Evolution

19th-century idealistic morphologists such natural-theological assumptions about a per- gram as one not dominated by a typological as Carl F. Kielmeyer and3/30/09 J. F. Meckel sonified God who had created a perfectly and linear-recapitulationist mindset but 0403NewsFocus.qxp 5:23 that PM Page 29 retained their teleology, their typological adapted nature. Bronn’s translation, though it rather as continuing to wrestle with the need emphasis on form, and their linear recapitula- altered key ideas to make Darwin comprehen- to account for variability and unpredictable tionism. This story, emphasizing the long per- sible to a German academic audience, was not change in terms of mechanistic laws of sistence of a German transcendental approach a conservative throwback. It represented the nature—among which Haeckel included, at to nature, has been deeply entrenched in the dynamic engagement of a leading paleontolo- the top of his list, natural selection. Haeckel’s Introduction history of biology. gist who had also long been working on many Darwinism thus shows continuity with early Gliboff challenges this history right from of the questions Darwin claimed as his own— 19th-century concerns, mediated through Stockholm. “When we started, 2 Darwin’s Inspiration, Darwin’s Legacy the beginning. ascription of simple theThe search profile was bigger, linear a critical yet generous equal, who saw himself Bronn. But those concerns were always more Andrew Sugden a magnolia she recapitulationism to the[flower],” views of Romantic as moving science forward through the modi- flexible than has been acknowledged, and recalls.

But 30 years ago,toshe embryologists, he notes, owes much a carica- fications he made to Darwin’s flawed theory. their articulation changed over time. Of and by others ture developed Karldiscovered Ernst vontiny Baer in a Bronn’s death in 1862 afforded him little course Haeckel’s Darwinism was not Articles ancient flowers by sieving polemical context, then adopted uncritically by chance to steer the conversation further.

Darwin’s own, but it was not an aberration or through sand and clay sedi3 ofOn influential historians suchthis as E. S. a distortion some true theory, any the Origin of Life onmore Earth ments. With technique, Russell and they Stephen Jay collected Gould. hunthan any otherCarl post-Darwinian additions or Zimmer have now

Science 9 January 2009 323: 198-199 Gliboff’s fresh reading the origadjustments were.

It was science moving on. dreds of of millimeter-size preserved in inal sources flowers, interpretssome Kielmeyer Gliboff ’s overall picture of scientific dimensions, from PorOn the of Art and Symbolism and Meckelthree as far less rigidly advance, in5 contrast to Origin Richards’s emphasis on other locations Balter typological intugal theirand orientation and with charisma andMichael passion, is one of scientists Cretaceous deposits 70 milScience 6 February 2009 323: 709-711 much more lion attentive to nature’s building and innovating incrementally, workto 120 million years old. variability thanThis has fossil been seen ing with what their predecessors have handed diversity 8 On the Origin of Photosynthesis before. Bothshows for these them and sculpting it into something new yet that early-19thangiosperms were

Mitch Leslie century naturalists and for their understandable to those around them. His senthriving, with several groups

Science 6 March 2009 323: 1286-1287 well-established, by 100 milintellectual heirs, Gliboff argues, sitive reading allows Out us of tothe see post-1859 past. page 10 lion years ago.

In some, the Tinyrational Amborellaactors sits the critical issue was to understand German evolutionists as rather 10 On theatOrigin of Flowering Plants flower parts are whorled the the nature’s manifold variety while like than irrationally stuck inbottom someof early-19th Elizabeth Pennisi angiosperm family tree. those of modern flowers; in seeking out underlying strict natucentury moment with unmodern commitothers they are spiraled, con Science 3 April 2009 324: 28-31 ral laws to account for it. ments. By challenging the very foundations of sidered by some researchers This provides newprimitive starting arrangement. Some the standard narrative of von German morpholas the amore from one 14 of the nonflowering seed plants Alexander Humboldt and the General Physics point for analyzing Darwin’s ogy, this careful, compelling account does at flower fossils havefirst prescribed numbers of “We are realizing that this or gymnosperms, whose heyday was 200 of the Earth petals, anotherpaleonmodern feature, whereas in ago.

Modern gymnosperms translator, the prominent least asyears much as Richards’s to undermine the

Stephen T. Jackson huge diversity is probably million the petal count include conifers, ginkgoes, andGerman the cycads, tologist H. G.others Bronn—a figure lit-varies. association of 19th-century

Science 1 May 2009 324:Darwin596-597 with their astout trunks and large fronds. tle attended toInin1998, the Chinese standardgeologist Ge Sun of the result of one innovaism with dangerously exceptional view of Jilin University in Changchun, China, came Before angiosperms came along, these story but the lynchpin of Glinature. But two books offer very different 16 themuch Making German Evolution: Translation and Tragedy across what seemed to be a much older tion piled on top of plants were more diverse and boff’s. Intriguingly andfossil, plausibly, reads. Is cycadlike scientific progress a matter Lynn K. Nyhartsuch flower. The called Archaefructus, was included species, as theof perGliboff argues that Bronn’s uselooked to be 144 mil- another innovation.” sonal anguish and triumph, or ofwoody intellectual

Science 27 February 2009 323: 1170-1171 an aquatic plant that extinct Bennettitales, and many REPORTS of terms likelion “vervollkommnet” chuggingcalled Our conceptofof which it should be years old. By 2002, Sun and David plants along?Gnetales, population expansion (23)Originality areboth. found the in the located acapacious presence lineages that co-occur in the ad- 6.2% divergence). In contrast, —Petersites Crane, of of thetwo Florida of Natural few representatives, including 18 Darwin’s (perfect) as Dilcher translations for Dar- Museum enough to include unstable for H.today albomarginatus and H. faber, the refugiaofare genetically joint jacent refugia.

In all species, average nethad nucle- outside (south of) History (FLMNH) in Gainesville University Chicago firs, area survive (see family tree, Peter J. Bowler win’s “improved” or “favored” well as incommon the Bahia in refugium area for were H. faber otide differences across from localities (22) reflects more similar to each other, although to a lesser p. as described an entire plant, roots to flow31).

Also the Jurassic and Notes Science 9 January 323: 223-226 were not abouthigh dragging Darwin within refugia (2.6 to extent in H. faber (0.1 to 1.6%). Signatures of andReferences H. semilineatus.

The lack gone; of 2009 signature geographic ers, entombed on astructure slab of rock unearthed in These fossils often spark debate because seed ferns, a group long their of 1. E. Haeckel, Generellenow Morphologie der Organismen backward into a German teleopage 16 Liaoning in northeastern China. specimens tend to be imperfectly preserved most(Georg famous member is Caytonia, which Reimer, Berlin, 1866).

22 The Red Queenpress and the Court Jester: Species Diversity logical view of nature (asArchaefructus has in putative 2. Theto reviewer previously served as astructures. reader for both In one wasn’t much and leave room for interpretation. To help seems have precarpel-like Fig. 2. sense, Genetic diversity C A B

and the Role of Biotic and Abiotic Factors Through Time books at the manuscript stage. been claimedtoby those who have refugial unstableplant areas before remedy that, Friis and her colleagues have These g roups’ perceived relevance to look at.(stable) “It’s versus a flowering

Michael J. Benton intowere the Brazilian Atlantic rainforest. there flowers,” Dilcher notes.

It lacked paid attention Bronn at all). begun to examine flowers using synchroflower evolution and their relationships to A painter, too. Haeckel’s oil landscape of highlands in Java, from (Top) Species-specific stability maps;have an tron radiation to generate a 3D image of angiosperms

Science 6 February 2009 323: 728-732 petals and sepals, it did have ping-ponged between 10.1126/science.1169621 Instead, Gliboff asserts, Bronn’sbut Wanderbilder (1905).

Science funding Climate regulation Human rights

CREDITS (TOP TO BOTTOM): COURTESY OF STEPHEN MCCABE, UC SANTA CRUZ; PHOTO BY JENNIFER SVITKO, COURTESY OF WILLIAM L. CREPET, CORNELL UNIVERSITY

CREDIT: ERNST HAECKEL/FROM WANDERBILDER (W. KOEHLER, GERA-UNTERMHAUS, 1905)

Contents

ORIGINS

Pernambuco modeledcarpel. refugiaWhen in black.

(A)Nixon H. enclosed Kevin and their inner structures, allowing the fossil to camps, depending on how the evolutionary refugium albomarginatus, (B) H. semilineatus, colleagues at Cornell University compared remain intact while Friis peers inside it trees were26 constructed.

Evidence for Ecological Speciation and Its Alternative (C) H. faber. Note the absence of large 1171 www.sciencemag.org SCIENCE VOL 323 27 FEBRUARY 2009 its stable traits regions with those same traits in 173 living from many angles (Science, 7 December In the mid-1980s, Peter Crane, now at

Dolph Schluter Bahia refugium in the southern portion plants, Archaefructus came out as a sister to 2007, p.

1546).

“We can get fantastic resothe University of Chicago in Illinois, pro* * Science 6 February 2009 323: 737-741 of the forest (south of the Bahia and living closerto tothe the com- lution,” says Friis. “It’s really exciting.” But posed a solution, the anthophyte hypothesis.

São angiosperms Paulo refugia)and relative mon ancestor than even Amborella. so far, the flowers Friis finds are Using several evidenceSpecies and noting central and northern areas.

Asterisks 30 lines The of Bacterial Sã o Paulo refugium Challenge: Making Sense of Genetic Archaefructus’s distinction that both Bennettitales and denote refugia inferred beyond was the short- too diverse to trace back to a Gnetales and Ecological Diversity current ranges Within of the target species. lived, however. months, better dat- particular ancestor. “From organize their male 5.4% and Christophe Fraser, together Eric J. Alm, Martin F. Polz, et al. localities sampled for found these fossils, we cannot ingSymbols of the indicate sediments in which it was female organs

Science 6 February 2009 323: 741-746 molecular analysis.

Scale bar, 400 km. yielded younger dates, putting this f irst say what is the basic in what could be con(Bottom) The 50%with majority-rule con- fossil form,” she says. flower squarely other early strued as a preflower, sensus Bayesian phylogenetic trees, 7% 36 Stability Predicts Genetic flower parts, about 125 million years old. he considered them, Diversity in the Brazilian Atlantic 7.8% rooted with sequences from the othAlso, a 2009 phylogenetic analysis of Before flowers along with angiosperms,

Forest Hotspot er two congeneric species studied 67 (root taxanot by shown).

Doyle Thick and Peter Endress they have yet as comprising a single

Ana Carolina Carnaval, Michael J. Hickerson, Célio F. B. Haddad, et al. page Although 36 internodes de- of the 5.3 – University Zurich, Switzerland, angiosperm entity called note cladesofwith posterior probability placed to find the oldest fossil

Science 6 February 323: 785-789 4% 2009 Larger than 5.8% life. Although merely thegreater fossilthan in with liliesindicate rather than at flowers, researchers anthophytes. For the next 90%.water Percentages 2.2 millimeters in diameter, this 3D corrected distances between theTamura-Nei base of the angiosperms, although this assume that the ancesdecade, most family trees fossil flower shows that grasses date Around the world, governments turn to AAAS as an objective, multidisciplinary scientific authority to clades (20).is contested. conclusion tral angiosperm evolved based image: on morphology supback to 94 million years ago.

Cover ©Tui De Roy/Minden Pictures/FLPA 5.6% educate public officials and judicial figures on today’s most pressing issues. Our goal is to promote Inset: George Richmond/Bridgeman Art Library, London (Superstock) informed policy decisions that benefit society. And this is just one of the ways that AAAS is committed to www.sciencemag.org SCIENCE VOL 324 3 APRIL 2009 29 advancing science to support a healthy and prosperous world. Join us. Together we can make a difference. aaas.org/plusyou/policy

© 2009 by The American Association for the Advancement of Science. All rights reserved.

On the Origin of of

Introduction

Life on Earth

Darwin’s Inspiration, Darwin’s Legacy

in some warm little pond, with all sorts of ammonia and phosphoric salts, light, heat, electricity, etc., present, that a protein compound was chemically formed ready to undergo still more complex changes, at the present day such matter would be instantly devoured or absorbed, which would not have been the case before living creatures were formed.”

Scientists today who study the origin of life do not share Darwin’s pessimism about our ability to reconstruct those early moments.

“Now is a good time to be doing this research, because the prospects for success are greater than they have ever been,” says John Sutherland, a chemist at the University of Manchester in the United Kingdom. He and others are addressing each of the steps involved in the transition to life: where the raw materials came from, how complex organic molecules such as RNA formed, and how the first cells arose. In doing so, they are inching their way toward making life from scratch. “When I was in graduate school, people thought investigating the origin of life was something old scientists did at the end of their career, when they could sit in an armchair and speculate,” says Henderson James Cleaves of the Carnegie Institution for Science in Washington, D.C. “Now making an artificial cell doesn’t sound like science fiction any more. It’s a reasonable pursuit.”

The day has passed delightfully. Delight itself, however, is a weak term to express the feeling of a naturalist who, for the first time, has wandered by himself in a Brazilian forest. The elegance of the grasses, the novelty of the parasitical plants, the beauty of the flowers, the glossy green of the foliage, but above all the general luxuriance of the vegetation, filled me with admiration. A most paradoxical mixture of sound and silence pervades the shady parts of the wood. The noise from the insects is so loud, that it may be heard even in a vessel anchored several hundred yards from the shore; yet within the recesses of the forest a universal silence appears to reign. To a person fond of natural history, such a day as this brings with it a deeper pleasure than he can ever hope to experience again.

—Charles Darwin, The Voyage of the Beagle, Feb 29th [1832]

The collection reprinted here is a sample of the articles published in 2009 by Science magazine in celebration of the Darwin bicentenary. We start with four of the essays from our “On the Origin of” series, prepared by Science’s news writers; further essays in this series are appearing monthly in Science throughout the year. A Perspective by Stephen Jackson then considers the legacy of Alexander von Humboldt, for whom, like Darwin, the South American tropics were a critical inspiration, and who died 150 years ago in the year of the publication of Darwin’s Origin. (Humboldt’s Personal Narrative of his tropical explorations was acknowledged by Darwin as ‘far exceed[ing] in merit anything I have read’ on the subject.) A book review by Lynn Nyhart explores two recent volumes on Ernst Haeckel’s work, his interpretations of Darwin and his contributions to evolutionary thought.

In the first of four Review articles reprinted here, Peter Bowler analyzes the originality of Darwin’s contribution to the understanding of the diversity and diversification of the living world. Michael Benton

2

examines the extent to which biotic and abiotic factors have shaped species diversity in the fossil record. Dolph Schluter reviews how research on speciation has shifted in focus from morphological evolution to reproductive isolation, tracing the links between Darwin’s ideas and current thinking. Christophe Fraser and colleagues discuss the contentious area of microbial species formation, an issue that would surely have vexed Darwin horribly had the bewildering diversity of microbes been known in his day.

Finally, with a focus on conservation, a Report by Ana Carnaval et al., who model evolutionary processes in endemic tree-frog species in the Brazilian Atlantic Forest, the very biodiversity hotspot that so inspired Darwin on his South American landfall, and that is now reduced to a collection of small fragments scattered along the coast. Darwin returned to the Brazilian coast on his final homeward leg, more than four years after his first landfall there. His enthusiasm for the tropical forested landscape was undiminished.

In my last walk I stopped again and again to gaze on these beauties, and endeavoured to fix in my mind for ever, an impression which at the time I knew sooner or later must fail … they will leave, like a tale heard in childhood, a picture full of indistinct, but most beautiful figures.

—Charles Darwin, The Voyage of the Beagle, August 1836 Andrew Sugden, Deputy Editor

AN AMAZON OF WORDS FLOWED FROM Charles Darwin’s pen. His books covered the gamut from barnacles to orchids, from geology to domestication. At the same time, he filled notebooks with his ruminations and scribbled thousands of letters packed with observations and speculations on nature. Yet Darwin dedicated only a few words of his great verbal flood to one of the biggest questions in all of biology: how life began.

The only words he published in a book appeared near the end of On the Origin of Raw ingredients Species: “Probably all the organic beings which Life—or at least life as we know it—appears to have ever lived on this earth have descended have emerged on Earth only once. Just about all from some one primordial form, into which life organisms use double-stranded DNA to encode was first breathed,” Darwin wrote. genetic information, for example. They copy Darwin believed that life likely emerged their genes into RNA and then translate RNA spontaneously from the chemicals into proteins. The genetic code it is made of today, such as carbon, THE YEAR OF they use to translate DNA into pronitrogen, and phosphorus. But he teins is identical, whether they are The English did not publish these musings. emus or bread mold. The simplest naturalist had The English naturalist had built explanation for this shared biology his argument for evolution, in is that all living things inherited it built his argularge part, on the processes he from a common ancestor— ment for evocould observe around him. He did namely, DNA-based microbes that lution, in large lived more than 3.5 billion years not think it would be possible to see life originating now because ago. That common ancestor was part, on the the life that’s already here would already fairly complex, and many processes he prevent it from emerging. scientists have wondered how it In 1871, he outlined the prob- This might have evolved from a simpler essay isobserve the first could lem in a letter to his friend, botanist in a monthly series, with predecessor. Some now argue that more on evolutionary around him. .

Joseph Hooker: “But if (and Oh! roots membrane-bound cells with only online at blogs. what a big if!) we could conceive sciencemag.org/origins RNA inside predated both DNA

DARWIN

CREDIT: KATHARINE SUTLIFF/SCIENCE

L

ike many other scientists raised in temperate latitudes, Charles Darwin was enthralled by his first glimpse of the tropical rain forest. His Beagle diary entry conveyed those immediate and thrilling first impressions, but the encounter with the Brazilian Atlantic Forest had an enduring influence on the development of his ideas over the following decades, with resounding echoes even today in 21st century evolutionary science.

198

9 JANUARY 2009

VOL 323

SCIENCE

and proteins. Later, RNA-based life may have evolved the ability to assemble amino acids into proteins. It’s a small step, biochemically, for DNA to evolve from RNA. In modern cells, RNA is remarkably versatile. It can sense the levels of various compounds inside a cell and switch genes on and off to adjust these concentrations, for example.

It can also join together amino acids, the building blocks of proteins. Thus, the first cells might have tapped RNA for all the tasks on which life depends.

For 60 years, researchers have been honing theories about the sources of the amino acids and RNA’s building blocks. Over time, they have had to refine their ideas to take into account an ever-clearer understanding of what early Earth was like.

In an iconic experiment in 1953, Stanley Miller, then at the University of Chicago, ignited a spark that zapped through a chamber filled with ammonia, methane, and other gases. The spark created a goo rich in amino acids, and, based on his results, Miller suggested that lightning on the early Earth could have created many compounds that would later be assembled into living things.

By the 1990s, however, the accumulated evidence indicated that the early Earth was dominated by carbon dioxide, with a pinch of nitrogen—two gases not found in Miller’s flask. When scientists tried to replicate Miller’s experiments with carbon dioxide in the mix, their sparks seemed to make almost no amino acids. The raw materials for life would have had to come from elsewhere, they concluded.

In 2008, however, lightning began to look promising once again. Cleaves and his colleagues suspected that the failed experiments were flawed because the sparks might have produced nitrogen compounds that destroyed any newly formed amino acids. When they added buffering chemicals that could take up these nitrogen compounds, the experiments generated hundreds of times more amino acids than scientists had previously found.

Cleaves suspects that lightning was only one of several ways in which organic compounds built up on Earth. Meteorites that fall to Earth contain amino acids and organic carbon molecules such as formaldehyde. Hydrothermal vents spew out other compounds that could have been incorporated into the first life forms. Raw materials were not an issue, he says: “The real hurdle is how you put together organic compounds into a living system.”

Step 1: Make RNA An RNA molecule is a chain of linked nucleotides. Each nucleotide in turn consists of three parts: a base (which functions as a

www.sciencemag.org

CREDITS (TOP TO BOTTOM): KATHARINE SUTLIFF/SCIENCE; GEORGE RICHMOND/BRIDGEMAN ART LIBRARY, LONDON (SUPERSTOCK)

NEWSFOCUS

3

4

EVOLUTIONARY ROOTS ORIGINS

soup. “We’ve got the molecules in our RNA, producing the first protocells. “The goal sights,” he says. is to have something that can replicate by itself, Sutherland can’t say for sure where these using just chemistry,” says Szostak. reactions took place on the early Earth, but he After 2 decades, he and his colleagues notes that they work well at the temperatures have come up with RNA molecules that can and pH levels found in ponds. If those ponds build copies of other short RNA molecules. dried up temporarily, They have been able to they would concentrate mix RNA and fatty “Now making an the nucleotides, making acids together in such a conditions for life even way that the RNA gets artificial cell doesn’t more favorable. trapped in vesicles. The Were these Darwin’s vesicles are able to add sound like science warm little ponds? “It fatty acids to their might just be that he fiction any more. It’s membranes and grow. wasn’t too far off,” says In July 2008, Szostak Sutherland. a reasonable pursuit.” reported that he had figured out how proto—HENDERSON JAMES CLEAVES, cells could “eat” and Step 2: The cell CARNEGIE INSTITUTION FOR SCIENCE bring in nucleotides to If life did start out with RNA alone, that RNA build the RNA. would need to make copies of itself without All living cells depend on complicated help from proteins. Online in Science this channels to draw nucleotides across their week (www.sciencemag.org/cgi/content/ membranes, raising the question of how a abstract/1167856), Tracey Lincoln and Ger- primitive protocell membrane brought in these ald Joyce of the Scripps Research Institute in molecules. By experimenting with different San Diego, California, have shown how that recipes for membranes, Szostak and his colmight have been possible. They designed a leagues have come up with protocells leaky pair of RNA molecules that join together and enough to let nucleic acids slip inside, where assemble loose nucleotides to match their they could be assembled into RNA, but not so partner. Once the replication is complete, old porous that the large RNA could slip out. and new RNA molecules separate and join Their experiments also show that these with new partners to form new RNA. In 30 vesicles survive over a 100°C range. At high hours, Lincoln and Joyce found, a population temperatures, protocells take in nucleotides of RNA molecules could grow 100 million quickly, and at lower temperatures, Szostak times bigger. found, they build RNA molecules faster.

Lincoln and Joyce kept their RNA moleHe speculates that regular temperature cules in beakers. On the early Earth, however, cycles could have helped simple protocells surreplicating RNA might have been packed in the vive on the early Earth. They could draw in first cells. Jack Szostak and his colleagues at nucleotides when they were warm and then use Harvard Medical School in Boston have been them to build RNA when the temperature investigating how fatty acids and other mole- dropped. In Szostak’s protocells, nucleotides cules on the early Earth might have trapped are arranged along a template of RNA. Strands of RNA tend to stick together at low temperatures. When the protocell warmed up again, the heat might cause the two strands to pull apart, allowing the new RNA molecule to function.

Now Szostak is running experiments to bring his protocells closer to life. He is developing new forms of RNA that may be able to replicate longer molecules faster. For him, the true test of his experiments will be whether his protocells not only grow and reproduce, but evolve.

“To me, the origin of life and the origin of Darwinian evolution are essentially the same thing,” says Szostak. And if Darwin was alive today, he might well be willing to write a lot Protocell. Researchers at more about how life began.

Harvard are trying to make

“letter” in a gene’s recipe), a sugar molecule, soup. “We’ve got the molecules in our RNA, producingVenus, the firstphallus, protocells. goal or“The pebble? and a cluster of phosphorus and oxygen sights,” he says. is to have something thatknow can replicate by itself, “I don’t much about Art, but I know what atoms, which link one sugar to the next. For Sutherland can’t say for sure where these using just chemistry,” says Szostak.the humorist and art critic I like,” quipped years, researchers have tried in vain to synthe- reactions took place on the early Earth, but he After 2 decades, and hisback colleagues Geletthe Burgess in 1906. For archaeosize RNA by producing sugars and bases, notes that they work well at the temperatures have come up with RNA molecules that logists, distinguishing art can from nonart is still joining them together, and then adding phos- and pH levels found in ponds. If those ponds build copies of quite otherashort RNA Take molecules. challenge. the 6-centimeter-long phates. “It just doesn’t work,” says Sutherland. dried up temporarily, have been able toas the Venus of piece They of quartzite known This failure has led scientists to consider they would concentrate mix RNA and fattyin 1999 next to a Tan-Tan. Found in Morocco “Now making an two other hypotheses about how RNA came to the nucleotides, communicate making acidsof together suchestimated a meaning, whether they be the rich trove stone intools to be be. Cleaves and others think RNA-based life conditions for life eventhat make way that the RNA gets words up our languages, the musibetween 300,000 and 500,000 years old, it artificial cell doesn’t may have evolved from organisms that used a more favorable. cal sounds that convey emotion, or the dra- resembles trapped in vesicles. a human figureThe with stubby arms different genetic material—one no longer Were these Darwin’s vesicles areBednarik, able to addan independent matic paintings that, 30,000 years after their and legs.

Robert sound like science found in nature. Chemists have been able to warm little ponds?

“It caused the discoverers of the Chau- archaeologist fatty acids to intheir creation, based Caulf ield South, use other compounds to build backbones for might just be vet thatCave he to break membranes and grow. down in tears.

Australia, insists that an ancient human fiction any more. It’s nucleotides (Science, 17 November 2000, wasn’t too far off,” While says sites like Chauvet might be vivid deliberately In Julymodified 2008, Szostak the stone to p. 1306). They’re now investigating whether Sutherland. had reasonable pursuit.” examples ofa what some researchers still make itreported look morethat likehe a person. these humanmade genetic molecules, called figured how protoconsider a “creative explosion” that began If so, this objetout d’art is so old JAMES CLEAVES, PNA and TNA, could have emerged on their Step 2: The cell cellscreated couldnot “eat” and when modern humans —HENDERSON colonized Europe that it was by our CARNEGIE INSTITUTIONnumFOR SCIENCE own on the early Earth more easily than RNA. If life did start out with bring inwhich nucleotides about 40,000 years ago, an increasing own species, first to According to this hypothesis, RNA evolved RNA alone, that build the RNA. berRNA of prehistorians are tracing our sym- appears in Africa nearly later and replaced the earlier molecule. would need to make copiesmuch of itself without All living cells depend complicated bolic roots further back in time—and 200,000 yearsonago, but But it could also be that RNA wasn’t put help from proteins.

Online channels to draw nucleotides across their in some cases,intoScience speciesthis ancestral to Homo by one of our ancestogether the way scientists have thought. “If week (www.sciencemag.org/cgi/content/ membranes, raising question sapiens. Like modern humans themselves, tors, the perhaps theof how a you want to get from Boston to New York, abstract/1167856), Traceybehavior Lincolnseems and Gerprimitive protocell membrane brought symbolic to have its origins large-brained H. in these there is an obvious way to go. But if you can’t ald Joyce of theinScripps in have molecules. with different Africa.Research Recent Institute excavations turnedBy experimenting heidelbergensis, get there that way, there are other ways you San Diego, California, have shown how that for membranes, Szostak up elaborate stone tools, beads,recipes and ochre thought by someand his colcould go,” says Sutherland. He and his col- might have been possible. designed a leagues have come up with protocells leaky dating backThey 100,000 or more years ago, anthropologists to leagues have been trying to build RNA from pair of RNA molecules thatresearchers join togetherare andstillenough to let nucleic slip inside, where although debating be theacids common simple organic compounds, such as formaldeloosewhich nucleotides to fmatch their demonstrate they could be assembled RNA, but not so Since their discovery

by Frenchassemble spelunkers of these inds really ancestor into of modhyde, that existed Earth life began. replication is complete,But old there’s porouswidethat the large RNA could slip out. in on 1994, thebefore magnificent lions, partner. horses, Once and the symbolic expression. ern humans and They find they rhinos make better progress toward andwalls new RNA molecules separatethat andthe joinbuildingTheir experiments also show that seem to leap from the of spread agreement blocks Neandertals.

That that these producing RNAChauvet if they combine compo- France with new partners to form newpreceded RNA. In 30 vesicles art. survivewould over a 100°C range. At high Cave inthesouthern have of symbolism full-blown mean that nents of sugars reigned and the components of bases hours,paintLincoln and Joyce a population take in nucleotides as the world’s oldest cave “When wefound, talk about beads andtemperatures, art, we are protocells art is an extremely together insteadings. of separately comRNA and molecules could growabout 100 million quickly, and at ancient lower temperatures, Szostak Expertly making composed in redofochre actually talking material technologies part of the Homo plete sugars andblack bases charcoal, first. timesdemonbigger. for symbolic expression that certainly found, they RNA molecules faster. the vivid drawings post-buildrepertoire.

“Ignoring the Symmetry in stone.

Over the past few years, they have docuLincoln Joycethe kept their RNA mole- thought He speculates regularwe temperature strate that the artistic gift stretches backanddate origins of symbolic and fewthat specimens have Some stone tools mented almost an entire route from years. prebiotic in beakers.communication, On the early Earth, however,by acycles simplepaleoart, protocells surmore than 30,000 Thesecules paintings potentially verycould wide have ofhelped very early require a mental molecules to RNA and aresure preparing to pub- replicating RNAmargin,” might have been packed in the Dietrich vive onStout the early Earth. They could drawimage in to create. are almost to be mentioned in any artisays archaeologist explaining them away, lish even more cle details of their success.

Dis- art. first cells.

Jack of Szostak and his colleagues at nucleotides when were warm andout then use or paper about the earliest But what University College London. orthey rejecting them covering these new reactions School in Boston of have been them build RNA when do they reallymakes tell usSutherabout theHarvard originsMedical of The evolution symbolism wastoonce of hand doesthe nottemperature serve this discipline well,” land suspect it wouldn’t have been that hard investigating how fatty acids andbeen other protocells, nucleotides artistic expression? thought to have asmolerapid as dropped. “flicking In on Szostak’s Bednarik wrote in a 2003 analysis of the for RNA to emergeThe directly from anhumans organicwhocules on the early Earth mightashave trapped Clive are arranged a template of RNA. StrandsAnthropology. prehistoric decorated a light switch,” archaeologist Gamble along Venus of Tan-Tan in Current of RNA tend to stick at low tempera- are skeptical, Chauvet’s walls by torchlight arrived at the of the Royal Holloway, University of LonYettogether many archaeologists When again, resemblance the cave with their artistic genius already in full don, put it some years ago. Buttures. given newthe protocell arguing warmed that theup stone’s to a heatappears might causehuman the two strands to pull flower. And so, most researchers agree that evidence that symbolic behavior figure might be apart, coincidence. Indeed, the newthe RNA molecule to Tan-Tan function.“figurine” is remthe origins of art cannot simply long before cave allowing paintings, debate over the Now Szostak is running to be pegged to the latest discovery THE YEAR OF Gamble now says that his muchiniscent of aexperiments similar controversy over a Recent protocells closer to life.

He is develof ancient paintings or sculpcited comment needsbring to behis modsmaller stone discovered in 1981 at the site of of RNARam that may able to ture. Some of the earliest art ified: “It’s a dimmer oping switchnew now,forms Berekhat in thebeIsraeli-occupied Golan excavations molecules him, likely perished over the ages; a stuttering candle.” replicate longerHeights.

Tofaster. someFor archaeologists, this have turned the truepintest of his experiments will be much remains to be found; and As they more precisely 250,000-year-old object resembles a woman, whether his protocells only up elaborate archaeologists don’t always point when symbolic behavior but othersnot argue thatgrow it wasand shaped by natural reproduce, but evolve. agree on how to interpret what is began, scientists are hoping they forces, and, in any case, looks more like a stone tools, “To me, the penguin origin ofor life and the origin of an exhaustive unearthed. As a result, instead of might one day crack the tougha phallus.

Even after beads, and Darwinian are essentially same chasing after art’s first appearest question of all: What was itsevolution microscopic studytheconcluded that the thing,” says AndRam if Darwin ance, many researchers seek to evolutionary advantage toSzostak.

Berekhat objectwas hadalive indeed been etched ochre dating today, he mightwith wellabe willing to write what a lot some consider understand its symbolic roots. humans?

Didatsymbols, as many tool to emphasize Researchers back 100,000Protocell. more about how life began.

After all, art is an aesthetic This essay continuesHarvard researchers suspect, serve as a its “head” and “arms,” many researchers have are trying to make our more See more ZIMMER simplesocial life forms, shown expression of something more monthlyorseries. glue that helped tribes of rejected it as a–CARL work of art. For some, proof of on humans’ evolutionary here in aearly computer image. to survive Carl Zimmer is the symbolic author of Microcosm: coli and the fundamental: the cognitive abil- journey humans and behaviorE.requires evidence that the years onlineago. at blogs.

New Science of Life.symbols had a commonly understood meanity to construct symbols that sciencemag.org/origins. reproduce?

www.sciencemag.org

SCIENCE

VOL 323

–CARL ZIMMER

Carl Zimmer is the author of Microcosm: E. coli and the New Science of Life.

9 JANUARY 2009

Art and Symbolism

199

CREDITS (TOP (LEFTTO TOBOTTOM): RIGHT): KATHERINE SUTLIFF/SCIENCE; THE BOXGROVE PROJECT CREDITS KATHARINE SUTLIFF/SCIENCE; THE BOXGROVE PROJECT; GEORGE RICHMOND/BRIDGEMAN ART LIBRARY, LONDON (SUPERSTOCK)

simple life forms, shown here in a computer image.

On the Origin of of

CREDIT: JANET IWASA

CREDIT: JANET IWASA

“letter” in a gene’s recipe), a sugar molecule, and a cluster of phosphorus and oxygen atoms, which link one sugar to the next. For years, researchers have tried in vain to synthesize RNA by producing sugars and bases, joining them together, and then adding phosphates. “It just doesn’t work,” says Sutherland.

This failure has led scientists to consider two other hypotheses about how RNA came to be. Cleaves and others think RNA-based life may have evolved from organisms that used a different genetic material—one no longer found in nature. Chemists have been able to use other compounds to build backbones for nucleotides (Science, 17 November 2000, p. 1306). They’re now investigating whether these humanmade genetic molecules, called PNA and TNA, could have emerged on their own on the early Earth more easily than RNA. According to this hypothesis, RNA evolved later and replaced the earlier molecule.

But it could also be that RNA wasn’t put together the way scientists have thought. “If you want to get from Boston to New York, there is an obvious way to go. But if you can’t get there that way, there are other ways you could go,” says Sutherland. He and his colleagues have been trying to build RNA from simple organic compounds, such as formaldehyde, that existed on Earth before life began.

They find they make better progress toward producing RNA if they combine the components of sugars and the components of bases together instead of separately making complete sugars and bases first.

Over the past few years, they have documented almost an entire route from prebiotic molecules to RNA and are preparing to publish even more details of their success. Discovering these new reactions makes Sutherland suspect it wouldn’t have been that hard for RNA to emerge directly from an organic

EVOLUTIONARY ROOTS

DARWIN

www.sciencemag.org SCIENCE VOL 323 9 JANUARY www.sciencemag.org SCIENCE VOL 323 2009 6 FEBRUARY 2009

199

5 709

ORIGINS

ORIGINS tools, which Wynn and many others argue are clear examples of an imposed form based on a mental template. Some have even argued that these skillfully crafted hand axes had symbolic meanings, for example to display prestige or even attract members of the opposite sex.

The half-million-year mark also heralded the arrival of H. heidelbergensis, which had a much larger brain than H. erectus. Not long afterward, our African ancestors began to create a wide variety of finely crafted blades and projectile points, which allowed them to exploit their environment in more sophisticated ways, and so presumably enhance their survival and reproduction.

Archaeologists refer to these tools as Middle Stone Age technology and agree that they did require mental templates. “The tools tell us that the hominid world was changing,” says Wynn. ing and were shared within groups of people. ity to hold an abstract concept in one’s head— As one moves forward in time, humans For example, the hundreds of bone and stone and, in the case of the tool, to “impose” a pre- appear able to imagine and create even more “Venus figurines” found at sites across Eura- determined form on raw material based on an elaborate tools, sharpening their evolutionsia beginning about 30,000 years ago were abstract mental template. That kind of ability ary edge in the battle for survival. By skillfully carved and follow a common motif. was probably not needed to make the earliest 260,000 years ago, for example, ancient They are widely regarded not only as sym- known tools, say Wynn and other researchers. humans at Twin Rivers in what is now Zambia bolic expression, but full-fledged art.

These implements, which date back 2.6 mil- could envision a complex finished tool and Thus many researchers are reluctant to lion years, consist mostly of rocks that have put it together in steps from different compoaccept rare, one-off discoveries like the Tan- been split in two and then sharpened to make nents. They left behind finely made blades Tan or Berekhat Ram objects as signs of simple chopping and scraping implements. and other tools that had been modified— symbolic behavior. “You can imagine [an Then, about 1.7 million years ago, large, usually by blunting or “backing” one edge— ancient human] recognizing a resemblance teardrop-shaped tools called Acheulean hand to be hafted onto handles, presumably made but [the object] still hav[ing] no symbolic axes appeared in Africa. Likely created by of wood. These so-called backed tools have meaning at all,” says Philip Chase, an anthro- H. erectus and probably used to cut plants and been widely regarded as evidence of sympologist at the University of Pennsylvania. butcher animals, these hand-held tools vary bolic behavior when found at much younger Thomas Wynn, an anthropologist at the greatly in shape, and archaeologists have sites. “This flexibility in stone tool manufacUniversity of Colorado, Colorado Springs, debated whether creating the earliest ones ture [indicates] symbolic capabilities,” says agrees: “If it’s a one-off, I don’t think it required an abstract mental template. But by archaeologist Sarah Wurz of the Iziko Musecounts. It’s not sending a message to anyone.” about 500,000 years ago, ancient humans were ums of Cape Town in South Africa. creating more symmetrical Late Acheulean Similar cognitive abilities were possibly Tools of the imagination required to make the famous Given how difficult it is to detect 400,000-year-old wooden spears “If it’s a one-off, I don’t think the earliest symbolic messages in from Schöningen, Germany. One the archaeological record, some recent study concludes that these it counts. It’s not sending a researchers look instead for proxy spears’ creators—probably membehaviors that might have bers of H. heidelbergensis—carmessage to anyone.” required similar cognitive abiliried out at least eight preplanned ties, such as toolmaking. Charles steps spanning several days, —THOMAS WYNN, UNIVERSITY OF COLORADO, Darwin himself saw an evoluincluding chopping tree branches COLORADO SPRINGS tionary parallel between toolwith hand axes and shaping the making and language, probably spears with stone flakes. the most sophisticated form of The idea that sophisticated symbolic behavior. “To chip a toolmaking and symbolic flint into the rudest tool,” Darwin thought require similar cognitive wrote in The Descent of Man, skills also gets some support demands a “perfect hand” as well from a surprising quarter: brainadapted to that task as the “vocal imaging studies. Stout’s team ran organs” are to speaking. positron emission tomography To many researchers, making scans on three archaeologists— sophisticated tools and using Symbolic start. Some scientists argue that this 77,000-year-old engraved ochre all skillful stone knappers—as symbols both require the capac- shows symbolic capacity. they made pre-Acheulean and

Late Acheulean tools. Both methods turned on visual and motor areas of the brain. But only Late Acheulean knapping turned on circuits also linked to language, the team reported last year.

6

710

6 FEBRUARY 2009

VOL 323

SCIENCE

www.sciencemag.org

Color me red At Twin Rivers, it’s not just the tools that hint at incipient symbolic behavior. Early humans there also left behind at least 300 lumps of ochre and other pigments in a rainbow of colors: yellow, red, pink, brown, purple, and blue-black, some of which were gathered far from the site. Excavator Lawrence Barham of the University of Liverpool in the United Kingdom thinks they used the ochre to paint their bodies, though there’s little hard evidence for this. Most archaeologists agree that body painting, as well as the wearing of personal ornaments such as bead necklaces, was a key way that early humans symbolically communicated social identity such as membership in a particular group, much as people today declare social allegiances and individual personalities by their clothing and jewelry.

Yet while the Twin Rivers evidence is suggestive, it’s hard to be sure how the ochre was actually used. There’s little sign that it was ground into powder, as needed for decoration, says Ian Watts, an independent ochre expert in Athens. And even ground ochre could have had utilitarian uses, says archaeologist Lyn Wadley of the University of Witwatersrand in Johannesburg, South Africa. Modern-day experiments have shown that ground ochre can be used to tan animal hides, help stone tools adhere to bone or wooden handles, and Eye of the beholder. Archaeologists debate whether this modified stone was meant to represent a woman. even protect skin against mosquito bites.

“We simply don’t know how ancient people used ochre 300,000 years ago,” Wadley consider diagnostic elements of symbolic says. And since at that date the ochre users behavior came together. And in work now were not modern humans but our archaic in press, the Blombos team reports finding ancestors, some experts are leery of assign- engraved ochre in levels dating back to ing them symbolic savvy.

100,000 years ago (Science, Yet many archaeologists are 30 January, p. 569). willing to grant that our species, There are other hints that the H. sapiens, was creating and sciencemag.org modern humans who ventured Hear author using certain kinds of symbols out of Africa around this time Michael Balter by 75,000 years ago and per- discuss the roots of art at might also have engaged in symhaps much earlier. At sites such www.sciencemag.org/ bolic behavior. At the Skhul rock as Blombos Cave on South darwin. shelter in Israel, humans left Africa’s southern Cape, people 100,000-year-old shell beads left sophisticated tools, including elabo- considered by some to be personal ornarately crafted bone points, as well as perfo- ments (Science, 23 June 2006, p. 1731). At rated beads made from snail shells and the 92,000-year-old Qafzeh Cave site pieces of red ochre engraved with what nearby, modern humans apparently strongly appear to be abstract designs. At this single preferred the color red: Excavators have site, a number of what many archaeologists studied 71 pieces of bright red ochre associ-

Online

CREDIT: FRANCESCO D’ERRICO AND APRIL NOWELL

CREDITS (TOP TO BOTTOM): FRENCH MINISTRY OF CULTURE AND COMMUNICATION/DRAC RHONE-ALPES/DEPARTMENT OF ARCHAEOLOGY; CHRIS HENSHILWOOD AND FRANCESCO D’ERRICO

A roaring start. Researchers agree that Chauvet Cave’s magnificent paintings, including these lions, are full-blown art.

www.sciencemag.org

SCIENCE

VOL 323

ated with human burials. Some researchers argue that this represents an early case of “color symbolism,” citing the universal importance of red in historical cultures worldwide and the apparently great lengths to which early humans went to gather red ochre. ”There is very strong circumstantial evidence for the very great antiquity of the color red as a symbolic category,” says anthropologist Sally McBrearty of the University of Connecticut, Storrs.

These finds of colorful ochre, fancy tools, and beads have convinced many researchers that the building blocks of symbolism had emerged by at least 100,000 years ago and possibly much earlier. But why? What selective advantages did using symbols confer on our ancestors?

To some scientists, the question is a no-brainer, especially when it is focused on the most sophisticated form of symbolic communication: language. The ability to communicate detailed, concrete information as well as abstract concepts allowed early humans to cooperate and plan for the future in ways unique to our species, thus enhancing their survival during rough times and boosting their reproductive success in good times. “What aspects of human social organization and adaptation wouldn’t benefit from the evolution of language?” asked Terrence Deacon, a biological anthropologist at the University of California, Berkeley, in his influential book The Symbolic Species: The Coevolution of Language and the Brain. Deacon went on to list just some of the advantages: organizing hunts, sharing food, teaching toolmaking, sharing past experiences, and raising children. Indeed, many researchers have argued that symbolic communication is what held groups of early humans together as they explored new environments and endured climatic shifts.

As for art and other nonlinguistic forms of symbolic behavior, they may also have been key to cementing these bonds, by expressing meanings that are difficult or impossible to put into words. In that way, artistic expression, including music, may have helped ensure the survival of the fittest. This may also explain why great art has such emotional force, because the most effective symbols are those that convey their messages the most powerfully— something the artists at Chauvet Cave seem to have understood very well.

6 FEBRUARY 2009

–MICHAEL BALTER

7

711

ORIGINS

On the Origin of of

Photosynthesis

8 1286

To catch a photon Over more than 200 years, researchers have ironed out most of the molecular details of how organisms turn carbon dioxide and water into food. Chlorophyll pigment and about 100 other proteins team up to put light to work. Plants, some protists, and cyanobacteria embed their chlorophyll in two large protein clusters, photosystem I and photosystem II. And they need both systems to use water as an electron source. Light jump-starts an electrical circuit in which The electron thief electrons flow from the photosystems Either way, it took some fancy fiddling to through protein chains that convert the primitive reaction make the energy-rich molecules THE YEAR OF centers to oxygen-generating Given its ATP and NADPH. These molephotosystems. Oxygenic photocules then power the synthesis synthesis was a huge upgrade, importance in of the sugars that organisms leading to a land of plenty, says making and depend on to grow and multiply. biochemist John Allen of Queen keeping earth Photosystem II—the strongest Mary, University of London. naturally occurring oxidant— “Water is everywhere, so the lush, photoregains its lost electrons by organisms never ran out of elecsynthesis ranks swiping them from water, genertrons. They were unstoppable.” ating oxygen as a waste product.

