phylogeographic or phylogenetic pattern. Using coalescent models and the geographic structure of genetic data, it is possible to test the data against specific historical demographic scenarios that invoke vicariance or dispersal (Knowles and Carstens 2007, Richards et al. 2007, Hickerson and Meyer 2008). Such statistical approaches, however, are designed to address data associated with shallow phylogenetic trees, mostly at the intraspecific level. For deeper evolutionary events, different biogeographic methods are preferred. The most commonly used methods for such deep historical inferences search for evidence of congruence among different lineages and then explain this congruence (or lack of congruence) with vicariance/dispersal scenarios (Nelson and Platnick 1981, Ronquist 1997, Ree and Smith 2008, Ree and Sanmart覺織n 2009).

IBS SPECIAL

Historical biogeography, conservation biology, evolutionary ecology, and global climate change biology all require information about how historical patterns and processes have shaped lineage diversification at various spatial and temporal scales. It is important to understand how specific historical processes, and specific biogeographic boundaries, may have differentially impacted lineages or various components of biotic assemblages. The convergence of molecular phylogeographic datasets with robust approaches for estimating lineage divergence times has enabled an outgrowth of comparative phylogeographic research that may address such questions about differential biological responses of lineages. It is becoming increasingly clear that large comparative phylogeographic datasets may provide an excellent way to use multiple independent lineages simultaneously to infer models of historical divergence across landscapes (Bermingham and Moritz 1998, Arbogast and Kenagy 2001, Hickerson and Meyer 2008). These, in turn, may represent broad and generalizable models for projection onto other unstudied taxonomic groups, and even larger biotic assemblages. This insight from comparative analyses are particularly important for areas with either vague geological

BIOGEOGRAPHIC BOUNDARIES

J. M. Daza and C. L. Parkinson (cparkins@mail.ucf.edu), Dept of Biology, Univ. of Central Florida, 4000 Central Florida Blvd., Orlando, FL 32816, USA. T. A. Castoe, Dept of Biochemistry and Molecular Genetics, Univ. of Colorado School of Medicine, Aurora, CO 80045, USA.

Juan M. Daza*, Todd A. Castoe* and Christopher L. Parkinson

BIOGEOGRAPHIC BOUNDARIES ON

SYMPOSIUM ISSUE

IBS SPECIAL

Here we explore the application of comparative phylogeography beyond the intraspecific level to interpret regional historical processes in Middle America, and formulate new hypotheses to describe spatial temporal lineage diversification on this broad regional scale. The core concept is that a biogeographic boundary may represent a spatial context over which a large number of lineage divergences may be temporally mapped (LeacheÂ´ et al. 2007). For a given area, or axis of vicariance, the distribution of divergence times across lineages holds important biological information which can be used to interpret historical scenarios, and also predict the breadth of impact of historical processes on other components of biological communities (Hickerson et al. 2006a, b, Hickerson and Meyer 2008). Given the overlap of divergence time estimates for multiple related lineages, common patterns can be identified which may represent deep-reaching historical processes. These can be contrasted with patterns unique to particular lineages or groups of lineages. Using related lineages, such that a single phylogenetic tree can be used for the entire analysis (as in the current study), allow the predictions of temporal congruence to be largely independent of errors in calibration points (required for absolute time estimation). This is because estimates of relative time within a single dated tree are particularly robust, making such systems particularly ideal for testing for temporal correspondence of events among lineages (regardless of the accuracy of calibration points). We applied this comparative approach to patterns of lineage diversification in snakes of Middle America the tropical region between Mexico and northwestern South America. A fairly large number of lineages of snakes that range throughout Middle America have been sampled for the same mitochondrial loci, making them a good system for the current study. The exaggerated relief, diversity of habitats, and the dynamic tectonic and climatic history of the Middle America have all contributed to its high endemicity and diversity (Whitmore and Prance 1987, Jackson et al. 1996). Middle America has experienced a complex tectonic and geological history, and lies at the active junction of four major tectonic plates and several tectonic blocks (IturraldeVinent 2006, Marshall 2007). Deciphering the events that have historically shaped present-day biological diversity is complicated due to the continual physiographical reshaping of the region since the Cretaceous. Details of most of the tectonic history of Middle America still remain fragmentary and controversial (Coney 1982, Iturralde-Vinent 2006, Mann et al. 2007). This region has been the subject of intense biogeographic study for 40 yr, although the geological and climatic complexities of the region have precluded any clear consensus model describing the historical processes that generated its high taxonomic diversity (Savage 1982, Campbell 1999). For this reason, Middle America is an ideal setting for applying comparative phylogeographic data to infer patterns of lineage diversification, and the degree to which divergences are temporally coincident. While many previous studies largely agree in identifying major biogeographic boundaries across Middle America (Marshall and Liebherr 2000, Perdices et al. 2005, Devitt 2006, Crawford et al. 2007, Castoe et al. 2009), there is no consensus of when these boundaries may have been relevant in splitting lineages. Furthermore, there is even less 344

resolution on how many times, through history, these boundaries were effective in dividing lineages. Thus, our two aims were to 1) determine the degree to which these ecologically diverse lineages appear to share overlapping divergence times over the same biogeographic break, and 2) to estimate the number of discrete times in history each boundary may have led to lineage diversification. To address these questions, we examined Bayesian posterior distributions of divergence time estimates for a total of five major biogeographic boundaries across Middle America that are shared by multiple snake lineages, totaling 28 individual phylogeographic breaks. We also used an approximate Bayesian computation approach, using a hierarchical coalescent model, to infer the discrete number of divergence episodes for the same biogeographic breaks (Hickerson et al. 2006b, 2007). We use these results to infer how the distributions of divergence times may be related to an interpretation of historical biogeographic events that have broadly impacted the fauna in the region.

Material and methods Target taxa Our phylogenetic sampling includes multiple clades of snakes, including viperids and elapids, as well as nonvenomous colubrids, that contain lineages distributed throughout Middle America. Previously, we had conducted a more restricted comparative study including three lineages of mesic highland-inhabiting viperid snakes in Middle America, and found evidence for shared divergences across three biogeographic boundaries in Middle America (Castoe et al. 2009). The current study includes expanded sampling of a greater ecological diversity of lineages, such as lowland groups (e.g. Micrurus, Bothriechis schlegelli, Porthidium, Leptodeira), habitat or dietary specialists (Micrurus spp., Leptodeira nigrofasciata) and habitat or dietary generalists (Bothrops asper, Leptodeira septentrionalis). Despite all lineages being snakes and thus sharing somewhat similar dispersal characteristics and life history traits, the lineages sampled do contain a diverse sampling of ecological groupings, and should be capable of providing a much broader perspective on co-diversification and speciation in Middle America than the previous study (Castoe et al. 2009). We assembled a single combined data set, incorporating 28 nodes that correspond to clear phylogeographic breaks across Middle America (Fig. 1; Devitt 2006, Castoe et al. 2007a, 2009, Daza et al. 2009). The first major lineage comprises the subfamily Crotalinae. This group of venomous snakes is particularly diverse in the Neotropical region, and their phylogenetic relationships have been studied extensively (Parkinson et al. 2000, 2002, Castoe et al. 2005, 2009, Castoe and Parkinson 2006). Sequences for all relevant nodes of pitvipers were obtained from several published trees: Agkistrodon, (Parkinson et al. 2000), Bothriechis schlegelii (WuÂ¨ ster et al. 2002), Crotalus durissus (WuÂ¨ ster et al. 2005); Lachesis (Zamudio and Greene 1997), and highland pitvipers (Castoe et al. 2005, 2009). The second lineage includes members of the family Elapidae, and specifically includes representatives of the monadal and triadal coralsnake lineages (Castoe et al. 2007a). Finally, we

SYMPOSIUM ON

compiled phylogenetic results of Neotropical colubrids from two sources: Devitt (2006) and Daza et al. (2009). The first includes the major lineages of the genus Trimorphodon (Colubrinae) and the second includes the major lineages of the genus Leptodeira (Dipsadinae). Phylogenetic reconstruction We assembled a molecular dataset that includes two mitochondrial protein-coding genes sequences from cytochrome b and NADH dehydrogenase subunit 4 (Supplementary material). Alignment of each gene was accomplished using Clustal W (Larkin et al. 2007) and corrected manually using GeneDoc 2.6 (Nicholas and Nicholas 1997). The dataset was partitioned by gene and codon position, and a different GTRGI model for each partition was implemented (as selected by MrModeltest 2.3 using AIC, Nylander 2004). We used the package Beast 1.4.8, a Bayesian approach to estimate simultaneously the phylogeny and both relative and absolute divergence times (Drummond and Rambaut 2007).

345

ISSUE

We estimated divergence times using two different approaches. First, we estimated relative divergence times (RT analysis thereafter) so that we could examine temporal congruence among nodes regardless of absolute time (and the calibration assumptions that accompany absolute time estimation). Second, we calibrated the molecular phylogenies using fossil and other calibrations to obtain absolute estimates of divergence dates (AT analysis). The strength of this two step approach is that we can first optimize rates using a Bayesian approach and obtain an ultrametric tree that relies only on the evolutionary process (and fitting of the relaxed clock model) and is unaffected by the uncertainty of the fossil record and other calibrations (Graur and Martin 2004, Heads 2005). This non-calibrated

IBS SPECIAL

Divergence time estimation

tree can be used to infer congruence in divergence time among lineages even when no nodal calibrations exist, and further used to evaluate the impact of adding calibration points on the correspondence of divergence time across nodes. Once inferences of temporal congruence are made, calibration points can then be added to estimate the absolute time scale of divergence events. We implemented the Bayesian relaxed molecular clock method with uncorrelated lognormal rates among branches (Drummond et al. 2006), assuming a birth death process for the speciation model. For the RT analysis we set the treeModel.rootHeight parameter to be 1 using a normal distribution with a mean 1.0 and SD 0.1 and used the programâ&#x20AC;&#x2122;s default priors. For the AT analysis we used a lognormal prior for the treeModel.rootHeight parameter with a mean 3.7 and SD 0.3, and the following additional constraints: for the tMRCA of Crotalus atrox and C. ruber we used a uniform prior between 2.5 and 4.5 Ma; for the tMRCA of Sistrurus Crotalus we used a uniform prior between 9.0 and 32.0; for the tMRCA of Agkistrodon contortrix we used a uniform prior between 5.0 and 32.0. The remaining priors were set to the program defaults for the AT analysis. To ensure convergence of our estimates, we initiated four independent runs in Beast with random starting trees, and ran each for 10 million generations. Chains were sampled every 1000 generations, and convergence and stationarity were verified by examining likelihood scores and parameter estimates using Tracer 1.4 (Rambaut and Drummond 2007). Based on examination of trial runs in Tracer (which burned in prior to 2 million generations), the conservative burnin period of three million generations was used for final runs, and we combined the posterior samples from all four runs, and report the results of this combined posterior sample. We used the program TreeStat 1.2 (Rambaut and Drummond 2008) to summarize the Markov chain results for posterior divergence date estimates, and used an R script to create posterior density plots for nodes of interest.

BIOGEOGRAPHIC BOUNDARIES

Figure 1. Map of Middle America showing the five major biogeographic boundaries analyzed in this study. [1] Middle South America transition, [2] Talamanca Cordillera, [3] Nicaraguan Depression, [4] Motagua Polochic Faults, [5] Isthmus of Tehuantepec.

IBS SPECIAL

ISSUE

SYMPOSIUM

BIOGEOGRAPHIC BOUNDARIES

Calibration points We used four calibration points to obtain absolute date estimates for the molecular phylogeny. We constrained the origin of Sistrurus to be at least 9.0 Ma (Parmley and Holman 2007), and the origin of Agkistrodon contortrix to be at least 5.0 Ma (Holman 2000). We also constrained the divergence between the species Crotalus ruber and C. atrox to be between 2.5 and 4.5 Ma based on phylogeographic information on the vicariance between mainland and Baja California peninsula desert regions (Castoe et al. 2007b, 2009). Finally, based on the oldest colubrid fossil known, the root of the tree (the tMRCA of Colubroidea) was set to have occurred before 40 Ma (Rage et al. 1992, Head et al. 2005). Shared divergence To make inferences about the degree to which lineage divergences were coordinated in time we used msBayes (Hickerson et al. 2006a) to estimate the number of independent/discrete lineage divergence times per biogeographic break. MsBayes implements an approximate Bayesian computation approach using a hierarchical coalescent model where hyper-parameter estimation is utilized to discriminate the differences between time of divergence among pairs of taxa and variance in coalescent times (Hickerson et al. 2006b, 2007). For these analyses we included only the nodes that had more than two samples per taxon pair, based on the requirements of the program. For each analysis (corresponding to each break) we drew one million samples from the hyper-prior and, using the hierarchical approximate Bayesian computation acceptance/ rejection algorithm, constructed the hyper-posterior from 2000 samples (tolerance 0.002). We contrasted the results obtained with msBayes and those based on posterior distributions of divergence dates and 95% credibility intervals obtained with Beast. Additionally, from posterior densities of individual lineage divergence times (from the Beast divergence dating analyses), we assemble pooled posterior densities for divergence times by combining data from multiple lineages (for a particular biogeographic break). For these pooled posterior densities, we summed the lineage-specific posterior density per unit time, across all lineages for each break. These distributions can be interpreted as the probability of divergence pooled over all lineages examined, and we discuss in the text how these may be useful particularly as informed priors for future studies. For interpreting co-divergence, however, these pooled posteriors may be somewhat misleading in that they may obscure multi-modal divergence posteriors of different lineages.

Results Our estimate of phylogeny is consistent with recent studies that have specifically analyzed phylogenetic relationships among the taxa included here (Fig. 2, WuÂ¨ ster et al. 2002, 2005, Devitt 2006, Castoe et al. 2007a, 2009, Daza et al.

346

2009). The ultrametric trees we obtained with the RT and AT analyses yielded similar results (Supplementary material). When standardizing the root of the RT tree to be the absolute date obtained with the AT analysis, we did not find any difference in the relative timing of phylogenetic events between the two trees. In other words, adding calibration points did not affect our inferences of relative divergence times, as compared between lineages/nodes of codistributed lineages. The AT analysis resulted in a tree with an overall depth of 41.8 Ma (95% credibility interval 30.9 55.69). The divergence between Colubridae and Elapidae was estimated to be 38.8 Ma and the split between Old World and New World Elapids was inferred at 21.5 Ma and the same divergence but within crotalines was estimated at 19.4 Ma. Divergence times were consistent with those from Sanders and Lee (2008), Castoe et al. (2009), Daza et al. (2009) and Kelly et al. (2009). In contrast, our estimated divergence times were younger than those from Devitt (2006), Burbrink and Pyron (2008), WuÂ¨ ster et al. (2008), and Vidal et al. (2009). The eight splits identified in the Middle South America transition spanned from the early Miocene to the Pleistocene (CI95% 0.8 22.8 Ma). The three lineage divergences across north and south areas of the Talamanca Cordillera occurred between 2.5 and 3.9 Ma (CI95% 1.4 5.4). The divergences across the Nicaraguan Depression spanned from 4.1 to 8.8 Ma (CI95% 2.4 11.9). The divergences across the Motagua Polochic faults were estimated to have occurred between 3.8 and 6.8 Ma (CI95% 2.4 9.9). Lastly, the five cladogenetic events identified across the Isthmus of Tehuantepec were estimated to be between 2.8 and 7.35 Ma (CI95% 1.5 10.1). Out of the five phylogeographic breaks analyzed, three of them showed a strong correspondence in divergence times among multiple lineages (Fig. 3). Across the Isthmus of Tehuantepec break, with the exception of a single divergence estimate (for Porthidium species), the lineages appeared to have diverged around the same time. The cladogenetic events occurring at the other biogeographic breaks were not entirely coincident in time, although as we discuss in detail below, a number of strong patterns of congruence are evident. The summary of estimated parameters using the Approximate Bayesian Computation algorithm is shown in Table 1. According to the msBayes results, the Talamanca Cordillera and the Motagua Polochic Faults have likely undergone a single vicariant event. The pooled posterior distributions in these two breaks also showed a single peak, and the widely overlapping 95% CIs further supports a shared divergence (Fig. 3, 4). Small values of V, a parameter that measures the incongruence among divergence times along the same barrier, were found for these two biogeographic boundaries. In contrast, V value was highest for the Middle American South American transition (V 3.46), followed by the divergences along the Nicaraguan Depression and the Isthmus of Tehuantepec (Table 1). Similarly, non-overlapping 95% CIs and multimodal pooled posterior distribution of dates were observed in these three phylogeographic breaks (Fig. 3).

5 1 1

3 4 5

1 2 5

1 5

3 4

OLIG 30

MIO 20

PI 10

PL 0

Million years before present Figure 2. Dated tree obtained using the relaxed molecular clock method implemented in Beast. Node heights represent mean node ages (based on the combined posterior of four independent runs). Bars on nodes represent the 95% credibility interval of divergence times. Stars depict calibration points and numbers indicate nodes (Fig. 1) utilized in the congruence tests.

347

ISSUE

EOC

IBS SPECIAL

BIOGEOGRAPHIC BOUNDARIES

SYMPOSIUM

Sinomicrurus kelloggii Sinomicrurus macclellandi Micrurus fulvius Micrurus diastema Micrurus mipartitus Micrurus surinamensis Leptomicrurus narducii Trimorphodon biscutatus Trimorphodon quadruplex Trimorphodon lyrophanes Trimorphodon lyrophanes Trimorphodon vilkinsonii Trimorphodon paucimaculatus Trimorphodon paucimaculatus Atractus wagleri Dipsas pratti Ninia atrata Imantodes inornatus Imantodes cenchoa Leptodeira nigrofasciata Mexico Leptodeira nigrofasciata Costa Rica Leptodeira frenata Leptodeira punctata Leptodeira splendida Leptodeira a. rhombifera Leptodeira a. cussiliris Leptodeira maculata Leptodeira s. polysticta Leptodeira s. polysticta Leptodeira s. ornata Leptodeira bakeri Leptodeira a. annulata Leptodeira septentrionalis Pacific CostaRica Leptodeira septentrionalis Atlantic CostaRica Ovophis monticola Gloydius ussuriensis Gloydius shedaoensis Ophryacus undulatus Ophryacus melanurus Lachesis muta Lachesis stenophrys Lachesis melanocephala Agkistrodon contortrix Agkistrodon piscivorus Agkistrodon taylori Agkistrodon bilineatus Sistrurus catenatus Crotalus tigris Crotalus atrox Crotalus ruber Crotalus durissus SWMex1 Crotalus durissus Veracruz1 Crotalus durissus Belize1 Crotalus durissus Chiapas1 Crotalus durissus Venezuela Crotalus durissus Brazil Bothriechis schlegelii Ecuador Bothriechis schlegelii Costa Rica Bothriechis supraciliaris Bothriechis nigroviridis Bothriechis lateralis Bothriechis marchi Bothriechis thalassinus Bothriechis bicolor Bothriechis aurifer Bothriechis rowleyi Rhinocerophis alternatus Bothriopsis taeniata Bothrops atrox Bothrops asper Pacific CostaRica Bothrops asper Atlantic CostaRica Porthidium ophryomegas Porthidium dunni Porthidium hespere Porthidium yucatanicum Porthidium nasutum Porthidium porrasi Porthidium lansbergi Porthidium arcosae Cerrophidion petlalcalensis Cerrophidion tzotzilorum Cerrophidion godmani Honduras Cerrophidion godmani Costa Rica Cerrophidion godmani Guatemala Cerrophidion godmani Guatemala Atropoides picadoi Atropoides indomitus Atropoides occiduus Atropoides olmec Atropoides nummifer Atropoides mexicanus

3000

Middle-South America transition (1)

2000

BIOGEOGRAPHIC BOUNDARIES

1000

1600

Talamanca Cordillera (2)

1200

Posterior probability of divergence dates

800 400

1000

Nicaraguan Depression (3)

800 600 400 200 1200 Motagua-Polochic Faults (4)

800

SYMPOSIUM

400

1600 Isthmus of Tehuantepec (5) 1200 800 400

Million years ago

IBS SPECIAL

ISSUE

Figure 3. Left: posterior density plots of divergence times for snake lineages across five biogeographic boundaries in Middle America. Right: line represents the pooled posterior distribution of divergence times for each biogeographic barrier and the bars are the 95% credibility intervals for each lineage pair. Black bars indicate the lineage pairs not included in the msBayes analysis.

Discussion Emerging hypotheses for Middle American speciation patterns Despite consensus in the identification of major biogeographic boundaries that have shaped Middle Americaâ&#x20AC;&#x2122;s

biodiversity (Savage 1982, Marshall and Liebherr 2000, Morrone 2001), there has been little quantitative insight as to when these barriers may have led to diversification, in what temporal order, and especially the degree to which divergences were temporally coordinated. In total, our dataset included 28 individual cladogenetic events that span five biogeographic boundaries, bringing a fair amount

Table 1. Statistics summary from the msBayes runs. n number of lineage pairs, C number of possible divergence times, V parameter indicating the degree of discordance among divergence times. Phylogeographic break

C mode

C mean

C CI95%

V mean

(1) (2) (3) (4) (5)

7 3 3 5 5

1.87 1.27 1.58 1.01 2.23

2.03 1.39 1.91 1.49 2.62

1.00 3.88 1.00 2.36 1.07 3.00 1.00 3.29 1.12 4.56

3.46 0.12 0.59 0.13 1.15

Middle South America transition Talamanca Cordillera Nicaraguan Depression Motagua Polochic Faults Isthmus of Tehuantepec

348

Middle-South America transition (1)

0.6

0.4

0.0 2

Talamanca Cordillera (2) 2 1 0

Estimates of relative and absolute divergence times

0.4

0.0 1

Motagua-Polochic Faults (4)

0 1

3 4 Isthmus of Tehuantepec (5)

0.4 0.2

0.0 1

Ψ (= number of possible divtimes)

Figure 4. Posterior distribution of the number of divergence times for snake lineages across five biogeographic boundaries in Middle America.

ISSUE

349

IBS SPECIAL

of evidence to bear on inferences of regional diversification. Analysis of this dataset contributes new findings that appear to reject previous hypotheses of temporal diversification and further clarify historical biogeographic patterns in Middle American taxa. It thus presents an example of how such a comparative spatio temporal approach may yield insight into the historical processes that have shaped a previously well studied yet poorly understood region. Our results show that a majority of divergences across these diverse snake lineages appeared to be essentially coincident in time and space (Fig. 3). These findings suggest coordinated vicariance as a dominating force in speciation in the Middle American snake lineages studied. We found that some boundaries show great synchrony among diverse lineages (breaks in Talamanca, MotaguaPolochic and Tehuantepec). Other biogeographic breaks show evidence of multiple divergence time periods, evidenced by comparisons of credibility intervals and from the Approximate Bayesian Computation analyses; these

Comparative phylogeographic data coupled with divergence time estimates can illuminate much about a region’s history. When divergence time estimates from independent studies are compared, however, we expect that substantial error in absolute divergence time estimates may often exist, due largely to differences in dating approaches and interpretations of the fossil and geological record (Heads 2005). In such comparative studies the precise absolute divergence times are often much less important than the estimates of the relative coordination of divergence events across lineages. This is particularly the case when inferring the number of discrete temporal windows of divergence, such as in the current study. To circumvent this issue here, we assembled multiple related lineages into a single dataset, and use this large combined dataset for jointly estimating divergence times and instances of co-divergence. Because the same calibration assumptions are applied to the entire tree, and also because relative divergence time estimates are highly robust within a tree, this approach can provide more precise estimates of the relative timing of divergence across lineages. Our absolute dates are consistent with our previous work with these snakes (Castoe et al. 2009, Daza et al. 2009), most likely because of the very similar divergence dating strategies and calibrations, and they are also consistent with other independent studies (Sanders and Lee 2008, Kelly et al. 2009). A few studies, however, on particular lineages we included in our dataset have estimated older node ages than we have here, particularly for deeper nodes. We interpret these discrepancies in two ways. First, fossil snakes are extremely scarce for certain taxonomic groups and usually the available and non-ambiguous ones are used as calibrations for fairly recent cladogenetic events since most of the fossils come from the Pliocene and Pleistocene (Holman 2000); using recent calibrations points to estimate older nodes has been identified as a potential source of error (Ho et al. 2008). Second, discrepancies are likely to occur when different calibrations points are used. For example Wu¨ ster et al. (2005, 2008) and Devitt (2006) incorporated geological information (the emergence of the Mexican transvolcanic axis and the Isthmus of Panama, respectively) instead of fossil data (as in our case) for dating Trimorphodon and Crotalus divergences, respectively. Given

BIOGEOGRAPHIC BOUNDARIES

Pr(Ψ D)

0.8

2 3 Nicaraguan Depression (3)

SYMPOSIUM

multi-modal periods of divergences appear to characterize the breaks in Panama and Nicaragua. Since the Miocene, Middle America has continually endured extensive terrain dynamics powered by tectonic activity, and we interpret our results as indicating that this dynamic process has been the dominant force in lineage diversification, and that such tectonically-driven vicariance explains the remarkably high degree of synchronization among such ecologically distinct lineages. In some cases, however, we do find evidence that intrinsic factors (e.g. dispersal and ecological features) may have also played roles in lineage divergence times, rather than purely extrinsic (e.g. tectonic) forces. Examples of this include divergences along the Isthmus of Panama, the divergence of Bothriechis across the Nicaraguan Depression, and the divergence of Porthidium along the Isthmus of Tehuantepec.

BIOGEOGRAPHIC BOUNDARIES ON

SYMPOSIUM ISSUE

IBS SPECIAL

the uncertainty in the fossil and geological record we would not necessarily expect multiple studies converge to the same dates (given the use of different calibrations). Because of the potential biases that different choices of calibrations may impose on estimates of shared divergence, our combination of all data into a single dataset, and our ability to rely on highly accurate inferences of relative divergence time across lineages (rather than calibration points), we expect our results of shared divergence to be particularly robust.

Divergence across the Middle America South America transition (1) The area between southern Honduras and northwestern Colombia is biogeographically important because it represents the intermediate land connection between the two main continental landmasses of the western Hemisphere, as well as the division between two oceans. The details of the dynamic connections between these landmasses from the Miocene onward, however, remain controversial. Recent phylogenetic and biogeographic evidence has uncovered complex patterns that suggest that biotic interchange between terrestrial fauna may have entailed multiple dispersal and vicariant events that occurred across a fairly broad time scale, far broader than the time surrounding the final closure of the Isthmus of Panama around 3.5 Ma (Collins et al. 1996, Bermingham and Martin 1998, Pennington and Dick 2004, Koepfli et al. 2007). Our analyses indicate that recurrent diversification has occurred since the middle Miocene (Fig. 3, 4). MsBayes suggests two main episodes of diversification, although there is no strong demarcation between these two episodes based on the 95% CIs of divergence times. Although this study is limited in taxonomic scope, it is the first to include explicit temporal evidence across multiple terrestrial lineages, showing evidence (independent of assumptions of fossil calibrations, etc.) for multiple episodes of lineage divergence among the continents. A similar disparate pattern has been recently found for divergences between marine geminate species on either side of the isthmus (Marko 2002, Hurt et al. 2009), suggesting that both terrestrial and marine species responded in a similar broad temporal fashion. Collectively, our data and othersâ&#x20AC;&#x2122; raise the question of whether pre-final closure dispersal/vicariant events of terrestrial lineages were all based on overwater dispersal, or instead, multiple transient landconnections joined parts of Lower Central America and South America prior to the final isthmus closure. Given the number of Pliocene and Miocene divergences associated with this region, the early transient land bridges hypothesis seems more likely, and warrants further evaluation with additional comparative data. It is notable that the final closure date for the Panamanian Isthmus at 3.5 Ma has been commonly used as a regional calibration point for previous marine and terrestrial biogeographic studies (Bermingham et al. 1997, WuÂ¨ ster et al. 2002, 2005, 2008). In the case of terrestrial studies, this practice is unsound because this time period probably represents a period of dispersal, rather than having any direct relevance to vicariance (and is thus not particularly useful in applying to divergence time estimates). More importantly, based on our results, we find evidence from multiple lineages that 350

divergence times across this boundary appear almost completely independent of this 3.5 Ma closure date (Fig. 3). Therefore, of all the biogeographic breaks we have examined here, this event represents one of the most problematic choices for use as a calibration point. Furthermore, recent evidence has shown that marine geminate species across both sides of the Isthmus diverged in a temporally staggered manner since the Miocene (Hurt et al. 2009), suggesting that this region represents a poor calibration point for both marine and terrestrial divergence times estimates. Divergence across the Talamanca Cordillera (2) The Talamanca mountain range and associated cordilleras running down the spine of Costa Rica and northwestern Panama represent a composite of Neogene and Quaternary mountains with an active geomorphological history since the Miocene (Marshall et al. 2003, MacMillan et al. 2004, Marshall 2007). Phylogenetically, lineages along the Pacific slope of Costa Rica/Panama and those in northern South America tend to be more closely related than are lineages on either side of the Talamanca ridge (Castoe et al. 2005, Weigt et al. 2005, Crawford et al. 2007, Daza et al. 2009). Combining the results from msBayes and the pooled posterior distributions of divergence times, our results favor a single vicariant event centered around 3.9 Ma (Fig. 3). The timing of this event near the final closure of the isthmus of Panama raises the question of whether this event was driven by the final tectonic uplifts of the Talamancan ridge (MacMillan et al. 2004) or possibly the large-scale changes in habitat distributions brought about through changes in ocean currents and weather patterns accompanying the closure of the isthmus of Panama. Divergence across the Nicaraguan Depression (3) The Nicaraguan Depression is a lowland corridor running from the Caribbean to the Pacific near the border between Costa Rica and Nicaragua. Marine sediments indicate that a seaway existed multiple times here during the Pliocene, separating regions to the north and south (Coates and Obando 1996). There is also evidence implying that a continuous peninsular landmass connected Honduras with modern day Costa Rica during the Miocene (Kirby and MacFadden 2005, Kirby et al. 2008), contrasting a hypothesis that this region comprised a set of islands interconnected by shallow waters during the Miocene (Coates and Obando 1996). Two lineages of highland pitvipers (Atropoides and Bothriechis) show largely overlapping early divergences over this area, whereas a third highland pitviper lineage (Cerrophidion) and the lowland lineage (the colubrid Leptodeira septentrionalis) show substantially later divergences. The posterior distributions cluster in a staggered manner that broadly extends from 4 10 Ma (Fig. 3), countering a hypothesis of a single coordinated divergence event. This multi-modal pattern of divergence is also evident in the msBayes results that show diffuse posterior density across a broad range of discrete divergence events from one to five, although a majority of posterior density is

351

ISSUE

Mexico’s Isthmus of Tehuantepec has long been considered a biogeographic break for both highland and lowland species (Marshall and Liebherr 2000, Parkinson et al. 2000, Morrone and Ma´rquez 2001). Geological evidence suggests that from the late Miocene through late Pliocene, an extensive downdropping of the eastern block along the

In this study we investigated Middle American regional historical biogeography by focusing on particular spatial areas known to be major biogeographic boundaries, and characterizing these boundaries by synthesizing information about how multiple lineages temporally diverged across them. The large number of independent lineage diversification events examined provides new data for testing existing hypotheses of regional patterns of lineage diversification, and further evidence for generating new hypotheses of Neotropical diversification. We expect that our estimates of divergence, and the degree of synchronization, represent sound testable hypotheses for unstudied taxa or communities, certainly in cases where we found divergence to be highly correlated across lineages. Combining ABC statistical methods for inferring the coordination of divergences across lineages (Hickerson et al. 2006b, 2007, Leache´ et al. 2007, Hickerson and Meyer 2008) with analyses of posterior distributions of divergence times based on robust probabilistic methods from a combined phylogenetic dataset provided an ideal complementary strategy for dissecting shared divergence patterns. Because the use of pooled posterior distributions may obscure relevant lineage-specific responses to biogeographic boundaries, they should not be used as the only evidence for shared divergence among co-distributed taxa. However, in the absence of any other information about lineage divergence (i.e. when calibration points are scarce), pooled posterior distributions of divergence times are useful as an a priori expectation of divergence time for unstudied species, or even as a Bayesian prior in phylogenetic analyses.

IBS SPECIAL

Divergence across the Isthmus of Tehuantepec (5)

Conclusion

BIOGEOGRAPHIC BOUNDARIES

Recent studies have uncovered a sharp phylogeographic break along the axis where the Maya and Chortis tectonic blocks (in northern Middle America) come in contact and form a long NE SW trending basin along the Motagua Polochic Fault zone (Perdices et al. 2005, Devitt 2006, Concheiro-Pe´rez et al. 2007). The continued tectonic activity uplifting highlands on either side of this basin, and its further entrenchment, appears to have generated divergence events in both lowland and highland species. Based on the pooled posterior distribution of divergence times, credibility intervals and msBayes results (Fig. 3, 4), we find a clear pattern of concentrated temporal divergence across multiple species that span this area, suggesting that this zone acted as a barrier to many different lineages over this period from 3 8 Ma (Fig. 3). Our phylogeographic analysis suggests the primarily lowland snake genera, Trimorphodon and Leptodeira, diverged across this barrier in near concert with the highland lineages Bothriechis, Atropoides and Cerrophidion (Fig. 2, 3). Terrestrial fossil information for Middle America is scarce, therefore the regional calibration for dating purposes needs to rely either on the fossil record from relatively distant lineages, or be based on estimated evolutionary rates. Here, we find evidence that the Motagua Polochic Fault phylogeographic break may be a reasonably sound calibration point when no other information for regional calibrations is available. For example, the results of our pooled posterior distribution for the shared divergence across this break (Fig. 3) could be readily incorporated as a prior distribution for species divergence times in a Bayesian analysis when other useful calibration points are lacking, or a null hypothesis for other statistical tests in future studies.

Divergence across the Motagua Polochic Faults (4)

Tehuantepec fault zone resulted in a considerable reduction of the highlands and probably a marine embayment (Barrier et al. 1998). Given the cumulative evidence of diversification across multiple lineages on both sides of the Isthmus, a broad-reaching vicariant event during the Pliocene has been suggested as being responsible for the divergence of numerous lineages (Marshall and Liebherr 2000, Hasbu´ n et al. 2005, Mulcahy et al. 2006, Castoe et al. 2009). Our posterior distributions for divergence times strongly support this model, inferring a highly constrained temporal window at the end of the Pliocene when a majority of diversification events (4 of 6) occurred (Fig. 3). This window is consistent with proposals that events during the Pliocene severed gene flow among lineages straddling the isthmus (Hasbu´ n et al. 2005, Mulcahy et al. 2006, Leo´ n-Paniagua et al. 2007). However, the 95% credibility intervals (Fig. 3) and the msBayes results (Fig. 4) suggest that a second period of divergence also occurred earlier in the Miocene across the isthmus. Two genera, Crotalus and Porthidium, apparently diverged earlier, suggesting that a different geological/climatic event at the end of the Miocene (e.g. vegetation shifts; Cerling et al. 1997) may have been responsible for divergence in these two arid-adapted groups. Our data are thus consistent with hypotheses of broad vicariance across the isthmus due to Pliocene downdropping and seaway formation across the isthmus, but further suggest a more ancient divergence here affecting at least arid-adapted species.

SYMPOSIUM

centered over 2 events (Table 1, Fig. 4). A reasonable a priori expectation for divergences across this boundary may include rapid and highly coordinated divergence across multiple lineages due to the geo-tectonic model including seaway formation in the Pliocene. Instead, our data point to multiple periods (or one long broad period) of vicariance (and probably also dispersal) across the Nicaraguan Depression, rejecting a model centered on a single discrete barrier to gene flow coordinating divergences across lineages. Our data do fit an alternative model, that of Kirby and MacFadden (2005), which suggests a dynamic landmass may have transiently existed across the Nicaraguan Depression during the second half of the Miocene. This particular example highlights the important synergistic role in generating and testing hypotheses that comparative phylogeographic studies can have in conjunction with geological-tectonic data.

BIOGEOGRAPHIC BOUNDARIES ON

SYMPOSIUM

Advances in estimation and comparison of divergence times, coupled with the growing interest in phylogeographic research, will surely continue to illuminate new understanding of the roles that historical processes have played in generating the planet’s biodiversity. We found widespread evidence for a surprisingly high number of lineages showing coordinated divergence, and these divergences often fit previous expectations based on geological and tectonic evidence. In other cases, however, (e.g. Nicaraguan Depression) we found substantial evidence supporting one geological model (dynamic transient land connections) over other models. Overall, our findings are highly encouraging, and strongly implicate the existence of an underlying and unifying model of Middle American biogeography that is tractable to assemble and eventually comprehend. The level of detailed information emerging from comparative phylogeographic studies, augmented with information from the fossil, geological, tectonic, and climatic records, hold great promise for accelerating insight into how biodiversity was established on the planet, and also how it may be shaped by climate change and anthropogenic disturbance.

Acknowledgements We acknowledge participants of the Boundaries Symposium at the International Biogeography Society Meeting, including B. Riddle, D. Hafner, J. Morrone, T. Pennington, J. Klicka, and R. Whittaker, for stimulating discussion, suggestions, and encouragement. We thank J. Castoe, A. Fenwick, A. P. Jason de Koning, S. Johnson, H. Kalkvik, J. Reece, K. Kozak, R. Tursi, and two anonymous reviewers for constructive comments on the manuscript, and T. Hether for his help with the R script. We acknowledge the support of a National Inst. of Health Training Grant (LM009451) to TAC, and a National Science Foundation (NSF) Collaborative Research grant to CLP (DEB 0416000). Many sequences used in this study were from tissues generously donated by J. A. Campbell and E. N. Smith, who’s fieldwork was supported by the NSF (grants DEB 0613802 and DEB 9705277 to J. A. Campbell and DEB 0416160 to E. N. Smith) and the Inst. Bioclon (to E. N. Smith).

IBS SPECIAL

ISSUE

References Arbogast, B. S. and Kenagy, G. J. 2001. Comparative phylogeography as an integrative approach to historical biogeography. J. Biogeogr. 28: 819 825. Barrier, E. et al. 1998. Neotectonic evolution of the Isthmus of Tehuantepec (southeastern Mexico). Tectonophysics 287: 77 96. Bermingham, E. and Martin, A. P. 1998. Comparative mtDNA phylogeography of neotropical freshwater fishes: testing shared history to infer the evolutionary landscape of lower Central America. Mol. Ecol. 7: 499 517. Bermingham, E. and Moritz, C. 1998. Comparative phylogeography: concepts and applications. Mol. Ecol. 7: 367 369. Bermingham, E. et al. 1997. Fish biogeography and molecular clocks: perspectives from the Panamanian Isthmus. In: Kocher, T. D. and Stepien, C. A. (eds), Molecular systematics of fishes. Academic Press, pp. 113 128. Burbrink, F. T. and Pyron, R. A. 2008. The taming of the skew: estimating proper confidence intervals for divergence dates. Syst. Biol. 57: 317 328. Campbell, J. A. 1999. Distribution patterns of amphibians in Middle America. In: Duellman, W. E. (ed.), Patterns of

352

distribution of amphibians: a global perspective. Johns Hopkins Univ. Press, pp. 111 210. Castoe, T. A. and Parkinson, C. L. 2006. Bayesian mixed models and the phylogeny of pitvipers (Viperidae: Serpentes). Mol. Phylogenet. Evol. 39: 91 110. Castoe, T. A. et al. 2005. Modeling nucleotide evolution at the mesoscale: the phylogeny of the Neotropical pitvipers of the Porthidium group (Viperidae: Atropoides, Cerrophidion, Porthidium). Mol. Phylogenet. Evol. 37: 881 898. Castoe, T. A. et al. 2007a. Higher-level phylogeny of Asian and American coralsnakes, their placement within the Elapidae (Squamata), and the systematic affinities of the enigmatic Asian coralsnake Hemibungarus calligaster (Wiegmann, 1834). Zool. J. Linn. Soc. 151: 809 831. Castoe, T. A. et al. 2007b. Phylogeographic structure and historical demography of the western diamondback rattlesnake (Crotalus atrox): a perspective on North American desert biogeography. Mol. Phylogenet. Evol. 42: 193 212. Castoe, T. A. et al. 2009. Comparative phylogeography of pitvipers suggests a consensus of ancient Middle American highland biogeography. J. Biogeogr. 36: 88 103. Cerling, T. E. et al. 1997. Global vegetation change through the Miocene/Pliocene boundary. Nature 389: 153 158. Coates, A. G. and Obando, J. A. 1996. The geologic evolution of the Central American isthmus. In: Jackson, J. B. C. et al. (eds), Evolution and environment in tropical America. Univ. of Chicago Press, pp. 21 56. Collins, L. S. et al. 1996. Earliest evolution associated with closure of the Tropical American Seaway. Proc. Nat. Acad. Sci. USA 93: 6069 6072. Concheiro-Pe´rez, G. A. et al. 2007. Phylogeny and biogeography of 91 species of heroine cichlids (Teleostei: Cichlidae) based on sequences of the cytochrome b gene. Mol. Phylogenet. Evol. 43: 91 110. Coney, P. J. 1982. Plate tectonic constraints on the biogeography of Middle America and the Caribbean region. Ann. Mo. Bot. Gard. 69: 432 443. Crawford, A. J. et al. 2007. The role of tropical dry forest as a long-term barrier to dispersal: a comparative phylogeographical analysis of dry forest tolerant and intolerant frogs. Mol. Ecol. 16: 4789 4807. Daza, J. M. et al. 2009. Complex evolution in the Neotropics: the origin and diversification of the widespread genus Leptodeira (Serpentes: Colubridae). Mol. Phylogenet. Evol. 53: 653 667. Devitt, T. J. 2006. Phylogeography of the western lyresnake (Trimorphodon biscutatus): testing aridland biogeographical hypotheses across the Nearctic-Neotropical transition. Mol. Ecol. 15: 4387 4407. Drummond, A. J. and Rambaut, A. 2007. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evol. Biol. 7: 214. Drummond, A. J. et al. 2006. Relaxed phylogenetics and dating with confidence. PLoS Biol. 4: 1 12. Graur, D. and Martin, W. 2004. Reading the entrails of chickens: molecular timescales of evolution and the illusion of precision. Trends Genet. 20: 80 86. Hasbu´ n, C. R. et al. 2005. Mitochondrial DNA phylogeography of the Mesoamerican spiny-tailed lizards (Ctenosaura quinquecarinata complex): historical biogeography, species status and conservation. Mol. Ecol. 14: 3095 3107. Head, J. J. et al. 2005. First report of snakes (Serpentes) from the Late Middle Eocene Pondaung formation, Myanmar. J. Vertebr. Paleontol. 25: 246 250. Heads, M. 2005. Dating nodes on molecular phylogenies: a critique of molecular biogeography. Cladistics 21: 62 78.

BIOGEOGRAPHIC BOUNDARIES IBS SPECIAL ISSUE

353

Marshall, J. S. et al. 2003. Landscape evolution within a retreating volcanic arc, Costa Rica, Central America. Geology 31: 419 422. Morrone, J. J. 2001. Biogeografı´a de Ame´rica Latina y el Caribe. Manuales y Tesis, SEA. Morrone, J. J. and Ma´rquez, J. 2001. Halffter’s Mexican transition zone, beetle generalized tracks, and geographical homology. J. Biogeogr. 28: 635 650. Mulcahy, D. G. et al. 2006. Historical biogeography of lowland species of toads (Bufo) across the Trans-Mexican Neovolcanic Belt and the Isthmus of Tehuantepec. J. Biogeogr. 33: 1889 1904. Nelson, G. J. and Platnick, N. I. 1981. Systematics and biogeography: cladistics and vicariance. Columbia Univ. Press. Nicholas, K. B. and Nicholas, H. B. Jr 1997. GeneDoc: a tool for editing and annotating multiple sequence alignments. Distributed by the authors. Nylander, J. A. A. 2004. MrModeltest v2. Distributed by the author. Parkinson, C. L. et al. 2000. Phylogeography of the pitviper clade Agkistrodon: historical ecology, species status, and conservation of cantils. Mol. Ecol. 9: 411 420. Parkinson, C. L. et al. 2002. Multigene phylogenetic analysis of pitvipers, with comments on their biogeography. In: Schuett, G. W. et al. (eds), Biology of the vipers. Eagle Mountain Publ., USA, pp. 93 110. Parmley, D. and Holman, J. A. 2007. Earliest fossil record of a pigmy rattlesnake (Viperidae: Sistrurus Garman). J. Herpetol. 41: 141 144. Pennington, R. T. and Dick, C. W. 2004. The role of immigrants in the assembly of the South American rainforest tree flora. Phil. Trans. R. Soc. B 359: 1611 1622. Perdices, A. et al. 2005. Evolutionary history of the synbranchid eels (Teleostei: Synbranchidae) in Central America and the Caribbean islands inferred from their molecular phylogeny. Mol. Phylogenet. Evol. 37: 460 473. Rage, J.-C. et al. 1992. A colubrid snake in the late Eocene of Thailand: the oldest known Colubridae (Reptilia, Serpentes). C. R. Acad. Sci. Paris 314: 1085 1089. Rambaut, A. and Drummond, A. J. 2007. Tracer v1.4. <http:// beast.bio.ed.ac.uk/Tracer>. Rambaut, A. and Drummond, A. J. 2008. TreeStat v1.2: tree statistic calculation tool. <http://tree.bio.ed.ac.uk/software/ treestat/>. Ree, R. H. and Smith, S. A. 2008. Maximum likelihood inference of geographic range evolution by dispersal, local extinction, and cladogenesis. Syst. Biol. 57: 4 14. Ree, R. H. and Sanmartı´n, I. 2009. Prospects and challenges for parametric models in historical biogeographical inference. J. Biogeogr. 36: 1211 1220. Richards, C. L. et al. 2007. Distribution modelling and statistical phylogeography: an integrative framework for generating and testing alternative biogeographical hypotheses. J. Biogeogr. 34: 1833 1845. Riddle, B. R. et al. 2008. The role of molecular genetics in sculpting the future of integrative biogeography. Prog. Phys. Geogr. 32: 173 202. Ronquist, F. 1997. Dispersal vicariance analysis: a new approach to the quantification of historical biogeography. Syst. Biol. 46: 195 203. Sanders, K. L. and Lee, M. S. Y. 2008. Molecular evidence for a rapid late-Miocene radiation of Australasian venomous snakes (Elapidae, Colubroidea). Mol. Phylogenet. Evol. 46: 1180 1188. Savage, J. M. 1982. The enigma of the Central American herpetofauna: dispersal or vicariance? Ann. Mo. Bot. Gard. 69: 464 547.