But water clings to its elechigh on the However, some bacteria trons. With its oxidizing power, top- 10 list of don’t rely on water as an elec- This essay is the third photosystem II can wrench them monthly series. More tron source, using hydrogen sul- in aevolutionary away, but the reaction centers in on evolution online at fide or other alternatives. These blogs.sciencemag.org/ nonoxygenic photosynthesizers milestones. nonconformists, which today origins. cannot. Biochemists James

6 MARCH 2009

Allen (no relation to John Allen) and JoAnn Williams of Arizona State University, Tempe, and colleagues are working out how a bacterial reaction center could have evolved photosystem II’s appetite for electrons.

Taking a hands-on approach, they have been tinkering with the reaction center of the purple bacterium Rhodobacter sphaeroides to determine if they can make it more like photosystem II. First they targeted bacteriochlorophyll, the bacterial version of chlorophyll that’s at the core of the reaction center, and altered the number of hydrogen bonds. Adding hydrogen bonds hiked the molecule’s greed for electrons, they found.

The water-cleaving portion of photosystem II sports four manganese atoms that become oxidized, or lose electrons. So the team equipped the bacterial reaction center with one atom of the metal. In this modif ied version, the added manganese also underwent oxidation, the researchers reported in 2005.

James Allen says that their creations aren’t powerful enough to split water. But eventually, they hope to engineer a reaction center that can oxidize less possessive molecules, such as hydrogen peroxide, that would have been present on the early Earth. Even if the researchers never replicate photosystem II, “if we def ine the intermediate stages, we’ve accomplished a lot,” he says.

VOL 323

SCIENCE

www.sciencemag.org

CREDIT: NEALE CLARK/GETTY IMAGES

DARWIN

CREDIT: K. SUTLIFF/SCIENCE

Try to picture the world without photosynthesis. Obviously, you’d have to strip away the greenery—not just the redwoods and sunflowers, but also the humble algae and the light-capturing bacteria that nourish many of the world’s ecosystems. Gone, too, would be everything that depends on photosynthetic organisms, directly or indirectly, for sustenance—from leaf-munching beetles to meat-eating lions. Even corals, which play host to algal partners, would lose their main food source.

Photosynthesis makes Earth congenial for life in other ways, too. Early photosynthesizers pumped up atmospheric oxygen concentrations, making way for complex multicellular life, including us. And water-dwellers were able to colonize the land only because the oxygen helped create the ozone layer that shields against the sun’s ultraviolet radiation.

Oxygen-producing, or oxygenic, photosynthesis “was the last of the great inventions of microbial metabolism, and it changed the planetary environment forever,” says geobiologist Paul Falkowski of Rutgers University in New Brunswick, New Jersey.

Given its importance in making and keeping Earth lush, photosynthesis ranks high on the top-10 list of evolutionary milestones. By delving into ancient rocks and poring over DNA sequences, researchers are now trying to

piece together how and when organisms first began to harness light’s energy. Although most modern photosynthesizers make oxygen from water, the earliest solar-powered bacteria relied on different ingredients, perhaps hydrogen sulfide. Over time, the photosynthetic machinery became more sophisticated, eventually leading to the green, well-oxygenated world that surrounds us today. In the lab, some biochemists are recapitulating the chemical steps that led to this increased complexity.

Other researchers are locked in debates over just when this transition happened, 2.4 billion years ago or much earlier.

Looking so far into the past is difficult.

The geological record for that time is skimpy and tricky to interpret. Eons of evolution have blurred the molecular vestiges of the early events that remain in living organisms. But “it’s a terribly important problem,” says biochemist Carl Bauer of Indiana University, Bloomington, one well worth the travails.

live in habitats such as scalding hot springs, don’t generate oxygen. Their photosynthetic proteins huddle in relatively simple “reaction centers” that may have been the predecessors of the two photosystems.

Envisioning the steps that led to this complex biochemistry is mind-boggling.

Similarities between proteins in photosynthetic and nonphotosynthetic bacteria suggest that early microbes co-opted some photosynthesis genes from other metabolic pathways. But protophotosynthesizers might also have helped each other piece these pathways together by swapping genes. Biochemist Robert Blankenship of Washington University in St. Louis, Missouri, and colleagues say they’ve uncovered traces of these lateral gene transfers by comparing complete bacterial genomes.

For example, their 2002 study of more than 60 photosynthetic and nonphotosynthetic bacteria (Science, 22 November 2002, p. 1616) suggested that bugs had passed around several photosynthesis genes, including some involved in synthesizing the bacterial version of chlorophyll.

Gene-sharing might also explain the puzzling distribution of the photosystems, Blankenship says. A cell needs both photosystems to carry out oxygenic photosynthesis.

Yet modern nonoxygenic bacteria have the presumptive predecessor either of photosystem I or of photosystem II, never both. To explain how the two protein complexes wound up together, Blankenship favors “a large-scale lateral [gene] transfer” or even a fusion of organisms carrying each photosystem. However, other researchers remain skeptical, arguing that one photosystem evolved from the other, possibly through the duplication of genes, creating an ancient cell with both. No one knows for sure.

Catching rays. Long before plants got in on the act, photosynthetic cyanobacteria living in pools like this one in Yellowstone National Park were changing the composition of the atmosphere.

says astrobiologist Roger Buick of the University of Washington, Seattle. These hints could push the origin back 600 million years or more.

One line of evidence is oil biomarkers that researchers think are the remains of cyanobacteria. They’ve turned up in rocks that are up to 2.7 billion years old. And in western Australia, thick shale deposits that are 3.2 billion years old hold rich bacterial remains but no traces of sulfur or other possible electron sources, suggesting that the microbes were using water to make energy.

Geologist Euan Nisbet of Royal Holloway, University of London, and colleagues found additional support for an early origin when they went searching for traces of RuBisCO, a key photosynthetic enzyme.

RuBisCO feeds carbon dioxide into the reactions that yield sugars. The enzyme version found in oxygenic photosynthesizers Oxygenic photosynthesis “was the plays favorites: It prefers carbon dioxide that contains the carbonlast of the great inventions of micro12 isotope over the bulkier carbon-13. In 2007, Nisbet and bial metabolism, and it changed the his colleagues found disproportionately low carbon-13 values planetary environment forever.” indicative of RuBisCO activity —Paul Falkowski, Rutgers University when they analyzed organic matter in rocks from three sites Something in the air about 2.4 billion years ago, geologists see in Zimbabwe and Canada that are between How the photosystems got their start is cru- the first unmistakable signs of significant, 2.7 billion and 2.9 billion years old. Nisbet cial for understanding the origin of photo- sustained levels of atmospheric oxygen. concludes that oxygen-making photosynthesis. But the question that’s drawn the These signs include red beds, or layers synthesis began at least 2.9 billion years ago. most attention—and provoked the most tinged by oxidized iron, i.e., rust. Further The early-origin case isn’t ironclad. For wrangling—is when photosynthesis began. support that the GOE marks an atmospheric example, a 2008 paper that has some Most researchers accept that nonoxygenic revolution comes from a technique that researchers fuming claims that the oil biophotosynthesis arose first, probably shortly detects skewed abundances of sulfur iso- markers are contaminants that seeped in after life originated more than 3.8 billion topes that occur if the air lacks oxygen. from younger rocks. Advocates also have to years ago. “Life needs an energy source, and These imbalances persisted until the GOE, explain why it took hundreds of millions of the sun is the only ubiquitous and reliable when they vanished. years for oxygen to build up in the air. energy source,” says Blankenship.

Hard-liners construe these data to mean Although the last word on the origins of The sharpest disputes revolve around that oxygenic photosynthesis could not have oxygen-making photosynthesis isn’t in, when organisms shifted to oxygenic photo- emerged until shortly before the GOE. But researchers say they are making progress. One synthesis. At issue is how to interpret a other scientists disagree. “We are finding thing is for certain, however: Without this watershed in the fossil record known as the more and more hints that oxygenic photo- innovation, Earth would look a lot like Mars.

–MITCH LESLIE great oxidation event (GOE). In rocks from synthesis goes deeper into the fossil record,” www.sciencemag.org

SCIENCE

VOL 323

6 MARCH 2009

9 1287

NEWSFOCUS

ORIGINS embryo that serves as its food supply.

Darwin was perplexed by the diversity of flowering plants; they were too numerous and too varied, and there were too few fossils to sort out which were more primitive.

Throughout much of the 20th century, magnolia relatives with relatively large flowers were leading candidates for the most primihow flowers got started—and from which tive living flowers, although a few ancestor. Today, researchers have analytical researchers looked to small herbs instead. tools, fossils, genomic data, and insights that In the late 1990s, molecular systematics Darwin could never have imagined, all of came to the rescue, with several reports prewhich make these mysteries less abom- senting a fairly consistent picture of the inable. Over the past 40 years, techniques lower branches of the angiosperm tree. An for assessing the relationships between obscure shrub found only in New Caledonia organisms have greatly improved, and gene emerged as a crucial window to the past. sequences, as well as morphology, now help Amborella trichopoda, with its 6-millimeter researchers sort out which angiosperms greenish-yellow flowers, lives deep in the arose early and which arose late. New fossil cloud forests there. In multiple gene-based finds and new ways to study them—with assessments, including an analysis in 2007 synchrotron radiation, for example—pro- of 81 genes from chloroplast genomes vide a clearer view of the detailed anatomy belonging to 64 species, Amborella sits of ancient plants. And researchers from var- at the base of the angiosperm family tree, ious fields are figuring out genomic changes the sister group of all the rest of the that might explain the amazing success of angiosperms. this fast-evolving group.

Given that placement, Amborella’s tiny These approaches have given researchers flowers may hint at what early blossoms a much better sense of what early flowers were like. It’s one of “the most similar living were like and the relationships among them. flower[s]” to the world’s first flower, says But one of Darwin’s mysteries remains: the James Doyle of the University of Californature and identity of the angiosperm ances- nia, Davis. The petals and sepals of its sintor itself. When flowering plants show up in gle-sex flowers are indistinguishable and the fossil record, they appear with a bang, vary in number; so too do the numbers of with no obvious series of intermediates, as seed-producing carpels on female flowers Darwin noted. Researchers still don’t know and pollen-generating stamens on male which seed- and pollen-bearing flowers. The organs are spirally organs eventually evolved into THE YEAR OF arranged, and carpels, rather the comparable flower parts. than being closed by fused tisFor more on “We’re a bit mystif ied,” says sue as in roses and almost all botanist Michael Donoghue of familiar flowers, are sealed by a flower origins, Yale University. “It doesn’t secretion. listen to a appear that we can locate a close Most genetic analyses showed relative of the flowering plants.” that water lilies were the next podcast by branch up the angiosperm tree, author Elizabeth Seeking the first flower followed by a group represented One of two major living groups by star anise, which also has a Pennisi at of seed plants, angiosperms have primitive look about it, says “covered” seeds that develop This essay is the fourth Doyle, “though each of these www. encased in a protective tissue in a monthly series. has deviations from the ancesFor more on evolutionary sciencemag. called a carpel (picture a bean topics tral type.” online, see the at pod). That’s in contrast to the Origins blog org/ nonflowering gymnosperms, blogs.sciencemag.org/ Fossil records origins.

For more on multimedia/ such as conifers, which bear flower Although some fossil pollen origins, listen by author naked seeds on scales. An to a podcast dates back 135 million years, no podcast.

Elizabeth Pennisi at angiosperm’s carpel sits at the www.sciencemag.org/ credible earlier fossil evidence center of the flower, typically multimedia/podcast. exists. In Darwin’s day, and for surrounded by pollen-laden stamany decades afterward, palemens. In most flowers, the carpel and stamens obotanists primarily found leaves or pollen are surrounded by petals and an outer row of but almost no fossil flowers. They had the leaflike sepals. Seeds have a double coating wrong search image, says Else Marie Friis of as well as endosperm, tissue surrounding the the Swedish Museum of Natural History in

On the Origin of of

to British botanist Joseph Dalton Hooker, lamenting an “abominable mystery” that threw a wrench into his theory of evolution: How did flowering plants diversify and spread so rapidly across the globe? From rice paddies to orange groves, alpine meadows to formal gardens, prairies to oakhickory forests, the 300,000 species of angiosperms alive today shape most terrestrial landscapes and much of human life and culture. Their blooms color and scent our world; their fruits, roots, and seeds feed us; and their biomass provides clothing, building materials, and fuel. And yet this takeover, which took place about 100 million years ago, apparently happened in a blink of geological time, just a few tens of millions of years.

The father of evolution couldn’t quite fathom it. Darwin had an “abhorrence that evolution could be both rapid and potentially even saltational,” writes William Friedman in the January American Journal of Botany, which is devoted to this “abominable mystery.” Throughout his life, Darwin pestered botanists for their thoughts on the matter, but they couldn’t give him much help.

Now, 130 years later, evolutionary biologists are still pestering botanists for clues about what has made this plant group so successful, as well as when, where, and

28

10

3 APRIL 2009

DARWIN

VOL 324

SCIENCE

www.sciencemag.org

CREDITS (TOP TO BOTTOM): COURTESY OF STEPHEN MCCABE, UC SANTA CRUZ; PHOTO BY JENNIFER SVITKO, COURTESY OF WILLIAM L. CREPET, CORNELL UNIVERSITY

IN 1879, CHARLES DARWIN PENNED A LETTER

CREDIT: WITH THE KIND PERMISSION OF THEPERMISSION DIRECTOR AND THE DIRECTOR BOARD OFAND TRUSTEES, ROYAL BOTANIC GARDENS, KEW CREDITSREPRODUCED (TOP TO BOTTOM): REPRODUCED WITH THE KIND OF THE THE BOARD OF TRUSTEES, ROYAL BOTANIC GARDENS, KEW; GEORGE RICHMOND/BRIDGEMAN ART LIBRARY, LONDON (SUPERSTOCK)

Flowering Plants

Stockholm. “When we started, the search profile was bigger, a magnolia [flower],” she recalls. But 30 years ago, she and others discovered tiny ancient flowers by sieving through sand and clay sediments. With this technique, they have now collected hundreds of millimeter-size flowers, some preserved in three dimensions, from Portugal and other locations with Cretaceous deposits 70 million to 120 million years old.

This fossil diversity shows that angiosperms were thriving, with several groups well-established, by 100 million years ago. In some, the flower parts are whorled like those of modern flowers; in others they are spiraled, considered by some researchers as the more primitive arrangement. Some flower fossils have prescribed numbers of petals, another modern feature, whereas in others the petal count varies.

In 1998, Chinese geologist Ge Sun of Jilin University in Changchun, China, came across what seemed to be a much older flower. The fossil, called Archaefructus, was an aquatic plant that looked to be 144 million years old. By 2002, Sun and David Dilcher of the Florida Museum of Natural History (FLMNH) in Gainesville had described an entire plant, from roots to flowers, entombed on a slab of rock unearthed in Liaoning in northeastern China.

In one sense, Archaefructus wasn’t much to look at. “It’s a flowering plant before there were flowers,” Dilcher notes. It lacked petals and sepals, but it did have an enclosed carpel. When Kevin Nixon and colleagues at Cornell University compared its traits with those same traits in 173 living plants, Archaefructus came out as a sister to living angiosperms and closer to the common ancestor than even Amborella.

Archaefructus’s distinction was shortlived, however. Within months, better dating of the sediments in which it was found yielded younger dates, putting this f irst flower squarely with other early fossil flower parts, about 125 million years old.

Also, a 2009 phylogenetic analysis of 67 taxa by Doyle and Peter Endress of the University of Zurich, Switzerland, placed the fossil in with water lilies rather than at the base of the angiosperms, although this conclusion is contested.

Out of the past.

Tiny Amborella sits at the bottom of the angiosperm family tree.

from one of the nonflowering seed plants or gymnosperms, whose heyday was 200 million years ago. Modern gymnosperms include conifers, ginkgoes, and the cycads, with their stout trunks and large fronds.

Before angiosperms came along, these plants were much more diverse and included cycadlike species, such as the extinct Bennettitales, and many woody plants called Gnetales, of which —Peter Crane, a few representatives, including the University of Chicago joint firs, survive today (see family tree, p. 31). Also common in the Jurassic were These fossils often spark debate because seed ferns, a group now long gone; their specimens tend to be imperfectly preserved most famous member is Caytonia, which and leave room for interpretation. To help seems to have precarpel-like structures. remedy that, Friis and her colleagues have These g roups’ perceived relevance to begun to examine flowers using synchro- flower evolution and their relationships to tron radiation to generate a 3D image of angiosperms have ping-ponged between their inner structures, allowing the fossil to camps, depending on how the evolutionary remain intact while Friis peers inside it trees were constructed. from many angles (Science, 7 December In the mid-1980s, Peter Crane, now at 2007, p. 1546). “We can get fantastic reso- the University of Chicago in Illinois, prolution,” says Friis. “It’s really exciting.” But posed a solution, the anthophyte hypothesis. so far, the flowers Friis finds are Using several lines of evidence and noting too diverse to trace back to a that both Bennettitales and particular ancestor. “From Gnetales organize their male these fossils, we cannot and female organs together say what is the basic in what could be conform,” she says. strued as a preflower, he considered them, Before flowers along with angiosperms, Although they have yet as comprising a single to find the oldest fossil angiosperm entity called Larger than life. Although merely flowers, researchers anthophytes. For the next 2.2 millimeters in diameter, this 3D assume that the ancesdecade, most family trees fossil flower shows that grasses date tral angiosperm evolved based on morphology supback to 94 million years ago.

“We are realizing that this huge diversity is probably the result of one innovation piled on top of another innovation.”

www.sciencemag.org

SCIENCE

VOL 324

3 APRIL 2009

11

29

ORIGINS

ORIGINS

Angiosperms

such a causal relationship is not settled. Later, animals that ate fruit and SEED PLANT Paleozoic seed ferns dispersed seeds likely helped evolvPHYLOGENY ing species expand quickly into new Asterids » territory. Some think the answer lies Eudicots in genes: duplications that gave the angiosperm genome opportunities to Rosids try out new floral shapes, new chemical attractants, and so forth. This Monocots flexibility enabled angiosperms to exploit new niches and set them up for long-term evolutionary success.

Magnolias » “My own view is that in the past, we have looked for one feature,” says Crane. Now, “we are realizing that Water lilies this huge diversity is probably the result of one innovation piled on top ? of another innovation.”

Archaefructus » The latest insights into diversification come from gene studies. From Amborella 2001 to 2006, Pamela Soltis of the FLMNH and Claude dePamphilis ? of Pennsylvania State University, Caytonia University Park, participated in the Floral Genome Project, which ?

Bennettitales searched for genes in 15 angiosperms.

Now as a follow-up, the Ancestral Angiosperm Genome Project looks at gene activity in five early angioCycads sperms and a cycad, a gymnosperm.

DePamphilis and his colleagues Ginkgoes » matched all the genes in each species against one another to deter? Gnetales mine the number of duplicates. They then looked at the number of differences in the sequences of each gene Conifers » pair to get a sense of how long ago Extinct taxa the duplication occurred. In most early angiosperms, including water lilies and magnolias, they saw many Shifting branches. As this simplified family tree shows, gene studies have helped clarify the relationships of many simultaneous duplications—but not living angiosperms, but fitting in extinct species is still a challenge, and some nodes are hotly debated. in Amborella, they reported in the January 2009 American Journal of Botany, The Floral Genome Project also looked are not as well-defined as they are in Araconfirming earlier reports. The data suggest to see whether the genetic programs guiding bidopsis. This sloppiness may have made that a key genome duplication happened flower development were consistent development flexible enough to undergo after the lineage leading to Amborella split throughout the angiosperms. “We found that many small changes in expression patterns off but before water lilies evolved. “We’re there are fundamental aspects that are con- and functions that helped yield the great beginning to get the idea that polyploidiza- served in the earliest lineages,” says Soltis. diversity in floral forms. tion may have been a driving force in creat- “But there are differences in how the genes In his letter to Hooker, Darwin wrote ing many new genes that drive floral devel- are deployed.” that he would like “to see this whole probopment,” dePamphilis says.

Take the avocado, a species on the lower lem solved.” A decade ago, Crepet thought Others have noted that a duplication branches of the angiosperm tree. In most Darwin would have gotten his wish by now. occurred in the evolution of grasses, and the angiosperms, the flower parts are arranged in That hasn’t happened, but Crepet is optiFloral Genome Project confirms that yet concentric circles, or whorls, around the mistic that he and his colleagues are on the another duplication paved the way for eudi- carpels, with stamens innermost, then petals, right track, as analyses of various kinds of cots, the group that includes apples, roses, and finally sepals. Each tissue has its own data become more sophisticated. “We are beans, tomatoes, and sunflowers. “There are distinct pattern of gene expression, but not less likely to go around in circles in the next some real ‘hot spots’ in angiosperm evolu- in the avocado. Genes that in Arabidopsis 10 years,” he says. “I believe a solution to tionary history,” says dePamphilis, who is are active only in, say, the developing petals the problem is within reach. … The mystery working to fully sequence the genome of spill over in avocado to the sepals. Thus in is solvable.”

–ELIZABETH PENNISI Amborella with his colleagues. the more primitive plants, petals and sepals

Flowers, food, fuel. at the the diversity diversityofofangiosperms. angiosperms.Given Given that they represent nine in 10and landsense plants, it’s no surprise as fuel. Darwin Darwin marveled marveled at mainstays of both our welfare of beauty.

Clockwisethat fromthey left:serve aspens, that they represent ninewelfare in 10 land no surprise that they as aspens, orchids,orchids, grasses,grasses, sunflowers, tulips, apples, mainstays of both our and plants, sense ofit’sbeauty.

Clockwise fromserve top left: sunflowers, tulips,walnuts. apples, walnuts.

12 30

Inside and out. Synchrotron radiation helped produce a 3D rendering (gold) of this fossil male flower (right) and insights into its internal structure.

3 APRIL 2009

VOL 324

SCIENCE

www.sciencemag.org

CREDITS: (ARCHAEFRUCTUS IMAGE) DAVID DILCHER; (ALL OTHER IMAGES) PHOTOS.COM

Seeds of success The angiosperm’s ancestor may be missing, but what is very clear—and was quite annoying to Darwin—is that the angiosperm prototype so readily proved a winner. Seed ferns and other gymnosperms arose about 370 million years ago and dominated the planet for 250 million years. Then in a few tens of millions of years, angiosperms edged them

out. Today, almost nine in 10 land plants are angiosperms.

The exact timing of the angiosperms’ explosion and expansion is under debate, as is the cause. At least one estimate based on the rate at which gene sequences change—that is, the ticking of the molecular clock— pushes angiosperm evolution back to 215 million years ago. “There appears to be a gap in the fossil record,” says Donoghue, who also notes that molecular dating methods “are still in their infancy” and, thus, could be misleading. He and others think that flowering plants lingered in obscurity for tens of millions of years before radiating toward their current diversity.

Whatever the timing, there was something special about the angiosperm radiation. During the 1980s and again in 1997, Cornell’s Karl Niklas compiled a database showing the first and last occurrences of fossil plants. When he and Crepet used that and more recent information to look at species’ appearances and disappearances, they found that new angiosperms appeared in bursts through time, whereas other plants, such as gymnosperms, radiated rapidly only at first. Moreover, angiosperms proved less likely to disappear, somehow resisting extinction, says Crepet.

Once the angiosperms arrived, how did they diversify and spread so quickly? Darwin suspected that coevolution with insect pollinators helped drive diversification, though

CREDITS CREDITS (TOP (TOP TO TO BOTTOM): BOTTOM): PHOTOS.COM; PHOTOS.COM; ELSE ELSE MARIE MARIE FRIIS FRIIS

work points you in another direction,” Crane says.

And if the molecular work is correct, then the field doesn’t know in which direction to turn, because in most analyses the genetic data don’t place any living plant close to angiosperms. The angiosperms group together, the living gymnosperms group together, and there’s nothing in between. “The nonangiosperm ancestor just isn’t there,” says paleobotanist William Crepet of Cornell. “I’m starting to worry that we will never know, that it transformed without intermediates.”

Living gymnosperms

ported this idea. Crane and others carefully dissected and described fossils of these groups, looking for the precursors to carpels, the seed’s double coat, and other distinctive angiosperm traits.

But they have run into problems. “We do not really know how to compare them because the structures are very differentlooking; figuring out what’s homologous is quite a difficult thing,” says Crane. He and his colleagues argue, for example, that the seeds in the Bennettitales have two coverings, which may be a link to angiosperms.

But in the January American Journal of Botany, Gar Rothwell of Ohio University, Athens, and two colleagues disagree, saying that what Crane calls the outer layer is the only layer, and f ind fault with the hypothesis in general.

To make matters worse for anthophyte proponents, genebased evolutionary trees break up this grouping, pulling the Gnetales off any angiosperm branch and placing them among or next to the other gymnosperms.

“The molecular work points in one direction; the paleobotanical

www.sciencemag.org

SCIENCE

VOL 324

3 APRIL 2009

13 31

TIVES PERSPECTIVES

E

HISTORY HISTORY OF OF SCIENCE SCIENCE HISTORY OF SCIENCE

atmospheric, oceanic, geological, ecological, and cultural phenomena across the globe. Humboldt’s obsession with geographically referenced measurements and collections was central to his vision. He recognized that spatial arrays of observations could be aggregated to reveal patterns that would in turn reveal underlying processes—such as the distribution of incident radiation, the transport of heat and materials in winds and ocean currents, the influence of temperature on plant form, and the effect of latitude and continentality on mountain snowline.

He expanded this vision in the succeeding years, establishing international cooperative networks of meteorological and geomagnetic measurement stations, inventing isotherms and other graphical devices to portray spatial patterns, and noting that plant form is often better predicted by local environment than by taxonomic affinity (a paradox resolved by Darwin).

Humboldt’s genius lay in his geographical vision, and in his intuition that Earth’s land surface, oceans, atmosphere, and inhabitants form an integrated whole, with linkages among the various components (4, 5). Humboldt’s general physics of the Earth envisioned climate as a major control of Earth-surface phenomena, with vegetation serving as both an index of climate and a proximal control of microclimate, animal habitat, and cultural practices (6–8).

Humboldt’s dream of systematic observational arrays across the globe took hold in the 19th century. Throughout the century, countless Humboldt-inspired explorations were launched, each involving systematic measurement and mapping of physical, biological, and often cultural features of landscapes and oceans (8–10).

These surveys were relentlessly inductive,

von Humboldt and vonAlexander Humboldt and the General Physics l Physics of the Earthof the Earth

In early In the the early 19th 19th century, century, Alexander Alexander von von In the early 19th century, Alexander von In the early 19th century, Alexander von Humboldt laid the foundations Humboldt laid the foundations for today’s today’s Humboldt laid the foundations for for today’s Humboldt laid the foundations for today’s Earth Earth system system sciences. sciences.

Earth system sciences.

Earth system sciences.

Stephen Stephen T. T. Jackson Jackson Stephen T. Jackson

with geographi14 596 596 ents596 and collec-

better predicted by local environment than by ically producing detailed descriptive reports 1 2009 VOL www.sciencemag.org MAY 2009 resolved VOL 324 324 bySCIENCE SCIENCE www.sciencemag.org 11 MAY MAY 2009 VOL 324 SCIENCE www.sciencemag.org taxonomic affinity (a paradox with little integration within or among the com-

RIGINAL COPY IN THE NATURAL HISTORY MUSEUM, PARIS

A

CREDIT: CREDIT: REPRODUCED REPRODUCED FROM FROM AN AN ORIGINAL ORIGINAL COPY COPY IN THE THE NATURAL NATURAL HISTORY HISTORY MUSEUM, MUSEUM, PARIS PARIS CREDIT: REPRODUCED FROM AN ORIGINAL COPY ININ THE NATURAL HISTORY MUSEUM, PARIS

A

sss scientists scientists are are celebrating celebrating the the 200th 200th scientists are celebrating the 200th anniversary of Charles Darwin’s anniversary of of Charles Charles Darwin’s Darwin’s birth birth anniversary birth rating the 200thand and the the 150th 150th anniversary anniversary of of the the pubpuband the 150th anniversary of the publication s Darwin’s birthof lication of his his On On the the Origin Origin of of Species, Species, lication of his On the Origin of Species, Darwin’s rsary of the pub- ideas Darwin’s ideas continue continue to to shape shape and and enrich enrich Darwin’s ideas continue to shape and enrich the sciences (1).

6 May 2009 marks the sciences sciences (1).

(1). 66 May May 2009 2009 marks marks the the 150th 150th the the 150th gin of Species, anniversary of the death of another 19th-cenanniversary of the death of another 19th-cenanniversary of the death of another 19th-cenhape andtury enrich tury figure—Alexander figure—Alexander von von Humboldt— Humboldt— tury figure—Alexander von Humboldt— marks the 150th whose scientific whose scientific legacy legacy also also flourishes flourishes in in the the whose scientific legacy also flourishes in the 21st century.

Humboldt helped create nother 19th-cen21st century. century. Humboldt Humboldt helped helped create create the the 21st the intellectual n Humboldt— intellectual world world Darwin Darwin inhabited, inhabited, and and his his intellectual world Darwin inhabited, and his writings inspired Darwin to embark writings inspired Darwin to embark on on flourisheswritings in the inspired Darwin to embark on H.M.S. Beagle.

More pertinent to our time, H.M.S. Beagle.

More pertinent to our time, H.M.S. Beagle.

More pertinent to our time, lped create the Humboldt Humboldt established established the the foundation foundation for for the the Humboldt established the foundation for the habited, Earth and his system sciences: the integrated system Earth system system sciences: sciences: the the integrated integrated system system Earth of on to embark on of knowledge knowledge on which which human human society society may may of knowledge on which human society may depend depend in the the face face of of global global climate climate change. change. ent to our time,in depend in the face of global climate change.

Darwin, Like Darwin, Humboldt Humboldt undertook undertook aaa Like Darwin, Humboldt undertook undationmajor forLike the voyage that would shape major voyage voyage that that would would shape shape his his ideas ideas his ideas tegrated major system and thinking.

Humboldt spent 5 years and thinking.

Humboldt spent 5 years (1799 and thinking. Humboldt spent 5 years (1799 (1799 man society maywith to to 1804) 1804) with botanist botanistAimé Aimé Bonpland Bonpland explorexplorto 1804) with botanist Aimé Bonpland exploring Venezuela, the northern limate change. ing Venezuela, Venezuela, the the northern northern Andes, Andes, and and cencening Andes, and central tral Mexico, Mexico, with visits visits to to Tenerife, Tenerife, Cuba, Cuba, and and dt undertook a with tral Mexico, with visits to Tenerife, Cuba, and the United theideas United States.

States. They They collected collected botanical, botanical, United States.

They collected botanical, hape histhe zoological, zoological, geological, geological, and and ethnological ethnological specspecgeological, and ethnological specnt 5 yearszoological, (1799 imens, made extensive atmospheric imens, made made extensive extensive atmospheric atmospheric and and geogeo- Intellectual imens, and geoIntellectual riches. riches. The The central central portion portion of of Humboldt’s Humboldt’s Physical Physical Tableau Tableau of of the the Andes Andes and and Neighboring Neighboring Intellectual riches.

The central portion of Humboldt’s Physical Tableau of the Andes and Neighboring Bonplandphysical explor- measurements, physical measurements, and and recorded recorded the the Countries, published as part of (2, 3), shows Chimborazo in profile, with vegetation zones, plant species, physical measurements, and recorded the Countries, published as part of (2, 3), shows Chimborazo in profile, with vegetation zones, plant species, and and Countries, published as part of (2, 3), shows Chimborazo in profile, with vegetation zones, plant species, and geographic Andes, and cen- location geographic location of of their their thousands thousands of of snowline geographic location of their thousands of snowline depicted depicted at at appropriate appropriate elevations. elevations. In In the the original, original, the the profile profile is flanked on on both both sides sides by by tables tables snowline depicted at appropriate elevations.

In the original, the profile isis flanked flanked on both sides by tables specimens specimens and tens tens of of thousands thousands of of measuremeasure- describing specimens and tens of thousands of measuredescribing elevational elevational patterns patterns in in temperature, temperature, humidity, humidity, light light refraction refraction and and intensity, intensity, agriculture, agriculture, fauna, fauna, and and erife, Cuba, and and describing elevational patterns in temperature, humidity, light refraction and intensity, agriculture, fauna, and ments.

Humboldt spent the next 22 years and other physical, chemical, and biological features. ments.

Humboldt spent the next 22 years and other physical, chemical, and biological features. ments.

Humboldt spent the next 22 years and other physical, chemical, and biological features. ected botanical, most most of of his his inherited inherited fortune fortune in in Paris, Paris, prepremost of his inherited fortune in Paris, prehnological specparing and publishing 45 volumes of paring and publishing 45 volumes of aaa nevernever- distribution distribution of of incident incident radiation, radiation, the the transport transport serving serving as as both both an an index index of of climate climate and and aaa proxproxparing and publishing 45 volumes of neverdistribution of incident radiation, the transport serving as both an index of climate and proxspheric and geo-report finished on his travels. of heat and materials in winds and ocean curimal control of microclimate, animal habitat, finished report on his travels. of heat and materials in winds and ocean curimal control of microclimate, animal habitat, finished report on his travels. of heat and materials in winds and ocean curimal control of microclimate, animal habitat, Intellectual riches.

The central Humboldt’s Physical TableauPhysical of the Andes and Neighboring Countries, as part Intellectual riches.

The portion centralofportion of Humboldt’s Tableau of the Andes andpublished Neighboring these the first was slim work rents, the of on and cultural practices (6–8). d recordedOf of (2, 3), shows Chimborazo in profile, with vegetation zones, plant species, and snowline depicted appropriate elevations.

Ofthe these volumes, volumes, thepublished first was aaa as slim work rents, the influence influence of temperature temperature on plant plant andatzones, cultural practices (6–8).

Of these volumes, the first was slim work rents, the influence of temperature on plant and cultural practices (6–8).

Countries, part of (2, 3), shows Chimborazo in profile, with vegetation plant species, and entitled on the Plants form, and effect of and continenHumboldt’s dream systematic In the the profile of is flanked sides bythe tables describing elevational patterns in temperature, humidity, lightof entitled Essay onoriginal, the Geography Geography of Plantson both form, and the effect of latitude latitude andprofile continenHumboldt’s dream of systematic observaobservaentitled Essay on the Geography of Plants form, and the effect of latitude and continenHumboldt’s dream of systematic observair thousands ofEssay snowline depicted at appropriate elevations.

In the original, the is flanked on both sides by tables refraction and intensity, agriculture, fauna, and other physical, chemical, and biological features.

(2, tional (2, 3).

3). The The modest modest title title belies belies the the intellectual intellectual tality tality on on mountain mountain snowline. snowline. tional arrays arrays across across the the globe globe took took hold hold in in the the (2, 3).

The modest title belies the intellectual tality on mountain snowline. tional arrays across the globe took hold in the nds of measuredescribing elevational patterns in temperature, humidity, light refraction and intensity,century. agriculture, fauna,the and richness He richness within. within. In In the the text text and and accompanying accompanying He expanded expanded this this vision vision in in the the succeeding succeeding 19th 19thcentury. century.Throughout Throughoutthe thecentury, century,countless countless richness within.

In the text and accompanying He expanded this vision in the succeeding 19th Throughout century, countless ext 22 years and other physical, chemical, and color the Humboldt lays out years, establishing color plate plate (see (see the figure), figure), Humboldt lays outbiological years, features. establishing international international cooperative cooperative Humboldt-inspired Humboldt-inspiredexplorations explorationswere werelaunched, launched, color plate (see the figure), Humboldt lays out years, establishing international cooperative Humboldt-inspired explorations were launched, aaa vision of a comprehensive “general physics networks of meteorological and geomagnetic each involving systematic measurement and e in Paris, previsionsof ofscientists comprehensive “general physics networks of meteorological and geomagnetic each involving systematic measurement and vision aa comprehensive “general physics networks of meteorological and geomagnetic each involving systematic measurement and arenothing celebrating the a200th and geophysical measurements, inhabited, and his writings inspired isotherms Darwin to atmospheric the Earth” aimed at less than synmeasurement stations, inventing mapping of physical, biological, and often culumes of aof neverdistribution of incident radiation, the transport serving as both an index of climate and a proxof the Earth” aimed at nothing less than a synmeasurement stations, inventing isotherms mapping of physical, biological, and often culof the Earth” aimed at than a birth syn- embark measurement stations, isotherms mapping of physical, biological, and often culDarwin’s on H.M.S. Beagle.inventing More pertinent to our and of their anniversary ofnothing Charlesless recorded the geographic location thesis atmospheric, oceanic, geological, and other graphical devices to portray spatial tural of and oceans (8–10). thesis of ofand atmospheric, oceanic, geological, and other graphical devices tocontrol portray spatial tural features features of landscapes landscapes and oceans (8–10). thesis of atmospheric, oceanic, geological, and other graphical devices to portray spatial tural features of landscapes (8–10). of150th heat anniversary and materials windstime, and ocean curimalthe of microclimate, habitat, foundation for thousands the of thein publiHumboldt established ofanimal specimens andand tensoceans of thousands ecological, and cultural phenomena across the patterns, and noting that plant form is often These surveys were relentlessly inductive, typecological, and cultural phenomena across the patterns, and noting that plant form is often These surveys were relentlessly inductive, typecological, and cultural phenomena across the patterns, and noting that plant form is often These surveys were relentlessly inductive, of his On the Origin of Species, Darwin’s system of(6–8).

22 the Earth on system sciences: thecultural integratedpractices cation measurements. Humboldt spent the nexttypwas a slim work rents, the influence of temperature plant and globe.

Humboldt’s obsession with geographibetter predicted by local environment than by ically producing detailed descriptive reports globe.continue Humboldt’s obsession withthe geographibetter predicted by local environment thandeby years icallyand producing detailed descriptive reports globe.

Humboldt’s obsession with geographibetter predicted by local environment than by ically producing detailed descriptive reports shape and enrich sciences in Paris, ideas to of knowledge on which human society may most of his inherited fortune raphy ofcally Plants form, and the effect of latitude and continenHumboldt’s dream of systematic observareferenced measurements and taxonomic affinity (a resolved with within or among the cally referenced measurements and colleccollectaxonomic affinity (a paradox paradox resolved by by preparing with little little integration integration within or amongof theacomcomcally referenced measurements and collectaxonomic affinity (a paradox resolved by with little integration within among the comof pend of global climate change.

(1).

6 May 2009 marks the 150th anniversary and publishing 45 or volumes nevin the face tions was central to his vision.

He recognized Darwin).

Humboldt’s genius lay in his geoponent entities.

However, for aaa few intellectus the intellectual tality on mountain snowline. tional arrays across the globe took hold in the tions was central to his vision.

He recognized Darwin).

Humboldt’s genius lay in his geoponent entities.

However, for few intellectutions was central to his vision.

He recognized Darwin).

Humboldt’s genius lay in his geoponent entities.

However, for few intellectuthe death of another 19th-century figure—AlexLike Darwin, Humboldt undertook a major er-finished report on his travels. that arrays observations could be graphical vision, and in intuition that nimble participants—including Charles that spatial spatial arrays ofexpanded observations could be invoyage graphical vision, and19th in his his intuition that ally allyOfthe nimble participants—including Charles d accompanying Heof this vision the succeeding century.

Throughout century, countless that spatial arrays of observations could be graphical and in his intuition that ally nimble participants—including Charles von Humboldt—whose scientific legacy his ideas and thinking. these volumes, the first was a slim work ander thatvision, would shape aggregated to reveal patterns that would in Earth’s land surface, oceans, atmosphere, and Darwin, T. H. Huxley, Matthew Maury, Asa aggregated to reveal patterns thatHumboldt would in Humboldt Earth’s land surface, oceans, atmosphere, and explorations Darwin,Essay T. H. H. Huxley, Matthewof Maury, Asa aggregated to reveal patterns that would in Earth’s land surface, oceans, atmosphere, and Darwin, T. Huxley, Matthew Maury, the establishing 21st century. the Geography PlantsAsa (2, flourishes in spent 5 years (1799 to 1804) with entitled on umboldt also lays out years, international cooperative Humboldt-inspired were launched, turn reveal underlying processes—such as the inhabitants form an integrated whole, with Gray, C. Hart Merriam, and Peter Kropotkin— turn reveal underlying processes—such as the inhabitants form an integrated whole, with Gray, C. Hart Merriam, and Peter Kropotkin— turn reveal underlying processes—such as the inhabitants form an integrated whole, with Gray, C. Hart Merriam, and Peter Kropotkin— Darwin intellectual richhelped create the intellectual world botanist Aimé Bonpland exploring Venezuela, 3).

The modest title belies the “general physics networks of meteorological and geomagnetic each components involving systematic measurement anddata and experilinkages among the various (4, these explorations provided linkages among theand various components (4, ness thesewithin. explorations provided data and and experiexperilinkages among the various components (4, these explorations provided data northern Andes, central Mexico, with In the text and accompanying color the g less than a syn- measurement stations, inventing isotherms mapping of physical, biological, and often cul5).

Humboldt’s general physics of the Earth ence that spurred the development of biogeog5).

Humboldt’s general physics of the Earth ence that spurred the development of biogeogCuba, and the United States. plate (see the figure), Humboldt lays out a vivisits to Tenerife, 5).

Humboldt’s general physics of the Earth ence that spurred the development of biogeogBotany Department and Program in Ecology, University of Botany Department Department and and Program Program in in Ecology, Ecology, University University of of Botany envisioned climate as aaa major control of raphy, oceanography, and other envinic, geological, and other graphical devices toThey portray spatial features of landscapes oceans (8–10).

Wyoming, envisioned climate astural major control of sion raphy, ecology, oceanography, and otherof envigeological, “general physics the collected botanical, zoological, of ecology, aand comprehensive envisioned climate as major control of raphy, ecology, oceanography, and other enviWyoming, Laramie, Laramie, WY WY 82071, 82071, USA. USA. E-mail: E-mail: jackson@ jackson@ Wyoming, Laramie, WY 82071, USA. E-mail: jackson@ uwyo.edu Earth-surface phenomena, vegetation ronmental sciences (11). uwyo.edu Earth-surface phenomena, with vegetation ronmental sciences (11).less omena across the patterns, and noting that plant and ethnological specimens, made extensive Earth” aimed at nothing than a synthesis of form is often Thesewith surveys were relentlessly inductive, typuwyo.edu Earth-surface phenomena, with vegetation ronmental sciences (11).

typically producing detailed descriptive reports with little integration within or among the component entities. However, for a few intellectually nimble participants—including Charles Darwin, T. H. Huxley, Matthew Maury, Asa Gray, C. Hart Merriam, and Peter Kropotkin— these explorations provided data and experience that spurred the development of biogeography, ecology, oceanography, and other environmental sciences (11).

Unfortunately, the conceptual unification among the sciences of the Earth that Humboldt sought never developed in the century following his death. Disciplinary specialization played a large role in eclipsing Humboldt’s integration, as did 20th-century trends toward reductionism, experimentalism, and fine-scale processes in many disciplines.

A new incarnation of Humboldt’s general physics of the Earth began to emerge with the plate tectonics revolution in the 1960s. Drawing on Humboldtian spatial arrays of observations, this theory provided a unified explanatory framework for disparate geophysical, geological, paleontological, and biogeographic phenomena.

Today, a second, even broader manifestation of Humboldt’s vision aspires to understand the interactions and feedbacks among the components of the Earth system, encompassing the lithosphere, atmosphere, hydrosphere, cryosphere, and biosphere as well as human societies and economies. This effort is often referred to as Earth system science, but it could just as well be designated “general physics of the Earth,” using the early-19th century definition of physics as the study of the material world and its phenomena (which we now call science).

Global environmental change may be the greatest challenge faced by human societies since the advent of agriculture. Humboldt advocated for science that spoke to human needs and concerns (5). It is fitting that on the 150th anniversary of his death, we recognize his role in fostering the sciences that speak to the most profound human concerns—sustainability of human societies and the ecosystems on which they depend.

References and Notes

1. P. J. Bowler, Science 323, 223 (2009).[Abstract/ Free Full Text] 2. A. de Humboldt, Essai sur la géographie des plantes (Levrault, Schell & Co., Paris, 1807).

3. An English translation of (2) is currently in press at the University of Chicago Press, Chicago.

4. A. de Humboldt, Personal Narrative of Travels to the Equinoctial Regions of the New Continent, During the Years 1799–1804, by Alexander de Humboldt and Aime Bonpland; with Maps, Plans, &c. (Longman, Hurst, Rees, Orme, & Brown, London, 1814 to 1829), vols.

1 to 7.

5. A. von Humboldt, Cosmos: A Sketch of the Physical Description of the Universe (Johns Hopkins Univ. Press, Baltimore, 1997), vol. 1.

6. M. Nicolson, Hist. Sci. 25, 167 (1987). [ISI] 7. M. Nicolson, in Romanticism and the Sciences, A. Cunningham, N. Jardine, Eds. (Cambridge Univ. Press, Cambridge, 1990), p. 169.

8. M. Dettelbach, in Cultures of Natural History, N. Jardine, et al., Eds. (Cambridge Univ. Press, Cambridge, 1996), p. 287.

9. S. F. Cannon, Science in Culture: The Early Victorian Period (Dawson, New York, 1978).

10. W. H. Goetzmann, New Lands, New Men: America and the Second Great Age of Discovery (Viking, New York, 1986).

11. P. J. Bowler, The Norton History of the Environmental Sciences (Norton, New York, 1993).

In the early 19th century, Alexander von Humboldt laid the foundations for today’s Earth system sciences.