SYMPOSIUM

Hickerson, M. J. and Meyer, C. P. 2008. Testing comparative phylogeographic models of marine vicariance and dispersal using a hierarchical Bayesian approach. BMC Evol. Biol. 8: 322. Hickerson, M. J. et al. 2006a. Comparative phylogeographic summary statistics for testing simultaneous vicariance. Mol. Ecol. 15: 209 223. Hickerson, M. J. et al. 2006b. Test for simultaneous divergence using approximate bayesian computation. Evolution 60: 2435 2453. Hickerson, M. J. et al. 2007. msBayes: pipeline for testing comparative phylogeographic histories using hierarchical approximate Bayesian computation. BMC Bioinform. 8: 268. Ho, S. Y. W. et al. 2008. The effect of inappropriate calibration: three case studies in molecular ecology. PLoS One 3: e1615. Holman, J. A. 2000. Fossil snakes of North America: origin, evolution, distribution, paleoecology. Indiana Univ. Press. Hurt, C. et al. 2009. A multilocus test of simultaneous divergence across the Isthmus of Panama using snapping shrimp in the genus Alpheus. Evolution 63: 514 530. Iturralde-Vinent, M. A. 2006. Meso-Cenozoic Caribbean paleogeography: implications for the historical biogeography of the region. Int. Geol. Rev. 48: 791 827. Jackson, J. B. C. et al. (eds) 1996. Evolution and environment in tropical America. Univ. Chicago Press. Kelly, C. M. R. et al. 2009. Phylogeny, biogeography and classification of the snake superfamily Elapoidea: a rapid radiation in the late Eocene. Cladistics 25: 38 63. Kirby, M. X. and MacFadden, B. 2005. Was southern Central America an archipelago or a peninsula in the middle Miocene? A test using land-mammal body size. Palaeogeogr. Palaeoclimatol. Palaeoecol. 228: 193 202. Kirby, M. X. et al. 2008. Lower Miocene stratigraphy along the Panama Canal and its bearing on the Central American peninsula. PLoS One 3: e2791. Knowles, L. L. and Carstens, B. C. 2007. Estimating a geographically explicit model of population divergence. Evolution 61: 477 493. Koepfli, K.-P. et al. 2007. Phylogeny of the Procyonidae (Mammalia: Carnivora): molecules, morphology and the Great American Interchange. Mol. Phylogenet. Evol. 43: 1076 1095. Larkin, M. A. et al. 2007. Clustal W and Clustal X version 2.0. Bioinformatics 23: 2947 2948. Leache´, A. D. et al. 2007. Two waves of diversification in mammals and reptiles of Baja California revealed by hierarchical Bayesian analysis. Biol. Lett. 3: 646 650. Leo´ n-Paniagua, L. et al. 2007. Diversification of the arboreal mice of the genus Habromys (Rodentia: Cricetidae: Neotominae) in the Mesoamerican highlands. Mol. Phylogenet. Evol. 42: 653 664. MacMillan, I. et al. 2004. Middle Miocene to present plate tectonic history of the southern Central American Volcanic Arc. Tectonophysics 392: 325 348. Mann, P. et al. 2007. Overview of plate tectonic history and its unresolved tectonic problems. In: Bundschuh, J. and Alvarado, G. E. (eds), Central America: geology, resources, and hazards. Taylor and Francis, pp. 205 241. Marko, P. B. 2002. Fossil calibration of molecular clocks and the divergence times of geminate species pairs separated by the Isthmus of Panama. Mol. Biol. Evol. 19: 2005 2021. Marshall, C. J. and Liebherr, J. K. 2000. Cladistic biogeography of the Mexican transition zone. J. Biogeogr. 27: 203 216. Marshall, J. S. 2007. The geomorphology and physiographic provinces of Central America. In: Bundschuh, J. and Alvarado, G. E. (eds), Central America: geology, resources, and hazards. Taylor and Francis, pp. 75 122.

Download the Supplementary material as file E6281 from <www.oikos.ekol.lu.se/appendix>.

IBS SPECIAL

ISSUE

SYMPOSIUM

BIOGEOGRAPHIC BOUNDARIES

Vidal, N. et al. 2009. Snakes (Serpentes). In: Hedges, S. B. and Kumar, S. (eds), The timetree of life. Oxford Univ. Press, pp. 390 397. Weigt, L. A. et al. 2005. Biogeography of the tu´ ngara frog, Physalaemus pustulosus: a molecular perspective. Mol. Ecol. 14: 3857 3876. Whitmore, T. C. and Prance, G. T. (eds) 1987. Biogeography and Quaternary history in tropical America. Clarendon Press. Wu¨ ster, W. et al. 2002. Origin and evolution of the South American pitviper fauna: evidence from mitochondrial DNA sequence analysis. In: Schuett, G. W. et al. (eds), Biology of the vipers. Eagle Mountain Publ., USA, pp. 111 128.

354

Wu¨ ster, W. et al. 2005. Tracing an invasion: landbridges, refugia, and the phylogeography of the Neotropical rattlesnake (Serpentes: Viperidae: Crotalus durissus). Mol. Ecol. 14: 1095 1108. Wu¨ ster, W. et al. 2008. A nesting of vipers: phylogeny and historical biogeography of the Viperidae (Squamata: Serpentes). Mol. Phylogenet. Evol. 49: 445 459. Zamudio, K. R. and Greene, H. W. 1997. Phylogeography of the bushmaster (Lachesis muta: Viperidae): implications for neotropical biogeography, systematics and conservation. Biol. J. Linn. Soc. 62: 421 442.

Ecography 33: 355 361, 2010 doi: 10.1111/j.1600-0587.2010.06266.x # 2010 The Author. Journal compilation # 2010 Ecography Subject Editor: Douglas A. Kelt. Accepted 4 February 2010

Transition zones, located at the boundaries between biogeographic regions, represent events of biotic hybridization, promoted by historical and ecological changes. They deserve special attention, because they represent areas of intense biotic interaction. In its more general sense, the Mexican Transition Zone is a complex and varied area where Neotropical and Nearctic biotas overlap, from southwestern USA to Mexico and part of Central America, extending south to the Nicaraguan lowlands. In recent years, panbiogeographic analyses have led to restriction of the Mexican Transition Zone to the montane areas of Mexico and to recognize five smaller biotic components within it. A cladistic biogeographic analysis challenged the hypothesis that this transition zone is biogeographically divided along a north-south axis at the Transmexican Volcanic Belt, as the two major clades found divided Mexico in an east-west axis. This implies that early Tertiary geological events leading to the convergence of Neotropical and Nearctic elements may be younger (Miocene) than those that led to the east-west pattern (Paleocene). The Mexican Transition Zone consists of five biogeographic provinces: Sierra Madre Occidental, Sierra Madre Oriental, Transmexican Volcanic Belt, Sierra Madre del Sur, and Chiapas. Within this transition zone, at least four cenocrons have been identified: Paleoamerican, Nearctic, Montane Mesoamerican, and Tropical Mesoamerican. Future studies should continue refining the identification of cenocrons and the reconstruction of a geobiotic scenario, as well as integrating ecological biogeographic studies, to allow a more complete understanding of the patterns and processes that have caused the biotic complexity of this transition zone.

355

ISSUE

and geographic conditions are favorable, organisms actively expanded their geographic distribution according to their dispersal capabilities, acquiring what we now view as their ancestral distribution (the role of dispersal). When the organisms have occupied all the available space, their distribution may stabilize, allowing the isolation of populations in different sectors of the area, and the differentiation of new species through the appearance of geographic barriers (the role of vicariance). To analyze the resulting complex patterns, biogeographers need to define specific questions and determine the most appropriate methods to answer them; importantly, this should be integrated within a coherent framework. Evolutionary biogeography integrates distributional, phylogenetic, molecular, and paleontological data to discover biogeographic patterns and assess the historical changes that have shaped them. It follows five steps (Fig. 1), each corresponding to particular questions, methods, and techniques (Morrone 2009). Panbiogeography and methods for identifying areas of endemism are used to identify biotic components, which are the basic units of evolutionary biogeography. Cladistic biogeography uses phylogenetic data to test the historical relationships between these biotic components. Based on the results of the panbiogeographic

IBS SPECIAL

Transition zones are located at the boundaries between biogeographic regions or realms (Darlington 1957), and represent events of biotic ‘‘hybridization’’, promoted by historical and ecological changes that allowed the mixture of different cenocrons (Morrone 2009). They deserve special attention, because boundaries between biogeographic regions are not static lines, but rather areas of intense biotic interaction (Ruggiero and Ezcurra 2003). The Mexican Transition Zone is a complex area where Neotropical and Nearctic biotas overlap, from the southwestern USA to Mexico and part of Central America, extending south to the Nicaraguan lowlands (Darlington 1957, Halffter 1962, 1964, 1972, 1974, 1976, 1978, 1987). Several authors have recognized the special status of this transitional biota from different perspectives. In recent years, several studies have been published under an implicit evolutionary biogeographic approach focused on this interesting area. I integrate these studies into a coherent framework that helps explain the biotic evolution of the Mexican Transition Zone. During the 19th and 20th centuries, biogeographers debated the mechanisms underlying biotic evolution, but in recent years some authors have concluded that both dispersal and vicariance are relevant processes (Brooks and McLennan 2001, Morrone 2009). Under favorable climatic

BIOGEOGRAPHIC BOUNDARIES

J. J. Morrone (juanmorrone2001@yahoo.com.mx), Museo de Zoologıá ‘‘Alfonso L. Herrera’’, Depto de Biologıá Evolutiva, Fac. de Ciencias, Univ. Nacional Autońoma de Me´xico (UNAM), Apartado postal 70-399, 04510 Mexico, D.F., Mexico.

Juan J. Morrone

SYMPOSIUM

Fundamental biogeographic patterns across the Mexican Transition Zone: an evolutionary approach

BIOGEOGRAPHIC BOUNDARIES ON

SYMPOSIUM ISSUE

IBS SPECIAL

Figure 1. Steps of an evolutionary biogeographic analysis (Morrone 2009).

and cladistic biogeographic analyses, a regionalization or biogeographic classification may be achieved. Intraspecific phylogeography, molecular clocks, and fossils may be incorporated to help identify the different cenocrons that become integrated in a biotic component. Finally, the geological and biological knowledge available can be integrated to construct a geobiotic scenario that may help explain the way the biotic components analyzed evolved. This approach does not imply that every biogeographer must follow all the steps, but that anyone may articulate a specific biogeographic question and choose the most appropriate method to answer it, and given some time, as the different analyses accumulate, coherent theories are formulated by their integrating. This approach, framed under integrative pluralism (Mitchell 2002), does not imply an eclectic or ‘‘anything goes’’ approach, but rather that the different methods give partial solutions when answering particular questions. Within the Mexican Transition Zone, there are studies corresponding to the five steps, which I examine and briefly discuss herein.

Identification of biotic components Biotic components are sets of spatio-temporally integrated taxa that coexist in given areas. During the 20th century, several authors recognized biogeographic provinces for 356

Mexico (Smith 1941, Goldman and Moore 1945, Cabrera and Willink 1973, Rzedowski 1978, Casas-Andreu and Reyna-Trujillo 1990, Ferrusquıá-Villafranca 1990, Ramı´rez-Pulido and Castro-Campillo 1990, Rzedowski and Reyna-Trujillo 1990), which can be considered preliminary as biotic components. Recent panbiogeographic analyses have tested these components and analyzed their interrelationships. Morrone and Ma´rquez (2001) analyzed 134 beetle (Coleoptera) species, documenting both a northern and a southern generalized track (Fig. 2A). The former comprised montane areas (Sierra Madre Occidental, Sierra Madre Oriental, Transmexican Volcanic Belt, Balsas Basin, and Sierra Madre del Sur), while the latter included the Sierra Madre de Chiapas and lowland areas in Chiapas, the Mexican Gulf, and the Mexican Pacific Coast, reaching south to the Isthmus of Panama). The northern track included the highest latitudinal and altitudinal mixture of Nearctic and Neotropical cenocrons, with a major Nearctic influence at higher altitudes and a higher Neotropical influence at lower altitudes. Owing to its mixed biota and its placement between the other regions, this generalized track has been considered to represent the Mexican Transition Zone in the strict sense (Morrone 2005, 2006). A previous study, based on mammals (Ortega and Arita 1998), arrived at similar conclusions. Escalante et al. (2004) analyzed the distributional patterns of 46 Mexican land mammal species belonging to the Nearctic biotic component to determine the southernmost boundary of the Nearctic region in the Mexican Transition Zone. They obtained six generalized tracks (Fig. 2B). The California generalized track lies in the northern part of the California Peninsula, in the state of Baja California, occupying the northern Sierras of Baja California, in the Sierra de San Pedro Ma´rtir, Sierra de Jua´rez, and the northwestern coastal chaparral. The CenterGulf generalized track crosses from northern Hidalgo and Veracruz, to southern Veracruz, Puebla, Tlaxcala, and the state of Mexico. The Center-North Pacific generalized track is represented by species distributed on the Sierra Madre Occidental and the Transmexican Volcanic Belt, crossing Durango, Jalisco, Michoacań, and the state of Mexico. The Center-South Pacific generalized track begins in southern Sinaloa, crosses Nayarit, Jalisco, and Michoacań, where it bifurcates: one part crosses the states of Mexico, Puebla, and Oaxaca, and ends in Chiapas, while the other crosses the southwestern portion of the state of Mexico and Guerrero, ending in western Oaxaca. The Isthmus of Tehuantepec generalized track begins in Guerrero and Veracruz, in both the Pacific and Gulf coasts, both parts join in Oaxaca, and then continue to Chiapas. The Chiapas generalized track lies in the Altos de Chiapas pine-oak and tropical montane cloud forests. Intersection of these six generalized tracks led Escalante et al. (2004) to identify nine nodes (Fig. 2B): three in the Transmexican Volcanic Belt, one in the southern Sierra Madre Oriental, one in the eastern Sierra Madre del Sur, one in the highlands of Chiapas, and three in the boundaries between two provinces. They concluded that taxa isolated in the highlands of Chiapas (as well as Guatemala) at the end of the Pleistocene may represent the southernmost Nearctic relicts in Mesoamerica, and that the other biogeographic provinces, together with the Sierra

SYMPOSIUM

BIOGEOGRAPHIC BOUNDARIES

IBS SPECIAL

ISSUE

357

Figure 2. (A) two generalized tracks identified by Morrone and MaÂ´rquez (2001); (B) six generalized tracks and nine nodes identified by Escalante et al. (2004); (C) general area cladogram obtained by Escalante et al. (2007); (D) provinces of the Mexican Transition Zone. chi, Chiapas; smoc, Sierra Madre Occidental; smor, Sierra Madre Oriental; sms, Sierra Madre del Sur; tvb, Transmexican Volcanic Belt.

BIOGEOGRAPHIC BOUNDARIES ON

SYMPOSIUM ISSUE

IBS SPECIAL

Madre Occidental and Balsas Basin provinces, represent the Mexican Transition Zone in the strict sense. Several other studies have contributed to the recognition of similar generalized tracks and nodes, and to the identification of smaller generalized tracks (Luna-Vega et al. 1999, 2000, 2001, Morrone et al. 1999, Morrone and Escalante 2002, Escalante et al. 2003, 2005, Ma´rquez and Morrone 2003, Ochoa et al. 2003, Morrone and Gutie´rrez 2005, Espinosa et al. 2006, Contreras-Medina et al. 2007a, Marin˜ o-Pe´rez et al. 2007, Toledo et al. 2007, Garcı´aMarmolejo et al. 2008).

Testing relationships among biotic components Once biotic components have been identified, they can be tested using cladistic biogeographic analyses. Some authors have provided cladistic biogeographic analyses of Mexico (Liebherr 1991, 1994, Marshall and Liebherr 2000, Flores Villela and Goyenechea 2001, Espinosa et al. 2006, Contreras-Medina et al. 2007b). Escalante et al. (2007) analyzed 40 plant and animal taxa distributed in Mexico and extending to both the Nearctic and Neotropical regions. Each taxonomic cladogram was transformed into a taxon-area cladogram by replacing its terminal taxa with the areas in which they occur. A paralogyfree subtrees analysis (Nelson and Ladiges 1996) allowed the construction of a general area cladogram (Fig. 2C), which showed two main clades. The Mexican Gulf, Tamaulipas, and Yucatan provinces are included in one clade, which forms the lowland region of eastern Mexico along the Caribbean coastline as far north as southern USA, possibly extending to Florida. The other clade includes the remaining provinces of central and western Mexico. The eastern boundary of the second clade the Sierra Madre Oriental, the Sierra Madre del Sur and Chiapas confines the provinces of the first clade. Within the second clade, a subclade consisting of the Balsas Basin, Chiapas, and Sierra Madre del Sur provinces principally forms the montane areas south of the Transmexican Volcanic Belt. This analysis challenges the commonly held opinion that Mexico is biogeographically divided along a north-south axis, as both main clades divide the country in an east-west axis. Escalante et al. (2007) concluded that the oldest eastwest division found did not contradict the currently recognized north-south axis that roughly divides Mexico into northern and southern portions on both sides of the Transmexican Volcanic Belt. The newly recognized biogeographical divide implies that early Tertiary geological events leading to the convergence of Neotropical and Nearctic elements in the Mexican Transition Zone may be younger (Miocene) than those that led to the east-west pattern (Paleocene). Additionally, the first clade may be recognized formally as a Caribbean region, separate from both the Neotropical and Nearctic regions, which may represent an older region that has existed independently since the Paleozoic. Previous biotic diversification studies of the Mexican Transition Zone need to be revised, because the division between the Nearctic and Neotropical regions in fact incorporates two biotic divisions, a north-south from the Miocene, and an east-west from the Paleocene. Given 358

the composite biotic nature of the Mexican Transition Zone, a complex pattern was expected to emerge; however, the results showed a single general area cladogram, with an east-west divide instead of the classical north-south division, implicit in previous explanations such as the Great American Biotic Interchange.

Regionalization Corroborated biotic components may be ordered hierarchically and used to provide a biogeographic classification. The current regionalization of Mexico (Morrone 2001, 2005, 2006) recognizes 14 biogeographic provinces: California, Baja California, Sonora, Mexican Plateau, Tamaulipas, Yucatań Penıńsula, Sierra Madre Occidental, Sierra Madre Oriental, Transmexican Volcanic Belt, Balsas Basin, Sierra Madre del Sur, Mexican Pacific Coast, Mexican Gulf, and Chiapas. Morrone (2005, 2006) assigned the Sierra Madre Occidental, Sierra Madre Oriental, Transmexican Volcanic Belt, Balsas Basin, and Sierra Madre del Sur provinces to the Mexican Transition Zone. Escalante et al. (2004) argued that Chiapas could be added to the Mexican Transition Zone, and EspinosaOrganista et al. (2008) considered the Balsas Basin to belong to the Neotropical region. The five biogeographic provinces of the Mexican Transition Zone (Fig. 2D), mainly recognized by species of plant and animal taxa (Morrone 2001), are as follows: 1) Sierra Madre Occidental province. Western Mexico, in the states of Chihuahua, Durango, Zacatecas, Sonora, Sinaloa, Nayarit, and Jalisco, above 1000 m altitude. This province has the highest Nearctic influence. 2) Sierra Madre Oriental province. Eastern Mexico, in the states of San Luis Potosı´, Coahuila, Hidalgo, Nuevo Leo´ n, Veracruz, Puebla, and Quere´taro, above 1500 m elevation. Biogeographic districts within the Sierra Madre Oriental province have been recognized by Espinosa-Organista et al. (2004). 3) Transmexican Volcanic Belt province. Central Mexico, in the states of Guanajuato, Mexico, Distrito Federal, Jalisco, Michoacań, Puebla, Oaxaca, Tlaxcala, and Veracruz. Biogeographic districts within the Sierra Madre Oriental province have been recognized by Torres Miranda and Luna Vega (2006). 4) Sierra Madre del Sur province. South central Mexico, from southern Michoacań to Guerrero, Oaxaca, and part of Puebla, above 1000 m altitude. 5) Chiapas province. Southern Mexico, Guatemala, Honduras, El Salvador, and Nicaragua; basically corresponds to the Sierra Madre de Chiapas, from 500 to 2000 m altitude.

Identification of cenocrons After establishing and testing the biotic components, timeslicing, intraspecific phylogeography, and molecular clocks can help establish when the cenocrons assembled within them. Cenocrons are sets of taxa that share the same biogeographic history, which constitute identifiable subsets within a biotic component by their common biotic origin and evolutionary history. From the aforementioned studies, it is evident that the complex biota of the Mexican Transition Zone consists of several cenocrons. Halffter’s

Construction of a geobiotic scenario Acknowledgements I thank Dave Hafner and Brett Riddle for inviting me to participate in the Symposium ‘‘Pattern and process at biogeographic boundaries’’ of the 4th Biennial Conference of the International Biogeography Society, Me´rida, Yucata´n, Mexico, 9 January, 2009. Tania Escalante, Douglas Kelt, Isolda Luna, Adolfo Navarro and two anonymous reviewers provided useful suggestions that helped improve the manuscript. I thank CONACyT project 80370 for financial support.

References Becerra, J. X. 2005. Timing the origin and expansion of the Mexican tropical dry forest. Proc. Nat. Acad. Sci. USA 102: 10919 10923.

359

ISSUE

Once biotic components and cenocrons have been identified, it is possible to construct a geobiotic scenario. By integrating biological and non-biological data, a plausible scenario can be developed to explain the episodes of vicariance/biotic divergence and dispersal/biotic convergence that have shaped biotic evolution. The east-west pattern detected by Escalante et al. (2007) corroborates the geological reconstructions of the Palaeocene to Miocene terrane migration and may help explain Mexican biotic complexity (Iturralde-Vinent 1998, Kerr et al. 1999). The collision of the Caribbean migrating plate 60 Ma predates the beginnings of the 49 Ma North-South American plate convergence, the latter event triggering the Great American Biotic Interchange during the Oligocene (ca 30 Ma) to

IBS SPECIAL

Future studies should continue refining the identification of biotic components and cenocrons and the reconstruction of a geobiotic scenario. Biogeographic regionalization of the Mexican Transition Zone should include more detailed analyses, especially to recognize biogeographic districts within the provinces. The complex biota of the Mexican Transition Zone should be dissected more thoroughly, by analyzing more taxa (especially non-insect invertebrates) from different methodological perspectives. On the other hand, I believe that integration of ecological biogeographic studies will allow a more complete understanding of the patterns and processes that have caused the biotic complexity of this remarkable transition zone. Ecological models might be particularly useful in providing clues to understand the biotic diversification in the Mexican Transition Zone. In special, island biogeography (Whittaker and Ferna´ndez-Palacios 2007) contains models and theories that may be applied to biotic components, which can be treated as islands.

BIOGEOGRAPHIC BOUNDARIES

Discussion

Miocene. Currently there is a geophysical debate over the origin and migration of the Caribbean plate since the late Mesozoic (Kerr et al. 1999, Mu¨ ller et al. 1999). The ‘‘Pacific model’’ states that the Caribbean plate originated in the Pacific Ocean and gradually moved eastward, passing between the North and South American plates prior to collision during the Miocene, into its present position (Kerr et al. 1999). The Caribbean plate may have carried in an evolving biota that remained isolated during the migration. This might explain the existence of a unique and older Caribbean biota that shares complex relationships with both Neotropical and Nearctic biotas. Other geological events, especially as related to the development of the Sierras Madre and the volcanism of the Transmexican Volcanic Belt (Ferrusquıá-Villafranca 1993, Ferrusquıá-Villafranca and Gonza´lez-Guzmań 2005), are relevant to explain the vicariant events that led to in situ differentiation within the Mexican Transition Zone (Morrone 2005). Given the Miocene age (ca 15 Ma) of the Transmexican Volcanic Belt, it seems very likely that the split between both subclades is a result of intense volcanic activity that led to a geographical barrier between northern and southern highland provinces.

SYMPOSIUM

(1987) distributional patterns based on species distributions, their closest relatives, species richness, degree of species sympatry, geological history, and diversity of habitats occupied (Liebherr 1991, 1994) may be considered as cenocrons (Morrone 2005). The Montane Mesoamerican cenocron includes Neotropical taxa that evolved in situ in humid montane habitats throughout Mexico and Central America. They have their highest species diversity in Central America, with species in the tropical and cloud forests in the mountains of Oaxaca, and further north and west along the Atlantic and Pacific watersheds. They have South American affinities and are hypothesized to have diversified in the Mexican Transition Zone in the Oligocene. The Paleoamerican cenocron includes Neotropical taxa that are restricted to Mexican montane areas, with ecological preferences for deserts, grasslands, and rain forests; they may also have some species in Central America. Their closest relatives are Old World temperate and tropical taxa. They underwent diversification prior to the Pliocene closure of the Isthmus of Tehuantepec. The Nearctic cenocron includes taxa that diversified in the mountains of Mexico during the Pliocene-Pleistocene. The Isthmus of Tehuantepec basically constitutes their southern limit, but these taxa may have a few species in Central America. Taxa generally occupy temperate conifer forests and grasslands above 1700 m of elevation. Their closest relatives are found further north, in the Nearctic region, along the Rocky Mountain Cordillera and areas across the USA and Canada. The Tropical Mesoamerican cenocron includes Neotropical taxa that evolved in humid lowland habitats throughout Mexico and Central America. They have South American affinities and are hypothesized to have diversified in the Mexican Transition Zone more recently than the taxa assigned to the other cenocrons, in the Pleistocene. Several recent phylogeographic and molecular clock studies of taxa from the Mexican Transition Zone (Sullivan et al. 1997, 2000, Cuenca et al. 2003, Garcı´a-Moreno et al. 2004, Becerra 2005, Hasbun et al. 2005, Mateos 2005, Wuster et al. 2005, Leo´ n-Paniagua et al. 2007) may help refine previously identified cenocrons. These analyses can be used to identify cenocrons and determine how and when they have dispersed and integrated.

BIOGEOGRAPHIC BOUNDARIES ON

SYMPOSIUM ISSUE

IBS SPECIAL

Brooks, D. R. and McLennan, D. A. 2001. A comparison of a discovery-based and an event-based method of historical biogeography. J. Biogeogr. 28: 757 767. Cabrera, A. L. and Willink, A. 1973. Biogeografıá de Ame´rica Latina. Monografıá 13, Serie de Biologıá, OEA, Washington DC. Casas-Andreu, G. and Reyna-Trujillo, T. 1990. Herpetofauna (anfibios y reptiles). Mapa IV.8.6. In: Atlas Nacional de Me´xico, Vol. III. Inst. de Geografıá, UNAM, Mexico D.F. Contreras-Medina, R. et al. 2007a. Application of parsimony analysis of endemicity to Mexican gymnosperm distributions: grid-cells, biogeographical provinces and track analysis. Biol. J. Linn. Soc. 92: 405 417. Contreras-Medina, R. et al. 2007b. Gymnosperms and cladistic biogeography of the Mexican Transition Zone. Taxon 56: 905 915. Cuenca, A. A. et al. 2003. Long-distance colonization, isolation by distance, and historical demography in a relictual Mexican pinyon pine (Pinus nelsonii Shaw) as revealed by paternally inherited genetic markers (cpSSRs). Mol. Ecol. 12: 2087 2097. Darlington, P. J. Jr 1957. Zoogeography: the geographical distribution of animals. Wiley. Escalante, T. et al. 2003. Using parsimony analysis of endemicity to analyze the distribution of Mexican land mammals. Southwest. Nat. 48: 563 578. Escalante, T. et al. 2004. The diversification of the Nearctic mammals in the Mexican Transition Zone. Biol. J. Linn. Soc. 83: 327 339. Escalante, T. et al. 2005. Las provincias biogeogra´ficas del componente Mexicano de Montan˜ a desde la perspectiva de los mamı´feros continentales. Rev. Mex. Biodivers. 76: 199 205. Escalante, T. et al. 2007. Cladistic biogeographic analysis suggests an early Caribbean diversification in Mexico. Naturwissenschaften 94: 561 565. Espinosa, D. et al. 2006. Historical biogeographic patterns of the species of Bursera (Burseraceae) and their taxonomical implications. J. Biogeogr. 33: 1945 1958. Espinosa-Organista, D. et al. 2004. Identidad biogeogra´fica de la Sierra Madre Oriental y posibles subdivisiones bio´ ticas. In: Luna, I. et al. (eds), Biodiversidad de la Sierra Madre Oriental, Las Prensas de Ciencias. UNAM, Mexico, D.F., pp. 487 500. Espinosa-Organista, D. et al. 2008. El conocimiento biogeogra´fico de las especies y su regionalizacio´ n natural. In: Sarukhań, J. (ed.), Capital natural de Me´xico. Vol. I. Conocimiento actual de la biodiversidad. Conabio, Mexico, D.F., pp. 33 65. Ferrusquıá-Villafranca, I. 1990. Regionalizacio´ n biogeogra´fica. Mapa IV.8.10. In: Atlas Nacional de Me´xico, Vol. III. Inst. de Geografıá, UNAM, Mexico, D.F. Ferrusquıá-Villafranca, I. 1993. Geology of Mexico: a synopsis. In: Ramamoorthy, T. P. et al. (eds), Biological diversity of Mexico: origins and distribution. Oxford Univ. Press, pp. 3 107. Ferrusquıá-Villafranca, I. and Gonza´lez-Guzmań, L. I. 2005. Northern Mexico’s landscape, part II: the biotic setting across time. In: Cartron, J. L. et al. (eds), Biodiversity, ecosystems and conservation in northern Mexico. Oxford Univ. Press, pp. 39 41. Flores Villela, O. and Goyenechea, I. 2001. A comparison of hypotheses of historical biogeography for Mexico and Central America, or in search for the lost pattern. In: Johnson, J. D. et al. (eds), Mesoamerican herpetology: systematics, zoogeography, and conservation. The Univ. of Texas at El Paso, pp. 171 181. Garcıá-Marmolejo, G. et al. 2008. Establecimiento de prioridades para la conservacio´ n de mamı´feros terrestres neotropicales de Me´xico. Mastozool. Neotrop. 15: 41 65.

360

Garcıá-Moreno, J. et al. 2004. Genetic variation coincides with geographic structure in the common bush-tanager (Chlorospingus ophthalmicus) complex from Mexico. Mol. Phylogenet. Evol. 33: 186 196. Goldman, E. A. and Moore, R. T. 1945. The biotic provinces of Mexico. J. Mammal. 26: 347 360. Halffter, G. 1962. Explicacio´ n preliminar de la distribucio´ n geogra´fica de los Scarabaeidae mexicanos. Acta Zool. Mex. 5: 1 17. Halffter, G. 1964. La entomofauna americana, ideas acerca de su origen y distribucio´ n. Folia Entomol. Mex. 6: 1 108. Halffter, G. 1972. Ele´ments anciens de l’entomofaune neotropicale: ses implications biogeógraphiques. In: Biogeographie et liasons intercontinentales au cours du Me´sozoique. 17me Congr. Int. Zool., Monte Carlo, pp. 1 40. Halffter, G. 1974. Ele´ments anciens de l’entomofaune neotropicale: ses implications biogeógraphiques. Quaest. Entomol. 10: 223 262. Halffter, G. 1976. Distribucio´ n de los insectos en la Zona de Transicio´ n Mexicana: relaciones con la entomofauna de Norteame´rica. Folia Entomol. Mex. 35: 1 64. Halffter, G. 1978. Un nuevo patro´ n de dispersio´ n en la Zona de Transicio´ n Mexicana: el mesoamericano de montan˜ a. Folia Entomol. Mex. 39 40: 219 222. Halffter, G. 1987. Biogeography of the montane entomofauna of Mexico and Central America. Annu. Rev. Entomol. 32: 95 114. Hasbun, C. R. et al. 2005. Mitochondrial DNA phylogeography of the Mesoamerican spiny-tailed lizards (Ctenosaura quinquecarinata complex): historical biogeography, species status and conservation. Mol. Ecol. 14: 3095 3107. Iturralde-Vinent, M. 1998. Synopsis of the geological constitution of Cuba. Acta Geol. Hisp. 33: 9 56. Kerr, A. C. et al. 1999. New plate tectonic model of the Caribbean: implications from a geochemical reconnaissance of Cuban Mesozoic volcanic rocks. Geol. Soc. Am. Bull. 111: 1581 1599. Leo´ n-Paniagua, L. et al. 2007. Diversification of arboreal mice of genus Habromys (Rodentia: Cricetidae: Neotominae). Mol. Phylogenet. Evol. 62: 653 664. Liebherr, J. K. 1991. A general area cladogram for montane Mexico based on distributions in the Platynine genera Elliptoleus and Calathus (Coleoptera: Carabidae). Proc. Entomol. Soc. Wash. 93: 390 406. Liebherr, J. K. 1994. Biogeographic patterns of montane Mexican and Central American Carabidae (Coleoptera). Can. Entomol. 126: 841 860. Luna-Vega, I. et al. 1999. Historical relationships of the Mexican cloud forests: a preliminary vicariance model applying Parsimony Analysis of Endemicity to vascular plant taxa. J. Biogeogr. 26: 1299 1305. Luna-Vega, I. et al. 2000. Track analysis and conservation priorities in the cloud forests of Hidalgo, Mexico. Divers. Distrib. 6: 137 143. Luna-Vega, I. et al. 2001. Biogeographical affinities among Neotropical cloud forests. Plant Syst. Evol. 228: 229 239. Marin˜ o-Pe´rez, R. et al. 2007. Ana´lisis panbiogeogra´fico de las especies mexicanas de Pselliopus Bergroth (Hemiptera: Heteroptera: Reduviidae: Harpactorinae). Acta Zool. Mex. 23: 77 88. Ma´rquez, J. and Morrone, J. J. 2003. Ana´lisis panbiogeogra´fico de las especies de Heterolinus y Homalolinus (Coleoptera: Staphylinidae: Xantholinini). Acta Zool. Mex. 90: 15 25. Marshall, C. J. and Liebherr, J. K. 2000. Cladistic biogeography of the Mexican Transition Zone. J. Biogeogr. 27: 203 216. Mateos, M. 2005. Comparative phylogeography of livebearing fishes in the genera Poeciliopsis and Poecilia (Poeciliidae:

BIOGEOGRAPHIC BOUNDARIES

Ramı´rez-Pulido, J. and Castro-Campillo, A. 1990. Regionalizacio´ n mastofaunı´stica (mamı´feros). Mapa IV.8.8.A. In: Atlas Nacional de Me´xico, Vol. III. Inst. de Geografıá, UNAM, Mexico D.F. Ruggiero, A. and Ezcurra, C. 2003. Regiones y transiciones biogeogra´ficas: complementariedad de los ana´lisis en biogeografıá histo´ rica y ecolo´ gica. In: Morrone, J. J. and Llorente, J. (eds), Una perspectiva latinoamericana de la biogeografıá, Las Prensas de Ciencias. UNAM, Mexico, D.F., pp. 141 154. Rzedowski, J. 1978. Vegetacio´ n de Me´xico. Limusa, Mexico D.F. Rzedowski, J. and Reyna-Trujillo, T. 1990. To´ picos biogeogra´ficos. Mapa IV.8.3. In: Atlas Nacional de Me´xico, Vol. III. Inst. de Geografıá, UNAM, Mexico D.F. Smith, H. 1941. Las provincias bio´ ticas de Me´xico, segu´ n la distribucio´ n geogra´fica de las lagartijas del geńero Sceloporus. An. Esc. Nac. Cienc. Biol. 2: 103 110. Sullivan, J. et al. 1997. Phylogeography and molecular systematics of the Peromyscus aztecus species group (Rodentia: Muridae) inferred using parsimony and likelihood. Syst. Biol. 46: 426 440. Sullivan, J. et al. 2000. Comparative phylogeography of Mesoamerican highland rodents: concerted versus independent response to past climate fluctuations. Am. Nat. 155: 755 768. Toledo, V. H. et al. 2007. Track analysis of the Mexican species of Cerambycidae (Insecta, Coleoptera). Revta. Bras. Entomol. 51: 131 137. Torres Miranda, A. and Luna Vega, I. 2006. Ana´lisis de trazos para establecer a´reas de conservacio´ n en la Faja Volcańica Transmexicana. Interciencia 31: 849 855. Whittaker, R. J. and Fernańdez-Palacios, J. M. 2007. Island biogeography: ecology, evolution and conservation. Oxford Univ. Press. Wuster, W. et al. 2005. Tracing an invasion: land bridges, refugia, and the phylogeography of the Neotropical rattlesnake (Serpentes: Viperidae: Crotalus durissus). Mol. Ecol. 14: 1095 1108.

SYMPOSIUM

Cyprinodontiformes) in central Mexico. J. Biogeogr. 32: 775 780. Mitchell, S. D. 2002. Integrative pluralism. Biol. Philos. 17: 55 70. Morrone, J. J. 2001. Toward a cladistic model of the Caribbean: delimitation of areas of endemism. Caldasia 23: 43 76. Morrone, J. J. 2005. Hacia una sı´ntesis biogeogra´fica de Me´xico. Rev. Mex. Biodivers. 76: 207 252. Morrone, J. J. 2006. Biogeographic areas and transition zones of Latin America and the Caribbean Islands, based on panbiogeographic and cladistic analyses of the entomofauna. Annu. Rev. Entomol. 51: 467 494. Morrone, J. J. 2009. Evolutionary biogeography: an integrative approach with case studies. Columbia Univ. Press. Morrone, J. J. and Ma´rquez, J. 2001. Halffter’s Mexican Transition Zone, beetle generalized tracks, and geographical homology. J. Biogeogr. 28: 635 650. Morrone, J. J. and Escalante, T. 2002. Parsimony Analysis of Endemicity (PAE) of Mexican terrestrial mammals at different area units: when size matters. J. Biogeogr. 29: 1095 1104. Morrone, J. J. and Gutie´rrez, A. 2005. Do fleas (Insecta: Siphonaptera) parallel their mammal host diversification in the Mexican Transition Zone? J. Biogeogr. 32: 1315 1325. Morrone, J. J. et al. 1999. Preliminary classification of the Mexican biogeographic provinces: a parsimony analysis of endemicity based on plant, insect, and bird taxa. Southwest. Nat. 44: 507 514. Mu¨ ller, R. D. et al. 1999. New constraints on the Late Cretaceous/ Tertiary plate tectonic evolution of the Caribbean. In: Mann, P. (ed.), Caribbean Basin. Sedimentary basins of the world, 4. Elsevier, pp. 39 55. Nelson, G. and Ladiges, P. Y. 1996. Paralogy in cladistic biogeography and analysis of paralogy-free subtrees. Am. Mus. Novit. 3167: 1 58. Ochoa, L. et al. 2003. Contribucio´ n al atlas panbiogeogra´fico de Me´xico: los ge´neros Adelpha y Hamadryas (Nymphalidae), y Dismorphia, Enantia, Lienix y Pseudopieris (Pieridae) (Papilionoidea; Lepidoptera). Folia Entomol. Mex. 42: 65 77. Ortega, J. and Arita, H. T. 1998. Neotropical-Nearctic limits in Middle America as determined by distributions of bats. J. Mammal. 79: 772 781.

IBS SPECIAL ISSUE

361

Ecography 33: 362 368, 2010 doi: 10.1111/j.1600-0587.2010.06244.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Kenneth H. Kozak. Accepted 22 February 2010

Cross-species and assemblage-based approaches to Bergmann’s rule and the biogeography of body size in Plethodon salamanders of eastern North America ´ . Olalla-Ta´rraga, Luis M. Bini, Jose´ A. F. Diniz-Filho and Miguel A ´ . Rodrı´guez Miguel A

IBS SPECIAL

ISSUE

M. A´. Olalla-Ta´rraga (m.olalla@imperial.ac.uk), Div. of Biology, Imperial College London, Silwood Park campus, Ascot, Berkshire SL5 7PY, UK. L. M. Bini and J. A. F. Diniz-Filho, Depto de Ecologı`a, ICB, Univ. Federal de Goia´s, CP 131, 74.001-970, Goiaˆnia, GO, Brazil. M. A´. Rodrı´guez, Dept of Ecology, Univ. of Alcala´, Alcala´ de Henares, ES-28871, Spain.

Over the past few years, there has been a resurgence of interest in the investigation of spatial patterns in body size over large-scales to explore Bergmann’s rule at different phylogenetic scales across a range of taxa (Blackburn et al. 1999, Gaston et al. 2008). In its original formulation Bergmann’s rule predicts that among closely related endotherms those living in colder regions tend to be larger than those in warmer environments as a result of their reduced surface-to-volume ratios and, hence, better heat conservation (Bergmann 1847). Remarkably, the debate around the validity of this ecogeographical ‘‘rule’’ has long been fostered by the conflation of pattern (increasing body size towards colder regions) and mechanism (heat conservation) (Blackburn et al. 1999). However, there is increasing recognition that the geographical distribution of body size is markedly idiosyncratic and Bergmann’s physiological mechanism cannot explain the observed size clines everywhere (Rodrı´guez et al. 2008, Diniz-Filho et al. 2009). Accordingly, a differentiation has to be made between pattern and process, so that the latter is not explicitly inherent to the former. In doing such distinction we do not only spur the scientific debate around the validity of the ‘‘rule’’, but the search of alternatives to Bergmann’s original mechanism that are able to account for the observed size clines. Part of the controversy around Bergmann’s rule stems from the seminal papers of Ray (1960) and Lindsey (1966) suggesting that some ectothermic organisms display intra and interspecific body size variation as a response to environmental gradients, which apparently requires alternative explanations to the ones offered for endotherms (Cushman et al. 1993, Ashton and Feldman 2003, OlallaTa´rraga and Rodrı´guez 2007). Since then, workers have tried to identify ecological or evolutionary mechanisms accounting for body size clines in ectotherms, but we are far from a consensus on a unifying mechanism.

362

A critical step before searching for underlying mechanisms is indeed examining what the patterns look like in nature. Because Bergmann’s rule was originally formulated for endothermic vertebrates, an abundant number of studies have reported the existence of body size gradients in mammals and birds (Ashton et al. 2000, Ashton 2002a, Rodrı´guez et al. 2008, Diniz-Filho et al. 2009, Olson et al. 2009), whereas the geographical variation of body size for many ectothermic organisms remains mostly unknown (but see below). Several authors have documented the patterns and explored the causes of body size variation in ectotherms at different levels of biological organization (Ashton 2002b, Belk and Houston 2002, Ashton and Feldman 2003, Blanckenhorn and Demont 2004, OlallaTa´rraga et al. 2006, Olalla-Ta´rraga and Rodrı´guez 2007, Adams and Church 2008). Notwithstanding, the variety of methods and terminologies involved not only prevent a comparison of results from different studies but also indicate the lack of agreement on a single standard approach to investigate Bergmann’s rule. Gaston et al. (2008) recently attempted to clear up part of the confusion by identifying three kinds of approaches to studying spatial patterns in biological traits in general and Bergmann’s rule in particular: intraspecific, interspecific and assemblage-based. They stressed that the distinction between intra- and interspecific approaches seems to be clear, but there is some confusion on the methodological differences between interspecific (hereafter ‘‘cross-species’’) and assemblagebased studies. Beyond semantic considerations (e.g. the assemblage-based approach has also been termed as the ‘‘community’’, ‘‘interspecific’’ or ‘‘grid-based’’ approach) (Blackburn and Hawkins 2004, Olalla-Ta´rraga et al. 2006), the methodological disparities between cross-species and assemblage-based approaches may not be trivial in terms of interpreting patterns and processes (Ruggiero and Hawkins 2006). In our view, this requires particular attention.