15

BOOKS ET AL. mer students who challenged his ideas as they gained intellectual independence, and debated the pro-evolution (but anti-Haeckel) Jesuit priest and entomologist Erich Wasby Robert J. Richards mann—the list could go on University of Chicago and on. These were not isoPress, Chicago, 2008. duced important books. Robert lated episodes but rather 579 pp. $39, £27.

J. Richards, the director of the moments in a lifelong camISBN 9780226712147.

University of Chicago’s Fishpaign to advance his philosobein Center for the History of phy, which was accompanied H. G. Bronn, Science and Medicine and a by a bitter hostility to orgaErnst Haeckel, and much-published author on Darwin nized religion. the Origins of and German Romantic biology, Richards does not negGerman Darwinism has written a biography of Haeclect Haeckel’s science proA Study in Translation kel. Sander Gliboff, a professor per, treating us to fascinatand Transformation in Indiana University’s Departing and original discussions ment of History and Philosophy of his pathbreaking systemby Sander Gliboff of Science, places Haeckel at the atic and phylogenetic work MIT Press, Cambridge, MA, end of a study that examines the on radiolaria and other marine 2008. 271 pp. $35, £22.95. larger process through which organisms, the importance ISBN 9780262072939.

Darwin’s words were translated, of linguistic analysis to his and his ideas modified, in the phylogenetic trees of the context of German biology. Both illuminate races of humans, and his remarkable experithe twists and turns that evolutionary thought mental work with siphonophores. These contook in Germany, but they do so in dramati- stitute important contributions to our undercally different ways. standing of the technical development of Richards’s book, though over twice as long evolutionary biology. as Gliboff’s, is the more entertaining read of The big picture here, however, is an arguthe two. In his characteristically rich and ment about the power of personality—at least rolling prose, Richards weaves a compelling one personality—to shape the course of scistory of a life marked by tragedy and of an ence. In Richards’s presentation, German evointense, larger-than-life figure whose passions lutionism was profoundly shaped by both drove his scientific research and philosophy. In Haeckel’s charisma and his combativeness.

Richards’s rendering, the scientific Haeckel Perhaps the late-19th-century opposition of cannot be understood separately from the evolutionary science to Christianity would not man’s personality and private circumstances. have been so fiery, he suggests, had Haeckel His love of nature was surpassed only by his not continually fanned its flames. And love for his first wife, Anna Sethe, who died in although Richards absolves Haeckel of perabdominal agony on his 30th birthday. Over sonal responsibility for fascism and Nazism, the next year, he wrote his way through the in part by situating him firmly in his time and despair that enveloped him, producing his place, he does show how the scientist’s ardent foundational work, Generelle Morphologie temperament led him to the occasional intem(1). Although he remarried, the union was not perate statement that could be taken up by happy, and passionate love would elude him extreme thinkers. One cannot leave this book until his sixties, when he had a secret affair that without a deep appreciation for Haeckel as a ended tragically with the death of his lover. tragic figure and for the force of personality in Science remained his salvation and refuge. shaping the direction science may take.

His professional life was also filled with Gliboff’s account is of a completely differdrama, much of which centered on his philos- ent order. His is not a story of personalities or ophy of evolutionary monism—a science- private lives (although he mentions salient centered faith that became one of the most details), but of German academics seeking to successful alternatives to the Judeo-Christian live up to the highest (if changing) ideals of religion among those searching for a secular Wissenschaft and of the ways in which spirituality. Haeckel could not turn down a Darwin’s theory was translated into this envifight: He battled the physician-statesman ronment. He thus situates Haeckel at the end Rudolf Virchow over the role of evolution in of a revised intellectual history of 19ththe schools (Haeckel argued that it should century German evolutionism. Central to his replace religious education), sparred with reli- account is the idea of translation, which he giously conservative scientists and with for- uses both synchronically, especially in treatThe Tragic Sense of Life Ernst Haeckel and the Struggle over Evolutionary Thought

Making German Evolution: Translation and Tragedy Lynn K. Nyhart

I

n this year of Darwin anniversaries (the 200th year of his birth and the 150th anniversary of On the Origin of Species), The Tragic Sense of Life and H. G. Bronn, Ernst Haeckel, and the Origins of German Darwinism remind us that the history of evolutionary thought in the 19th century extended well beyond Darwin himself. Darwin did not launch his theory onto an unprepared public and scientific community, nor was the evolutionism that developed after 1859 a mere extension of his views—it was not even one thing. How, then, should we think about the history of evolution in the 19th century? What sorts of accounts best help us understand the reception of Darwin’s theory, its relations to earlier ideas about nature, the directions that evolutionary investigation subsequently took, and the relations of all of these to the broader social, cultural, and religious concerns scientists shared with their contemporaries?

These questions become especially pointed when one considers German Darwinism, and especially Germany’s best-known follower of Darwin, Ernst Haeckel. Most often remembered by biologists as the author of the biogenetic law (“ontogeny recapitulates phylogeny”), Haeckel has also been accused of promoting European fascism via his monistic philosophy and of presenting a eugenic, biologically determinist vision of humanity that led to Hitler’s “final solution.” Can one scientist be responsible for so much? Most historians would say no, arguing that it takes a community, rather than an individual, to make a movement; that single-cause explanations are insufficient to account for something as broad as fascism; and that an individual cannot be held responsible for the ways in which others (such as Hitler) took up his ideas and molded them to new agendas after his death. But that still leaves open the questions of how to write responsibly about what Haeckel actually believed and how we should situate him in the history of evolutionary thought.

The historians under consideration here have chosen two radically different strategies to understanding Haeckel’s place within German evolutionism, and both have proThe reviewer is at the Department of the History of Science, University of Wisconsin, Madison, WI 53706–1393, USA. E-mail: lknyhart@wisc.edu

16

1170

27 FEBRUARY 2009

VOL 323

SCIENCE

www.sciencemag.org

CREDIT: ERNST HAECKEL/FROM WANDERBILDER (W. KOEHLER, GERA-UNTERMHAUS, 1905)

HISTORY OF SCIENCE

BOOKS ETAL. ing the translation of Origin of Species into translations involved an attempt to recast existGerman, and (more intriguingly) diachroni- ing German terms in a newer, more up-to-date cally, as scientists reworked older words such mode that encompassed selection yet tamed as “perfection” and “type” to lend them new Darwin’s emphasis on unpredictability to meet meanings. Gliboff’s own clear, crisp prose is the more rigorous requirements of a German key to the success of this analysis, as he deftly academic scientist’s understanding of a “law” leads his reader through dense philosophical of organic nature. Simultaneously, Bronn and terminological thickets with nary a thorn sought to translate Darwin’s ideas about selecscratch. This is some of the best close reading tion into a language without an exact equivaI have seen. It also represents a profound chal- lent for the term, and for an academic audience lenge to our standard picture of 19th-century lacking the gentlemanly traditions of breeding German biology. pigeons and dogs so central to Darwin’s expoThe old story, crudely put, is that Haeckel’s sition. The selection metaphor was further version of evolution was a Darwinism in name fraught with an anthropomorphism foreign to only, best understood as an update on early- Germans, who were not brought up on British 19th-century idealistic morphologists such natural-theological assumptions about a peras Carl F. Kielmeyer and J. F. Meckel that sonified God who had created a perfectly retained their teleology, their typological adapted nature. Bronn’s translation, though it emphasis on form, and their linear recapitula- altered key ideas to make Darwin comprehentionism. This story, emphasizing the long per- sible to a German academic audience, was not sistence of a German transcendental approach a conservative throwback. It represented the to nature, has been deeply entrenched in the dynamic engagement of a leading paleontolohistory of biology. gist who had also long been working on many Gliboff challenges this history right from of the questions Darwin claimed as his own— the beginning. The ascription of simple linear a critical yet generous equal, who saw himself recapitulationism to the views of Romantic as moving science forward through the modiembryologists, he notes, owes much to a carica- fications he made to Darwin’s flawed theory. ture developed by Karl Ernst von Baer in a Bronn’s death in 1862 afforded him little polemical context, then adopted uncritically by chance to steer the conversation further. influential historians such as E. S. Russell and Stephen Jay Gould.

Gliboff’s fresh reading of the original sources interprets Kielmeyer and Meckel as far less rigidly typological in their orientation and much more attentive to nature’s variability than has been seen before. Both for these early-19thcentury naturalists and for their intellectual heirs, Gliboff argues, the critical issue was to understand nature’s manifold variety while seeking out underlying strict natural laws to account for it.

This provides a new starting point for analyzing Darwin’s first translator, the prominent paleontologist H. G. Bronn—a figure little attended to in the standard story but the lynchpin of Gliboff’s. Intriguingly and plausibly, Gliboff argues that Bronn’s use of terms like “vervollkommnet” (perfect) as translations for Darwin’s “improved” or “favored” were not about dragging Darwin backward into a German teleological view of nature (as has been claimed by those who have paid attention to Bronn at all). A painter, too. Haeckel’s oil landscape of highlands in Java, from Instead, Gliboff asserts, Bronn’s Wanderbilder (1905). www.sciencemag.org

SCIENCE

VOL 323

And so, finally, we come to Haeckel.

Gliboff ’s key insight here is that Haeckel originally read Bronn’s translation of Darwin, not Darwin in the original. Gliboff shows Haeckel as both echoing and responding to Bronn’s concerns, rather than either reflecting directly on Darwin’s original writing or reaching directly back to the Romantic embryologists. (Although Gliboff acknowledges the centrality of monism to Haeckel’s thought, he focuses on the working evolutionary theorist, not the popular ideologue.) Like Bronn himself, Haeckel made further amendments both terminological and intellectual, and Gliboff rereads Haeckel’s research program as one not dominated by a typological and linear-recapitulationist mindset but rather as continuing to wrestle with the need to account for variability and unpredictable change in terms of mechanistic laws of nature—among which Haeckel included, at the top of his list, natural selection. Haeckel’s Darwinism thus shows continuity with early19th-century concerns, mediated through Bronn. But those concerns were always more flexible than has been acknowledged, and their articulation changed over time. Of course Haeckel’s Darwinism was not Darwin’s own, but it was not an aberration or a distortion of some true theory, any more than any other post-Darwinian additions or adjustments were. It was science moving on.

Gliboff ’s overall picture of scientific advance, in contrast to Richards’s emphasis on charisma and passion, is one of scientists building and innovating incrementally, working with what their predecessors have handed them and sculpting it into something new yet understandable to those around them. His sensitive reading allows us to see post-1859 German evolutionists as rational actors rather than irrationally stuck in some early-19thcentury moment with unmodern commitments. By challenging the very foundations of the standard narrative of German morphology, this careful, compelling account does at least as much as Richards’s to undermine the association of 19th-century German Darwinism with a dangerously exceptional view of nature. But the two books offer very different reads. Is scientific progress a matter of personal anguish and triumph, or of intellectual chugging along? Our concept of it should be capacious enough to include both.

References and Notes 1. E. Haeckel, Generelle Morphologie der Organismen (Georg Reimer, Berlin, 1866).

2. The reviewer previously served as a press reader for both books at the manuscript stage.

27 FEBRUARY 2009

10.1126/science.1169621

17

1171

REVIEW REVIEW

up, in explaining this case bypresent in- distribuby something (later identified as genetic(the mutascience, shaped by the organisms’ own activities so- vancing through a predetermined sequence of ulation became divided oceanicofislands. tions), but it was effect), not aimed anytoo onepermitted direction stages within each family, driven by force derived dependent acts of migration opments that would push other naturalists toward tions into terms past migrations, called Lamarckian butinthis Here, Darwin followed Lyell in seeing that Darwin bioand, thus, left adaptive evolution essentially an evolutionary vision during the years he worked extinctions and (for but not multiple vectors of change.

Evolution hadopento be from individual development.

REVIEW opments that would push other naturalists toward geography must become a historical ended. He allowed a limited role for variation in isolation. By the late 1850s, the idea of profor Lyell) evolutionary adaptations. depicted as a branching tree in which each act of an evolutionary vision during the years he worked science, explaining present distribushaped by the organisms’ own activities (the sogressive evolutionhiswastheory widelytorecognized, andorderly the by geographical branching was thevariants result ofin a amore less was unpredict an pattern of proposals of the individual variants in a population was the relations among species. Several in Byevolution the late 1850s, the idea of protions in Populations terms pastdivided migrations, called Lamarckian effect), but this tooorpermitted was widely recognized, It has been ofargued that Darwin’s move of isolation. progressive of the individual population positive role of individual competition was being opments that would push other naturalists toward barriers will develop independently predictable migration of organisms to a new locarelations. away essentially undirected ruled out any possibility available in the 1830s deflected attention opments that would push other naturalists toward extinctions and (for Darwin but multiple vectors of change.

Evolution had to be gressive evolution was widely recognized, and the was inspired and the positive role of individual competition essentially undirected ruled out any possibilto a more historical viewpointnot articulated by thinkers such as argued Herbert Peter J. Bowler that evolution could be shaped by a predeter- from the model of the branching an evolutionary vision during the years was he Spencer worked asevolutionary each adapts to its new environtion.depicted At the same time, Darwin’s undermined Itby has been Darwin’s move to a tree (11). for Lyell)the adaptations. as a branching treeshaped intheory whichby each act of an evolutionary visionby during years he worked positive role of individual competition being was being articulated thinkers such asthat Herbert could be a predeGerman romanticism [e.g. (12)], but a ity that evolution (Fig. 1). But keymore aspects of the viewpoint Darwinian vision in isolation.

By the late 1850s, the idea of proment in its own way, and the posthe old idea that species were idealized types, Populations divided by geographical branching was the result of a more or less unhistorical was inspired by German William Sharp Macleay’s quinary or circular mined developmental trend.

There was no obin isolation.

By the late 1850s, the idea of proarticulated by thinkers such as Herbert Spencer Peter J. Bowler But key aspects of the Darwinwas no incentive was provided by Spencer (Fig. 1). termined developmental trend.

There more practical were truly original andwidely wouldrecognized, not have occurred Charles Darwin’svious theory of natural has been hailed of theofmost innovative supposed(Fig. barriers will develop of organisms a new order. locagressive evolution was and thea more practical sibility thatindependently barriers be crossed fixedpredictable elementsmigration in a clearly defined to natural romanticism [e.g.

(12)], but system classification that goal toward selection which it was aimed, and as it one gressive evolution wasthe widely recognized, and the can 1).every But keytruly aspects of the Darwinian vision and would not have on the ian vision were original obvious goal toward which it was aimed, and biogeographical insights gained as each adapts to its new environtion.

At the same time, Darwin’s theory undermined to any be other naturalist at the time.

Here, Wallace contributions todid modern science. an When first pattern proposed in 1859, however, it was widely rejected positive role of individual competition was being occasionally allows for the branchSpecies had to be treated as populations of varyincentive was provided by the biogeographical genus contained five species that could arnot produce orderly of relations positive role of individual competition was being were trulytooriginal andnaturalist would not have occurred Charles Darwin’s theory of natural selection has been hailed as one of the most innovative Here, produce anspecies orderlywere pattern of relations occurred any comparison: other at the time.

Beagle voyage (1831–36).

The Galapagos it did not ment in its own way, and the posthe old idea that idealized types, provides a good He too moved toward by his contemporaries, even by those who accepted the general idea of evolution.

This article articulated by thinkers such as Herbert Spencer ingHerbert process Spencer of evolution that Darwin individuals, with no fixed limit on the range Peter J. Bowler insights gained onHere, the Beagle voyagePeter (1831–36). rangeditinwas a circle; each family five genera, sonaturalist species. When The accusation that the theory however, articulated by thinkers such provided as J. Bowlering to any and other at the time.

Wallace contributions to between modern science. first proposed in 1859, widely rejected accusation that the theomost obvious example Wallace provides a good comparison: He too between species. inThe species sibility that barriersthe can be crossed fixed elements the idea of branching evolution drivenprovided by local identifies those depended aspects ofonDarwin’s work that led him to develop this revolutionary theory, hierarchy.provides (Fig. 1). aBut keycomparison: aspects of He the too Darwinian vision conceived in vision the late 1830s. It was of possible variation.a clearly defined natural order.

The Galapagos species the most obvious on of through the taxonomic Chambers’s variation indicated con- idea (Fig. 1).

But key aspects of the Darwinian good moved toward by his contemporaries, even “random” by those who accepted the the general evolution.

This article occasionally allows forwithin the branchSpecies hadon to be treated as variation populationsindicated of varythe idea of branching evolution group can be ry depended “random” of how the relations moved toward adaptation, but even he did not share Darwin’s including his studies of biogeography and animal breeding, and his recognition of the role played were truly original andevolution would notdriven have occurred Charles Darwin’s theory natural selection been as one ofrevolutionary the most byofapproaching the aproblem of the example of how the relations within aCharles group can Vestiges of theinnovative Natural of idea Creation cerns ofofof hisDarwin’s opponents onthat thishas score. were truly original andingwould have occurred Darwin’s theory of naturalwith selection beenthe hailed branching local identifies those aspects work led himAshailed toDarwin develop this theory, History the processnot that Darwin individuals, no fixedhas limit rangeas one of the most innovative but even heby did not the ing concerns of his opponents ononthis score. driven byoflocal adaptation, explained byevolution supposing that an original The Tree of Life insight that the work of the animal breeders throws by the struggle for existence. to any other naturalist at the time.

Here, Wallace contributions to modern science.

When first proposed in 1859, however, it was widely rejected origin of new species through a study be explained by share supposing that an original pop- to modern evolution terms of parallel lines adhimself made clear, variation was certainly caused to any other naturalistconceived at the time.

Here, contributions science.

When first proposed in 1859, however, it was widely rejected adaptation, but even he did not Darwin’s including his studies of biogeography and animal breeding, and hisdepicted recognition of theinrole played inbecame the late Wallace 1830s. Itup, wasin this case possible variation. made clear, variation was divided share insight that the work of the As of Darwin himself population light onDarwin’s of natural provides athe good comparison: He selection. toodivided moved toward by the his contemporaries, even by(later thoseidentified who accepted the general of evolution. article biogeography One innovation the heart of Darwin’s ulation became up, in this case invancing through This a predetermined sequence ofprocess by existence. something as genetic muta- idea provides a good comparison: Heoftoo moved towardthat by hisbycontemporaries, even by at those who accepted the theory general idea of evolution. This article by approaching the problem of theDarwin was insight that the work of the animal breeders throws by struggle for identified oceanimal breeders throws light ondriven the process of certainly caused by something by independent acts of his migration theory was both original disturbing. pre-existing onesthis in arevolutionary progressive lead- by light he those publication Darwin’s Theso Tree of Life the idea of process branching evolution by local identifies aspectsof ofCharles work that led him to develop theory, led driven to construct model oftoopentoday is(later hard us to this revolutionary theory, dependent acts ofand migration to oceanic islands. stages within eachsequence family, driven forceThe derived tions), but itDarwin’s was not aimed inOnany one direction of new species through a study the idea of branchingorigin evolution by local identifies those seems aspects ofobvious Darwin’s workthat thatit led him for to develop on the of natural selection. as genetic mutations), but itand was not radical aimed in anic islands.

Here, Darwin followed Lyell selection.

It was not just that the idea offollowed natural selection ing up opento and humans (5).individual But if the general idea of natural thehis Origin of thus, Species in 1859 evolution is widely adaptation, but even he did not share Darwin’s including studies of biogeography and animal breeding, his recognition of the role played ended, divergent evolution.

Wallace appreciate just how new how it of biogeography that Darwin was One innovation at the heart of Darwin’s theory Here, Darwin Lyell in seeing that biofrom development. and, left adaptive essentially including his studies of biogeography and animal breeding, and his recognition of the role played adaptation, but even he did not share Darwin’s The theory theory was was both both original original and disturbing. pre-existing ones innot a progressive sequence leadhe publication of Charles Darwin’s On he publication ofHeCharles Darwin’s On of vi- challenged Thethat and disturbing. one direction and, thus, left adaptive mustand become evolution entirely new, Darwin’s any seems in seeing that biogeography thework beliefofthat the world was designed evolution waswas not entirely new, Darwin’s vision supposed to existence. have initiated a revolution construct his model of openso time. obvious today thathad it isproposed hard for usevoto the animal breeders throws by thethe struggle developed a similar model tested at the Lamarck that must becomeby a historical ended. allowed for variation insight that the work ofled thetoanimal breeders throws the struggle was for existence. was not just that thethe idea ofgeography natural selection up to humans (5). But if the general idea of Itinsight Originfor of Species in 1859a islimited widelyrole ing in 1859 is widely that the idea of natural selection open-ended.

He allowed a Origin of Species sion of how the process worked certainly was.

It was not just lution essentially a historical science, explaining present ended, divergent evolution.

Wallace appreciate just how new and how radical it by a wise and benevolent God. There was a wider of how the process worked certainly was.

Alboth inthe science and in Western culture.

Yet there light on the process of natural selection. it during his explorations in South there might be natural processes adapting spescience, the organisms’ own activities (the was so- not entirely new, Darwin’s vision challenged the belief that the world light on the process of natural selection. wasexplaining designed present distribuevolution supposed shaped to havebyinitiated a revolution developedAmerica a similar model and testedArchipelago at thefortime.

Lamarck had proposed that theory was eventually paralleled the world was designed theDarwin organdistributions in terms past migrations, supposed to have initiated a revolution belief that limited role shaped Although elementtheory of the teleology or goal-directedness almost though thethe theory eventually paralleled by challenged have been frequent claims that Darwinism was was both original and disturbing. pre-existing inwas aworked progressive sequence publication of Charles Darwin’s Onthis too and theof Malay cieswas to of changes invariation their environment.

But tions inwas terms of past migrations, called Lamarckian effect), but permitted The theory was both original and disturbing. pre-existing ones in a progressive sequence leadhe publication Charles Darwin’s On by wise and benevolent God. There a wider of how the ones process certainly was.leadAl- by aThe both inhe science and in Western culture.

Yet there it during his explorations in South there might be natural processes adapting speinthe science and in Western culture.

Yet there its basic outa Darwin but not for by Wallace, Darwin had conceived by a wise and benevolent God. There was isms’ own activities (the so-called Lamarckian extinctions and (for both universally accepted at the time.

Most thinkers— Wallace, Darwin had conceived its basic outline somehow “in the air” at the time, merely waiting It was not just that the idea ofextinctions natural selection ing had up to if the general ideaby of element ofclaims Species in Darwinism 1859 is widely (modern Indonesia). was perhaps the first toisrealize that ing if adaptation and (for Darwin butthe notOrigin multiple vectors of change.

Evolution to humans be It was not just that the idea of natural selection up to humans (5).

But if the general idea of ofcies Species in 1859 widely of teleology or goal-directedness almost though the theory (5). wasBut eventually paralleled have been Origin frequent that was Americaevolutionary and the Malay Archipelago to in their environment. But Darwin that Darwinism was butchanges this too vectors have been frequent line thewas latenot 1830s, after his Darwin’s returnthe from the wider element of teleology or goal-directedness adaptations. including Jean-Baptiste Lamarck and Chambers— ineach theinlate 1830s, after his return from voyage for someone to put few as readily available points challenged the belief world was designed evolution entirely new, vision supposed to aclaims have a revolution Adrian Desmond andPopulaJames totothe local environment was themultiple onlyifmechanism for Lyell) evolutionary adaptations. a branching tree in which act of that the world was designed was not entirely new, Darwin’s vision challenged the belief Lyell) supposedeffect), have initiated a permitted revolution universally accepted atthat thethe time.

Most thinkers— Wallace, Darwin had conceived its basic outline somehow “in thedepicted air” at theinitiated time, merely waiting (modern Indonesia). was perhaps the first to realize that evolution adaptation took it for granted that theGod. development ofadivided life on by geographical oforhow H.M.S. Beagle.

He worked on it inwas. relative merely waitvoyage of H.M.S. Beagle.

He worked on itAlin almost accepted at the was time.

Most had tomajor be depicted as a process worked certainly was. Al- by a wise and benevolent divided byhave geographical barriers will somehow air” at the time, of change.

Evolution together in“in thethe right way [for instance (1)].

The universally tionsGod. by a wise and benevolent There wider of the process worked certainly both in science and in Western culture.

Yet there Moore recently proposed that of evolution, there would be implications Populations branching was the result of a more less unThere was a wider of how the both in science and in Western culture.

Yet there for someone to put a few readily available points in the late 1830s, after his return from the voyage including Jean-Baptiste Lamarck and Chambers— Adrian Desmond and James to the local environment was the only mechanism earth represents the unfolding of a Lamarck coherent plan isolation over the over next 20eventually years, until the arrivalthe of took fact for that Alfred Russel Wallace (Fig. 1) indepennext 20 and branching asofeach to its ing someone to claims put few readily available relative isolation thinkers—including branching tree in which each act of independently element teleology orthe goal-directedness almost though theory was paralleled by have been that Darwinism was toof Darwin’s hatred slavery prompted for the whole by which species are the clas-theory was eventually paralleled by element of teleology develop barriers will develop predictable migration of(1)]. organisms aH.M.S. new the locaor goal-directedness almost though have been frequent claims thatsystem Darwinism was Moore have recently proposed thatadapts of evolution, there would be major implications it forofgranted thatJean-Baptiste development of life on independently Beagle.

Hethe worked onyears, it in until relative together in frequent the right waya[for instance The aimed at a predetermined goal.

(This assumption Wallace’s paper inhad 1858 precipitated the flurry of universally dently together formulated athe theory natural selection in in right way [for instance in 1858 precipitated arrival ofover Wallace’s paper Chambers—took for granted the developthe result ofsystem aDarwin’s more or mentor lessspecies unpredictable in itsprompted own way, and(13). the points accepted at the time.

Most thinkers— Wallace, Darwin conceived its basic outline somehow “in the air” atWallace theof time, merely waiting his move toward evolutionism sified groups. in geology, as each adapts to its new environ-“in thewas tion.

At the same time, theory undermined hatred of thinkers— slavery for the whole bywaiting which are clasatDarwin’s theenvironment time.

Most Wallace, Darwin had conceived its basic outline universally accepted new somehow air” atinto the time, merely represents theit unfolding ofthat a coherent plan isolation the next 20 years, until the arrival of earth fact that Alfred Russel (Fig. 1)Darwin’s indepenis stillofpreserved in therepresents very term “evolution”; the activity leading toafter theleading publication of publication the Origin.

1858 is taken as evidence forthat this position.

But idealized Alfred Russel Wallace (Fig. life on earth the unfolding of of organisms to a new location.

At possibility that barriers can be crossed oc(1)].

The fact that the flurry of activity to the ment migration including Jean-Baptiste Lamarck and Chambers— in the late 1830s, his return from the voyage for someone to put a few readily available points Because many slaveholders argued Charles Lyell, had shown how his uniformitarian his move toward evolutionism (13). sified into groups.

Darwin’s mentor in geology, ment in its own way, and the posthe old idea species were types, including Jean-Baptiste Lamarck and Chambers— in the late 1830s, after his return from the voyage for someone to put a few readily available points dently formulated a theory of natural selection in Wallace’s paper in 1858 precipitated the flurry of aimed at a predetermined goal. (This assumption Latin evolutio tothe theadevelopment unrolling of a barriers scroll.)

Historians have He quarried Darwin’s notebooks Darwin hadthecreated the outlines the theory Because many slaveholders argued Charles Lyell, had shown how uniformitarian itpreserved for granted that of life on cantogether of H.M.S. Beagle. worked onofitthe in relative is of(1)]. natural of the Origin. aimed at predetermined goal. undermined for therace branching process 1) independently formulated theory atook coherent planrefers the same time, Darwin’s theory allows together in way [for The activity that the black was separately theory would allow the biogeographer to the re-Beagle. He worked on it in relative took it for granted thatcasionally sibility that be crossed fixed elements in aof clearly defined natural order. the development of life on of H.M.S. in the right way [for instance (1)].

Thehis still in the very term “evolution”; the leading to the publication Origin.

1858 is taken asright evidence forainstance this position.

But thatevolution the created black racefrom was plan separately theory would allow theofidealized biogeographer tofixed re- the next 20 years, until the arrival of earth represents the unfolding The represents explanatory framework centered onallows the for fact andHistorians letters tothe establish the complex process by 20 years earlier, and there were significant dif- isolation earth the unfolding of a coherent plan over next 20 years, until the arrival of fact that Alfred Russel Wallace (Fig. 1) indepenthe white, Darwin have quarried Darwin’s notebooks in the very in the selection in 1858 is taken as evidence for this (This assumption is still preserved old idea that species were types, of that Darwin conceived construct the migrations species on an everoccasionally the branchSpecies had to be treated as populations of varyof a coherent isolation over that Alfred Russel Wallace (Fig. 1) indepenHistorians have quarried Darwin’s notebooks Latin evolutio refers to the unrolling of a scroll.)

Darwin had created the outlines of the theory created from thetoby white, Darwin construct athe migrations of species on an everaimedexplanatory at a predetermined goal.ing (This assumption Wallace’s paper in 1858the precipitated flurryby of The dently formulated a theory natural selection in show that all races share changing Populations couldWallace’s sometimes elements clearly defined natural order.

Spelate 1830s.

It was approaching the probprocess of evolution that Darwin individuals, with no fixed limitand on the range goal.wanted (This assumption paper in 1858 precipitated the flurry of aimed at a predetermined dently formulated a theoryinearth. of natural selection in framework centered on the letters to establish complex the process 20 years earlier, ing and there of were significant difwanted toa common show that ancestry, all races share changing earth.

Populations could sometimes is still preserved in the very term “evolution”; 1858 is taken asofevidence this position. But activity leading to the publication of the Origin. and he through realized become by position. geographical barriers, soleading that to the publication of the Origin. be treated as populations of varying of new species haddivided to for lem ofterm the origin conceived in thethelate 1830s. possiblefor variation. is still preserved in the very “evolution”; the activity 1858 Itiswas taken ascies evidence this But a common ancestry, and he realized become divided by geographical barriers, so that of a scroll.)

Historians have quarried Darwin’s notebooks Latin evolutio refers to the unrolling Darwin had created the outlines of the theory that this could defended whatwhat waswas once a single could split into limit on the Historians range of have quarried Darwin’s notebooks Latin evolutio refers atothat was individuals, with fixed study of biogeography that be Darwin by approaching the problem of the thethis unrolling ofclaim a scroll.)

Darwin had created the outlines the species theory claim could be defended once no aofsingle species could split into Life significant dif- and letters to establish the complex process by The explanatory frameworkorigin centered on the 20 years earlier,The and Tree thereofwere bycentered extending the throughout multiple branches adapting to separate environof new species through a study variation. led construct model of open-ended, theidea and environletters to establish the complex process by The explanatory framework 20 years earlier,possible andmultiple there were significant by to extending the his ideaon throughout branches adapting difto separate theevolution. animal kingdom.

As a basis fora ments (10).

Evolution divergent of biogeography that Darwin was One innovation at the heart of Darwin’s theory developed divergent the animal kingdom.

AsWallace a basis for ments (10).

Evolutionwould wouldbecome become aa divergent his thinking, this to process, some over and and led to construct his model of openseems so obvious today that it is hard for us to his thinking, this thesis is sure to is sure process, somebranches branchessplitting splitting over The Treewith ofwith Life similar model and tested itthesis during his exmuch controversy, but if overover again, whereas to dead end Fig. generate generate much controversy, butand if the Malay again, whereas others came to a dead end ended, divergent evolution. Wallace appreciate just how new and how radical it One innovation at theothers heartcame of Darwin’s theory in South America plorations Fig.2.2.Tree Tree of Life, Darwin’s notebooks of Life, fromfrom Darwin’s notebooks (22). (22). acceptedaccepted it would the through extinction. itemphasize would emphasize the through developed a similar model and tested was at the time. Lamarck had proposed that today that it is hard for us to seems soextinction. obvious Archipelago (modern Indonesia). crucial crucial role played his move toward a model These rigidly structured models of taxoimage of treeofoflife had appeared appeared inin roleby played by his toward a model These rigidly structured models of taxoTheThe image of thethe tree had it during his explorations in South there might be natural processes adapting spehow radical it was The idea of common descent now seems so Adrian Desmond andmove James Moore have appreciate just how new and life of branching evolutionevolution based on based geographical and evolution mademade good good sense sense Darwin’s notebooksofofthe thelate late 1830s 1830s (Fig. 2)2) nomic of branching onhatred geographical nomicrelations relations evolution Darwin’s notebooks (Fig. America and the Malay Archipelago cies to changes in their environment. But Darwin wonder what alternative of slavat the time.

Lamarckindependently had proposed that there obvious that weand might recently proposed that Darwin’s diversity. to anyone embedded in a vision of nature as a and was proposed by Wallace in diversity. to anyone embedded in a vision of nature as a and was proposed independently by Wallace in (modern Indonesia). was perhaps the first to realize that if adaptation might be natural processes species models could have beengoverned proposed account for Thisery prompted move evolutionism model was so his radical that toward many late orderly system by to a divine a paper published in 1855. adapting Both realized that itto predictable, This modelmany was so radical thatargued many that late predictable, orderly system governed byproposals a divine (13). a paper published in 1855. Both realized that it plan.

Adrian Desmond and James to the local environment was the only mechanism But Darwin was Because slaveholders changes in their the relations among Severalto evolutionists were unable to accept Such a world view species. made it difficult ac- 19th-century explained whyenvironment. naturalists were able to arrange 19th-century evolutionists were unable to accept plan.

Such a world view made it difficult to acexplained why naturalists were able to arrange Moore have recently proposed that of evolution, there would be major implications it in full.

Mayr that the theory of from the that theinhistory of lifedeflected on earth attention might be away species groups within groups, using descentto cept the 1830s created perhaps theinto first to realize that if adaptation theErnst black raceargued was separately available it in full.

Ernstonewanted Mayr argued that that the theory of cept that the history ofbranching life on earth mightWilbe white, species groups ancestor within Darwin’s hatred of slavery prompted for the whole system by which species are clascommon descent was of Darwin’s greatest and unpredictable, dependant frominto aenvironment common to explain the descent under- essentially tree (11). the local wasgroups, the onlyusing mechanism from theirregular model of the Darwin to show all races common descent was one of Darwin’s greatest essentially andquinary unpredictable, dependant from a common to explain thehave underhis move toward evolutionism (13). sified into groups. Darwin’s mentor in geology, achievements, addition ancestry, to natural selection the Sharp hazardsirregular of migration, isolation, and local lying similarities.

Closely be related species di- onliam of evolution, thereancestor would major implications system share aincommon and he realized that Macleay’s or circular achievements, in addition to natural migration, isolation, and local lyingverged similarities. related ancestor, species whereas have di- on the hazards itself (14).

So it was, but I think Mayr over- selection Darwinofwas led toward his alternarecently Closely from a common Because many slaveholders argued Charles Lyell, had shown how his uniformitarian system by which species are clas- adaptation. for the whole of classification supposed that every genus con- this claim could be defended by extending the estimated the (14). rapidity whichbut otherI naturalmodel in Darwin part because more his interthe recently ancestry of more distantly related forms must tive itself Sowith it was, think Mayr overadaptation. was he ledwas toward alternaverged from a common ancestor, whereas that the black race was separately theory would allow the biogeographer to rementor in geology, five species that could be arranged inistsa were As a basified groups.

Darwin’s idea throughout animal kingdom. tained Fig. 1. Charles construct Darwin, Alfred Russel Wallace, and on Herbert Spencer. converted thethe theory.

Many of the adaptation thanbecause cosmic teleology, be into traced further down the family tree to ested theto rapidity with which other naturaltiveinmodel in part he was thanks more inter- estimated the ancestry of moreback distantly related forms must created from the white, Darwin the migrations of species an everfive genera, and so on through Charles Lyell, had shown how his uniformitar- tocircle; each family sis for his thinking, this thesis is sure to genernon-Darwinian theories of evolution proposed the influence of William Paley’s natural thefind the common point of origin. be traced further back down the family tree to ested in adaptation than cosmic teleology, thanks ists were converted to the theory. Many of the wanted to show that all races share changing earth.Russel Populations sometimes Fig. 1. Charles Darwin, Alfred Wallace,ofcould andand Herbert Spencer.

Darwin created the outlines thetheory complex process by term would allow the biogeographer Vestigesduring of the much controversy, but if late accepted it would letters to establishhis “evolution”; the Latinchallenged evolutio refers to the the taxonomic hierarchy.

Chambers’s position. ian theory ate“eclipse of Darwinism” 19th proposed selection replaced divine benevidea of point common descent now seems soto ology. theory of natural selection thisancestry, vision which developed (6–9).

Darwin ferences But between thehad ways in by which he and barriers, non-Darwinian theoriesinoftheevolution to theNatural influence of William Paley’s natural thefind theThe common of origin. a common and he realized become divided geographical sohethat century were introduced with the aim of subvertolence as an explanation of adaptation.

Unlike obvious that we might wonder what alternative and there were sigDarwin unrolling of a scroll.)

The explanatory framework of species on an everrole played by his move the theory 20 years earlier, which he developed his theory (6–9). reconstruct the migrations the Natural History of Creation depicted evoluemphasize the crucial of relations. a highly creative thinker who synthesized a of nature as an orderly patternthat Wallace formulated their ideas.

Insingle this essay, I wassplit during the “eclipse of Darwinism” in the late 19th ology. Natural selection replacednotdivine benevThe idea of common descent now seemsforso Macleay thisthis claim could be defended was once species of on natural selection challenged vision which heinto developed his theorywho (6–9).

Darwin theory ferences betweenwhat the ways in awhich and could ing the toward implications of the principle of common and Chambers, Darwin expect models earth. could have been proposed to account the theory of natural selection chalof parallel lines did advancing through based on nificant differences between the ways inhewhich was a highly creative thinker synthesized centered changing Populations could sometimes a model of branching evolution tion in terms Darwin’s world view was profoundly differnumber of key insights, some derived from his argue1.that Darwin was truly original in his thinkFig. Charles Darwin, Alfred Russel Wallace, and Herbert Spencer. olence as an explanation of adaptation. Unlike century were introduced with the aim of subvertweRussel might Wallace, wonder what alternative by extending the idea Fig. throughout 1. Charles obvious Darwin, that Alfred and Herbert Spencer. multiple to Iseparate naturethis as vision an orderly pattern of relations. was aenvironhighly creative thinker who synthesized a of Wallace formulated their branches ideas.

In adapting thisInessay, he and Wallace formulated their ideas. this esof key insights, some derived from his as an orderly pattern become divided by geographical barriers, so sequence of stages within each lenged of nature a predetermined geographical diversity. a number adaptation of pop- As a basis for work and others from currents circulat- ent because he argued that thethe ing, and I support this(10). claim by addressing the scientific models could have been proposed to account for Macleay and Chambers, Darwin did not expect ing the implications of the principle of common animal kingdom. ments Evolution become a divergent Darwin’s world view was profoundly differofwork key insights, some derived from his of relations. argue Darwin was truly original in would his thinkthat Darwin was truly original inand his number from currents circusplit family, from individual This model was so radical that many late say, I that argue scientific and others that what was once a single species could driven323 by force derived www.sciencemag.org ulationsoftonatural their local environment was the solethesisferences ing over in his cultural environment.

Few would now ent related issue of defining just why the theory was 224 9 JANUARY 2009 his VOL SCIENCE theory selection challenged this vision which he developed his theory (6–9).

Darwin ferences between the ways in which he his thinking, this is sure to process, with some branches splitting and he developed theory (6–9).

Darwin theory of natural selection challenged this vision between the ways in which he and which because heworld arguedview that the of popwork and others from currents circulating, and Iand support this claim by addressing the scientific thisideas. claim bythis addressing Darwin’s wasadaptation profoundly dif-it thinking, I support lating inthe hisclaim cultural environment.

Few would into multiple branches adapting to separate en- development.

19th-century evolutionists were unable to accept cause of transmutation.

Many people found accept that evolution by natural seso disturbing to his contemporaries. of nature as an orderly pattern of relations. was a highly creative thinker who synthesized a Wallace formulated their In essay, I generate controversy, butformulated if over again, whereas others was came toinga in dead Wallace their ideas. In this essay, I was a highly creative thinker who synthesized a of nature as an orderly pattern of relations. to their local environment wasmuch the sole his end cultural environment. now ulations related issueissue of defining just why the theory Fig. 2. Tree ofFew Life,would from Darwin’s notebooks (22). why now accept the claim evolution byfrom natural that the agent adaptation of emphasize the224 vironments (10). Evolution would become These rigidly structured models ofwww.sciencemag.org taxonomic it in full. Ernst Mayr argued that the theory of the Darwin related of defining justthe ferent he argued a 9 JANUARY 2009 VOL 323 SCIENCE hardDarwin’s tobecause see natural selection asaccepted the either lection was in insights, the air.that Darwin approached the cause wasthrough certainly not first sug- number world viewMany was profoundly of claim key some derived his argue that Darwin was truly original inthe histotheory thinkitofdifferwould extinction.

Darwin’s world view was profoundly differargue that Darwin was truly original in his think- number of key insights, some derived from his of transmutation. people found it accept the that evolution by natural seso disturbing to his contemporaries. in the air.

Darwin approached the was the process, with some branches splitting made good sense to any- common descent was one of Darwin’s greatest was so disturbing to his contemporaries. selection was populations to their local environment relations and evolution divergent divine benevolence or of a rationally structured subject in a way that was significantly different gest the idea of evolution as an alternative to ent because he argued that adaptation pop- toward scientific work from currents circulating,Darwin and I support this the hadlection crucial rolethe by his move model These rigidly structured of taxoTheclaim image of addressing the treetoofsuglife appeared inin and ing,a and I support this claim by addressing the scientific work and others from currents circulat- ent because he argued that the adaptation of pophard to see natural selection asplayed the agent ofofeither was the others air.

Darwin approached themodels was certainly notby the first was certainly not the first to sugsignificantly different subject in cultural a2)way that was sole cause of transmutation.

Many people found over and over again, whereas others came to a one embedded in a vision of nature as a predict- achievements, in addition to natural selection itcosmic teleology.

Selection adapted species to anon geographical fromin(Fig. any of the nomic other efforts being made tonow extheDarwin creation of defining species by God. J.the B. Lamarck’s ulations to their local environment was the sole ing his environment.

Few would related issue just why theory was of branching evolution based relations and evolution made good sense Darwin’s notebooks of the late 1830s related issue of defining just why the theory was ing in his cultural environment. Few would now ulations to their local environment was the sole gest the idea of evolution as an alternative to subject in a way that was significantly different divine benevolence or of a rationally structured as an alternative to accept made to exsee natural selection as agent of think Mayr gest thepublished idea ofhis evolution from any ofclaim the other being end through extinction. system divineofplan. self (14). Many So it was, but I found it hardoftoastransmutation. ever-changing and itpeople didthe so by killing plain the ofthat lifeefforts on earth.

He had theory, incontemporaries.

1809,proposed had been widely discause Many found thehistory evolution bymade natural se- ofcosmic so disturbing diversity. anyone embedded ina aunique nature a environment, and was Wallace transmutation. people it overestimataccept the claimable, thatorderly evolution by governed natural se-by a cause so disturbing to dead his contemporaries. teleology.

Selection adapted species to anit from any ofin the to other efforts being tovision exthe creation oftospecies

by God. J. B.independently Lamarck’s by or of a rationally The image of the tree of life had appeared a world view made it difficult to accept the creation of species by God. J. B. Lamarck’s plain the history of life on earth.

He had a unique either divine benevolence ed the rapidity with which other Such off useless variations in a ruthless “struggle for combination of scientific interests that alerted cussed, although generally rejected (2–4).

Robert hard to see natural selection as the agent of either lection was in the air.

Darwin approached the Darwin was certainly not the first to sugThis model wassoso late was certainly not the first to sug- lection was in the air. Darwin approached the hard to see natural selection as the agent of either naturalists were predictable, orderly system governed by a divine environment, paper published 1855. disBoth realized it of Darwin ever-changing and it did byradical killing that many plain thethat history life on earth.

He had a unique theory, publisheda in 1809, had beeninwidely theory, published in 1809, had been widely discombination ofignored scientific interests thatdifferent alerted of the late (Fig. earth might be essenstructured cosmic Selection Darwin’s as notebooks history of life on converted the theory.

Many of the non-Darexistence.”

This 19th-century didteleology. notofseem the kind structured ofadapted process him to topics by significantly other naturalists.

Heit difficult Chambers’s Vestiges of thewhy Natural History of able divine benevolence or a rationally subject in a way that was gest the idea of evolution as an alternative to evolutionists were unable to accept plan.

Such a world view made to acexplained naturalists were to arrange divine benevolence or of atorationally structured subject in 2) a waythat thatthe was significantly different gest the idea ofinevolution an alternative to 1830s cussed, although generally rejected (2–4). Robert combination of scientific interests that alerted off useless variations in a ruthless “struggle for generally (2–4).

Robbywidely other naturalists.