Here, we first highlight the conceptual basis of crossspecies and assemblage-based approaches to outline the advantages and limitations of each of them. We then illustrate the differences by using distributional and body size data for Plethodon salamander species in eastern North America as a case study. Recently, Adams and Church (2008) used a cross-species analysis to evaluate Bergmann’s rule in this group. From their results, complemented with a meta-analysis of the intraspecific relationship between body size and temperature for 59 amphibians (including 38 Plethodon species), they concluded that amphibians do not follow Bergmann’s rule and discarded the classical heat conservation hypothesis as a valid explanation in this taxon. We conduct complementary analyses to show how the combination of an assemblage-based approach with phylogenetic eigenvector regression (PVR) (Diniz-Filho et al. 2007) can be useful to generate further insights into potential evolutionary and ecological mechanisms. In the light of our findings, we argue that it is premature to dismiss the existence of geographic body size gradients in amphibians, much less the search for causal factors.

Cross-species vs assemblage-based approaches

We followed the methods described in Diniz-Filho et al. (2007). Initially, we created a 110 110 km equal area grid of 319 cells and used geographic range maps (IUCN, Conservation International and NatureServe 2006) for each of the 44 Plethodon salamanders described in eastern North America together with body size data for each species (kindly provided by Adams and Church 2008) to calculate 363

ISSUE

Phylogenetic components of body size and the assemblage approach in Plethodon

IBS SPECIAL

Cross-species approaches treat each species as an independent datum and use bivariate scatter-plots to examine covariation of body size and latitude (or temperature) across species. This method has also been termed as the ‘‘midpoint approach’’: it involves obtaining a single spatial measure for each species (usually the latitudinal midpoint of its geographic range) and then plotting these mid points against the species’ body sizes (see Blackburn and Hawkins 2004, and references therein). In contrast, assemblage-based studies explore geographical patterns within grids covering the study region and combine the species’ presences/ absences in the cells with their body sizes to obtain cellmean body size values (usually log-transformed geometric means) (Ruggiero and Hawkins 2006 and Gaston et al. 2008). Blackburn and Hawkins (2004) termed this the ‘‘community approach’’, as such investigations examine the spatial distribution of summary statistics for body size across faunal assemblages (grid cells) of a particular biogeographic region. Therefore, the units of analysis in cross-species approaches are single species, whereas in assemblage-based methods are measures of average body size of all the species occurring within grid-cells. The pros and cons of cross-species and assemblage-based approaches have been discussed elsewhere (Blackburn and Hawkins 2004, Ruggiero and Hawkins 2006, Meiri and Thomas 2007). Notably, both methods need to circumvent potential problems associated with different sources of pseudoreplication. In assemblage-based studies, the varying proximity among units of analysis (i.e. grid cells) makes them have different levels of spatial autocorrelation. This causes degrees of freedom estimated in the usual way to be inflated, which will lead to increased type I error if ignored. Spatial statistical techniques allow tackling this (Diniz-Filho et al. 2003). In contrast, in cross-species analyses, inflated type I errors associated to pseudoreplication may arise from the phylogenetic non-independence of the data, which

requires the use of phylogenetic comparative methods. Assemblage-based studies may also suffer from phylogenetic autocorrelation effects (Olalla-Ta´rraga and Rodrı´guez 2007), but it is only recently that workers have found methods to jointly deal with spatially and phylogenetically structured data (Diniz-Filho et al. 2007, Ku¨ hn et al. 2009, see below). According to several authors (Blackburn and Hawkins 2004, Ruggiero and Hawkins 2006), the main advantage of assemblage approaches is that they allow a direct evaluation of the environmental structure underlying broad-scale geographical patterns, a feature that is severely limited in the case of cross-species analysis as environmental gradients are reduced to a single point in geographical or environmental space. That is, cross-species methods ignore the geographical structure that exists in the data by reducing the multidimensional nature of geographic ranges to single values. This can have serious implications for the interpretation of ecological and evolutionary patterns and processes such as those associated to Bergmann’s rule, even leading to incorrect or equivocal conclusions (Blackburn and Hawkins 2004, Ruggiero and Hawkins 2006). Because ecological and evolutionary processes usually take place in a geographical context, spatially explicit approaches are necessary to gain a multidimensional perception of species’ trait gradients (such as body size gradients) and to make explicit their links to environmental variation (Ruggiero and Hawkins 2006). While Meiri and Thomas (2007) also stressed the abovementioned limitations of cross-species studies, they paid special attention to those of assemblage-based approaches. Specifically, they pointed towards the potential sensitivity of the analyses to the uneven variation of species richness across cells and the use of mean log masses (or lengths) within each cell as a measure of body size. They emphasized the need to control for the effects of richness (because species are not added to cells at random) and claimed that measures of central tendency other than the mean are preferred (because body size distributions are often rightskewed at broad spatial scales) (Brown 1995). As a solution, they have suggested including species richness as a covariate in multiple regression models and using median body size rather than the mean (Olson et al. 2009). However, far from being a pervasive problem with the assemblage-approach, it should be noted that these concerns only make sense in the absence of log-normal size distributions and the existence of species-poor cells. The solution to both potential problems is straightforward since we can evaluate to what extent the use of means or the spatial distribution of species richness affects the results.

ISSUE

IBS SPECIAL

average log10 body length (hereafter mean body size) in each grid cell (see Olalla-Ta´rraga et al. 2006, Olalla-Ta´rraga and Rodrı´guez 2007 for details). A single body size estimate was used for each species assuming that intraspecific spatial variation in body size is small relative to the interspecific variation (Gaston et al. 2008, Olalla-Ta´rraga et al. 2009). Almost 80% of the Plethodon salamanders in eastern North America show no relationship between temperature and body size at the intraspecific level, whereas 15% display a significant positive correlation (Adams and Church 2008). Because only three species display significant negative relationships between temperature and intraspecific size variation, contrary to the observed assemblage (interspecific) level pattern (see below), there are strong arguments to believe that our analyses are not only unaffected, but could be reinforced by the inclusion of intraspecific geographic size trends. Analyses based on log10 medians generated similar estimates to those obtained for log10 means (Supplementary material Table S1). Because species of Plethodon have been extensively used in behavioral, ecological and evolutionary studies, their ranges (extents of occurrence) are well known (see Highton 1995 and Lannoo 2005 and references therein). We then used phylogenetic eigenvector regression (PVR) (Diniz-Filho et al. 1998; see also Desdevises et al. 2003, Ku¨ hn et al. 2009). PVR estimates the phylogenetic signal in body size data by regressing this trait against a set of orthogonal eigenvectors extracted from a pairwise phylogenetic distance matrix that describes phylogenetic relatedness among species. Hence, we can partition the total body size of each species (T) into a phylogenetic component (P), which represents its predicted body size value according to the phylogeny, and a specific or ‘‘ecological’’ component (S), which corresponds to the model residuals and can be interpreted as the independent response of the species (see also Diniz-Filho et al. 2009 for a detailed explanation). Rohlf (2001) criticized the PVR method because all eigenvectors would be necessary to represent the entire phylogeny and, thus, estimates of P and S could be biased by eigenvector selection. Nonetheless, simulation studies have revealed correct type I error rates when estimating phylogenetic correlations using PVR (Martins et al. 2002, Diniz-Filho and Toˆ rres 2002). Moreover, using a few eigenvectors is enough to take phylogenetic autocorrelation into account and ensure species’ independence in respect to the analysed trait for further analyses. To check this assumption, we used Moran’s I autocorrelation coefficients in the S-component (see also Diniz-Filho and Toˆ rres 2002). We based our PVR analysis on a species-level phylogenetic tree for Plethodon (Wiens et al. 2006) and selected the first five eigenvectors for the analysis, by successively adding eigenvectors to remove residual autocorrelation (Diniz-Filho and Toˆ rres 2002). These eigenvectors described 95.3% of the variation in phylogenetic structure and were subsequently used as predictors of body size. A multiple regression model of body size against the selected eigenvectors explained 63.2% of the variance in size, with significant contribution of all eigenvectors, thus indicating the existence of phylogenetic signal in the data. Also, it is worthwhile noting that there is no autocorrelation 364

in model residuals (Moran’s I 0.07; p 0.486), so that the set of eigenvectors we used were appropriate to describe phylogenetic patterns in body size and ensure species’ independence in respect to trait variation. Using the predicted and residual values of this regression (i.e. the P and S component values obtained for each species), we generated mean-P and mean-S values for each cell (Fig. 1a and b respectively). Finally, we built regression models to evaluate the relationships between the spatial patterns observed for mean-P and mean-S components and environmental variation. We processed three environmental variables in ArcGIS 9.2 at the resolution of the equal area grid cells to incorporate them as predictors into the analyses: mean annual temperature, annual precipitation, and the Global Vegetation Index. Following a standard procedure in macroecology, the original data in raster format were upscaled by averaging all 0.5 degrees pixels to the resolution of our grid cells (110 110 km) (see Olalla-Ta´rraga et al. 2006 and Olalla-Ta´rraga and Rodrı´guez 2007, for further details on data description and sources). These variables were selected on the basis that they are related to three hypotheses proposed for explaining body size gradients in amphibians, heat balance, water availability and primary productivity, respectively (Olalla-Ta´rraga and Rodrı´guez 2007). Because most Plethodon species occur in forested areas of the Appalachian and Ouachita mountains (Highton 1995), we also added to our models range in elevation within each cell as a measure of mesoscale climatic variation (Olalla-Ta´rraga et al. 2006, see also Rodrı´guez et al. 2008). Statistical analyses were performed with SAM (Spatial Analysis in Macroecology; Rangel et al. 2006), STATISTICA (StatSoft 2003) and PDAP (Phenotypic Diversity Analysis Programs; Garland et al. 1993). In contrast to Adams and Church’s (2008) cross-species analysis, our assemblage approach identified a strong gradient of body size variation in Plethodon assemblages in geographical space, with decreasing size northwards. Thus, Plethodon salamanders in eastern North America follow the converse to Bergmann’s rule. This pattern is partly influenced by the occurrence of Plethodon cinereus, a small-bodied species, towards the northern distributional limits of this clade. A large proportion of the current distribution range of P. cinereus lies in northern latitudes that were covered by the Laurentide ice sheet in the Pleistocene and could only be colonized after the glacials retreated. Hence, a third of the data consist of monospecific assemblages of P. cinereus which may lead to the suggestion that the extreme northerly distribution of P. cinereus is an artifact and the species should be removed from the analysis. Even though we favor the view that there are no obvious biological reasons for the exclusion of P. cinereus, we conducted separate analyses excluding the species (Supplementary material Table S2). By doing so, despite an extreme reduction in statistical power, the model remained significant (although GVI and precipitation become more important than temperature perhaps because the range of variation for this variable was also extremely reduced, with minimum average temperature increasing from 28C to 48C). We found that our three-variable models including mean annual temperature, annual precipitation and the

Figure 1. Geographical patterns of the average phylogenetic component of body size (a) and the specific component (b) resulting from PVR for all the 44 Plethodon salamander species in eastern North America. Numbers included in the legend of each map are snout-to-vent length values in millimeters. Overall, there are marked gradients in latitudinal space, but local geographic mismatches between phylogenetic and specific components.

the regression models (Olalla-TaÂ´rraga et al. 2009), an approach analogous to PVR in a spatial context. Also, we built spatial correlograms based on Moranâ&#x20AC;&#x2122;s I coefficients (not shown) to evaluate if this approach removed the spatial autocorrelation of the residuals of our multiple-regression models at all distance classes, which would indicate that the fitted models adequately describe the spatial variation in body size across all spatial scales (Diniz-Filho et al. 2003). The inclusion of spatial filters to remove spatial autocorrelation in the residuals of our regression models did not change the relative importance of the environmental predictors as indicated by the standardized partial regression coefficients (Table 1). Range in elevation contributed little to the increase of the explanatory power (adjusted R2) of the models (e.g. between 0.7 and 1.5%), hence we did not included this variable in our models. Although our results support Adams and Churchâ&#x20AC;&#x2122;s (2008) contention that this group does not follow Bergmannâ&#x20AC;&#x2122;s rule, it contradicts their conclusion that there

Model

Env Env Spa Env Rich Env Env

Weighting variable

None None None 1 Log10(1/SR2) Species richness

Standardized coefficients (P component)

Standardized coefficients (S component)

Temp

Precip

GVI

Temp

Precip

GVI

0.868* 0.851* 0.756* 0.777* 0.684*

0.250* 0.064 0.230 0.380* 0.430*

0.03 0.150* 0.031 0.047 0.117*

0.539 0.715 0.478 0.458 0.364

0.823* 0.717* 0.825* 0.776* 0.746*

0.03* 0.119 0.014 0.11* 0.16*

0.064* 0.01 0.077 0.108* 0.116*

0.706 0.746 0.743 0.610 0.523

365

ISSUE

Table 1. Multiple regression models of environmental variables against average cell values of phylogenetic (P) and specific (S) components obtained from PVR analysis for Plethodon species in eastern North America. For each model we show their corresponding coefficients of determination (R2) and standardized partial regression coefficients of the predictors. Additionally, we provide results from weighted-LS regression analyses using species richness in each cell (S) and (1 Log10[1/S2]) as weighting factors. Env Spa models include non-redundant spatial filters as predictors, whereas Env Rich models include species richness. Predictor variables are: Temp mean annual temperature; Precip annual precipitation; GVI global vegetation index. An asterisk means that the standardized regression coefficients were significant at a probability level pB0.001.

IBS SPECIAL

Global Vegetation Index explained 53.9 and 70.6% of the variation in mean-P and mean-S, respectively (Table 1). Mean annual temperature was the primary predictor of the spatial pattern of mean-P and mean-S components, being positively correlated with both. To adjust mean body sizes in grids by the number of species (Meiri and Thomas 2007), we repeated these analyses using weightedLeast Square regression. Specifically, we used species richness values (S) (Supplementary material Fig. S1) and (1 Log10[1/S2]) as weighting factors to recalculate standardized regression coefficients (Olalla-TaÂ´rraga and RodrÄąÂ´guez 2007). Following Olson et al. (2009) we also built models using species richness as a covariate. Our results remained robust in all cases and mean annual temperature consistently had the highest standardized coefficient in the models for mean-P and mean-S (Table 1). Aditionally, to address the statistical problems associated with spatial autocorrelation in the model residuals, we included eigenvector-based spatial filters as predictors in

is no spatial pattern in body size at any level of biological organization. Moreover, our results reveal a strong positive association between the geographical patterns in the phylogenetic (P) and ecological (S) components of body size for species assemblages and mean annual temperature. This also runs counter to Adams and Church’s (2008) analysis that was unable to detect a signal of mean annual temperature in body size variation, but it is similar to an assemblage-based study for the complete urodele fauna of Europe and eastern North America (Olalla-Ta´rraga and Rodrı´guez 2007).

IBS SPECIAL

ISSUE

The heat balance hypothesis Olalla-Ta´rraga and Rodrı´guez (2007) proposed the heat balance hypothesis to account for body size gradients exhibited by both endotherms and ectotherms at an assemblage level. For ectothermic groups with thermoregulating abilities, whose body size is not large enough to limit heat gain in low energy environments (e.g. snakes), this hypothesis parallels the traditional heat conservation mechanism originally conceived by Bergmann for endotherms. In contrast, for thermoconformers, which have limited abilities to control heat exchange and depend more closely on the thermal conditions of the environment, the hypothesis predicts a reversed Bergmann pattern as animals would minimize body size in colder regions to reduce heating times and maximize the time available for foraging and reproduction (Olalla-Ta´rraga and Rodrı´guez 2007). Hence, the heat balance hypothesis stems from older ideas presented by Cowles and Bogert (1944) and espoused by Ashton and Feldman (2003) that took into account the importance of surface-to-volume ratios for the thermal adjustments and activity times in reptiles (Olalla-Ta´rraga et al. 2006), but complementary develops expectations for ectotherms with a high degree of thermoconformism (Olalla-Ta´rraga and Rodrı´guez 2007). By doing the distinction between behavioral thermoregulators and thermoconformers, the heat balance hypothesis does not only consider the role of body size for the thermal ecology of ectotherms, but indirectly incorporates the importance of factors that constrain thermoregulatory behaviors in these organisms. For instance, the common absence of behavioral thermoregulation in salamanders is likely due to hydric limitations on heat exchange with the environment resulting from high rates of cutaneous evaporative water loss (Feder 1982, Feder and Lynch 1982, see also OlallaTa´rraga et al. 2009). Consistent with the predictions of the heat balance hypothesis, Olalla-Ta´rraga and Rodrı´guez (2007) found that anurans (behavioral thermoregulators) follow Bergmann’s rule in Europe and North America, whereas urodeles (mostly thermoconformers) do the opposite. In the latter case, the clear converse Bergmann’s gradients may be also determined by the fact that the elongate body plans of urodeles generate a larger surface to volume ratio, which also contributes to increased heating rates in low energy environments. Our results for Plethodon salamanders are also consistent with the heat balance hypothesis. The pattern of the specific component (S) is strongly related to thermal availability in 366

the environment, which suggests independent adaptive responses of species to current climatic conditions (DinizFilho et al. 2007). That is, species smaller than expected by their phylogenetic relatedness are more frequent in cold environments, whereas species larger than expected are mostly distributed in warmer areas. However, the individual responses of species to temperature appear not to be the only cause for the overall gradient in body size, as we also found a climatic signal in the phylogenetic component (P). This is likely to be related to the particularities of recent diversification in this woodland-adapted salamander group since the early Pliocene. Several authors have suggested that most extant species of Plethodon salamanders in eastern North America originated as a result of rapid speciation events during the last five Mya following a history of shifts in forest cover along the Appalachian mountains associated to climatic changes (Highton 1995, Kozak et al. 2006, Wiens et al. 2006). According to this, continuous altitudinal shifts in forest cover through the Pliocene and Pleistocene may have favoured allopatric diversification events that generated many morphologically-similar cryptic species that, at the same time, tended to conserve their ancestral climatic niches (Highton 1995, Kozak and Wiens 2006, Wiens et al. 2006). Because rapid evolutionary radiation usually involves marked changes in the diversity of morphological and ecological characters, Plethodon appears to be a particular case of rapid speciation (Kozak et al. 2006). Under this hypothesis, the observed relationship between the spatial pattern in the phylogenetic component of body size and mean annual temperature may simply reflect the interplay between rapid diversification and a structured trend across the phylogeny to maintain ancestral temperature-body size relationships. Even if true, this pattern in the phylogenetic structure of the data is not strong enough to obscure the signal of unique responses of each species to current climate variation found in the specific component. Bearing in mind the above, why do we find a pattern that Adams and Church (2008) did not detect? Statistically, Adams and Church’s (2008) cross-species analyses are properly done. However, we favor the view that cross-species analyses provide different information and are not as useful as the assemblage approach under almost all possible scenarios. We would expect both approaches to converge into similar results only when distribution ranges for the whole set of analysed species within the clade are either extremely restricted or mostly unknown. On the contrary, when species ranges are well known and comprise enough environmental variation, we suggest examining the occurrence of interspecific body size gradients using spatially-explicit assemblage approaches (i.e. using complete geographic information, see Blackburn and Hawkins 2004, Ruggiero and Hawkins 2006 and above). Of course, this is not to say that a cross-species approach is not helpful in detecting correlated evolution between traits across species. For instance, cross-species analyses conducted in a phylogenetic framework have proven to be useful to detect lineagelevel patterns in the relationships between a number of life history attributes and body size (see e.g. Freckleton et al. 2002 and references therein). The problem arises when we use as a trait a variable that does not actually represent the

whole geographic range of a species (either the latitudinal midpoint or an overall measure of temperature as Adams and Church did). In doing so, we are restricting a multidimensional trait (i.e. a species distribution) to a single point. Therefore, a more in depth approach to study Bergmann’s rule requires employing alternative protocols that simultaneously consider the distribution of species into a geographic and phylogenetic framework, such as the assemblage approach we have described here.

Concluding remarks We have shown how an assemblage approach complemented with phylogenetic eigenvector regression can be useful to explore both the ecological and evolutionary mechanisms associated with Bergmann’s rule in a spatially explicit context. Contrary to a cross-species approach, our assemblage-based analysis detected strong geographical trends in interspecific variation of body size of Plethodon salamanders in eastern North America. These arguments lead us to call for caution in making generalizations on the validity of Bergmann’s rule or its reverse in particular taxa, unless we take into account the limitations of our database, the clade we are using as model system and the statistical techniques we employ to analyze the data. This is especially true if we are also interested in examining potential explanations. Although our analyses favor the heat balance hypothesis as a mechanism, considering the idiosyncratic patterns detected for different taxa, the disparities between phylogenetic and non-phylogenetic studies and the apparent discrepancies at different taxonomic scales, a general explanation for Bergmann’s rule still remains elusive.

References

367

ISSUE

Adams, D. C. and Church, J. O. 2008. Amphibians do not follow Bergmann’s rule. Evolution 62: 413 420. Ashton, K. G. 2002a. Do amphibians follow Bergmann’s rule? Can. J. Zool. 80: 708 716. Ashton, K. G. 2002b. Patterns of within-species body size variation of birds: strong evidence for Bergmann’s rule. Global Ecol. Biogeogr. 11: 505 523. Ashton, K. G. and Feldman, C. R. 2003. Bergmann’s rule in nonavian reptiles: turtles follow it, lizards and snakes reverse it. Evolution 57: 1151 1163. Ashton, K. G. et al. 2000. Is Bergmann’s rule valid for mammals? Am. Nat. 156: 390 415.

IBS SPECIAL

Acknowledgements We are grateful to D. C. Adams for kindly providing us his body size database. Comments by D. C. Adams, B. A. Hawkins, J. Hortal, K. H. Kozak, S. Meiri, A. Mooers and three anonymous reviewers greatly improved a previous draft. J. C. Nabout helped us to create the phylogenetic distance matrix and R. Vidanes helped with the distributional data. Research funding was provided by the European Commission (FP7 Marie Curie Intra-European Fellowship for Career Development to M.A´ .O.-T.), the Spanish Ministry of Science and Innovation (grant CGL2006-03000/BOS to M.A´ .R. and FPU fellowship AP2005-0636 to M.A´ .O.-T) and CNPq productivity grants to L.M.B. and J.A.F.D.-F.

Belk, M. C. and Houston, D. D. 2002. Bergmann’s rule in ectotherms: a test using freshwater fishes. Am. Nat. 160: 803 808. ¨ ber die Verha¨ltnisse der Wa¨rmeo¨ konomie Bergmann, C. 1847. U der Thiere zu ihrer Gro¨ sse. Gottinger Studien 3: 595 708. Blackburn, T. M. and Hawkins, B. A. 2004. Bergmann’s rule and the mammal fauna of northern North America. Ecography 27: 715 724. Blackburn, T. M. et al. 1999. Geographic gradients in body size: a clarification of Bergmann’s rule. Divers. Distrib. 5: 165 174. Blanckenhorn, W. U. and Demont, M. 2004. Bergmann and converse Bergmann latitudinal clines in arthropods: two ends of a continuum? Integr. Comp. Biol. 44: 413 424. Brown, J. H. 1995. Macroecology. Univ. Chicago Press. Cowles, R. B. and Bogert, C. M. 1944. A preliminary study of the thermal requirements of desert reptiles. Bull. Am. Mus. Nat. Hist. 83: 265 296. Cushman, J. H. et al. 1993. Latitudinal patterns in European ant assemblages: variation in species richness and body size. Oecologia 95: 30 37. Desdevises, Y. et al. 2003. Quantifying phylogenetically structured environmental variation. Evolution 57: 2647 2652. Diniz-Filho, J. A. F. and Toˆ rres, N. M. 2002. Phylogenetic comparative methods and the geographic range size-body size relationship in new world terrestrial Carnivora. Evol. Ecol. 16: 351 367. Diniz-Filho, J. A. F. et al. 1998. An eigenvector method for estimating phylogenetic inertia. Evolution 52: 1247 1262. Diniz-Filho, J. A. F. et al. 2003. Spatial autocorrelation and red herrings in geographical ecology. Global Ecol. Biogeogr. 12: 53 64. Diniz-Filho, J. A. F. et al. 2007. Seeing the forest for the trees: partitioning ecological and phylogenetic components of Bergmann’s rule in European Carnivora. Ecography 30: 598 608. Diniz-Filho, J. A. F. et al. 2009. Climate history, human impacts and global body size of Carnivora at multiple evolutionary scales. J. Biogeogr. 36: 2222 2236. Feder, M. E. 1982. Thermal ecology of neotropical lungless salamanders (Amphibia: Plethodontidae): environmental temperatures and behavioral responses. Ecology 63: 1665 1674. Feder, M. E. and Lynch, J. F. 1982. Effect of elevation, latitude, season and microhabitat on field body temperatures of neotropical and temperate zone salamanders. Ecology 63: 1657 1664. Freckleton, R. P. et al. 2002. Phylogenetic analysis and comparative data: a test and review of evidence. Am. Nat. 160: 712 726. Garland, T. Jr et al. 1993. Phylogenetic analysis of covariance by computer simulation. Syst. Biol. 42: 265 292. Gaston, K. J. et al. 2008. Ecogeographical rules: elements of a synthesis. J. Biogeogr. 35: 483 500. Highton, R. 1995. Speciation in eastern North American salamanders of the genus Plethodon. Annu. Rev. Ecol. Syst. 26: 579 600. IUCN, Conservation International and NatureServe 2006. Global Amphibian Assessment. <www.globalamphibians.org>, ver. 1.1, accessed 15 October 2006. Kozak, K. H. and Wiens, J. J. 2006. Does niche conservatism promote speciation? A case study in North American salamanders. Evolution 60: 2604 2621. Kozak, K. H. et al. 2006. Rapid lineage accumulation in a non-adaptive radiation: phylogenetic analysis of diversification rates in eastern North American woodland salamanders (Plethodontidae: Plethodon). Proc. R. Soc. B 273: 539 546.

Ku¨ hn, I. et al. 2009. Combining spatial and phylogenetic eigenvector filtering in trait analysis. Global Ecol. Biogeogr. 18: 745 758. Lannoo, M. J. 2005. Amphibian declines: the conservation status of United States species. Univ. California Press. Lindsey, C. C. 1966. Body sizes of poikilotherm vertebrates at different latitudes. Evolution 20: 456 465. Martins, E. P. et al. 2002. Adaptive constraints and the phylogenetic comparative method: a computer simulation test. Evolution 56: 1 13. Meiri, S. and Thomas, G. H. 2007. The geography of body size challenges of the interspecific approach. Global Ecol. Biogeogr. 16: 689 693. Olalla-Ta´rraga, M. A´ . and Rodrı´guez, M. A´ . 2007. Energy and interspecific body size patterns of amphibian faunas in Europe and North America: anurans follow Bergmann’s rule, urodeles its converse. Global Ecol. Biogeogr. 16: 606 617. Olalla-Ta´rraga, M. A´ . et al. 2006. Broad-scale patterns of body size in squamate reptiles of Europe and North America. J. Biogeogr. 33: 781 793. Olalla-Ta´rraga, M. A´ . et al. 2009. Geographic body size gradients in tropical regions: water deficit and anuran body size in the Brazilian Cerrado. Ecography 32: 581 590.

IBS SPECIAL

ISSUE

Download the Supplementary material as file E6244 from <www.oikos.ekol.lu.se/appendix>.

368

Olson, V. A. et al. 2009. Global biogeography and ecology of body size in birds. Ecol. Lett. 12: 249 259. Rangel, T. F. L. V. B. et al. 2006. Towards an integrated computational tool for spatial analysis in macroecology and biogeography. Global Ecol. Biogeogr. 15: 321 327. Ray, C. 1960. The application of Bergmann’s and Allen’s rules to the poikilotherms. J. Morphol. 106: 85 108. Rodrı´guez, M. A´ . et al. 2008. Bergmann’s rule and the geography of mammal body size in the western Hemisphere. Global Ecol. Biogeogr. 17: 274 283. Rohlf, F. J. 2001. Comparative methods for the analysis of continuous variables: geometric interpretations. Evolution 55: 2143 2160. Ruggiero, A. and Hawkins, B. A. 2006. Mapping macroecology. Global Ecol. Biogeogr. 15: 433 437. StatSoft 2003. STATISTICA (data analysis software system), version 6. <www.statsoft.com>. Wiens, J. J. et al. 2006. Rapid diversification, incomplete isolation, and the ‘‘speciation clock’’ in North American salamanders (genus Plethodon): testing the hybrid swarm hypothesis of rapid radiation. Evolution 60: 2585 2603.

Ecography 33: 369 379, 2010 doi: 10.1111/j.1600-0587.2010.06315.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editors: David Nogue´s-Bravo and Carsten Rahbek. Accepted 7 March 2010

Biogeography of body size in Pacific island birds Alison G. Boyer and Walter Jetz A. G. Boyer (alison.boyer@yale.edu) and W. Jetz, Dept of Ecology and Evolutionary Biology, Yale Univ., 165 Prospect St., New Haven, CT 06520-8106, USA.

Many insular vertebrates have undergone rapid and dramatic changes in body size compared to their mainland counterparts. Here we explore the relationship between two well known patterns of island body size the tendency for large-bodied species to dwarf and small-bodied species to get larger on islands, known as the ‘‘island rule’’, and the scaling of maximum and minimum body size of island assemblages with island area. Drawing on both fossil and modern data, we examined the relationship between body size and island area in Pacific island birds, both within clades and at the island assemblage level. We found that the size of the smallest bird on each island decreased with island area while the maximum body size increased with island area. Similarly, within clades the body size of small-bodied groups decreased and large-bodied groups increased from small to large islands, consistent with the island rule. However, the magnitude of size change within clades was not sufficient to explain the overall scaling of maximum size with island area. Instead, the pattern was driven primarily by the evolution of very large, flightless birds on large islands. Human-mediated extinctions on islands over the past few millennia severely impacted large, flightless birds, to the effect that this macroecological pattern has been virtually erased. After controlling for effects of biogeographic region and island area, we found island productivity to be the best predictor of maximum size in flightless birds. This result, and the striking similarities in maximum body size between flightless birds and island mammals, suggests a common energetic mechanism linking body size and landmass area in both the island rule and the scaling of island body size extremes.

369

ISSUE

island area, Filin and Ziv (2004) found the degree of body mass change in island birds and mammals to be inversely related to island area, where both insular dwarfing of large-bodied forms and gigantism in small-bodied forms were more pronounced on the smallest islands. These findings suggest that island rule-like body size changes could produce a scaling relationship between island area and maximum and minimum size in island assemblages. Alternatively, it is also conceivable that the scaling of size extremes occurs without the island rule and is driven by very large or small single-island endemics. Given these direct connections and likely causal overlap, a joint assessment is likely to integrate and advance our understanding of the structure of island communities. We hypothesize that area-scaling of size extremes may arise from an interaction between how island area and body size affect immigration, survival (emigration and extinction), or in situ evolution. However, size-area scaling could also be a simple result of sampling effects. Following well-established empirical and theoretical evidence in island biogeography (MacArthur and Wilson 1967, Rosenzweig 1995, Kalmar and Currie 2007, Kreft et al. 2008, Whittaker et al. 2008), larger islands are expected to harbor higher species richness. Even with just slight body size variance in the regional pool, species-rich assemblages on large islands should show larger body size

IBS SPECIAL

The body sizes of insular vertebrates often differ dramatically from those of their mainland counterparts. Body size is a fundamental ecological parameter that reflects many other ecological characteristics associated with resource requirements, life-history, and ecological interactions (Heaney 1978, Damuth 1981, Lomolino 1985, Brown 1995, McNab 2002) across species and assemblages. Mammal and bird body size on islands often follows a predictable pattern, known as the ‘‘island rule’’, where species that are large on the mainland tend to decrease in size on islands and small species increase in size (Clegg and Owens 2002, Lomolino 2005, Raia and Meiri 2006; see also Meiri et al. 2008). In a related pattern, the maximum body size of vertebrates on islands and other landmasses appears to increase with island area while minimum body size decreases, producing a divergent scaling of body mass extremes over island area referred to here as ‘‘area-scaling of body size extremes’’ (Marquet and Taper 1998, Burness et al. 2001, Okie and Brown 2009). The island rule is based on a species-focal, evolutionary perspective, where body size changes result from differential selection on body size in the island environment. In contrast, studies of body size extremes involve attributes of a whole assemblage and causal mechanisms may involve both species-level evolutionary processes as well as communitylevel processes. In a key study linking the island rule with

ISSUE

IBS SPECIAL

maxima and smaller minima than assemblages of fewer species on small islands. In this case a simple null model that simulates random colonization from the regional pool (following Simberloff 1978) should be sufficient to explain the area-extremes pattern. However, we expect variation above and beyond this sampling effect (see Marquet and Taper 1998) and we explore several other hypotheses for area-scaling that make additional, directional predictions. Area-scaling of body size extremes could arise in the absence of evolutionary processes through size-neutral (or potentially size-selective) immigration. Island biogeography theory and empirical studies have shown that colonization rate declines as a function of isolation (Heaney 1986, Whittaker and Ferna´ndez-Palacios 2007). Under the ‘‘target effect’’, larger islands may receive more immigrants than smaller islands of similar isolation (Whitehead and Jones 1969, Lomolino 1990), but no straightforward body size trends are obvious. While some larger-bodied species tend to disperse longer distances (e.g. birds: Sutherland et al. 2000; mammals: Van Vuren 1998), possible interactions between body size and island area are unknown. If immigration has played an important role in area-scaling of body size extremes, we would expect an effect of island isolation where less isolated islands have larger body size maxima and smaller minima than similar-sized, more isolated islands. Body size and island characteristics such as area should have clear ramifications on body size extremes by influencing the persistence of species. Extinction probabilities are strongly negatively associated with population size (Lande 1993, IUCN 2001) and limited land area may necessarily limit population size on islands. Population size per unit area (density) is constrained by body size such that smallbodied species may attain high or low densities but large species usually exhibit lower density (Damuth 1981, Silva et al. 1997, Jetz et al. 2004a). Assuming species of different body sizes require similar minimum viable population sizes (Traill et al. 2007), this means that on small islands largebodied species may be more prone to extinction and less likely to persist than small-bodied species. Population densities are additionally affected by environmental conditions. More productive areas are expected to facilitate larger energy flux through consumers and support a larger number of individuals and populations (Wright 1983, Evans et al. 2006, Hurlbert and Jetz 2010), and strong positive relationships between energy availability and population density have been demonstrated in, among other groups, ants, birds, and lizards (Evans et al. 2005, Meehan 2006, Buckley and Jetz 2007). Higher environmental temperature and precipitation are associated with higher environmental productivity (Woodward et al. 1995). We therefore predict, above and beyond sampling and area effects, wider body mass extremes on islands with increasing temperature and precipitation. Evolutionary change in body size on islands is a third possible mechanism for between-island variation in body mass extremes. Immigrant species may undergo in situ evolutionary changes in body size after establishment on an island, consistent with the island rule (Anderson and Handley 2002, McNab 2002). If island rule-like body size changes produce the area-scaling of maximum and minimum size in island assemblages, we would expect to see 370

strong effects of island area on the body sizes of closelyrelated species on different islands. Fourth, in situ diversification could produce area-scaling of size extremes because larger islands may support more speciose adaptive radiations (Kisel and Barraclough 2010), resulting in more species, and wider body mass extremes. This expectation would be partly captured by the null model based on species richness, or at least the same variables associated with higher species richness (area and productivity as above) may also promote diversification. In both evolutionary hypotheses we would expect to observe wider size extremes on older islands, relative to similarly-sized young islands, due to a longer period for in situ evolution. In summary, we predict an increase of body mass maxima and decrease of minima with island area, beyond the effect expected under random colonization. If mass- and area-selective immigration govern body size distributions then we expect size extremes to vary with island isolation. If energetic constraints on population densities and body sizes are important we predict higher temperature and precipitation to be associated with wider size extremes. Strong positive effects of island age on body size extremes would indicate in situ evolutionary processes generating the areascaling pattern. Finally, if island rule type body size changes contribute to the area-scaling of maximum and minimum size in island assemblages, we would expect to see a strong association of body size and island area within lineages. Although island body size patterns have been studied primarily in mammals, birds colonized and thrived on many oceanic islands, and island birds provide an independent opportunity to examine the constraints on body size evolution. Abundant zooarchaeological and fossil remains of birds from Pacific islands reveal a wide ecological and taxonomic diversity of birds that once dominated terrestrial vertebrate communities on oceanic islands (Worthy and Holdaway 2002, Steadman 2006). Fossil evidence shows that flightlessness on islands evolved in at least eight orders of birds (Dinornithiformes, Anseriformes, Psittaciformes, Strigiformes, Columbiformes, Gruiformes, Ciconiiformes, and Passeriformes), and rails, geese, ducks, pigeons, and ibises repeatedly evolved the flightless condition after colonization of numerous isolated islands (McNab 1994b, Slikas et al. 2002, Steadman 2006). On islands lacking mammalian predators, reduction of flight ability may have been selected for because of the associated decrease in individual metabolic requirements (McNab 2002), and a corresponding increase in population size (McNab 1994a). While this pattern is not strictly consistent with the island rule, it does appear to be energetically driven (Lomolino 2005). Island birds show several predictable patterns in response to island living, including the classical island rule (Clegg and Owens 2002), lower metabolic rates compared to mainland populations (McNab 1994b), niche expansion (Scott et al. 2003) and density compensation (MacArthur et al. 1972, Wright 1980). In addition, McNab (1994b) presented data suggesting a loose relationship between island area and body size in island birds. Here we use species-occurrence and body size data from fossil and extant assemblages to reconstruct the biogeography of body size in Pacific island birds. We examine the scaling of minimum and maximum body size with island area in both prehuman and modern avifaunas and test for

the additional effects of island age, isolation, environment and other island attributes on body size extremes. We also examine patterns of body size variation with island area within lineages in an effort to determine the relationship between the island rule and area-scaling of body size extremes.

Methods We gathered species occurrence and body size data for all known land birds from 48 Pacific islands, incorporating both fossil and modern data. Only islands with at least 10 fossil specimens were included in the dataset, as a lowerbound on fossil sampling effort. The data set spanned the tropical Pacific including Melanesia (10 islands), western Polynesia (14 islands), eastern Polynesia (9 islands), the Marianas (5 islands), New Zealand (4 islands), and the Hawaiian Islands (6 islands). Today, these islands range in size from 5 km2 to almost 146 000 km2 and cover a total area of 317 200 km2. On these islands, birds were the dominant terrestrial vertebrates due to the limited over-water dispersal abilities of non-volant mammals. We focused on the terrestrial environment, so all seabirds and shorebirds were excluded from analysis. Species lists of breeding birds for each island were primarily gathered from Worthy and Holdaway (2002) for the New Zealand region, Olson and James (1991) for the Hawaiian islands, and Steadman (2006) for other islands, but were supplemented by information from a variety of published sources (Supplementary material Appendix S1). Invasive species were excluded. The total historic and extant breeding avifauna across the study islands consists of 583 species. Physical attributes of each island, including land area (km2), maximum elevation (m), geology, distance from nearest continent (km) and isolation index (a composite measure incorporating distance from the nearest continent, island group, and island) were gathered from the United Nations Environment Program Islands Directory (1998). Mean annual temperature (8C) and mean annual precipitation (mm) for each island were extracted from WorldClim climatic layers (Hijmans et al. 2005).

We examined the relationship between island area and the mass of the largest and smallest species on each island with least-squares linear regression. Although the fossil record for many islands is far from complete, we were confident in characterizing maximum body sizes on each island for the following reasons: 1) large-bodied birds have a much higher likelihood of being preserved as fossils than small-bodied birds (Duncan et al. 2002, Boyer 2010), 2) large-bodied birds and non-passerine taxa are often discovered and described early in paleontological study of an island (Steadman 2006), and 3) the discovery of a larger species on one or a few islands would be unlikely to change our results since strong area-size relationships have been observed for both the largest and the second-largest species in island mammals (Okie and Brown 2009). For minimum size on each island, all volant species identified as smallest are extant and their size estimates were not based on fossils. Minimum size of flightless species was based primarily on fossil evidence, and the future discovery of smaller flightless species could alter our results for this group. To facilitate interpretation, for area-scaling analyses, we excluded all predatory birds, such as owls, hawks, and herons, due to the substantially different population densities of carnivorous vertebrates (Juanes 1986, Jetz et al. 2004a), and because size evolution in different trophic levels of island mammals has been shown to result 371

ISSUE

Available body size data for Pacific island species were gathered from the literature (Supplementary material Appendix S1). However, few body size estimates for Pacific island endemics, especially extinct species, have been published. For these species, we developed body mass estimates based on the allometry of hind limb skeletal measurements in over 600 avian skeletal specimens (277 species from 13 orders of birds and 21 Passerine families) following the methods of Campbell and Marcus (1992). Only specimens with associated live-masses were measured and females and left leg were used preferentially. Specimens were selected to include a broad range of body sizes and an effort was made to include as many Pacific island genera as possible. We examined the scaling of mass with hindlimb diameter using linear regression of natural log-transformed

Analyses

IBS SPECIAL

Body size estimates

bone diameter (mm) on natural log-transformed mass (g). Body mass (M) was predictable from femur diameter (ln(M) 2.44 ln(femur) 2.50, R2 0.95, n 580), tibiotarsus diameter (ln(M) 2.41 ln(tib.) 2.72, R2 0.96, n 561), and tarsometatarsus diameter (ln(M) 2.30 ln(tars.) 2.97, R2 0.93, n 547). Allometry of femur and tibiotarsus diameters did not differ between flightless and volant birds, although flightless birds showed a slightly steeper scaling of tarsometatarsus diameter (t-tests, femur: t 0.19, DF 576, p 0.85; tibiotarsus: t 1.92, DF 557, p 0.05; tarsometatarsus: t 5.56, DF 543, p B0.001). We calculated the average prediction error (APE) of the regression equations using independent measurements of 33 specimens of disparate sizes and taxonomy. APE reflects the mean percent difference between the mass estimate and the actual mass for each specimen. We found an APE of 5.96% for estimates based on the tibiotarsus, a value well within the range acceptable for studies utilizing body size estimates (Damuth and MacFadden 1990). To estimate body size of extinct species, we measured the hindlimb of 591 fossil specimens of 155 Pacific island species, and obtained measurements from the literature for 77 more taxa. We applied the formulas listed above, using the tibiotarsus preferentially over the other bones, to obtain mass estimates. Although a multivariate model incorporating all three hindlimb elements may have been preferred, this was not an option for the majority of fossil specimens due to the lack of associated skeletal material. One mass estimate was made for each specimen, and population-level size estimates were based on the mean of specimens of each species from each island.

ISSUE

IBS SPECIAL

from different underlying mechanisms and result in different body size patterns (Raia and Meiri 2006). Insectivores were included in the analysis; they comprised the majority (37 of 46) body size minima, but no body size maxima were insectivorous. Body mass (g) and area (km2) were log10-transformed prior to analysis. Because flightless birds, which in most cases evolved in situ on each island, might be expected to reflect a stronger relationship between island area and size than volant (flying) species, we analyzed flightless and volant species separately. We repeated the analysis including only those species still living today (extant species). Because large islands and landmasses support both higher species richness and also a greater total resource supply, it is important to determine the influence of sampling effects on size-area scaling. We developed a simple null model to evaluate whether the scaling of body size extremes with island area is simply a result of the greater species richness on larger islands given the regional species pool. Ideally the species pool should reflect the potential source pool of species that are capable of dispersing to, establishing on, and/or evolving on Pacific islands. Understanding of the potential source pools for Pacific island birds is severely hampered by the lack of phylogenetic studies of island birds (Steadman 2006). Without clear, defensible source pools for each island or island group, it is unclear whether localized (based on biogeographic regions) or extensive source pools (including source areas from which colonists originated in the past Australia, New Guinea, Asia, and North America) would represent the most accurate source pool for our null model. As an intermediate solution, we used the total species list from across the 48 islands in our dataset (502 species) as the regional source pool. For each of the islands we drew from the species pool at random without replacement the number of species known from that island and recorded their masses. In separate model runs, body masses were drawn from the Pacific-wide list of species either 1) unweighted, and 2) weighted by the number of island occurrences (Gotelli and Graves 1996, Jetz et al. 2004b). Expected null values for the size-area scaling relationship were estimated by fitting regressions to the simulated maximum and minimum body size over empirical island area. This procedure was repeated 1000 times to produce a distribution of expected slopes for the null model. Null models were run for the full pre-human avifauna, the extant avifauna, as well as for pre-human flightless and volant species. We compared empirical slopes to the distribution of null model slopes using one-tailed t-tests. We compared the effect of area on maximum body size in birds to that of other environmental factors on each island in a generalized linear modeling framework (function glm in R v2.8.1; R Foundation for Statistical Computing 2008). Categorical predictors of maximum body size for each island included biogeographic region (eastern Polynesia, Hawaii, Melanesia, Marianas, New Zealand, and western Polynesia) and geological type (continental, coralline, volcanic-coral mix, and volcanic, see also Kreft et al. 2008). Continuous environmental predictors were mean annual temperature (8C), annual precipitation (mm), maximum island elevation (m), species richness, island area (km2), an estimate of island age or time since last emergence 372

(millions of years), distance to nearest continent (km), and isolation (UNEP Island Directory Isolation Index). All continuous predictors were log10-transformed before analysis, except isolation, which was normally distributed. Predictor variables were mostly weakly correlated, but 5 out of 28 variable combinations had Pearson correlations of 0.70 or greater (Supplementary material Table S1). We ran single-predictor and multi-predictor generalized linear models (GLM) to test the influence of predictors on maximum body size (Mmax) in flightless and volant island birds. To control for the effect of species richness, null model predictions of body size maxima were included as a covariate in ‘‘richness-controlled’’ glm models. Because biogeographic region showed a strong and significant effect on Mmax in pairwise models, we also developed multipredictor mixed-effects models where biogeographic region (Region LME) was treated as a random effect. Goodness of fit of each model was measured by Akaike information criterion (AIC; Burnham and Anderson 2002) where smaller AIC values indicate a better model fit. To test for within-clade patterns of body size evolution, we explored the relationship between log10 body mass and log10 island area within avian genera and families in a nested linear mixed-effects model. Separate models were constructed for small- and large-bodied birds, with the cutoff point at 60 g corresponding roughly to the mode of the global body size distribution of birds (Blackburn and Gaston 1994, Maurer 1998). Scaling intercept was allowed to vary between clades as a random effect of family nested within order. This mixed effects model was compared to a cross-species, GLM model of body size over area where all data points were equally weighted. In an additional analysis we compared the scaling of maximum size in flightless island birds to that observed in island mammals. We gathered data from the literature (sources given in Supplementary material Appendix S1) on the largest mammal species found on seven Caribbean islands, New Guinea, Madagascar, and five continental landmasses during the late-Pleistocene. We compared our results to the area-scaling relationship observed by Okie and Brown (2009) for mammals on islands of the Sunda shelf.

Results The maximum size of birds in the pre-human avifauna of Pacific islands was geographically structured (Fig. 1). The large islands of New Zealand in particular, but also New Caledonia and the Hawaiian islands emerge as harboring particularly large-bodied species, while small, isolated islands in the Marquesas and Henderson island had much smaller species. Before human impacts, maximum body size (Mmax) of birds strongly increased with island area (Table 1, Fig. 2a). After severe extinctions of Pacific island birds over the past few millennia, the modern avifauna shows a much weaker but still statistically significant area-scaling of maximum size (Table 1, Fig. 2b). In the pre-human avifauna, the scaling of Mmax appeared somewhat nonlinear, with a steeper slope above 102 km2. The structure of this increase was further enlightened when flightless and volant species were considered separately (Fig. 2c and d). Flightless species showed a much steeper scaling of Mmax with area

Figure 1. Map of Pacific islands showing the maximum body size of birds before human colonization on each island.

extant and volant species the observed slope was steeper than the null expectation (Table 1, Fig. 2b, d). Maximum body size in flightless island birds was positively correlated with species richness, area, and elevation, and was negatively related to temperature and distance from mainland (Supplementary material Table S2). Biogeographic regions Hawaii, Melanesia, and New Zealand harbored significantly larger flightless birds, as did continental islands in comparison to other geologic types. After controlling for richness, these effects remained significant (Table 2). In volant birds, maximum body size was related to species richness, area, temperature and distance (Supplementary material Table S2). Eastern Polynesia and the Marianas had smaller volant birds than other biogeographic regions. After correcting for richness, none of these effects remained significant (Table 2). After correcting for richness using the null model, a multi-predictor model incorporating area, temperature and precipitation provided the best fit to maximum size of flightless birds, explaining 89% of variation in Mmax (Table 3). This model remained the best fit within mixed-effects models incorporating biogeographic region. In volant birds, maximum size was best explained by area alone, but models including island

Table 1. Scaling of log10 maximum and minimum body mass with log10 island area in non-predatory Pacific island birds. Linear regression statistics (from Fig. 2) and null model slopes (above: species pool unweighted, below: species pool weighted by number of occurrences) are provided. No unweighted null model was run for flightless species due to very high levels of endemism. pnull was measured by one-tailed t-tests between empirical and null model slopes. Maximum body size Slope (SE)

Pre-human (n 46)

2.17

0.44 (0.05)

Extant (n 46)

2.49

0.13 (0.05)

Flightless (n 32) Volant (n 46)

1.2 2.7

0.69 (0.09) 0.14 (0.04)

*** 0.61 *

0.15

*** 0.68 *** 0.29

Null slope (SE) 0.24 0.23 0.10 0.08 0.31 0.11 0.08

(0.09) (0.10) (0.03) (0.03) (0.14) (0.06) (0.04)

pnull

Int.

Slope (SE)

* * ns ns * ns ns

1.25

0.11 (0.03)

*** 0.29

1.5

0.16 (0.06)

2.37 1.25

0.09 (0.09) 0.11 (0.03)

ns 0.04 *** 0.29

0.19

Null slope (SE) 0.04 0.03 0.03 0.03 0.03 0.04 0.04

(0.01) (0.01) (0.01) (0.01) (0.11) (0.02) (0.02)

pnull * * * * ns * *

Significance of p-values: ns 0.10, *B0.05, **B0.01, *** B0.001.

373

ISSUE

Int.

Minimum body size

IBS SPECIAL

than did volant species (Table 1). Minimum body size decreased with island area, with similar slopes observed in pre-human, modern, flightless, and volant birds (Table 1, Fig. 2). Size extremes in island birds were correlated with species richness (Supplementary material Table S2, Fig. S1), but the overall observed scaling of maximum and minimum avian body size with island area was not a simple sampling artifact of greater numbers of species on large islands (Table 1, Supplementary material Fig. S1) regardless of whether the species pool was weighted by number of island occurrences and was steeper than expected from such a null model (Fig. 2a). Observed size maxima were significantly lower than that predicted based on species richness for flightless species (paired t-test, DF 28, t 4.42, pB0.001; Supplementary material Fig. S2) but not in volant species (paired t-test, DF 39, t 1.49, p 0.14; Supplementary material Fig. S2). For flightless species, areascaling of Mmax was significantly steeper than expected, but the scaling of minimum size was not significantly different than that expected under random sampling (Table 1). The scaling of maximum size in extant and volant species did not differ from the null model, but for minimum size in

Figure 2. Scaling of body size extremes with island area for non-predatory birds on 48 Pacific islands. Solid lines show linear regression fits to the maximum (squares) and minimum (circles) body mass of (a) the pre-human avifauna, (b) extant birds, (c) flightless, and (d) volant species on each island. Null model predictions based on sampling (unweighted species pool) are shown with dashed lines (with confidence intervals given by dotted lines). Regression statistics are given in Table 1.

age, elevation, precipitation, and distance were also wellsupported (Table 3). We found considerable support for island rule type body size patterns within genera and families of Pacific island birds (Fig. 3). Several small-bodied groups showed a negative relationship between island area and body size, where large islands supported smaller species than small

islands, and for large-bodied groups the trend was the opposite. Accounting for different intercepts among clades, mixed effects models showed a trend reminiscent of the island rule: large-bodied groups decreased in size on smaller islands and small-bodied groups increased in size (Table 4). We tested the effect of biogeographic region in the withinclade analysis by including it (as a categorical fixed effect) in

IBS SPECIAL

ISSUE

Table 2. Relationships between environmental predictors and log10 maximum body size (Mmax) in Pacific island birds, while controlling for effects of species richness (see Supplementary material Table S2 for regressions on raw data). Regression statistics are given for each predictor; smaller AIC values indicate better fit. Sample sizes: flightless n 32 and volant n 46. In the two categorical variables Geology and Region, each level was related to the baseline category (Continental and E. Polynesia, respectively). All continuous predictors, except isolation, were log10-transformed before analysis. Predictor

Category

Island area Island age Island elevation Precipitation Temperature Distance Isolation Geology Continental Coralline Volc. and coral Volcanic Region E. Polynesia Hawaii Melanesia Marianas New Zealand W. Polynesia

Flightless Mmax Int.

Slope (SE)

AIC

Int.

Slope (SE)

AIC

0.61 1.23 1.59 2.35 4.16 2.33 1.33 1.92 0.39

0.43 (0.10) 0.16 (0.17) 0.46 (0.21) 1.15 (0.83) 2.98 (1.21) 0.96 (0.44) 0.00 (0.004) 1.69 (0.49) 1.54 (0.49) 1.13 (0.38) 1.06 (0.24) 0.72 (0.26) 0.27 (0.24) 1.58 (0.43) 0.56 (0.25)

*** ns * ns * * ns ** ** ** *** * ns ** *

37.55 52.56 48.51 51.43 38.52 48.56 53.5 45.6 38.51

0.08 0.70 0.73 0.92 0.95 0.24 1.09 0.44 1.19

0.07 (0.04) 0.01 (0.07) 0.01 (0.06) 0.06 (0.32) 0.52 (0.45) 0.07 (0.22) 0.00 (0.001) 0.11 (0.17) 0.16 (0.17) 0.04 (0.15) 0.23 (0.15) 0.22 (0.20) 0.14 (0.13) 0.32 (0.23) 0.19 (0.14)

ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns

0.2 2.88 5.05 2.87 3.15 2.79 2.48 4.72 2.44

Significance of p-values: ns 0.10, * B 0.05, ** B 0.01, *** B 0.001.

374

Volant Mmax

Table 3. Multi-predictor models of log10 maximum body size (Mmax) in the pre-human avifaunas of Pacific islands. AIC values are given for both general linear models (GLM) and linear mixed-effect models incorporating random effects of biogeographic region (Region LME); R2 values are given for GLM models. The combined model includes: Area Age Elev. Dist. Precip. Temp. Geology. All models control for the effects of species richness by including the species richness null model (unweighted species pool) as a fixed effect. All continuous predictors were log10-transformed before analysis. Models with AIC values within 2 AIC of the lowest for each model type are shown in bold. Model

Flightless Mmax GLM

Intercept Area Area Age Area Elev. Area Age Elev. Area Precip. Area Temp. Area Temp. Precip. Area Dist. Area Geology Combined

Volant Mmax Region LME

GLM

Region LME

AIC

51.50 37.55 39.46 39.54 41.46 34.13 33.58 22.80 39.44 38.85 28.60

0.80 0.80 0.80 0.80 0.84 0.82 0.89 0.80 0.83 0.92

51.19 46.05 49.53 49.21 52.62 41.15 41.41 31.85 47.27 45.34 40.87

0.91 0.20 2.15 1.10 2.71 2.04 2.65 4.52 2.15 3.82 7.20

0.41 0.41 0.46 0.47 0.41 0.35 0.35 0.41 0.45 0.54

the nested taxonomic model. The addition of region to the model improved fit for both large-bodied (Area: AIC 541 vs Area Region: AIC 494) and small-bodied species (Area: AIC 281 vs Area Region: AIC 294). We conclude that region has a significant effect on the within-clade scaling of body size with island area. However, within-clade evolution did not appear sufficient to account for the scaling of Mmax with area in the prehuman avifauna, as the slope within clades of large-bodied species was more shallow than the slope in the cross-species model (Table 4). The upper boundary on body size appeared to increase with the accumulation of additional clades and singleton taxa on larger islands (Fig. 3c, d).

AIC 8.54 11.16 15.47 15.80 20.70 13.13 10.78 12.91 13.12 23.18 30.83

In mammals, maximum body size exhibits a strong scaling with landmass area on Caribbean islands and continental landmasses (linear regression, n 14, slope 0.66, SE 0.05, pB0.001; Fig. 4), and the slope is statistically indistinguishable from the observed scaling in Pacific island flightless birds (slope 0.69; t-test, DF 39, p 0.36). The area-scaling of Mmax in flightless birds is also consistent with the size-area scaling exponents of 0.56 and 0.62 observed for the largest and second largest mammal species on islands of the Sunda shelf (t-tests, DF 39 and 39, p 0.17 and 0.29; Okie and Brown 2009), and the sizearea scaling exponent of 0.52 observed for top endothermic herbivores of landmasses around the world (t-test, p 0.11; Burness et al. 2001).

Discussion

ISSUE

375

IBS SPECIAL

Figure 3. Scaling of body size with island area within (a) genera and (b) families of Pacific island birds. Regression lines for significant (p B0.10; black lines) and non-significant (gray lines) are shown without species data points for purposes of clarity. Singleton genera and families are shown with open circles. Heavy dashed lines indicate the overall area-size scaling for maximum and minimum size in the pre-human avifauna (from Fig. 2a). Number of (c) genera and (d) families on each island increases with area (cubic spline fits).

The body size of Pacific island birds before human arrival was strongly structured by island area and geography. We found a strong increase in body size maxima and a decrease in size minima with increasing island area in Pacific island birds (Fig. 2) and the striking similarity of this relationship to that in mammals suggests a common mechanism. Interspecific patterns reminiscent of the island rule (intraspecific in its original form) were present in Pacific island bird clades (Fig. 3), and these patterns matched the overall area-scaling of minimum size, but within-clade patterns were not sufficient to account for the steep area-scaling of maximum size. Area-scaling of maximum body size on Pacific islands was driven by the pattern within flightless species, as 12 of 46 largest species were flightless and these flightless species were found on larger islands. Although area was the best single predictor of maximum size in flightless birds, Region also had a strong effect (Table 2), reflecting the complex biogeographic histories of Pacific islands. Area-scaling of maximum size in flightless birds is not a sampling artifact, where large islands would be expected to support larger-bodied species as a result of greater species richness. On the contrary, flightless birds have a very high degree of endemism, having primarily evolved their body

Table 4. Results of cross-species and nested mixed-effects models on the scaling of body size with island area in pre-human Pacific island birds. Sample sizes (n), slopes, standard error of slopes (SE), and p-values are given for each model. In the nested taxonomic model, scaling intercept was allowed to vary between clades as a random effect of Family nested within Order. Cutoff between large- and small-bodied species was 60 g. Significance of p-values: ns 0.10, * B 0.05, **B 0.01, ***B 0.001. Cross-species model Slope

Slope

0.015 0.104 0.010

0.017 0.016 0.010

ns *** ns

0.003 0.033 0.031

0.004 0.011 0.007

ns ** ***

size in situ on each island, and thus their body size is tightly linked to local conditions. Under these circumstances, one might expect a strong signal of island age on the maximum size of flightless birds. However, age did not account for much variation in size, neither in single-predictor models (Table 2) nor in multi-predictor models incorporating effects of area and biogeographic region (Table 3). Perhaps this should not be surprising given that loss of flight (Slikas et al. 2002) and dramatic changes in body size (Lister 1989, Keogh et al. 2005) are thought to occur very rapidly on islands. We found little effect of distance or island isolation on maximum size in flightless birds. It seems clear that flightless birds have very low dispersal capabilities. While the phylogenetic affinities of many flightless island birds are subjects of debate (Worthy 2001, Worthy and Holdaway 2002), it is thought that many flightless rails may have originated from multiple colonizations of one or a few widespread, volant rails (Slikas et al. 2002, Kirchman and Steadman 2006). In this case, there is little reason to expect a strong influence of isolation on size in flightless birds. Islands are often likened to replicated ‘‘natural experiments’’ where evolutionary processes can be studied (MacArthur and Wilson 1967, Carlquist 1974, Williamson 1981, Whittaker and Ferna´ndez-Palacios 2007). Repeated on islands around the world, evolution of large body size in flightless birds is one such natural experiment. The scaling of Mmax with island area in Pacific island flightless birds parallels that observed for island mammals, with empirical scaling exponents in the range 0.52 0.69. Might these exponents simply reflect a common effect of resource limitation in the evolution of island communities? Following allometric theory and empirical support from mammals,

IBS SPECIAL

ISSUE

All species (n 1197) Large-bodied species (n 668) Small-bodied species (n 529)

Figure 4. Scaling of maximum body size with island area in birds (circles) and mammals (squares). Area scaling of maximum body size in flightless Pacific island birds (solid line), Sunda shelf mammals (dashed line; data from Okie and Brown 2009), and Caribbean and continental mammals (dotted line).

376

Nested taxonomic model

the body-mass scaling of area A required to support an individual I of mass M can be approximated as A/I 8Mb (Jetz et al. 2004a), where the exponent b represents the body mass scaling exponent or field of basal metabolic rate B, B8Mb. In mammals b varies around 0.75 (Savage et al. 2004, Anderson and Jetz 2005) although the exponent may be lower in some clades (White et al. 2009). In birds b is usually slightly smaller (Savage et al. 2004, Jetz et al. 2008) and the body mass dependence of area needs is weaker (Brown and Maurer 1987). Assuming that minimum viable number of individuals, Imin, (minimum viable population size; Traill et al. 2007) as well as the proportion of habitable island area are invariant with regard to body size, then the area (i.e. the ‘‘minimum dynamic area’’, sensu Pickett and Thompson 1978) necessary to accommodate the total minimum viable population size, Imin, should be approximated by A 8Imin Mb. Conversely, the maximum size Mmax supported on an island of area A would scale as Mmax 8 A1/b. For both mammals and birds we might thus expect a maximum body size-island area scaling exponent, 1/b, in the range of 1.25 1.50. The empirical values for area-scaling exponents we found are significantly lower than these predictions. There are several non-mutually exclusive explanations for this discrepancy (outlined in Okie and Brown 2009), including: 1) the proportion of each island that is occupied by a species is itself a function of body mass or island area (as may be expected from potential body mass-occupancy and scaleproportional occupancy relationships, Hartley et al. 2004, Hurlbert and White 2007), or 2) minimum viable population size (Imin) increases with body size (as suggested by Brook et al. 2006), or 3) individual area requirements scale much more steeply than 3/4 (Okie and Brown 2009). While previous studies of size-area scaling in island mammals have been based on extinction-structured ‘‘relaxation’’ faunas (Marquet and Taper 1998, Okie and Brown 2009), the flightless Pacific island avifauna is derived from different historical biogeographic processes. While observed maximum size in flightless birds could be interpreted as an equilibrium at the largest sustainable maximum size for each island, it is unlikely that all islands supported a species at the maximum sustainable size due to biogeographic limitations and phylogenetic constraints on body plan and dietary plasticity. Higher energy availability per unit area on highly productive (warm and wet) islands would be expected to decrease per individual area needs and thus facilitate the persistence of larger species. We found that, in addition to area, temperature and precipitation formed the best model of maximum size in flightless birds (Table 3). This

avifauna has been through a severe extinction filter, we caution that basing biogeographic and macroecological theory on the modern island avifauna could be misleading. In summary, we found that the increase of body mass maxima and decrease of minima with island area arise from a combination of evolutionary, ecological, and historical biogeographic processes. Area-scaling of maximum size in flightless island birds was not an artifact of the species area relationship and sampling effects, and was not governed by size- or area-selective immigration or limited by island age. Within-clade, cross-species ‘‘island rule’’ type patterns were present, but only accounted for part of the variation. Instead, our results suggest that the effect of island area on maximum body size may reflect body size mediated constraints of energy availability on population survival. Further exploration of the energetic limitation of island populations will be necessary to elucidate the common mechanism linking body size and landmass area in both the island rule and the scaling of island body size extremes. Acknowledgements J. Belmaker, N. Cooper, L. Buckley, A. Hurlbert, D. McGlinn and J. Stegen provided comments that improved the manuscript. A.G.B. was supported by a National Science Foundation Postdoctoral Research Fellowship in Biological Informatics (DBI-0805669). This study was partially supported by a National Science Foundation Grant BCS0648733 to W.J. Data collection was made possible by grants from the Univ. of New Mexico and the Frank M. Chapman Fund of the American Museum of Natural History.

References

377

ISSUE

Alcover, J. A. et al. 1998. The extent of extinctions of mammals on islands. J. Biogeogr. 25: 913 918. Anderson, K. J. and Jetz, W. 2005. The broad-scale ecology of energy expenditure of endotherms. Ecol. Lett. 8: 310 318. Anderson, R. P. and Handley, C. O. 2002. Dwarfism in insular sloths: biogeography, selection, and evolutionary rate. Evolution 56: 1045 1058. Biber, E. 2002. Patterns of endemic extinctions among island bird species. Ecography 25: 661 676. Blackburn, T. M. and Gaston, K. J. 1994. The distribution of body sizes of the world’s bird species. Oikos 70: 127 130. Blackburn, T. M. et al. 2004. Avian extinction and mammalian introductions on oceanic islands. Science 305: 1955 1958. Boyer, A. G. 2008. Extinction patterns in the avifauna of the Hawaiian islands. Divers. Distrib. 14: 509 517. Boyer, A. G. 2010. Consistent ecological selectivity through time in Pacific island avian extinctions. Conserv. Biol. 24: 511 519. Brook, B. W. et al. 2006. Minimum viable population size and global extinction risk are unrelated. Ecol. Lett. 9: 375 382. Brown, J. H. 1995. Macroecology. Univ. Chicago Press. Brown, J. H. and Maurer, B. A. 1987. Evolution of species assemblages: effects of energetic constraints and species dynamics on the diversification of the North American avifauna. Am. Nat. 130: 1 17. Buckley, L. B. and Jetz, W. 2007. Insularity and the balance of environmental and ecological determinants of population density. Ecol. Lett. 10: 481 489. Burness, G. P. et al. 2001. Dinosaurs, dragons and dwarfs: the evolution of maximal body size. Proc. Nat. Acad. Sci. USA 98: 14518 14523.

IBS SPECIAL

relationship appears to result from the interaction of temperature with precipitation. In richness-controlled, multi-predictor models including both temperature and precipitation (Table 3) the effect of temperature was positive. However, we found no significant pairwise effect of precipitation on Mmax in flightless or volant species. The average annual temperature on islands in our dataset ranges from 8.4 to 26.98C and precipitation ranges from 71.5 to 417.7 cm yr 1. We note that temperature and precipitation are only proxies for productivity, and that our dataset may not span enough variation in these variables for a significant signal. Future investigations will benefit from a careful assessment of the interaction between habitable area, occupancy, and estimates of island-wide net productivity which were not available for this study. The additional evaluation of higher trophic levels will also be critical, as energy and thus space needs can vary manifold between primary and tertiary consumers of the same body size (Damuth 1981, Jetz et al. 2004a, Nagy 2005). The minimum size of volant birds on Pacific islands decreases with island area with an exponent steeper than predicted based on sampling alone (Table 1). There is little variation in minimum size across the range of island areas, with the majority (75%) of islands having a species in the range of 6 10 g. There could be problems in determining the smallest bird on each island because of incomplete sampling of small and inconspicuous birds both in modern surveys and the fossil record. However, the negative relationship between island area and body size within clades of small-bodied birds (Table 4) suggests that size in smallbodied birds does respond to island conditions. Similar patterns have been observed in small-bodied landbirds of New Zealand (Cassey and Blackburn 2004), islands off the coast of Australia (Scott et al. 2003), and in a sample of global islands (Clegg and Owens 2002) where size changes were thought to reflect release from interspecific competition on islands. Body size changes in large-bodied and small-bodied forms may be driven by different ecological factors (Heaney 1978, Clegg and Owens 2002, Cassey and Blackburn 2004). A reconciliation of the area-scaling of size extremes with the island rule will involve integrating information from both small- and large-bodied forms. Extinction has substantially altered the biogeography of body size in island birds. The scaling of maximum size has become much weaker in the modern avifauna and today is only marginally different than expected under random sampling. Human colonization of islands is linked to severe extinction episodes on islands worldwide (Pimm et al. 1994, Steadman 1995, Burney 1997, Alcover et al. 1998, Biber 2002, Blackburn et al. 2004, Duncan and Blackburn 2004), and these extinctions were often strongly size-biased and removed many large-bodied and flightless bird species (Duncan et al. 2002, Roff and Roff 2003, Boyer 2008, 2010). The loss of island megafauna has potentially resulted in major changes in ecosystem function (Hansen and Galetti 2009). In New Zealand and the Hawaiian islands, two of the most well-studied island groups, coevolution of plants with now-extinct browsing birds has been documented (James and Burney 1997, Worthy and Holdaway 2002), and in Tonga extinction of large, frugivorous pigeons may have disrupted seed dispersal in several tree species (Meehan et al. 2002). Because today’s Pacific island

ISSUE

IBS SPECIAL

Burney, D. A. 1997. Tropical islands as paleoecological laboratories: gauging the consequences of human arrival. Hum. Ecol. 25: 437 457. Burnham, K. P. and Anderson, D. R. 2002. Model selection and multimodel inference: a practical information-theoretic approach, 2nd ed. Springer. Campbell, J. K. E. and Marcus, L. 1992. The relationship of hindlimb bone dimensions to body weight in birds. In: Campbell, K. E. Jr (ed.), Papers in avian paleontology honoring Pierce Brodkorb. Natural History Museum of Los Angeles County, Los Angeles, CA, pp. 395 412. Carlquist, S. 1974. Island biology. Columbia Univ. Press. Cassey, P. and Blackburn, T. M. 2004. Body size trends in a Holocene island bird assemblage. Ecography 27: 59 67. Clegg, S. M. and Owens, I. P. F. 2002. The ‘island rule’ in birds: medium body size and its ecological explanation. Proc. R. Soc. B 269: 1359 1365. Damuth, J. 1981. Population density and body size in mammals. Nature 290: 699 700. Damuth, J. and MacFadden, B. J. 1990. Body size in mammalian paleobiology: estimation and biological implications. Cambridge Univ. Press. Duncan, R. P. and Blackburn, T. M. 2004. Extinction and endemism in the New Zealand avifauna. Global Ecol. Biogeogr. 13: 509 517. Duncan, R. P. et al. 2002. Prehistoric bird extinctions and human hunting. Proc. R. Soc. B 269: 517 521. Evans, K. L. et al. 2005. The roles of extinction and colonization in generating species energy relationship. J. Anim. Ecol. 74: 498 507. Evans, K. L. et al. 2006. Abundance, species richness, and energy availability in the North American avifauna. Global Ecol. Biogeogr. 15: 372 385. Filin, I. and Ziv, Y. 2004. New theory of insular evolution: unifying the loss of dispersability and body-mass change. Evol. Ecol. Res. 6: 115 124. Gotelli, N. J. and Graves, G. R. 1996. Null models in ecology. Smithsonian Inst. Hansen, D. M. and Galetti, M. 2009. The forgotten megafauna. Science 324: 42 43. Hartley, S. et al. 2004. Coherence and discontinuity in the scaling of species’ distribution patterns. Proc. R. Soc. B 271: 81 88. Heaney, L. R. 1978. Island area and body size of insular mammals: evidence from the tri-colored squirrel (Callosciurus prevosti ) of southeast Asia. Evolution 32: 29 44. Heaney, L. R. 1986. Biogeography of mammals in SE Asia: estimates of rates of colonization, extinction and speciation. Biol. J. Linn. Soc. 28: 127 165. Hijmans, R. J. et al. 2005. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25: 1965 1978. Hurlbert, A. H. and White, E. P. 2007. Ecological correlates of geographical range occupancy in North American birds. Global Ecol. Biogeogr. 16: 764 773. Hurlbert, A. H. and Jetz, W. 2010. More than ‘‘More Individuals’’: the nonequivalence of area and energy in the scaling of species richness. Am. Nat., online early. IUCN 2001. IUCN Red List categories and criteria: version 3.1. IUCN Species Survival Commission, IUCN, Gland, Switzerland and Cambridge, UK. James, H. F. and Burney, D. A. 1997. The diet and ecology of Hawaii’s extinct flightless waterfowl: evidence from coprolites. Biol. J. Linn. Soc. 62: 279 297. Jetz, W. et al. 2004a. The scaling of animal space use. Science 306: 266 268.

378

Jetz, W. et al. 2004b. The coincidence of rarity and richness and the potential signature of history in centres of endemism. Ecol. Lett. 7: 1180 1191. Jetz, W. et al. 2008. Environment, migratory tendency, phylogeny, and basal metabolic rate in birds. PLoS One 3: e3261. Juanes, F. 1986. Population density and body size in birds. Am. Nat. 128: 921 929. Kalmar, A. and Currie, D. J. 2007. A unified model of avian species righness on islands and continents. Ecology 88: 1309 1321. Keogh, J. S. et al. 2005. Rapid and repeated origin of insular gigantism and dwarfism in Australian tiger snakes. Evolution 59: 226 233. Kirchman, J. J. and Steadman, D. W. 2006. Rails (Rallidae: Gallirallus) from prehistoric archaeological sites in western Oceania. Zootaxa 1316: 1 31. Kisel, Y. and Barraclough, T. G. 2010. Speciation has a spatial scale that depends on levels of gene flow. Am. Nat. 175: 316 334. Kreft, H. et al. 2008. Global diversity of island floras from a macroecological perspective. Ecol. Lett. 11: 116 127. Lande, R. 1993. Risks of population extinction from demographic and environmental stochasticity and random catastrophes. Am. Nat. 142: 911 927. Lister, A. M. 1989. Rapid dwarfing of red deer on Jersey in the Last Interglacial. Nature 342: 53 542. Lomolino, M. V. 1985. Body size of mammals on islands: the island rule re-examined. Am. Nat. 125: 310 316. Lomolino, M. V. 1990. The target area hypothesis: the influence of island area on immigration rates of non-volant mammals. Oikos 57: 297 300. Lomolino, M. V. 2005. Body size evolution in insular vertebrates: generality of the island rule. J. Biogeogr. 32: 1683 1699. MacArthur, R. H. and Wilson, E. O. 1967. The theory of island biogeography. Princeton Univ. Press. MacArthur, R. H. et al. 1972. Density compensation in island faunas. Ecology 53: 330 342. Marquet, P. A. and Taper, M. L. 1998. On size and area: patterns of mammalian body size extremes across landmasses. Evol. Ecol. 12: 127 139. Maurer, B. A. 1998. The evolution of body size in birds. II. The role of reproductive power. Evol. Ecol. 12: 935 944. McNab, B. K. 1994a. Energy conservation and the evolution of flightlessness in birds. Am. Nat. 144: 628 642. McNab, B. K. 1994b. Resource use and the survival of land and freshwater vertebrates on oceanic islands. Am. Nat. 144: 643 660. McNab, B. K. 2002. Minimizing energy expenditure facilitates vertebrate persistence on oceanic islands. Ecol. Lett. 5: 693 704. Meehan, H. J. et al. 2002. Potential disruptions to seed dispersal mutualisms in Tonga, western Polynesia. J. Biogeogr. 29: 695 712. Meehan, T. D. 2006. Energy use and animal abundance in litter and soil communities. Ecology 87: 1650 1658. Meiri, S. et al. 2008. The island rule: made to be broken? Proc. R. Soc. B 275: 141 148. Nagy, K. 2005. Field metabolic rate and body size. J. Exp. Biol. 208: 1621 1625. Okie, J. and Brown, J. H. 2009. Niches, body sizes, and the disassembly of mammal communities on the Sunda Shelf islands. Proc. Nat. Acad. Sci. USA 106: 19679 19684. Olson, S. L. and James, H. F. 1991. Descriptions of thirty-two new species of birds from the Hawaiian Islands: parts I and II. Ornithol. Monogr. 45: 1 88.

Pickett, S. T. A. and Thompson, J. N. 1978. Patch dynamics and the design of nature reserves. Biol. Conserv. 13: 27 37. Pimm, S. L. et al. 1994. Bird extinctions in the central Pacific. Phil. Trans. R. Soc. B 344: 27 33. R Foundation for Statistical Computing 2008. R: a language and environment for statistical computing, version 2.8.2. <www.R-project.org>. Raia, P. and Meiri, S. 2006. The island rule in large mammals: paleontology meets ecology. Evolution 60: 1731 1742. Roff, D. A. and Roff, R. J. 2003. Of rats and Maoris: a novel method for the analysis of patterns of extinction in the New Zealand avifauna before European contact. Evol. Ecol. Res. 5: 759 779. Rosenzweig, M. L. 1995. Species diversity in space and time. Cambridge Univ. Press. Savage, V. M. et al. 2004. The predominance of quarter-power scaling in biology. Funct. Ecol. 18: 257 282. Scott, S. N. et al. 2003. Morphological shifts in island-dwelling birds: the roles of generalist foraging and niche expansion. Evolution 57: 2147 2156. Silva, M. et al. 1997. Differences in population density and energy use between birds and mammals: a macroecological perspective. J. Anim. Ecol. 66: 327 340. Simberloff, D. 1978. Using island biogeographic distributions to determine if colonization is stochastic. Am. Nat. 112: 713 726. Slikas, B. et al. 2002. Rapid, independent evolution of flightlessness in four species of Pacific Island rails (Rallidae): an analysis based on mitochondrial sequence data. J. Avian Biol. 33: 5 14. Steadman, D. W. 1995. Prehistoric extinctions of Pacific island birds: biodiversity meets zooarchaeology. Science 267: 1123 1131. Steadman, D. W. 2006. Extinction and biogeography of tropical Pacific birds. Univ. Chicago Press. Sutherland, G. D. et al. 2000. Scaling of natal dispersal in terrestrial birds and mammals. Conserv. Ecol. 4: 16.

Traill, L. W. et al. 2007. Minimum viable population size: a meta-analysis of 30 years of published estimates. Biol. Conserv. 139: 159 166. United Nations Environment Programme 1998. Island directory basic environmental and geographic information on the significant islands of the world. <http://islands.unep.ch/ isldir.htm>, accessed April 2008. Van Vuren, D. 1998. Mammalison dispersal and reserve design. In: Caro, T. (ed.), Behavioral ecology and conservation biology. Oxford Univ. Press, pp. 369 393. White, C. R. et al. 2009. Phylogenetically informed analysis of the allometry of mammalian basal metabolic rate supports neither geometric nor quarter-power scaling. Evolution 63: 2658 2667. Whitehead, D. R. and Jones, C. E. 1969. Small islands and the equilibrium theory of insular biogeography. Evolution 23: 171 179. Whittaker, R. J. and FernaÂ´ndez-Palacios, J. M. 2007. Island biogeography, 2nd ed. Oxford Univ. Press. Whittaker, R. J. et al. 2008. A general dynamic theory of oceanic island biogeography. J. Biogeogr. 35: 977 994. Williamson, M. H. 1981. Island populations. Oxford Univ. Press. Woodward, F. I. et al. 1995. A global land primary productivity and phytogeography model. Global Biogeochem. Cycles 9: 471 490. Worthy, T. H. 2001. A giant flightless pigeon gen. et sp. nov. and a new species of Ducula (Aves: Columbidae), from Quaternary deposits in Fiji. J. R. Soc. N. Z. 31: 763 794. Worthy, T. H. and Holdaway, R. N. 2002. The lost world of the Moa. Indiana Univ. Press. Wright, D. H. 1983. Species energy theory: an extension of species-area theory. Oikos 41: 496 506. Wright, S. J. 1980. Density compensation in island avifaunas. Oecologia 45: 385 389.

Download the Supplementary material as file E6315 from <www.oikos.ekol.lu.se/appendix>.

IBS SPECIAL ISSUE

379

Ecography 33: 380 391, 2010 doi: 10.1111/j.1600-0587.2010.06273.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Jeremy T. Kerr. Accepted 6 March 2010

Determinants of palm species distributions across Africa: the relative roles of climate, non-climatic environmental factors, and spatial constraints Anne Blach-Overgaard, Jens-Christian Svenning, John Dransfield, Michelle Greve and Henrik Balslev A. Blach-Overgaard (anne.overgaard@biology.au.dk), J.-C. Svenning, M. Greve and H. Balslev, The Ecoinformatics and Biodiversity Group, Dept of Biological Sciences, Aarhus Univ., Ny Munkegade 114, DK-8000 Aarhus C, Denmark. J. Dransfield, Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3AB, UK.

IBS SPECIAL

ISSUE

Most of the Earthâ&#x20AC;&#x2122;s biodiversity resides in the tropics. However, a comprehensive understanding of which factors control range limits of tropical species is still lacking. Climate is often thought to be the predominant range-determining mechanism at large spatial scales. Alternatively, speciesâ&#x20AC;&#x2122; ranges may be controlled by soil or other environmental factors, or by non-environmental factors such as biotic interactions, dispersal barriers, intrinsic population dynamics, or timelimited expansion from place of origin or past refugia. How species ranges are controlled is of key importance for predicting their responses to future global change. Here, we use a novel implementation of species distribution modelling (SDM) to assess the degree to which African continental-scale species distributions in a keystone tropical group, the palms (Arecaceae), are controlled by climate, non-climatic environmental factors, or non-environmental spatial constraints. A comprehensive data set on African palm species occurrences was assembled and analysed using the SDM algorithm Maxent in combination with climatic and non-climatic environmental predictors (habitat, human impact), as well as spatial eigenvector mapping (spatial filters). The best performing models always included spatial filters, suggesting that palm species distributions are always to some extent limited by non-environmental constraints. Models which included climate provided significantly better predictions than models that included only non-climatic environmental predictors, the latter having no discernible effect beyond the climatic control. Hence, at the continental scale, climate constitutes the only strong environmental control of palm species distributions in Africa. With regard to the most important climatic predictors of African palm distributions, water-related factors were most important for 25 of the 29 species analysed. The strong response of palm distributions to climate in combination with the importance of nonenvironmental spatial constraints suggests that African palms will be sensitive to future climate changes, but that their ability to track suitable climatic conditions will be spatially constrained.

It is well known that the tropics harbour a major fraction of the Worldâ&#x20AC;&#x2122;s biodiversity; yet, little is known regarding the drivers of tropical species distributions (Gentry 1988). In fact, while understanding the controls of geographic distributions of species is a central issue in ecology and biogeography (Gaston 2009a, b), we still do not have a thorough understanding of the limiting factors for the distribution of any species, despite decades of research (Gaston 2009b). While substantial theory on species distributions exists, it needs better validation from empirical work (Gaston 2009a). In particular, a better understanding of non-equilibrium range dynamics is needed (Holt et al. 2005), as this will have crucial importance for robust conservation planning in the face of global change (Midgley and Thuiller 2005). Notably, tropical biodiversity is foreseen to be critically threatened not only by climate change, but also by land-use changes and contingent habitat loss 380

and fragmentation (Bradshaw et al. 2009). At the same time, the distributions of tropical species are particularly poorly documented (Collen et al. 2008). Hence, there is a special need for understanding the determinants of the current distributions of tropical species. Species distribution studies have primarily focused on the role of the environment, with climate often being assumed to be the main range-limiting factor (Gaston 2003), especially at large spatial scales (Pearson and Dawson 2003). Tropical regions experience little intra-annual climatic variability, particularly in terms of temperature. As a result, tropical species may have evolved narrow climatic tolerances; hence, steep climatic gradients have been hypothesised to provide stronger barriers to range expansion in the tropics than elsewhere (Janzen 1967). Consistent with this hypothesis, elevational range size has been found to increase with latitude in most vertebrate

381

ISSUE

Here, we assess the controls of the African continentalscale species distributions of a keystone tropical organism group, the palms (Arecaceae). Palms constitute important components of tropical and subtropical ecosystems across the World, with only few species occurring in warmtemperate regions (Dransfield et al. 2008). Both cultivated and wild palms are a much utilised resource in rural communities throughout the tropics (Dransfield et al. 2008). The species richness of palms is highest in tropical Asia ( 1200 species) and the Americas (730 species), while only 65 species occur in Africa (Dransfield et al. 2008). Wild palms, nevertheless, constitute a keystone resource for African rural communities (Cunningham and Milton 1987, Dransfield 1988), as well as for wildlife (Yamakoshi 1998). The majority of African palms inhabit the humid rain forests, or low-lying swamplands, while only rather few species are associated with dry, open habitats (e.g. savannas or even deserts), and then often in the riparian zone or where the water table is locally high (Dransfield 1988, Tuley 1995). Recently, it has been found that climate and soil constitute important controls of continental- and regional-scale palm species distributions in the New World, although purely spatial constraints are generally of similar or greater importance (Bjorholm et al. 2008). However, the continental-scale controls of African palm distributions have not previously been quantitatively investigated. The smaller-scale distributions of New World palms have been related to topography, hydrology, human impact and dispersal, with the relative importance of these factors varying among studies (Clark et al. 1995, Svenning 1999, Normand et al. 2006, Svenning et al. 2006). Few analogous studies have been made in Africa, but there is evidence that similar factors influence the local distribution of palms here (Mwaura and Kaburu 2009). Here we used a novel implementation of species distribution modelling (SDM) to estimate the degree to which the continental-scale distributions of African palm species are controlled by 1) climate, 2) non-climatic environmental factors such as habitat and human impact, or alternatively, 3) non-environmental spatial constraints, potentially reflecting the effects of biotic interactions and/or dispersal limitations. Distributions were analysed by the SDM algorithm Maxent in combination with climatic and non-climatic environmental predictors (habitat, human impact), as well as spatial eigenvector mapping (Borcard and Legendre 2002, Griffith 2003). The inclusion of eigenvectors (spatial filters, Griffith 2003) in SDM has recently been shown to effectively capture non-environmental constraints caused by dispersal-limited non-equilibrium range dynamics (De Marco et al. 2008). Our specific study questions concerning the range-limiting factors for the African palm species were: 1) what is the relative importance of environmental factors and non-environmental spatial constraints? 2) What is the relative importance of climatic and non-climatic environmental factors? 3) What is the relative importance of water- vs. temperature-related factors as climatic controls? By providing answers to these questions, we contribute to an improved quantitative basis for understanding what determines species distributions as well as for predicting the extent to which keystone tropical

IBS SPECIAL

groups investigated (McCain 2009). In contrast to climate, non-climatic environmental factors such as topography, land use and soil type are thought to increase in importance at increasingly finer scales (Pearson and Dawson 2003). Numerous studies have shown that topographic and edaphic factors often limit the distribution of tropical species at local to landscape scales (Svenning 1999, Jones et al. 2006). Although there are indications that nonclimatic environmental factors may also be important range determinants at larger scales (Tuomisto 2007, Buermann et al. 2008), the relative importance of climatic and nonclimatic environmental factors for determining the distributions of tropical species at continental scales is not well understood. Whilst it is often argued or assumed that species range limits are determined by environmental constraints (McInnes et al. 2009), range limits may also be caused by non-environmental constraints, notably biotic interactions and dispersal limitations (Soberon 2007, Gaston 2009a), or a combination of both (Tuomisto et al. 2003). A species may be excluded from a region due to the presence of a competitor, by the absence of a mutualist or prey species, or by more complex biotic interactions (Case et al. 2005). Biotic interactions have been proposed to be particularly strong in the tropics, where they may limit species distributions more strongly than at higher latitudes (Brown et al. 1996). More generally, there is some evidence that speciesâ&#x20AC;&#x2122; low-latitude range limits are more often determined by biotic interactions than their high-latitude range limits (Normand et al. 2009). Alternatively, species distributions may simply be spatially constrained by limited dispersal (Gaston 2009a). It is widely accepted that species distributions can be constrained by major dispersal barriers (e.g. oceans), while the extent to which dispersal limits distributions at smaller scales is contentious (Gaston 2009b). Nevertheless, there is increasing evidence that dispersal can indeed be an important constraint on species distributions within continents and smaller regions (Svenning and Skov 2004, Normand et al. 2006, Munguia et al. 2008). Dispersal limitation may reflect physical dispersal barriers such as mountain ranges within regions (Brown et al. 1996, Gaston 2003), or time-limited expansion from place of origin or refugia (Svenning and Skov 2004, Paul et al. 2009). Ranges may also simply fail to expand due to intrinsic population dynamics: e.g. Allee effects (growthrate depressions at low population densities) may potentially restrict species range expansions even in the absence of external drivers (Holt et al. 2005). The extent to which species distributions are limited by biotic interactions or dispersal have strong implications for nature conservation, especially in relation to climate change (Thomas et al. 2004); hereunder for predictive distribution modelling, which generally relies on speciesâ&#x20AC;&#x2122; distributions being at least close to equilibrium with the contemporary abiotic environment (Guisan and Thuiller 2005). If these assumptions are violated, model calibration may be inaccurate, potentially jeopardising the predictive ability of models, e.g. in relation to the consequences of future climate change (Guisan and Thuiller 2005).

organism groups, such as the palms, will be sensitive to future climate changes.

Materials and methods Palm data

Predictor variables were compiled to represent potential climatic and non-climatic range controls, dividing the latter into three subgroups (habitat, human impact, and spatial constraints). Variable selection was initially based on correlation tests using Pearson’s correlations and one-way ANOVA tests to minimise potential collinearity issues (for further explanation, see Supplementary material Appendix S1 and Table S2). We used seven climatic variables commonly used in species distribution modelling to represent the climatic controls, hereof three variables representing temperature (annual mean temperature, AMT; temperature seasonality, TSEA; minimum temperature of the coldest month, TMIN) and four variables primarily representing water availability (annual precipitation, PREC; precipitation seasonality, PSEA; precipitation of driest quarter, PDRY; water balance, WATBAL, computed as the annual sum of monthly differences between potential evapotranspiration and precipitation). These variables were all from the Worldclim data set (Hijmans et al. 2005), except WATBAL, which was computed following Skov and Svenning (2004) using the CRU CL 2.0 data set (New et al. 2002) (Supplementary material Fig. S1). The non-climatic factors were represented by five habitat, two human impact and fourteen spatial constraint variables. The habitat variables to represent the environmental variability of the area were: slope (SLOPE), derived from the U.S. Geological Survey GTOPO30 digital elevation model, and soil type (SOIL) from the Harmonised World Soil Database, which for Africa is derived from regional SOTER (soil and terrain) studies and the Digital Soil Map of The World at 30ƒ (Fisher et al.