He species to an environment, and it thethetheory Wallace in thetially on winian of evolution during the cussed, although him to topics and wasbyproposed independently and unpredictable, theories thatmight would beever-changing instituted by a benevolent God, certainly drew on ideas discussed at exthe Creation of 1844 sparked arejected debate over thegroups, pos- using cosmic teleology. adapted species to an from any of theignored other efforts being made to the creation of species J. B. Lamarck’s itSelection innot full.

Ernst that of of species cept that history of life on earth be species into groups within cosmic teleology.

Selection adapted species toproposed an any of otherirregular efforts being made to ex- dependant creation God. J. B. Lamarck’s byfrom existence.”

This did seem the Mayr kind ofargued process todescent topics ignored by the other naturalists.

He Chambers’s Vestiges ofby theGod. Natural History of him especially because its essentially “selfish” nature time, but was forced by his scientific interests to sibility that newVestiges species were produced from the Natural History discussed at the off useless variations in a ruthadlate 19th century ert Chambers’s of certainly drew on ideas widely did so by killing a paper published in 1855.

Both realized that it the hazards of migration, isolation, and local “eclipse of Darwinism” in the ever-changing environment, and it did so by killing plain the history of life on earth.

He had a unique theory, published in 1809, had been widely discommonby descent was oneGod, of Darwin’s greatest irregular and unpredictable, a common explain the undertheory, published in 1809, had been widely dis- plain the history of life on earth. He had a unique ever-changing environment, and it did so by killing would be instituted a benevolent certainly drew onessentially ideas widely discussed at the thatdependant Creation of 1844from sparked a debateancestor over the to posmeant that avariations parasitic way of life was anot perfectly use those sources ofthe inspiration in migration, a that highly origthe This did seem explained why naturalists were able to arrange the of Creation of 1844 sparked a debate over time, but was forced by his scientific interests less “struggle for existence.” aptation.

Darwin was led toward his alternative were introduced with the aim of subverting off useless in a ruthless “struggle for combination of scientific interests alerted cussed, although generally rejected (2–4).

Robert achievements, in addition to natural selection on hazards of isolation, and local lying similarities.

Closely related species have dioff useless variations in a ruthless “struggle for combination of scientific interests that alerted cussed, although generally rejected (2–4).

Robert sibility that new species were produced from time, but was forced by his scientific interests to especially because its essentially “selfish” nature natural adaptive response in some circumstances. inaluse way. to those sources ofbyinspiration in a led highly groups, because he wasHe more existence.” interested inThis implications possibility that new species produced from the process that would instituted species principle of common descent model inbypart existence.” did not seem theitbe kind process him to topics ignored other He Chambers’s Vestiges ofrecently thewere Natural ofancestor, itself (14). was, but I by think Mayr over- Vestiges adaptation.

Darwin was towardmeant hiskind alternaverged from aHistory common whereas did not seemof thethe kind of process him descent to topics ignored other naturalists.

Chambers’s ofinto the groups Naturalwithin History of using School of Philosophy and Anthropological Studies, The Queen's thatofaThis parasitic way ofSo life was aofperfectly use those sources of inspiration innaturalists. a highly origMore seriously for the idea of cosmic teleTo some Darwin may have atbeen ones in a Road progressive sequence explaincertainly the underto thebe instituted American neo-Lamarckians Edward original way. a more benevolent especially its essena common adaptation than discussed cosmic teleology, pre-existing (15). Theby that would beGod, instituted by abecause benevolent God, certainly drew extent, on ideas widely discussed thewasnatural Creation 1844 sparked a debate over the posUniversity ofof Belfast, University Belfast, Belfast, Northern estimated the rapidity with which other naturaltive model in part because he interthe ancestry of more distantly related forms that would a benevolent God, drew on ideas widely at the thanks Creation of 1844from sparked a debateancestor over thetoposadaptive response in some circumstances. inal way.must School Philosophy andspecies Anthropological The Queen's ology,“selfish” Darwin’s supposition that“selfish” the production merely “ahead of hisby time,” anticipating develIreland,of BT7 1NN, UK. E-mail: especially because its essentially nature time, but was forced his scientific interests to sibility that were produced from if Studies, the general idea To some extent, Darwin may have been nature meant that a parasitic way have proposed that leading up tonew humans (5).p.bowler@qub.ac.uk But lying similarities.

Closely related species influence of William Paley’s natural theology.

Drinker Cope and Alpheus Hyatt tially ists were converted to the theory.

Many of the ested in adaptation than cosmic teleology, thanks be traced further back down the family tree to especially because its essentially “selfish” nature time, but was forced by his scientific interests to sibility that new species were produced from More seriously for the idea of cosmic teleTo some extent, Darwin may have been University of Belfast, University Road Belfast, Belfast, Northern meant that aa parasitic way of that life was a response perfectly use those sources inspiration inofa William highlydeveloriganticipating develadaptive Natural selection divine benevolence merely “ahead of his time,” ofnatural life Darwin’s was perfectly natural diverged recently from a common the evolution of each should be seen as a non-Darwinian theories of evolution proposed toof the influence Paley’sology, thethep.bowler@qub.ac.uk common point of origin. merely meant that a parasitic way of life was agroup perfectly use ancestor, those sources of inspiration in areplaced highly origsupposition the production “ahead of his time,” anticipating Ireland, BT7 1NN, UK.find E-mail: natural adaptive response some circumstances. inal way. that to- divine Macleay opments other naturalists in some circumstances. of parallel moved through the same whereas the ancestry of more distantly related as an explanation of adaptation. Unlike during thein“eclipse of Darwinism” in the late 19th ology.push Natural selection replaced benevThe idea of Studies, common descent now seems so would natural adaptive series response in some lines circumstances.

School of Philosophy and Anthropological The Queen's www.sciencemag.org SCIENCE VOL 323have9 been JANUARYMore 2009seriously for the idea of cosmic tele- 223School of Philosophy and Anthropological Studies, The Queen's inal way.

Toansome extent, Darwin may More seriously for the idea of cosmic teledown the his theory of developmental stages, an updated ward evolutionary the years forms must be traced further back and Chambers, Darwin did not expect hierarchy century were introduced with the aimUniversity of subvertolencevision as anduring explanation of he adaptation.

Unlike obviousRoad thatBelfast, we might alternative More seriously for the idea of cosmic teleTo some extent, Darwin may have been University of Belfast, University Belfast,wonder Northern what of Belfast, University Road Belfast, Belfast, Northern ology, Darwin’s ing supposition that the the of production “ahead of hisBytime,” anticipating develIreland, BT7 1NN, UK. E-mail: p.bowler@qub.ac.uk 223of www.sciencemag.org SCIENCE VOL 323 the 9Darwin JANUARY 2009 the implications the principle common Macleay and Chambers, not expect models could have been proposed tomerely account isolation. the late 1850s, idea did that production ology, Darwin’s supposition family to find the common point of origin. to predict an orderly pattern of relations. of the suggested in Chambers’s worked infor ology, Darwin’sversion supposition thatidea the production merely “ahead of his time,” anticipating develIreland, BT7 1NN, UK. E-mail:tree p.bowler@qub.ac.uk

Darwin’s Originality Darwin’s Originality Darwin’s Originality

REVIEW

Darwin’s Originality

18

224

www.sciencemag.org SCIENCE VOL323 323 SCIENCE 9 JANUARY 2009 9 JANUARY 2009 VOL www.sciencemag.org

223

CREDIT: SYNDICS OF CAMBRIDGE UNIVERSITY LIBRARY

CREDIT: SYNDICS OF CAMBRIDGE UNIVERSITY LIBRARY

CREDIT: SYNDICS OF CAMBRIDGE UNIVERSITY LIBRARY

T

CREDIT (LEFT TO RIGHT): STAPLETON COLLECTION/CORBIS; HULTON-DEUTSCH COLLECTION/CORBIS; MICHAEL NICHOLSON/CORBIS

CREDIT (LEFT STAPLETON TO RIGHT): STAPLETON COLLECTION/CORBIS; HULTON-DEUTSCH COLLECTION/CORBIS; MICHAEL NICHOLSON/CORBIS CREDIT (LEFT TO RIGHT): COLLECTION/CORBIS; HULTON-DEUTSCH COLLECTION/CORBIS; MICHAEL NICHOLSON/CORBIS CREDIT (LEFT TO RIGHT): STAPLETON COLLECTION/CORBIS; HULTON-DEUTSCH COLLECTION/CORBIS; MICHAEL NICHOLSON/CORBIS

T T T T

www.sciencemag.org

SCIENCE

VOL 323

9 JANUARY 2009

223

19

REVIEW

CREDIT: AMERICAN MUSEUM OF NATURAL HISTORY

CREDIT: AMERICAN MUSEUM OF NATURAL HISTORY

Vestiges. The similarities linking the species in directed variation among individual organisms. tion thinking, and although he may have exa genus were due not to a recent common ances- Although convinced that the degree of variabil- aggerated the extent to which Darwin himself try, but to parallel trends independently reaching ity was artificially enhanced under domestica- made the conceptual transition, the subsequent the same stage of development. Like Chambers, tion, Darwin, nevertheless, accepted that there development of the selection theory brought this they endorsed the recapitulation theory (ontog- must be some equivalent variability in every implication out more clearly.

In the debates that followed the publication eny recapitulates phylogeny, in the terminology wild population. The analogy with artificial seintroduced by Ernst Haeckel) and saw evolution lection then allowed him to depict natural selec- of On the Origin of Species, the analogy with role as the addition of preordained stages to ontog- tion as a parallel process in which a few variant artificial selection continued to play a keyREVIEW eny. Adaptation was not crucial once the basic individuals, in this case with characters useful to by forcing even Darwin’s critics to think about descent (15). The American neo-Lamarckians scale and that could be investigated directly. interbreeding individuals. Traditionally, species character of the group was established, and the species rather than the human breeder, sur- the problems of heredity and variation in a new Edward Drinker Cope and Alpheus Hyatt pro- There was a well-developed network of breeders were treated as idealized types with a fixed esof the ofgroup andbe reproduce. withalthough harmfultheir characthe linear, orthogenetic vive way (18). sence,Opponents any variationsuch fromas theFleeming norm beingJenkin, trivial by this Those time, and ideas about posedevolution that the evolution each group should byand thevariation strugglewere for existence, on large might eventually generate ters through are eliminated who saw selection working and impermanent.

The breeders knewvariations that they heredity distinctly pregenetseen as abizarre series ofnonadaptive parallel lines moved could produce huge changes in structure by still acical (likewill Darwin’s own), they a very clear the same hierarchy of developmental an the up- breeder to extinction—the the- stages, any had animal just as not permit or “sports of nature,” were, nevertheless, characters as a prelude variations over a number of appreciation of have how they produced changes in cumulating datedDarwin version of the idea suggested could make no intoChambers’s he working ory of “racial senility.” reproduce if it does not the character within normal the framework defined by this Vestiges. The similarities linking the species in a their artificially small populations. The insight generations. When Darwin linked this informasense of the theory proposed

by Cope and Hyatt, wants. It was the breeders who taught Darwin analogy. For supporters such as Francis Galton, genus were due not to a recent common ancestry, that they worked by selection may have been tion with his conviction that species could change is not directed preor- among to was clarify nature because he could notbutimagine evolutionary artificial selection variation indefinitely overhelped time, he driventhe toward a (this is toward the pointsome of contention to parallelantrends independently that reaching the important trends. But Like the Chambers, build onnotebooks), his exist- but of the popway dainedthey goal, allowing him toDarwin’s heredity selection, process driven by predetermined new form of and species concept inpaving which the experts studying theboth same stage of development. becomes paramount. natural rangegeof breeders certainly evolution taught himmust one be thing. for He theulation the in recapitulation theorying (ontogeny flourished the late 19th fact that such theoriesendorsed conviction that adaptive revolutionary impact The of Mendelian becomes of theheredity species’ character, that in process. a domesticated population there phylogeny, the terminology intro- realized the in theory The notion of part “hard” was incentury demonstratesrecapitulates just how radical an open-ended, branching netics.variability duced by Ernst Haeckel) and saw evolution as is always a fund of apparently purposeless and not the result of accidental deviations from a At the same time, the breeders’ attitude to- troduced in opposition to the “soft” form of of open-ended, divergent evolution was to the the addition of preordained stages to ontogeny. undirected variation among individual organisms. fixed norm. This is what Mayr called the transiDarwin toward the view pushedconvinced inheritance implied by the Lamarckian process. naturalists of the time. tion from typological thinking to population thinkthat the degree of variabilAdaptation was not crucial once theward basic variation char- Although of domestication, inter- The undirected natureheofmay variation was clarified that species is artificially just a population ing, and although have exaggerated the ity was enhanced under acter of the group was established, and the the linear, extent to the which Darwin himselfpopulations made the conDarwin, nevertheless, accepted that there mustboth be through might even- individuals.

Artificial Selection orthogenetic evolution of the group breeding study of large by Traditionally, species transition, the the subsequent of some equivalent variability every wild tually generate bizarre nonadaptivewere characters These non-Darwinian models were ultimately idealized types with in a fixed es- popustudies of the Galtonceptual and through breeding development treated as as a prelude to extinction—the theory of “racial lation. The analogy with artificial selection then the selection theory brought this implication out synthesis of the selection sence, any variation from the norm being trivial geneticists. Although it took some time for the marginalized by thesenility.”

Darwin could make no sense of the allowed him to depict natural selection as a par- more clearly. century. geneticists to accept studies theory and genetics theory in theproposed early 20th impermanent.

The breeders that they In the debates the that situation, followed thetheir publication in which aknew few variant individuals,

by Cope and Hyatt,and because he allel process ultimately Darwin’s Genetic mutations seemed toimagine be essentially plu- process coulddriven produceinhuge changes in structure of mutation of On the Origin ofendorsed Species, the analogyclaim with this case with characters usefulby to acthe species could not an evolutionary artificial selection to play a keyaffect role rather than the human survive re- the by providing predetermined But the fact that such normal justtrends. the source variations overbreeder, a number of and that could only way the continued environment ralistic and undirected, cumulating by forcing was evenby Darwin’s criticsModern to thinkevoluabout Those withlinked harmfulthis characters flourishedmechanism in the late 19th century dem- produce. that Darwin’s infor-are elimof “random” variationtheories generations.

When Darwin the population selection. onstrates just how radical the theory of open- inated by the struggle for existence, just as the the problems of heredity and variation in a new required as its raw ended, material.

This later devel- mation with his conviction that species could tionary developmental biology has reopened the divergent evolution was to the naturalists breeder will not permit any animal to reproduce if way (18). Opponents such as Fleeming Jenkin, over he was driven evolution can opment highlights the importance of another change indefinitely whether variation selection working and on large variations it does not have time, the character he wants.

It wasquestion the whoofsaw of the time. species in variation which is be form ofwho as open-ended Darwin and his still folinsight gained by Darwin in the late 1830s, his toward a newbreeders or “sports of nature,”aswere, nevertheless, taughtconcept Darwin that not quite Artificial Selection within framework defined by this toward some preordained goal, allowing of the animal the populationdirected The natural believed.

Butthethe non-Darwinian vision decision to investigate the work becomes paramount. lowersworking analogy.unfolding For supporters as Francis Galton, him to build on hispart existing conviction that adaptThese non-Darwinian models ultimately theirwererange to ansuch orderly, predictable breeders (Fig. 3) and his recognition that of variability becomes of the species’ of evolution marginalized by the synthesis of the selection ive evolution must be an open-ended, branching artificial selection helped to clarify the nature method of artificial theory selection offered a useful character, not the result of accidental deviations plan has been essentially marginalized by accepof both heredity and selection, paving the way and genetics in the early 20th century. process. how the equivalent natural norm.

This is what Mayr called the from a fixed tance of the insights onimpact which Darwin based way of understandingGenetic for thekey revolutionary of Mendelian At the same time, the breeders’ attitude mutations seemed to be essentially pluexactand role played by Dar- justtransition to populaof natural selection. process operated. Theralistic typological genetics.

The notion of “hard” heredity was toward variationthinking pushed Darwin toward his the theory undirected, providing the sourcefrom of “random” variation that Darwin’s mecha- view that the species is just a population of introduced in opposition to the “soft” form of of his win’s study of breeding in the formulation nism by required as its raw material. theory is much debated historians (16–17), This later development highdoubt of how important but there can be little lights the importance of another artificial andby natural the analogy between insight gained Darwinselecin the In thistocase, tion became in his later thinking. late 1830s, his decision investigate because the workeven of the animal Darwin was truly unique, Wallace breeders (Fig. 3) and his recogdid not take this step and dissociated himself nition that their method of artifiselection expressed from the link with artificial cial selection offered a useful way in Darwin’s later writings. of understanding how the equivDarwin turned to alent the breeders in search of a natural process operated.

The exactcould role played

by Darwin’s be changed— clue as to how a population study ofwhere breeding in the formulamodifications here at least was a situation tion of his theory is much debated were actually being produced a human by historianson(16–17), buttime there scale and that couldcanbebe investigated directly. little doubt of how impornetwork of breedThere was a well-developed tant the analogy between artificial natural selection became in his their ideas about ers by this time, andand although thinking.

In this pregeneticase, Darwin heredity and variationlater were distinctly was truly unique, because even they a very clear cal (like Darwin’s own), Wallace did had not take this step and they produced in appreciation of how dissociated himself changes from the link The expressed insight their artificially small populations. with artificial selection Darwin’s later writings. that they worked byinselection may have been Darwin turned to theamong breeders of contention important (this is the point in search of a clue as to how a notebooks), but the experts studying Darwin’s population could be changed— He realbreeders certainly taught him one thing. here at least was a situation where ized that in a domesticated population there beis modifications were actually ing produced on a human time Fig. 3. Pigeons (23). purposeless and unalways a fund of apparently

20

www.sciencemag.org

SCIENCE

VOL 323

9 JANUARY 2009

225

Traditionally, species scale and that for could investigated directly. descent (15). The neo-Lamarckians TheAmerican Struggle for Existence thatbetransition in the late 19thinterbreeding century. It individuals. the Men Who Discovered It (Doubleday, New were treated as idealized types with a fixed esThere was a well-developed network of breeders Edward DrinkerOne Cope and Alpheus Hyatt proYork, 1958). of the most disturbing aspects of Darwin’s did, however, highlight the harsher aspects of 2. P. Corsi, Agebeing of Lamarck: sence, any variation from theThe norm trivial Evolutionary by for this existime, and their of ideas about The posed that the evolution of each group should to the be struggle struggle. potential theory was its appeal the although consequences Theories in France, 1790–1830 and impermanent.

The breeders knew that they (Univ. of heredity were distinctly pregenetseen as a series tence of parallel moved through that equates with and the variation clearly as thelines natural process implications were drawn out even more California Press, Berkeley, 1988). could produce huge changes in structure by ical (like Darwin’s own), they had a very clear the same hierarchy of developmental stages, an upbreeder’s activity as a selecting agent. This very when Galton argued that it would be necessary 3. A. Desmond, The Politics acof Evolution: Morcumulating normal variations over a number of appreciation of how they produced changes in dated version of harsh the idea suggested in Chambers’s vision of nature certainly threatened the to apply artificial selection to the human race phology, Medicine and Reform in Radical generations.

When Darwin linked this informatheir artificially small populations.

The insight Vestiges. The similarities linking the species in a London (Univ. of Chicago Press, Chicago, traditional belief in a benevolent Creator. The in order to prevent “unfit” individuals from that species could change genus were due not to a recent common ancestry, that they worked by selection may have been tion with his conviction 1989). term “struggle for existence” occurs in Thomas reproducing and undermining the biological P. J.heBowler, Evolution: The aHistory of an was driven toward but to parallel trends independently reaching the important (this is the point of contention among indefinitely over4.time, Robert Malthus’s An Essay on the Principle health of the population. This was the eugenics (Univ.inofwhich California Press, Berkeley, ed. concept the popsame stage of development. Like Chambers, they experts studying Darwin’s notebooks), but the new form of speciesIdea of Population, although used in the context of program, and in its most extreme manifestation 3, 2003).The natural range of endorsed the recapitulation theory (ontogeny breeders certainly taught him one thing. He ulation becomes paramount. tribal groups competing for limited resources. at the hands of the Nazis, it led not just to the 5. J. A. Secord, Victorian Sensation: The Exof the species’ character, recapitulates phylogeny, in the terminology intro- realized that in a domesticated population there variability becomes part traordinary Publication, Reception, and Secret Darwin saw that population pressure would lead sterilization but also to the actual elimination of deviations duced by Ernst Haeckel) and saw evolution as is always a fund of apparently purposeless and not the result of accidental Authorship of Vestigesfrom of theaNatural History to competition between individuals and was per- those unfortunates deemed unfit by the state. Did Mayr called theChicago transi- Press, Chicago, the addition of preordained stages to ontogeny. undirected variation among individual organisms. fixed norm. This is what of Creation (Univ. of the first to realize that it might represent Darwin’s emphasis the natural tion elimination from typological 2000). thinking to population thinkAlthough convinced that the degree on of variabilAdaptation was haps not crucial once the basic charby which the population could change variants help to create a 6. heD.may a means of maladaptive Kohn, Ed.,exaggerated The Darwinian ing, and although have the Heritage: A acter of the group was established, and the linear, ity was artificially enhanced under domestication, TheevenprocessDarwin, workednevertheless, by climate throughoftime (19, 20). of opinion in must which atrocities Centennial extent to which Darwin himselfRetrospect made the(Princeton con- Univ. Press, accepted that there be such orthogenetic evolution the group might NJ, 1985). theequivalent popu- became possible? the least fitcharacters variants within subsequent development of some variability in every wild popu- ceptual transition, thePrinceton, tually generate eliminating bizarre nonadaptive 7. J. Browne, Charles Darwin: Voyaging (Jonahas to be admitted that,then by making death it- theory lation and allowing theof better adapted to survive the selection brought this implication lation.

The analogyItwith artificial selection as a prelude to extinction—the theory “racial than Cape, London, 1985).out breed.make Thisno wassense whatofthethe philosopher in nature, Darwinmore introduced self a creative andcould clearly. 8. J. Browne, Charles Darwin: The Power of allowedHerhim to depict natural force selection as a parsenility.” Darwin refer he to as the Spencer a new and profoundly disturbing insightIninto thethe debates that followed the Cape, publication allel“survival process in which a few variant individuals, theory proposedbert

by Cope andwould Hyatt,later because Place (Jonathan London, 2002). selection of an theevolutionary fittest.” Strictly speaking, an insight seems to have 9. ofM. J. S. Hodge, G. Radick, Eds., The Camof resonated On the Origin Species, the analogy with in this case withworld, characters useful that to the species could not imagine process drivennatural bridge Companion Darwin with thebreeder, thinkingsurvive of many not underrequires differential selection continued to play atokey role (Cambridge than the human andwho re- didartificial by predetermined trends.only But the fact that reproduction such ratheramong Univ. Press, Cambridge, 2003). variants, thought that theproduce. pressureThose of with his theory.

Darwin-even Darwin’s standharmful or accept the details by forcing critics to think about characters areofelimtheories flourished in the but lateDarwin 19th century dem10. M. J. S. Hodge, Stud. Hist. Biol. 6, 1 (1982). wastheory necessary to makeinated it effective.

Darwinism competition ism wasfornotexistence, “responsible” socialthe problems of 11. heredity and variation in a new Naturalists: by the struggle just asforthe onstrates just how radical the of open P. F. Rehbock, The Philosophical input frombreeder Malthus, henot permit seems that without or eugenics in anytosimple way. some wayall,(18).

OpponentsThemes such as Fleeming Jenkin, will any animal reproduce if After ended, divergentItevolution was to thethe naturalists in Early Nineteenth-Century British geneticists endorsed eugenics by analogy would not have come up with the theory. early saw selection working on large variationsPress, Madison, it does not have the character he wants.

It was the who of the time.

Biology (Univ. of Wisconsin 1983).

The idea of struggle was pervasive in thewho lit- taught even whileordismissing with Darwin animal that breeding “sports of nature,” were, nevertheless, still breeders variation is not J. Richards, The Meaning Artificial Selection exploitedtoward in some evolution. erature of the period, but could be directed naturalpreordained selection asgoal, the mechanism working within 12. the R. framework defined by thisof Evolution: The allowing of Morphological Construction For supporters such as Francis Galton,and Ideological to build conviction thattoadaptSpencer had on his many different In the 1850s, him Andexisting the Nazis wanted purify a analogy. fixed racial These non-Darwinian models ways. were ultimately Reconstruction of Darwin’s Theory (Univ. of artificial selection helped clarify the nature ivebe evolution bewhich an open-ended, branching marginalized byalready the synthesis of competition the selectioncould they certainly did not want to admit seen how turned must type, ChicagotoPress, Chicago, 1992). of both heredity and the way process. less had evolved gradually from an ape ancestry. theory and genetics the different, early 20thand century. into ainvery in some respects But 13.

A.selection, Desmond,paving J. R. Moore, Darwin’s Sacred for primarily the revolutionary impact Mendelian At the the breeders’ attitude Genetic mutations seemed to be essentially Cause: Race, of Slavery and the Quest for Humechanism of pluprogress (21).

Forsamebytime, disturbing, proposing that evolution worked man Lane,was London, 2009). genetics.DarThe notion ofOrigins “hard”(Allen heredity variation pushed toward the variants, ralistic and undirected, providing just the source individuals Spencer, the interaction betweentoward through the Darwin elimination of useless 14.

E. Mayr, One Long Argument: introduced in opposition to the “soft” form of Charles Darview that the species is just a population of of “random” variation that Darwin’s mechastimulated their efforts to adapt to the chang- win created an image that could all too easily be win and the Genesis of Evolutionary Thought nism required asing its raw material. social and physical environment. He then exploited by those who wanted the human race (Harvard Univ. Press, Cambridge, MA, 1991).

This later development high- concept of the “inheritance to conform to their own pre-existing ideals. In 15. P. J. Bowler, The Eclipse of Darwinism: Antiinvoked Lamarck’s lights the importance of another of acquired characteristics” to explain how these the same way, his popularization of the struggle Darwinian Evolution Theories in the Decades insight gained by Darwin in the accumulated over many gen- metaphor focused attention onto the individualAround 1900 (Johns Hopkins Univ. Press, self-improvements Baltimore, MD, 1983). late 1830s, his decision inves- to biological evolution and so- istic aspects of Spencer’s philosophy. erations,toleading 16. R. J. Richards, Stud. Hist. Philos. Sci. 28, 75 tigate the workcial ofprogress. the animal Modern science recognizes the importance Spencer’s self-improvement mod(1997). breeders (Fig. 3)el and his recogof progress became immensely popular in the of Darwin’s key insights when used as a way 17. M. Ruse, J. Hist. Ideas 36, 339 (1975). nition that their later method of artifi19th century, and because it too seemed to of explaining countless otherwise mysterious 18. J. Gayon, Darwinism’s Struggle for Survival: cial selection offered a useful wayas the motor of change, it was aspects of the natural world. But some of those Heredity and the Hypothesis of Natural Sestruggle rely on of understandingoften howconfused the equiv-with the Darwinian mechanism. insights came from sources with profoundly dislection (Cambridge Univ. Press, New York, 1998). alent natural process operated.

In fact, Spencer thought that all humans will turbing implications, and many historians now 19. P. J. Bowler, J. Hist. Ideas 37, 631 (1976).

The exact role played by Darwin’s eventually acquire the faculties needed to inter- recognize that the theory, in turn, played into 20. A. Desmond, J. R. Moore, Darwin (Michael study of breeding in the formulaact harmoniously with one another. But his occa- the way those implications were developed by Joseph, London, 1991). tion of his theory is much debated sional use of highly individualistic language al- later generations. This is not a simple matter of 21. M. Francis, Herbert Spencer and the Invention by historians (16–17), but there of Modern Life (Acumen, Stocksfield, UK, lowed him to be perceived as the apostle of free science being “misused” by social commentacan be little doubt of how impor2007).

Much of what later became known tors, because Darwin’s theorizing would almost 22. C. Darwin, Transmutation Notebook B, from tant the analogy enterprise. between artificial was, in fact, Spencerian certainly have been different had he not drawn as “social Natural Selection portfolio p. 36 (Cambridge and natural selection becameDarwinism” in his social Lamarckism expressed in the terminology inspiration from social, as well as scientific, inUniv. Library, Cambridge, 1838); P. H. Barlater thinking. In this case, Darwin rett, P. J. Gautrey, S. Herbert, D. Kohn, S. with of struggle popularized by Darwin. fluences.

We may well feel uncomfortable was truly unique, because even Smith, transcribers and Eds., Charles DarThis point is important in the context of those aspects of his theory today, especially in Wallace did not take this step and win’s Notebooks p. 180 (British Museum of by modern opponents of of their subsequent applications to human the charge raised light dissociated himself from the link Natural History, Cornell Univ. Press, Ithaca, Darwinism that the theory is responsible for the affairs. But if we accept science’s power to upwith artificial selection expressed NY, 1987); available at the Darwin Digital Library, http://darwinlibrary.amnh.org/. in Darwin’s laterappearance writings. of a whole range of unpleasant social set the traditional foundations of how we think on struggle. Darwin exploited the about the world, we should also accept its po- 23. G. Neumeister, Das Ganze der Taubenzucht policies Darwin turned to the based breeders (B. F. Voigt, Weimar, 1876). idea as of to thehow struggle in search of a clue a for existence in a way that tential to interact with moral values. nearly was unique until paralleled by Wallace population could be changed— years later. here at least was 20 a situation whereTheir theory certainly fed into the movements modifications were actually be-that led toward various kinds of References and Notes 1. L. Eiseley, Darwin’s Century: Evolution and not the only Darwinism, it was Fig. 3. Pigeons (23).vehicle ing produced onsocial a human time but www.sciencemag.org

SCIENCE

VOL 323

9 JANUARY 2009

225

21

Speciation

disentangle key aspects of clade histories. Clades are monophyletic, including all descendants of an ancestor, whereas taxa may be monophyletic or vestigations, provide an outline of how of these paraphyletic, excluding some descendants the and other Comparative studies correspond to the predictions ancestor. macroecological studies disentangle key aspects of clade histories.

Clades of the RedtoQueen, Court Jester,that andsister multilevel add rigor analyses showing clades REVIEW are monophyletic, including all descendants of an (Table 1), and outline some phymixed models may vary whereas in rate of taxa evolution, timing of increases ancestor, may be monophyletic or of spelogenetic studies ofand themorphospace macroevolution in species richness occupation, paraphyletic, excluding some descendants of the cies diversity. of evolutionary novelties across and distributions ancestor.

Comparative macroecological studies lineages Here, I that will sister explore the add rigorand to subclades. analyses showing clades The Global global, Patterntaxic of Diversification largest-scale investigations, provide may vary in rate of evolution, timing of increases Michael J. Benton Through an outline Time of how these and other studies correin species richness and morphospace occupation, A key thetheorigin of modern spond toquestion the predictions of Red Queen, Court and distributions ofabout evolutionary novelties across Evolution may be dominated by biotic factors, as in the Red Queen model, or abiotic factors, today’s 10 million species biodiversity is how Jester, and multilevel mixed models (Table 1), lineages and subclades. Here, I will exploreand the as in the Court Jester model, or a mixture of both. The two models appear to operate microbial arose from a single ultimate speciesofofthe outline some phylogenetic studies macrolargest-scale global, taxic investigations, provide predominantly over different geographic and temporal scales: Competition, predation, and other Michael J. Benton life 3500 of million ago 1).correTwo evolution diversity. an outline ofspecies howyears these and (Ma) other (Fig. studies biotic factors shape ecosystems locally and over short time spans, but extrinsic factors such as models for global diversification are termed the spond to the predictions of the Red Queen, Court disentangle key aspects of clade histories.

Clades The climate and oceanographic events as shape larger-scale patterns regionally and Global Pattern ofmodel Diversification (5–7)(Table and the exEvolution may be dominatedand by tectonic biotic factors, in the RedareQueen model,including or abiotic factors, of an saturation/equilibrium monophyletic, all descendants Jester, and multilevel mixed models 1), and globally, through millions years. studies that speciesor pansion Through model Time (8–11). The equilibrium model as in the and Court Jester thousands model, or and a mixture of of both.

ThePaleobiological two ancestor, models whereas appear tosuggest operate taxa may be monophyletic outline some phylogenetic studies of the macrodiversity is driven largely by abiotic factors such as climate, landscape, or food supply, and paraphyletic, excluding some descendants of the A key question about the origin of modern biopredominantly over different geographic and temporal scales: Competition, predation, and other at has prevailed, among marine paleobiologists evolution of species diversity. ancestor.

Comparative macroecological studies comparative approaches intospans, clade but dynamics. diversity isa how today’s 10 million species arose biotic factorsphylogenetic shape ecosystems locallyoffer and new over insights short time extrinsic factors such as least, for long time, and represents a classic add rigor to analyses showing that sister clades from a singlePattern ultimate ofitmicrobial life climate and oceanographic and tectonic events shape larger-scale regionally andof increases Red The Global of species Diversification viewpoint because implies priQueen may varypatterns in rate of evolution, timing 3500 million years ago (density (Ma) (Fig. 1). Two modportant not export organism-level processes to marily here are two ways of viewing evolution, globally, andare through thousands and millions Paleobiological studies suggest that species Through Time richness morphospace occupation, here two ways of viewing evolu-of years. scales, andinto itspecies is likely thatand evolution operates in dependence) on biotic controls distributions evolutionary els fordiversity. global diversification termed biothe global andsupply, it isnovelties likely that global through the largely spectacles of either Red diversity is driven abiotic factors such asregional climate,orand landscape, oroffood and across oftheeither way (3).scales, tion, through the by spectacles a pluralistic A key question about the originare of modern lineages and Here,way I will(3). explore the saturation/equilibrium model (5–7) and the exevolution operates insubclades. a pluralistic Queen or Queen the Court The Red Queen comparative phylogenetic approaches offer new insights intoare clade dynamics.

There two broad methodologies for provide stud- diversity There isare two versions of the species equilibrium the Red or Jester. the Court Jester.

The how today’s 10 million arose largest-scale global, taxic investigations, pansion (8–11).

The equilibrium model has There are two broad methodologies for studies model (1) stems from Darwin, who viewed evoenton taxic and model, time when the global diversity differing in the Red Queen model (1) stems from Darwin, who ies of species from a model single ultimate species of microbial life an outline of howthrough these and time, other studies correprevailed, among marine paleobiologists at least, of species diversity through time, taxic and phylution as primarily a balance of biotic pressures, spond to the predictions of the Red Queen, Court of bi- phylogenetic involves viewedhere evolution primarily a balance The taxic approach ecosystem saturated.

3500 million yearsbecame ago (Ma) (Fig. 1).Sepkoski’s Two modportant not to (4). export organism-level processes to marine are twoasways of viewing evolution, y be dominated by biotic factors, as in the Red Queen model, or abiotic factors, Jester,The and multilevel mixed models 1), and for a long time, and represents a classic Red Queen logenetic (4). taxic approach treatmost notably competition, and characterels for global are termedthree the regional species, or global scales, itinvolves is (Table likely that coupled the ofit was either Red oticorpressures, notably competition, it treating genera, orandfamilies as indelogisticdiversification model (5) identified urt Jester model, a through mixture ofmost both.spectacles The two models appear the toand operate outline some or phylogenetic studies of the macro- viewpoint because it implies primarily biotic coning species, genera, families as independent ized by the Red Queen’s statement to Alice in saturation/equilibrium model (5–7) and the exevolution operates in a pluralistic way (3).

Queenand or temporal the Jester.

The Red Queen and byCourt thescales: Red Queen’s statement and counting their occurrences in the Cambrian, most of the wasgeographic characterized pendent equilibria, y over different Competition, predation, other entities evolution of species diversity. trols (density dependence) global diversity. entities and counting their occurrences Through the Looking-Glass thatwho “it allevothe shape ecosystems locally and overfrom short time spans, buttakes extrinsic factors asThere pansion model (8–11).

Theaon equilibrium model has areand twoother broadfactors. methodologies foragainst studies Paleozoic, model (1)in stems Darwin, viewed that “it such The phylogenetic beginning in to Alice Through the Looking-Glass against time and perhaps third, oceanographicrunning and tectonic events shape larger-scale patterns regionally and There are twocontinuing versions of the present equilibrium time and other factors.

The phylogenetic apyou can do, to keep in the same place.”

The Global Pattern oftime, Diversification prevailed, among marine paleobiologists at (Fig. least, of speciesuses diversity through taxic and phylution all as primarily a balance of biotic pressures, trees to takes the running you can do, to keep in approach cladograms or molecular the Pliocene and to the through thousands and millions of years.

Paleobiological studies suggest that species Through Time differing the timea classic when the uses cladograms or molecular trees to model, The Court Jester model (2) isit that evolution, proach for a long time, andinrepresents Red global Queen logenetic (4).

The taxicofapproach involves treatmost notably competition, was place.”

Courtand Jester model (2)supply, is and disentangle aspects clade histories.

Clades same riven largely the by abiotic factors suchThe as climate, landscape, orcharacterfood A key key question about the origin of modern bio- 2A). These three equilibrium levels correspond speciation, and extinction rarely happen except in viewpoint because it implies primarily biotic coning species, genera, or families as independent ized by the Red Queen’s statement to Alice in phylogenetic that approaches offer new insights into dynamics.rarely andclade extinction are monophyletic, allmillion descendants of to three sets of phyla, the Cambrian, Paleozoic, evolution, speciation, diversity is including how today’s 10 species arose response to changes“it in the physical trols Modern, (density dependence) on global diversity. entities counting their occurrences against Throughexcept theunpredictable Looking-Glass takes all the an from a singleof ultimate species of microbial life and whereas taxa may be monophyletic happen in responsethat to unpredictable ancestor, that interacted and successively Fig. 1. and Operation Red Queen environment, recalling the capricious behavior 3500 million years ago (Fig. 1).A Two modre two ways of viewing evolution, portant not to export organism-level to There each are two versions of the timeparaphyletic, and other factors.

The(Ma) phylogenetic ap- replaced running you canphysical do, to keep in the same place.” processes changes in the environment, recalling some descendants other through the equilibrium Cambrianor excluding (biotic causation) and Court Jester els for global diversification are termed the h the spectacles of Court either theJester Red regional or global scales, and it is likely that uses of licensed foolmodel of Medieval times.

Neither model, differing in the time when the global proach cladograms or molecular trees Genus to Ordovician Thethe (2) islicensed that evolution, (abiotic causation) models at differbehavior of the fool of intervals, the capricious of the ancestor.

Comparative macroecological and Permo-Triassic saturation/equilibrium model (5–7) and the life ex-span or the Court Jester.

Thewas Red Queen evolution operates in a and pluralistic way (3). model proposed asrarely exclusive, both speciation, and extinction happen except in ent geographic and temporal scales (3-6 My) was proposed as for studies rigormodel to analyses showing that sister Medieval times. model pansion (8–11). The equilibrium model has reaching higher equilibrium levels after each ms from Darwin, who viewed evo-Neither There are two broad methodologies studies add Darwin and Van Valenspecies (1) allowed extrinsic Court response unpredictable changes infor theValen physical prevailed, paleobiologists at least, marily a balance of biotic to pressures, diversity through time, and phy(A).

Themay Red Queen maymarine at timing vary inamong rate ofprevail evolution, of long-term replacementJester event. The second model exclusive, and both ofDarwin andprimarily Van (1)taxic clades Fig. 1.

Operation of Red Queen influences on evolution in their biotic, a long time, and represents a classic Red Queen competition, environment, and it was characterlogenetic (4).

The taxic approach involves treat- for recalling the capricious behavior organismic and species level on short Species A allowed for extrinsic influences on evolution in and morphospace equilibrium level from increases in species richness (6, 7) identifies a single (biotic causation) and Court Jester Queen views. because it implies primarily biotic conRed Queen’s Red statement to Alice in species, genera, or families as independent life span of the licensed fooling ofRed Medieval times.

Neither time scales,viewpoint whereas the Court Jester Genus biotic, Queen views. primarily occupation, and distributions of evolutionary the early Paleozoic, perhaps 400 Ma, to the trols (density dependence) on global diversity.

Looking-Glasstheir that “it takes all the entities and counting their occurrences against (abiotic causation) models at differ(1-2 My) Species in as a Red Queen world de- holds his own on larger scales. The life span model wasdiversity proposed and both There aretemporal two versions of the equilibrium can do, to keep inSpecies the same place.” time other factors.

The phylogenetic ap- across diversity in and a exclusive, Red Queen world novelties lineages andscales subclades.

Here, present (Fig. 2B). In both models, the equilibria ent geographic and (3-6 My) pends primarily on intrinsic factors, such as body Red stippled green shape shows an time areawhen the global Darwin Van Valen (1)uses allowed for extrinsic model, the ester model (2) is that and evolution, proach cladograms or molecular trees toexplore Court Apparent Milankovitch intrinsic factors, as I(A). thediffering largest-scale global, to biodiversity saturation in which depends primarily on willThe correspondQueen?

Red Queen mayinprevail at taxic insize, breadth physiological tolerance, orsuch adaptRed Queen effects might be d extinction rarely happen except in Jester Red Queen influences onofevolution in their primarily biotic, where Scale (100 ky) organismic and species level on short Species tolerance, body in size, breadth of aphysiological from scale effect ability to hard times.

In Court Jester world, spenpredictable changes the physical erroneously, but these are Red Queen views. Fig. 1. Operation of Red Queen identified life span time whereas the Court Jester Fig. 1. Operation of Red Queen (biotic causation) orcapricious adaptability to hardontimes.

In a Court Jester likely recalling thecies behavior A scales, diversity depends fluctuations in climate, the result of spatial averaging of Locality Average Biotic Entire (1-2 My) and world Court Jester Species Neither diversity(biotic in a causation) Red Queen de- holds his own on larger scales. The ed fool of Medieval and Court Jester (abiotic causation) at difGenus world,times. species depends onmodels fluctuations species region/ models continent landscape, and diversity food supply.

In reality, of as course, responses to climate change (abiotic causation) at body differ- regional primarily on intrinsic factors, such life span proposed as pends exclusive, and both Red and stippled green shape shows an area range climate ferent geographic temporal scales (A).

The Red landscape, food supply.

Inand realin climate, entand geographic and temporal scales Apparent (3-6 My)complex physical perturMilankovitch both aspects might prevail in different ways at and other zoneand species level Van Valen (1) size, allowed for extrinsic Queen? breadth of physiological tolerance, orprevail adapt-at where Red Queen effects might Court be Queen Red Queen (A). The Red Queen Scale (100 ky) may prevail at organismic might prevail in difity, course, both that may be in opposite whataspects could perhaps bemay called the bations Jesterdievolution in different theirof primarily biotic, fromthe scale effect ability totimes, hard times.

In a Court organismic andJester speciesworld, level onspeshort Specieserroneously, but these are on short time scales, whereas Court Jester holds identified Geographical scale iews. could ferent ways and at different times, the what and so cancel each other, life span multilevel mixed model.

Traditionally, biologists time scales, whereas Court Jester rections, cies diversity depends on fluctuations in climate, his own on larger scales.

The stippled green shape likely (1-2 theMy) result of spatial averaging of Locality Average Biotic Entire iversity in a Red Queen world deholds his own onmixed larger scales.

The suggesting no controlling effect of the perhaps be called multilevel model. have tended to think in a Red Queen, Darwinian, species region/ continent shows an area where Red Queen effects might be foodthe supply.

In reality, of course, regional responses to climate change ly on intrinsiclandscape, factors, such and as body Red stippled green shape shows aninarea environment on evolution. range climate Apparent B biologists have tended to think in Traditionally, Milankovitch intrinsic, biotic factors way, and geologists a physical identified erroneously, but these are likely the reof physiological tolerance, or adaptboth aspects might prevail in different ways and at and other complexQueen? physical perturwhere Red Queen effects might be Physical-environmental Red Queen zone Scale (100 ky) disruptions facadifferent Red Queen, Darwinian, intrinsic, Court Jester, extrinsic, physical factors way. from scale d times. In a Court Jester world, speerroneously, but these are bations that may be in opposite di- effect times, whatidentified could perhaps bebiotic called the Genus sult of spatial averaging of regional responses to elicit biotic responses along the Geographical depends on fluctuations inand climate, likely theinresult of spatial averaging of may tors way, geologists a Court Jester, exclimate change and other complexscale physical perturlife span Much of the divergence between the Red Locality Average Biotic Entire rections, and so cancel each other, multilevel mixed model.

Traditionally, biologists species and region/ continent(3-6 My) red line separating Red Queen d food supply.Queen In reality, of course, regional responses to climate change bations that may be in opposite directions, and so and Court Jester world views may depend suggesting no controlling effect physical factors way. trinsic, range ofclimate the have tended to think in a Red Queen, Darwinian, might prevail in different ways and at and other complex physical pertur- Court Jester outcomes (B). The usage zone cancel each other, suggesting Court no controlling effect scale (2) (Fig. 1): Biotic interactions Much of the divergence between the drive Red physical environment on evolution. bations thatand may be in opposite s, what couldon perhaps be biotic called the intrinsic, factors way, geologists in dia here Speciesof the physical environment is the microevolutionary Red Queen, Geographical scale B Jester on evolution. Physicalmuch of the local-scale success or failure of inrections, and so cancel each other, Physical-environmental disruptions world views may Queen and Court Jester xed model. Traditionally, biologists Court Jester, extrinsic, physical factors way. life span opposed to the macroevolutionary Genus environmental disruptions may elicit biotic resuggesting no controlling effect of the as o think in a Red Queen,on Darwinian, dividuals, populations, and1): species (Red Queen), may elicit biotic responses along the (1-2 My) depend scale (2) (Fig. Biotic interactions life span Much of the divergence between Red Red Queen that posits constant exphysical environment onthe evolution. sponses along the red line separating Red Queen ic factors way, and geologists in a B line separating Red Queen and (3-6 My) red but perhaps these processes areviews overwhelmed by drive much of Jester the local-scale success or tinction Queen and Court world may depend Physical-environmental disruptions risk, a view that has been largely and Court Jester extrinsic, physical factors way.

Red outcomes (B). The usage here is Genus Milankovitch Court Jester outcomes (B).

The usage substantial tectonic and climatic processes atdrive time maypopulations, elicit biotic responses along the rejected and species failure of (2) individuals, Court on between scale (Fig. 1): Biotic interactions life span the divergence the Red (31). Illustration based on (2). the ky) microevolutionary Red Queen, as opposed to Queen Scale (100 5 redperhaps line separating Redprocesses and Species here is(3-6 theMy) microevolutionary Red Queen, scales above 10 but years (Court Jester).

ItQueen isofimJester ourt Jester world viewsof may (Red Queen), these much thedepend local-scale success or failure inthe macroevolutionary Red Queen that posits conCourt Jester outcomes (B).

The usage life span as opposed to the macroevolutionary Court (Fig. 1): Biotic interactions drive by substantial tectonic and are overwhelmed stant extinction risk, a view that has More been largely Less dividuals, populations, and species (Red Queen), Species (1-2 My) Jester local-scale success or failure of in- here is the microevolutionary Red Queen, Red Queen 5 life span that posits constant exDepartment of these Earth Sciences, University of Bristol, Bristol as opposed to the macroevolutionary rejected (31).

Illustration based on (2). at time scales above 10 climatic processes but perhaps processes are overwhelmed by pulations, and species (Red Queen), Environmental scale (1-2risk, My) a view that has been largely tinction Red Queen that posits constant exBS8 1RJ, UK. tectonic E-mail: mike.benton@bristol.ac.uk Red Milankovitch not to export (Court Jester).

It isclimatic important substantial processes at time rejected (31). Illustration based on (2). hese processesyears are overwhelmed by and tinction risk, a view that has been largely Queen Scale (100 ky) Red 5 Milankovitch ctonic and climatic processes organism-level to regional scales aboveat time 10processes years (Court Jester).or It global isonimrejected (31).

Illustration based (2).

Queen Scale (100 ky) 5

SPECIALSECTION

REVIEW

Red Queen and the Court Jester: ies Diversity and the Role of Biotic Abiotic Factors Through Time

Temporal scale

Temporal scale Temporal scale

Temporal scale

Temporal scale Temporal scale

T T T

10 years (Court Jester). It is im-

728

6 FEBRUARY 2009

Department of Earth Earth Sciences, University of Bristol, Bristol Sciences, University of Bristol, BS8 1RJ, UK. E-mail: mike.benton@bristol.ac.uk mail: mike.benton@bristol.ac.uk

22

728

6 FEBRUARY 2009

VOL 323

Bristol

SCIENCE

VOL 323 Less

SCIENCE

Less www.sciencemag.org More

Environmental scale

www.sciencemag.org

6 FEBRUARY 2009

VOL 323

SCIENCE

www.sciencemag.org

More

Environmental scale

Number of genera (corrected; blue)

400 200 0

200

Number of genera 400 (corrected;600 blue)

600

3000 2000 1000

Number genera 2000 of 3000 (empirical; red)

1000

0

0

tion

new taxa could become established only by on diversity, and other correction SPECIALSECTION plants, insects, or vertebrates, because regimes marine ecosystem became saturated. these be groups seem to have Sepkoski’s coupled logistic model (5) as toradiated produce data in driving others to extinction.

Key evidence may so complex explosively, without diversity plateaus,signals identified threeand equilibria, in the Camorigination extinction rates which geologic and are is that both plants, insects, orbiologic vertebrates, because marine ecosystem became saturated. particularly the past 100 million brian,been most density-dependent of the Paleozoic, and pernot obviously separated. appear to have these ingroups seem to have radiated Sepkoski’s coupled logistic model (5) (5– years (My) (10). haps a third, beginning in the Pliocene Life on land today be as much as 7), limiting risesequilibria, in diversity andCampromoting explosively, without diversity plateaus, identified three in Resolution between themay equilibriand continuing to the the present (Fig. particularly in as thelife past 100sea, million brian, most 2A). of the Paleozoic, andevents. per- levels after extinction inand the so it 25 asand diverse rapid recovery um times models, between these These three equilibrium years (My) haps a third,correspond beginning theglobal Pliocene expansionist models, might seem tointhree sets of phyla, the Alternative models for diversimay be wrong to(10). generalize from marine straightforward, but thebetween solution Paleozoic, and (Fig. Modern, that Resolution the and continuing to the present allowing global paleontological studies to all delife.equilibriPerhaps fication areCambrian, expansionist, pendsand on adequate assessment of the these interacted and successively um models, and between 2A).

These threetoequilibrium levelsreplaced similar patterns of and exspecies diversity rise, with damping, but land sea show quality of the fossil record. The longeach other through the Cambrianexpansionist models, might seem correspond to three sets limit of phyla, the Denwithout a predictable (8–11).

(8, ponential increase in species numbers term saturation model for global diverOrdovician and Permo-Triassic interCm O S D C P Tr J K Pg Ng butdiffer the solution deCambrian, Paleozoic, and Modern, thatextincof origination and sity dependence 9, 11),straightforward, or(Fig. perhaps in their key sification 2, bluethey curve) arises vals, reaching higher equilibrium levels 500 400 300 200 100 0 pends on adequate the interacted replaced from extensive attempts to correct data as aof after successively each replacement event. with the seaassessment acting giant tion rates and does notlong-term preclude expansionist rules (13, 14), Time (Ma) of error the fossil Thediverlongeach other through the Cambriansets forquality sampling (6,where 7),record. whereThe second model 7) identifiesby a models because they may be(6,dampened Gaussian petri dish, species as theterm multiple-equilibria and expansingle equilibrium level from the food early saturationand model for global diverOrdovician and Permo-Triassic interof limiting factors such as shortage density-dependent, 2. Patterns animal genus the past sity is equilibrial O of S marine D C P Trdiversification J K through Pg Ng originally used raw data, Paleozoic, perhaps 400levels Ma, to the Fig. Fig. 2. Cm Patterns of marine animal genus diversification through sion models sification (Fig. 2, blue curve) arises vals, reaching higher equilibrium 530 My, the Phanerozoic.

The two lines compare current estimates from by or space, orpresent active(Fig. predation, as well asthe and thecorrection land witnessing 400Phanerozoic.

300 The200 100 0 without (5, 8–11),continuing although (damp2B). In both models, the empirical past500 530(uncorrected) My, the two(red linesline) compare current from extensive attempts to correct data after each long-term replacement event. the Sepkoski database and samplingclimate andequilibria other physical factors.

Further, ened) exponential rise in diversity more recent analyses return a some- as ever correspond to biodiversity estimates from the empirical (uncorrected) database Time (Ma) standardized (corrected) analysis of the Paleobiology Sepkoski Database (blue line). sets for sampling error (6, 7), whereThe second model (6, 7) identifies a what dampened butecospace congruent signal saturation in whichlogistic new taxamodel could The the coupled sectors of are conquered it seems that (redempirical line) and sampling-standardized (corrected) analysisreached of the new curve (red line) suggests that global marine diversity as the multiple-equilibria expansingle equilibrium from the when corrections are imposed.

Cor-and become established onlyearly by driving aPaleobiology for global diversificamay be partly anlevel artifact of taxonomic (9, 14).

Any model Database (blue line).

The empirical curve (red line) possible plateau throughanimal the Paleozoic (450 to 250 Ma) and has risen, Fig. 2.

Patterns of marine genus diversification through the past sion models used Paleozoic, perhaps 400 itMa, toworked the out rection for encompass samplingoriginally is the clearly es- raw data, others extinction.

Key evidence is apparently exponentially, ever since.

The sampling-standardized curve suggests that global marine diversity reached a possible plateau scale (Fig. 2, red to curve); was tion must independent evi530 My, the Phanerozoic. The two lines compare current estimates from sentialwithout (6, 7), and future investigation that both and the extinction correction (5, 8–11),of although present (Fig. 2B).

In origination bothlevels models, (blue line)the suggests that (450 globaltomarine diversity reached near-modern through Paleozoic 250database Ma) and(red has risen, apparently familial but does not for increasing complexity organat ordinal and dence the empirical (uncorrected) Sepkoski line) and samplingappropriate indepenrates appear to to biodiversity have been density- levels some 400 Ma and there has been only modest increase since then. must determine more recent analyses return a someequilibria correspond exponentially, ever analysis since. The sampling-standardized curve(blue (blueline). specific work convincingly at generic isms, increases in the occupation standardized (corrected) ofD, the Paleobiology Database dent proxies for preservation and hu- of novel dependent (5–7), limitingorrises di- Cm, Cambrian; C, Carboniferous; Devonian; J, Jurassic; K, Cretaceous; what dampened but congruent signal saturation in which new taxa could in The line) suggests that global marine diversity reached near-modern curveO,(red line) suggests that global marineS,diversity man error; some current proxies (suchwithin parand promoting rapid recovery levels (10, versity 11), there are problems with empirical explosive evolution Ng, Neogene; Ordovician; P, Permian; Pg, Paleogene; Silurian;reached Tr, ecospace, when corrections are imposed.

Corbecome established only by driving levels some 400 Ma and there has been only modest increase as number of fossiliferous localities) after extinction events. a possible plateau through Triassic.

Based on (6). the Paleozoic (450 to 250 Ma) and has risen, assumptions (11, 12), and and addition of novel clades key numerical ticular clades, since then.

Cm, Cambrian; C, Carboniferous; D, Devonian; J, Ju-curve rection dependent for sampling is clearly esothers to extinction.

Keymodels evidence is apparently are themselves on diversity, Alternative for global diexponentially, ever since.

The sampling-standardized the background assumption of a global rassic; K, Cretaceous; Ng, Neogene; O, Ordovician; P, Permian; Pg, without the loss of precursors (9, 11, 15), sential (6, may 7), and investigation that both origination extinction allowing and other correction regimes be sofuture complex partly that an artifact taxonomic scale (Fig. 2, red near-modern versification and are expansionist, global (blue line) suggests globalof marine diversity reached carrying capacitydiversity is doubtful (8, 11). all data of which have happened many times in Paleogene; S, Silurian; Tr,worked Triassic.out Based on (6). as to produce in which geologic and biologic it was atonly ordinal and familial to rise, with10, damping, but with-400curve); must determine appropriate indepenrates appearspecies to have been densitylevels some Ma and there has been modest increase since then. hard to export loMy. Further, it has past 500 separated. not obviously out proved a predictable limit depen- levels but does not work convincingly at generic signals are the dent proxies for preservation and hudependent (5–7), limiting rises in(8–11). di-theDensity Cm, Cambrian; C, Carboniferous; D, Devonian; J, Jurassic; K, Cretaceous; or specific (10,extensive 11), there are problems to withcor- Life on land today may be as much as 25 of origination andspeciose extinctionterrestrirates does notcurve) ariseslevels from attempts gistic model to the much more error; some current proxies (such versity and dence promoting rapid recovery Ng, Neogene; O,keyOrdovician; P, Permian; Pg, Paleogene; Tr, man timesLarge-Scale as diverse as life in the sea, soSpecies it may be 12),(6, and7), thewhereas back-S, Silurian; preclude one expansionist models because they mayrect Diversity datanumerical sets forassumptions sampling(11, error al realm, whether considers plants, insects, as Controls number ofonfossiliferous localities) after extinction events.

Triassic.

Based on (6). assumption of a global carrying capacity is wrong to generalize from marine paleontologground be dampened by limiting factors such as shortage because these groups seem to and expansion models Taxic paleobiological studies have provided the multiple-equilibria or vertebrates, arelife. themselves on diversity,a Alternative models fororglobal diPerhaps dependent land and sea of food or space, active predation, as well as by doubtful (8, 10, 11). Further, it has proved hard to ical studies to all without diversity pla(5, evidence about controls, mainly have radiated explosively, originally used raw data, without correction great deal of otherpatterns correction regimes increase may beinso complex an artifact of taxonomic scale (Fig.more 2, redshowand versificationclimate are expansionist, allowing similar of exponential export the logistic model to the much and other physical factors.global Further, itpartly years Bioticthey factors, such teaus, particularly in thewith pastdamping, 100 million 8–11), although more recent analyses return abiotic, on species diversity.

(8,data 9, in 11), or perhaps speciose terrestrial realm, whether oneand considers seems to that the coupled logistic model may becurve); as tonumbers produce which geologic and biologic it was worked out at ordinal familialaspecies species diversity rise, but withdampened but congruent signal whenor Court (My)a (10). as body size, colonizing ability somewhat signals notdiet, obviously separated. levels but does not the work atintrinsic) generic out predictable (8–11). Density depenTable 1.limit Macroevolutionary phenomena and their support for either Redconvincingly Queen (biotic, Jesterare (physical, extrinsic) models.

Manyor ecologiResolution equilibrium models, Correction for samcorrections are (10, imposed. specialization, appear to have effect on couldbetween fit either worldview, and are noted “multilevel mixed.” on land today may be aslittle much as 25 specific levels 11), there are problems with cal Life dence of origination andthe extinction ratessodoes not asor expansionist models, investhe diversity of asmodern organisms, although and between these and plingnumerical is clearlyassumptions essential (6,(11, 7),12), and and future times as diverse life in the sea, so it may be key the backpreclude expansionist models because they may Red Queen Court Jester Multilevel mixed solution ground life-history characmight seem by straightforward, tigationassumption must determine appropriate indepenand r-selected wrong to generalize from marine paleontologof a global carrying capacity is abundance be dampened limiting factorsbut suchthe as shortage Interspecific competition Waxing and waning of cladesfor in association with and humanVicariance and dispersal in major phylogenetic splitslarge (17) preservation error; depends adequate assessment quality (short dent proxies ical studies to gestation all life. period, Perhaps landlitter andsize, sea (8, 10, 11).

Further, it has proved hard to teristics of food oron space, or active predation,ofasthe well as by doubtful tectonic and oceanographic events (2, 17) record.displacement The long-term saturation as number of fossome current proxies (such and short interbirth intervals) sometimes correof the fossil show similar patterns of exponential increase in export the logistic model to the much more climate andCharacter other physical factors. Further, it Mass extinctions and smaller extinction events Latitudinal diversity gradient (22–24) (Fig. may 2, triggered blue withnumbers high species model that for global diversification siliferous localities) themselves species (8, richness 9, 11), (16). or perhaps they speciose whether onedependent considers late seems the coupled logistic model be by extrinsicterrestrial causes suchrealm, asare eruptions, 0

The Red Queen and the Court Jester: Species Diversity and the Role of Biotic The Abiotic Red Queen andThrough the Court Jester: and Factors Time Species Diversity and the Role of Biotic and Abiotic Factors Through Time

Number of genera (empirical; red)

Speciation

climate change, anoxia, impact (10, 11)

Table 1. Macroevolutionary phenomena Coordinated and their support for either the Red Queen (biotic, intrinsic) or Court Jester (physical, extrinsic) models. Many Evolutionary arms races (1) turnovers, originations, and extinctions Occupation of new ecospace (25) could fit either worldview, and so are noted as “multilevel mixed.” in response to physical perturbations– termed Red Queen

“coordinated stasis” or “turnover pulse” hypothesis (2, 29, 30) Court Jester Nonconstant probability of extinction (3, 11)with Waxing and waning of clades in association

of ecological InterspecificConstancy competition guilds through time (25) tectonic and oceanographic events (2, 17) Incumbency advantage Lack of evidence for a global carrying capacity and Character displacement Mass extinctions and smaller extinction events (3, 24) equilibrium levels (8, 10) triggered extrinsic causes such“evolutionary as eruptions, Lack ofby cohesiveness of the great climatefaunas” change, (12)anoxia, impact (10, 11) Species turnovers, richness–energy relationshipand (18,extinctions 19) Evolutionary arms races (1) Coordinated originations,

Constancy of ecological guilds through time (25) Incumbency advantage (3, 24)

in response to physical perturbations– termed Inverse relationship between global temperature and “coordinated stasis” or “turnover pulse” hypothesis biodiversity (21) (2, 29, 30) Lack of clear correlation of species richness with Nonconstant extinction (3, 11) body probability size or other of biotic factors (16) Lack of evidence for a global carrying capacity and www.sciencemag.org SCIENCE VOL 323 equilibrium levels (8, 10) Lack of cohesiveness of the great “evolutionary faunas” (12) Species richness–energy relationship (18, 19)

Multilevel mixed Subdivision of niches/specialization (10, 25)phylogenetic splits (17) Vicariance and dispersal in major Declining global extinction rates through time (1, 5)

Latitudinal diversity gradient (22–24)

Onshore-offshore patterns and disturbance (3) Resource use: stenotopes more speciose Occupation of new are ecospace (25) than eurytopes (29, 30)

Subdivision of niches/specialization (10, 25) Declining global extinction rates through time (1, 5) 729 Onshore-offshore patterns and disturbance (3)

6 FEBRUARY 2009

Resource use: stenotopes are more speciose than eurytopes (29, 30)

Inverse relationship between global temperature and biodiversity (21) Lack of clear correlation of species richness with body size or other biotic factors (16)

23 www.sciencemag.org

SCIENCE

VOL 323

6 FEBRUARY 2009

729

their continuing diversification in the Late Jurassic and Cretaceous was mainly indistinguishable from a random walk. In particular, dinosaurs did not participate in the Cretaceous Terrestrial Revolution, some 130 to 100 Ma, when flowering plants, leaf-eating insects, social insects, squamates, and many other modern groups radiated substantially.

A

245.9

237

Anisian

shifts should mainly occur low in a clade’s history: Clade shapes vary from bottom-heavy to topheavy, and diversification shifts may be concentrated low (dinosaurs and bats) or high (insects and ants) in a clade (26).

In the future, the identification of diversification shifts across numerous taxa may provide evidence

232.5 228

Lad

203.6 199.6

Crn

Norian

Rh. EJ

PTEROSAURIA

“Dinosauromorphs” ORNITHISCHIA

Dinosauria

SAUROPODOMORPHA THEROPODA PHYTOSAURIA

Crurotarsi

AETOSAURIA CROCODYLOMORPHA

“Rauisuchids” POPOSAUROIDEA ORNITHOSUCHIDAE

B 0.12

Principal coordinate 2

0.06 0

Dinosaur morphospace

-0.06 -0.12

Pterosaur morphospace

-0.18 -0.24 -0.3 Crurotarsan morphospace -0.36 -0.24

-0.16

-0.08

0

0.08

0.16

0.24

0.32

Principal coordinate 1

Fig. 3. Phylogenetic relationships and morphospace occupation for Triassic archosaurs. (A) Framework phylogeny for Triassic crurotarsans scaled to the Triassic time scale. Numbers at top refer to millions of years before the present; gray bars represent the observed durations of major lineages; vertical dashed lines denote two extinction events, at the Carnian-Norian and Triassic-Jurassic boundaries; arrowheads indicate lineages that survived the latter event. Lad, Ladinian; Crn, Carnian; Rh, Rhaetian; EJ, Early Jurassic. (B) Empirical morphospace for Late Triassic archosaurs, based on the first two principal coordinates. Large circles, dinosaurs; ovals, pterosaurs; squares, poposauroids; hexagons, phytosaurs; stars, aetosaurs; crosses, crocodylomorphs; smaller black dots, “rauisuchids”; larger black dots, nondinosaurian dinosauromorphs, Scleromochlus. Based on (28).

Geographic and tectonic history has generated patterns of species diversity through time.

The slow dance of the continents as Pangaea broke up during the www.sciencemag.org past 200 My has affected SCIENCE VOL 323 modern distribution patterns. Unique terrestrial faunas and floras, notably those of Australia and South America, arose because those continents were islands for much of the past 100 My. Further, major geologic events such as the formation of the Isthmus of Panama have permitted the dispersal of terrestrial organisms and have split the distributions of marine organisms. A classic example of vicariance is the fundamental division of placental mammals into three clades, Edentata in South America, Afrotheria in Africa, and Boreoeutheria in the northern hemisphere, presumably triggered by the split of those continents 100 Ma (17). Other splits in species trees may relate to dispersal events, or there may be no geographic component at all.

Species richness through time may correlate with energy. The species richness–energy relationship (18) posits correlations with evapotranspiration, temperature, or productivity, and studies of terrestrial and marine ecosystems have shown that these factors may explain as much as 90% of current diversity, although relationships between species diversity and productivity change with spatial scale (19). Over long time spans, there are strong correlations between plankton morphology and diversity and water temperature: Cooling sea temperatures through the past 70 My, and consequent increasing ocean stratification, drove a major radiation of Foraminifera, associated with increasing body size (20). More widely, there is close tracking be-

24

Court Jester worldviews. If the majority of diversification shifts are coordinated, and associated with particular climatic, tectonic, and geographic drivers, then the Court Jester model of macroevolution would prevail. This would link most increases in species diversity to particular largescale radiation events, such as the Cretaceous Terrestrial Revolution (26), or recoveries after mass extinctions. If, on the other hand, the majority of diversification shifts are unique to particular clades, and not coordinated temporally with others, then the Red Queen worldview might be considered.

Comparing Sister Taxa A powerful element of the comparative phylogenetic approach to species diversity relationships through time Fig. 3. Phylogenetic and moris the opportunity to compare sister taxa. Sisters phospace occupation fortheirTriassic archosaurs. arose from a single ancestor, and so trajectoriesFramework occupy the same amount of time, and (A) phylogeny for Triassic cruthey started with the same genotype and pherotarsans scaled tosubsequent the Triassic time scale. notype. Any similarities in their evolution probably reflect this phylogenetic signal of Numbers at top refer to millions of years a common origin, but differences reflect independent aspects separate histories. before theof their present; gray bars represent the Comparisons of sister taxa have allowed tests observed durations of the resource-use hypothesis (29),of thatmajor general- lineages; verists are less speciose and have longer species tical dashed lines denote two extinction durations than specialists. Specialists divide the physical environment small patches, each events, at theintoCarnian-Norian and Triassicoccupied by a species, and each probably more Jurassic boundaries; arrowheads indicate subject to environmental crises than their generalist relatives. Classic examples in support of latter the lineages that survived the event.

Lad, resource-use hypothesis come from studies of NeLadinian; Carnian; Rh, Rhaetian; EJ, Early ogene mammalsCrn, (29). For example, two antelope subgroups, the tribes Alcelaphini and Aepycerotini, Jurassic. morphospace for Late diverged 6 to (B) 8 Ma.Empirical The former is now highly speciose, witharchosaurs, some 7 living and 25 extinct speTriassic based on the first two cies, and the latter is represented by two species, principal coordinates.

Large only one, the impala Aepyceros, surviving.

The circles, dinoslowly evolving Aepycerotini consists of few saurs; ovals, pterosaurs; squares, poposauspecies at any time, and each of those is long lived, whereas the speciosephytosaurs; Alcelaphini consistsstars, aetosaurs; roids; hexagons, of many short-lived species. The ecological crosses, crocodylomorphs; habits of both clades differ: The impala has a smaller black broad, generalist diet, whereas the specioseblack aldots, “rauisuchids”; larger dots, nondicelaphines show more dietary specialization. In nosaurian dinosauromorphs, wider studies of many clades of Neogene African Scleromochlus. and South American mammals (30), the resourceBased onwas (28). use hypothesis supported, and some subsidiary predictions confirmed: Specialists are more common than generalists, carnivores include more generalists than herbivores, and there are more specialists in habitats that underwent recent environmental change (tropical rain forests and deserts). The resource-use model then stresses the role of climate and tectonic movements in determining species diversity rather than biological controls such as competition and predation.

tween temperature and biodiversity on the global scaleOutlook for both marine and terrestrial organisms Paleontologists and evolutionary ecologists have debated species diversity and largely familial independently. richness were (21), where generic relatively low during warm “greenhouse” phas6 FEBRUARY 2009 731 es of Earth history, coinciding with relatively high origination and extinction rates.

A much-studied manifestation of energy and temperature gradients is the latitudinal diversity gradient (LDG), namely the greater diversity of life in the tropics than in temperate or polar regions, both on land and in the sea. There are two explanations (22): (i) the time and area hypothesis, that the tropical belt is older and larger than temperate and polar zones, and so tropical clades have had longer to speciate, or (ii) the diversification rate hypothesis, that there are higher rates of speciation and lower rates of extinction in the tropics than elsewhere. There is geological and paleontological evidence for a mixture of both hypotheses (23, 24).

Species diversity may increase by the occupation of new ecospace. The number of occupied guilds, that is, broad ecological groupings of organisms with shared habits, has increased in several steps through time, from 20 in the early Paleozoic to 62 in post-Paleozoic marine faunas (25). Further, marine animals have shown several step increases in tiering, the ability to occupy and exploit different levels in the habitat: At times, burrowers have burrowed deeper, and reef-builders have built taller and more complex reefs. Analogous, if even more dramatic, expansions of ecospace have occurred on land, with numerous stepwise additions of new habitats, from the water-margin plants and arthropods of the early Paleozoic to the forests

and upland habitats of the later Paleozoic when land animals first burrowed, climbed, and flew, through the introduction of herbivory, giant size, endothermy, and intelligence among vertebrates, and the great blossoming of flowering plants (with associated vast expansions in diversity of plant-eating and social insects and modern vertebrates) during the Cretaceous Terrestrial Revolution 100 Ma (26).

The other mode of species increase globally or regionally is by niche subdivision, or increasing specialization. This is hard to document because of the number of other factors that vary between ecosystems through time. However, mean species number in communities (alpha diversity) has increased through time in both marine (15, 25) and terrestrial (10) systems, even though niche subdivision may be less important than occupation of new ecospace in increasing biodiversity. Further, morphological complexity may be quantified, and a comparative study of crustaceans shows, for example, that complexity has increased many times in parallel in separate lineages (27).

Phylogenetic Studies of Clade Histories Species are not randomly distributed; they have an evolutionary history, and so occur as twigs on a great phylogenetic tree. Studying species as members of clades is a fruitful approach to understanding the drivers and controls on speciation. Key questions include (i) Do species diversify early in a clade’s history? (ii) How do diversity and disparity (variance in characters or morphology) covary? (iii) Do major lineages within a clade follow similar, or different, patterns, and if different, why? (iv) Do evolutionary radiations follow the acquisition of new characters or emptying of ecospace? (v) How do major clades of apparent competitors interact over long spans of geologic time? and (vi) How do sister clades vary in species diversity and why?

For such analyses, the ideal is a complete species tree, a phylogenetic tree that contains all species living and extinct, plotted accurately against geologic time (4). Simple to say; hard to achieve. More commonly, incomplete trees have been used, with the risk of error in calculations of evolutionary rates or comparisons of subclades. In paleontology, it has proven much easier to work with higher taxa such as genera or families because species fossil records are less complete than those of higher taxa, and yet it is not clear how higher-level patterns relate to those at species level. Many key questions can be tackled by comparing a real tree to a hypothetical tree that follows an equal-rate Markov (ERM) model, equivalent to tree growth after a random walk, where equal chances of speciation and of extinction are shared by all species (4).

Major biotic replacements, where one clade replaces another, have been a focus of debate about the roles of competition and progress in

macroevolution, and dinosaurs provide a classic example. The standard view was that dinosaurs originated in the Late Triassic, some 230 Ma, by a process of competition in which they prevailed over their precursors, the crocodile-like crurotarsans and others, because of superior adaptations.

A comparative phylogenetic study (28) shows, however (Fig. 3), that the Dinosauria expanded in two steps, one after an extinction event 225 Ma that removed dominant herbivores, and the second following the end-Triassic extinction 200 Ma that removed most of the crurotarsans. Dinosauria remained at moderate diversity and low disparity, and at lower disparity than the crurotarsans they supposedly out competed, during the 25 My between the events, suggesting that there was no insistent competition driving other groups to extinction but rather that the dinosaurs occupied new ecospace opportunistically, after it had been vacated.

A further study on Dinosauria explored the subsequent evolution of the clade (26). Classic views that the dinosaurs arose with a flourish, and then finally gave way in the Cretaceous to the superior mammals, or that they dwindled to extinction because of “racial senility,” had long been abandoned. The dinosaurs seemed to be radiating actively in the Cretaceous, with many new clades appearing through their last 55 My, and especially in their final 15 My. The new study (26) shows that most diversification shifts (departures from ERM assumptions) fall in the first one-third of the history of the clade and that their continuing diversification in the Late Jurassic and Cretaceous was mainly indistinguishable from a random walk. In particular, dinosaurs did not participate in the Cretaceous Terrestrial Revolution, some 130 to 100 Ma, when flowering plants, leaf-eating insects, social insects, squamates, and many other modern groups radiated substantially. There is no geometric reason that diversification shifts should mainly occur low in a clade’s history: Clade shapes vary from bottom-heavy to top-heavy, and diversification shifts may be concentrated low (dinosaurs and bats) or high (insects and ants) in a clade (26).

In the future, the identification of diversification shifts across numerous taxa may provide evidence for the relative importance of the Red Queen and Court Jester worldviews. If the majority of diversification shifts are coordinated, and associated with particular climatic, tectonic, and geographic drivers, then the Court Jester model of macroevolution would prevail. This would link most increases in species diversity to particular large-scale radiation events, such as the Cretaceous Terrestrial Revolution (26), or recoveries after mass extinctions. If, on the other hand, the majority of diversification shifts are unique to particular clades, and not coordinated temporally with others, then the Red Queen worldview might be considered.

Comparing Sister Taxa A powerful element of the comparative phylogenetic approach to species diversity through time is the opportunity to compare sister taxa. Sisters arose from a single ancestor, and so their trajectories occupy the same amount of time, and they started with the same genotype and phenotype.

Any similarities in their subsequent evolution probably reflect this phylogenetic signal of a common origin, but differences reflect independent aspects of their separate histories.

Comparisons of sister taxa have allowed tests of the resource-use hypothesis (29), that generalists are less speciose and have longer species durations than specialists. Specialists divide the physical environment into small patches, each occupied by a species, and each probably more subject to environmental crises than their generalist relatives. Classic examples in support of the resource-use hypothesis come from studies of Neogene mammals (29). For example, two antelope subgroups, the tribes Alcelaphini and Aepycerotini, diverged 6 to 8 Ma. The former is now highly speciose, with some 7 living and 25 extinct species, and the latter is represented by two species, only one, the impala Aepyceros, surviving. The slowly evolving Aepycerotini consists of few species at any time, and each of those is long lived, whereas the speciose Alcelaphini consists of many short-lived species.

The ecological habits of both clades differ: The impala has a broad, generalist diet, whereas the speciose alcelaphines show more dietary specialization. In wider studies of many clades of Neogene African and South American mammals (30), the resource-use hypothesis was supported, and some subsidiary predictions confirmed: Specialists are more common than generalists, carnivores include more generalists than herbivores, and there are more specialists in habitats that underwent recent environmental change (tropical rain forests and deserts). The resourceuse model then stresses the role of climate and tectonic movements in determining species diversity rather than biological controls such as competition and predation.

Outlook Paleontologists and evolutionary ecologists have debated species diversity largely independently.

The realization that the Red Queen and Court Jester models may be scale-dependent, and that evolution may be pluralistic (3), opens opportunities for dialog. Taxic studies in paleontology continue to have great value in highlighting correlations between species richness and other factors, but comparative phylogenetic methods will illuminate questions about clade dynamics, species richness, and the origin of novelties.

Further, methods are shared by paleontologists and neontologists, and this allows direct communication on the patterns and processes of

macroevolution. Viewed close up, evolution is all about biotic interactions in ecosystems (Red Queen model), but from further away, the large patterns of biodiversity are driven by the physical environment (Court Jester model).

References and Notes

1. L. M. Van Valen, Evol. Theory 1, 1 (1973).

2. A. D. Barnosky, J. Vert. Paleont. 21, 172 (2001).

3. D. Jablonski, Evolution 62, 715 (2008).

4. A. Purvis, Annu. Rev. Ecol. Syst. 39, 301 (2008).

5. J. J. Sepkoski Jr., Paleobiology 10, 246 (1984).

6. J. Alroy et al., Science 321, 97 (2008).

7. J. Alroy, Proc. Natl. Acad. Sci. U.S.A. 105, 11536 (2008).

8. T. D. Walker, J. W. Valentine, Am. Nat. 124, 887 (1984).

9. M. J. Benton, Science 268, 52 (1995).

10. M. J. Benton, B. C. Emerson, Palaeontology 50, 23 (2007).

11. S. M. Stanley, Paleobiology 33, S1 (2007).

12. J. Alroy, Evol. Ecol. Res. 6, 1 (2004).

13. G. J. Eble, Geobios 32, 223 (1999).

14. M. J. Benton, Geol. J. 36, 211 (2001).

15. R. K. Bambach, A. H. Knoll, J. J. Sepkoski Jr., Proc. Natl. Acad. Sci. U.S.A. 99, 6854 (2002).

16. N. J. B. Isaac, K. E. Jones, J. L. Gittleman, A. Purvis, Am. Nat. 165, 600 (2005).

17. D. E. Wildman et al., Proc. Natl. Acad. Sci. U.S.A. 104, 14395 (2007).

18. G. Hunt, T. M. Cronin, K. Roy, Ecol. Lett. 8, 739 (2005).

19. G. G. Mittelbach et al., Ecology 82, 2381 (2001).

20. D. N. Schmidt, H. R. Thierstein, J. Bollmann, R. Schiebel, Science 303, 207 (2004).

21. P. J. Mayhew, G. B. Jenkins, T. G. Benton, Proc. R. Soc. London B. Biol. Sci. 275, 47 (2008).

22. G. G. Mittelbach, Ecol. Lett. 10, 315 (2007).

23. D. Jablonski, K. Roy, J. W. Valentine, Science 314, 102 (2006).

24. J. W. Valentine, D. Jablonski, A. Z. Krug, K. Roy, Paleobiology 34, 169 (2008).

25. R. K. Bambach, A. M. Bush, D. H. Erwin, Palaeontology 50, 1 (2007).

26. G. T. Lloyd et al., Proc. R. Soc. London B. Biol. Sci. 275, 2483 (2008).

27. S. J. Adamowicz, A. Purvis, M. A. Wills, Proc.

Natl. Acad. Sci. U.S.A. 105, 4786 (2008).

28. S. L. Brusatte, M. J. Benton, M. Ruta, G. T. Lloyd, Science 321, 1485 (2008).

29. E. S. Vrba, Am. J. Sci. 293A, 418 (1993).

30. A. M. Bofarull et al., BMC Evol. Biol. 8, 97 (2008).

31. S. Finnegan, J. L. Payne, S. C. Wang, Paleobiology 34, 318 (2008).

32. Thanks to S. J. Braddy, P. C. J. Donoghue, D. Jablonski, A. Purvis, M. Ruta, D. N. Schmidt, and anonymous referees for comments and to S. Powell for drafting the figures. Supported by the Natural Environment Research Council and the Royal Society.

25

theories of the process (8, 11). I leave out discussion of sympatric and allopatric speciation but SPECIALSECTION Under this the Darwinian instead identify likelihoodperspective, of ecologicallinkand was relatively ing speciation with adaptation mutation-order speciation when there is gene 61. R. C. Albertson, J. T. Streelman, T. D. Kocheer, J. Hered.

52. O. Seehausen et al., Nature 455, 620 (2008).

41. D. Jablonski, Evolution 62, 715 (2008). straightforward, requiring onlyaaspecial test of whether flow.

I ignore reinforcement, type of 94, 291 (2003).

53. R. Lande, Genetics 128, 443 (1991).

42. R. J. Whittaker, K. A. Triantis, R. J. Ladle, J. Biogeogr. 35, 62. K. Schwenk, N. Brede, B. Streit, Philos.species Trans. R. Soc. 54. D. A. Joyce et al., Nature 435, 90 (2005).

977 (2008). phenotypic differences between were natural selection thought to favor stronger preLondon Ser. B 363, 2805 (2008).

55. G. Fryer, Environ. Biol. Fishes 45, 109 (1996).

43. R. Gillespie, Science 303, 356 (2004). selection. example, at the caused by natural mating reproductive isolationFor once postzygotic 63. D. Garant, S. E. Forde, A. P. Hendry, Funct. Ecol. 21, 434 56. J. W. Valentine, D. Jablonski, in Evolution, Time and 44. J. Todd Streelman, P. D. Danley, Trends Ecol. Evol. 18, Advancement of American Association for the isolation has evolved. I also ignore speciation by (2007).

Space, R. W. Sims, J. H. Price, P. E. S. Whalley, Eds. 126Schluter (2003).

Dolph Science 1939 [the last polyploidy, evenspeciation though crucial 64. O. Seehausen, Curr. Biol.selection 16,symposium R334 might (2006).be (Academic Press, London, 1983), pp. 201–226.

45. D. Ackerly, D. Schwilk, C. Webb, Ecology 87, S50 (2006).

65.

Darwin,stages.

Voyage on of the Boogle (Collier, Newthe York,bio57. C. O. Webb, D. D. Ackerly, M. A. McPeek, M. J. Donoghue, in 46. C. E. Parent, B. J. Crespi, Evolution 60, 2311 (2006). theC.early symposium speciation before major Natural drives the origin as Darwin initially Mechanisms of 1909), p. 383.concept (7)], an extensive comRev. Ecol.

Syst. 33,claimed.

475 (2002).

47. J. A. selection Coyne, T. D.commonly Price, Evolution 54, 2166 (2000). of species, Annu. logical species speciation

by D. selection fall into mutation-order.

Under 66. We thankand A. Birand, R. Glor, A.from Harrison, L. Harmon, 58. ecological J. J. Wiens, C.and H. Graham, Annu. Rev. Ecol.

Syst. 36, 519 Speciation 48. J. B. Losos, Schluter, Nature 408,two 847 broad (2000). categories: Adaptation inparative and biogeographic studyand showcased D Jablonski, L. Mahler, C. Marshall, the anonymous 49. D. Lack,speciation, Darwin’s Finches (Cambridge is Univ.

Press, by divergent(2005). ecological divergence driven natural selection between environments, Darwin to Dobzhansky stances in which derived morphological and life reviewers for useful comments.

Supported by the NSF 59. S. Skulason, T. B. Smith, Trends Ecol. Evol. 10, 366 Cambridge, 1947). whereas under mutation-order speciation, divergence occurs when different mutations arise and (grant DEB-0519777) and thearisen NIH between (grant (1995).

50. U. Dieckman, M. Doebeli, J. A. J. Metz, D. Tautz, Adaptive Appreciation of connection adaptaofthe fishes had overGM56693). and over history forms are fixed in separate adapting to similar60. selection pressures. of parallel evolution H. D. Bradshaw, D. W.Tests Schemske, Nature 426, 176 Speciation (Cambridgepopulations Univ. Press, Cambridge, 2004). tion and speciation began withtype Darwin when a (12).

The reagain from the same ancestral of51.reproductive trait-based assortative and reproductive isolation by active 10.1126/science.1157966 (2003).

U. K. Schliewen,isolation, D. Tautz, S. Pääbo, Nature 368, 629 (1994). mating, morphological concept of nonparasitic species largely prepeated, parallel origin of lamprey selection have demonstrated that ecological speciation is a common means by which new species vailed.

In Onfrom the Origin of Species, Darwin wrote in streams the same migratory, parasitic arise. Evidence for mutation-order speciation by natural selection is more limited and has been that “I look at thethat term“Again species, asagain one arbitrarily and the paraancestor showed best documented by instances of reproductive isolation resulting from intragenomic conflict. given forofthe sake ofevolved convenience to aout setdisof nonparasitic sitic lampreys have into theories the process (8, 11).

I leave However, we still have not identified all aspects of selection, and identifying the underlying genes REVIEW individuals closely resembling each other...” and forms...correlated with in small streams, cussion of sympatric and life allopatric speciation but for reproductive isolation remains challenging.

“The amount of difference is one very important where a identify suitable foodlikelihood supply inofthe way of large instead the ecological and criterion in settling whether(12). two forms should be or seasonal” When correlated fish is scarce mutation-order speciation when there is gene ries: ecological speciation and mutation-order tttook evolutionary biologists nearly 150 years, ranked as species or varieties” (1).

Under this flow. environmental I ignore reinforcement, a special type of took evolutionary biologists nearly 150 would nevertheless be advantageous in both of with repetition is factors, such speciation is thought defined as accumulation of Ecological refers to the but at last can we agree Darwin the speciation. natural selection to the favor stronger precanwith agree with that Darwin of view, years, butwe at last their environments.

Thespeciation relative importance unlikely to result from chance; environmental origin of species, “that mystery of mysteries” evolution of reproductive isolation between pop- sufficiently many differences between populamating reproductive isolation oncebepostzygotic pressures must therefore the cause that the origin of species, “that mystery of these two categories of mechanism for the ori- selection (1), really does occur by means of natural selection ulations or subsets of a single population by ad- tions to warrant their Iclassification as separate isolation has evolved. also ignore speciation by mysteries” (1), really does occur by means of gin of species in nature is unknown. of speciation. “As a result of our recent studies Dolph Not Schluter (2–5). all species appear to evolve by aptation to different environments or ecological taxonomic species.

Darwin understood the impolyploidy, even though selection might be crucial In this review, I summarize progress in un- on being added to fishes...weight is constantly natural selection (2–5). Not all species appear to selection, but the evidence suggests that most of niches (2, 8, 9). Natural selection is divergent, portance of reproductive barriers between species in the early stages. but the evidence suggests features of speciation evolve by selection, derstanding the general the theory that speciation is...under the rigid Natural origin of species, initially claimed.

Mechanisms of (1), but the study of speciation after the pubactingas in Darwin contrasting directions between environthem do.selection The effortcommonly leading updrives to thisthe conclusion to by selection. speciation control of the environment” (12). However, this that most of do. The leading Iand do mutation-order. not differentiateUnder speciation bythem selection fall effort into broadupcategories: ecological which drives the fixation of different lication involved many experimental and two conceptual ad- ments, of this work focused mainly on the evoSpeciation Adaptation from of referring to the origin morphothis conclusion involved many isexperimental by sexual selection herebetween becauseenvironments, natural selec- case is onlyand ecological speciation, divergence selection alleles,natural each advantageous in one environment lution vances, including a revision of the driven notionbyofdivergent oftospecies differences, particularly of Darwin Dobzhansky advances, including a revision drives the divergence of mate preferences, logical species. and conceptual tion whereas under speciation,ofdivergence occurs when different mutations in the other.

Following G. S. arise Maniand and morphological speciation itself, mutation-order 80 years after publication On but not but also of between behavioral and Appreciation oftraits the connection adapta80 years orTests mutation-order mechaThe turning point for speciation studies of notion of Species, speciation byselection either ecological arethe fixed in populations adapting toon similar pressures. of parallel evolution B. C. Clarke (10), I define mutation-order specia- other the Origin of separate the to aitself, definition basedafter phenotypic traits. tion and speciation beganconcept with Darwin when a to mating, (8, 11).byI came of speciation publication ofisolation On the instead Origin of the Species, nisms, in most theories of the process with the Darwinian modern of reproductive isolation, trait-based assortative reproductive isolation byisolation active tion as and the evolution of reproductive reproductive of morphological Under this linking morphological concept of perspective, species largely of prethe a definition based on reproductive isolation inleave out discussion of sympatric and allopatric “Species separation is defined as a stage selection have is a common means which of newdifferent species speciation with adaptation was relatively straightthe chance occurrence andbyfixation differences (6, 7).demonstrated that ecological speciation vailed.

In On the Origin of Species, Darwin wrote differences (6, 7). speciation but instead identify the likelihood of evolutionary process at which physiological stead of morphological arise.

Evidence for mutation-order speciation by natural selection is more limited and has been The main question today is how does selec- alleles between populations adapting to similar forward, requiring test of whether that “I look at theonly termabecome species, as one phenotyparbitrarily main question todayare is of how does selec-isolation and mutation-order speciation when ic mechanisms developed” (6) ecological isolating bestThe documented by instances resulting from intragenomic conflict. selection pressures.

Reproductive isolation evolves tion lead to speciation?

What thereproductive mechanisms differences between species were caused by given for the sake of convenience to of What are the mechareinforcement, a spetoa set mean tion lead to speciation? there is gene flow.

I ignore (here, “physiological” is interpreted However, we still have not identified all aspects of selection, and identifying the underlying genes of natural selection, what genes are affected, and because populations fix distinct mutations that natural selection.

For example, at the American individuals closely resembling each other...” and favor nisms naturalatisolation selection, what genes are af- would cial type of naturalbeselection thought evolved reproductive isolation between populafor reproductive remains challenging. nevertheless advantageous in to both of Association how doofchanges these genes yield the habitat, forofthedifference Advancement Science 1939 “The amount is oneofvery important from[the geographical barriers to fected, and mechanical, how do changes at these genes yield their stronger premating reproductive isolation once tions, as distinct environments.