IBS SPECIAL

ISSUE

In general, there is a dearth of high-quality tropical biodiversity data; however, we have assembled a highquality database on African palm species occurrences (n 1920 unique species records, Fig. 1A), which is representative of the distribution of each palm species, and in general well distributed across each species’ range, i.e. with no apparent geographical sampling bias. The occurrence data came primarily from herbarium collections from the Royal Botanic Gardens, Kew; the Nationaal Herbarium Nederland; the Missouri Botanical Garden and herbarium collections accessed through the Global Biodiversity Information Facility (GBIF) data-portal (Botanic Garden and Botanical Museum Berlin-Dahlem; Herbarium of the Aarhus Univ.; Museum National d’Histoire Naturelle; European Environment Agency; Fairchild Tropical Botanic Garden; National Herbarium of New South Wales; Royal Museum of Central Africa). Additional data was obtained through literature surveys, and from private databases and observations. For the present study, we selected all African palm species with ]20 unique georeferenced observations (n 29) (Supplementary material Table S1). The remaining 36 palm species are particularly rare (20) or in general widespread, yet, inadequately sampled (6), or with a questionable taxonomy (10). The selected palm species represent the variation in range sizes and ecological requirements amongst the African palms.

Predictor variables

Figure 1. (A) Georeferenced point localities for all African palm species (n 1920), (B) White’s vegetation map of Africa (White 1983), overlaid with physical barriers for plant migration according to Richards (1973) and Tuley (1995) (a, the Dahomey Gap; b, the Cameroon Range), and the major forest refugia recognised for Africa for the Last Glacial Maximum (LGM): 1, Upper Guinea; 2, Cameroon-Gabon; 3, Congo Basin; and 4, Eastern Congo DRC (Hamilton and Taylor 1991, Maley 1991, Morley 2000).

382

introduce bias in the modelling towards either environment predictors or spatial filters. Hence, as a fixed number of predetermined environmental predictors were used, we also used the same number of spatial filters (first 7 or first 14 filters, depending on the environmental predictor set). These filters may represent relatively broad- to mediumscale spatial patterns (Supplementary material Fig. S4) such as those potentially resulting from major dispersal barriers (e.g. the Dahomey Gap in western Africa; Mayr and Oâ&#x20AC;&#x2122;Hara 1986), dispersal-limited expansions from glacial refugia (Svenning and Skov 2004), other time-limited range expansions (Paul et al. 2009), or other non-environmental range constraints (Holt et al. 2005). Our approach is in line with De Marco et al. (2008) who included the first five spatial filters to explicitly include broad-scale spatial constraints into species distributions modelling. All layers were reprojected to the Lambert Azimuthal Equal Area projection and resampled or aggregated to 1-km grid size using the nearest neighbour or bilinear resampling techniques for categorical and continuous variables respectively. All GIS operations were conducted in ArcGIS 9.2 (ESRI, Redlands, CA, USA). Species distribution modelling

383

ISSUE

Species distribution modelling was implemented using the maximum entropy approach (Maxent) of Phillips et al. (2006) due to its high performance compared to other predictive algorithms in a recent comparative methodological study (Elith et al. 2006). To address the study questions, we ran Maxent with nine models based on different combinations of the predictors for each species. The combinations were: I) a full environmental model (clim hab hum) that included all the climatic, habitat and human impact predictors, II) a combined climate and habitat model (clim hab) that included just the climatic and habitat predictors, III) an environment and filter model (clim hab hum filters) that included all environmental predictors and all spatial filters. Additionally, we tested the performance of IV) a climate and filter model (clim filters) that included the seven climatic predictors, and for balance, just the first seven filters to examine a simpler and potentially more parsimonious model compared to the clim hab hum filters model. The remaining models included single groups of predictors in isolation V) climate (clim), VI) habitat (hab), VII) human impact (hum) variables, or VIII IX) only the spatial filters (seven or fourteen filters, for use in different model comparisons) (Supplementary material Table S3). The default settings for Maxent were used, with the allowed response types (linear, quadratic, product, threshold and hinge functions of the variables) for a species being determined by its number of point localities, as these settings have been shown to provide good predictive performance over a range of datasets (Phillips and Dudik 2008). To assess the relative influence of each predictor variable for a given species in the best performing model, we examined the contribution of each predictor to the final regularised training gain when all variables of the particular model were included in the Maxent run. The climatic variable with the highest gain (highest contribution to the

IBS SPECIAL

2008). Three additional habitat variables were derived from remote sensing: 1) The Global Land Cover 2000 (GLC), assembled from various remote sensing data sources (SPOT vegetation, JERS and ERS radar, and The Defence Meteorological Satellite Program data) at 1-km spatial resolution (Mayaux et al. 2004); 2) Generalised Global Vegetation Index (GVI), obtained from Advanced Very High Resolution Radiometer (AVHRR) data as a series of monthly mean values for 1985 1988 (Kineman and Hastings 1992), which were averaged to derive an annual mean at 10? resolution; 3) The Vegetation Continuous Field product (VCF) derived from Moderate Resolution Imaging Spectroradiometer (MODIS) data which provides a measure of percent tree cover for the year 2001 at 500-m resolution (Hansen et al. 2002) (Supplementary material Fig. S2). Current anthropogenic factors are also believed to control species ranges; hence, human impact was represented by: 1) human population density (POP), obtained from population statistics at 2.5? resolution for the year 2000 (CIESIN and CIAT 2005); 2) Human Influence Index (HMNINFL), consisting of values representing low (0) to high (64) human influence, and estimated for each grid cell based on several proxy data layers for human influence such as population density, land transformation, accessibility from road, rivers and coastlines, and electric power infrastructure (Sanderson et al. 2002) (Supplementary material Fig. S3). To account for spatial constraints on the palm distributions, hereunder dispersal limitation-generated patterns, we used spatial filters (eigenvectors) (Griffith 2003) as spatial constraint variables. Spatial filters are orthogonal variables (Supplementary material Table S2), which represent the spatial relationship amongst spatial units (here: grid cells), and capture the geometry of the study area at a range of scales (Diniz-Filho and Bini 2005). The spatial filters were computed in SAM 3.0 (Rangel et al. 2006) by constructing a pairwise distance matrix amongst all grid cells using their geographical coordinates (latitude and longitude). The distance matrix was subsequently truncated at a given distance (default settings in SAM 3.0), and from this modified distance matrix 991 spatial filters were computed using principal coordinate analysis (Borcard and Legendre 2002). Only positive filters were retained. Due to computational limitations the filters were computed at a coarse resolution (100 100-km), but subsequently interpolated to 1 1-km resolution using the Inverse Weighted Distance method in GIS. This method was applied due to its inherent property that if point x0 (point to be interpolated) coincide with point xi (known value), the interpolation z(x0) always takes on the value of z(xi) (Webster and Oliver 2007). Griffith (2003) has proposed several strategies for selecting suitable spatial filters for regression analyses including: 1) maximisation of R2 by using the stepwise regression procedure, 2) minimisation of regression residual spatial autocorrelation to obtain independent variable components, and 3) a significant correlation between the response and each spatial filter to use only meaningful eigenvectors. However, none of these strategies seem applicable in this particular case, where a key aim was to assess the relative importance of environmental factors and non-environmental spatial constraints for species distributions. In this case it was of paramount importance not to

prediction for each species) was subsequently used in isolation in a model for all the point localities to also estimate its overall contribution to the prediction for a given species without the influence of the other predictors. This was done to ensure that our findings were not affected by the remaining collinearity among the predictors. If pairs of variables are highly correlated, the importance of one may be wrongly diminished by Maxent when both are included in the predictor set (Phillips et al. 2004). Statistical analyses Model performance was assessed by dividing the species occurrence data into random training (80%) and test (20%) datasets, and using 10 000 randomly selected pseudoabsences from the whole study area. Random selection of pseudo-absences has recently been found to outperform selection of pseudo-absences in low suitability areas (Wisz and Guisan 2009). A given model was calibrated on the training data and evaluated on the test data using two threshold-independent assessment measures: the Area Under the receiver operating characteristics Curve (AUC) and the point-biserial correlation (COR). AUC provides a measure of the accuracy of predictive distribution models (Lobo et al. 2008), a value of 0.5 indicating that the model is no better than random, while AUC values ]0.750 are considered in the ‘‘best’’ model category (Phillips and Dudik 2008). However, comparing models across species using AUC scores is problematic as AUC is influenced by species’ prevalence, nonetheless, AUC can safely be applied to evaluate model performance within species (Lobo et al. 2008). Another issue regarding AUC, notably for presenceonly methods, is that in theory the maximum achievable AUC can only be 1 a/2, where a is the fraction of grids covered by a species’ distribution. However, a is often unknown (as in our case), and in practice the maximum AUC computed this way can be exceeded by the test AUC (AUC based on the test data set) (Phillips et al. 2006), although the reasons for this are largely unexplained in the

literature. We minimised these problems by only comparing AUC values among models within species (see below), thereby keeping prevalence constant. The point-biserial correlation (COR) was calculated as Pearson’s correlation between model predictions (suitability scores) and presence (1)/pseudoabsence (0) in the test data set (Elith et al. 2006, Phillips and Dudik 2008). We primarily assessed the relative importance of the different predictor sets by comparing the performance of the various models using Wilcoxon sign rank tests with species as the sampling unit (i.e. a paired test). Two-tailed tests were used for model comparisons between the models based on non-overlapping sets of predictors, while onetailed tests were used for comparisons of nested models. These tests were conducted in JMP 7.0 (SAS Inst., Cary). We additionally addressed the study questions by testing if certain groups of predictor variables contributed the strongest predictor variables for a species more often than expected by chance. This was assessed by classifying the most important variable (variable with the highest contribution to the training gain) for each species in the best performing model based on the Wilcoxon sign rank test (clim filters) into various pairs of mutually exclusive groups (most important variable: environment vs spatial filters; most important climate variable: water vs temperature), and testing the group frequencies against the null expectation based on the number of variables in a group using Goodness of fit- (G-) tests (Sokal and Rohlf 1995).

Results The AUC and COR measures provided highly consistent estimates of model performance (Table 1). Overall both environmental factors and spatial constraints were important controls of palm species distributions across Africa. The clim hab hum filters model had superior predictive ability to the clim hab hum model, and was also significantly better than the filters model according to

IBS SPECIAL

ISSUE

Table 1. Wilcoxon sign rank test comparisons of the predictive ability among species distribution models based on different sets of predictors for 29 African palm species with ]20 unique occurrence records. The tests compare two measures of model predictive ability: the Area Under the receiver operating Curve (AUC) and the point-biserial correlation (COR). Three model comparisons (MC1-3) are shown. Model

AUC

COR

Median

[min-max]

Median

[min max]

MC1

Clim hab hum filters Clim hab hum Filters

0.984a(a) 0.975b 0.979ab

0.749 1.000 0.608 0.999 0.897 1.000

0.181a(a) 0.156b 0.151b

0.022 0.433 0.022 0.346 0.044 0.408

MC2

Clim filters Clim Filters

0.983a(a) 0.976b 0.976b

0.855 1.000 0.759 1.000 0.858 1.000

0.153a(a) 0.135b 0.139b

0.050 0.416 0.037 0.330 0.039 0.390

MC3

Clim hab hum Clim Clim hab Hab Hum

0.975a 0.976a 0.974a 0.931b 0.805c

0.608 0.999 0.759 1.000 0.608 1.000 0.630 0.998 0.518 0.959

0.156a 0.135ab 0.146b 0.101c 0.035d

0.022 0.346 0.037 0.330 0.018 0.354 0.009 0.250 0.001 0.084

Clim hab hum, (all environmental layers: climate, habitat, human impact); Filters, first 14 (MC1) or 7 (MC2) spatial filters; Clim, climate; Clim hab, climate habitat; Hum, human impact. Different superscript letters (a d) indicate models which were significantly different (only tested within groups, MC1-3). All significantly different models differed at p B0.0001 except for clim filters vs filters at p 0.0057 and clim hab hum vs hab at p 0.0023 (in the AUC test), and clim hab hum filters vs filters at p 0.0021 and clim hab hum vs clim hab at p 0.0182 (in the COR test). The superscript (a) indicates no significant difference between the clim hab hum filters and clim filters models in AUC and COR.

384

COR (but not AUC; Table 1). Furthermore, the clim filters model provided superior predictive ability to the clim and filters models, while the latter two models performed equally well (Table 1). These results indicate that palm distributions are not only limited by environmental factors, but also by spatial constraints, as was also apparent from the mapped predicted distributions for most species (Fig. 2, Supplementary material Fig. S5, S6). However, despite the fact that filters were an important component in structuring African palm distributions, the most important range predictor for a given species was significantly far more often a climatic variable (25 spp.) than a spatial filter (4 spp.; Table 2). Among the environmental factors, only climate was important in determining distributions. There were no significant differences between the clim model, with only climatic predictors, and the more complex environmental models (clim hab hum, clim hab) (Table 1). The clim model and the two more complex models including climatic variables were all superior to the non-climatic hum and hab models (Table 1). Hence, the simplest of these three climate models, the clim model with just seven climate predictors, was sufficient to account for the environmental control of the palm species ranges, with no significant additional effect of the non-climatic environmental factors. Supporting this conclusion, the clim hab hum filters model also did not exhibit significantly better performance than the clim filters model (Table 1). As climate and spatial filters were sufficient to optimally predict palm species distributions across Africa; we assessed the relative importance of water- vs temperature-related factors using the clim filters model. Water-related variables strongly predominated over temperature-related variables as the most influential individual predictor (Table 2, Fig. 3A F). In terms of the specific species responses to climate, most of the species showed a positive response to increasing precipitation or water balance (Fig. 3A, D), with many speciesâ&#x20AC;&#x2122; responses peaking at approximately 2000 3000 mm (Fig. 3B, C), i.e. tropical rain forest climate. Just a few palm species were associated with low precipitation regimes and negative water balance (Fig. 3C).

385

ISSUE

Our analyses showed that the continental-scale distributions of the African palms are not only limited by environmental controls, but also by spatial constraints (Table 1). Of the environmental controls, climate exhibited the greatest influence, while habitat or human impact had negligible impact on the distributions (Table 1). In terms of climate, the African palms were more dependent on water-related variables than temperature (Table 2). We found that environmental factors (just climatic factors, in fact) had a strong effect on the African palm distributions. Hereby our results agree with most current thinking which emphasises climate as the main range determinant at large spatial scales (Pearson and Dawson 2003). Other studies have also shown that palm species distributions are sensitive to climate at multiple spatial scales (Salm 2007, Walther et al. 2007, Bjorholm et al. 2008). In addition, the paleo-record documents that past

IBS SPECIAL

Discussion

climate change has had profound effects on the diversity and distribution of African palms (Morley 2000, Pan et al. 2006). Nevertheless, by including spatial filters in the modelling (De Marco et al. 2008) we found evidence that African palm species distributions are also often strongly spatially constrained beyond the effects of environmental control, i.e. in contrast to the equilibrium postulate (the assumption that species are in equilibrium with the contemporary abiotic environment; Guisan and Thuiller 2005). Notably, considering the predicted distributions, the clim model clearly overpredicted the distributions for most species relative to the clim filters model (Fig. 2, Supplementary material Fig. S5). We cannot firmly establish the mechanisms behind the purely spatial range constraints, but in many cases they are consistent with dispersal limitation in relation to dispersal barriers or glacial refugia. For example, the Mediterranean Chamaerops humilis (Tuley 1995) was predicted to also occur widely across southern Africa by the clim model, while the inclusion of spatial filters (clim filters) removed this overprediction (Fig. 2). The effect of spatial filters, in this case, probably reflects dispersal limitation caused by the broad band of unsuitable non-Mediterranean climate between northern and southern Africa, hindering the southwards dispersal of this species. We note that numerous plant species that are native to either northern or southern Africa have become naturalised or even invasive after introduction to the other region (Fox 1990). The overpredictions of the clim model in comparison to the predictions of the clim filters model for several other species could be explained by geographic features that have been proposed as dispersal barriers in the phytogeographical literature (Richards 1973, Tuley 1995). Podococcus barteri, Eremospatha wendlandiana, Laccosperma robustum and Raphia palma-pinus all fail to fill both parts of their disjunct suitable areas in the Guineo-Congolian phytoregion (White 1983), in contrast to many other forest palm species of this region (Fig. 2, Supplementary material Fig. S5). The suitability disjunction corresponds to an intrusion of savanna in the distribution of tropical lowland rain forest in Ghana, Togo and Benin (Fig. 1B), known as the Dahomey Gap (Mayr and Oâ&#x20AC;&#x2122;Hara 1986). The Dahomey Gap likely arose in the Holocene (cf. Maley 1991), but was also present during dry Pleistocene glacials (Dupont and Weinelt 1996), and has been suggested as a barrier for the dispersal for many other organisms in the region (Richards 1973). Another proposed dispersal barrier is the Cameroon Range (Fig. 1B), which forms a divide between the coastal West African and Congolian floras (Tuley 1995), and appears to constitute a dispersal barrier to some palm species (notably, Eremospatha cabrae and Eremospatha cuspidata) according to our results (Supplementary material Fig. S5). The discontinuous distribution patterns of many forest plants and animals in Africa are thought to reflect retractions to forest refugia during cold and dry glacial periods of the Pleistocene in combination with limited postglacial range expansion (Mayr and Oâ&#x20AC;&#x2122;Hara 1986, also cf. Hamilton and Taylor 1991). The predicted distributions by the clim filters and clim hab hum filters models (Fig. 2, Supplementary material Fig. S5, S6) for some species (E. wendlandiana, Laccosperma acutiflorum,

ISSUE

IBS SPECIAL Figure 2. Presence localities and predicted distributions of five selected African palm species based on Maxent models fitted to all the presence localities per species using only climate predictors (clim), climate predictors and spatial filters (clim filters), or just seven spatial filters (filters). AUC scores are displayed in the upper right-hand corner of each predicted distribution map. Corresponding maps for the remaining 24 palm species are provided in Fig. S5 in the Supplementary material.

386

Table 2. Goodness of fit- (G-) tests for whether the observed frequency of the most important predictor variable for each species (based on the best performing model (clim filters)) across two pairs of two mutually exclusive predictor classes deviates from random expectation (expected frequency n [number of predictor variable in the class/number of predictor variables in the two classes combined], n 29 species). The number of predictor variables in a class is given in parentheses. The G-tests were implemented using Williamsâ&#x20AC;&#x2122; correction for the two-cell case (Gadj). Classes Environment (7)$ Filters (7) Water balance (4) Temperature (3)

Observed

Expected

25 4 25 4

14.5 14.5 16.6 12.4

Gadj 16.65*** 11.30***

$, just the seven climatic variables, given that the analyses were based on the clim filters model; ***, p B0.001.

ISSUE

387

IBS SPECIAL

Laccosperma laeve, Oncocalamus mannii, Oncocalamus tuleyi, Podococcus acaulis and P. barteri) coincided broadly with recognised glacial refugia for rain forest in Africa. Despite some disagreement in the literature, there is a general consensus that the following four regions acted as refugia: Upper Guinea, Cameroon-Gabon, the Congo Basin, and Eastern Congo DRC (Fig. 1B) (Hamilton and Taylor 1991, Maley 1991, Morley 2000). The distributions of E. wendlandiana, P. acaulis, P. barteri, O. mannii and O. tuleyi are restricted to the vicinity of the Cameroon-Gabon refuge and L. acutiflorum and L. laeve to both the Upper Guinea and Cameroon-Gabon refuges despite suitable areas occurring much more widely. Similar refuge-restricted patterns are also known from other regions and taxa, e.g. European trees (Svenning and Skov 2007, Svenning et al. 2008). The extent to which a species fills its full potential distribution (range-filling, Svenning and Skov 2004) has for other plant groups been shown to be related to speciesâ&#x20AC;&#x2122; dispersal ability or time for dispersal (Schurr et al. 2007, Paul et al. 2009). For African palms, such assessments have not yet been conducted, but palms are generally thought to be poor dispersers (Dransfield 1981). A number of studies have found evidence that New World palms are dispersal limited at both continental (Bjorholm et al. 2008) and smaller scales (Charles-Dominique et al. 2003, Normand et al. 2006, Svenning et al. 2006). Palms are mainly dispersed by mammals or birds, but there is substantial variability among palm species in the types of dispersers within these groups (Zona and Henderson 1989), and thereby potentially also in dispersal ability. As spatial filters simply capture spatial range constraints of any nature, the evidence for constraints beyond those caused by environmental limitation could also reflect mechanisms other than dispersal, e.g. biotic interactions. However, we found little evidence that biotic interactions limit the African palm species distributions at large scale. Notably, even within the larger genera (Hyphaene, Eremospatha and Laccosperma), many, and often the majority, of the species tend to cooccur at least sometimes in the same regions. The most obvious potential exception would be Hyphaene petersiana and Hyphaene thebaica (Fig. 2, Supplementary material Fig. S5, S6). We found that climatic factors were important determinants of African palm species distributions, as already

discussed, while the non-climatic environmental factors had no discernable influence. These results agree with the a priori notion that climate is the main environmental control of species ranges at large scales, while non-climatic environmental factors such as human impact and soil type mainly are important at smaller scales (Pearson and Dawson 2003). At local to landscape scales, habitat and human impact are important determinants of palm species distributions in the New World (Clark et al. 1995, Svenning 1999, Normand et al. 2006, Svenning et al. 2006). There is no reason to think that African palm species distributions would not be similarly influenced by these factors, and some evidence for this does indeed exist (Cunningham and Milton 1987, Tuley 1995). However, at the continental scale the importance of these variables is not strong enough to be apparent with the data presently available. Yet, we note that soil, notably the response to soils associated with good water retention and high water tables, appeared as an important predictor for dry-climate palms in the hab, clim hab and clim hab hum models (results not shown) indicating that hydrology is an important aspect for these palms. Especially for Hyphaene petersiana does soil constitute a stronger predictor than any climatic variables, according to a recent study of this species (Blach-Overgaard et al. 2009). However, these relationships are not frequent enough to affect the overall model comparisons, where the simple climatic model performed as well as models including both climatic and non-climatic environmental predictors. The limited overall importance of habitat factors such as soil is likely caused by the 1-km spatial resolution of the present study, which may cause much of the variation in habitat and human influence to be within grid cells. In terms of the climatic predictors, water-related variables were found to strongly predominate over temperature variables as range-controlling factors for the African palms (Table 2). The superior importance of water-related factors is consistent with the idea that water availability has greater importance than temperature in subtropical and tropical zones for controlling patterns of species diversity (Hawkins et al. 2003). Moreover, water is in general of greatest importance for palm occurrences worldwide (Dransfield et al. 2008); even dry-climate palms are dependent on locally high water tables to persist in dry environments (Tuley 1995). Similarly to our findings, Salm et al. (2007) found that a water-related climatic variable was the main predictor of palm species distributions and species richness in Brazil, whereas temperature was only of secondary importance. We note that our findings on the relative importance of the various factors will most likely be specific to our study group, the African palms, and further studies will be needed to assess to what extent they can be generalised to other species groups with other geographies, ecologies, and evolutionary histories. Our variable selection approach limited collinearity problems. Nevertheless, there were fairly high correlations among some of the predictor variables (Supplementary material Table S2). These correlations had no overall influence on our major conclusions. Maxent may only use one of the variables in a pair of highly correlated variables, and thus has an ability to deal with such cases; however overall model performance will not be affected (Phillips et al. 2004). More importantly, in our case, the only highly

ISSUE

IBS SPECIAL

Figure 3. Response curves to the most important climatic predictors for 28 African palm species. For each species, the response curve for the climatic predictor with the highest relative contribution in the clim filters Maxent model is shown. For 25 species this climatic variable had the highest relative contribution among all 14 predictors, while for three species (Eremospatha haullevilleana, Hyphaene petersiana and Raphia sudanica) it ranked second in importance after a spatial filter. The response curves were estimated by Maxent modelling based on all presence localities per species using just the selected climatic predictor. For the 29th species, Hyphaene thebaica, precipitation seasonality was the second-most important predictor after a filter (response curve not shown).

correlated predictors were WATBAL and PREC (Supplementary material Table S2), which both belong to the same predictor category (water-related climate variables); ensuring that any collinearity effects on the relative importance of variables would not be relevant for any of our study questions. More broadly considering our methodology, in the present study we have achieved new insights in the determinants of species distributions in a keystone tropical organism group by applying the novel implementation of SDM developed by De Marco et al. (2008). While our 388

study therefore illustrates the potential of this method, we note that other methodologies also may provide important insights on broad-scale species distribution patterns. Notably, species distributions can also be modelled as whole assemblages, e.g. using constrained ordinations (Svenning and Skov 2005) or using Community Dissimilarity Modelling (Ferrier and Guisan 2006). The use of spatial filters to represent non-environmental spatial constraints on the species distributions is easily implemented in many of these approaches. Several alternative methods for representing spatial structure in species distributions are also

Conclusions

Acknowledgements We thank Jeremy Kerr and two anonymous reviewers for helpful comments and suggestions on previous drafts of this manuscript. This work was financed by the Fac. of Science at Aarhus Univ. to ABO and MG, and the Danish Natural Science Research Council through grants #272-07-0242 to JCS and #272-06-0476 to HB. We thank the Royal Botanic Gardens, Kew (William Baker) for access to the herbariumâ&#x20AC;&#x2122;s palm collections and digitised palm database, and Jan Wieringa and James C. Solomon for providing the electronic palm databases of the Nationaal Herbarium Nederland and Missouri Botanical Garden, respectively. We are also grateful to Terry Sunderland, Ross Bayton, Sebastien Barot, Antje Ahrends and Jon Lovett for providing access to their private databases and information on palm observations across Africa. Finally, we also thank Signe Normand for initial computation of WATBAL and GVI, and Ib Friis for constructive comments on the predicted palm distributions.

References Bjorholm, S. et al. 2008. To what extent does Toblerâ&#x20AC;&#x2122;s 1st law of geography apply to macroecology? A case study using American palms (Arecaceae). BMC Ecol. 8: 11. Blach-Overgaard, A. et al. 2009. Climate change sensitivity of the African ivory nut palm, Hyphaene petersiana Klotzsch ex Mart. (Arecaceae) a keystone species in SE Africa. IOP Conf. Ser.: Earth Environ. Sci. 8: 012014. Boko, M. et al. 2007. Africa. Climate change 2007: impacts, adaptation and vulnerability. In: Canziani, P. M. L. et al. (eds), Contribution of Working Group II to the forth assessment report of the Intergovernmental Panel on Climate Change. Cambridge Univ. Press, pp. 433 467. Borcard, D. and Legendre, P. 2002. All-scale spatial analysis of ecological data by means of principal coordinates of neighbour matrices. Ecol. Model. 153: 51 68. Bradshaw, C. J. A. et al. 2009. Tropical turmoil: a biodiversity tragedy in progress. Front. Ecol. Environ. 7: 79 87. Brown, J. H. et al. 1996. The geographic range: size, shape, boundaries, and internal structure. Annu. Rev. Ecol. Syst. 27: 597 623.

389

ISSUE

The current study implements a novel SDM approach (De Marco et al. 2008) to determine the factors which best determine species distributions by assessing the relative importance of climate, non-climatic environmental factors and non-environmental spatial constraints for the continental-scale distributions of African palms. In contrast to general theory and the typical assumption of many species distribution modelling studies (Pearson and Dawson 2003, Guisan and Thuiller 2005); our results clearly show that

although climate constitutes an important control of African palm distributions, they are also strongly controlled by spatial constraints. We note that the spatial constraints were, to a large extent, consistent with dispersal limitation in relation to physical barriers (often broad zones of unsuitable climate) as well as time-limited expansions from past refugia. Our results highlight the need for SDM approaches that are robust to violations of the equilibrium postulate, at least when the goal is accurate prediction into new temporal or spatial domains. Consistent with theory (Pearson and Dawson 2003), non-climatic environmental constraints such as human impact or soil type had no discernible influence at the large scale of the present study. The climatic control mainly represented sensitivity to the climatic water regime. From a conservation approach, the findings that many African palm species are not only sensitive to the climatic water regime, but also have distributions that are strongly constrained by purely spatial factors are worrying since Africa is predicted to experience changed precipitation patterns with strong drying in many regions during the 21st century (Boko et al. 2007), and thus the ability of the palms to track suitable climatic conditions may be spatially constrained.

IBS SPECIAL

available (Dormann et al. 2007); although unlike spatial filters most cannot be implemented in presence-only based modelling. Since the distributions of African palms are strongly influenced by climate and are often also limited by nonenvironmental spatial constraints, which in many cases are consistent with dispersal limitation effects, African palms are likely to be at risk from future climate change. Continental-based climate assessments show that Africa is likely to experience marked climatic changes over the 21st century with drying and warming in most subtropical regions and slight increases in precipitation in the tropics (Boko et al. 2007). The African palms exhibit varying responses to the climate variables studied (Fig. 3). Hence, we should also expect the response of the African palm species to future climate change to vary among species. Nevertheless, the majority of species have their distributions strongly controlled by the climatic water regime, with most preferring a wet rain forest climate, and must therefore be expected to be sensitive to increasing drying. Climate change will also affect the hydrology, notably in the drier regions, and will therefore constitute a risk for the hydrology-dependent dry-climate palms (Blach-Overgaard et al. 2009). These conclusions are supported by the paleorecord, which links past palm species extinctions in Africa to increasing drought (Morley 2000). When species are exposed to major climate changes, they are forced to either adapt to the changed conditions, or to track a suitable climate by migration. However, the often strong nonenvironmental spatial constraints on the African palm species distributions suggest that the scope for large-scale range shifts to track the changing climate will be limited. Hereby, dispersal limitation will probably aggravate the risk posed by climate change to African palm species. The risks to the African palm flora are likely to be exacerbated by the same factors which cause Africa to be listed as the continent that is most vulnerable to future climate change, namely the additional stresses that plague the region (e.g. poverty, political instability, disasters) (Boko et al. 2007). For example, the capacity for evolutionary adaptation will be limited in populations already reduced by habitat degradation and direct over-utilisation (Cunningham and Milton 1987, Dransfield et al. 2008), while migration rates will be reduced in strongly fragmented landscapes. As the African palms also provide important resources to rural communities and wildlife in Africa (Dransfield 1988, Yamakoshi 1998), declining palm populations would exacerbate the climate-change induced problems for society and ecosystems in the region.

ISSUE

IBS SPECIAL

Buermann, W. et al. 2008. Predicting species distributions across the Amazonian and Andean regions using remote sensing data. J. Biogeogr. 35: 1160 1176. Case, T. J. et al. 2005. The community context of species’ borders: ecological and evolutionary perspectives. Oikos 108: 28 46. Charles-Dominique, P. et al. 2003. Colonization front of the understorey palm Astrocaryum sciophilum in a pristine rain forest of French Guiana. Global Ecol. Biogeogr. 12: 237 248. CIESIN and CIAT 2005. Gridded population of the World version 3 (GPWv3): population density grids. Socioeconomic data and applications center (SEDAC), Palisades, NY, Columbia Univ. Press. Clark, D. A. et al. 1995. Edaphic and human effects on landscapescale distributions of tropical rain forest palms. Ecology 76: 2581 2594. Collen, B. et al. 2008. The tropical biodiversity data gap: addressing disparity in global monitoring. Trop. Conserv. Sci. 1: 75 88. Cunningham, A. B. and Milton, S. J. 1987. Effects of basketweaving industry on Mokola Palm and dye plants in northwestern Botswana. Econ. Bot. 41: 386 402. De Marco, P. et al. 2008. Spatial analysis improves species distribution modelling during range expansion. Biol. Lett. 4: 577 580. Diniz-Filho, J. A. F. and Bini, L. M. 2005. Modelling geographical patterns in species richness using eigenvectorbased spatial filters. Global Ecol. Biogeogr. 14: 177 185. Dormann, C. F. et al. 2007. Methods to account for spatial autocorrelation in the analysis of species distributional data: a review. Ecography 30: 609 628. Dransfield, J. 1981. Palms and Wallace’s line. In: Whitmore, T. C. (ed.), Wallace’s line and plate tectonics. Clarendon Press, pp. 43 56. Dransfield, J. 1988. The palms of Africa and their relationships. In: Goldblatt, P. and Lowry, P. P. (eds), Modern systematic studies in African botany. Missouri Botanical Garden Press, pp. 95 103. Dransfield, J. et al. 2008. Genera Palmarum. The evolution and classification of palms. Royal Botanic Gardens, Kew. Dupont, L. M. and Weinelt, M. 1996. Vegetation history of the savanna corridor between the Guinean and the Congolian rain forest during the last 150 000 years. Veg. Hist. Archaeobot. 5: 273 292. Elith, J. et al. 2006. Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29: 129 151. Ferrier, S. and Guisan, A. 2006. Spatial modelling of biodiversity at the community level. J. Appl. Ecol. 43: 393 404. Fisher, G. et al. 2008. Global agro-ecological zones assessment for agriculture (GAEZ 2008). IIASA, Laxenburg, Austria and FAO, Rome, Italy. Fox, M. D. 1990. Mediterranean weeds: exchanges of invasive plants between the five Mediterranean regions of the world. In: Di Castri, F. et al. (eds), Biological invasions in Europe and the Mediterranean Basin. Kluwer, pp. 179 200. Gaston, K. J. 2003. The structure and dynamics of geographic ranges. Oxford Univ. Press. Gaston, K. J. 2009a. Geographic range limits of species. Proc. R. Soc. B 276: 1391 1393. Gaston, K. J. 2009b. Geographic range limits: achieving synthesis. Proc. R. Soc. B 276: 1395 1406. Gentry, A. H. 1988. Changes in plant community diversity and floristic composition on environmental and geographical gradients. Ann. Missouri Bot. Gard. 75: 1 34. Griffith, D. A. 2003. Spatial autocorrelation and spatial filtering: gaining understanding through theory and scientific visualization. Springer.

390

Guisan, A. and Thuiller, W. 2005. Predicting species distribution: offering more than simple habitat models. Ecol. Lett. 8: 993 1009. Hamilton, A. C. and Taylor, D. 1991. History of climate and forests in tropical Africa during the last 8 million years. Clim. Change 19: 65 78. Hansen, M. C. et al. 2002. Towards an operational MODIS continuous field of percent tree cover algorithm: examples using AVHRR and MODIS data. Remote Sens. Environ. 83: 303 319. Hawkins, B. A. et al. 2003. Energy, water, and broad-scale geographic patterns of species richness. Ecology 84: 3105 3117. Hijmans, R. J. et al. 2005. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25: 1965 1978. Holt, R. D. et al. 2005. Theoretical models of species’ borders: single species approaches. Oikos 108: 18 27. Janzen, D. H. 1967. Why mountain passes are higher in the tropics. Am. Nat. 101: 233 249. Jones, M. M. et al. 2006. Effects of mesoscale environmental heterogeneity and dispersal limitation on floristic variation in rain forest ferns. J. Ecol. 94: 181 195. Kineman, J. and Hastings, D. 1992. Monthly generalized global vegetation index from NESDIS NOAA-9 weekly GVI data (APR 1985 DEC 1988). Digital raster data on a 10-minute Cartesian Orthonormal Geodetic (lat/long) 1080 2160 grid (Platte Carree projection). Global Ecosystems Database Ver. 2.0. NOAA National Geophysical Data Center, Boulder, CO. Lobo, J. M. et al. 2008. AUC: a misleading measure of the performance of predictive distribution models. Global Ecol. Biogeogr. 17: 145 151. Maley, J. 1991. The African rain forest vegetation and paleoenvironments during Late Quaternary. Clim. Change 19: 79 98. Mayaux, P. et al. 2004. A new land-cover map of Africa for the year 2000. J. Biogeogr. 31: 861 877. Mayr, E. and O’Hara, R. J. 1986. The biogeographic evidence supporting the Pleistocene forest refuge hypothesis. Evolution 40: 55 67. McCain, C. M. 2009. Vertebrate range sizes indicate that mountains may be ‘higher’ in the tropics. Ecol. Lett. 12: 550 560. McInnes, L. et al. 2009. Where do species’ geographic ranges stop and why? Landscape impermeability and the Afrotropical avifauna. Proc. R. Soc. B 276: 3063 3070. Midgley, G. F. and Thuiller, W. 2005. Global environmental change and the uncertain fate of biodiversity. New Phytol. 167: 638 641. Morley, R. J. 2000. Origin and evolution of tropical rain forests. Wiley. Munguia, M. et al. 2008. Dispersal limitation and geographical distributions of mammal species. J. Biogeogr. 35: 1879 1887. Mwaura, F. and Kaburu, H. M. 2009. Spatial variability in woody species richness along altitudinal gradient in a lowland-dryland site, Lokapel Turkana, Kenya. Biodivers. Conserv. 18: 19 32. New, M. et al. 2002. A high-resolution data set of surface climate over global land areas. Clim. Res. 21: 1 25. Normand, S. et al. 2006. Geographical and environmental controls of palm beta diversity in paleo-riverine terrace forests in Amazonian Peru. Plant Ecol. 186: 161 176. Normand, S. et al. 2009. Importance of abiotic stress as a rangelimit determinant for European plants: insights from species responses to climatic gradients. Global Ecol. Biogeogr. 18: 437 449.

Pan, A. D. et al. 2006. The fossil history of palms (Arecaceae) in Africa and new records from the Late Oligocene (28-27 Mya) of north-western Ethiopia. Bot. J. Linn. Soc. 151: 69 81. Paul, J. R. et al. 2009. Evolutionary time for dispersal limits the extent but not the occupancy of species’ potential ranges in the tropical plant genus Psychotria (Rubiaceae). Am. Nat. 173: 188 199. Pearson, R. G. and Dawson, T. P. 2003. Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful? Global Ecol. Biogeogr. 12: 361 371. Phillips, S. J. and Dudik, M. 2008. Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecography 31: 161 175. Phillips, S. J. et al. 2004. A maximum entropy approach to species distribution modeling. In: Proceedings of the 21st International Conference on Machine Learning. ACM Press, pp. 655 662. Phillips, S. J. et al. 2006. Maximum entropy modeling of species geographic distributions. Ecol. Model. 190: 231 259. Rangel, T. F. L. V. B. et al. 2006. Towards an integrated computational tool for spatial analysis in macroecology and biogeography. Global Ecol. Biogeogr. 15: 321 327. Richards, P. W. 1973. Africa, the ‘‘odd man out’’. In: Meggers, B. J. et al. (eds), Tropical forest ecosystems in Africa and South America: a comparative review. Smithsonian Inst. Press, pp. 21 26. Salm, R. et al. 2007. Cross-scale determinants of palm species distribution. Acta Amazon. 37: 17 26. Sanderson, E. W. et al. 2002. The human footprint and the last of the wild. Bioscience 52: 891 904. Schurr, F. M. et al. 2007. Colonization and persistence ability explain the extent to which plant species fill their potential range. Global Ecol. Biogeogr. 16: 449 459. Skov, F. and Svenning, J.-C. 2004. Potential impact of climatic change on the distribution of forest herbs in Europe. Ecography 27: 366 380. Soberon, J. 2007. Grinnellian and Eltonian niches and geographic distributions of species. Ecol. Lett. 10: 1115 1123. Sokal, R. R. and Rohlf, F. J. 1995. Biometry. W. H. Freeman. Svenning, J.-C. 1999. Microhabitat specialization in a species-rich palm community in Amazonian Ecuador. J. Ecol. 87: 55 65.

Svenning, J.-C. and Skov, F. 2004. Limited filling of the potential range in European tree species. Ecol. Lett. 7: 565 573. Svenning, J.-C. and Skov, F. 2005. The relative roles of environment and history as controls of tree species composition and richness in Europe. J. Biogeogr. 32: 1019 1033. Svenning, J.-C. and Skov, F. 2007. Ice age legacies in the geographical distribution of tree species richness in Europe. Global Ecol. Biogeogr. 16: 234 245. Svenning, J.-C. et al. 2006. The relative roles of environment, history and local dispersal in controlling the distributions of common tree and shrub species in a tropical forest landscape, Panama. J. Trop. Ecol. 22: 575 586. Svenning, J.-C. et al. 2008. Postglacial dispersal limitation of widespread forest plant species in nemoral Europe. Ecography 31: 316 326. Thomas, C. D. et al. 2004. Extinction risk from climate change. Nature 427: 145 148. Tuley, P. 1995. The palms of Africa. The Trendrine Press. Tuomisto, H. 2007. Interpreting the biogeography of South America. J. Biogeogr. 34: 1294 1295. Tuomisto, H. et al. 2003. Dispersal, environment, and floristic variation of western Amazonian forests. Science 299: 241 244. Walther, G. R. et al. 2007. Palms tracking climate change. Global Ecol. Biogeogr. 16: 801 809. Webster, R. and Oliver, M. A. 2007. Geostatistics for environmental scientists. Wiley. White, F. 1983. Vegetation of Africa a descriptive memoir to accompany the Unesco/AETFAT/UNSO vegetation map of Africa. U.N. Educational, Scientific and Cultural Organization. Wisz, M. and Guisan, A. 2009. Do pseudo-absence selection strategies influence species distribution models and their predictions? An information-theoretic approach based on simulated data. BMC Ecol. 9: 8. Yamakoshi, G. 1998. Dietary responses to fruit scarcity of wild chimpanzees at Bossou, Guinea: possible implications for ecological importance of tool use. Am. J. Phys. Anthropol. 106: 283 295. Zona, S. and Henderson, A. 1989. A review of animal-mediated seed dispersal of palms. Selbyana 11: 6 21.

Download the Supplementary material as file E6273 from <www.oikos.ekol.lu.se/appendix>.

IBS SPECIAL ISSUE

391

Ecography 33: 392 401, 2010 doi: 10.1111/j.1600-0587.2010.06269.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: David M. Green. Accepted 14 March 2010

The relationship between geographic range size and life history traits: is biogeographic history uncovered? A test using the Iberian butterflies Enrique Garcia-Barros and Helena Romo Benito E. Garcia-Barros (garcia.barros@uam.es) and H. Romo Benito, Dept of Biology, Univ. Auto´noma de Madrid, ES-28049 Madrid, Spain.

IBS SPECIAL

ISSUE

The geographic range of a species is influenced by past phylogenetic and biogeographic patterns. However, other historical interactions, including the interplay between life history and geography, are also likely involved. Therefore, the range size of a species can be explained on the basis of niche-breadth or dispersal related hypotheses, and previous work on European butterflies suggests that both, under the respective guise of ecological specialisation and colonising ability may apply. In the present study, data from 205 species of butterflies from the Iberian peninsula were processed through multiple regression analyses to test for correlations between geographic range size, life history traits and geographic features of the species distribution types. In addition, the percentage of variance explained by the subsets of variables analyzed in the study, with and without control for phylogenetic effects was tested. Despite a complex pattern of bivariate correlations, we found that larval polyphagy was the single best correlate of range size, followed by dispersal. Models that combined both life history traits and geographic characteristics performed better than models generated independently. The combined variables explained at least 39% of the variance. Bivariate correlations between range size and body size, migratory habits or egg size primarily reflected taxonomic patterning and reciprocal correlations with larval diet breadth and adult phenology. Therefore, aspects of niche breadth i.e. potential larval diet breadth emerged as the most influential determinants of range size. However, the relationships between these types of ecological traits and biogeographic history must still be considered when associations between life history and range size are of interest.

Range size not only represents a proxy for species rarity (IUCN 2001), but also a ‘‘parameter’’ for patterns of diversity (Rosenzweig 1975, 2003, Gaston and Blackburn 1999, He and Legendre 2002, Gaston 2003, Storch et al. 2007). The geographic range of a species is determined by the interaction between habitat suitability and connectivity, species niche requirements and population dynamics (Gregory and Gaston 2000). Regardless of the complexity of the processes involved (Brown 1984, Gaston et al. 1997, 2000, Cowley et al. 2001b, Gilbert and Lechowicz 2004), these interactions operate via the biological properties of individual species. Hence, the relationships should be identifiable as correlations between species range size and ecological features (Brown 1984, 1995, Gaston and Blackburn 1994, Gaston et al. 2000, Diniz-Filho et al. 2005). Large-scale inter-specific analyses of present species ranges are often hampered by three factors: 1) phylogenetic patterning; 2) geographic turnover in species life histories; and 3) the biogeographic history underlying present ranges. Taxonomic variability and phylogenetic relationships across a species range are routinely addressed (Carrascal et al. 2008, Gove et al. 2009, Calosi et al. 2010), however geographic turnover and biogeographic history remain 392

largely unresolved. First, species life histories exhibit geographic variability across a species range. Consequently, analyses of ‘‘mean’’ trait values measured across geographic gradients may suggest a misleading niche breadth-based explanation (Gaston et al. 2007). A provisional solution is to focus on intermediate scale patterns. Finally, evidence for hidden historical causes in observable ranges can be tested by comparing apparent relationships between range size and ecological factors with simple biogeographic features of the species range, such as the chorotype and range position (as shown by recent work on plant distributions; Weiser et al. 2007, Gove et al. 2009). Butterflies are highly sensitive to environmental changes, particularly changes impacting vegetation because butterfly larvae are specialised herbivores. Several Lepidopteran life history traits display notable interspecific variation, as well as intraspecific differences along geographic gradients (Nylin 2009). Studies assessing butterfly range size and ecology across species have been reported in temperate areas, most notably the British Isles (Hodgson 1993, Quinn et al. 1998, Dennis et al. 2000, 2004, 2005, Cowley et al. 2001a, b), among others (Hughes 2000, Komonen et al. 2004). Dennis et al. (2004) provides evidence that

Figure 1. Main Iberian faunal regions based on butterfly distributions (adapted from Romo 2008).

correlates range size with population density (depending on the geographic scale; Hughes 2000), development time, number of broods per year, adult mobility, and resource use specificity. Furthermore, host plant type is correlated with phenology (Cizek et al. 2006) and adult mobility with larval polyphagy, resource availability and range position (Komonen et al. 2004). However, Dennis et al. (2005) points out that range size and larval host range are associated because of a reciprocal dependence on other

life history and resource variables. Therefore, the available evidence suggests a complex pattern of interrelated factors, and those compatible with a dispersal- or niche- related explanation of range size are most relevant. The role of history in the evolution of the ecological links within the land biotas remains untested, and evidence suggests that the geographic structure of the West-Palaearctic butterfly fauna has been strongly modelled by postglacial events (Dennis et al. 1991, 1998, Schmitt 2007). The present study served to determine how the integration of data on species ranges and ecology might modify former explanations of range size in these insects (Dennis et al. 2004, 2005). In addition, to what degree the correlations among variables were supported from data derived from faunal regions of different climate and physiography. To answer these questions, we tested the relationships between life histories and geographic ranges of Iberian butterflies. Obscure historical patterns, which might be identifiable on a topological basis, were addressed by analyzing relationships between range size and variables that describe species range (with and without controlling for phylogenetic relatedness). We subsequently compared the explanatory power of each of the two subsets of variables (life history and geography), and tested for a relationship between subsets in terms of their shared variance.

Table 1. Bivariate correlations (R) between occupancy (AREA) and the independent variables, based on the raw species data (columns a, b) and on the standardised independent contrasts (columns c, d), and from the full data set (a, c) or the subset of species with known egg sizes (b, d). Sample sizes are a 205, b 155, c 146, d 119. ns p 0.05, * pB0.05, ** p B0.01, *** p B0.001, **** p B0.0001. Variable

Raw species values (a)

Independent contrasts (c)

(d)

0.221*** 0.148* 0.345***

0.188* 0.125 ns 0.179* 0.354***

0.004 ns 0.104 ns 0.178*

0.053 ns 0.076 ns 0.246** 0.255**

VOLT NMON MMON SDMN OWSA OWSE OWSL OWSP

0.357**** 0.782**** 0.239*** 0.724**** 0.173* 0.050 ns 0.169* 0.246***

0.420**** 0.781**** 0.266*** 0.697**** 0.144 ns 0.116 ns 0.140 ns 0.289***

0.445*** 0.647**** 0.178* 0.565**** 0.010 ns 0.089 ns 0.024 ns 0.159 ns

0.308*** 0.743**** 0.287** 0.663**** 0.063 ns 0.007 ns 0.118 ns 0.212*

LPDB LHPL LHPF LHTH LHTW LHPM

0.624**** 0.139* 0.158* 0.032 ns 0.158* 0.029 ns

0.651**** 0.182* 0.091 ns 0.033 ns 0.091 ns 0.011 ns

0.681**** 0.045 ns 0.256 ns 0.014 ns 0.146 ns 0.002 ns

0.596**** 0.152 ns 0.084 ns 0.040 ns 0.127 ns 0.007 ns

EGGB MIRW MIRS MIRO MIRM MIRX

0.070 ns 0.138* 0.026 ns 0.073 ns 0.133 ns 0.107 ns

0.065 ns 0.215** 0.072 ns 0.117 ns 0.203* 0.161*

0.096 0.093 0.075 0.089 0.015 0.096

LATM DISP ALTM ENDM REGA REGB REGC REGD REGE

0.241*** 0.761**** 0.466**** 0.495**** 0.315*** 0.671**** 0.479**** 0.592**** 0.641****

0.316*** 0.499**** 0.508**** 0.444**** 0.281*** 0.679**** 0.469**** 0.604**** 0.648****

0.309*** 0.634**** 0.406*** 0.419*** 0.162 ns 0.535**** 0.407*** 0.538**** 0.555****

AWL ASXD EGGS MIGR

0.081 0.063 0.054 0.029 0.109 0.040

ns ns ns ns ns ns

ISSUE

ns ns ns ns ns ns

IBS SPECIAL

(b)

0.412**** 0.724**** 0.461**** 0.328**** 0.166 ns 0.544**** 0.501**** 0.542**** 0.541****

393

Consequently, species occupancies may be underestimated for widespread and overestimated for rare species (Romo and GarcÄąÂ´a-Barros 2005, Romo et al. 2006). Life history variables Variable selection and measurements were chosen based on the available study material and literature. Care was taken to obtain only information specific to the study area (sources in GarcÄąÂ´a-Barros et al. 2004, with information from the French Pyrenees from Lafranchis 2000 incorporated when required), with the exception of relative egg size (as detailed below). Life history variables were divided into to three subsets: adult features (size, migratory habits and fecundity); phenology; and larval features (larval host specificity, gregariousness and ant relationships). Multistate categorical factors were recoded as dummy binary variables to facilitate the identification of the precise nature of any significant effects. The variables and abbreviations used throughout the remaining text are as follows. Adult features

IBS SPECIAL

ISSUE

Figure 2. Relationship between dispersion (DISP) and the species size range (AREA). (A) after the species values: DISP 0.033 0.0004 AREA 0.038 AREA1/2 (R 0.779, pB0.00001). (B) measured on the independent contrasts: DISP 0.575 AREA, R 0.634, p B0.00001.

Methods

AWL: adult size (forewing length in mm, log-transformed), estimated as the median of male and female mean wing lengths from sample cabinet specimens (n 20 000: GarcÄąÂáBarros unpubl.). When B10 individuals of each sex were available, measurements from the literature were obtained and included in the data set (Manley and Allcard 1970, FernaÂńdez-Rubio 1991, Maravalhas 2003). ASXD: sexual dimorphism in adult size range, residuals from a linear regression of female size onto male size, both log-transformed (female AWL 0.022 1.031 male AWL; R 0.988; pB0.00001). MIGR: migratory habits, where the species were classified as â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;sedentaryâ&#x20AC;&#x2122;â&#x20AC;&#x2122; or â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;migratoryâ&#x20AC;&#x2122;â&#x20AC;&#x2122; (including eumigrants, intra-regional migrants and long-distance dispersers; Templado 1976, Eitschberger et al. 1991). EGGS: relative egg size, estimated as the residuals from a regression of log-transformed egg volume (from GarcÄąÂá-Barros 2000b) on AWL (R 0.628, p B0.00001 using raw data; and R 0.428, p B0.00001 using independent contrasts, see below). Estimated egg size does not always include Iberian samples (consequently, a combination with local adult sizes resulted in some level of measurement error), and was available from only 154 species. Therefore, it was necessary to duplicate some analyses, as detailed below.

Species and range sizes The data matrix was comprised of 205 butterfly species (superfamilies Hesperioidea and Papilionoidea) from the Iberian Peninsula (Iberia, the continental territories of Portugal and Spain). Non-native, long-distance migrant species (such as monarch butterflies, Danaus spp.), and species with missing life history data were excluded. Each species range (AREA, the dependent variable) was measured as the number of 10 km square cells (UTM military grid) occupied (data from GarcÄąÂ´a-Barros et al. 2004, updated for this study by the authors). Area coverage at this grid size does not adequately represent the region and sampling was concentrated on the most diverse areas.

394

Phenology

VOLT: voltinism pattern (one life-cycle per year, or more than one in at least part of the territory). NMON: number of months in which the adults were recorded (95% confidence limits). MMON: mean month of adult occurrence (with months scored as 1 12). SDMN: standard deviation of MMON. The last three variables (NMON, MMON and SDMN) were calculated from pooled historical records with reliable dating (data from the authors database, details in GarcÄąÂ´a-Barros et al. 2004). Four binary variables were used to code for the over wintering stage: OWSE (egg), OWSL (larva), OWSP (pupa) and OWSA (adult).

Figure 3. Relationships between range size (AREA) and larval diet breadth (LPDB) based on species raw data (A, C) and independent contrasts (B, D). A and B: bivariate plots; C and D: relationships after controlling for effects of other variables with significant effects, based on the residuals of the linear regressions of AREA and LPDB on the remaining variables in the models of Table 2 (column A) and 3 (column A).

Larval, and larval-host features

described by three binary factors (adapted from Fiedler 1991): weak (including facultative myrmecophily, MIRW), strong (MIRS) and obligate (MIRO). The three variables were also combined as a single semi-quantitative variable (MIRX, with values from 0 no relationship with ants to 4 obligate ant-dependence), as well as an all-or-nothing binary variable (MIRM, where â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;1â&#x20AC;&#x2122;â&#x20AC;&#x2122; means any level of association and â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;0â&#x20AC;&#x2122;â&#x20AC;&#x2122; no association). Geographic variables (species range attributes)

395

ISSUE

The X and Y UTM coordinates of each species data subsets were used to estimate LATM (mean Y value, a surrogate of mean latitude) and DISP (dispersion of the cells occupied, calculated as the geometric average of the variances of X and Y). ALTM (mean altitude, in m a.s.l.) was derived from the original database records. ENDM: degree of endemism, quantified as the percentage of 18 30Ć&#x2019; cells occupied by each species in Iberia over the number of European species (from Kudrna 2002). Five variables: REG (A, B, C, D, E) were adopted to qualify the species according to their presence/absence in each of five Iberian regions (A to E, Fig. 1, formerly defined by Romo 2008).

IBS SPECIAL

LPDB (larval diet breadth) was measured as the number of plant families (F), genera (G) and species (S) used as larval hosts using the formula LPDB (S G Fa)1/3, where a (G F)/2S. The index assumes higher values for any number of hosts when the hosts represent a greater number of genera and families (i.e. increased LPDB); LPDB was better correlated to range size (r 0.62) than either the number of host species, genera or families (respectively R 0.589, R 0.567 and R 0.316; pB0.001). LHPM: coarse taxonomy of the larval hosts (monocotyledons or dicotyledons). Plant structure was broadly described by two binary variables, LHTW (woody plants) and LHTH (herbaceous plants), and plant parts eaten were described as LHPL (leaves), and LHPF (flowers, fruits or buds). Larval gregariousness was measured by coding the egglaying mode of each species (EGGB: eggs laid singly, or arranged in egg clusters), since the habit of laying the eggs in clusters is almost universally associated to gregarious larval habits. The larvae of some butterflies (namely, the Lycaenidae) may show symbiotic or parasitic relationships with specific ants, whose presence may condition the success of the butterfly population. The strength of myrmecophily was

Figure 4. Relationships between range size (AREA) and the concentration of speciesâ&#x20AC;&#x2122; ranges in the study relative to Europe (ENDM) based on raw data (A, C) and independent contrasts (B, D). A and B: bivariate plots; C and D: relationships after controlling for the effects of other significant variables (residuals from the linear regressions of AREA and ENDM on the remaining variables, models in Table 2 and 3 for columns A and B, respectively).

IBS SPECIAL

ISSUE

Table 2. Explanation of AREA based on species means and three subsets of independent variables: life history plus geography (A), life-history (B) and geography (C). The figures shown are estimated coefficients and signs (Coef.), Wald statistics and significance levels (P) of the variables selected in each model. Whole model results are shown at the lower part of the table. ** p B0.01, *** p B0.001, **** p B 0.0001. Model: Variable

(A) All variables Coef.

MIGR NMON MMON SDMN OWSP LPDB ENDM ALTM LATM REGA REGB REGC REGE DF Deviance Deviance/DF Loglikelihood R2

396

0.285 0.039 0.346 0.016 0.154 0.840 0.262 0.393 0.210

Wald (P) 89.163**** 7.274*** 26.068**** 13.269*** 6.921*** 31.840**** 21.117**** 18.935**** 16.507**** 195 193.374 0.992 303.885 0.820****

(B) Life history Coef. 0.186 0.120 0.001 0.358 0.159 0.063

Wald (P) 10.096** 252.368**** 11.912**** 164.216**** 10.787*** 16.724**** 198 320.105 1.617 367.251 0.811****

Wald (P) 77.010**** 8.725** 16.512**** 48.338**** 35.102**** 27.171**** 35.644**** 197 307.555 1.561 360.975 0.715****

Table 3. Explanation of AREA based on the independent contrasts and variable selection. Variable subsets: life history plus geography (A), life history (B) and geography (C). The data shown are estimated coefficients and signs (Coef.), Wald statistics and significance levels (P) of the variables selected in each model. Whole model fitting results are shown at the lower part of the table. ** pB0.01, *** p B0.001, **** pB0.0001. Model: Variable SDMN LPDB ENDM REGD

(A) All variables

(B) Life history

Coef.

Wald (P)

Coef.

0.514 0.381 1.122

17.492**** 7.209** 16.952****

0.289 0.292

DF Deviance Deviance/DF Loglikelihood R2

143 235.083 1.644 128.977 0.614****

Wald (P) 7.221** 7.997** 144 240.065 1.667 137.855 0.572****

Independent contrasts Phylogenetically independent contrasts were calculated using the package COMPARE (Martins 2004). Branch lengths were set to the same length (1.00) except at polytomies, where we set a low value (0.00001; as suggested by Martins 2004). Single contrasts at these nodes were computed by initially resolving the relationships arbitrarily, then the average value per polytomy was calculated and contrasts computed. The adopted phylogenetic topology (Supplementary material) followed the sources quoted by GarcÄąÂ´a-Barros (2000a) and Cizek et al. (2006); and updated after Als et al. (2004), Wahlberg et al. (2005), Braby et al. (2006), Lukhtanov et al. (2006), PenË&#x153; a et al. (2006), Simonsen et al. (2006), Weingartner et al. (2006) and Kodandaramaiah and Wahlberg (2009).

Coef.

Wald (P)

1.338

28.393 145 241.998 1.669 136.855 0.268****

logarithmic link function. The variables were entered in descending order according to their bivariate relationship with AREA (Table 1), and a backward selection was fit at each step to discard redundant terms. Waldâ&#x20AC;&#x2122;s statistic was applied to measure the contribution of each variable, and model strength with the standard deviation (whole model R2 values were added from GRM fitting). First, one model was fit using all variables (life-history and geographic). Two additional models were subsequently calculated with the two sets of variables independently. Finally, we applied variance partitioning (Legendre and Legendre 1998, DinizFilho and Bini 2008) to measure the amount of variance explained by life history, geographic features, or shared by both sets of variables (the variables included in this specific analysis were those that exhibited significant effects in any of the three formerly described models, i.e. â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;allâ&#x20AC;&#x2122;â&#x20AC;&#x2122;, â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;life historyâ&#x20AC;&#x2122;â&#x20AC;&#x2122;, and â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;geographyâ&#x20AC;&#x2122;â&#x20AC;&#x2122;).

Procedure Preliminary analyses

Pure life-history Pure geographic Shared (life history geographic) Residual (not explained) Total Model (R)

Species values

Contrasts

0.161 0.456 0.655 0.139 1.000 0.901

0.188 0.089 0.390 0.333 1.000 0.666

397

ISSUE

Table 4. Variance partitions among the life history and geographic variables, based on the species values and on the standardised independent contrasts. The subset of variables for each column consisted of those with significant effects in any of the models described in Table 2 (raw data) and Table 3 (contrasts). The figures represent the proportion of variance explained (over 1.000).

Pairwise correlations are shown in Table 1. Egg size (EGGS) was significantly correlated with AREA. One additional stepwise regression restricted to data from species with known EGGS values did not select egg size (using either the log-transformed or the independent contrasts). The variables selected without controlling for phylogeny included NMON, SDMN, REGB, REGA, REGC, REGE and LPDB (deviance/DF 0.873; log-likelihood 232.728; R2 0.837; p B0.0001), and variables from the set of contrasts included MIGR, LPDB, ALTM, LATM, REGA and REGD (deviance/DF 1.701; loglikelihood 74.425; R2 0.597; pB0.0001). Consequently, we excluded EGGS from subsequent analyses, which resulted in a somewhat lower number of species and contrasts analyzed. LATM displaced dispersion (DISP) (with which it was correlated; R 0.427, pB0.0001; R 0.393, p B 0.0001, based on the contrasts) in the analysis, probably due to heteroscedascity in the X and Y values. Therefore, to gain insights into the correlation in dispersion patterns, we tested the relationship between DISP to AREA with a polynomial function (species means) and a linear function (contrasts) (Fig. 2). We subsequently analysed the residuals following the methods described above.

IBS SPECIAL

The raw data values for AREA, SDMN and ALTM were square root transformed, MMON cubed, and AWL, EGGS and LPDB were log10-transformed to approach normality. Two series of computations were conducted, using species means and independent contrasts. Contrast-based regressions were forced through the origin (Felsenstein 1985). The bivariate relationships between AREA and the independent variables were determined. Subsequently, regression models were selected using a manual stepwise procedure (GLM module of Statistica: StatSoft 2004), assuming a Poisson distribution of the residuals and a

Table 5. Bivariate correlations (R) between the geographical dispersion of the species data points not explained by occupancy (residuals of the regression of DISP on AREA) and the potentially explanatory variables, derived from raw data (n 205) and independent contrasts (n 149). ns p 0.05, * p B0.05, ** p B 0.01, *** p B0.001, **** p B0.0001. Variable

Species values ns ns ns ns

Contrasts

AWL ASXD EGGS MIGR

0.097 0.009 0.040 0.036

0.131 0.083 0.086 0.061

ns ns ns ns

VOLT NMON MMON SDMN OWSA OWSE OWSL OWSP

0.061 ns 0.161* 0.246** 0.023 ns 0.007 ns 0.016 ns 0.098 ns 0.183*

0.083 ns 0.178* 0.062 ns 0.159 ns 0.028 ns 0.064 ns 0.031 ns 0.041 ns

LPDB LHPL LHPF LHTH LHTW LHPM

0.061 ns 0.055 ns 0.257*** 0.284**** 0.270*** 0.236**

0.125 ns 0.031 ns 0.077 ns 0.239** 0.129 ns 0.201*

EGGB MIRW MIRS MIRO MIRM MIRX

0.015 ns 0.191* 0.034 ns 0.046 ns 0.203* 0.157 ns

0.011 ns 0.119 ns 0.196* 0.049 ns 0.181* 0.232**

LATM ALTM ENDM REGA REGB REGC REGD REGE

0.438 0.257*** 0.226** 0.020 ns 0.115 ns 0.177* 0.463**** 0.329****

0.280*** 0.282*** 0.229** 0.125 ns 0.363**** 0.476**** 0.551**** 0.411****

IBS SPECIAL

ISSUE

Results Pairwise analyses found 24 variables significantly correlated with AREA (Table 1) across species means, and 14 across contrasts (Table 1, Fig. 3 and 4). The selected variables are detailed in Table 2 and 3. A maximum of six ‘‘life history’’ variables were selected when this subset of variables was analysed. However, in the combined analyses (life history plus geography), the geographical variables tended to displace the life history traits from the models, with the exception of LPDB (Fig. 3) and, depending on the analysis, SDMN. Seven of the geographical variables exhibited significant effects when species means were analysed (ENDM, ALTM, LATM, REGA, REGB, REGC and REGE; Table 2). However, significant effects were not evident when contrasts were analysed, with the exception of ENDM and/ or REGD, which remained significant (Table 3, Fig. 4). Variance partitioning (Table 4) indicated that irrespective of the source of data (species mean values or contrasts), a large part of the explanation of the variance is shared by life history plus geography. Finally, dispersion (with fixed effects for AREA) was positively correlated with geography (REGD and ENDM) and negatively with larval host type (LHTH i.e. ‘‘not 398

Table 6. Stepwise selection results summarising the relationships between range dispersion (with fixed AREA effects, residuals from the function in Fig. 2) and the remaining variables, based on species raw values and on independent contrasts. The values shown are estimated coefficients and signs (Coef.), Wald statistics, and significance (P) of selected variables. Whole model fitting statistics are indicated at the lower part of the table. * p B0.05, ** pB 0.01, *** p B0.001, **** pB0.0001. Variable

REGD ENDM LHTH OWSP DF Deviance Deviance/DF Loglikelihood R2

Species data

Contrasts

Coef.

Wald (P)

Coef.

0.177 0.164 0.099 0.0633

52.588**** 17.754**** 10.433** 4.503*

0.369

200 5.2751 0.026 84.2684 0.299****

Wald (P) 6.187* 145 180.325 1.244 222.579 0.052**

feeding on herbaceous plants’’), and with ‘‘overwintering as a pupa’’ (OWSP) (Table 5, 6).

Discussion The following summarizes the results of our study: 1) significant pairwise correlations between range size and several life history variables were observed; 2) these correlations generally tend to loose strength after controlling for phylogeny; 3) the variables retained by stepwise regression selection were associated with larval host diet breadth, adult phenology, and geographic features of species distribution; and 4) an integral component of the explanation provided by the models was of a mixed (life history and geographic) nature. Both niche theory and dispersal-related expectations are supported by the bivariate correlations across the species means results generated in this study. Comparatively, widespread species are larger, more sexually dimorphic, and exhibit increased fecundity and dispersal ability. Species with more scattered/patchy distributions tend to show longer periods of adult occurrence, their larvae typically feed on dicotyledonous woody plants, and demonstrate some degree of ant-dependence as well as a restricted Iberian range with low mean latitude and altitude. The generally (though not universally, Table 1) lower correlations from independent contrasts do not simply reflect the reduced degrees of freedom (as argued by Dennis et al. 2004), but also the different weight of taxonomic patterning across variables (Brooks and McLennan 2002, Diniz-Filho and Torres 2002). Our comparative results were partly hindered by poor phylogenetic resolution and unknown branch lengths, two reasons why (despite the robust contrasts method under such conditions; Martins and Garland 1991, Garland et al. 1992) we concentrated on identifying the sign and strength of the relationships rather than their shape (Quader et al. 2004). Therefore, within these limitations we confidently concluded that the bivariate correlations between range size and relative egg size, dispersive adults, long period of adult occurrence, larval diet

taxa, done within a phylogeographic framework. The geographic distribution of species is a product of speciation, extinction and the temporal dynamics of its range (Gaston 1998, 2003). Consequently, the range of a species should be examined and explained in terms of the species features, its interactions with the environment (Kean and Barlow 2004) and the historical factors, which likely set limits on other interactions. However, little attention has been drawn to the relationships between different macroecological patterns (Blackburn and Gaston 2001). Two incidental findings of our work relate to endemism and Rapoport’s pattern. The significance of endemism has a practical application in Iberia: Iberian endemics tend to be rare in Iberia (for comparable results from other regions and taxa see Gregory and Blackburn 1998 and Carrascal et al. 2008). Based on our knowledge of butterfly species and their distributions in the study area, most of the butterflies with small ranges in this region are restricted to relatively high elevations on the main mountain chains, which cover a small portion of the peninsula. These species may be particularly sensitive e.g. in terms of global warming and other environmental impacts. For the same reason, the concentration of mountain ranges in the northern half of the peninsula explains the negative pairwise correlations between range size and both latitude and altitude, incongruent with Rapoport’s pattern (where ranges should be wider at higher latitudes or altitudes; Rapoport 1975, Lomolino et al. 2006). In summary, our results identified niche breadth (via larval polyphagy) as primarily correlated with range size, together with interactions between non-explicit historical causes (represented by chorotypes) and life histories. A more thorough ‘‘dissection’’ of the biological correlations of range size, as well as an integrated multi-scale protocol is required before more specific explanations for range size may be achieved. Acknowledgements We thank I. Echavarren for assistance during the early stages of the study, and C. Stefanescu and M. L. Munguira for contributing to the revision of life history data. This study was partly funded by project GL2006-10196 (M. E. C.).

References

399

ISSUE

Als, T. D. et al. 2004. The evolution of alternative parasitic life histories in large blue butterflies. Nature 432: 386 390. Austin, M. P. and Smith, T. M. 1989. A new model for the continuum concept. Vegetatio 83: 35 47. Blackburn, T. M. and Gaston, K. J. 2001. Linking patterns in macroecology. J. Anim. Ecol. 70: 338 352. Braby, M. F. et al. 2006. Molecular phylogeny and systematics of the Pieridae (Lepidoptera: Papilionoidea): higher classification and biogeography. Zool. J. Linn. Soc. 147: 239 275. Brooks, D. R. and McLennan, D. A. 2002. The nature of diversity. An evolutionary voyage of discovery. Chicago Univ. Press. Brown, J. H. 1984. On the relationship between abundance and distribution of species. Am. Nat. 124: 255 279. Brown, J. H. 1995. Macroecology. Chicago Univ. Press. Calosi, P. et al. 2010. What determines a species’ geographical range? Thermal biology and latitudinal range size relationships in European diving beetles (Coleoptera: Dytiscidae). J. Anim. Ecol. 79: 194 204.

IBS SPECIAL

breadth and some geographic variables were not taxonomically driven artefacts. Among the life history variables, multiple selection on independent contrasts supports a causal interpretation of range size in terms of niche breadth (potential or realised larval polyphagy may assist the offspring to thrive in more diverse habitats) and, at least partially, dispersion in the adult ‘‘time window’’. However, the second variable may actually represent a geographic turnover of species phenologies, as the Iberian lands cover a remarkable array of climate and vegetation types. Phenological variation along a geographic gradient should be higher among populations than within them, therefore the observed correlation may not denote an ability of widespread species to occupy varied habitats, but the degree to which widespread species phenologies are tuned to local conditions. Depending on the methods and variables used to study British butterflies, this explanation is supported by some reports but not others (Hodgson 1993, Dennis et al. 2004, 2005). Correlations between range size and body size or dispersal ability (formerly documented from various taxa, Gaston and Blackburn 1994, Brown 1995, Purvis et al. 2001 and Diniz-Filho et al. 2005, but cf. Hillebrand et al. 2001, Wilkinson 2001, Fernandez and Vrba 2005 or Rundle et al. 2007) are not supported by the most parsimonious interpretation of our results. The relationship between Iberian butterfly range and adult size is taxonomic, range size is weakly related to migratory status, and both relationships break down under multivariate selection protocols. However, we cannot strictly discard alternative explanations of dispersal type, namely because butterfly wing length (which we tested) might not the best surrogate for dispersal ability (as shown for damselflies by Rundle et al. 2007). Furthermore, functional and genetic links among species life history attributes (Stearns 1977, Roff 2002) generally result in complex patterns of cross correlations (see butterfly examples in Garcı´a-Barros 2000a or Cizek et al. 2006). If distribution patterns are to be evaluated in terms of the realized niche (Austin and Smith 1989, Kockemann et al. 2009) and niche is assessed based on environmental variables (Thuiller et al. 2003), ‘‘complex’’ variables (describing organism-habitat interactions) should perform better in multivariate tests than ‘‘proximate’’ variables (describing features of the organism, or of the habitat, Dennis et al. 2004, 2005). Therefore, the distinction between ‘‘causal’’ and ‘‘most parsimonious’’ solutions is of interest. However, a more exhaustive analysis of the life histories of Iberian butterflies at local or regional levels and comparable studies across different regions is warranted to provide resolution at this scale. Results of the study suggested that the explanation for range size could be improved by including additional variables to represent processes not explicitly tested. For example, some geographic variables exhibited a high weight and proportion of variance shared with life history data. Range size might be the result of interactions between species life history patterns and geographic history. Comparable conclusions have been drawn from recent work on plant biogeography (Svenning and Skov 2005, Weiser et al. 2007, Gove et al. 2009). Resolving these processes for further analyses is not a simple task, as it requires a comparative approach to the life histories of each of the

ISSUE

IBS SPECIAL

Carrascal, L. M. et al. 2008. Explanations for bird species range size: ecological correlates and phylogenetic effects in the Canary Islands. J. Biogeogr. 35: 2061 2073. Cizek, L. et al. 2006. Host plant defences and voltinism in European butterflies. Ecol. Entomol. 31: 337 344. Cowley, M. J. R. et al. 2001a. Density distribution relationships in British butterflies. I. The effect of mobility and spatial scale. J. Anim. Ecol. 70: 410 425. Cowley, M. J. R. et al. 2001b. Density distribution relationships in British butterflies. II. An assessment of mechanisms. J. Anim. Ecol. 70: 426 441. Dennis, R. L. H. et al. 1991. A multivariate approach to the determination of faunal structures among European butterfly species (Lepidoptera: Rhopalocera). Zool. J. Linn. Soc. 101: 1 49. Dennis, R. L. H. et al. 1998. Faunal structures among European butterflies: evolutionary implications of bias for geography, endemism and taxonomic affiliations. Ecography 21: 181 203. Dennis, R. L. H. et al. 2000. Ecological correlates of island incidence and geographical range among British butterflies. Biodivers. Conserv. 9: 343 359. Dennis, R. L. H. et al. 2004. Host plants and butterfly biology. Do host-plant strategies drive butterfly status? Ecol. Entomol. 29: 12 26. Dennis, R. L. H. et al. 2005. Does diet breadth control herbivorous insect distribution size? Life history and resource outlets for specialist butterflies. J. Insect Conserv. 9: 187 200. Diniz-Filho, J. A. F. and Torres, N. M. 2002. Phylogenetic comparative methods and the geographic range size body size relationship in new world terrestrial Carnivora. Evol. Ecol. 16: 351 367. Diniz-Filho, J. A. F. and Bini, L. M. 2008. Macroecology, global change and the shadow of forgotten ancestors. Global Ecol. Biogeogr. 17: 11 17. Diniz-Filho, J. A. F. et al. 2005. Macroecology, geographic range size body size relationships and minimum viable population analysis for new world Carnivora. Acta Oecol. 27: 25 30. Eitschberger, U. et al. 1991. Wanderfalter in Europa (Lepidoptera). Appeal for international cooperation in the research of the migration of insects. Atalanta 22: 1 67. Felsenstein, J. 1985. Phylogenies and the comparative method. Am. Nat. 125: 1 15. Fernandez, M. H. and Vrba, E. S. 2005. Body size, biomic specialization and range size of African large mammals. J. Biogeogr. 32: 1243 1256. Fernańdez-Rubio, F. 1991. Guıá de mariposas diurnas de la Penıńsula Ibe´rica, Baleares, Canarias, Azores y Madeira. Pira´mide, Madrid. Fiedler, K. 1991. Systematic, evolutionary and ecological implications of myrmecophyly within the Lycaenidae (Insecta: Lepidoptera: Papilionoidea). Bonner Zool. Monogr. 31: 1 210. Garcıá-Barros, E. 2000a. Body size, egg size, and their interspecific relationships with ecological and life history traits in butterflies (Lepidoptera: Papilionoidea, Hesperioidea). Biol. J. Linn. Soc. 70: 251 284. Garcıá-Barros, E. 2000b. Egg size in butterflies (Lepidoptera: Papilionoidea and Hesperiidae): a summary of data. J. Res. Lepid. 35: 90 136. Garcıá-Barros, E. et al. 2004. Atlas of the butterflies of the Iberian Peninsula and Balearic Islands (Lepidoptera: Papilionoidea & Hesperioidea). Monogr. S.E.A. 11: 1 228. Garland, T. J. et al. 1992. Procedures for the analysis of comparative data using phylogenetically independent contrasts. Syst. Biol. 41: 18 32.

400

Gaston, K. J. 1998. Species-range size distributions: products of speciation, extinction and transformation. Phil. Trans. R. Soc. B 353: 219 230. Gaston, K. J. 2003. The structures and dynamics of geographic ranges. Oxford Univ. Press. Gaston, K. J. and Blackburn, T. M. 1994. Are newly described bird species small-bodied? Biodivers. Lett. 2: 16 20. Gaston, K. J. and Blackburn, T. M. 1999. A critique for macroecology. Oikos 84: 353 368. Gaston, K. J. et al. 1997. Interspecific abundance-range size relationships: an appraisal of mechanisms. J. Anim. Ecol. 66: 579 601. Gaston, K. J. et al. 2000. Abundance occupancy relationships. J. Appl. Ecol. (Suppl. 1) 37: 39 59. Gaston, K. J. et al. 2007. The scaling of spatial turnover: prunning the thicket. In: Storch, D. et al. (eds), Scaling diversity. Cambridge Univ. Press, pp. 181 222. Gilbert, B. and Lechowicz, M. J. 2004. Neutrality, niches and dispersal in a temperate forest understorey. Proc. Nat. Acad. Sci. USA 101: 7651 7656. Gove, A. D. et al. 2009. Dispersal traits linked to range size through range location, not dispersal ability, in Western Australian angiosperms. Global Ecol. Biogeogr. 18: 596 606. Gregory, R. D. and Blackburn, T. M. 1998. Macroecological patterns in British breeding birds: covariation of species’ geographical ranges at differing spatial scales. Ecography 21: 527 534. Gregory, R. D. and Gaston, K. J. 2000. Explanations of commonness and rarity in British breeding birds: separating resource use and resource availability. Oikos 88: 515 526. He, F. L. and Legendre, P. 2002. Species diversity patterns derived from species area models. Ecology 85: 1185 1198. Hillebrand, H. et al. 2001. Differences in species richness patterns between unicellular and multicellular organisms. Oecologia 126: 114 124. Hodgson, J. G. 1993. Commoness and rarity in British butterflies. J. Appl. Ecol. 30: 407 427. Hughes, J. B. 2000. The scale of resource specialization and the distribution and abundance of lycaenid butterflies. Oecologia 123: 375 383. IUCN 2001. IUCN Red List categories and criteria: version 3.1. IUCN Species Survival Commission, Gland and Cambridge. Kean, J. and Barlow, N. 2004. Exploring rarity using a general model for distribution and abundance. Am. Nat. 163: 407 E416. Kockemann, B. et al. 2009. The relationships between abundance, range size and niche breadth in central European tree species. J. Biogeogr. 36: 854 864. Kodandaramaiah, U. and Wahlberg, N. 2009. Phylogeny and biogeography of Coenonympha butterflies (Nymphalidae: Satyrinae) patterns of colonization in the Holarctic. Syst. Entomol. 34: 315 323. Komonen, A. et al. 2004. The role of niche breadth, resource availability and range position on the life history of butterflies. Oikos 105: 41 54. Kudrna, O. 2002. The distribution atlas of European butterflies. Oedippus 20: 1 342. Lafranchis, T. 2000. Les papillons de jour de France, Belgique et Louxembourg et leurs chenilles. Biotope, Me`ze. Legendre, P. and Legendre, L. 1998. Numerical ecology, 2nd ed. Elsevier. Lomolino, M. V. et al. 2006. Biogeography, 3rd ed. Sinauer. Lukhtanov, V. A. et al. 2006. Rearrangement of the Agrodiaetus dolus species group (Lepidoptera, Lycaenidae) using a new cytological approach and molecular data. Insect Syst. Evol. 37: 325 334.

Manley, W. B. L. and Allcard, H. G. 1970. A field guide to the butterflies and burnets of Spain. E. W. Classey, Hampton. Maravalhas, E. 2003. As borboletas de Portugal. The butterflies of Portugal. Apollo Books. Martins, E. P. 2004. COMPARE, version 4.6b. Computer programs for the statistical analysis of comparative data. Dept Biology, Indiana Univ., <http://compare.bio.indiana. edu/>. Martins, E. P. and Garland, T. Jr 1991. Phylogenetic analyses of the correlated evolution of continuous characters: a simulation study. Evolution 45: 534 557. Nylin, S. 2009. Gradients in butterfly biology. In: Settele, J. et al. (eds), Ecology of butterflies in Europe. Cambridge Univ. Press, pp. 198 216. Pen˜ a, C. et al. 2006. Higher level phylogeny of Satyrinae butterflies (Lepidoptera: Nymphalidae) based on DNA sequence data. Mol. Phylogenet. Evol. 40: 29 49. Purvis, A. et al. 2001. Past and future carnivore extinctions: a phylogenetic perspective. In: Gittleman, J. L. et al. (eds), Carnivore conservation. Cambridge Univ. Press, pp. 11 34. Quader, S. et al. 2004. Nonlinear relationships and phylogenetically independent contrasts. J. Evol. Biol. 17: 709 715. Quinn, R. M. et al. 1998. The distribution of butterflies and their foodplants. Ecography 21: 279 288. Rapoport, E. 1975. Areografıá: estrategias geogra´ficas de las especies. Fondo de Cultura Econo´ mica. Roff, D. A. 2002. Life history evolution. Sinauer. Romo, H. 2008. Diversidad geogra´fica de las mariposas diurnas Ibero-baleares. Ecosistemas 17: 106 111. Romo, H. and Garcıá-Barros, E. 2005. Distribucio´ n e intensidad de los estudios faunı´sticos sobre mariposas diurnas en la Penıńsula Ibe´rica e Islas Baleares (Lepidoptera, Papilionoidea y Hesperioidea). Graellsia 61: 37 50. Romo, H. et al. 2006. Identifying recorder-induced geographic bias in an Iberian butterfly database. Ecography 29: 873 885. Rosenzweig, M. L. 1975. On continental steady states of species diversity. In: Cody, M. L. and Diamond, J. M. (eds), Ecology and evolution of communities. Harvard Univ. Press, pp. 121 140. Rosenzweig, M. L. 2003. How to reject the area hypothesis of latitudinal gradients. In: Blackburn, T. M. and Gaston, K. J.

(eds), Macroecology. Concepts and consequences. Blackwell, pp. 87 106. Rundle, S. D. et al. 2007. Range size in North American Enallagma damselflies correlates with wing size. Freswater Biol. 52: 471 477. Schmitt, T. 2007. Molecular biogeography of Europe: Pleistocene cycles and postglacial trends. Front. Zool. 4: 11. Simonsen, T. J. et al. 2006. Morphology, molecules and fritilllaries: approaching a stable phylogeny for Argynnini (Lepidoptera: Nymphalidae). Insect Syst. Evol. 37: 405 418. StatSoft 2004. STATISTICA (data analysis software system), version 6.1. Statsoft, Tulsa, <www.statsoft.com>. Stearns, S. C. 1977. The evolution of life history traits: a critique of the theory and a review of the data. Annu. Rev. Ecol. Syst. 8: 145 171. Storch, D. et al. 2007. Scaling species richness and distribution. Uniting the species area and the species energy relationships. In: Storch, D. et al. (eds), Scaling biodiversity. Cambridge Univ. Press, pp. 300 321. Svenning, J.-C. and Skov, F. 2005. The relative roles of environment and history as controls of tree species composition and richness in Europe. J. Biogeogr. 32: 1019 1033. Templado, J. 1976. Sobre la variacio´ n geogra´fica de los Ropalo´ ceros Ibe´ricos (Lepidoptera). Graellsia 31: 79 92. Thuiller, W. et al. 2003. Large-scale environmental correlates of forest tree distributions in Catalonia (NE Spain). Global Ecol. Biogeogr. 12: 313 325. Wahlberg, N. et al. 2005. Phylogenetic relationships and historical biogeography of tribes and genera in the subfamily Nymphalinae (Lepidoptera: Nymphalidae). Biol. J. Linn. Soc. 86: 227 251. Weingartner, E. et al. 2006. Speciation in Pararge (Satyrinae: Nymphalidae) butterflies North Africa is the source of ancestral populations of all Pararge species. Syst. Entomol. 31: 1 13. Weiser, M. D. et al. 2007. Latitudinal patterns of range size and species richness of New World woody plants. Global Ecol. Biogeogr. 16: 679 688. Wilkinson, D. M. 2001. What is the upper size limit for freeliving microorganisms. J. Biogeogr. 28: 285 291.

Download the Supplementary material as file E6269 from <www.oikos.ekol.lu.se/appendix>.

IBS SPECIAL ISSUE

401

Ecography 33: 402 407, 2010 doi: 10.1111/j.1600-0587.2010.06323.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Helmut Hillebrand. Accepted 8 January 2010

Dispersion fields, diversity fields and null models: uniting range sizes and species richness Michael Krabbe Borregaard and Carsten Rahbek

IBS SPECIAL

ISSUE

M. Krabbe Borregaard (mkborregaard@bio.ku.dk) and C. Rahbek, Center for Macroecology, Evolution and Climate, Dept of Biology, Univ. of Copenhagen, Universitetsparken 15, DK-2100 Copenhagen, Denmark.

One of the cornerstones of macroecological research is the ongoing effort to understand large-scale patterns of species richness. These patterns are emergent properties of the distributional ranges of individual species, as they are formed by the overlap of species ranges in a given area. As such, the sizes of ranges, and the processes controlling their geographical location, are key determinants of richness patterns. However, a satisfactory link between range size distributions, the spatial location of individual ranges, and species richness has been slow to emerge. To forge this link, a central part is the role of species associations in determining the composition of species in a defined area. Overlap between species ranges may arise because species have similar ecologies (Webb 2000), for historical reasons of dispersal (Svenning et al. 2008), or may simply be random, because ranges are constrained by the shape of the geographical domain (Colwell and Lees 2000, Jetz and Rahbek 2001). However, although species interactions have consistently been shown to affect species co-occurrence at local scales (Gotelli and McCabe 2002), the importance of species associations for distributional overlap at biogeographical scales remains a central question for macroecology (Gotelli et al. 1997, in press). Species’ ranges are usually continuous at large scales, and as a consequence, the species richness values of closely located sites are not independent. This means that the spatial pattern of species richness cannot be explained by analyzing sites as a set of independent points in a regression (Legendre 1993). Spatial regression methods solve the statistical issue of autocorrelation (Rangel et al. 2006), but a simplistic use of these statistics risks missing the main point. Spatial patterning is not a statistical issue it is an inherent quality of biogeographical data (Rahbek and Graves 2000, Diniz-Filho et al. 2003). In the light of this, recent attention has focused on developing conceptual and analytical tools for macroecological analysis that deal explicitly with species’ ranges. One important advance is the concept of the ‘‘dispersion field’’, developed by Graves and Rahbek (2005). The dispersion field is the set of geographical ranges of all species that occur in a given site. Just as the continental 402

species richness pattern is created by the overlap of all species in a continent, the dispersion field can be visualized as the pattern created by overlapping the ranges of all species occurring in a given cell (Fig. 1). These dispersion fields have striking geometric shapes, and have a number of promising applications. First, it has been argued that the geometric shape of dispersal fields are an approximation to the regional species source pool (Graves and Rahbek 2005). The source pool plays a key role in theories of community assembly, but the concept has been consistently difficult to pin down (Gotelli and Graves 1996). Even more importantly, dispersion fields visualize the species associations that create richness patterns. Hence, they provide an opportunity for more stringent tests of ecological hypotheses for species richness than standard regression methods. A growing research paradigm in macroecology is to replace curve-fitting methods with mechanistic models of range placement (Rahbek et al. 2007, Rangel et al. 2007, Gotelli et al. 2009). Such models also generate predictions on the structure of dispersion fields. Comparing both richness patterns and dispersion fields to modeled patterns thus constitutes an opportunity for validation of these models at two hierarchical levels, a standard for pattern-oriented modeling (Grimm et al. 2005). A promising approach for investigating the link between range sizes and richness patterns has recently been developed by Arita et al. (2008). This approach is based on dispersion fields, and starts with the presence absence matrix of sites versus species. In this matrix, the columns are sites, rows are species, and the matrix elements represent the presence (1) or absence (0) of a given species in a particular site (Gotelli 2000). The strength of the presence absence matrix is that it combines information on species richness (which are the column sums), range sizes (which are the row sums), and the co-occurrence of species (which can be measured by the degree of co-variance in the matrix). However, the presence absence matrix is not easy to visualize graphically.

Figure 1. Illustration of the dispersal fields and diversity fields of Graves and Rahbek (2005) and Arita et al. (2008), respectively. Top left: a schematic illustration of elliptical ranges. Vertical lines indicate two focal cells (marked as red squares) for the dispersion fields shown in the bottom panels. A red ellipse illustrates a focal range for one diversity field. Ranges of the same color belong to the same diversity field. One range belongs to both diversity fields. Top right: the richness map resulting from overlap of the ranges in top left panel. One range is illustrated as a red outline. The richness values within this outline constitute the diversity field. Bottom left: the dispersion field for the leftmost focal cell in the top left panel. Bottom right: the dispersion field for the rightmost focal cell in the top panel.

One approach is to create ‘‘range-diversity plots’’ (Arita et al. 2008), which are scatter plots that combine information from the columns and rows of the presenceabsence matrix (Fig. 2, from Arita et al. 2008). The dispersion field of a site can be calculated from the presence absence matrix, as the mean range size of all species that occur at the site. A ‘‘by sites’’ range-diversity plot is then created by plotting this value against the species richness of the site (similar to Fig. 2 of Graves and Rahbek 2005). Because dispersion fields allow range sizes

to be expressed in the same currency as richness values (i.e. individual sites), the range-diversity plot makes it possible to investigate their relationship directly. Analogously to the ‘‘dispersion field’’ of Graves and Rahbek (2005), Arita et al. (2008) developed the concept of the ‘‘diversity field’’, which is the set of richness values of sites within the range of a given species (Fig. 1 upper right). The diversity field is illustrated in a ‘‘by species’’ rangediversity plot, which plots the mean species richness of sites occupied by a species against the range size of that

IBS SPECIAL ISSUE

Figure 2. The range-diversity plot presented by Arita et al. (2008). Left: the range-diversity plot by species for North American mammals. See text or (Arita et al. 2008) for a description of the constraint lines in the plot. Right: the range-diversity plot for sites; same dataset as left panel. Whole plot taken directly from Arita et al. (2008).