The relative importance of speciation behavioral, chemical, physiological, symposium last major symposium criterion in settling whether two forms should be chemical, these isolationofhas evolved. for I also species were deinterbreeding).

Subsequently, the other habitat, behavioral, mechanical, postzygotic categories mechanism the ignore origin on and that are the reproducspeciation before the varieties” biological species concept ecological speciation and mutation-order t took incompatibilities evolutionary biologists nearly 150 years, ries: two ranked as species or (1).

Under this that even though selection (7)], physiological, and other incompatibilities speciation bynature polyploidy, finedanasextensive “groups of interbreeding natural popuof species in is unknown. tive barriers between new species?

As a start, the comparative and biogeographic but at last we can agree with Darwin that the speciation. Ecological speciation refers to the view, speciation is defined as the accumulation of that are instances reproductively isolated from are theways reproductive between new be crucial in the early stages. lationsshowcased In this review, I summarize progress in undermany which barriers new arisespeby might in which derived morevolution of reproductive isolation between pop- study origin ofbyspecies, “thatspecies mysterymight of mysteries” sufficiently many differences between populathe many ways by which new standing the general features of speciation by se- phological this of point on,had the other such and groups” (7). From cies? As can a start, selection be grouped into two broad categolife history forms fishes (1), really does occur by means of natural selection ulations or subsets of a single population by ad- tions to warrant their classification as separate Speciation and Adaptation from Darwin can be grouped the evoluspecies might arise by selection study of speciation was the study of lection.

I do not differentiate speciation by sexual arisen over and over again from the same ancestral (2–5). Not all species appear to evolve by aptation to different environments or ecological taxonomic species. Darwin understood the imto Dobzhansky into two broad categories: ecological speciation tion of reproductive isolation (3).between Progress up to here natural selection the type (12). repeated, parallel origin ofspecies nonniches (2, 8, because 9). Natural selection is drives divergent, selection, but the evidence suggests most of selection portance ofThe reproductive barriers Biodiversity Research Centre and Zoologythat Department, divergence of of mate by either eco- parasitic lamprey in streams frombetween theafter same and speciation.

Appreciation the preferences, connectionbetween between adapthen in understanding link morphoUniversity British Columbia, Vancouver, V6T spe1Z4, acting in contrasting directions environthemmutation-order do.ofThe effort leading up toEcological thisBCconclusion (1), but the study of the speciation themigrapuborwhich mutation-order most tory, parasitic showed that was “Again and Canada.

E-mail: schluter@zoology.ubc.ca toexperimental the evolution reproductive with Darwin when adaptation largely ciation refers tation speciation logical speciation ments,and drives began themechanisms, fixation of in different involved many andofconceptual ad- logical lication of thisancestor workand focused mainly on the evoor subsets of a under isolation between populations a morphological concept of species largely forgotten, its contributions uncertain vances, including a revision of the notion of alleles, each advantageous in one environment lution of species differences, particularly the of On the Following Origin of Species, Darwin single population adaptation to different concept. traits but also of behavioral and but not in In theSCIENCE other.

S. Mani and new speciation itself, 80byyears after publication of enOn prevailed. morphological www.sciencemag.org VOL G. 323 6 FEBRUARY 2009 737 lookI at the mutation-order term species, as one other Thephenotypic biologicaltraits. species concept must surely vironments ecological niches (2, 8, 9).

Natuthat “I(10), B. C. Clarke define speciathe Origin oforthe Species, to a definition based on wrote divergent, acting contrasting arbitrarily difficultperspective, to investigate any given forofthe sake of convenience made this it more ral selection isisolation tion as the evolution reproductive isolation by haveUnder reproductive instead of in morphological Darwinian linking selection.

T. directions set of individuals resembling each link between speciation natural thea chance occurrence closely and fixation of different differencesbetween (6, 7). environments, which drives to speciation with adaptationand was relatively straightand “The amount ofadapting difference is one Dobzhansky the The fixation differenttoday alleles, eachdoes advanta(13) only suggested the genes unalleles between populations to similar mainofquestion is how selec- other...” forward, requiring a test ofthat whether phenotypin to onespeciation? environment the other. very criterion in settling whether two derlying differences between in orimportant geous selection pressures.

Reproductive isolation evolves tion lead Whatbut arenot thein mechanisms ic differences between speciespopulations were caused by andgenes B. C.areClarke (10), speciesmutations or varieties” were unlikely to be the Following G. S. Mani should be ranked traits because populations fix as distinct that dinary of natural selection, what affected, andI forms natural phenotypic selection. For example, at the American the habitat, evolu- (1). as the changed define speciation as the Under this view,be speciation is defined of reproductive isolation. He would nevertheless advantageous in both of basis how domutation-order changes at these genes yield Association for the Advancement oflater Science 1939 tion of reproductive the chance accumulation of sufficiently many differences mind, but at the time and the their environments.

The relative importance of his behavioral, mechanical,isolation chemical,byphysiological, speciation symposium [thethis lastviewpoint, major symposium and fixation of that different be- between to mechanism warrant their greater of studying reprooccurrence these twopopulations categories of forclassificathe origin generally and other incompatibilities are thealleles reproducon speciation beforedifficulty the biological species concept of species in nature is unknown. tive barriers betweenadapting new species?

As a start, the tion (7)], an extensive comparative and biogeographic to similar selecspecies. Darwin un- ductive morphology, must have tween populations as separate taxonomic isolation than In thisthe review, I summarize progress inbarriers under- discouraged manypressures. ways by which new species might evolves arise by derstood study showcased instances in which derived mortion Reproductive isolation importance of reproductive many from pursuing the connecstanding species the general of speciation by se- tion. selectionpopulations can be grouped into twomutations broad categophological and no liferesearch history effort forms followed of fishes that had because fix distinct that between (1), features but the study of speciation Virtually lection. do not differentiate speciation bymainly sexual tested arisen the overrole andofover again from the same ancestral of this work focused adaptation in speciation. after theIpublication selection here because natural selection drives the type (12). The repeated, parallel origin of nonBiodiversity Research Centre and Zoology Department, on the evolution of species differences, particudivergence of mate preferences, by either eco- Models parasitic lamprey in streams from the same migraof Speciation

by Selection of morphological traits but also of behavUniversity of British Columbia, Vancouver, BC V6T 1Z4, larly logical or mutation-order mechanisms, in most tory, parasitic ancestor showed “Again and Canada. E-mail: schluter@zoology.ubc.ca The topic of natural selection that in speciation is ioral and other phenotypic traits.

REVIEW

Evidence for Ecological Speciation and Its Alternative

Evidence for Ecological Speciation and Its Alternative

II

I

26

www.sciencemag.org

SCIENCE

VOL 323

6 FEBRUARY 2009

737

Speciation

way becau parasitic lampreys have evolved ecologiinto non- A Gambusia models of speciation, once again receiving attention. The two most again theBoth ulations d forms...correlated life in small general hypotheses involving selection are parasitic cal and mutation-order,with are theoretically mutations where a suitable food supply in the way ecological and mutation-order speciation. Eco- streams, plausible, and only data can determine order. Div of large fish is scarce or seasonal” (12). When corimportance nature.

The logical speciation is defined as the evolution related their relative chastic bu with environmental factors,insuch repetition of reproductive isolation between populations is unlikely key is to out bychance; whichenvironmental mechanism from gene to figure result from reproductive isolation firstbeevolved by divergent natural selection arising from dif- selection both smal pressures must therefore the cause(3). of infinite) p result ofgenetic our recentdifferences studies on Once “As the aearliest ferences between ecological environments (2, speciation. can be ec is constantly being added to the 8, 9, 14). It predicts that reproductive isolation fishes...weight have accumulated between populations mutationspeciation is...under the rigidmutations control of either process, subsequent should evolve between populations adapting to theory by that ecology d the environment” (12). However, this case is only contrasting environments but not between pop- referring might beorigin favored in one population gence as s to the of morphological species. and turning not the other because studies of epistatic ulations adapting to similar environments. The The evolution point for speciation came with genetic background basic idea has been around for a while (7), al- withinteractions ductive iso the modern concept of speciation “Species tion of eco is defined as a stage of the evoluthough it was tested only recently. The agents of separation (10). Hence, epistasis, including that postzygot process Dobzhansky-Muller at which physiologicalincomisolat- B Mimulus producing divergent selection are extrinsic and can include tionary Specia ing mechanisms become developed” (6) (here, abiotic and biotic factors such as food resources, patibilities in hybrids between species tragenomi “physiological” is interpreted to mean evolved climate, habitat, and interspecies interactions reproductive (3), can result from either ecological or otic drive isolation between populations, as speciation. such as disease, competition, and behavioral in- distinct mutation-order sterility (F from geographical barriers to interbreedSpeciation can be rapid terference. Ecological speciation can lead to the ing). Subsequently, mutationspecies were under definedboth as cause, by c of interbreeding populations alleles that are evolution of any type of reproductive isolation, “groups speciation models,natural because tions cau reproductively isolated from other such to fixation by natural selection including premating isolation, hybrid sterility, are driven countering groups” (7).

From this point on, the study of and intrinsic hybrid inviability as well as extrin- in both cases. However, under the muthe same i speciation was the study of the evolution of sic, ecologically based pre- and postzygotic iso- reproductive tation-order process, the same alleles, Speciation isolation (3). Progress up to then in if present,the would be favored in every lation. Speciation by sexual selection is ecologi- understanding mutation-o link between morphological thelargely early forgotten, stages of population, at least in cal speciation if ecologically based divergent se- speciation vergence and adaptation was gamete re under the new concept. lection drives divergence of mating preferences, its contributions divergence.uncertain For this reason, mutationFig. 1. (A) Example of ecological speciation. Repeatedly and fixation o The species concept when must surely 1.

(A) Example of ecological speciation.

Repeatedly and speciation is difficult there Fig. for example by sensory drive (15). orderbiological independently, the mosquito fish, Gambusia hubbsi, inhabiting more because difficult togene investigate any link independently, mosquito fish, Gambusia hubbsi, inhabitIn accordance with (10), mutation-order spe- haveismade flow increasgeneitflow, blue holes in the the Bahamas has evolved a larger caudal region and geous mu between speciation and natural selection. T. ing bluehead holesin inthethepresence Bahamas evolved a larger of has predators (top) than incaudal their ulations, ciation is defined as the evolution of reproduc- Dobzhansky es the possibility that favorable muta- smaller (16). Div (13) suggested that the genes under- region and smaller head in the presence ofprobability predatorsof(top) absence (bottom) (29).

In laboratory trials, the in one population will than in their absence (bottom) (29). In laboratory trials, two tive isolation by the fixation of different advan- lyingtions occurring differences between populations in ordinary the other learn individuals mating was higher when they were from different tageous mutations in separate populations ex- phenotypic spread traits to other werepopulations, unlikely to be preventing the basis of probability of twothe individuals matingenvironment was higher (and whensimilar they havior und populations having same predation isolation.

He later hisresultmind, were divergence (17, 18).

Anychanged process periencing similar selection pressures. Whereas reproductive from different populations the same preda- tion, not m body shape) than when they were having from opposite predation at the viewpoint, theincluding generally tion environment similar shape)(29)]. than (B) when they efficient s environments.

[Photo(and credit: Brianbody Langerhans Example flow, different alleles are favored between populations but ing in time low this levels of geneand from opposite predation environments.

[Photo credit: is the cul difficultyfacilitates of studyingsubsequent reproductivediverisola- were of reproductive isolation evolving under the mutation-order under ecological speciation, the same alleles greater selection, Langerhans (29)]. (left) (B) Example of reproductive isolation tion than morphology, must have discouraged Brian mechanism.

Male-fertile and male-sterile (right) flowers of mutation-o would be favored in different populations under gence by the mutation-order process evolving thean mutation-order mechanism. between Oregon population of monkeyMale-fertile flowers (M. al scenario many from pursuing the connection. Virtually no F2 hybridsunder speciation mutation-order speciation. Divergence occurs research (19).effort In contrast, andhaving male-sterile (right)male flowers of F2 hybrids guttatus) a cytoplasmic sterility element andbetween nuclear Both followed ecological that tested the role of (left) anyway because, by chance, the populations do adaptation can proceed with or without gene flow, an restorer and population a closely related species flowers (M. nasutus) having neither Oregon of monkey (M. guttatus) hav- ecologica in speciation.

(46,a47).

Both flowers have M. guttatus cytoplasm.

The are theore cytoplasmic maleshown sterility element and nuclear restorer not acquire the same mutations or fix them in the although it is easiest when gene flow ing flower on the left also has the nuclear restorer,having whereasneither the one(46, on only data of Speciation by Selection and a closely related species (M. nasutus) same order. Divergence is therefore stochastic Models is absent. the right, undeveloped anthers, lacks the cytoplasm. restorer. [Photo 47).

Both with flowers shown have M. guttatus The ative imp The topic of natural selection in speciation is once Experiments with laboratory popu- credit: Andrea but the process is distinct from genetic drift. It on theCase left (47)] also has the nuclear restorer, whereas the key is to again receiving attention. The two most general flower can occur in both small and large (though not lations of Drosophila and yeast dem- one on the right, with undeveloped anthers, lacks the restormechanism hypotheses involving selection are ecological and infinite) populations. Selection can be ecologi- mutation-order onstrate the plausibility of ecological [Photoincluding credit: Andrea Case isolation, (47)] isolation, premating hybrid first evolved (3). Once the speciation.

Ecological speciation er. cally based under mutation-order speciation, but is defined speciation.

In thoseofinstances as the evolution reproductivewhen iso- sterility, and intrinsic hybrid inviability as well as ences have accumulated b between populations divergent natural ecology does not favor divergence as such. It lation measurable pre- andby postmating re- extrinsic, ecologically based pre- and postzygotic either process, subsequen isolation.

Speciation by sexual is favored arising isolation from differences between JYAlpha), and selection sexual isolation (ds2)inbe-one populatio evolved, it wasecogreater be- (Odsh, can lead to the evolution of any type of repro- selection productive speciation if ecologically basedMost diver- of because of epistatic inte environments (2, 8, 9, to 14).different It predictsenvironments that ecological tween Drosophila species. these genes ductive isolation, with the exception of ecologi- logical tween lines subjected reproductive isolation should evolve between gent selection drives divergence of mating background (10). Hence, e than between lines raised under homogeneous show molecular signatures of positive selection, cally based pre- and postzygotic isolation. producing Dobzhansky-M populations adapting to contrasting environments preferences, for example by sensory drive (15).

Speciation resulting from intragenomic con- but conditions 21). adapting Laboratory experiments provingwith natural rolespe(3), provided In accordance (10), selection’s mutation-order in hybridsthat between specie not between (20, populations to similar under flict such as meiotic drive or cytoplasmic male environments. on various maintained fixation occurred before complete reproductive ciation is defined as the evolution of reproductive either ecological or mutati Themicrobes basic idea has been around for homoisolation rather afterward. sterility (Fig. 1B) is likely to be mutation-order a while geneous conditions many have by the fixation of than different advanta- The top-down Speciation can be rapid (7), although it wasfor tested onlygenerations recently. isolation geous in separate expebecause alleles ar agents of divergent selection are extrinsic and with genetic divergence consistent the mutations identifying (i) themodels, phenotypic speciation because, by chance, the initial muta- Thedetected approach involvespopulations riencing selection pressures.

Whereas(ii) natural include abiotic andprocess biotic factors food on (22),such butaseffects re- similar those selection traits in both traits under divergent selection, tions causing drive and those countering it are can mutation-order resources, climate, habitat, and interspecies inter- different alleles are favored between populations the mutation-order proces unlikely to be the same in separate populations. productive isolation have not been explored. associated with reproductive isolation, and (iii) actions such as disease, competition, and behav- under ecological speciation, the same alleles present, would be favored Two approaches investigate Speciation by sexual selection is mutation-order ioral interference. the genes underlying traits and reproductive isoin different populations under least in the early stages o Ecological speciation the can mechanisms lead would be favored in nature.mutation-order The lation.

Step (iii) has beenoccurs challenging undermutation-order both speciation if divergence of mate preferences or to the of speciation natural speciation.

Divergence any- reason, spe evolution ofbyany type selection of reproductive gamete recognition occurs by the fixation of bottom-up approach involves (i) genetic map- approaches but is needed to understand how sealternative advantageous mutations in different ping of reproductive isolation between closely lection has led to reproductive isolation.

6 FEBRUARY 2009 VOL 323 SCIENCE www.sciencemag.org 738 populations, as by sexual conflict (16). Diver- related species, (ii) testing whether discovered gence in song and other learned components genes exhibit a genomic signature of positive Ecological Speciation of behavior under purely social selection, not selection, and (iii) identifying the phenotype Evidence for ecological speciation has accumumolded by selection for efficient signal trans- and source of fitness effects of alternative alleles lated from top-down studies of adaptation and mission (5), is the cultural equivalent of the at selected loci. The approach has been hugely reproductive isolation [reviewed in (2, 8, 9)]. mutation-order process. Additional scenarios successful in identifying major genes implicated We now know of many real species that have, in hybrid inviability (Hmr, Lhr, Nup96), sterility at least in part, evolved by divergent natural seare elaborated in (5).

27

Number of studies

y to mate if they are of nents of reproductive isolation lacking identifiable less of relatedness as causes (Fig. 2). The unidentified component of speciation, if built by selection and not genetic affinity. is also prematComponents – divergent selection plants (28), Littorina marine snail ecotypes n which rentially inhabiting different zones of the intertidal 15 basis of (24), and mosquito fish inhabiting blue e under 10 holes with and without fish predators in the nt selecBahamas (29) (Fig. 1A). In these studies, it ortative 5 was shown that males and females are more insects, likely to mate if they are of the same ecoin fish, 0 or preftype, regardless of relatedness as indicated notypic by phylogenetic affinity.

Components – unknown cause flowerEcological speciation is also supported 15 be under by examples of premating reproductive environ10 isolation in which individuals choose or 30, 31)]. preferentially encounter mates on the basis vergent 5 directly of phenotypic traits that are under ecologiuced fitcally based divergent selection. Examples 0 he enviinclude assortative mating by host choice -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 migrant in insects, body size and coloration in fish, 31, 32)] Cumulative reproductive isolation beak size in birds, pollinator preferences for ybrids in of of the the magnitude of reproductive isolation Fig. 2.2. Estimates Estimates magnitude of reproductive [extrin- Fig. specific phenotypic floral traits, and variaresulting from divergent selection components (top), compared isolation resulting from divergent selection components 3)]. For tion in flowering time—traits inferred to be with othercompared componentswith lacking identifiable causes (bottom). (top), other components lacking l perenDivergent selection components includeDivergent those attributable to under divergent selection between environidentifiable causes (bottom). selection s of the active selection on traits (immigrant inviability and extrinsic components include those attributable to active ments [see examples in (8, 30, 31)]. uttatus) postzygotic and(immigrant to trait-basedinviability assortative and mating (habitat selectionisolation) on traits extrinsic Ecologically based divergent selection North preference, floralisolation) isolation, and breeding time). The assortative unattributed postzygotic and to trait-based s when components include intrinsic hybrid inviability, sexual selection has also been directly measured, as shown mating (habitat preference, floral isolation, and breeding of the against by reduced fitness of each ecotype in the enpollen competition, and reduced hybrid fecundity. time).hybrids, The unattributed components include intrinsic mple of Data were taken from (32, 31) (table S1). A negative value vironment of the other [immigrant inviabilhybrid inviability, sexual selection against hybrids, pollen ypic dif- indicates that hybrids had higher fitness than the parental species competition, and reduced hybrid fecundity. Data were ity; reviewed in (31, 32)] and by reduced fitutes di- for at least one component of postzygotic isolation. One data value taken from (32, 31) (table S1). A negative value indicates ness of hybrids in the parental environments because of –2.66 was left out of the bottom panel. that hybrids had higher fitness than the parental species [extrinsic postzygotic isolation (33)]. For for at least one component of postzygotic isolation. One example, each of the coastal perennial and value of –2.66 was left out of the bottom panel.

739 VOL 323 data 6 FEBRUARY 2009

lection between environments. The connections between selection on ordinary phenotypic traits and reproductive isolation are often strong and straightforward. It follows that much of the genetic basis of reproductive isolation should involve ordinary genes that underlie differences in phenotypic traits. But we still know little about the genetics of ecological speciation.

One line of evidence comes from tests of parallel speciation, whereby greater reproductive isolation repeatedly evolves between independent populations adapting to contrasting environments than between independent populations adapting to similar environments (20, 23). A major challenge in applying the test to natural populations is to eliminate the possibility that each ecotype has originated just once and has spread to multiple locales. This is difficult because gene flow of neutral markers between closely related but nearby populations can result in the false appearance of multiple independent origins of these populations when evaluated by phylogenies (3, 24). However, multiple origins are supported in several examples of parallel speciation, including the sympatric benthiclimnetic species pairs of threespine stickleback in young lakes of British Columbia (25, 26), the repeated origin of divergent marine and stream populations of threespine stickleback around the Northern Hemisphere (27), ecotypes of Timema walking stick insects living on different host

28

inland annual races of the monkey flower (Mimulus guttatus) along the west coast of North America has low fitness when transplanted to the habitat of the other (31). This is an example of active selection on phenotypic differences, and it also constitutes direct reproductive isolation because it is an evolved barrier to gene flow between parental populations. Multiple traits are probably involved, including flowering time and tolerance of salt and drought. This type of reproductive isolation is context-dependent and is weakened in intermediate environments.

On the other hand, active selection favors the evolution of ever-greater differences between populations, which may strengthen the barrier to gene flow (20).

It is unclear how much reproductive isolation typically evolves by ecologically based divergent selection in nature. We can approximate an answer from estimates of the combined contribution of active selection on traits and trait-based assortative mating, as compared with other forms of reproductive isolation (Fig. 2 and table S1). These estimates are incomplete because individual studies may lack data on components of reproductive isolation, separate components may not be independent, and the strength of barriers between species may not be symmetric (34). Nevertheless, compilation of the data shows that the amount of reproductive isolation attributable to active selection and traitbased assortative mating is at least as strong, on average, as the amount from components

of reproductive isolation lacking identifiable causes (Fig. 2). The unidentified component of speciation, if built by selection and not genetic drift, could be the result of either ecological or mutation-order mechanisms.

These examples indicate a growing knowledge of the mechanisms of selection and its consequences for reproductive isolation. At this point, the most glaring deficiency is our knowledge of the impact of selection on genes.

Optimistically, progress is being made with genetic mapping to identify quantitative trait loci (QTLs) and genes or regulatory control regions that affect individual phenotypic traits on which components of reproductive isolation depend.

Examples include the yup QTL, which affects flower color differences between the monkey flowers, Mimulus cardinalis and M. lewisii (35).

Swapping alleles of this QTL between the species with repeated backcrossing resulted in shifts in pollinator preference and, hence, indirectly affected premating isolation. Survival and salt tolerance of second-generation hybrids between the sunflowers Helianthus annuus and H. petiolaris transplanted to the salt marsh habitat of their hybrid descendent species (H. paradoxus) mapped strongly to a QTL identified as the salt tolerance gene CDPK3 (36).

Another form of investigation involves the analysis of genome scans of ecologically different populations and species. These scans compare allelic variation within and between populations at many marker loci spaced throughout the genome (37). Markers that show excessive differentiation between populations (outliers) may indicate selection on nearby genes. The method is particularly informative when applied to populations with ongoing hybridization, because outlier markers may identify points in the genome that resist the homogenizing influence of gene flow, perhaps indicating genomic regions under divergent selection. However, sets of genes that diverged under a mutation-order process can produce the same pattern (17, 18), which makes analysis of such studies more difficult. Clues to whether ecologically based divergent selection is involved are gained if outliers at the same genomic locations turn up repeatedly in scans between populations that inhabit contrasting environments (38) and by identifying phenotypic traits under divergent selection that map to those locations in the genome (36, 39, 40). As genomic resources increase for more species, it will be possible to measure natural selection directly on genomic regions of interest by transplanting otherwise relatively homogenous experimental populations containing alternative alleles into the environments of the parent species (35).

Mutation-Order Speciation Mounting evidence for divergent selection in speciation does not diminish the potential role

of mutation-order divergence. It may be that the mutation-order process is more difficult to detect, or perhaps we have not looked hard enough at species with only small ecological differences (5). We still do not know much about the selective factors causing mutation-order speciation.

Evidence for mutation-order speciation comes from instances in which reproductive isolation apparently evolved as a by-product of conflict resolution between genetic elements within individuals (intragenomic conflict), such as cytoplasmic male sterility in hermaphroditic plants (Fig. 1B), and genetic elements conferring meiotic drive. Under both mechanisms, a mutation arises that can distort representation in gametes and spreads in a selfish manner, even though such an element reduces overall fitness of the organism that bears it. This, in turn, places selection on mutations in other genes that counter the selfish element’s effects and restore more equal genetic representation in gametes.

Distorter and restorer mutations are unlikely to be the same in different populations regardless of environment; thus the process leads to divergence. The mismatch between the distorter in one population and the restorer in the other can result in hybrid sterility or inviability and, thus, reproductive isolation (3, 41). Numerous examples of selfish elements, such as those observed in cytoplasmic male sterility of plants, support these hypotheses (42, 43). In addition, partial reproductive isolation generated by meiotic drive has been identified in Drosophila [reviewed in (3, 41)]. Sexual conflict is also expected to lead to mutation-order speciation, but there are few compelling examples (3). The contribution by these mechanisms to speciation is still uncertain, however. The alleles responsible for meiotic drive and cytoplasmic male sterility may be prevented from spreading to fixation because selection on such elements is frequencydependent (43) and because restorer alleles arise and weaken selection on the distorter elements (44). Second, if divergent populations come into secondary contact, the alleles within each population causing cytoplasmic male sterility or meiotic drive (and the corresponding restorer alleles) will spread between the populations by gene flow, eliminating that component of reproductive isolation (43). Hence, for these mechanisms to contribute to speciation, the fitness of hybrids must be reduced to very low levels, or other incompatibilities must arise that interact with these genes to prevent their spread after secondary contact.

Conclusions Our understanding of the role of natural selection in speciation has come a long way since Darwin’s time. If he were here to witness, he would most likely be staggered by the discoveries of genes and molecular evolution and astonished at the prospect that evolutionary conflict

between genes could generate reproductive isolation (45). Mostly, I expect that he would be chuffed by mounting evidence for the role of natural selection on phenotypic traits in the origin of species. This is really what On the Origin of Species was all about. Between 1859 and the present, the general acceptance of the biological species concept altered the focus of speciation studies. Yet, the discovery that reproductive isolation can be brought about by ecological adaptation in ordinary phenotypic traits bridges Darwin’s science of speciation and our own.

The most obvious shortcoming of our current understanding of speciation is that the threads connecting genes and selection are still few. We have many cases of ecological selection generating reproductive isolation with little knowledge of the genetic changes that allow it. We have strong signatures of positive selection at genes for reproductive isolation without enough knowledge of the mechanisms of selection behind them. But we hardly have time to complain. So many new model systems for speciation are being developed that the filling of major gaps is imminent. By the time we reach the bicentennial of the greatest book ever written, I expect that we will have that much more to celebrate.

References and Notes

1. C. Darwin, On the Origin of Species by Means of Natural Selection (J. Murray, London, 1859).

2. D. Schluter, The Ecology of Adaptive Radiation (Oxford Univ. Press, Oxford, 2000).

3. J. A. Coyne, H. A. Orr, Speciation (Sinauer Sunderland, MA, 2004).

4. P. R. Grant, B. R. Grant, How and Why Species Multiply: The Radiation of Darwin’s Finches (Princeton Univ. Press, Princeton, NJ, 2008).

5. T. Price, Speciation in Birds (Roberts & Company, Greenwood Village, CO, 2008).

6. T. G. Dobzhansky, Genetics and the Origin of Species (Columbia Univ. Press, New York, 1937).

7. E. Mayr, Systematics and the Origin of Species (Columbia Univ. Press, New York, 1942).

8. D. Schluter, Trends Ecol. Evol. 16, 372 (2001).

9. H. D. Rundle, P. Nosil, Ecol. Lett. 8, 336 (2005).

10. G. S. Mani, B. C. Clarke, Proc. R. Soc. London Ser. B Biol. Sci. 240, 29 (1990).

11. M. Turelli, N. H. Barton, J. A. Coyne, Trends Ecol. Evol. 16, 330 (2001).

12. C. L. Hubbs, Am. Nat. 74, 198 (1940).

13. T. Dobzhansky, Am. Nat. 74, 312 (1940).

14. Additional references are available as supporting material on Science Online.

15. J. A. Endler, Am. Nat. 139, S125 (1992).

16. W. R. Rice et al., Proc. Natl. Acad. Sci. U.S.A. 102, 6527 (2005).

17. N. Barton, B. O. Bengtsson, Heredity 57, 357 (1986).

18. A. S. Kondrashov, Evolution 57, 151 (2003).

19. L. H. Rieseberg, personal communication (2008).

20. W. R. Rice, E. E. Hostert, Evolution 47, 1637 (1993).

21. J. R. Dettman, C. Sirjusingh, L. M. Kohn, J. B. Anderson, Nature 447, 585 (2007).

22. Z. D. Blount, C. Z. Borland, R. E. Lenski, Proc. Natl. Acad. Sci. U.S.A. 105, 7899 (2008).

23. D. Schluter, L. M. Nagel, Am. Nat. 146, 292 (1995).

24. R. K. Butlin, J. Galindo, J. W. Grahame, Philos. Trans. R. Soc. London Ser. B Biol. Sci. 363, 2997 (2008).

25. H. D. Rundle, L. Nagel, J. W. Boughman, D. Schluter, Science 287, 306 (2000).

26. E. B. Taylor, J. D. McPhail, Proc. R. Soc. London Ser. B Biol. Sci. 267, 2375 (2000).

27. J. S. McKinnon et al., Nature 429, 294 (2004).

28. P. Nosil, S. P. Egan, D. J. Funk, Evolution 62, 316 (2008).

29. R. B. Langerhans, M. E. Gifford, E. O. Joseph, Evolution 61, 2056 (2007).

30. S. Via, Trends Ecol. Evol. 16, 381 (2001).

31. D. B. Lowry, J. L. Modliszewski, K. M. Wright, C. A. Wu, J. H. Willis, Philos. Trans.

Royal Soc. London Ser. B Biol. Sci. 363, 3009 (2008).

32. P. Nosil, T. H. Vines, D. J. Funk, Evolution 59, 705 (2005).

33. H. D. Rundle, M. C. Whitlock, Evolution 55, 198 (2001).

34. N. H. Martin, J. H. Willis, Evolution 61, 68 (2007).

35. H. D. Bradshaw, D. W. Schemske, Nature 426, 176 (2003).

36. C. Lexer, Z. Lai, L. H. Rieseberg, New Phytol.

161, 225 (2004).

37. M. A. Beaumont, Trends Ecol. Evol. 20, 435 (2005).

38. P. Nosil, D. J. Funk, D. Ortiz-Barrientos, Mol. Ecol. 18, 375 (2009).

39. S. M. Rogers, L. Bernatchez, Mol. Biol. Evol.

24, 1423 (2007).

40. S. Via, J. West, Mol. Ecol. 17, 4334 (2008).

41. H. A. Orr, J. P. Masly, N. Phadnis, J. Hered.

98, 103 (2007).

42. D. A. Levin, Syst. Bot. 28, 5 (2003).

43. L. H. Rieseberg, J. H. Willis, Science 317, 910 (2007).

44. D. C. Presgraves, Trends Genet. 24, 336 (2008).

45. D. C. Presgraves, Curr. Biol. 17, R125 (2007).

46. L. Fishman, J. H. Willis, Evolution 60, 1372 (2006).

47. A. L. Case, J. H. Willis, Evolution 62, 1026 (2008).

48. We thank M. Arnegard, R. Barrett, A. Case, H. Hoekstra, M. Noor, P. Nosil, S. Otto, T. Price, L. Rieseberg, S. Rogers, A. Schluter, S. Via, M. Whitlock, J. Willis, and a reviewer for assistance and comments. This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada and the Canada Foundation for Innovation.

Supporting Online Material

www.sciencemag.org/content/full/323/5915/737/ DC1 Tables S1 to S3 References 10.1126/science.1160006

29

The Bacterial Species Challenge: Making Sense of Genetic and Ecological Diversity

T T

T

30

www.sciencemag.org

SCIENCE

VOL 323

6 FEBRUARY 2009

741

lg in o

lyt icu s

V. ordalii

oi en

si s

B

S. pseudopneumoniae S. pneumoniae 3% divergent 1.2% divergent

um

A

V. r

The Bacterial Species Challenge: Making Sense of Genetic and Ecological Diversity

Speciation

V. a

values that place two bacterial isolates into the same or different species. The overall genetic SPECIALSECTION relatedness mayhuman be measured by the pneumoniaeof isisolates a major pathogen, S. extent of DNA hybridization between them, and mitis is a commensal bacteria with a history S. Rogers, A. Schluter, Via,more M. Whitlock, J. Willis, and 40. S. Via, J. West, Mol. Ecol. 17, 4334 (2008).

32. P. Nosil, T. H. Vines, D. J. Funk, Evolution 59, 705 (2005). those that show 70%S. or DNA hybridpseudoof taxonomic and S. work a reviewer foruncertainty assistance and (11), comments.

This 41. H. A. Orr, J. P. Masly, N. Phadnis, J. Hered. 98, 103 33. H. D. Rundle, M. C. Whitlock, Evolution 55, 198 (2001). ization are defined as the same species (2, 10). pneumoniae is abyrecently described of was supported grants from the Naturalorganism Sciences and (2007).

34. N. H. Martin, J. H. Willis, Evolution 61, 68 (2007).

Such cutoffs imply that sequences thatthecluster Engineering Research Council of Canada and Canada 42. D. A. Levin, Syst. Bot. 28, 5 (2003).

35. H. D. Bradshaw, D. W. Schemske, Nature 426, 176 (2003). status that nonetheless corresponds to uncertain Foundation foraInnovation.

43. L. H. Rieseberg, J. H. Willis, Science 317, 910 (2007).

36. C. Lexer, Z. Lai, L. H. Rieseberg, New Phytol. 161, 225 together with certain of There similarity data (12). are a distinct cluster in theseamount 44. D. C. Presgraves, Trends Genet. 24, 336 (2008).

(2004). must be differences from the Material same species, and Supporting Online of moreover sequence striking in the amount 45. D. C. Presgraves, Curr. Biol. 17, R125 (2007).

37. M. A. Beaumont, Trends Ecol. Evol. 20, 435 (2005). that this cutoff value is applicable to all groups www.sciencemag.org/cgi/content/full/323/5915/737/DC1 diversity observed within homologous house46.2L. Fishman, J. H. Willis,1 Evolution 60, 1372 (2006).

38. P. Nosil, D. J. Funk,1D. Ortiz-Barrientos, Mol. Ecol. 18, 2,3,4 1 Tables S1 to S3or archaea. Recent MLSA studies, of bacteria Christophe Fraser, * Eric J. Alm, Martin F. Polz, William Hanage 47. A.Brian L. Case,G.J. Spratt, H. Willis, Evolution 62,P. 1026 (2008).

375 (2009). keeping References genes in these named species, ranging which use the concatenated sequences of mul48. We thank M. Arnegard, R. Barrett, A. Case, H. Hoekstra, 39. S. M. Rogers, L. Bernatchez, Mol. Biol. Evol. 24, 1423 from 1.2% for S. pneumoniae to 3.0% for S. 10.1126/science.1160006 M. Noor, P. Nosil, S. Otto, of T. Price, Rieseberg, housekeeping genes to discern clustering The (2007).

Bacteria and Archaea are the most genetically diverse superkingdoms life, L.and techniques tiple pseudopneumoniae and up to 5.0% for S. mitis. patterns among populations of closely related for exploring that diversity are only just becoming widespread. Taxonomists classify these The distance between two randomly selected organisms into species in much the same way as they classify eukaryotes, but differences in their taxa, suggest that species defined by taxonois similar to the average S. mitisthat genotypes values place two bacterial isolates into disthe mists in many cases correspond to well-resolved biology—including horizontal gene transfer between distantly related taxa and variable rates of REVIEW S. pseudotance between S. pneumoniae and same or different overall genetic clusters. species.

However,The these studies also homologous recombination—mean that we still do not understand what a bacterial species is. This sequence pneumoniae (5.1%) (2). This implies relatedness ofgenotypes isolates may be cutoff measured by the show that there is no universal or descripis not merely a semantic question; evolutionary theory should be able to explain why species theclusters of that ahybridization fixed level of sequence diverthat extent ofuse DNA between them, and tor of characterizes a species.

Furexist at all levels of the tree of life, and we need to be able to define species for practical species would gence for differentiating those that show 70% or more DNA hybridthermore, inspection of the clusters does tend not applications in industry, agriculture, and medicine. Recent studies have emphasized the need to to either S. pneumoniae and pseudoization arerejoin defined aswhich the same (2, 10). always clearly reveal level species in theS.hierarchy combine genetic diversity and distinct ecology in an attempt to define species in a coherent and or breakthan up sequences S. mitis that 1) nearly pneumoniae, Such imply that that cluster is morecutoffs fundamental any otherso(Fig. (7). convincing fashion. The resulting data may help to discriminate among the many theories of of its own.

This is every isolate was a species together with a certain amount of similarity As an example, Fig. 1A shows the relationprokaryotic species that have been produced to date. clearly unsatisfactory. must be frommultiple the sameisolates species,of and ships among threemoreover closely that this streptococcal cutoff value is species. applicableStreptococcus to all groups he species debate in microbiology is not reflect only a tiny subset of those characters that related 1 2 traordinary variety of 1different microbial life, muchin1ofthe it pneumoniae he species isbacnotF. Polz, Habitats Differentiation of bacteriaand or archaea.

Recent MLSA S. studies, Christophe Fraser, * Ericindesire J.microbiology Alm,to2,3,4 Martin Brian G. Spratt, Williamresources P. Hanage is aEcological major human pathogen, mitis allow bacteria to use only about adebate human catalog desiremanner, to catalog Beyond this, taxa too fraction similar of to is A natural criterion to identify clusters of uncultured (1). only diversity about a inhuman which use the concatenated sequences of taxomulaclear commensal bacteria with a history of and only capture a small terial a consistent but environment, rRNA se- nomic importance, we might want in consistent manbe distinguished and evolutionary tiple housekeeping genes topseudopneumoniae discern clustering The Bacteria anddiversity Archaea areabecause the mostofgenetically diverse superkingdoms life, andbytechniques uncertainty (11), andwhich S. true diversity in circumscribed this ofsuperkingdom of life. is also abacterial fundamental argument what it the patterns among of uncertain closely related butabout is alsoour a fundamental argument ecological features that quences have Taxonomists revealed further diversity through is to species, is populations to find ner, for exploring that diversity of are onlyevolutionary justbecause becomingMore widespread. classify these acall recently described organism status recently, molecular methods [particularly reveals ignorance how taxa,nonetheless suggestthem that species defined bycluster taxonoabout our ignorance of way how analysis (MLSA) and that Among of whatform, it into reveals distinguish from close multilocus organisms species much thebacterial same classifysequence eukaryotes, butand differences inRNA their corresponds to arelatives. distinct in DNA-DNA hybridization ribosomal(2) forces shape, and in extinguish ge-as they mists data in many correspond biology—including horizontal gene between distantly related taxa(1), andand variable of andoftransfer extinguish diversity evolutionary forces form, shape, metagenomic studies pathogens, thecases ability causeto differences awell-resolved distinctive these (12).

There aretostriking in (rRNA) sequencing] have helped to this definerates species, netic lineages, of the mechanisms differensequence clusters.

However, studies also homologous recombination—mean thatcommon we still not these understand what a by bacterial species This of dobut has beenthese used to within define bacterial genetic lineages, of thesharing mechanisms needs to be explained theory.limitations Thus, is. practidisease amount ofhistorically sequence diversity observed methods have serious and the tiation between subpopulations show thatbut there is no universal ora descripis not merely semantic question; evolutionary should beassign able of to explain whyobservations species lack and constitute minute differentiation subpopulations sharing species, pathogens cal difficulties, housekeeping genescutoff inonly these named cannot reliably a theory, large collection of similar homologous descent, and ofabetween the process of adaptation to new theory tor of clusters that characterizes a species.

Furexist at all levels of the tree of life, and we need to be able to define species for practical species, ranging from 1.2% for S. pneumoniae to andenvironments. of the process of adaptaas yet unclassified microbial of of vast to amounts common speciesof(e.g., rRNA sequences are too fraction of overall bacterial diversity. Mapping niches anddescent, changing Animal spe- strains thermore, inspection the clusters does applications in industry, agriculture, andand medicine.

Recent studies emphasized therRNA need to for S. pseudopneumoniae and up toresources 5.0% not for to resolve similar species). se- 3.0% cies are defined by and theirchanging morphological be- conserved environments. of how tion to new niches diversity have allhave fueled the controversy bacterial diversity ontoofenvironmental always reveal which level in randomly theofhierarchy combine traits genetic diversity ecology to define species a coherent and S. mitis.clearly The between two sequence surveys have, however, the extrahavioral andare by defined their and ability or inability to in an indicates thatdistance closely related groups bacteria Animal species bydistinct their morphooneattempt defines a bacterial speciesinrevealed (3–8). is more fundamental than any other (Fig. 1)fine(7). convincing fashion.

The resulting data may help to discriminate among the many theories of lected S. mitis genotypes is similar to the average ordinary variety of microbial life, much of it interbreed, but such categories cannot easily be can be ecologically divergent. For example, logical and behavioral traits and by their abilAsresource anbetween example, 1A has shows relationprokaryotic species thatorhave been produced to date.

S. Fig. pneumoniae andthe S.observed pseudouncultured (1). Beyond this, taxa too similar to be distance applied to the Bacteria Archaea (or indeed to Genetic Clustering but such categories been ity or inability to interbreed, scale partitioning ships among multiple of three closely genotypes (5.1%) (2).

This implies and circumscribed rRNA se- pneumoniae many taxonomists or Ar- distinguished Darwin commented that “all trueby classification cannoteukaryotic easily be microbes). applied toInstead, the Bacteria among coastal Vibrio isolates populations coexisting related streptococcal species.

Streptococcus reflect only a tiny subset of those characters that he species debate in microbiology is not that the use of a fixed level of sequence divergence quences have revealed further diversity through have been forced to rely on biochemical tests and chaea (or indeed to many eukaryotic microbes). is genealogical” [(9), p. 404]. Taxonomists have in the water column (13). Partitioning was pneumoniae is a major human pathogen, mitis allowused bacteria to userelatedness different(MLSA) resources in and the for about a human desire to catalog bacdifferentiating species would tend toS.either sequence analysis limited morphological characteristics for this taxonomists have been forced to purrely multilocus to define(2)cutoff were collected from thus sequence discovered because strains Instead,only is a commensal bacteria with a history of taxoenvironment, and only capture a small fraction of terial diversity in a consistent manner, but rejoin S. pneumoniae and S. pseudopneumoniae, metagenomic studies (1), and this diversity needs pose.

Naturally, biochemical characters have been on biochemical tests and limited morphological values that place two bacterial isolates into the distinct, ecologically informative samples, and nomic uncertainty andnearly S.ofpseudopneumoniae thebetrue diversity this superkingdom of life. is also a fundamental argument of what it to break up S. mitis(11), so that every isolate was explained theory.

Thus,overall practical dif- or selected for the convenience of because taxonomists; they biocharacteristics for this purpose.

Naturally, same or differentbyin species.

The genetic the phylogenetic structure the ecologically isspecies a recently described organism of uncertain status molecular [particularly reveals about our ignorance of how evolutionary More recently, of its own.

This is clearly unsatisfactory. lack theory,may andmethods observations ofofisolates be measured of byvast the adifferentiated populations was superimposed chemical characters have been selected for the ficulties, relatedness that nonetheless corresponds to a distinct cluster in 1forces form, shape, and extinguish bacterial geDNA-DNA hybridization and ribosomal RNA amounts of as yet unclassified microbial diversity Department of of Infectious Disease Epidemiology, taxonomists; they reflect Imperial only a extent of DNA hybridization between them, and on their habitats. Habitats were defined using convenience 2 Habitats these dataand (12).Ecological There are Differentiation striking differences in (rRNA) sequencing] have helped to define species, netic lineages, of the mechanisms of differenCollege London, London W2 1PG, UK. Department of Civil have all fueled the controversy of how one detiny subset of those characters that allow bac- those that show 70% or more DNA hybridiza- an empirical modeling approach. This analysis and Environmental Engineering, Massachusetts Institute of theclear amount of sequence diversity observed within but these methods have(3–8). serious limitations and A tiation between subpopulations sharing common natural criterion to identify clusters of fines a bacterial species defined as the same species (2, 10). Such revealed high levels of specialization for some teria to use differentMAresources in 3the environtion are Department of Technology, Cambridge, 02139, USA. homologous housekeeping genes we in these named cannot reliably assign a large collection of similar evolutionary descent, and of the process of adaptation to new importance, which might want only capture a smallInstitute fraction of the cutoffs imply that sequences that cluster together populations (e.g., V. ordalii is only found as ment, and Biological Engineering, Massachusetts of TechnolGenetic species, rangingisfrom 1.2% for S. pneumoniae to strains toClustering species (e.g., rRNA sequences are too to niches and changing environments.