403

ISSUE

IBS SPECIAL

species (Fig. 2 left). ‘‘By sites’’ and ‘‘by species’’ rangediversity plots are complementary, and together visualize the presence absence matrix. The main strength of range-diversity plots is that they clarify the connection between richness values and range sizes: range sizes and species richness values of a community are linked because they are the marginal sums of the presence absence matrix. Because of this link, the points in range-diversity plots become constrained to certain regions of the plot area. Thus, the dispersion of points over the plot region can be interpreted to yield information on the ecological processes structuring the assemblage. To facilitate the interpretation of range-diversity plots, Arita et al. (2008) mathematically developed a set of constraint lines for point dispersion. First, they use a thick, solid line to mark regions of the plot that cannot be occupied by points. These areas represent impossible combinations of e.g. mean range and species richness, and are calculated as a mathematical function of the minimum and maximum values of range size or species richness (Fig. 2). Additionally, Arita et al. (2008) added thin lines that connect areas of equal covariance. They argue that what creates the dispersion of points in range-diversity plots is covariance in the presence-absence matrix, which reflects associations between species (in the ‘‘by species’’ plot) or similitude between sites (in the ‘‘by sites’’ plot). Accordingly, species with a similar degree of covariance between their geographic distribution and the distributions of all other species should align along these lines, when observing a ‘‘by species’’ plot. In the range-diversity plots presented by Arita et al. (2008), the points are widely dispersed across the plot area (Fig. 2). The point clouds have characteristic shapes, and the points all lie within a region that is clearly smaller than the permissible area delineated by the thick line. Also, most points are located to the right of the ‘‘fill’’ line (showing the grand mean), which indicates positive covariance for both sites and species. Arita et al. (2008) interpret the patterns of point dispersion in range-diversity plots as the results of ecological processes. For instance, they argue that the general orientation of points in the ‘‘by sites’’ plot (Fig. 2 right) is created by a combination of Rapoport’s rule and the latitudinal gradient of species richness. However, to argue that patterns are created by ecological processes, and not by mathematical constraints on point dispersion, requires that the mathematical constraints are well described. Also, for range-diversity plots to be a useful tool, they should reveal patterns that result from a relationship between range sizes and richness values. If the patterns in range-diversity plots are just functions of the range size and richness frequency distributions themselves, it would be simpler to investigate these distributions separately. Thus, to evaluate the analytical power of range-diversity plots, the pertinent questions are: a) is the entire area within the solid line available for points, or is point dispersion constrained by other factors? And b) is the pattern of point dispersion created by associations among species/sites, or does it result from some other aspect of the calculation of range-diversity plots? 404

To answer these questions, we constructed rangediversity plots for a high-quality dataset of the birds of South America (Rahbek and Graves 2001). This dataset contains 2869 species in 1676 one-degree grid cells. Thus, they contain fewer sites but more species than the dataset for North American mammals used by Arita et al. (2008). Of these 2869 species, 643 have ranges that extend into Central and North America (this was generally only a small part of their range). For these species, we considered only the range within South America, even though this leads to a, for most species, minor underestimation of their actual range. Repeating the analyses using only South American endemics does not affect any conclusions here (unpubl.).

The range-diversity plot by species The ‘‘by species’’ range-diversity plot shows a pattern similar to Arita et al.’s (2008) pattern for North American mammals, although there are more points in the lower left corner, indicating a group of small-ranged species that occur in grid cells with low overall diversity (compare Fig. 2 left with Fig. 3 top left). The points fall within a conical shape, with most points located towards the right part of the plot. A general difficulty for the interpretation of range-diversity plots is that no standard statistical tests exist for testing point dispersion or the degree to which covariance lines explain the location of points. Still, for South American birds the points do not appear to follow the iso-covariance lines even at visual inspection. Instead the points appear to be constrained along a straight line at the right edge of the cloud of points. However, this constraint lies far from the solid line indicating the permissible area. In addition to the permissible area constraint developed by Arita et al. (2008), the potential combinations of range size and mean diversity will also be constrained by the empirical distribution of richness values. A simple way of describing the mechanism for this is to start with a species with a range size of only one grid cell. The highest possible value of mean species richness at occupied sites (the x axis of the range-diversity plot) will then equal the highest species richness value on the map. If the range size is two grid cells, the highest mean richness value is the mean of the two most species-rich areas, and so forth. The highest richness reflected in our bird data is a grid cell close to Quito in Ecuador, and it is extremely high: 845 bird species within a one-degree latitudinal-longitudinal square. In the second most speciose cell, richness already decreases to 782 species, and the maximum value for mean species richness quickly decreases as more cells are included. We added this constraint to the plot (blue and red in Fig. 3). The fit of the constraint line to the point cloud is visually striking. This indicates that the richness frequency distribution is what constrains the point dispersion in the ‘‘by species’’ plot for South American birds. The range size distribution is also very clearly visible, as it is the distribution of points along the y-axis. As Arita et al. (2008) clearly point out, ranges are proportions, and proportion data tend to stack up as they near the limits (0 and 1). At continental scales, this effect results in a highly skewed range-size distribution, with a large number of very

Figure 3. Top left: the range-diversity plot by species for South American birds. Plots show isoclines connecting areas of equal covariance and the mathematical constraint lines developed by Arita et al. (2008). An envelope delimiting the possible values is added in blue and red (see text for explanation of this null constraint). Top right: same as top left, except that range size values have been logit-transformed to remove the effect of skew in the range size frequency distribution. Bottom left: the range-diversity plot by sites for South American birds. The maximal constraint line of a null model similar the one shown in the top left panel is shown in red. Bottom right: same as bottom left, with points added indicating the results of two null models. The green points indicate a simple null model, accounting for only the higher contribution of larger-ranging species to the calculations. The blue points indicate a ‘‘spreading dye’’ model, which also incorporates the effects of the continuous nature of individual ranges.

The ‘‘by sites’’ plot (Fig. 3 bottom) for South American birds is very different, both from the ‘‘by species’’ plot and from Arita et al.’s (2008) plot for North American mammals (Fig. 2). The most striking differences are the absence of points around and to the left of the ‘‘fill’’ line,

405

ISSUE

The range-diversity plot by sites

and the pattern of point dispersion: the points fall into two separate point clouds with diverging shapes. In Fig. 3 (bottom left), we have also added constraint lines based on the range-size frequency distribution. These constraints follow a rationale similar to that presented for the ‘‘by species’’ plot: if a site contains only one species, the highest possible mean range value is the range size of the largestranging species of the assemblage; if it contains two species, the highest value is the mean of the two largest range sizes, and so on. The right edge of the right-most group of points seems to follow the null constraint line (though at a small distance), but the fit is less convincing than for the ‘‘by species’’ plot. All of the points in the ‘‘by sites’’ plot for South American birds occur to the right of the ‘‘fill’’ line. Arita et al. (2008) argued that all points are expected to cluster around this line in the absence of biological processes, which means that deviations from this line indicate similarity between sites. However, unless the range size distribution is completely symmetric, most points are expected to lie to the right of the ‘‘fill’’ line, simply as a consequence of sampling effects. The reason is that largerranging species exist at more sites, and thus contribute a range size value to more data points in the ‘‘by sites’’ plot (for a discussion of the statistical consequences of range sizes see Jetz and Rahbek 2002, Lennon et al. 2004). As a result, each site does not randomly sample the range size frequency distribution of the entire assemblage. Thus, the mean range size of each site will generally be higher than the grand mean range size of the assemblage. Figure 3 (bottom right)

IBS SPECIAL

small-ranged species (Gaston 1996, Graves and Rahbek 2005). The strong skew moves most points towards the bottom of the range-diversity plot, which potentially obscures any pattern between the points. To increase the linearity of proportion data, logit-transformation is often recommended prior to analysis (Sokal and Rohlf 1995). For South American birds, logit-transforming the range sizes yields a much more uniform pattern (Fig. 3 top right). The fit of the null model is even more apparent, and the points are evenly dispersed within the area between the null model lines. Although there are points everywhere between the red and blue lines, there remains an overweight of points to the right of the ‘‘fill’’ line (Fig. 3 top right). Still, there is no strong evidence for an effect of biological inter-dependence between range size and mean species richness. Thus, the results indicate that the ‘‘by species’’ range-diversity plot for the South American bird fauna does not convey any additional information above that of the range size distribution.

ISSUE

IBS SPECIAL

demonstrates this effect using a simple null model, which allocates sites randomly to each species while maintaining the empirical range size distribution (shown as green points). Even though the effect seems to explain why the points are located in the right side of the plot, it cannot account for the dispersion of points. The empirical points are widely dispersed in the plots, whereas the null model points are tightly clustered. The wide dispersion of points in the ‘‘by sites’’ plot reveals a high level of spatial structure of the assemblage. Sites with high similarity are grouped together, whereas sites with a very different species composition are located further apart. The complex pattern of points indicates a possible role for ecological assembly processes in structuring species composition. However, a high degree of site similarity, and thus point dispersion, could also be generated simply by range cohesiveness. Species ranges usually consist of several closely located cells, and thus random overlap is expected to lead to adjacent cells being highly similar. To investigate this effect, we created 2869 random species range maps using a spreading dye algorithm (Jetz and Rahbek 2001). This algorithm randomly places cohesive ranges on the geographic domain, while maintaining the empirical range size frequency distribution. We then generated rangediversity plots for this dataset, adding the points to Fig. 3 (blue points, bottom right panel). The set of points generated by this null model are much more dispersed across the plot. The center of gravity for the points is close to the points created without range cohesion (in green), and thus it does not seem that range cohesion in itself is responsible for the covariance of sites (for a discussion of range cohesion and the ‘‘by species’’ range-diversity plot, see Villalobos and Arita in press). Incorporating range cohesion clearly generates a more realistic level of point dispersion. However, it still does not capture the empirical pattern. The empirical pattern thus probably reflects historical or ecological processes that have generated two disparate areas in South America: one large area where species demonstrate a high degree of nestedness in their distribution, and one in which there is very little nestedness. The present analysis identified structural constraints on the point dispersion in range diversity plots, beyond those considered by Arita et al. (2008). For our dataset of South American birds, these constraints were more important for structuring range-diversity plots than were the constraint lines described by Arita et al. (2008). Given the diversity of data sets that can be analyzed with range-diversity plots, a promising research avenue is to investigate how these results generalize. For instance, spatial scale is known to be an important determinant of ecological patterns (Rahbek 2005, Nogues-Bravo et al. 2008). At finer scales, where range cohesion is lower and biotic interactions are more pronounced, it is likely that other processes drive the patterns in range-diversity plots (H. Arita pers. comm.). The diverse patterns observed in ‘‘by sites’’ plots, which are robust even after accounting for mathematical constraints, highlight the potential of these plots for generating and testing hypotheses on how species’ distributions create patterns of species richness. 406

Acknowledgements The authors thank the Danish National Research Foundation for its support of the Center for Macroecology, Evolution and Climate. We also thank Hector Arita for insightful comments and constructive criticism on the manuscript, which significantly improved the final version.

References Arita, H. T. et al. 2008. Species diversity and distribution in presence absence matrices: mathematical relationships and biological implications. Am. Nat. 172: 519 532. Colwell, R. and Lees, D. C. 2000. The mid-domain effect: geometric constraints on the geography of species richness. Trends Ecol. Evol. 15: 70 76. Diniz-Filho, J. A. F. et al. 2003. Spatial autocorrelation and red herrings in geographical ecology. Global Ecol. Biogeogr. 12: 53 64. Gaston, K. J. 1996. Species-range-size distributions: patterns, mechanisms and implications. Trends Ecol. Evol. 11: 197 201. Gotelli, N. J. 2000. Null model analysis of species co-occurrence patterns. Ecology 81: 2606 2621. Gotelli, N. J. and Graves, G. R. 1996. Null models in ecology. Smithsonian Inst. Press, Washington DC, USA. Gotelli, N. J. and McCabe, D. J. 2002. Species co-occurrence: a meta-analysis of J. M. Diamond’s assembly rules model. Ecology 83: 2091 2096. Gotelli, N. J. et al. 1997. Co-occurrence of australian land birds: Diamond’s assembly rules revisited. Oikos 80: 311 324. Gotelli, N. J. et al. 2009. Patterns and causes of species richness: a general simulation model for macroecology. Ecol. Lett. 12: 873 886. Gotelli, N. J. et al. in press. Macroecological signals of species interactions in the danish avifauna. Proc. Nat. Acad. Sci. USA. Graves, G. R. and Rahbek, C. 2005. Source pool geometry and the assembly of continental avifaunas. Proc. Nat. Acad. Sci. USA 102: 7871 7876. Grimm, V. et al. 2005. Pattern-oriented modeling of agentbased complex systems: lessons from ecology. Science 310: 987 991. Jetz, W. and Rahbek, C. 2001. Geometric constraints explain much of the species richness pattern in african birds. Proc. Nat. Acad. Sci. USA 98: 5661 5666. Jetz, W. and Rahbek, C. 2002. Geographic range size and determinants of avian species richness. Science 297: 1548 1551. Legendre, P. 1993. Spatial autocorrelation: trouble or new paradigm? Ecology 74: 1659 1673. Lennon, J. J. et al. 2004. Contribution of rarity and commonness to patterns of species richness. Ecol. Lett. 7: 81 87. Nogues-Bravo, D. et al. 2008. Scale effects and human impact on the elevational species richness gradients. Nature 453: 216 219. Rahbek, C. 2005. The role of spatial scale and the perception of large-scale species-richness patterns. Ecol. Lett. 8: 224 239. Rahbek, C. and Graves, G. R. 2000. Detection of macroecological patterns in south american hummingbirds is affected by spatial scale. Proc. R. Soc. B 267: 2259 2265. Rahbek, C. and Graves, G. R. 2001. Multiscale assessment of patterns of avian species richness. Proc. Nat. Acad. Sci. USA 98: 4534 4539. Rahbek, C. et al. 2007. Predicting continental-scale patterns of bird species richness with spatially explicit models. Proc. R. Soc. B 274: 165 174.

Rangel, T. F. L. V. et al. 2006. Towards an integrated computational tool for spatial analysis in macroecology and biogeography. Global Ecol. Biogeogr. 15: 321 327. Rangel, T. F. L. V. et al. 2007. Species richness and evolutionary niche dynamics: a spatial pattern-oriented simulation experiment. Am. Nat. 170: 602 616. Sokal, R. R. and Rohlf, F. J. 1995. Biometry: the principles and practice of statistics in biological research. W. H. Freeman.

Svenning, J. C. et al. 2008. Postglacial dispersal limitation of widespread forest plant species in nemoral europe. Ecography 31: 316 326. Villalobos, F. and Arita, H. T. in press. The diversity field of new world leaf-nosed bats (phyllostomidae). Global Ecol. Biogeogr. Webb, C. O. 2000. Exploring the phylogenetic structure of ecological communities: an example for rain forest trees. Am. Nat. 156: 145 155.

IBS SPECIAL ISSUE

407

Ecography 33: 408 419, 2010 doi: 10.1111/j.1600-0587.2010.06434.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Jens-Christian Svenning. Accepted 20 May 2010

Contrasting environmental and regional effects on global pteridophyte and seed plant diversity Holger Kreft, Walter Jetz, Jens Mutke and Wilhelm Barthlott H. Kreft (hkreft@uni-goettingen.de), Biodiversity, Macroecology and Conservation Biogeography Group, Fac. of Forest Sciences and Forest Ecology, Univ. of Go¨ttingen, Bu¨sgenweg 2, DE-37077 Go¨ttingen, Germany, and Nees Inst. for Biodiversity of Plants, Univ. of Bonn, Meckenheimer Allee 170, DE-53115 Bonn, Germany, and Div. of Biological Sciences, Univ. of California San Diego, 9500 Gilman Drive MC 0116, La Jolla, CA 92093-0116, USA. W. Jetz, Div. of Biological Sciences, Univ. of California San Diego, 9500 Gilman Drive MC 0116, La Jolla, CA 92093-0116, USA, and Dept of Ecology and Evolutionary Biology, Yale Univ., 165 Prospect Street, New Haven, CT 065208106, USA. J. Mutke and W. Barthlott, Nees Inst. for Biodiversity of Plants, Univ. of Bonn, Meckenheimer Allee 170, DE-53115 Bonn, Germany.

IBS SPECIAL

ISSUE

Pteridophytes (ferns and fern-allies) represent the second-largest group of vascular plants, but their global biogeography remains poorly studied. Given their functional biology, pteridophytes are expected to show a more pronounced relation to water availability and a higher dispersal ability compared to seed plants. We test these assertions and document the global pattern of pteridophyte richness across 195 mainland and 106 island regions. Using non-spatial and spatial simple and multiple regression models, we analyze geographic trends in pteridophyte and seed plant richness as well as pteridophyte proportions in relation to environmental and regional variables. We find that pteridophyte and seed plant richness are geographically strongly correlated (all floras: r 0.68, mainland: r 0.82, island floras: r 0.77), but that the proportions of pteridophytes in vascular plant floras vary considerably (0 70%). Islands (mean 15.3%) have significantly higher proportions of pteridophytes than mainland regions (mean 3.6%). While the relative proportions of pteridophytes on islands show a positive relationship with geographic isolation, proportions in mainland floras increase most strongly along gradients of water availability. Pteridophyte richness peaks in humid tropical mountainous regions and is lowest in deserts, arctic regions, and on remote oceanic islands. Regions with Mediterranean climate, outstanding extra-tropical centres of seed plant richness, are comparatively poor in pteridophytes. Overall, water-energy variables and topographical complexity are core predictors of both mainland pteridophyte and seed plant richness. Significant residual richness across biogeographic regions points to an important role of idiosyncratic regional effects. Although the same variables emerge as core predictors of pteridophyte and seed plant richness, water availability is clearly a much stronger constraint of pteridophyte richness. We discuss the different limitations of gametophytes and sporophytes that might have limited the ability of pteridophytes to extensively diversify under harsh environmental conditions. Our results point to an important role of taxon-specific functional traits in defining global richness gradients.

The disproportionately higher species richness of some places compared to others is among the most prominent, yet unresolved questions in biogeography and ecology. An increase of species richness from the poles to the equator known as the latitudinal diversity gradient has been documented for a variety of different groups of organisms (reviewed in Hillebrand 2004), and a growing number of studies document strong associations between regional species numbers and different climatic and other environmental variables (among many others: Currie and Paquin 1987, O’Brien 1993, 1998, Kerr and Packer 1997, Jetz and Rahbek 2002, Hawkins et al. 2003, Kreft and Jetz 2007, Field et al. 2008). Contemporary climatic factors, particularly ambient energy and water availability may constrain the number of individuals and thereby species that can coexist in a region or may limit the number of species that 408

are able to tolerate specific local conditions or might have prevented many clades to adapt to colder and drier climates (Wright 1983, O’Brien 1993, 1998, Mittelbach et al. 2001, Jetz and Rahbek 2002, Hawkins et al. 2003, Currie et al. 2004, Wiens and Donoghue 2004). Second, variation in the degree of topographic heterogeneity and associated spatial variability in climate may cause differences in richness through local species turnover (Ricklefs 1987, Kerr and Packer 1997). Finally, climate and habitat heterogeneity may drive regional differences in speciation rates (Qian and Ricklefs 2000, Ricklefs 2004). Alternative hypotheses emphasize the importance of the idiosyncratic biogeographic history of regions caused, e.g. by plate tectonics, mountain uplift, glacial extinction, and postglacial dispersal (Ricklefs 1987, 2004, Jansson 2003, Wiens and Donoghue 2004, Fine and Ree 2006, Hughes and

409

ISSUE

2006). Additionally, important insights also come from tropical elevational gradients consistently demonstrating peaks of pteridophyte diversity in mid-elevation montane forests at ca 1800 2400 m a.s.l. (Kessler 2000, Kessler et al. 2001, Hemp 2002, Bhattarai et al. 2004a, Kluge et al. 2006). Common explanations for this consistent pattern are maximal levels of humidity or a favourable combination of high humidity and mild temperatures (Kessler et al. 2001, Bhattarai et al. 2004b), while others have attributed midelevation peaks in richness to geometric constraints on range locations (Watkins et al. 2006). Together these results point to a strong effect of water availability on the broadscale geographic distribution of pteridophyte richness. However, these studies mainly investigate tropical or subtropical pteridophyte floras not accounting for the full global spectrum of ambient energy, water availability, or habitat heterogeneity (Kerr and Packer 1997, Hawkins et al. 2003, Kreft and Jetz 2007). However, the global distribution of pteridophyte diversity and its relationship with environmental and regional factors remains unquantified. In this paper, we analyze the global distribution of pteridophyte richness and relate it to abiotic variables. Our analyses are based on a comprehensive global data set of the diversity of pteridophytes and seed plants in 301 regional floras worldwide. A simple expectation about the geographic variation in pteridophyte richness might be that diversity patterns of pteridophytes and seed plants are largely indistinguishable either resulting from similar ecological responses to climatic constraints or from the similar timing of the relevant portions of pteridophyte and seed plant clades. This would then predict a strong correlation between pteridophytes and seed plants, similar proportions of pteridophytes in all floras worldwide (13 000 pteridophyte spp./320 000 vascular plant spp. : 4.1%), and consequently similar relationships between richness and environment in both groups. Such relationships could also be expected (all other things being equal) under a simplistic scenario of deep-time ‘‘evolutionary niche conservatism’’ (Wiens and Donoghue 2004), as both pteridophytes and seed plants originated under tropical humid conditions and additional adaptations were necessary for the colonization of and diversification in drier and colder habitats. However, pteridophytes and seed plants (gymnosperms and angiosperms) show major differences in their morphology, ecophysiology, and reproductive biology (Page 2002). Unlike seed plants, pteridophytes have two independent, free-living life stages, the haploid gametophyte and the diploid sporophyte. The free-living gametophytes are fertilized by motile spermatozoids that need a water film for movement. Another major difference is that pteridophytes have small, wind-dispersed diaspores and are not dependent on animal vectors for pollination and seed dispersal (Tryon 1970, 1986). Animal pollination and dispersal, on the other hand, are considered key innovations for an explosive diversification within angiosperms (Willis and McElwain 2002). All these differences are likely to have an influence on how pteridophytes and seed plants have responded to opportunities of diversifying into habitats with more extreme climatic conditions. This makes pteridophytes and seed plants suitable model groups to explore potential influences of taxon-specific ecological and life-history differences on the distribution of species

IBS SPECIAL

Eastwood 2006, Jansson and Davies 2008). An intrinsic complication for understanding the specific role of historical and present factors arises from the fact that many attributes of past and modern environments important for species richness are strongly collinear (Endler 1982). Although there is little doubt that contemporary environmental conditions constrain regional species richness (Hawkins et al. 2003, Currie and Francis 2004, Field et al. 2008), it has been shown that the species richness of various groups of organisms responds differently to these gradients (Currie 1991, Jetz et al. 2009, Kissling et al. 2009). This suggests that further insights into the ecological and evolutionary drivers of cross-taxon differences in richness might contribute towards a more mechanistic understanding of richness gradients. Here, we use the global pteridophyte flora as a model and compare it to seed plants in order to investigate how contemporary environment interacts with taxon-specific ecological and evolutionary constraints in shaping the global distribution of species richness. With an estimated 13 000 species (Smith et al. 2006), pteridophytes (ferns and fern allies, a paraphyletic group including monilophytes and lycophytes) form the second largest group within the estimated ca 320 000 vascular plant species (Prance 2001) and are at the base of the phylogeny of vascular plants (Pryer et al. 2001). With an age of ca 400 million yr, pteridophytes are considerably older than flowering plants and dominated landscapes during the Carboniferous and Permian (Niklas et al. 1983). In contrast to the common notion of extant pteridophyte diversity being a remnant of a very old diversification (Niklas et al. 1983), recent studies into the molecular evolution and phylogeny of pteridophytes (reviewed in Pryer et al. 2004, Smith et al. 2006) have revealed that the extant diversity of pteridophytes is a product of a rather recent diversification that paralleled the rise of angiosperms, which now dominate global landscapes (Schneider et al. 2004a). This has challenged traditional perspectives on the evolution and biogeography for pteridophytes. According to Schneider et al. (2004a), two evolutionary scenarios are invoked to explain the secondary diversification in pteridophytes: first, the rise of the angiosperms in the Cretaceous might have created habitats with greater structural complexity and new ‘‘niche space’’ in which various clades of pteridophytes could have diversified. Alternatively, major changes in the global environment (tectonic uplifts, changes in climate and CO2 concentration) could have triggered the diversification of ferns. Recent regional to continental-scale studies have provided key insights into the biogeography of pteridophytes. Aldasoro et al. (2004) found strong relationships between pteridophyte richness and humidity as well as with the distance to proposed glacial rainforest refugia in SubSaharan Africa. These findings are congruent with a study of pteridophytes in Ugandan rainforests where additionally strong controls by soil properties were found (Lwanga et al. 1998). Strong controls of diversity and composition by soil properties and micro-topography have also been found in Amazonian pteridophyte communities (Tuomisto and Poulsen 1996, 2000, Tuomisto et al. 2003). A strong effect of water availability on pteridophyte richness has been also demonstrated for the Australian flora (Bickford and Laffan

richness. In this study, we ask the following questions: 1) how is pteridophyte diversity distributed at a global scale? 2) What are core abiotic correlates of this pattern? 3) How does the relative proportion of pteridophytes in vascular plant floras vary geographically and along environmental gradients? 4) Given their stronger dependence on water, are pteridophytes greater in proportional richness in humid environments? 5) Given their different dispersal strategies, what is the difference in relative richness on islands compared to mainland regions and along gradients of isolation?

Methods Diversity data Complementary to a database that was assembled to estimate global patterns of overall vascular plant diversity and analyze its correlates (Barthlott et al. 1996, 2005, Kier et al. 2005, Mutke and Barthlott 2005, Kreft and Jetz 2007, Kreft et al. 2008), we collected information on species numbers of pteridophytes and seed plants for 301 geographic regions worldwide (195 mainland and 106 island floras; compare Fig. 1). The primary source for species numbers was the Checklist of world ferns (Hassler and Swale 2001, 2004) which includes ferns (i.e. monilophytes) as well as lycophytes. Additionally, floras, checklists, and compilations of species richness accounts were exploited

(compare Supplementary material Table S3). The data set covers a broad spectrum of abiotic conditions (Fig. 1). The number of seed plant species was derived by subtracting the number of pteridophyte species from the total number of vascular plant species. Abiotic data Boundaries of all geographic regions were digitalized in a geographic information system, and AREA (km2) as well as latitude and longitude of the centroid were calculated for each geographic unit. We analyzed a number of potential abiotic determinants of pteridophyte and seed plant richness, which have been shown to be strong predictors of overall vascular plant richness (Francis and Currie 2003, Hawkins et al. 2003, Kreft and Jetz 2007, Field et al. 2008). As climatic variables, we analyzed variables associated with ambient heat: mean annual temperature (8C; TMP), and potential evapotranspiration (mm yr 1; PET); water availability: mean annual precipitation (mm yr 1; PRE), number of days per year with precipitation 0.1 mm (n; WET); and integrative variables of water-energy dynamics: actual evapotranspiration (mm yr 1; AET), and water balance (mm yr 1; WTB). Mean values of climatic variables for all investigated geographic regions were derived from a global high-resolution climatology at 10? resolution (New et al. 2002). AET, PET, and WTB data were

(a)

Number of pteridophyte species

ISSUE

0–5

5–20

20–50

50–100

100–200

200–500

500–750 750 –1500

1500–1875

IBS SPECIAL

(b)

Proportion of pteridophytes (%) 0–2

3–5

6–10

11–15

16–20

21–30

31–35

36–40

41–71

Figure 1. Spatial location of 301 floras analyzed in this study (195 mainland and 106 island floras). Circles represent mass centroids of each geographic unit. (a) Number of pteridophyte species per geographic unit, (b) relative proportion of pteridophytes in vascular plant floras.

410

Statistical analysis

Results Global distribution of pteridophyte richness Pteridophytes occurred in almost all terrestrial habitats except for some parts of sandy deserts (most arid parts of the Arabian Peninsula), Antarctica, and some remote atolls (Fig. 1). Pteridophyte richness was highly unevenly distributed across the globe with a pronounced latitudinal gradient and peaked in tropical, humid regions containing complex topography (tropical Andes, Mesoamerica, Himalaya and south east Asia; Fig. 1b). Compared to other tropical regions, the African tropics showed only moderate levels of species richness. Lowest richness was encountered in hot and dry climates as well as in high arctic regions. There was also an apparent difference between the species richness of pteridophytes in mainland and island regions. Islands and especially those of oceanic origin were characterized by fewer pteridophyte species as compared to neighbouring mainland regions (Fig. 1b). Some large tropical land-bridge islands such as New Guinea, Borneo, and Sumatra, on the other hand, had very high species numbers and were among the globally most diverse regions in terms of pteridophyte richness. 411

ISSUE

We analyzed the association of pteridophyte and seed plant richness both in absolute terms and as proportion of overall vascular plant richness with environmental variables using single- and multi-predictor statistical models. Since the geographic units differed in size, regression analyses of species richness were performed including AREA as a covariate. An inherent feature of macroecological data sets is the presence of spatial autocorrelation in the response variable (e.g. species richness), in predictor variables (e.g. climate, topography), and most importantly in model residuals

(Legendre 1993, Lennon 2000). Since spatial autocorrelation was present in our data, we confirmed our results from traditional Generalized Linear Models (GLMs) using Simultaneous Autoregressive models (SARs) of the ‘‘errormodel’’ type which have been shown to represent a powerful statistical approach in dealing with spatial autocorrelation in macroecological data sets (Lichstein et al. 2002, Tognelli and Kelt 2004, Kreft and Jetz 2007, Kissling and Carl 2008). Since to date no theory-based selection procedure exists to choose among different SAR types in ecology, all three different types of SAR were evaluated (Lichstein et al. 2002, Kissling and Carl 2008). Model selection was based on model fit using Akaike information criterion (AIC) and on the reduction of spatial autocorrelation in model residuals (Kissling and Carl 2008). SARs of the ‘‘error model’’ type with a weighted neighbourhood structure and a lag-distance of 1500 km best accounted for the spatial structure in our data set. Spatial autocorrelation was evaluated using Moran’s I correlograms and global Moran’s I values. The goodnessof-fit of statistical models was assessed using AIC (Johnson and Omland 2004), a measure that evaluates the relative statistical support based on model fit and complexity. More complex models were given relatively more statistical support, if the DAIC was 2. Additionally, we report r2values, which for spatial models represent the non-spatial trend of the fitted model without including the spatial signal (Haining 1990). For SARs, these represent squared Pearson correlation coefficients between the nonspatial component of the SAR prediction and observed values. Statistical analyses were performed in R ver. 2.7 (R Development Core Team 2005) using the spdep library (Bivand 2006) for spatial analyses.

IBS SPECIAL

obtained from a global database provided by UNEP-GRID at 0.5 degree resolution (Ahn and Tateishi 1994, Tateishi and Ahn 1996). Two variables were investigated describing the heterogeneity of the landscape. First, as a proxy of ecosystem diversity (ECODIV), we counted the number of different ecosystems in each geographic unit from the GLC2000 global land cover data set at 30ƒ resolution (European Commission Joint Research Centre 2002). Topographic diversity (TOPODIV) measured as elevational range (difference between maximum and minimum elevation) was derived from the GTOPO data set at a 30ƒ resolution (USGS 1996). This measure encapsulates the diversity of landscapes within a region and additionally captures the potential for spatio-temporal climatic dynamics (Rahbek and Graves 2001, Jetz and Rahbek 2002). This measure is a strong correlate of overall vascular plant richness at broad geographic scales (Kreft and Jetz 2007). Each geographic region was further assigned to one major biome (Olson et al. 2001).The extraction of predictor variables across the boundaries of all geographic units was performed in ArcGIS/ArcINFO. To account for potential regional effects and different biogeographic histories in our analyses, we investigated biogeographic realms (REALM) (compare Ricklefs et al. 2004, Kreft and Jetz 2007, Hortal et al. 2008, Qian 2008). Therefore, each geographic region was assigned to one of the following five biogeographic realms following the delineations of Olson et al. (2001): Afrotropics (AFT), Indo-Malaya (IND), Nearctic (NEA), Neotropics (NET), Palaearctic (PAA). Due to insufficient data points, Australia was not considered in this part of the analysis. Since most of the environmental variables were not available for small oceanic islands, we restricted these analyses to the mainland part of the data set. To test for a potential relationship between isolation and pteridophyte proportions, we calculated the Euclidean distance to the next continental landmass for all islands assuming that continental floras are the main sources for island colonisations (Whittaker and Ferna´ndez-Palacios 2007). Dependent (species richness of pteridophyte and seed plants, pteridophyte proportions) and continuous independent variables were log10-transformed for analysis to meet the assumption of normality of model residuals and to improve the linearity of models. Proportions are often analyzed using arcsine transformation (Sokal and Rohlf 1981), but in our data this did not improve the behaviour of model residuals and model fits.

Proportional richness of pteridophytes richness across islands, mainlands, and environments

Seed plant species

In general, there was a strong positive relationship between pteridophyte richness and seed plant richness (all floras: r 0.68, mainland: r 0.82, island floras: r 0.77; all pB0.001; Fig. 2a, b). However, the relative proportions of pteridophytes varied considerably and ranged between 0 and 70% and differed strongly between mainland and islands floras (Fig. 1a, 2c). Islands had an average pteridophyte proportion of 15.3%912.6 (SD), whereas mainland regions had only 3.6%92.5 (p BB0.001; Mann-Whitney U test). On islands, the proportion of pteridophytes increased with geographic isolation (r 0.41, Fig. 2d). The proportion of pteridophytes in mainland floras strongly co-varied with environmental gradients (Table 1, Fig. 3). Biome membership was a strong predictor of the relative proportion of pteridophytes and explained 53% of the variance (Table 1, compare Supplementary material Table S1 for SAR results, Fig. 3j). Highest proportions were found in montane biomes (mean 8.0%) and tropical moist broad-leaf forests (mean 6.8%), lowest proportions in arid biomes, like deserts (1.4%) and Mediterranean climate regions (2.1%) (Fig. 3j). The generally lower proportions of pteridophytes in more arid biomes was further evidenced by the strong positive relationship between pteridophyte proportion and variables representing water availability (Fig. 3, Table 1). Pteridophyte proportions were also significantly higher in regions with greater topographical complexity (Table 1, Fig. 3h). In a multi-predictor (a)

1000 100 10

mainland floras slope: 0.54; r = 0.82

Proportion pteridophytes (%)

ISSUE

IBS SPECIAL

(b)

10000

10 100 1000 Pteridophyte species

(c)

island floras slope: 0.79; r = 0.77

10 100 1000 Pteridophyte species

(d)

island floras r = 0.41

50 40 30 20 10 0 mainland

island

100

1000

Isolation (km)

Figure 2. Relationship between the species richness of pteridophytes and seed plants of (a) mainland (n 195) and (b) island floras (n 106). (c) Box-and-whisker plots of pteridophyte proportions in mainland and island floras. Islands have significantly higher proportions (p BB0.001; Mann-Whitney U test). (d) Relationship between geographic isolation measured as distance to the nearest mainland (km) and proportions of pteridophytes in island floras. Linear fits with according correlation coefficients and 95%-confidence intervals are shown.

412

context, a model containing main effects of PET, PRE and TOPODIV had the highest relative support and together explained between 54 and 55% of the global variance in pteridophyte proportions in GLM and SAR, respectively (Table 2, Supplementary material Table S2). Environmental predictors of mainland pteridophyte species richness AREA had only a moderate effect on pteridophyte richness (Table 1). The slopes of the species area relationship were very similar in both statistical approaches (GLM: 0.189 0.05, SAR: 0.1790.04) and comparable to previously reported values for plants (Rosenzweig 1995, Kier et al. 2005). Both pteridophytes and seed plant richness patterns were strongly related to environmental conditions (Table 1). In pteridophytes, mean annual precipitation (PRE) was the strongest single-predictor (Table 1). Other terms representing water availability, single variables describing waterenergy dynamics, or interaction terms between temperature and water-related variables also showed high or even slightly higher correlations with pteridophyte richness (Table 1). Variables representing only ambient energy either showed non-significant or minor effects on pteridophyte richness. Variables describing the heterogeneity of the landscape had significant positive yet weaker effects on pteridophyte richness (Table 1). High correlations were again observed with BIOME membership (r2 0.60). REALM was a significant predictor of pteridophyte richness accounting for about half of the variance. There were noteworthy differences in the effects of certain predictor variables or terms on the richness of pteridophytes and seed plants, respectively (Table 1). Most importantly, variables representing water availability and terms encapsulating both water and ambient energy had consistently stronger effects on pteridophyte than on seed plant richness. Terms representing ambient energy or heterogeneity in turn tended to be stronger predictors for seed plant richness. The effect of regional factors tended to be more pronounced in pteridophytes. We continued to construct an ad-hoc statistical model of pteridophyte richness that incorporated AREA as a covariate. We then added terms from each of the categories ambient energy, water, and heterogeneity and tested all possible combination. In the context of a model including AREA and TOPODIV, the main effect terms of PET and WET were the strongest predictors of pteridophyte richness. Models of alternative combinations of the variables TEMP, PET, PRE, AET, or WTB, as well as models including interactions yielded lower statistical support. The resulting multi-predictor model including AREA, PET, WET, and TOPODIV accounted for 77% (GLM) and 78% (SAR) of the global variance in pteridophyte richness (Table 2, Supplementary material Table S2). Other variables as well as interaction terms failed to enter the model due to insufficient support based on AIC values. After controlling for the above mentioned dissimilarities in area size and contemporary environmental conditions (Table 2), significant differences in residual species richness remained across the biogeographical realms (compare

Table 1. Single predictor non-spatial Generalized Linear Models (GLM) of pteridophyte proportions, pteridophyte species richness, and seed plant species richness for mainland floras. Pteridophyte proportions were log10(x 1)-transformed. While there was no significant effect of AREA on pteridophyte proportions, richness models included AREA as a covariate to control for disparities in area size. Variable abbreviations: AREA area size of the geographic unit in km2, TEMP mean annual temperature, PET potential evapotranspiration (mm yr 1), PRE mean annual precipitation (mm), WET annual number of days with precipitation, AET actual evapotranspiration (mm yr 1), WTB water balance (mm yr 1), ECODIV number of land cover classes (n), TOPODIV elevational range (m), BIOME biome membership, REALM biogeographic realm. All continuous explanatory variables were log10-transformed. Explanatory variables

Pteridophyte proportion t

Null AREA

1.61

0.01

TEMP PET

1.39 0.15

PRE WET AET WTB TEMP PRE TEMP WET PET PRE PET WET

Pteridophyte richness

Seed plant richness

AIC

27 28

3.15

0.05

319 311

6.78

0.19

155 116

0.01 0.00

27 25

Ambient energy 3.12 0.06 3.04 0.10

311 303

6.95 6.99

0.26 0.31

101 87

14.22 12.38

0.52 0.45

165 139

9.4 7.73

90 175

11.51 9.83

0.57 0.43

3 53

12.47 9.06

0.45 0.30 0.53 0.47 0.54 0.49

140 94 166 144 172 149

7.72 5.88

Water-energy 0.66 0.36 0.71 0.67 0.72 0.71

116 236 92 115 83 87

10.91 8.71

0.57 0.34 0.62 0.67 0.64 0.67

3 79 22 53 31 50

TOPODIV ECODIV

1.94 0.52

0.02 0.00

29 26

0.37 0.75

Heterogeneity 0.20 0.11

281 302

3.45 3.71

0.38 0.25

68 104

BIOME REALM

0.53 0.42

147 123

170 177

0.61 0.53

2 20

AIC

Fig. 4b). Including the variable REALM in the multipredictor model of pteridophyte richness yielded considerably strong statistical support indicated by a sharp drop in AIC values (GLM: DAIC 88; SAR: DAIC 82) and explained an additional 8% of the variance. The Neotropical, Indomalayan, and Nearctic realm had significantly higher richness than expected from their current environment than Afrotropics and the Palaearctic realms (Fig. 4b, Tukey HSD test on ANOVA results). Predicting global geographic trends

The use of data derived from checklists and regional floras is not without problems, but given the lack of comprehensive

specimen data and range maps this represents the only source for global-scale macroecological analyses at present. One potential problem is that geographic units differ in size. While AREA had no significant effect on pteridophyte proportions, unsurprisingly AREA did have a significant effect on species numbers. However, this effect was comparatively weak for pteridophytes (r2 0.05, GLM slope of the species area relationship: 0.18), and moderate for seed plants (r2 0.19, slope: 0.23). Environmental and regional variables, on the other hand, surfaced as much stronger predictors (Table 1). A second potential problem is that the global inventory of both pteridophyte and seed plant diversity is far from completed introducing the risk of biased results. It is reasonable to assume that there is a geographic trend in the completeness of floristic accounts. Temperate floras are most likely better documented then highly-diverse humid tropical ones. A more complete floristic inventorying of the latter could even reinforce some of the relationships (namely the already strong correlations between richness and water-energy variables). On the other hand, we consider pteridophyte proportions to be rather robust towards incompleteness unless there is a severe geographic bias in the collection activity between pteridophytes compared to seed plants, with pteridophytes being systematically better collected in island floras, or relatively more under-collected in tropical regions. This said, we have no indication that this is actually the case. Inevitably, the data used stem from different time periods. The effect on our analyses is hard to evaluate, but an important impact would require differential effects on ferns 413

ISSUE

Discussion

Region 0.60 0.55

IBS SPECIAL

Model predictions across an equal area grid of ca 12 000 km2 used the same environmental predictors and model coefficients obtained from the GLM. They illustrated the geographic interplay of the various predictors of pteridophyte proportions (Fig. 5a) and richness (Fig. 5b) as identified from the multivariate statistical modelling. Highest richness values were predicted for montane regions in the humid tropics: New Guinea, northern Andes, Borneo, or the Mesoamerican Isthmus. Lowest values were predicted for the most arid parts of deserts (Sahara, Arabian Peninsula, Thar Desert, and Taklamakan) as well as for high arctic tundra regions.

Water 0.70 0.54

% pteridophytes (log10)

1.2

(a)

(b)

0.8

0.4

0.0 103

104

105

106

% pteridophytes (log10)

AREA (km2) 1.2

100

WET (d yr–1)

(f)

0.4

0.0 10

1000

100

1.2

100

1000

AET (mm yr–1)

(g)

100

1000

WTB (mm yr–1)

(h)

(i)

0.8

0.4

0.0 5

102

ECODIV (n) 1.2

103

104 0

TOPODIV (m)

ABSLAT

(j)

(k)

0.8

0.4

0.0 1

BIOME

ISSUE

0.8

IBS SPECIAL

1000

(e)

PRE (mm yr–1) % pteridophytes (log10)

100

PET (mm yr–1)

(d)

% pteridophytes (log10)

(c)

AFT IND NEA NET PAA

REALM

Figure 3. Single-predictor relationships between selected environmental variables and proportions of pteridophytes (y-axis; log10 x 1scale) for mainland floras. In (a) to (i) linear fits and 95%-confidence intervals are shown (a) AREA area of the geographic unit (km2), (b) PET potential evapotranspiration (mm yr 1), (c) WET wet days per years (d yr 1), (d) PRE mean annual precipitation (mm yr 1), (e) AET actual evapotranspiration (mm yr 1), (f) WTB water balance (mm yr 1; x 1600), (g) ECODIV number of different ecosystems, (h) TOPODIV maximum elevational range (m), (i) ABSLAT absolute geographic latitude, (j) BIOME: 1 tropical and subtropical moist broadleaf forests, 2 tropical and subtropical dry broadleaf forests, 3 tropical and subtropical coniferous forests, 4 temperate broadleaf and mixed forests, 5 temperate coniferous forests, 6 boreal forests/taiga, 7 tropical and subtropical grasslands, savannas, and shrublands, 8 temperate grasslands, savannas, and shrublands, 9 montane grasslands and shrublands, 10 tundra, 11 Mediterranean forests, woodlands and scrub, 12 deserts and xeric shrublands. (k) REALM biogeographic realms: AFT Afrotropic, IND Indomalaya, NEA Nearctic, NET Neotropic, PAA Palearctic. Hatched lines in (j) and (k) mark the global mean.

and seed plants. To assess the completeness and whether there are differences between pteridophytes and seed plants, one could investigate the rates at which new species are added to checklists for available time-series of floristic accounts for the same region or based on specimen data. We consider this beyond the scope of our study and thus hypothesize that the data adequately reflect our current knowledge on diversity patterns in both groups. 414

At first glance, pteridophyte and seed plant richness show very similar geographic patterns. Pteridophytes exhibit a pronounced latitudinal diversity gradient with pteridophyte richness peaking in equatorial regions (Fig. 1a, 3f, 5b) a pattern that is well supported for the majority of plant and animal groups (Humboldt 1808, Pianka 1966, Rohde 1992, Rosenzweig 1995, Hillebrand 2004, Mittelbach et al. 2007). Furthermore, pteridophyte richness varies

Table 2. Non-spatial Generalized Linear Models (GLM) multi-predictor environmental model of pteridophyte proportions and pteridophyte richness [log10(x 1) transformed]. For variable descriptions and transformations, see Table 1 and Methods. The Akaike information criterion (AIC) for the null model containing only the intercept is 27 and 319, respectively. Explanatory variables

Pteridophyte proportions

Pteridophyte richness

Slope

(Intercept) AREA PET PRE WET TOPODIV

( 0.49) 0.12 0.42 0.07

2.71** 2.12* 14.69*** 2.92**

( 6.19) 0.20 1.02 1.42 0.34

15.89*** 6.23*** 10.42*** 21.18*** 7.51***

AIC Model r2 Moran’s I

172 0.54 0.24

43 0.77 0.24

***pB0.001, **pB0.01, *p B0.05.

systematically along gradients of climate and topography. Notably, the combination of PET and WET, which has been previously shown to be the strongest set of predictors for overall vascular plant richness in a previous study on a similar data set (Kreft and Jetz 2007), also emerged as the best variable combination to describe the water-energyrichness relationship for pteridophytes (Table 2). This is not surprising, since the availability of ambient heat and water represent ecophysiological constraints to plant life in general (O’Brien 2006) and is thus expected to affect both groups. Additionally, topographic heterogeneity facilitates the species richness of pteridophytes. These relationships and their geographic interplay are captured in the global richness predictions derived from the environmental model (Fig. 5b). Similar to vascular plants in general (Barthlott et al. 2005, Mutke and Barthlott 2005), centres of pteridophyte diversity are especially concentrated in tropical montane regions. Notably, the tropical Andes or the Indomalayan region known centres for flowering plants (Barthlott et al. 2005, Mutke and Barthlott 2005) also exhibit highest pteridophyte species numbers (Fig. 1a). High correlations between species richness and contemporary environment, as found in the present study, have been interpreted as leaving little room for historical explanations (Francis and Currie 2003). However, this view disregards that climate history and thus rates of speciation, (b) ab

Residuals GLM

0.5

0.0

AFT

IND

NEA NET PAA

Biogeographic realm

AFT

IND

NEA NET PAA

Biogeographic realm

Figure 4. Box-and-whisker plots of the residual variation across biogeographic realms of (a) proportions of pteridophytes and (b) pteridophyte species richness after accounting for differences in contemporary environment (GLM models in Table 2). Characters indicate results from Tukey HSD tests on ANOVA results.