Animalofspecall species, to find ecological features that MIT ogy, MAthis 02139, USA. 4Broad Institute superkingdom of life. More true Cambridge, diversity in with a certain amount of similarity must be from single free-swimming cells), whereas others are 3.0% for S. pseudopneumoniae and up to 5.0% for conserved to resolve that similar rRNA se- distinguish ciesHarvard are defined by their morphological and be- Darwin them from close relatives.

Among commented “allspecies). true classification and University, Cambridge, MA 02139, USA. DNAa wide recently, molecular methods [particularly the same species, and moreover that this more generalist (Fig. 1B) and can colonize S. mitis.

The randomly sequence surveys have, revealed thecutoff extrahavioral traits and by their ability or inability to is thedistance ability between to cause two a distinctive disgenealogical” [(9), however, p. 404]. Taxonomists have pathogens, *To whom correspondence should be addressed.

E-mail: DNA hybridization ribosomal RNAeasily (rRNA) value is applicable tomicrobial all groups of much bacteria or variety ofmitis surfaces, including organic particles lected S. genotypes is similar to the average ordinary variety of life, of it interbreed, but suchand categories cannot be thus c.fraser@imperial.ac.uk used sequence relatedness to define cutoff ease has historically been used to define species, helped define (or species, Recent MLSAthis, studies, which usetothe zooplankton water column Most sequencing] distance between inS.the pneumoniae and (13).

S. pseudouncultured (1). Beyond taxa too similar be and applied to thehave Bacteria or to Archaea indeedbut to archaea. and can- concatenated of multiple Vibrio (5.1%) populations are implies deeply these have serious Instead, limitations the predicted pneumoniae genotypes (2). This distinguished sequences and circumscribed byhousekeeprRNA se- of many methods eukaryotic microbes). taxonomists among cases 741 not assign large collection www.sciencemag.org oftests similar geneshave to SCIENCE discern patterns divergent eachlevel other, and in many VOLdiversity 323 6through FEBRUARY 2009 that the usefrom of a fixed of sequence divergence quences revealedclustering further havereliably been forced to arely on biochemical and ing to species (e.g., rRNA sequences are too populations of closely related taxa, suggest that are congruent with named species; however, V. strains limited morphological characteristics for this pur- multilocus sequence analysis (MLSA) (2) and for differentiating species would tend to either similarcharacters species). have rRNAbeen se- species defined by taxonomists many needs cases splendidus is a notableand exception and splits into conserved to resolve rejoin S. pneumoniae S. pseudopneumoniae, metagenomic studies (1), and this in diversity pose. Naturally, biochemical ex- correspond sequence clusters. related groups with distinct quence have, however, revealed thethey to well-resolved or break up closely S. mitis so that nearly every isolate was to be explained by theory. Thus, practical dif- numerous selectedsurveys for the convenience of taxonomists; is ecological indicating However, these studies and alsoobservations show that there presumably a species of preferences, its own. This is clearly unsatisfactory. ficulties, lack of theory, of vast no universal or descriptor of clusters that recent ecological radiation from a sympatric 1 amounts of ascutoff yet unclassified microbial diversity Department of Infectious Disease Epidemiology, Imperial Habitats population and Ecological a species.

Furthermore, inspection (13). Differentiation Thus, genetic clusters characterizes College London, London W2 1PG, UK. 2Department of Civil have all fueled the controversy of how one de- ancestral and Environmental Engineering, Massachusetts Institute of always clearly reveal ecology can be discerned. of the clusters does not that correlate with A clear natural criterion to identify clusters of fines a bacterial species (3–8).

Technology, Cambridge, MA 02139, USA. 3Department of What do the genetic data tell us about mechwhich level in the hierarchy is more fundamen- evolutionary importance, which we might want Biological Engineering, Massachusetts Institute of TechnolGenetic Clustering than any other (Fig. 1) (7). and that the anisms of population tal to call species, is to finddifferentiation ecological features ogy, Cambridge, MA 02139, USA. 4Broad Institute of MIT As ancommented example, Fig. the relation- evolutionary queshistory the microbes distinguish them fromof close relatives. inAmong Darwin that1A “allshows true classification and Harvard University, Cambridge, MA 02139, USA. pathogens, abilityare to organized cause a distinctive disis genealogical” [(9), p.isolates 404]. Taxonomists have tion? of three closely ships among multiple That the bacteria into genetic *To whom correspondence should be addressed. E-mail: c.fraser@imperial.ac.uk ease hasishistorically been usedinteresting to define species, thus usedstreptococcal sequence relatedness define cutoff clusters not, per se, a very obserrelated species. toStreptococcus REVIEW

us

orvegic

vibrio n Entero nsis

V. calvie S. mitis 5% divergent

r ua

s i nu ge i/lo r e ch

ia

st ae is V. V. f eri ch fis .

V

s

teu

rs pe

u

s V.

a*

cid

e an

p V.

S. oralis 0.005

Vibrio splendidus

0.005

Fig. 1. Multilocus sequence analysis of closely related species. (A) Radial minimum evolution tree constructed using MEGA4, showing clusters among 97 isolates of four Streptococcus species identified as indicated. The tree was built using concatenates of six housekeeping loci, resulting in a total of 2751 positions in the final data set (2). Distances were calculated as the percentage of variant nucleotide sites. The mean distance within the clusters, calculated by MEGA4, is shown. To the right, the pneumococcal cluster is shown at larger scale, and putative subclusters are indicated in dark gray, purple, and green. (B) Ecological

associations of Vibrionaceae sequence clusters (13). Habitats (colored dots) were estimated as differential distributions of groups of closely related strains among samples (size fractions enriched in different environmental resources). Clusters associated with named species are evident, and in most cases species show a clear predilection for one of the habitats. The exception is V. splendidus, which breaks up into many closely related ecological populations. Asterisk denotes that trees based on additional loci indicate that the placement of V. panecida within V. splendidus may be an artifact of horizontal gene transfer at the Hsp60 locus.

not, per se, The a veryfirst interesting observation; many or makes strict thereference accumulation of neutral diversity.

The but pathogens only a minute fraction of diversity. of a population proposed mechanism was unifying biological to the vation; many constitute or most models most models of a population first proposed wasniche basedpartitioning, on artificial overall bacterial of mutation bacterial based selection reproducing experimentswith witha principles reproducing withdiversity. a small Mapping amount of on artificial of mechanism selection and amount of mutation will eventually selection experiments with bacteria grownasfora diversity onto environmental resourcesconsisting indicates small will eventually produce populations under the ecotype has rightly become popular bacteria grown for extended periodsproduce populations consisting of clusters of related orga- framework extended periods under stable in chethat closely of related groups of bacteria can be ecorelated organisms, irrespective of stable which to conditions discuss bacterial of clusters conditions in chemostats, which showed within nisms, irrespective the details of the the evolutionmostats, which showedand repeated selective sweeps logically For example, fine-scale re- repeated forces or ecologiselective of sweeps in which whole evolution, the detailsdivergent. of the evolutionary speciation, ecology. ary forceswas or ecological more in The which the whole waspredicts thoughtthat to source partitioningAhas been observedobservaamong genome fixation ecotype modelgenome (4, 16, 24) cal differentiation. more substantial thought todifferentiation. hitchhike to A is thatmutation there is very little common hitchhike ancestry to fixation along with an advantageous coastal Vibrio populations coexisting in the water substantial there is very little neutral diversity will be preserved among bacalong with observation an advantageous (periodic tion is that diversity many populations microbes, mutation (periodic selection) (22). Selective column Partitioning discovered of was microbes, frombecause which neutral (22). inSelective sweeps ofcan purge terial (which sweeps should in many(13). populations populations within niches selection) from which we maydiversity infer some features of the be canmonophyletic), purge almost alland genetic the popstrains were collected distinct, ecologically in the population we may infer some from features of the selective almost all genetic thusdiversity predictsinthat ecolandscape.a candidate Neutral diversity is the ulationare andcoherent thus constitute a candidate informative samples,diversity and the phylogenetic struclandscape. Neutral is the amount of selective for types self-contained genemechapools. and thus constitute mechanism polymorphism nisma result, for reducing variation that (23).ecotypes ture of the ecologically differentiated populations in noncoding re- amount been suggested polymorphism that is evident reducingofneutral variationthat (23).is evident in non- As it has neutral was their habitats. substitutions.

Habitats were coding regions or results in synonymous sub- should be considered as putative or actual spegionssuperimposed or results inonsynonymous Niches and Ecotypes stitutions. common measure of neutral cies, defined using measure an empirical modeling approach. andOne Ecotypes One common of neutral diversity is the Niches depending on the level of genetic differdiversity is the effective population size N To extendfrom this model, one can population. consider multiple This analysis revealed high levels of specialization , dee the ancestral This effective population size Ne, defined as the size To extend this model, one can consider multiple entiation ecological niches characterized by of theproviding selective for some populations (e.g., V. ordalii is only found fined as the size of a population evolving in the of a population evolving in the absence of se- ecological niches characterized by the selective model therefore has the advantage theyunderstanding confer to specific genes.

This is as singlethat free-swimming cells), others are of selection that would generate as much genes.

This aadvantages lection would generate aswhereas much neutral di- absence advantages they confer to specific mechanistic of the evolutionmore generalist (Fig. 1B) and can colonize a wide neutral diversity as is actually observed. Esti- the ecotype model, where genes adapted to is the ecotype model, where genes adapted to ary processes, as well as an organizing principle versity as is actually observed. Estimates of N variety of surfaces, including5 organic9 particles ande mates of Ne for bacteria range from 105 to 109 specific niches cause selective sweeps within for bacteria range from 10 to 10 (14–18). To specific niches cause selective sweeps within for classifying species, that is based on experizooplankton in the water column (13). Most of the (14–18). To put this into context, the numbers of those niches but not in other niches. In this way put this into context, the numbers of Vibrio cells those niches but not in other niches. In this mental observations of bacterial populations. predicted Vibrio populations are deeply divergent Vibrio cells per cubic meter of seawater in tem- the population will undergo adaptation and difHowever, these observations of repeated per cubic meter of seawater in temperate coastal way the population will undergo8 adaptation from each other, and in 8many cases are congruent perate coastal regions range from 10 to 109 (19), ferentiation while maintaining relatively low levels regions range from 10 to 109 (19), which sug- and differentiation while maintaining relatively selective sweeps were made in chemostats, of neutral diversity, as selective sweeps confined with named species; however, V. splendidus is a which suggests vast census population sizes census and population sizes (>1020closely ). This low neutral diversity, as selective ungests vast natural environments arethe markedly to each ecotype regularly purge population notable exception splits into numerous (>1020levels ). Thisofobservation—a mismatch of many whereas mismatch of many orders of regularly observation—a sweeps confined to each ecotype stable and diverse.

How would one detect the related groups with distinct ecological preferences, orders of magnitude between effective population of any diversity that might have accumulated size purge sweeps in diversity natural bactemagnitude indicating between effective population the census population of any diversity selective (Fig. 2A). of Crucially, what neutral we do presumably recent ecological radiation size and population size (truethat of might most presence accumulated (Fig. 2A). originally Crucially,used what examples and census population size (true of (13). mostThus, bac- have populations?

Thetomost conclusive observe is predicted be associated with adaptfrom a sympatric ancestral population bacteria studied to date)—was to rial date)—was originally to neutral is argue predicted diversity we do observe not The fromability bacteria but from RNA viruses, teria studied ive traits. of such selective sweeps to genetic clusterstothat correlate with ecologyused can be counter claims of neutrality and instead that come The which mutate at much highersize rateshas than DNAcounter claims of neutrality and instead argue to associated limit the effective population been recdiscerned. all be genetic variationwith wasadaptive adaptive traits.

(20, 21).

21). ability been established from thatWhat all genetic wastell adaptive (20, of there such are selective limit the based It has(17, ognizedlife forforms. some time 23), and this model do thevariation genetic data us about mechHowever, several sweeps different to mechanisms population size has (Fig. been2).recognized sequences the However, are several different and mechacollected over many by years that and has been substantially developed Cohan anisms of there population differentiation the effective that can explain this mismatch explain thismicrobes mismatchin(Fig. 2). anddriving this model has population for Whatever some time (17, 23), are structure the human influenza nisms that can colleagues (4, 16, 24).ofBecause it links patternsviof evolutionary history of the question? mechanisms the differWhatever driving driven by repeated selecsubstantially Cohan is predominantly genetic differentiation with adaptation, and makes That bacteria aremechanisms organized intoare genetic clustersthe is been entiation of bacteria developed into clusters,bythey mustand re- rus differentiation of bacteria into clusters, they colleagues (4, 16, 24). Because it links patterns tive sweeps (25) and that the resulting effective must restrict the accumulation of neutral of genetic differentiation with adaptation, and population size Ne (<100) is very much smaller 742 6 FEBRUARY 2009 VOL 323 SCIENCE www.sciencemag.org

31

iples of ype has within ciation,

cts that ng bacould be cotypes As a reshould ies, dentiation l therehanistic sses, as ssifying bserva-

A

E2

E1

B new and empty

* * *

* *

colonization

* *

old and void

stable ecotype concept

C

D

del with ot niche bottlewhole le indill other induce ks will pulation ded into etween e popupatches acterial eria are ig. 2B)

Population size

20

ated sewhereas ble and ence of lations? ot from utate at orms. It ollected cture of y driven hat the 100) is acteria. genetic models karyotic alogies, iven by ngitudions, as shes in changes ological

colonization

Bacteria

Phage

15 10 5 0 0

20

40

60

80

100

Time

Fig. 2. Different models of microbial evolution that lead to low values of Ne. (A) The ecotype model of bacterial population differentiation. The tree shows a single bacterial lineage that differentiates into two sublineages (E1 and E2) that differ in some aspect of their ecology. Periodic selection (a selective sweep) occurs at the points marked by asterisks and eliminates almost all of the diversity that has arisen since the last episode of periodic selection, which is shown by the dashed branches (diversity purged by periodic selection) or solid branches (existing diversity) on the tree. As the two populations are ecologically distinct (i.e., ecotypes), periodic selection in one sublineage does not influence diversity in the other sublineage and vice versa. Each ecotype can therefore diverge to become separate species. Reproduced from (24) with permission. (B) A metapopulation. Patches of varying size (gray circles) are vacant (empty) or may be colonized by a single genotype randomly acquired from another patch. Strains may diversify within a patch (as shown by different colors representing distinct genotypes), which may colonize empty patches as described above. A characteristic of this sort of metapopulation is patch turnover, in which patches occasionally become unable to support colonization and their inhabitants are removed (solid gray circles). (C) A neutral model with small population size. Different genotypes (different colors) arise by mutation or recombination and increase or decrease in the population by random drift. For some purposes, this simple model is an adequate effective description of the more complex processes represented in (A), (B), and (D), and of other more complex evolutionary models not described in this review. (D) Predator-prey dynamics and population bottlenecks. Regular population bottlenecks can drastically shrink the effective population size. In this case, bacteria-phage predator-prey dynamics are simulated with a classical Lotka-Volterra model, which can generate oscillations in population size of any amplitude. Population sizes and time axes are in arbitrary units for illustrative purposes only.

Metapopulation structure, in which the population is divided into patches and where individuals disperse between patches, can generate very low effective population sizes if patches turn over (i.e., if patches are only intermittently able to support bacterial growth, and if a small number of bacteria are dispersed to colonize empty patches) (Fig. 2B) (27). This structure well describes the situation for parasites, which can colonize a host but are then forced to move on because the host develops immunity or dies (17). It also describes any situation where bacteria use a limited resource intensively for short bursts, followed by dispersal to new resource patches (e.g., colonization of organic particles in seawater by Vibrio populations). This metapopulation model is fundamentally different from the ecotype model because it does not predict an association between neutral diversity and adaptive traits.

The relevance of the metapopulation model to the species question is that, although highly idealized and simplified, it may capture some of the effects of complexity and instability of actual ecosystems on population structure. Selective sweeps are predicted to be inevitable in simple, stable environments but not in complex metapopulations [a point partly addressed in (28)]. A metapopulation may evolve, differentiate, and adapt without global selective sweeps. Diversity lost by a local selective sweep in one patch may be rescued and reintroduced from other patches. The ecotype model, with its predicted monophyletic relationship between niche and genotype, may therefore not be an appropriate model of speciation in complex ecosystems.

Choosing Between Models It has proven difficult to discriminate between models of population differentiation that focus on ecotypes or metapopulations. For example, the ecotypic structure of a soil Bacillus has been modeled to predict a priori which sequence clusters were ecotypes, and hence which ones should be associated with specific ecological properties (16). Some clusters are associated with certain phenotypic traits, such as a propensity to grow on shady north-facing slopes or sunny south-facing slopes. However, this model fitted no better (and in fact slightly worse) than a version of the model with several subpopulations and diversity generated only by neutral drift. This version of the model was dismissed because of its association with a very low estimate of population size (14). However, estimates of effective population size Ne are often grossly disconnected from census population sizes. It has proven very challenging to find models that successfully explain low estimated values of Ne while providing better predictions than models based on simple neutral drift. The analysis of Bacillus partly did this by predicting more ecotypes in the model than were observed

(27). This structure well describes the situation does not predict an association between neutral than observed for bacteria. The use of longitu- Bottlenecks, Metapopulations, and Local for parasites, which can colonize a host but are diversity and adaptive traits. dinal ecological and genetic data to distinguish Extinctions then forced to move on because the host develops The relevance of the metapopulation model to between competing models of evolution has a The essential element of the ecotype model immunity or dies (17). It also describes any the species question is that, although highly ideneutral diversity long pedigree eukaryotic On with to limiting situation whereinbacteria use abiology limited (26). resource alizedrespect and simplified, it may capture some is of not the the basis offor these any inference of a niche per se, rather the intensively shortanalogies, bursts, followed by dispersal effects adaptation of complexity andbut instability of effecactual structure driven(e.g., by selective sweeps population bottleneck caused by structure. the replacement of to new resource patches colonization of tive ecosystems on population Selective data from natupopulation by descendants from would require good longitudinal the whole organic particles in seawater by Vibrio popula- sweeps are predicted to be inevitable in simple,a extinction ral bacterial populations, as well as is observations individual andbutthenotresulting tions).

This metapopulation model fundamen- single stable environments in complex metaof episodic crashes in diversity causally associof all other lineages (Fig. 2A).

Other tally different from the ecotype model because it populations [a point partly addressed inmecha(28)]. ated with genetic changes and not associated nisms that induce or involve regular population with changes in ecological covariates. bottlenecks will also restrict neutral diversity.

743 ww.sciencemag.org SCIENCE VOL 323 6 FEBRUARY 2009

32

Relative r of of change distance between Relative rate r rate change in in distance between -9)-9) clusters per generation (x10 clusters per generation (x10

SPECIALSECTION

and diversity generated only by neutral drift. This ples shown in Fig. 3 differ only in the rate of to recombination will depend more on the accuto recombination will depend more on accuples shown in Fig. 3 differbetween only inthe theclusters, rate of mulation and diversity by neutral drift.of This of differences at neutral locithethan at recombination version of thegenerated model wasonly dismissed because its homologous mulation loci. of differences neutral lociathan at homologous recombination between the clusters, version of the model waslow dismissed because of its all The modelsatalso assumed homoother parameters being held constant.

As re- adaptive association with a very estimate of populaadaptive loci.

The models also assumed a homoall other parameters being held constant.

As reassociation with a very low estimate of population size (14). However, estimates of effective combination increases, we see a distinction be- geneous distribution of polymorphisms across tion size (14).

However, estimates of effective combination increases, we see a distinction be- geneous distribution of polymorphisms across population size N e are often grossly disconnected population size Ne ecological are often grossly Recombination rate relative to mutation using established criteria, a hypothfrom census population sizes.

It hasdisconnected proven very Recombination rate relative to mutation from census population sizes. has successfully proven very that can be esis challenging to tested. find models Itthat 0.10 clonal 3 challenging to find 0.10 clonal This problem of lowmodels power selection 3 explain low estimated values tothat ofdetect Nesuccessfully while pro0.53 threshold explain low estimated Ne while protovalues reject neutrality) a (or, more accurately, viding better predictions than of models basedis on 0.53 threshold viding better predictions than models on 2.00 sexual simple neutral drift. in The analysis of based Bacillus genetics that very general problem population 2 2.00 sexual simple neutral drift.

The analysis of Bacillus 2 partlynot didnegate this by ecotypes in in does thepredicting importancemore of adaptation partly did this by predicting more ecotypes in the model than were observed established evolution, but rather suggests using that more work theneeded model than were aobserved usingthat established 1 ecological criteria, hypothesis can be model-based methods is if we want 1 ecological criteria, a hypothesis that can be tested. to discriminate among different biologically tested.

This problem of low power to detectdata. selection plausible explanations of genetic In 0 problem of low powerneutrality) to detect selection (or, This more accurately, is a very scheme for performing Table 1 we proposetoa reject 0 (or, more accurately, to reject neutrality) is a very general problem in population that does analyses that could be used genetics to test, develop, general problemimportance in population genetics that does not negate of adaptation evoand validatethedifferent competing modelsinmore -1 not negate the importance ofthat adaptation in evolution, but rather suggests more work is -1 systematically. lution, but rather suggests that more work is "Speciation point" for needed if we want model-based methods to dis"Speciation point" for sexual species needed if among we wantdifferent model-based methods to discriminate biologically plausible -2 sexual species Homologous Recombination criminate among different biologically plausible -2 explanations genetic data.

In Table 1 we proOne specific of challenge to models that invoke explanations of genetic data.

In Table 1that we propose a scheme for performing analyses could involves a feature of bacterial ecotypic structure poseused a scheme performing analyses could -3 be to test,fordevelop, and validatethat different evolution—homologous recombination—that -3 be used to test, develop, and validate different competing models more systematically.

0% 10% 20% 30% we have notmodels yet discussed.

Bacterial reproduccompeting more systematically.

0% 10% 20% 30% Mean genetic distance between clusters the obligate reassortment tion does not involve Homologous Recombination Mean genetic distance between clusters of genetic material observed in most higher orHomologous Recombination 3. The dynamics of cluster divergence. The figure summarizes some key results from (15) in a One specific challenge to models that invoke Fig. Fig. 3. The dynamics of cluster Thetwo figure summarizes some key results from (15) in ganisms.

However, recombination does occur in phase-space plot of the geneticdivergence. dynamics of populations, with recombination occurring be-a One specific challenge to amodels invoke ecotypic structure involves feature that of bacterial phase-space plot of the genetic dynamics of two populations, with recombination occurring bearchaea (29) and typically bacteria ecotypicand structure involves a feature of involves bacterial tween them at a rate that is varied for the three different simulations. The y axis shows the rate of evolution—homologous recombination—that tween them at a rate that isbetween varied for the three as different simulations.

The y distance axis shows the(xrate of change of genetic distance the clusters a function of the genetic itself axis). short piece of DNA with the replacement of a evolution—homologous recombination—that we have not yet discussed. Bacterial reproduction When changethe of rate genetic distance between the clusters as a function of the genetic distance itself (x axis). of change is positive, the populations will diverge genetically; when negative, they strain. the homologous segment from reassortment another we have yet discussed.

Bacterial reproduction does notnot involve the obligate of converge.

When the rate change of is positive, the each populations willisdiverge when negative, they The of direction change for scenario shown genetically; by arrows color-coded to each Recombination becomes less probable with does not involve the obligate reassortment of genetic material observed in most higher orga- scenario. converge. For Thelow direction of change for each scenario is shown by arrows color-coded to each recombination rates, the populations are effectively clonal and always diverge sequence divergence between the increasing genetic However, material observed in mostdoes higher orgascenario.

For As lowthe recombination populations are effectively clonal and alwaysslow diverge nisms. recombination occur in (green line). recombinationrates, rate the increases, the cohesive effects of recombination the nisms.

However, recombination does occur in 31), which reduces donor and the recipient (30, (green line).

As the recombination rate increases, the cohesive effects of recombination slow the bacteria and archaea (29) and typically involves rate of divergence, until a threshold is passed (red line) and the populations become effectively bacteria and and typically involves but does not archaea eliminate recombination between rate of in divergence, is passed (red line) and For the recombination populations become effectively the replacement of a(29) short piece of DNA with sexual the sense until that athethreshold populations no longer diverge. rates above this the replacement of a short pieceanother of such DNAstrain. with sexual in the sense that the populations no longer diverge. For recombination rates above this closely related species.

Because of interthe homologous segment from level, the fate of the two populations will depend on how genetically distinct they are at the outset. the homologous segment from another strain. any given isolate within species recombination, level, of the populations will then depend on how genetically distinct are at the outset.

Recombination becomes less probable with in- If theythe arefate within thetwo “speciation point,” recombination will cause themthey to merge.

If they are Recombination becomes lesstobetween probable with in- farther If they are within the “speciation point,” then recombination will cause them to merge.

If they are at donor least acreasing speciessequence is almost certain containthe divergence away than this “speciation point,” they will continue to diverge from each other. These creasing sequence divergence farther are away than using this “speciation they some material isbetween characteristic of curves derived the model point,” described in will (15).continue to diverge from each other. These and thegenetic recipient (30, 31),that which reducesthe butdonor does and the recipient (30, species.

31), which reduces but does closely related Hence, whereas it curves are derived using the model described in (15). other was once thought that bacteria do not form spe6 FEBRUARY 2009 VOL 323 SCIENCE www.sciencemag.org 744 cies in the eukaryotic sense because they do not 6 FEBRUARY 2009 VOL 323 SCIENCE www.sciencemag.org 744 recombine at all (32), one current view is that process that reduces the rate of recombination Illegitimate Recombination and Gene they do not form species because they recom- between them—for example, a period of allopa- Content Variation bine too much (5). try or ecological differentiation. The speciation Illegitimate recombination or gene acquisition In asexual clonal organisms, even in the point is the amount of divergence between clus- is another unusual feature of bacteria. In this absence of any selective pressure, clusters will ters that needs to accumulate to prevent them case, genes or clusters of genes are acquired spontaneously split into multiple lineages or from returning to a single cluster if the barriers that typically have no homolog(s) in the recipi“daughter” clusters (15). However, under cer- to recombination are removed. A recent study ent strain. The importance of this phenomenon tain circumstances recombination can prevent hypothesized that two related Campylobacter is evident in the clear and ubiquitous signature this, and we can hence divide the bacteria into species are currently undergoing this process of such events in the growing body of genomic “sexual” and “nonsexual” species. This effect, of merging into a single species as a result of data. These are identified by differences in the described at greater length elsewhere (15), is changes in their environment (33). characteristics of the acquired DNA and that of The above insights were reached using mod- the host strain, for example, in base composition summarized in Fig. 3, which shows the rate at which two clusters diverge over time—that els based on the assumption that genetic varia- or codon usage; in most cases, the donor of the is, the increase in the mean genetic distance tion is neutral. Although this is obviously not DNA in question is unknown. Gene acquisition between them. If this becomes negative, then always an appropriate assumption, it is plau- leads to genomes being punctuated by stretches the two clusters will stop diverging and instead sible that the number of loci explicitly involved of foreign DNA. The largest of these (which converge. The three examples shown in Fig. 3 in adaptive ecological differentiation will be may be many kilobases in length) were initially differ only in the rate of homologous recom- small, and thus that in an unstable landscape, termed “pathogenicity islands,” because the new bination between the clusters, all other param- genomic barriers to recombination will depend functions encoded by the imports were often eters being held constant. As recombination in- more on the accumulation of differences at neu- involved in virulence, but a better term is “gecreases, we see a distinction between a “clonal” tral loci than at adaptive loci. The models also nomic islands” as the phenomenon is far from organism in which clusters are predicted to di- assumed a homogeneous distribution of poly- limited to pathogens (36, 37). Although it is verge (the green line) and a “sexual” organism morphisms across the genome, and violation hard to quantify the selective impact of import(the blue line) in which they are predicted to be of this may alter the tempo and mode of these ing any given gene(s) into a new background, held together by recombination. For “sexual” processes (34, 35). the occasional ability to gain a new adaptation species, the divergence of clusters requires a in this fashion—such as a new metabolic capa-

33

the clear and ubiquitous signature of such events tionary fate of such genes may hence be only acteristics, including the extent of variation in The importance of this phenomenon is evident in between them by mobile elements. The evolu- different genera on the basis of their specific charin the growing body of genomic data. These are loosely coupled with that of any particular gene content and recombination. In any case, no the clear and ubiquitous signature of such events tionary fate of such genes may hence be only acteristics, including the extent of variation in identified by differences in the characteristics of species or strain in which they are found, and biologist would deny the importance of ecology in the growing body of genomic data. These are loosely coupled with that of any particular gene content and recombination. In any case, no are maintained selection by and the biologist to what would we observe, it may not be easy to the acquired DNA that of inthethehost strain, for of they species or strain in through which they are found, deny thebut importance of ecology identified by and differences characteristics incorporate it in a fashion that example, in base composition or cobut it may not beineasy to is the acquired DNA andof that of the host strain, for they are maintained through selection by the to what we observe, bility ormost a new mode transticular species ortaxonomists. strain which convenient for Nonedon usage; in cases, the donor of incorporate it inand a fashion is example, in base composition or cothey arethat mainmission for a pathogen—may they are found, theless, population geneticists the DNA in question is unknown. convenient for taxonomists.

None-may don usage; in most cases, the donor of be of enormous importance in tained through selection by the have little choice geneticists but to tackle Gene the acquisition to genomes theless, population may the DNA inleads question is unknown. terms of speciation. habitat to which each host strain is question defining being Gene punctuated by leads stretches of have little of choice but bacterial to tackle species the acquisition to genomes Perhaps even more striking adapted. In the case of very mobile question of defining bacterial species punctuated by of stretches of or, at the very least, populations. foreignbeing DNA. The largest these elements—for example, plasmids is the amount of variation in or, at the veryareleast, populations.

DNA. Thekilobases largest ofintheseEcological Whether estimating effective (whichforeign maycontent be many revealed by mulencoding we resistance to antibiotics gene Ecological factor b Whether we size are estimating effective may be termed many kilobases in population from neutral diversity length)(which were initially “pathofactor b tiple genomes from the same or heavy metals—the ecological population size from neutral diversity length) were because initially termed “pathoor choosing an appropriate set of genicity islands,” the that new which implies determined by these species, specificity orstrains choosing anforappropriate set of at genicity islands,” because the new to test positive selection functions encoded by the occurs importsatwere gene acquisition a accessory loci may have no link strains to of testinterest, for positive selection at functionsin encoded by the imports were a locus species definitions often involved virulence, but a bethigh frequency. observe using surprisingly the of sequence a to locus interest, we species definitions often involved in virulence, butIta betare implicit ingenes much(Fig. of the ter term is “genomic islands” as the is now to speak housekeeping 4).analytical are implicit in much of the analytical ter termcommonplace is “genomic islands” as the toolkit of population genetics. phenomenon is far from limited to genome, which of the “core” toolkit of population genetics. phenomenon is far from limited to Distinguishing among mechapathogens (36, 37).

Although it is hard encodes fundamental funcIdentifying Mechanisms and Distinguishing among mechapathogens (36, 37). Although it is hard nisms population differentiation to quantify the selective impact of imtions shared all members Delineating Speciesdifferentiation nisms ofofpopulation to quantify the by selective impact of imbacteria ultimately comes down portingporting given gene(s) ainto new What do ultimately we want comes from bacterial ofany a species (and, gene(s) it into should goa new inin bacteria down to to any given Ecological Ecologicalfactor factoraa species?

Do weofneed theoretical withoutthesaying, otherability related testing ability of different models background, occasional to to testing thethe ability different models background, the occasional ability 4. Differences between schematic toto 4. Differences betweencore coreand andauxiliary auxiliary genes. genes. This schematic consistency evenvariable atvariable the expense species), which bolted explainhighly highly patterns gain a new in thisinis fashion— explain patterns gain aadaptation newonto adaptation this fashion—Fig.Fig. illustrates the relationships between three species in “ecotype space,” within illustrates the relationships between three species in “ecotype space,” of taxonomic practicality, incorpothe “auxiliary” or “accessory” within and between genetic-ecological such as a new metabolic capability and between genetic-ecological such as a new metabolic capability shown in two dimensions,and anda amobile mobilegene gene common common to herehere in two dimensions, to all all three. three. clusters “clonal” “sexual” genome, composed of genes rating both (Fig. 1).1).It It isand stillstill unclear or amode new mode of transmission clusters (Fig. is unclear or a new of transmission for afor ashown The areas occupied by the species are shown as solid in red, blue, The areas occupied by the species are shown as solid lines in red, blue, whether populations into a single theoand operons may or may these patterns areare mainpathogen—may of enormous whether these patterns mainpathogen—may bethat ofbe enormous im- im- and green. The part of the ecological space where the shared mobile gene and green. The part of the ecological space where the shared mobile gene tained framework?

One unifying present all isolates. It reticalby notinbeterms gene flow or or selection, portance in terms of speciation. tained by gene flow selection, portance of in speciation. is selected in each species shownbybyaadashed dashedpurple purple line line and is selected in each species is isshown and overlaps overlaps theoretical is consider seems likelymore that more such striking auxiliaof to population Perhaps even andwhat whattheconcept theeffect effect of population Perhaps even striking is theis the all three species ranges. Examples of (circular) genomes from each species and all three species ranges.

Examples of (circular) genomes from each species structure is.

The joint distribution of of ofhelp variation in gene content with and without the purple mobile element are also illustrated. Note as the arena within which genes toindetermine the species ryof structure is.

The joint distribution amountamount variation gene content that with and without the purple mobile element are also illustrated. Note that genetic and ecological data cancan be be revealed by multiple genomes individuals are similar enough, specific ecological properties for each species, the locus is not selected for all isolates, and its evolugenetic and ecological data revealed by multiple genomes fromfromfor each the locus is not selected for all isolates, and its evolu- or interbreed above that for Vibrio the same species, For implies that tionaryspecies, enough, organism. example, of the is uncoupled from that of each host species, because if one used, used,asasdescribed described above forindiVibrio the same species, whichwhich implies that tionary fatefate is uncoupled from that of each host species, because if one species to genes define compete populations gene acquisition at a sur- undergoes a selective sweep vidual (13), variant dia group of related occurs Leptospirilor goes extinct, the mobile gene may be gene acquisition occurs at a sur- undergoes a selective sweep or goes extinct, the mobile gene may be species (13), to define populations making a strong success. theoretprisingly high frequency.

It is now reintroduced from one of the other species. Examples of such distributed without reproductive been hypothrectly for lum has recently without makingto aeither strong theoretprisingly high frequency.

It ofisthe now of the determinants other species.inExamples suchb-lactamase distributed ical commitment of these commonplace to speak “core”reintroduced loci includefrom drugone resistance pathogensof(e.g., Practical advances building on esized to adapt to different ical commitment to either of these commonplace to speak of thefundamental “core” loci genes) include drug resistance determinants in pathogens (e.g., b-lactamase alternatives.

One clear result from genome, which encodes and heavy metal resistance in environmental organisms. These this or other theoretical concepts areas of an acid mine drainage alternatives.

One clear result from genome, which encodes fundamental and metal resistance in environmental of the hereare is functions by allof members genes mayheavy be transferred among strains and species by organisms. conjugative These plas- allwill onlystudies come discussed when these system byshared shuffling chro- of agenes) all of the studies discussed here functions shared by all members of a genes may be transferred among strains and species by conjugative plasthat the underlying ques- is species it should go without mids or other mobile elements (including transducing phage). developed into theoretical explicit models enriched in mosome(and, segments that the underlying theoretical quesspeciessaying, (and, it should go without mids or other mobile elements (including transducing phage). tions concerning species will not be other related species), onto and model-based algorithms that noncore genes (38, 39). We tions concerning species will not saying,which other related species), onto answered in the absence of more detailed genetichabitat to which each host strain is adapted.

In is bolted the “auxiliary” or “accessory” are tested and refined on a wide be should, however, be aware mapping. some guidethe case of very mobile elements—for examcomposed genesormay and operons answered in the absenceMoreover, of more genetictodifferent which each host strain is adapted.

In environmental which genome, is bolted thein“auxiliary” “accessory” it maydetailed be sensible that changes coreof genes also leadthat to habitat tic of levels of ecological specificity, range of data.

Alternatively, lines for the types of ecological studies that will ple, plasmids encoding resistance to antibiotics or may or may not be present in all isolates.

It seems environmental mapping.

Moreover, some guidethe case of very mobile elements—for examgenome, composed of genes and operons that ecological differentiation, a phenomenon well ranging from highly conserved core functions to suggest an ad hoc application of principles be most informative are emerging.

Most imporheavy metals—the ecological specificity deterthat such auxiliary genes help to determine lines for the types of ecological studies that ple, plasmids encoding resistance to antibiotics or may orlikely may not be present in all isolates.

It seems documented in experimental studies of bacteria that are essential for growth in all environments to different genera on the basis of their specificwill the ecological data be imporrelmined by these loci may have no linka tant, the specific ecological properties the(40). organism. heavy be most informative arecollected emerging. metals—the ecological deterlikely that suchinauxiliary genes help toofdetermine including the extentmust ofMost variation growing structured environments to loci that are accessory involved withspecificity adaptation to characteristics, evant to the niche boundaries of the populations to the sequence clusters we observe using houseFor example, a group of related Leptospirillum tant, thecontent ecological data collected be relby these accessory mayniche-specific have no link in the specific ecological properties of the In must any case, Estimates vary, depending on organism. the genomes mined Some loci narrow gene and recombination. specific habitat. if genetic groups do map exgenes (Fig. 4).we observe using house- studied. has recently been of hypothesized to adapt to dif- keeping evant toAnd thewould niche boundaries of not the populations sequence For example, a group butrelated as littleLeptospirillum as 40% of genes to the biologist deny the importance of ecolthat are available, genes may beclusters distributed across species, being no onto sampling categories (as is likely ferent areas of an acid mine drainage system by keeping genes (Fig. 4). studied.

And geneticbut groups notbemap has recently hypothesized to adapt to dif-of a transferred between them by mobile elements. clusively to what weifobserve, it maydonot easy exgenomes may bebeen present in all sequenced ogy to be the case), more complex statistical models shuffling of chromosome segments enriched in Identifying Mechanisms and clusively onto sampling categories (as is likely in a fashion that is convenient fate of such genes may hence to incorporate it may consider genes Delineating The evolutionary ferent named areas ofspecies an acid(41). mineWe drainage system by Species will be needed to identify and describe the undernoncore genes (38, 39). We should, however, be Identifying Mechanisms population gea named species as being characteristaxonomists.

Nonetheless, be only loosely coupledand with that of any par- for within to be the case), more complex statistical models shuffling of chromosome segments enriched in aware that changes in core genes may also lead to What do we want from bacterial species? Do we lying niche structure. Longitudinal studies that Species neticists may have little choice but to tackle will be needed to identify and describe the the undernoncore genes (38, 39). We should, however, bewell Delineating the dynamics of ecological associations ecological differentiation, a phenomenon need theoretical consistency even at the expense measure question of defining bacterial species or, thethat lying niche structure.

Longitudinal studies aware documented that changes in core genes may also lead to What do we want from bacterial species?

Do we 18. H.time Ochman, C. Wilson, in Escherichia coli andathow willA.also be helpful to determine in experimental studies of bacteria andofvalidating taxonomic practicality, incorporating Table 1. A proposed strategy for developing models of bacterial evolution both that over populations. are estimatvery least, Salmonella typhimurium: Cellular andwe Molecular Biology, measure the dynamics of ecological associations ecological a phenomenon well diversity need theoretical consistency evenfoundation at into the expense mightdifferentiation, eventually be used to classify genetic dataand and“sexual” provide populations a firm for a transient natural habitatsWhether are, and thus how likely growing in structured environments (40). “clonal” a single F. C. Neidhart, Ed. (ASM Press, Washington, DC, diver1987), size from neutral ing population overeffective time will also be helpful to determine how documented in species experimental studiesonofthe bacteria taxonomicframework? practicality, both bottlenecks bacterial are to result.

Finally, whole-genome Estimates vary,concept. depending genomes of theoretical Oneincorporating unifying theoretical pp. 1649–1654. to sity or choosing appropriate set thus of transient natural habitats are, and how likely growing structured (40). a within single sequences from etan entire populations of strains environthatinare available,environments but as little as 40% of genes “clonal” conceptand is to“sexual” considerpopulations species as theinto arena 19. J. R. Thompson al., Appl.

Environ.

Microbiol.

70, 4103 1. Collect according to systematic stratification.

Focus onOne longitudinal for positive a locuswhole-genome of interest, (2004). bottlenecks are selection to result.atFinally, Estimates vary, samples depending on the genomesecological theoretical framework? unifyingstudies, theoretical test geographical studies, and measurement of physical and chemical gradients affecting bacterial 20. M. Kimura, Trends Biochem.

Sci. 1, N152 (1976). definitions are implicit in much the sequences from entire populations of ofenvironthat are available, but as little as 40% of genes concept is to consider species as the arena within species 21. R. Milkman, Trends Sci. 1,genetics.

N152 (1976). growth. Consider biotic factors www.sciencemag.org such as the presence of other competing parasitic6 phage.

745 SCIENCEbacteria VOLor323 FEBRUARY 2009 of Biochem. population analytical toolkit

Speciation

22. K. C. Atwood, L. K. Schneider, F. J. Ryan, Proc. Natl.

2. For each isolate, sequence as much as possible and affordable (16S rRNA, MLSA, auxiliary Distinguishing among mechanisms of popAcad. Sci. U.S.A. 37, 146 (1951). genes, full genomes, etc.). bacteria ultimately ulation www.sciencemag.org SCIENCE VOL 323 6 FEBRUARY 23. B.2009 R. differentiation Levin, Genetics 99, 1in(1981).

3. Use empirical classification algorithms that use genetic and ecological data to jointly map isolates.

24. F. M.down Cohan, to E. B.testing Perry, Curr.

17, R373 differcomes theBiol. ability of(2007).

25. models A. Rambaut al., Naturehighly 453, 615 (2008). patterns 4. To guide model formulation, use population genetic tests on observed clusters, focusing on ent toetexplain variable 26. R. A. Fisher, E. B. Ford, Heredity 1, 143 (1947). tests for selection, population structure, and gene flow. and between genetic-ecological within 27. M. Slatkin, Theor. Popul.

Biol. 12, 253 (1977).clusters 5. Generate evolutionary models and simulate populations. whether these (Fig. It is still 28. J.1).

Majewski, F. M.unclear Cohan, Genetics 152, 1459patterns (1999).

6. Test, then reject or adapt, evolutionary models according to agreement between simulations 29. maintained E. J. Feil, B. G. Spratt, Annu. Rev. Microbiol.

55, 561 (2001). and are by gene flow or selection, 30. J. Majewski, F. M. Cohan, Genetics 153, 1525 (1999). and real populations; if necessary, return to step 1. what the effect of population structure is. The 31. J. Majewski, P. Zawadzki, P. Pickerill, F. M. Cohan, 7. For successful models, develop model-based methods for interpreting pure genetic data jointC.distribution of genetic ecological data G. Dowson, J. Bacteriol.

182,and 1016 (2000). (without ecological covariates) and test on new data. above for12th Vibrio specan 32. be S. T.used, Cowan,asin described Microbial Classification, Symposium 8. If one or more validated models emerge, use these to classify genetic data and to develop the Society for General Microbiology, G. C. Ainsworth, without making cies of(13), to define populations bacterial species concepts.

P. H. A. Sneath, Eds. (Cambridge Univ. Press, Cambridge, a strong theoretical commitment to either of 1962), pp. 433–455.

34 mental bacteria will be useful in dissecting the roles of the auxiliary and core genome in

References and Notes 1. J. Handelsman, Microbiol. Mol. Biol. Rev. 68, 669 (2004).

33. S. K. Sheppard, N. D. McCarthy, D. Falush, M. C. J. Maiden, Science 320, 237 (2008).

34. A. C. Retchless, J. G. Lawrence, Science 317, 1093 (2007).

these alternatives. One clear result from all of the studies discussed here is that the underlying theoretical questions concerning species will not be answered in the absence of more detailed genetic-environmental mapping. Moreover, some guidelines for the types of ecological studies that will be most informative are emerging. Most important, the ecological data collected must be relevant to the niche boundaries of the populations studied. And if genetic groups do not map exclusively onto sampling categories (as is likely to be the case), more complex statistical models will be needed to identify and describe the underlying niche structure. Longitudinal studies that measure the dynamics of ecological associations over time will also be helpful to determine how transient natural habitats are, and thus how likely bottlenecks are to result. Finally, whole-genome sequences from entire populations of environmental bacteria will be useful in dissecting the roles of the auxiliary and core genome in ecological differentiation. If after this process it emerges that some model or models are consistently validated for different study systems, these would inevitably form a good basis for identifying fundamental levels of clustering, or species.