415

ISSUE

–0.5

IBS SPECIAL

(a)

extinction, and dispersal are broadly collinear with contemporary conditions (Endler 1982). This fact makes it extraordinarily difficult to disentangle the specific roles of past and modern conditions on global diversity gradients (Ricklefs 2004). The significant differences in residual richness across biogeographic regions (Fig. 4) demonstrate that an interpretation that is purely based on contemporary environmental determinism is not tenable. For instance, it has been long recognized that the Afrotropics are noticeably poor in pteridophyte species (Tryon 1986). Such cross-realm differences in pteridophyte diversity above and beyond contemporary climate are supported by our results. One plausible explanation for the lower diversity of the African pteridophyte flora might be that it has suffered disproportionately from extinctions during Pleistocene dry periods (Richards 1973, Tryon 1986), an interpretation that is congruent with the results from Aldasoro et al. (2004) who found pteridophyte richness to be higher in closer proximity to Pleistocene rainforest refugia. On a much longer time scale, African rainforests have also suffered greater area losses than the Neotropical or Indomalayan rainforests during the last 55 million yr which might have resulted in overall reduced possibilities for diversification and increased extinction rates (Fine and Ree 2006). Additionally, the smaller extent of humid tropical mountain systems in Africa might have resulted in lower speciation rates. Such historical processes are likely to have left prominent traces in the current distributions of pteridophytes and are important for a unified understanding and interpretation of their diversity patterns. Comparison of our environment-based spatial prediction of pteridophyte richness (Fig. 5b) with expert-drawn (Barthlott et al. 2005) or modelled world maps of vascular plant richness (Kreft and Jetz 2007) revealed important differences between these two plant groups, despite the obvious similarities in their richness gradients. Notably, there was a much steeper increase of species densities of pteridophytes towards the equator compared to overall vascular plant and seed plant richness. In addition, looking at the proportions of pteridophytes in vascular plant floras, two pronounced trends surfaced: 1) compared to seed plants, pteridophytes are more diverse in tropical humid and mountainous regions and less diverse in arid regions and deserts (Fig. 3). 2) Pteridophytes are comparatively over-represented on islands (Fig. 2c). In the following, we

(a)

Proportion of pteridophytes (%) 0–1

1–2

2–3

3–4

4–5

5–6

6–8

(b)

Species richness of pteridophytes <2

2–20

20–50

50–100 100–150 150–200 200–300 300–400 400–500

>500

IBS SPECIAL

ISSUE

Figure 5. Environment-based spatial prediction of (a) proportions of pteridophytes (%) and (b) pteridophyte species richness. Predictions were derived from the multivariate GLM across an equal-area grid of ca 12 000 km2 using the same environmental predictor variables and model parameters as in Table 2.

discuss why pteridophytes may have been less successful than seed plants in diversifying under arid conditions or more successful under tropical humid conditions and what makes pteridophytes so successful on islands compared to seed plants. The strong positive correlations between pteridophyte proportions and variables of water availability (Fig. 3, Table 1, 2) suggest that intrinsic traits related to water availability might disproportionately limit the ability of pteridophytes to persist or undergo extensive diversifications under arid conditions. Importantly, pteridophytes are not completely absent from most arid habitats; and some pteridophyte clades have independently evolved remarkable drought adaptations (Kessler and Siorak 2007). The range of drought adaptations includes desiccation tolerance, xeromorphism, dense pubescence, curling and shedding of leaves, dormancy, shortened gametophytic phase, or geophytic life style (Kornas´ 1977, Nobel 1978, Given 1993). On the other hand, certain morphological and life-history strategies to cope with arid conditions common in flowering plants such as pronounced succulence, deep-rooted perennials, and annual life style are largely absent in pteridophytes. Additionally, most pteridophytes show only poorly controlled evaporative potential throughout most of their life-cycle thereby depending on soil-water availability and high air humidity (Page 2002). In arid environments, this often restricts pteridophytes to the most mesic microsites such as rocky crevices or below rock outcroppings (Nobel 1978). It has been frequently argued that the independent gametophyte stage of pteridophytes is also a disadvantageous trait in arid environments (Page 2002). Unlike seed plants, pteridophytes are not independent of 416

liquid water for fertilization, but require a water film for sexual fertilization by the free-swimming spermatozoids. Therefore, parallels to the amphibian life-cycle have been drawn (‘‘return to the water to breed’’; Page 1985, 2002). Consequently, pteridophytes are restricted to habitats where such conditions frequently occur or require special adaptations to tune the timing of spermatozoid release to periods of the year when conditions are favourable. Furthermore, the ecological potential of the gametophyte itself might be limited. The small gametophyte lacks vascular tissue, produces rhizoids instead of true roots, has poorly developed or non-existent cuticles, and little ability for internal water storage (Raghavan 1989). On the other hand, it has recently been shown that gametophytes of some species are remarkably desiccation-tolerant (Watkins et al. 2007), but how this generalizes across all extant pteridophytes is yet unknown. Mediterranean climate regions (hot and arid summers, cool and humid winters) are well known extra-tropical centres of plant diversity (Cowling et al. 1996) and have a much higher plant diversity than expected from their latitudinal position or contemporary climate (Kreft and Jetz 2007). These same regions, however, are not outstanding diversity centres for pteridophytes (Fig. 3j). In addition to the constraints of harsh summer aridity, factors such as co-evolutionary processes between flowering plants and animal pollinators and seed dispersers discussed as potential drivers of the high Mediterranean flowering plant diversity (Cowling et al. 1996, Linder 2003) do not exist in pteridophytes. Despite the environmental constraints in arid regions, the question why pteridophytes are disproportionately more

in pteridophyte proportions on islands and might thus partly compensate for the higher colonization rates of pteridophytes. The well-investigated vascular plant flora of New Zealand offers additional insights (McGlone et al. 2001). Here, pteridophytes and orchids, both characterized by dust-like diaspores but the latter depending on animal pollinators, are well-represented (McGlone et al. 2001). However, both groups have a significantly lower tendency towards endemism compared to other vascular plants (McGlone et al. 2001). This may suggest that there is a significant gene flow between island and mainland populations in some pteridophyte and orchid species. Relatively frequent long-distance dispersal events might thus be more important in the assembly of insular pteridophyte and orchid floras than animal interactions. In summary, our results provide evidence that taxonspecific functional ecological and evolutionary constraints are reflected in the current distribution of species richness at broad geographic scales. Although similar environmental factors control the diversity patterns of pteridophytes and seed plants, our results suggest an important role for taxonspecific ecological and evolutionary limitations and advantages in defining global richness gradients. Acknowledgements Data compilation was partly funded by the BIOLOG-BIOTA project by the German Federal Ministry of Education and Research (BMBF) and the Academy of Sciences and Literature, Mainz. H. K. received funding and travel support from the German National Academic Foundation and from a Feodor-Lynen postdoctoral fellowship by the Alexander von Humboldt-Foundation. Thanks to Frank A. La Sorte and Jonathan Belmaker for comments on an earlier version of the manuscript. The referees and the editor Jens Christian Svenning are acknowledged for providing very helpful comments and suggestions.

References

417

ISSUE

Ahn, C. H. and Tateishi, R. 1994. Development of a global 30minute grid potential evapotranspiration data set. J. Jap. Soc. Photogr. Remote Sens. 33: 12 21. Aldasoro, J. J. et al. 2004. Diversity and distribution of ferns in sub-Saharan Africa, Madagascar and some islands of the South Atlantic. J. Biogeogr. 31: 1579 1604. Barthlott, W. et al. 1996. Global distribution of species diversity in vascular plants: towards a world map of phytodiversity. Erdkunde 50: 317 328. Barthlott, W. et al. 2005. Global centres of vascular plant diversity. Nova Acta Leopoldina 92: 61 83. Benzing, D. H. 1987. Vascular epiphytism: taxonomic participation and adaptive diversity. Ann. Mo. Bot. Gard. 74: 183 204. Benzing, D. H. 1990. Vascular epiphytes. Cambridge Univ. Press. Bhattarai, K. R. et al. 2004a. Fern species richness along a central Himalayan elevational gradient, Nepal. J. Biogeogr. 31: 389 400. Bhattarai, K. R. et al. 2004b. Relationship between plant species richness and biomass in an arid sub-alpine grassland of the central Himalayas, Nepal. Folia Geobot. 39: 57 71. Bickford, S. A. and Laffan, S. W. 2006. Multi-extent analysis of the relationship between pteridophyte species richness and climate. Global Ecol. Biogeogr. 15: 588 601.

IBS SPECIAL

diverse in tropical regions is equally important. A main driver of the extraordinarily high species richness in tropical montane habitats is undoubtedly the pronounced tendency of pteridophytes towards an epiphytic life style (Benzing 1990). Approximately 20% of all pteridophyte species are epiphytes, twice as many as found in seed plants (Schuettpelz and Pryer 2009). In some species-rich families like Hymenophyllaceae, Aspleniaceae, or Polypodiaceae the proportion of epiphytic species even ranges between 59 and 93% (Madison 1977, Kress 1986, Gentry and Dodson 1987). Tropical montane forests in turn are known as globally outstanding centres of epiphyte diversity (Gentry and Dodson 1987, Kreft et al. 2004, KuÂ¨ per et al. 2004). Furthermore, the bulge in epiphyte diversity in midelevation cloud forests has been primarily attributed to the high humidity and absence of temperature constraints, which both facilitate the richness of an epiphytic life style (Gentry and Dodson 1987). Epiphytic habit as an independently evolved key innovation has triggered an explosive radiation in many tropical fern clades (Schneider et al. 2004b, Schuettpelz and Pryer 2009). Adaptive features for their great success as epiphytes include: 1) small, winddispersed diaspores, 2) independence from pollinators, 3) tendency of poikilohydry, 4) occurrence of xeric leaves, thick cuticles, succulent rhizomes, leaf scales and absorbing foliar trichomes, and 5) shade tolerance (Benzing 1987). The shift to more complexly structured canopies in angiosperm-dominated forests might have additionally triggered epiphytic pteridophyte diversification by niche differentiation (Schneider et al. 2004b). Additionally, many terrestrial pteridophyte species are low-light specialists (Page 2002). In combination with their tendency towards pronounced edaphic specialization (Tuomisto and Poulsen 1996, Tuomisto 2006), this might have led to extensive diversifications in dark habitats such as the forest floor of tropical montane and lowland forests leading to relatively lower competition for light and space with more lightdemanding seed plant species. Another striking difference between pteridophytes and seed plants emerges from the comparison of pteridophyte proportions in island and mainland floras. Island floras are generally characterized by lower species numbers than comparable mainland areas (Kreft et al. 2008) and this trend is also apparent for pteridophytes (Fig. 1a). The proportion of pteridophytes, however, is significantly higher on islands (Fig. 1b) and there is a globally positive relationship between the geographic isolation and pteridophyte proportions (Fig. 2d). This points to a greater colonization success in pteridophytes; and at least two life-history traits might explain this relationship: first, pteridophytes produce large numbers of small, airborne diaspores (20 60 mm; Tryon 1970) and are thus likely to have greater colonization rates than seed plants. It has been demonstrated that recurrent colonization events in pteridophytes are possible even over very long distances (Wagner 1995). Second, their independence of animal vectors for pollination and seed dispersal might provide pteridophytes a greater chance for establishing viable populations on newly colonized islands. Furthermore, it has been shown that speciation rates are generally higher for seed plants than for pteridophytes (Tryon 1970, Smith 1972). The higher diversification rates of seed plants in turn lead to a decrease

ISSUE

IBS SPECIAL

Bivand, R. 2006. Spdep: spatial dependence: weighting schemes, statistics and models. R package ver. 0.3 17. Cowling, R. M. et al. 1996. Plant diversity in mediterraneanclimate regions. Trends Ecol. Evol. 11: 362 366. Currie, D. J. 1991. Energy and large-scale patterns of animal- and plant-species richness. Am. Nat. 137: 27 49. Currie, D. J. and Paquin, V. 1987. Large-scale biogeographical patterns of species richness of trees. Nature 329: 326 327. Currie, D. J. and Francis, A. P. 2004. Regional versus climatic effect on taxon richness in angiosperms: reply to Qian and Ricklefs. Am. Nat. 163: 780 785. Currie, D. J. et al. 2004. Predictions and tests of climate-based hypotheses of broad-scale variations in taxonomic richness. Ecol. Lett. 7: 1121 1134. Endler, J. A. 1982. Problems in distinguishing historical from ecological factors in biogeography. Am. Zool. 22: 441 452. European Commission Joint Research Centre 2002. GLC 2000: global land cover mapping for the year 2000. <wwwgem.jrc.it/glc2000/>. Field, R. et al. 2008. Spatial species-richness gradients across scales: a meta-analysis. J. Biogeogr. 36: 132 147. Fine, P. V. A. and Ree, R. H. 2006. Evidence for a time-integrated species area effect on the latitudinal gradient in tree diversity. Am. Nat. 168: 796 804. Francis, A. P. and Currie, D. J. 2003. A globally consistent richness climate relationship for angiosperms. Am. Nat. 161: 523 536. Gentry, A. H. and Dodson, C. H. 1987. Diversity and biogeography of Neotropical vascular epiphytes. Ann. Mo. Bot. Gard. 74: 205 233. Given, D. R. 1993. Changing aspects of endemism and endangerment in Pteridophyta. J. Biogeogr. 20: 293 302. Haining, R. 1990. Spatial data analysis in the social and environmental sciences. Cambridge Univ. Press. Hassler, M. and Swale, B. 2001. World fern statistics by country. <http://homepages.caverock.net.nz/ bj/fern/ferndist.htm>. Hassler, M. and Swale, B. 2004. Checklist of world ferns. <http://homepages.caverock.net.nz/ bj/fern/>. Hawkins, B. A. et al. 2003. Energy, water, and broadscale geographic patterns of species richness. Ecology 84: 3105 3117. Hemp, A. 2002. Ecology of the pteridophytes on the southern slopes of Mt. Kilimanjaro. Plant Ecol. 159: 211 239. Hillebrand, H. 2004. On the generality of the latitudinal diversity gradient. Am. Nat. 163: 192 211. Hortal, J. et al. 2008. Regional and environmental effects on the species richness of mammal assemblages. J. Biogeogr. 35: 1202 1214. Hughes, C. and Eastwood, R. 2006. Island radiation on a continental scale: exceptional rates of plant diversification after uplift of the Andes. Proc. Nat. Acad. Sci. USA 103: 10334 10339. Humboldt, A. v. 1808. Ansichten der Natur mit wissenschaftlichen Erla¨uterungen. Cotta. Jansson, R. 2003. Global patterns in endemism explained by past climatic change. Proc. R. Soc. B 270: 583 590. Jansson, R. and Davies, T. J. 2008. Global variation in diversification rates of flowering plants: energy vs. climate change. Ecol. Lett. 11: 173 183. Jetz, W. and Rahbek, C. 2002. Geographic range size and determinants of avian species richness. Science 297: 1548 1551. Jetz, W. et al. 2009. Global associations between terrestrial producer and vertebrate consumer diversity. Proc. R. Soc. B 276: 269 278. Johnson, J. B. and Omland, K. S. 2004. Model selection in ecology and evolution. Trends Ecol. Evol. 19: 101 108.

418

Kerr, J. T. and Packer, L. 1997. Habitat heterogeneity as a determinant of mammal species richness in high-energy regions. Nature 385: 252 254. Kessler, M. 2000. Elevational gradients in species richness and endemism of selected plant groups in the central Bolivian Andes. Plant Ecol. 149: 181 193. Kessler, M. and Siorak, Y. 2007. Desiccation and rehydration experiments on leaves of 43 Pteridophyte species. Am. Fern J. 97: 175 185. Kessler, M. et al. 2001. A comparison of the tropical montane pteridophyte floras of Mount Kinabalu, Borneo, and Parque Nacional Carrasco, Bolivia. J. Biogeogr. 28: 611 622. Kier, G. et al. 2005. Global patterns of plant diversity and floristic knowledge. J. Biogeogr. 32: 1107 1116. Kissling, D. and Carl, G. 2008. Spatial autocorrelation and the selection of simultaneous autoregressive models. Global Ecol. Biogeogr. 17: 59 71. Kissling, W. D. et al. 2009. The global distribution of frugivory in birds. Global Ecol. Biogeogr. 18: 150 162. Kluge, J. et al. 2006. What drives elevational patterns of diversity? A test of geometric constraints, climate and species pool effects for pteridophytes on an elevational gradient in Costa Rica. Global Ecol. Biogeogr. 15: 358 371. Kornas´, J. 1977. Life-forms and seasonal patterns in the pteridophytes in Zambia. Acta Soc. Bot. Poloniae 47: 669 690. Kreft, H. and Jetz, W. 2007. Global patterns and determinants of vascular plant diversity. Proc. Nat. Acad. Sci. USA 104: 5925 5930. Kreft, H. et al. 2004. Diversity and biogeography of vascular epiphytes in western Amazonia, Yasunı´, Ecuador. J. Biogeogr. 31: 1463 1476. Kreft, H. et al. 2008. Global diversity of island floras from a macroecological perspective. Ecol. Lett. 11: 116 127. Kress, W. J. 1986. The systematic distribution of vascular epiphytes: an update. Selbyana 9: 2 22. Ku¨ per, W. et al. 2004. Large-scale diversity patterns of vascular epiphytes in Neotropical montane rain forests. J. Biogeogr. 31: 1477 1487. Legendre, P. 1993. Spatial autocorrelation: trouble or new paradigm? Ecology 74: 1659 1673. Lennon, J. J. 2000. Red-shifts and red herrings in geographical ecology. Ecography 23: 101 113. Lichstein, J. W. et al. 2002. Spatial autocorrelation and autoregressive models in ecology. Ecol. Monogr. 72: 445 463. Linder, H. P. 2003. The radiation of the Cape flora, southern Africa. Biol. Rev. 78: 597 638. Lwanga, J. S. et al. 1998. Assessing fern diversity: relative species richness and its environmental correlates in Uganda. Biodivers. Conserv. 7: 1378 1398. Madison, M. 1977. Vascular epiphytes: their systematic occurrence and salient features. Selbyana 2: 1 13. McGlone, M. S. et al. 2001. Endemism, species selection and the origin and distribution of the vascular plant flora of New Zealand. J. Biogeogr. 28: 199 216. Mittelbach, G. G. et al. 2001. What is the observed relationship between species richness and productivity? Ecology 82: 2381 2396. Mittelbach, G. G. et al. 2007. Evolution and the latitudinal diversity gradient: speciation, extinction and biogeography. Ecol. Lett. 10: 315 331. Mutke, J. and Barthlott, W. 2005. Patterns of vascular plant diversity at continental to global scales. Biol. Skrifter 55: 521 538. New, M. et al. 2002. A high-resolution data set of surface climate over global land areas. Clim. Res. 21: 1 25. Niklas, K. J. et al. 1983. Patterns in vascular land plant diversification. Nature 303: 614 616.

Schneider, H. et al. 2004a. Ferns diversified in the shadow of angiosperms. Nature 428: 553 557. Schneider, H. et al. 2004b. Unraveling the phylogeny of polygrammoid ferns (Polypodiaceae and Grammitidaceae): exploring aspects of the diversification of epiphytic plants. Mol. Phylogenet. Evol. 31: 1041 1063. Schuettpelz, E. and Pryer, K. M. 2009. Evidence for a Cenozoic radiation of ferns in an angiosperm-dominated canopy. Proc. Nat. Acad. Sci. USA 106: 11200 11205. Smith, A. R. 1972. Comparison of fern and flowering plant distributions with some evolutionary interpretations for ferns. Biotropica 4: 4 9. Smith, A. R. et al. 2006. A classification for extant ferns. Taxon 55: 705 731. Sokal, R. R. and Rohlf, F. J. 1981. Biometry: the principles and practice of statistics in biological research. W. H. Freeman. Tateishi, R. and Ahn, C. H. 1996. Mapping evapotranspiration and water balance for global land surfaces. ISPRS J. Photogr. Remote Sens. 51: 209 215. Tognelli, M. F. and Kelt, D. A. 2004. Analysis of determinants of mammalian species richness in South America using spatial autoregressive models. Ecography 27: 427 436. Tryon, R. M. 1970. Development and evolution of fern floras of oceanic islands. Biotropica 2: 76 84. Tryon, R. M. 1986. The biogeography of species, with special reference to ferns. Bot. Rev. 52: 117 156. Tuomisto, H. 2006. Edaphic niche differentiation among Polybotrya ferns in western Amazonia: implications for coexistence and speciation. Ecography 29: 273 284. Tuomisto, H. and Poulsen, A. D. 1996. Influence of edaphic specialization on pteridophyte distribution in neotropical rain forests. J. Biogeogr. 23: 283 293. Tuomisto, H. and Poulsen, A. D. 2000. Pteridophyte diversity and species composition in four Amazonian rain forests. J. Veg. Sci. 11: 383 396. Tuomisto, H. et al. 2003. Dispersal, environment, and floristic variation of western Amazonian forests. Science 299: 241 244. USGS 1996. GTOPO30 Digital Elevation Model. <http:// edcdaac.usgs.gov/gtopo30/gtopo30.asp>. Wagner, W. H. 1995. Evolution of Hawaiian ferns and fern allies in relation to their conservation status. Pac. Sci. 49: 31 41. Watkins, J. E. et al. 2007. Ecological and evolutionary consequences of desiccation tolerance in tropical fern gametophytes. New Phytol. 176: 708 717. Watkins, J. E. J. et al. 2006. Species richness and distribution of ferns along an elevational gradient in Costa Rica. Am. J. Bot. 93: 73 83. Whittaker, R. J. and Ferna´ndez-Palacios, J. M. 2007. Island biogeography. Oxford Univ. Press. Wiens, J. J. and Donoghue, M. J. 2004. Historical biogeography, ecology and species richness. Trends Ecol. Evol. 19: 639 644. Willis, K. J. and McElwain, J. C. 2002. The evolution of plants. Oxford Univ. Press. Wright, D. H. 1983. Species energy theory: an extension of species area-theory. Oikos 41: 496 506.

IBS SPECIAL

Nobel, P. S. 1978. Microhabitat, water relations, and photosynthesis of a desert fern, Notholaena parryi. Oecologia 31: 293 309. O’Brien, E. M. 1993. Climatic gradients in woody plant species richness: towards an explanation based on an analysis of southern Africa’s woody flora. J. Biogeogr. 20: 181 198. O’Brien, E. M. 1998. Water-energy dynamics, climate, and prediction of woody plant species richness: an interim general model. J. Biogeogr. 25: 379 398. O’Brien, E. M. 2006. Biological relativity to water energy dynamics. J. Biogeogr. 33: 1868 1888. Olson, D. M. et al. 2001. Terrestrial ecoregions of the world: a new map of life on earth. BioScience 51: 933 938. Page, C. N. 1985. Pteridophyte biology, the biology of the amphibians of the plant world. Proc. R. Soc. Edinburgh B 86: 439 442. Page, C. N. 2002. Ecological strategies in fern evolution: a neopteridological overview. Rev. Palaeobot. Palynol. 119: 1 33. Pianka, E. R. 1966. Latitudinal gradients in species diversity: a review of concepts. Am. Nat. 100: 33 46. Prance, G. T. 2001. Discovering the plant world. Taxon 50: 345 359. Pryer, K. M. et al. 2001. Horsetails and ferns are a monophyletic group and the closest living relatives to seed plants. Nature 409: 618 622. Pryer, K. M. et al. 2004. Phylogeny and evolution of ferns (monilophytes) with a focus on the early leptosporangiate divergences. Am. J. Bot. 91: 1582 1598. Qian, H. 2008. Effects of historical and contemporary factors on global patterns in avian species richness. J. Biogeogr. 35: 1362 1373. Qian, H. and Ricklefs, R. E. 2000. Large-scale processes and the Asian bias in species diversity of temperate plants. Nature 407: 180 182. R Development Core Team 2005. R: a language and environment for statistical computing. R foundation for Statistical Computing. Raghavan, V. 1989. Developmental biology of fern gametophytes. Cambridge Univ. Press. Rahbek, C. and Graves, G. R. 2001. Multiscale assessment of patterns of avian species richness. Proc. Nat. Acad. Sci. USA 98: 4534 4539. Richards, P. W. 1973. Africa, the ‘Odd Man Out’. In: Meggers, B. J. et al. (eds), Tropical forest ecosystems in Africa and South America: a comparative review. Smithsonian Inst. Press, pp. 21 26. Ricklefs, R. E. 1987. Community diversity: relative roles of local and regional processes. Science 235: 167 171. Ricklefs, R. E. 2004. A comprehensive framework for global patterns in biodiversity. Ecol. Lett. 7: 1 15. Ricklefs, R. E. et al. 2004. The region effect on mesoscale plant species richness between eastern Asia and eastern North America. Ecography 27: 129 136. Rohde, K. 1992. Latitudinal gradients in species diversity: the search for the primary cause. Oikos 65: 514 527. Rosenzweig, M. L. 1995. Species diversity in space and time. Cambridge Univ. Press.

ISSUE

Download the Supplementary material as file E6434 from <www.oikos.ekol.lu.se/appendix>.

419

Ecography 33: 420 424, 2010 doi: 10.1111/j.1600-0587.2010.06304.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: Thiago Rangel. Accepted 29 April 2010

mmSAR: an R-package for multimodel species area relationship inference Franc¸ois Guilhaumon, David Mouillot and Olivier Gimenez F. Guilhaumon (Francois.Guilhaumon@univ-montp2.fr) and D. Mouillot, Laboratoire Ecosyste`mes Lagunaires, Unite´ Mixte de Recherche 5119, Centre National de la Recherche Scientifique-IFREMER-UM2, Univ. Montpellier 2, cc 093, Place Euge`ne Bataillon, FR-34095 Montpellier Cedex 5, France. O. Gimenez, Centre d’Ecologie Fonctionnelle et Evolutive, Unite´ Mixte de Recherche 5175, Campus Centre National de la Recherche Scientifique, 1919 Route de Mende, FR-34293 Montpellier Cedex 5, France.

IBS SPECIAL

ISSUE

The species area relationship (SAR) is one of the most fundamental tools in ecology. After almost a century of quantitative ecology, however, the quest for a ‘‘best SAR model’’ still remains elusive, with a substantial uncertainty about the best fitting SAR model frequently being observed. Recent research has required that this uncertainty be addressed, and a multimodel SAR framework has been devised. Here we introduce the mmSAR R-package, which is a flexible and scalable implementation of the multimodel SAR framework for species-area datasets, and provide some examples of its use. This R-package provides functions for fitting SAR models, performing model selection, and the build up of multimodel SARs.

One of the most ancient and ubiquitous patterns that has been recognized in ecology is the increase in species richness (S) with increasing sampling area (A): the species area relationship (SAR). The SAR has been mystifying ecologists for more than 150 years (De Candolle 1855, MacArthur and Wilson 1967, Connor and McCoy 1979, Drakare et al. 2006, Southwood et al. 2006) and its modelling remains a central issue for theoretical ecologists and conservationists (Rosenzweig 1995, Smith 2010). Inference about the SAR is mandatory in the wide range of conservation applications that require the comparison of diversity patterns when regions differ in area, such as global scale conservation priority-setting schemes (Brooks et al. 2006, Lamoreux et al. 2006, Wilson et al. 2007). In theoretical studies, SARs are considered to be fundamental properties of biological systems and are, for example, explained in terms of species abundances and spatial distribution of individuals (He and Legendre 2002, Martin and Goldenfeld 2006) and constitute a cornerstone for macroecological investigations (Sˇizling and Storch 2004, Drakare et al. 2006). After Arrhenius (1921), the SAR has mainly been modelled using a power law (S cAz, where c and z are constants to be estimated). Despite this historical hegemony, however, several studies have highlighted other functional forms for SARs (Gleason 1922, Coleman et al. 1982, Lomolino 2000, Tjørve 2003, 2009). Moreover, quantitative studies focusing on comparisons among models have indicated that the power law SAR is not ubiquitous (Connor and McCoy 1979, Flather

420

1996, Stiles and Scheiner 2007), stressing the importance of testing the relative fit of various different models in SAR analyses (Smith 2010). Furthermore, recent analyses have often demonstrated substantial uncertainty in selecting the best SAR model for a given dataset (Stiles and Scheiner 2007, Guilhaumon et al. 2008). The multimodel selection framework (Burnham and Anderson 2002) is an approach that can account for such uncertainties in inferring the SAR, allowing the investigator to perform inferences while incorporating variability in both model selection and parameter estimation (multimodel SARs; Guilhaumon et al. 2008). Here, we introduce the mmSAR R-package for the freeware and open-source R software (R Development Core Team 2009). mmSAR is a flexible and scalable implementation of the multimodel SAR framework for species area datasets and provides several functionalities: fitting several relevant SAR models, performing a selection among this set of models, averaging the prediction of the SAR obtained from different models to establish a consensual inference and to provide robust confidence intervals. The present software note describes the different components of the multimodel SAR framework, as well as their implementation in the mmSAR R-package (Fig. 1). We illustrate the framework with the results of an analysis of a species area dataset for the plants of the Galapagos Islands (Preston 1962). The users interested in the methodological details of the multimodel SAR framework are referred to Guilhaumon et al. (2008).

Figure 1. Main components of mmSAR (A) and sample ‘‘use cases’’ (B). B1 simple nonlinear SAR model fitting. B2 multimodel SAR calculation.

Model fitting

The components of the mmSAR implementation of the multimodel SAR framework are presented in Fig. 1. Apart from the species area dataset itself, mmSAR provides R objects to handle SAR models, facilitating the fit of SAR models through non-linear regression and the construction of consensual prediction for the SAR with associated confidence intervals (Fig. 1A). Different applications can be envisaged with the mmSAR components, from simple model fitting to selection and average across sets of models (Fig. 1B).

mmSAR performs nonlinear regressions to obtain model parameter estimates by minimizing the residual sum of squares with an unconstrained Nelder Mead optimization algorithm. Assuming normality of the observations, this approach produces optimal maximum likelihood estimates of model parameters (Burnham and Anderson 2002). To avoid numerical problems, such as local minima, and speed up the convergence process, starting values used to run the optimization algorithm are carefully chosen. For directly interpretable parameters (e.g. an asymptote), corresponding values in the datasets are used (e.g. the observed maximum of species richness in the case of an asymptote), otherwise the standard procedures described by Ratkowsky (1983, 1990) are implemented. Finally, mmSAR gives the option to provide custom starting values, allowing users to implement exhaustive searches for best fits. We provide example fits of the eight SAR models implemented in mmSAR to the Galapagos Islands dataset in Fig. 2A1 A8.

Models

Regression validation Regressions are usually evaluated by statistical examination of normality and homoscedasticity of residuals. In mmSAR, two tests for the normality of the residuals are available: the Lilliefors extension of the Kolmogorov normality test, which is advocated when sample size is large or when the data show a substantial variability (e.g. continental scale studies) and the Shapiro Wilk test for normality, which focuses on skewness and kurtosis of the empirical distribution of the residuals and is useful for small sample size or when data results from small scale sampling. mmSAR tests

421

ISSUE

For a given dataset, a multimodel SAR inference is made simultaneously using the predictions of several non-linear regression models. Obtaining a consistent set of models is one of the most important challenges in informationtheoretic analyses (Burnham and Anderson 2002). mmSAR proposes a comprehensive set of SAR models (Table 1), including five convex models (power, exponential, negative exponential, Monod and rational function) and three sigmoid models (logistic, Lomolino, and cumulative Weibull). This includes convex, sigmoid, asymptotic, and non asymptotic functions, thus encompassing the various shapes attributed to SARs in the literature. Note that the linearized forms (via logarithmic transformations) of the power and exponential models, which require using log(S ) in place of S, were not implemented in mmSAR, otherwise precluding comparisons across the entire set of models. In mmSAR, models are implemented as R objects and new non linear SAR models should easily be specified by the user and added to the available collection.

IBS SPECIAL

The multimodel SAR framework

Table 1. Functional forms for the SAR implemented in mmSAR. In these equations, S and A represent, respectively, species richness and area, while c, z, f and d are fitted parameters. The parameter d is an upper asymptote, except for the rational function for which the upper asymptote is z/d. Name Power Exponential Negative exponential Monod Rational function Logistic Lomolino Cumulative Weibull

Code Power Expo Negexpo Monod Ratio Logist Lomolino Weibull

Formula z

S cA S c zlog(A) S d(1 exp( zA)) S d/(1 cA 1) S (c zA)/(1 dA) S d/(1 exp( zA f )) S d/1 (zlog(f/A)) S d(1 exp( zAf ))

for homoscedasticity by evaluating the correlation between residual magnitude and areas or fitted values (Pearsonâ&#x20AC;&#x2122;s product moment correlation coefficient). Generally, a model is considered not to be valid for a given dataset if one of the tests of normality or homoscedasticity is significant at the 5% level.

Model selection

2 2 2 2 3 3 3 3

Shape Convex Convex Convex Convex Convex Sigmoid Sigmoid Sigmoid

Asymptotic nature No No Yes Yes Yes Yes Yes Yes

mmSAR, these probabilities are materialized by Akaike weights (Burnham and Anderson 2002) derived from information criteria (IC) such as the Akaike information criterion (AIC) or its correction for small sample bias (AICc) and the Bayesian information criterion (BIC). AIC and other model selection criteria that estimate Kullback Leibler information are used widely in the ecological literature, but other criteria such as the BIC are also commonly used to carry out model selection (see Burnham and Anderson 2002 for a review of model selection and multimodel inference). AIC and BIC do not share the same conceptual bases and penalize differently for the dimension of the models (BIC tends to select models with fewer parameters than AIC), and although the results of (mm)SAR analyses are generally robust as regards the criterion used for model selection (Guilhaumon et al.

IBS SPECIAL

ISSUE

The information-theoretic framework for model-selection is based on the evaluation of multiple working hypotheses (Burnham and Anderson 2002). This evaluation of competing hypotheses, which are each represented by a different model, is achieved through the estimation, for each, of the probability to be the best in explaining the data. In

Number of parameters

Figure 2. mmSAR results for a species area dataset describing the plants of the Galapagos Islands (Preston 1962). (A1 A8) Fit of the eight SAR functional forms implemented in mmSAR (see Table 1 for equation descriptions). (B) Results of a model selection procedure (the eight of a bar indicates the probability (i.e. Akaike weights derived from the AICc criterion in this example) of the model being the best in fitting the data). (C) Model fits (dashed lines), multimodel SAR (black line) and associated non parametric confidence interval (grey shading).

422

2008), mmSAR implements both Kullback Leibler and Bayesian strategies for model selection. For a fitted model i, its weight wi is given by: wi

e 1=2Di M X e 1=2Dr

(1)

r 1

where M is the number of models in the set and Di is defined as Di ICi ICmin with ICmin the IC value for the best model. Akaike weights are a straightforward means of interpreting the IC values of each model, as model likelihood, and provide the basis of multimodel inference. For the Galapagos Islands data set, the best fitting model was exponential but three others models (power, negative exponential, and Monod) had almost equivalent probabilities in explaining the data (AICc Akaike weights in Fig. 2B). The four remaining models (rational function, logistic, Lomolino, and cumulative Weibull) have negligible likelihood and should contribute only marginally to the multimodel SAR (AICc Akaike weights in Fig. 2B).

Model averaging and confidence interval building In the model selection framework, model selection uncertainty arises when the dataset support several models with a similar strength (i.e. for a given dataset, no wi is higher than 0.9; Burnham and Anderson 2002), as this is the case with the data from the Galapagos Islands (Fig. 2B). In such cases, it is not adequate to rely exclusively on the best model only; multimodel inference can construct a more robust final inference (Burnham and Anderson 2002). As advocated for differently parameterized models, mmSAR implements model averaging and considers the weighted average of all valid model predictions (see Regression validation), with respect to model weights, to construct multimodel SARs: SÂŻË&#x2020;

M X

SË&#x2020;i wi

(2)

i 1

Guilhaumon, F., Mouillot, D. and Gimenez, O. 2010. mmSAR: an R-package for multimodel species area relationship inference. Ecography 33: 420 424 (Version 0). Acknowledgements We thank two anonymous reviewers for comments or helpful discussions and are grateful to David Mckenzie for editing and correcting the language.

IBS SPECIAL ISSUE

where SÂŻË&#x2020; is the multi-model averaged species richness and SË&#x2020;i is the species richness inferred from model i, M is the number of valid models. The multimodel SAR for the Galapagos Islands data set is presented in Fig. 2C. Finally, in mmSAR, confidence intervals incorporating uncertainty regarding both model selection and parameter estimation can be constructed using the percentile method and a non-parametric bootstrap scheme (Efron 1979, Buckland et al. 1997). For a given species area dataset, a large number of bootstrap samples are obtained in the following manner: 1) one of the SAR models included in the analysis is selected with a probability equal to its weight as calculated from eq. 1. 2) The selected model is fitted to the observed dataset under study. 3) The vectors of inferred species richness (regression line) and residuals are obtained from the regression and the residuals are standardized. 4) The residuals are sampled with replacement until sample size reaches that of the dataset, to form a vector of modified residuals. 5) The vector of modified residuals is added to the

vector of inferred species richness, to form the resample (bootstrap set of pseudo responses). A collection of multi-model SARs inferred from each of the resamples is gathered by applying the whole procedure of model selection and averaging, while the bootstrap estimates of species richness are sorted in ascending order to provide the percentile confidence intervals (Buckland et al. 1997): the limits of an approximate (1 a)100% confidence interval are given by picking the rth and sth values in the ordered vector of bootstrap estimates, such that r (b 1)a and s (b 1)(1 a). For the Galapagos Islands dataset, the number of resamples was fixed to 9999, thus the limits of the 95% confidence interval for a point estimate of species richness (Fig. 2C) are given by the 250th and the 9750th values. The mmSAR R-package may have potential uses in both theoretical and conservation analyses. For example, in theoretical applications such as investigations about how SARs may differ among different systems, model selection patterns (i.e. relative likelihoods of different SAR shapes) can be compared for the different systems. Allowing one, for example, to state about the saturation or non saturation of species richness with increasing area. These kind of analyses may help to extend discussions beyond the comparison of slopes of log-linear power SARs (Guilhaumon et al. 2008). In conservation applications, multimodel non-parametric confidence intervals can inform about the reliability of the multimodel SAR for a given dataset but also have more practical applications. For example, these confidence intervals were used by Guilhaumon et al. 2008 to rank regions of a dataset with respect to their biological richness. By positioning the observed richness of each region in the associated vectors of ordered bootstrap species richness estimates (the higher the position of the observed species richness in the vector of bootstrap estimates the higher the ecoregion in the ranking), these authors were able to devise a hotspot ranking methodology that was robust to the underlying form of SARs. The mmSAR R-package and a detailed userâ&#x20AC;&#x2122;s guide is available at the R-Forge website <http://mmsar.r-forge. r-project.org>. To cite mmSAR or acknowledge its use, cite this Software note as follows, substituting the version of the application that you used for â&#x20AC;&#x2DC;â&#x20AC;&#x2DC;Version 0â&#x20AC;&#x2122;â&#x20AC;&#x2122;:

References Arrhennius, O. 1921. Species and area. J. Ecol. 9: 95 99. Brooks, T. M. et al. 2006. Global biodiversity conservation priorities. Science 313: 58 61. Buckland, S. T. et al. 1997. Model selection: an integral part of inference. Biometrics 53: 603 618.

423

IBS SPECIAL

ISSUE

Burnham, K. P. and Anderson, D. R. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Springer. Coleman, B. D. et al. 1982. Randomness, area and species richness. Ecology 63: 1121 1133. Connor, E. F. and McCoy, E. D. 1979. The statistics and biology of the species area relationship. Am. Nat. 113: 791 833. De Candolle, A. 1855. Geógraphie botanique raisonneé; ou exposition des faits principaux et des lois concernant la distribution geógraphique des plantes de l’e´poque actuelle. Maisson. Drakare, S. et al. 2006. The imprint of the geographical, evolutionary and ecological context on species area relationships. Ecol. Lett. 9: 215 227. Efron, B. 1979. Bootstrap methods: an other look at the jackknife. Ann. Stat. 7: 1 26. Flather, C. H. 1996. Fitting species accumulation functions and assessing regional land use impacts on avian diversity. J. Biogeogr. 23: 155 168. Gleason, H. A. 1922. On the relation between species and area. Ecology 3: 158 162. Guilhaumon, F. et al. 2008. Taxonomic and regional uncertainty in species area relationships and the identification of richness hotspots. Proc. Nat. Acad. Sci. USA 105: 15458 15463. He, F. L. and Legendre, P. 2002. Species diversity patterns derived from species area models. Ecology 83: 1185 1198. Lamoreux, J. F. et al. 2006. Global tests of biodiversity concordance and the importance of endemism. Nature 440: 212 214. Lomolino, M. V. 2000. Ecology’s most general, yet protean pattern: the species area relationship. J. Biogeogr. 27: 17 26. MacArthur, R. H. and Wilson, E. O. 1967. The theory of island biogeography. Princeton Univ. Press.

424

Martin, H. G. and Goldenfeld, N. 2006. On the origin and robustness of power-law species area relationships in ecology. Proc. Nat. Acad. Sci. USA 103: 10310 10315. Preston, F. W. 1962. The canonical distribution of commonness and rarity: part I. Ecology 43: 185 215. R Development Core Team 2009. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, <www.R-project.org>. Ratkowsky, D. A. 1983. Nonlinear regression modelling. A unified practical approach. Dekker. Ratkowsky, D. A. 1990. Handbook of nonlinear regression models. Dekker. Rosenzweig, M. L. 1995. Species diversity in space and time. Cambridge Univ. Press. ˇ .Sizling, A. and Storch, D. 2004. Power-law species area relationships and self-similar species distributions within finite areas. Ecol. Lett. 7: 60 68. Smith, A. B. 2010. Caution with curves: caveats for using the species area relationship in conservation. Biol. Conserv. 143: 555 564. Southwood, T. R. E. et al. 2006. Observations on related ecological exponents. Proc. Nat. Acad. Sci. USA 103: 6931 6933. Stiles, A. and Scheiner, S. M. 2007. Evaluation of species area functions using Sonoran Desert plant data: not all species area curves are power functions. Oikos 116: 1930 1940. Tjørve, E. 2003. Shapes and functions of species area curves: a review of possible models. J. Biogeogr. 30: 827 835. Tjørve, E. 2009. Shapes and functions of species area curves (II): a review of new models and parameterizations. J. Biogeogr. 36: 1435 1445. Wilson, K. A. et al. 2007. Conserving biodiversity efficiently: what to do, where, and when. PLoS Biol. 5: 1850 1861.