In the foregoing we have emphasized ecotype and metapopulation models, but there are others that deserve consideration—notably the epidemic clonal model (42) and the impact of phage epidemics causing classic Lotka-Volterra boom-bust dynamics (43) illustrated in Fig. 2D—and it is possible, even likely, that more than one of these mechanisms may be relevant to any given problem in speciation and cluster formation. Distinguishing among these mechanisms is the bacterial species challenge (Table 1), described in 1991 by John Maynard Smith as follows: “Ecotypic structure, hitch-hiking, and localized recombination can explain the observed patterns of variation. The difficulty, of course, is that the model is sufficiently flexible to explain almost anything. To test the hypothesis of ecotypic structure, we need to know the distribution of electrophoretic types [i.e., genotypes] in different habitats” (17).

Much research on bacterial species to date has come from studies on pathogens, where the correct identification of species is crucial for accurate clinical diagnoses. However, for pathogens the identification of the multiple ecological

niches within (for example) the nasopharynx or gut is difficult, and studies of the relationships between bacterial populations and ecology may be more fruitful for some environmental species where the categorization of niches is a more tractable enterprise. Hopefully, we will soon obtain richer data sets that map bacterial diversity onto ecology and provide a way to distinguish among various models of population differentiation and speciation, including those based on ecotypes or metapopulations.

References and Notes

1. J. Handelsman, Microbiol. Mol. Biol. Rev. 68, 669 (2004).

2. W. P. Hanage, C. Fraser, B. G. Spratt, Philos.

Trans. R. Soc. London Ser. B 361, 1917 (2006).

3. M. Achtman, M. Wagner, Nat. Rev. Microbiol.

6, 431 (2008).

4. F. M. Cohan, Annu. Rev. Microbiol. 56, 457 (2002).

5. W. F. Doolittle, R. T. Papke, Genome Biol. 7, 116 (2006).

6. D. Gevers et al., Nat. Rev. Microbiol. 3, 733 (2005).

7. M. F. Polz, D. E. Hunt, S. P. Preheim, D. M. Weinreich, Philos. Trans. R. Soc. London Ser. B 361, 2009 (2006).

8. D. B. Rusch et al., PLoS Biol. 5, e77 (2007).

9. C. Darwin, The Origin of Species (Penguin Classics, London, 1985).

10. E. Stackebrandt et al., Int. J. Syst. Evol.

Microbiol. 52, 1043 (2002).

11. R. Facklam, Clin. Microbiol. Rev. 15, 613 (2002).

12. J. C. Arbique et al., J. Clin. Microbiol. 42, 4686 (2004).

13. D. E. Hunt et al., Science 320, 1081 (2008).

14. See supplementary information in (16) for a statistical comparison of the ecotype model with an effective neutral model and an implicit estimate of Ne. 15. C. Fraser, W. P. Hanage, B. G. Spratt, Science 315, 476 (2007).

16. A. Koeppel et al., Proc. Natl. Acad. Sci. U.S.A. 105, 2504 (2008).

17. J. Maynard Smith, Proc. R. Soc. London Ser. B 245, 37 (1991).

18. H. Ochman, A. C. Wilson, in Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology, F. C. Neidhart, Ed. (ASM Press, Washington, DC, 1987), pp.

1649–1654.

19. J. R. Thompson et al., Appl. Environ.

Microbiol. 70, 4103 (2004).

20. M. Kimura, Trends Biochem. Sci. 1, N152 (1976).

21. R. Milkman, Trends Biochem. Sci. 1, N152 (1976).

22. K. C. Atwood, L. K. Schneider, F. J. Ryan, Proc. Natl. Acad. Sci. U.S.A. 37, 146 (1951).

23. B. R. Levin, Genetics 99, 1 (1981).

24. F. M. Cohan, E. B. Perry, Curr. Biol. 17, R373 (2007).

25. A. Rambaut et al., Nature 453, 615 (2008).

26. R. A. Fisher, E. B. Ford, Heredity 1, 143 (1947).

27. M. Slatkin, Theor. Popul. Biol. 12, 253 (1977).

28. J. Majewski, F. M. Cohan, Genetics 152, 1459 (1999).

29. E. J. Feil, B. G. Spratt, Annu. Rev. Microbiol.

55, 561 (2001).

30. J. Majewski, F. M. Cohan, Genetics 153, 1525 (1999).

31. J. Majewski, P. Zawadzki, P. Pickerill, F. M. Cohan, C. G. Dowson, J. Bacteriol. 182, 1016 (2000).

32. S. T. Cowan, in Microbial Classification, 12th Symposium of the Society for General Microbiology, G. C. Ainsworth, P. H. A. Sneath, Eds. (Cambridge Univ. Press, Cambridge, 1962), pp. 433–455.

33. S. K. Sheppard, N. D. McCarthy, D. Falush, M. C. J. Maiden, Science 320, 237 (2008).

34. A. C. Retchless, J. G. Lawrence, Science 317, 1093 (2007).

35. K. Vetsigian, N. Goldenfeld, Proc. Natl. Acad.

Sci. U.S.A. 102, 7332 (2005).

36. A. Tuanyok et al., BMC Genomics 9, 566 (2008).

37. U. Dobrindt, B. Hochhut, U. Hentschel, J. Hacker, Nat. Rev. Microbiol. 2, 414 (2004).

38. P. Wilmes, S. L. Simmons, V. J. Denef, J. F. Banfield, FEMS Microbiol. Rev. 33, 109 (2009).

39. V. J. Denef et al., Environ. Microbiol. 11, 313 (2008).

40. E. Bantinaki et al., Genetics 176, 441 (2007).

41. R. A. Welch et al., Proc. Natl. Acad. Sci. U.S.A. 99, 17020 (2002).

42. J. Maynard Smith, N. H. Smith, M. O’Rourke, B. G. Spratt, Proc. Natl. Acad. Sci. U.S.A. 90, 4384 (1993).

43. K. H. Hoffmann et al., FEMS Microbiol. Lett.

273, 224 (2007).

44. We thank T. Connor and S. Deeny for useful discussions. Supported by University Research Fellowships from the Royal Society (C.F. and W.P.H.), a program grant from the Wellcome Trust (B.G.S.), grants from the U.S. Department of Energy Genomes to Life program (M.F.P. and E.J.A.), and the NSF/ National Institute of Environmental Health Sciences Woods Hole Centre for Oceans and Human Health, the NSF Biological Oceanography Program, and the Moore Foundation (M.F.P.).

745

35

Berlin,

nu. Rev.

9). nández,

4). nsect y, J. Rolff, 18.

004). , Proc. R.

in ot

hreat, sparse ators, species

pot. region, his sets ty

ause of n idenal proions of m sub-

36

L L

processes invoked to explain regions Supporting must OnlinebeMaterial www.sciencemag.org/cgi/content/full/323/5915/782/DC1 of high endemism (5, 6). Recent studies from Materials and Methods subtropical biomes have usefully employed post SOM Text models of species and habihoc palaeoclimate Figs. S1 and S2 Table to S1 provide insights about processes shaping tats References and species diversity (5, 7). Building genetic Audio S1 to S4 on them, we first map the palaeodistribution of 22 July 2008;species accepted to 28 identify November 2008 temporally stable endemic 10.1126/science.1163583 (refugial) and unstable (recently colonized) regions for species occurrence, which are then validatedbiomes with multispecies tropical have usefullymolecular employeddata. post Gohoc ing beyond the traditional species-by-species palaeoclimate models of species and habitats to approach, the molecular analyses contrast the provide insights about processes shaping genetic data to the spatially exfit of assemblage-level and species diversity (5, 7). Building on them, we by the plicit demographic scenarios suggested first map the palaeodistribution of endemic climate-based models. species to identify temporally stable (refugial) apply this approach to one of regions the world’s andWe unstable (recently colonized) for yet notoriously endangered most speciesspecies-rich, occurrence, which are then validated with Brazilian and understudied ecosystems: multispecies molecular data. Goingthebeyond the Atlantic Originally extending for traditionalrainforest. species-by-species approach, the mokm2 along thetheBrazilian coast and 1,300,000 lecular analyses contrast fit of assemblagethis biome reaching and Argentina, level datainto to Paraguay the spatially explicit demographic (8). has been reduced to by lessthe than 8% of its range scenarios suggested climate-based models.

Today’s fragments harbor one of the perWe apply this approach to one of largest the world’s of endemic species in the world, with centages most species-rich, yet notoriously endangered and understudied the of Brazilian Atlantic vertebrates still many species ecosystems: and even genera 1 Museum of Vertebrate Zoology, University of California, Berkeley, CA 94720–3160, USA. 2Biology Department, Queens College, City University of New York, Flushing, NY 11367, USA. 3Departamento de Zoologia, Instituto de Biociências, UNESP, Rio Claro, SP 3526-4100, Brazil.

4 Departamento de Zoologia, Instituto de Biociências, Universidade de São Paulo, SP 055008-090, Brazil.

*To whom correspondence should be addressed. E-mail: carnaval@berkeley.edu

6 FEBRUARY 2009

We use molecular genetic data from multiple, largely codistributed species to test whether spatial modeling of species-specific Late Quaternary refugia sheds light on historical processes and hence improves prediction of genetic endemism and diversity in tropical Brazil (11). We focus on three common species of tree frogs that are widely distributed along the Brazilian Atlantic forest: Hypsiboas albomarginatus, H. semilineatus, and H. faber. Given their life history traits, amphibians are useful indicators of environmental changes through time (12). Whereas H. albomarginatus and H. semilineatus occur in low and mid altitudes and are mostly restricted to the evergreen or semideciduous components of the Atlantic Forest in eastern Brazil, H. faber has a broader altitudinal range and also inhabits mixed and deciduous areas, occupying interior and coastal sites in the Atlantic Forest south to Paraguay and Argentina (figs. S1 and S2) (13).

The comparative phylogeographic approach is a powerful test of assemblage-scale responses to former environmental change and thereby provides a means for critical assessment of the scenarios produced by modeling of species’ distributions under palaeoclimates (7).

The palaeomodeling method intersects predicted species’ distributions under current conditions and climatic extremes of the Late Quaternary (6000 years before present, or 6 kybp, and 21 kybp) to predict areas of stability (regions in which species are predicted to occupy irrespective of time period) and unstable areas (7, 14). Because the stability maps raise specific hypotheses about regional differences

785

(AUC) values (16) 0.968, 0.989, and 0.994; maximum Kappa (17) 0.81, 0.925, and 0.94 in H. albomarginatus, H. faber, and H. semilineatus, respectively (fig. S2)]. Stability maps, depicting the intersection of distribution models for each taxon under current, 6 kybp, and 21 kybp climates, predict for all species a large central refugium throughout the Late Quaternary (“Bahia refugium”) (Fig. 2). A second, much smaller refugium is predicted in the northeasternmost portion of the forest (“Pernambuco refugium”).

In H. faber, a third, southeastern refugium of intermediate size is also predicted (“São Paulo refugium”). This is not surprising, given that this species occupies a broader environmental niche.

In contrast to the central and northern regions, populations south of the Bahia or São Paulo refugia appear much less stable, despite the more extensive (preclearing) range of the forest in southern and southeastern Brazil. We hypothesize that these areas received a significant influx of migrants from adjacent, large refugial populations after the LGM. These palaeomodel results are congruent with the fossil pollen record, which documents a replacement of forests by grasslands in the southern Atlantic forest during the LGM (14, 18) and suggests the occurrence of small forest refugia in the southernmost range of the putative Bahia refugium (19). The results also agree generally with forest models published previously (14), although the central refugium extends farther south in the frog-based models. Such differences are expected because the forest and its associated species may differ slightly in their climatic tolerances and realized

Distribution Modeling

Stability Predicts Genetic Diversity in the Brazilian Atlantic Forest Hotspot

Hypothesis Formulation

9. J. A. Thomas, G. W. Elmes, J. C. Wardlaw, Proc. R. Soc. in persistence diversity, theypost leadhoc to tropical biomesand havehence usefully employed London Ser. B 265, 1895 (1998).

10. G. W. Elmes, J. C. Wardlaw, K. Schönrogge, J. A. Thomas, phylogeographic predictions for both individual palaeoclimate models of species and habitats to Entomol. Exp. Appl. 110, 53 (2004). (codistributed taxa; species insights and assemblages provide about processes shaping genetic 11. Materials and methods are available as supporting driven by the model Fig. 1).

Field sampling is and species diversity (5, 7).

Building on them, we material on Science Online. first map the palaeodistribution endemic and predictions to cover both predictedofrefugia 12. P. J. DeVries, R. B. Cocroft, J. A. Thomas, Biol. J. Linn.

Soc. 49, 229 (1993). species identify colonized) temporallyareas, stableparticularly (refugial) unstableto(recently 1 2 Ana Carnaval, Michael Hickerson, Célio F. B. Haddad,3 13. F.Carolina Roces, J. Tautz, J. Acoust.*Soc. Am. 109,J.3080 (2001). and unstable previously (recently colonized) regions undersampled areas.for If emphasizing 4 1 Miguel T. Rodrigues, 14. R. Hickling, R. L. Brown, Craig J. Acoust.Moritz Soc. Am. 108, 1920 species occurrence, which are thencurrent validated with patterns the approach correctly predicts (2000). multispecies molecular Going beyond the of biodiversity at the data. regional scale, species 15. H. Markl, B. Hölldobler, Behav. Ecol. Sociobiol. 4, 183 Biodiversity hotspots, representing regions with high species endemism and conservation threat, (1978). traditional species-by-species the moshould consistently show (i) approach, higher genetic dihave globally.

Yet, biodiversity distribution data from within hotspots are too sparse 16. H.been Markl,mapped Z. Vgl. Physiol.

60, 103 (1968). lecular analyses contrast the fit of assemblageand among populations in refugia versity within 17. effective H. Markl, Science 149, 1392in(1965). for conservation the face of rapid environmental change. Using frogs as indicators, level data to the spatially explicit demographic 18. D. A. Grasso, Mori, F. Leunder Moli, M.paleoclimates, Giovannotti, ecological nicheA. models and simultaneous Bayesian analyses of multispecies relative to unstable areas, because of long-term scenarios suggested by the climate-based A. Fanfani, Ital. J. Zool. 65, 167 (1998). genetic persistence and population structure; (ii)models. molecular data, we compare alternative hypotheses of assemblage-scale response to late 19. T. C. Scott-Phillips, J. Evol. Biol. 21, 387 (2008).

We apply this approachexpansion to one of in theunstable world’s signature of population Quaternary climate change. reveals a Apollo hotspot within the Brazilian Atlantic forest hotspot.

20. K. G. Schurian, K. Fiedler, Nachr.This Entomol.

Vereins most species-rich, yet notoriously endangered and We show that the southern Atlantic forest was climatically unstable relative to the central region, areas, reflecting multispecies colonization from 14, 339 (1994). understudied ecosystems: the Brazilian Atlantic 21.

C. J. Hill, J. Aust.

Entomol.

Soc. 32, 283 (1993). which served as a large climatic refugium for neotropical species in the late Pleistocene. This sets adjacent refugial regions after the Last Glacial 22. D. R. Nash, T. D. Als, R. Maile, G. R. Jones, J. J. Boomsma, Maximum (LGM, 21 kybp); (iii) absence of new priorities for conservation in Brazil and establishes a validated approach to biodiversity 1 Science 319, 88 (2008).

Museum patterns of Vertebrate Zoology, University of California, genetic of isolation-by-distance in unprediction in other understudied, species-rich regions.

2 23. P. J. DeVries, Am. Mus. Nov. 3025, 1 (1991).

Berkeley, CA 94720–3160, Biology Department, areas, given that USA. colonization has been stable 24. K. Fiedler, B. Hölldobler, P. Seufert, Experientia 52, 14 Queens College, City University of New York, Flushing, NY too recent to3 permit restoration of equilibrium (1996).

Quaternary fluctuations refugia models have been dismissed because of 11367, USA. Departamento de Zoologia, Instituto de 25. M. ate A.ate Travassos, N. E. climate Pierce,climate Anim.

Behav.

60, helped 13 being described (8, 9). Our ultimate goal is to Biociências, Quaternary fluctuations and between migration genetic drift (15);Brazil.

UNESP, Rioand Claro, SP 3526-4100, to shape present-day diversity in temper- conflicting evidence (2, 3) or circularity in iden- 4Departamento de Zoologia, Instituto de Biociências, (2000). diversity in pinpoint regions for inventory work and habitat (iv) strong phylogeographic structure between helpedettoal.,shape present-day 26. N. ate E. Pierce Annu.

Rev. Entomol.

47, 733 (2002). and boreal systems (1), providing a tifying putative refugia (4), but historical pro- Universidade de São Paulo, SP 055008-090, Brazil. boreal systems (1), protection before we lose a substantial fraction refugia, reflecting assemblage-wide, long-term 27. J. C.temperate Downey, C.and Allyn, Bull. Mus. Entomol.

14, progeneral contextA.for understanding current pat- cesses must be invoked to explain regions of whom correspondence be addressed.

1 (1973). in isolated areas. E-mail: population persistence should viding a general context for understanding cur- of described and undocumented diversity. The *To terns ofthank endemism.

In tropics, high endemism (5, 6). Recent studies from sub- carnaval@berkeley.edu 28. We N.ofElfferich andthe P. J.In DeVries forPleistocene introducing Pleis- approach differs from previous methods by diDistribution models developed under current rent patterns endemism. the tropics, us to ant-butterfly acoustics; G. W. Elmes, J. C. Wardlaw, tocene refugia models have been dismissed be- rectly modeling historical processes, as opposed climatic conditions accurately predict distribuV. La Morgia, M. B. Bonsall, and referees for of conflicting evidence (2, 3)for ordesigning circularity to observed biodiversity (10), with the tions2009 of each of the target species along the At- 785 cause www.sciencemag.org SCIENCE patterns VOL 323 6 FEBRUARY comments and advice; and M. Charles the acousticalputative equipment. refugia (4), but historical aim of informing conservation. in identifying lantic rainforest domain [area-under-the-curve

is higherlarger thaninthe because of the in the southernmost range by of magnitude critical assessment of the scenarios produced by testsmall the other central species (Bahia) refugium which refugia documents a replacement of forests phylogeographic approach is a powerful of forest modeling of species’ distributions under assemblage-scale responses to palaeoformer environ- grasslands in the southern Atlantic forest during relative to the less stable (southern) portion of the climates (7). mental change and thereby provides a means for the LGM (14, 18) and suggests the occurrence of forest. Diversity of H. faber in this southern area higher than the otherCurrent species&because of the the southernmost rangeMap of ofispredicted critical assessment of the scenarios preproduced by small forest refugia inSpecies The palaeomodeling method intersects of species’ under palaeooccurrence stable areas historical climate dicted species’modeling distributions underdistributions current condiclimates (7). data Inclimatic H. albomarginatus andLate H. Quaternary faber, the niches. (putative refugia) data tions and extremes of the palaeomodeling method intersects preSpecies Map of predicted Current & São Paulo refugium extension the The predicted (6000 yearsofbefore present, or 6 kybp, and 21 kybp) occurrence stable areas historical climate dicted species’ distributions under current condiwestward theofand neighboring Cerrado data (putative refugia) data to predict into areas stability (regions which tions climatic extremes of in thebiome Late Quaternary reflects overprediction (fig.irrespective S2) (6000 years present, or 6(14). kybp, and species model are predicted tobefore occupy of21 kybp) Spatially explicit hypotheses Re: to predict areas ofthrough stability (regions Models habitat stability time period)ofand unstable areas (7, 14). fluctuatBecausein which species predicted to occupy irrespective of predict patterns of phyloing climatesmaps correctly Spatially explicit hypotheses Re: the stability raiseare specific hypotheses about and unstable areas (7, 14). Because rainforest geography in time the period) Brazilian Atlantic regional differences in persistence and hence dithe stability maps raise specific hypotheses about spatial-temporal patterns (Fig. 2 and S3 todifferences S5). In all species, high distribution of congruence versity, theyfigs. lead to phylogeographic predictions regional in persistence and hence diof colonization genetic diversity across taxa spatial-temporal patterns levels of individual divergence and population structure are for both species and assemblages (codistribution of congruence versity, they lead to phylogeographic predictions and/or vicariance of colonization genetic diversity across taxa forFig. both1). individual species and distributedacross taxa; Field sampling isassemblages driven (corefugia (Tamura-Nei corrected observed and/or vicariance distributed taxa; Fig. 1).

Field sampling is driven by the model to coverBahia both predicted and Perdistances (20):predictions 4 to 7% between by the model predictions to coverareas, both predicted Sampling across predicted stable and unstable areas refugia and unstable colonized) nambuco refugia, 1% (recently between the nearby Bahia Sampling across predicted stable and unstable areas refugia and unstable (recently colonized) areas, particularly emphasizing previously undersamSão Paulo refugia in H. faber).

Similarly, in and particularly emphasizing previously undersampled areas. the approach correctly predicts curthereIfare multiple, divergent withall taxa pled areas. If the approachclades correctly predicts current patterns of biodiversity at the regional scale, rent patterns of biodiversity at the regional in the Bahia region, agreeing with model-based scale, Genetic tests of of Assemblage-scale Genetic tests Assemblage-scale should consistently showarea. (i) higher species should consistently show (i)inhigher genetic Descriptive Descriptive this In genetic predictions of species a large refugium stability/expansion, hypothesis testingtesting stability/expansion, hypothesis phylogeography diversity within populations and among populations in refugia phylogeography diversity within and among in refugia divergence times (HABC) H. faber, divergent clades are also represented divergence times (HABC) relative to unstable areas, because of long-term perrelative to unstable areas, because of long-term perin the São Paulo region, of sistence andmatching population predictions structure; (ii) genetic sigandrefugium population structure; (ii) sigin this area.

Allgenetic taxain show asistence mid-sized nature of population expansion unstable areas, Fig. 1. Proposed method of biodiversity prediction. Three stages are involved: biodiversity disnaturegenetic of population incolonization unstable areas, Fig. 1.tribution Proposed method of model-based biodiversityhypothesis prediction.

Three stages arehypothesis involved:testing biodiversity disreflectingexpansion multispecies from adjacent modeling (top), formulation (middle), and southernmost low diversity across the reflecting multispecies colonization from adjacent tribution modeling (top), model-based hypothesis formulation (middle), hypothesis testing and refugial regions after the Last Glacial Maximum model validation (bottom). range of the forest, an area predicted to be less refugialbyregions after the LastFurthermore, Glacial Maximum stable the palaeomodels. mito- model validation (bottom).

DNA (mtDNA) lineages found in this chondrial786 6 FEBRUARY 2009 VOL 323 SCIENCE www.sciencemag.org refugia (one in in and E). Using Bayes factor (25), we also detect scale, persistence of populations region are shared with adjacent 786 6 FEBRUARY 2009long-term VOL 323 SCIENCE www.sciencemag.org H. albomarginatus and H. semilineatus, two in isolated refugial areas, as opposed to post-LGM evidence for stability in both areas under the noH. faber). migration model [B(Z2 = 0, Z2 > 0) = 4.89], as colonization of refugial regions.

Metrics of genetic diversity confirm the To test for assemblage-wide colonization well as under a postisolation migration model above patterns (Table 1). In H. albomarginatus of predicted unstable areas, we group mtDNA [B(Z2 = 0, Z2 > 0) = 4.84].

Relative to nuclear loci, mtDNA data are and H. semilineatus, genetic diversity (21) is an sequences from the southernmost refugial sites order of magnitude larger in the central (Bahia) [population 1 (Fig. 3A)] and from localities in more variable and readily collected and often refugium relative to the less stable (southern) unstable areas south of the refugium [population provide key insights into biological response portion of the forest. Diversity of H. faber in 2 (Fig. 3A)] to contrast two alternative historical to environmental modification (1). Although this southern area is higher than the other spe- models across the three codistributed species, single-locus inference can be imprecise in the cies because of the presence of two lineages that while allowing the taxon-specific demographic face of coalescent variance and the possibility of co-occur in the adjacent refugia. In all species, parameters to vary. In H1, the long-term persis- selection (26), our method benefits from a mulaverage net nucleotide differences across locali- tence model, two contemporary populations split titaxon approach, while explicitly accounting ties (22) reflects high geographic structure with- from an ancestral population prior to the LGM for the stochasticity of a single-locus coalescent in refugia (2.6 to 6.2% divergence). In contrast, (120,000 to 1.2 million years before present, or across taxa. Combining data sets from several sites located outside (south of) the refugia are Mybp, Fig. 3A). In H2, the recent colonization codistributed groups into a single hierarchical genetically more similar to each other, although model, population 2 is modeled as being colo- Bayesian analysis allowed us to estimate conto a lesser extent in H. faber (0.1 to 1.6%). Sig- nized from refugial population 1 subsequent to gruence across species, while borrowing strength natures of population expansion (23) are found the LGM (0 to 20 kybp; Fig. 3A). The results from the full comparative phylogeographic samin the unstable area for H. albomarginatus and indicate that all three species colonized the ple (24). This can translate into higher analytical H. faber, as well as in the Bahia refugium area southern (unstable) areas after the LGM (Z2 = 3, power and be more informative than qualitative for H. faber and H. semilineatus. The lack of the number of species evolved under H2), even comparisons of species-specific analyses. By signature of population expansion in the south- when allowing for postisolation migration (Fig. capturing the historical signal that emerges from ernmost localities of H. semilineatus may reflect 3, B and C). When Bayes factor is used (25), larger, combined multispecies molecular data low statistical power because of the exception- there is strong support for recent colonization in sets, HABC will offer the possibility of looking ally low levels of diversity observed in this spe- all three species (Z2 = 3) under the no-migration at patterns of historical community assembly in cies. As predicted, isolation by distance is not model [B(Z2 = 3, Z2 < 3) = 35.16], and moderate codistributed nonmodel organisms for which observed in unstable regions, but is detected support under a postisolation migration model barcode-type DNA sequence information (e.g., within refugial areas for H. albomarginatus and [B(Z2 = 3, Z2 < 3) = 5.70]. mtDNA data) can be feasibly collected.

Using the same framework to test for longCollectively, the results identify the central H. faber.

The hierarchical approximate Bayesian com- term persistence of refugial populations, we region as a hotspot within the Atlantic rainforest putation (HABC) method (24) allows us to use compare mtDNA sequences between the pre- hotspot and a refuge for biodiversity during data from all three species at once to test for dicted Pernambuco refugium [population 1 (Fig. climatic extremes of the Late Pleistocene. assemblage-wide responses to Late Quaternary 3A)] and adjacent (northern) populations from This is not to say that southern areas entirely climate change. These analyses support both the Bahia refugium [population 2 (Fig. 3A)] to lacked forested habitats in the late Pleistocene: model-driven hypotheses of (i) simultaneous, contrast alternative historical models H1 and H2. The existence of species and genera endemic multispecies colonization of unstable areas from In this case, the HABC results infer long-term to the southern forests (27), as well as some adjacent refugial populations since the LGM, persistence of populations in isolated refugia for palaeoecological and genetic evidence (28), as opposed to long-term persistence of popu- all three species (Z2 = 0, i.e., Z1 = 3), even when offer evidence to the contrary. Rather, the lations in unstable areas, and (ii) assemblage- allowing for postisolation migration (Fig. 3, D phylogeographically validated palaeomodels Model Validation

’s hiermimicry lack the and of. species li (12): studied nsals or with ants e acouslthough s attracts s to siga basal t expect nea. ecies of icularly Diptera, ays the nt’s hicue has ate and

22 July 2008; accepted 28 November 2008 10.1126/science.1163583

Diversity Distribution Modeling Diversity

REPORTS

Audio S1 to S4

Hypothesis Formulation

7. K. Schönrogge et al., J. Chem. Ecol. 30, 91 (2004).

8. T. Akino, J. J. Knapp, J. A. Thomas, G. W. Elmes, Proc. R. Soc. London Ser. B 266, 1419 (1999).

Model Validation

is admitted and accepted as a member of a host society, it mimics adult ant acoustics (particularly queens) to advance its seniority toward the

37

presence of two lineages that co-occur inAthe ad-

6.2% divergence). In contrast, B sites located

population expansion (23) are found in the C

refugial (stable) versus unstable areasaverage unstable area H. albomarginatus and H. faber, refugiaare aregenetically geneticallyunstable jacent refugia. all species, average net nuclearea for H. for albomarginatus and H. faber, outside(south (south of) of) the the refugia jacent In refugia.

In all species, net nucle- outside in thedifferences Brazilian Atlantic rainforest. as well as in the refugium Bahia refugium area for H. faber similar althoughto to a lesseras well otide across across localities (22) (22) reflects as in the Bahia area for H. faber more similartotoeach each other, other, although a lesser otide differences localities reflects more (Top) Species-specific stability maps; H. semilineatus.

The lack of lack signature of extent faber (0.1 (0.1 to of ofand and high geographic structure within refugia H. semilineatus.

The of signature of ininH.H.faber to 1.6%).

1.6%).Signatures Signatures high geographic structure within refugia (2.6 (2.6 to to extent Pernambuco modeled refugia in black. (A) H. refugium albomarginatus, (B) H. semilineatus, Fig. 2. diversity Genetic diversity in putative C Fig.H.2.faber.

Genetic in putative A B (C) Note the absence of large C B refugial (stable) versus unstable areas A Bahia refugium refugialregions (stable) unstable areas stable the southern portion in theinversus Brazilian Atlantic rainforest. * in the the forest Brazilian Atlantic rainforest. * of of the Bahia andmaps; (Top)(south Species-specific stability (Top)Paulo Species-specific stability Pernambuco modeled refugia in black.

São refugia) relative tomaps; the(A) H. refugium Pernambuco albomarginatus, (B) H. semilineatus, modeled refugia in areas. black.

(A) H. central and northern Asterisks Sã o Paulo refugium refugium (C) H. faber. the absencethe of large albomarginatus, (B) Note H. semilineatus, denote refugia inferred beyond Bahia refugium stable regions in the southern portion (C) H. faber.

Note the absence of large current ranges of the target species.

5.4% * * of theinforest (south of the Bahia and Bahia refugium stable regions the southern portion Symbols indicate localities sampled São Paulo refugia) relativeforto the * * of the forest (south of the Bahia and molecularcentral analysis. bar, areas.

400 km. andScale northern Asterisks Sã o Paulo refugium São PauloThe refugia) relative the the (Bottom) 50% majority-rule condenote refugia inferredtobeyond central and northern areas.

Asterisks current ranges of the target species. sensus Bayesian phylogenetic trees, 5.4% 7% Sã o Paulo refugium 7.8% Symbols indicatefrom localities denote with refugia inferred beyond the for rooted sequences thesampled othmolecular Scale bar, 400 km. current of analysis. thespecies target species. er two ranges congeneric studied 5.4% (Bottom)localities The 50%sampled majority-rule Symbols for con(root notindicate shown).

Thick internodes de5.3 – sensus Bayesian phylogenetic trees, 7% molecular analysis.

Scale bar, 400 km. note clades withwith posterior probability 7.8% 4% 5.8% rooted sequences from the oth(Bottom) The 50% majority-rule congreater than Percentages indicate er 90%. two congeneric species studied sensus Bayesian trees, deTamura-Nei corrected distances (root not phylogenetic shown).

Thick between internodes 7% 5.3 – 7.8% rooted(20). with the probability othnotesequences clades withfrom posterior 4% clades 5.8% 5.6% greater than 90%.

Percentages indicate er two congeneric species studied Tamura-Nei corrected distancesdebetween (root not shown).

Thick internodes 5.3 – clades note clades with(20). posterior probability 4% 5.8% 5.6% greater than 90%. Percentages indicate Tamura-Nei corrected distances between clades (20).

5.6%

Mybp, Fig. 3A). In H2, the recent colonization model, population 2 is modeled as being colonized from refugial population 1 subsequent to

coalescent variance and the possibility of selection (26), our method benefits from a multitaxon approach, while explicitly accounting for the

southern regions. This reassures us that the processes uncovered by the amphibian data may be generalized to and help to explain patterns of

Fig. 3. HABC analyses.

(A) Simulated models H1 (long-term persistence) and H2 (recent colonization). In both cases, each species was modeled as two contemporary populations with mutation-drift parameters q1 and q2 that split from an ancestral population at a time t in the past. Ancestral population sizes are represented by (qt)1 and (qt)2; ybp, years before present. (B to E) Hyperposterior (bars) and hyperprior (dashed) densities of Z2 (number of species evolved under H2) given data from three codistributed frog species. (B) and (C) Models of refugial sites (population 1) and unstable, southern areas (population 2). (D) and (E) Models of Pernambuco refugium (population 1) and Bahia refugium (population 2). (B) and (D) Postisolation migration not included in model; (C) and (E) postisolation migration included in model.

At a broader level, the congruence between

An Online Reference, version 5.2, 15 July

(2000).

New York, 2008); electronic database accessible at http://research.amnh.org/ herpetology/amphibia/index.php.

A. C. Carnaval, C. Moritz, J. Biogeogr. 35, 1187 (2008).

M. Slatkin, Evolution 47, 264 (1993).

J. A. Hanley, B. J. McNeil, Radiology 143, 29 (1982).

J. Cohen, Ed. Psych. Meas. 20, 37 (1960).

H. Behling, R. R. B. Negrelle, Quat. Res. 56, 383 (2001).

H. Behling, H. W. Arz, J. Pätzold, G. Wefer, Palaeogeogr. Palaeoclimatol. Palaeoecol.

179, 227 (2002).

K. Tamura, M. Nei, Mol. Biol. Evol. 9, 678 (1993).

F. Tajima, Genetics 143, 1457 (1996).

M. Nei, Molecular Evolutionary Genetics (Columbia Univ. Press, New York, 1987).

J. C. Fay, C. I. Wu, Genetics 155, 1405 (2000).

M. J. Hickerson, C. P. Meyer, BMC Evol. Biol.

8, 322 (2008).

R. E. Kass, A. E. Raftery, J. Am. Stat. Assoc.

90, 773 (1995).

J. W. O. Ballard, M. Kreitman, Genetics 138, 757 (1994).

C. F. B. Haddad, L. F. Toledo, C. P. A. Prado, Anfibios da Mata Atlântica (Atlantic Forest Amphibians) (Editora Neotropica, São Paulo, Brazil, 2008).

M.-P. Ledru et al., Divers. Distrib. 13, 761 (2007).

K. C. M. Pellegrino et al., Biol. J. Linn. Soc. London 85, 13 (2005).

G. S. Cabanne, F. M. d’Horta, E. H. R. Sari, F. R. Santos, C. Y. Miyaki, Mol. Phylogenet.

Evol. 49, 760 (2008).

J. M. C. da Silva, M. Tabarelli, Nature 404, 72

providing access to collections MNRJ and MZUSP; O. Peixoto, M. Gomes, A. Muri, R. Kautskyi, S. Lima, E. Santos, J. S. Filho, J. V. Filho, G. Barros, J. Queiroz, R. Araújo, L. Japp, H. Japp, J. Giovanelli, J. Alexandrino, L. Toledo, O. Araújo, G. Egito, J. Zina, D. Loebmann, D. Pavan, R. Amaro, V. Verdade, F. Curcio, M. Dixo, and J. Cassimiro for field work assistance; W. Monahan and R. Hijmans for discussions about the modeling work; L. Smith and D. Turong for DNA-sequencing assistance; R. Pereira, R. Damasceno, S. Rovito, J. Kolbe, S. Singhal, R. Puschendorf, and A. Pounds for discussions about earlier versions of the manuscript. Funding was provided by the NSF (awards DBI 0512013 to A.C.C., DEB 0743648 to M.J.H., DEB 416250 and DEB 0817035 to C.M.), Fundação de Amparo à Pesquisa do Estado de São Paulo and Conselho Nacional de Desenvolvimento Científico e Tecnológico (grants to C.F.B.H. and M.T.R.). Sequences are deposited in GenBank (FJ502639-FJ502822). A.C.C. and C.M. designed the study; A.C.C., C.F.B.H., and M.T.R. collected the data; A.C.C., C.M. and M.J.H. analyzed the data; A.C.C. wrote the paper. All authors discussed the results and commented on earlier versions of the manuscript.

788 model-based demographic hypotheses 6 FEBRUARY 2009 VOL 323Museum SCIENCE 32. We thank U. Caramaschi and H. Zaher for 2008 (American of Naturalwww.sciencemag.org History, and q, and average D values of the former were obtained not only from the total

Table 1. Population genetic summary metrics used in model validation. n, Table 1. Population genetic summary metrics used in model validation. n, Sample size; S, number of segregating sites. sites.

The diversity parameter Sample size; S, number of segregating The diversity parameterq qand andmean mean Da across Dlocalities are given per base pair (bp). Hs test (23) is used to detect a across localities are given per base pair (bp). Hs test (23) is used to detect populationpopulation expansion.

BA, Bahia; SP, São refugia.

Because expansion.

BA, Bahia; SP, Paulo São Paulo refugia.

Becausepredicted predicted refugia oftenthan larger than predicted unstable (recently colonized)areas, areas,n, n, S, S, refugia were oftenwere larger predicted unstable (recently colonized)

a of the former were obtained not only from the total q, and average Da values numberofofsamples, samples, all possible combinations of spatially number but but also also from from all possible combinations of spatially contiguouslocalities localities distributed the geographic of the unstable contiguous distributed withinwithin the geographic extensionextension of the unstable area.Parentheses Parentheses encompass minimum and maximum values from subsamples. area. encompass minimum and maximum values from subsamples. valuesininbold bold highlight statistical significance 0.05 probability level.

PPvalues highlight statistical significance at 0.05 at probability level.

SS n, Table 1. Population genetic summary metrics usednin nmodel validation.

Species Species Area Area (min.; max.) (min.; max.)

Sample size; S, number of segregating sites. The diversity parameter q(min.; and mean (min.; max.) max.)

qq Mean Hs corr. a ofDthe were obtained notMantel’s onlycoef. from the coef. total q, and average Da values Mean Dformer Hs Mantel’s corr. a (min.; max.) value) (P value) (P value) number ofmax.) samples,(min.; but alsomax.) from(Pall possible of spatially (min.;max.) (min.; (P value)combinations

albomarginatus (BA) (bp). Hs test 36 D across H. localities are given per Stable base pair (23) is used to 207 detect H.a albomarginatus Stable (BA) 36 207155) (970 bp) (13; 23) (81; population expansion.

BA, Bahia; SP, São Paulo refugia.

Because predicted (970 bp) (13; 23) (81; 155) Unstable refugia were often larger than predicted unstable (recently27 colonized) areas,22n, S,

0.076 localities distributed 0.062 within–20.546 0.499 of the unstable contiguous the geographic extension 0.076 0.062 –20.546 (0.001) 0.499 (0.034; 0.072) encompass (0.020; 0.082) area.

Parentheses minimum(0.141) and maximum values from subsamples.

(0.034; (0.020; 0.082) (0.001) 0.003 0.001 –11.498(0.141) P values in0.072) bold highlight statistical significance at 0.05–0.140 probability level.

Unstable(south of BA) 27 22 0.003 0.001 –11.498 (0.580) –0.140 (0.004) H. semilineatus Stable (BA) 0.031 0.036 0.054 (south of BA) (0.004) (0.580) n 28 S71 q Mean Da –17.778 Hs Mantel’s corr. coef.

Species Area (718 bp) (6; 13) (14; 58) (0.009; 0.034) 0.041) H. semilineatus Stable (BA) 71max.)

0.031 0.036 –17.778 (min.;28 max.) (min.; (min.; max.) (0.007; (min.; max.) (0.029) (P value) (0.460) (P0.054 value) Unstable 15 0.0030.034) 0.004 0.041) 0.114(0.029) 0.436 (0.460) (718 bp) (6; 13) (14; 958) (0.009; (0.007; H. albomarginatus Stable (BA) 36 207 0.076 0.062 –20.546 (0.248) 0.499 (south of BA) (0.357) Unstable 15 9 0.003 0.004 0.114 0.436 H. faber Stable (BA) 28 0.0180.072) 0.026 0.082)–38.111(0.141) 0.803 (0.001) (970 bp) (13; 23) (81; 94 155) (0.034; (0.020; (south of BA) (0.357) (0.0003) (0.248) (771 bp) (13; 23) (42; 80) (0.012; 0.022) (0.001; 0.044) (0.003) Unstable 27 22 0.003 0.001 –11.498 –0.140 H. faber Stable Stable (BA) (SP) 28 15 94 0.018 0.026 0.803 48 0.023 0.028 –5.981–38.111 0.305 (0.580) (south of BA) (0.004) (771 bp) (13; 23) (42; 80) (0.012; 0.022) (0.001; 0.044) (0.115)(0.003) (0.221) (0.0003) H. semilineatus Stable (BA) 28 71 0.031 0.036 –17.778 0.054 40 0.015 0.016 –13.255–5.981 0.0001 0.305 Stable Unstable (SP) 15 18 48 0.023 0.028 (718 bp) (6; 13) (14; 58) (0.009; 0.034) (0.007; 0.041)(0.014)(0.029) (0.456) (0.460) (south of SP) (0.115) (0.221) Unstable 15 9 0.003 0.004 0.114 0.436 Unstable 18 40 0.015 0.016 –13.255 0.0001 (south of BA) (0.357) (0.248) (south of SP) (0.014) (0.456) H. faber Stable (BA) 28 94 0.018 6 FEBRUARY 0.026 –38.111 0.803 www.sciencemag.org SCIENCE VOL 323 2009 787 (771 bp) (13; 23) (42; 80) (0.012; 0.022) (0.001; 0.044) (0.003) (0.0003) Stable 48 related groups 0.023 0.028 in this region –5.981relative to the 0.305 much more distantly presented here show that the (SP) central region 15 estation more ex- 787 www.sciencemag.org SCIENCE VOL 323of Atlantic 6 FEBRUARY 2009 (0.115) (0.221) had much higher stability relative to the south. forest endemics. tensive forests in São Paulo and southern Brazil 40 efforts, molecular 0.015 studies, (9, 0.016 –13.255 0.0001 diversity Forest lizards (14, 29) andUnstable birds (30) also show 18 Because collection 31). Not only could much unique (south ofofSP) (0.014) (0.456)could high diversity in the central portion the biome and conservation priorities have been heavily be lost, but ongoing habitat destruction relative to southern areas, and provide evidence biased toward southern and southeastern Brazil quickly erase the signature of the historical for population expansion in southern regions. (8, 9, 31), we predict that genetic diversity and processes that led to it, preventing a full underwww.sciencemag.org SCIENCE VOL corridor 323 6ofFEBRUARY 2009 in the central the standing This reassures us that the processes uncovered of the mechanisms underlying local en- 787 narrow endemism by the amphibian data may be generalized to biome have been substantially underestimated. demism and, therefore, impeding more effective and help to explain patterns of diversity in other, This is serious, given the higher rate of defor- conservation measures.

38

joint, multispecies analyses of mtDNA diversity shows that palaeoclimatic niche models and assemblage-scale molecular genetic analyses can be used to forecast spatial patterns of diversity in poorly explored, highly threatened ecosystems. In a world of ever-accelerating environmental changes, this approach can help to guide research and conservation in other global hotspots or similarly complex tropical ecosystems.

References and Notes

1. G. Hewitt, Nature 405, 907 (2000). [CrossRef] 2. F. E. Mayle, D. J. Beerling, W. D. Gosling, M. B. Bush, Philos. Trans. R. Soc. London B Biol.

Sci. 359, 499 (2004).

3. E. P. Lessa, J. A. Cook, J. L. Patton, Proc.

Natl. Acad. Sci. U.S.A. 100, 10331 (2003).

4. B. W. Nelson, C. A. C. Ferreira, M. F. da Silva, M. L. Kawasaki, Nature 345, 714 (1990).

5. C. H. Graham, C. Moritz, S. E. Williams, Proc. Natl. Acad. Sci. U.S.A. 103, 632 (2006).

6. W. Jetz, C. Rahbek, R. K. Colwell, Ecol. Lett.

7, 1180 (2004).

7. A. Hugall, C. Moritz, A. Moussalli, J. Stanisic, Proc. Natl. Acad. Sci. U.S.A. 99, 6112 (2002).

8. L. P. C. Morellato, C. F. B. Haddad, Biotropica 32, 786 (2000).

9. M. T. Rodrigues, Conserv. Biol. 19, 659 (2005).

10. C. Kremen et al., Science 320, 222 (2008).

11. Materials and methods are available as supporting material on Science Online.

12. M. B. Araújo et al., Ecography 31, 8 (2008).

13. D. R. Frost, Amphibian Species of the World:

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

Supporting Online Material

www.sciencemag.org/cgi/content/ full/323/5915/785/DC1 Materials and Methods Figs. S1 to S6 Tables S1 and S2 References 8 October 2008; accepted 9 December 2008 10.1126/science.1166955

39

Our New Darwin shirt — yours free when you become a AAAS member.

Join Today!

Special offer: $99 Professional $50 Postdoc/Student* Become a AAAS member today, at this special introductory offer and receive a free Darwin shirt (a $22.50 value free). Your membership gives you all of the personal membership benefits of AAAS including 51 issues of Science magazine, career development tools, and much more — all while supporting our international, public policy, and educational programs — the ones that advance science and serve society.

Order online at promo.aaas.org/darwin24

Visit promo.aaas.org/giftdarwin to order now *International orders will receive Science Digital edition - other membership categories are available online.