Conservation and ecology of the endemic madeira laurel pigeon - Paulo Oliveira by PARQUE NATURAL MADEIRA

CONSERVATION AND ECOLOGY OF THE ENDEMIC MADEIRA LAUREL PIGEON, COLUMBA TROCAZ

PJ Oliveira

Phd

2003

CONSERVATION AND ECOLOGY OF THE ENDEMIC MADEIRA LAUREL PIGEON, COLUMBA TROCAZ

PAULO JORGE DOS SANTOS GOMES OLIVEIRA

A thesis submitted in partial fulfilment of the requirements of the Manchester Metropolitan University for the degree of Doctor of Philosophy

Department of Biological Sciences The Manchester Metropolitan University 2003

Conservation and ecology of the endemic Madeira laurel pigeon, Columba trocaz Abstract The Madeira laurel pigeon Columba trocaz is a conservation dependent endemic species which inhabits the relict laurel forest of Madeira island (330 10’ and 300 08’ N and 150 50’ and 170 20’ W). The problems raised by the conservation of this bird are very interesting and form something of a paradox – it is a vulnerable insular endemic species but it is also a significant pest of crops. This project was designed to: 1) gather information on the status, ecology and behaviour of the species, and relate the findings to general theories and existing knowledge, especially of pigeons on islands; 2) understand and predict when and where pigeons attack agricultural crops; 3) combine the information on status, ecology, behaviour and the pest problem into a strategy for the conservation and management of the species. Biometrics of Madeira pigeons were compared to other Atlantic Island and continental pigeons, and it was apparent that the two Canary Island species, Columba bollii and C. junoniae exhibit more marked insular traits than C. trocaz. Population estimates were made using distance sampling methods. It was found that the population has increased from 2700+ in 1986 to about 10300 in 1995 and as the Pigeon’s laurel forest habitat is now very well protected, it is now not at serious risk of extinction. Habitat use studies were carried using line-transects in forest sites where the availability of fruit (the most important element of the diet) was assessed. The findings provided strong evidence that bird movements and shifts in abundance are related to fruit usage and availability. The use of crops seems to be mostly opportunistic and is governed by the birds’ movements within the adjacent forest. It is hypothesised that fruit phenology will influence the use of agricultural areas only to the extent that it governs such movements; there is no strong evidence that crops are attacked only when the availability of natural foods is low. Cabbages are the main crop attacked by pigeons, and the damage on individual fields were recorded to help understand the factors which underlie their proneness to attack. In general it was found that (i) pigeons chose those fields nearest to the forest and far from human settlements or infrastructures; (ii) they prefer to feed on open fields and on high standing items; and (iii) they do not discriminate between the centre and the margin of the fields. This information was used to create a model to understand and predict the risk of attack for each field. The predictions were successfully tested in a different area and thus the model has important implications for improving crop protection measures. Diet studies were carried out through the analyses of faecal samples using microhistological techniques. The presence of fruits in the diet was concordant with their availability, and leaves and flowers became important when fruits were scarce. It was also found that most seeds were defecated intact except the case of Laurus azorica. The species was shown to have a much wider diet than generally supposed. The findings of this research project enables a better understanding of the ecology and behaviour of the Madeira laurel pigeon and of the evolutionary biology of island species in general. They also form the basis of specific recommendations to reduce the pest problem and ensure the future survival of the species.

In memory of

Daniel, Luis Guilherme, Diogo Cardoso and Luis Monteiro.

“Here’s to the hearts and the hand of the men that come with the dust and are gone with the wind” Bob Dylan

Your dreams will forever remain alive in the souls of those who remember your short lives. There is a part of each one of you in whom I became today, and in everything I have done so far...this thesis, and all that it means, is both for you and from you, thanks!

Acknowledgements •

Thanks to everybody that helped me but don’t need to see their names here...thanks to friendship;

•

This work was financed by the author, the Parque Natural da Madeira through an EC Life project and by the Universidade de La Laguna through a grant conceded to M. Nogales.

Contents

Chapter 1. General introduction Chapter 2. Insular traits and morphological comparisons within macaronesian pigeons Chapter 3. Population trends and status. Chapter 4. Effects of fruits abundance on habitat and agricultural fields use Chapter 5. Patterns of use of agricultural fields Chapter 6. Diet and fruit resource availability: a study using microhistological analysis Chapter 7. Winter feeding on agricultural fields: a complementary study of diet Chapter 8. General discussion Chapter 9. Management recommendations. Up date of the Madeira laurel pigeon Action Plan. Chapter 10. References Appendix 1. Photographic annex

1. General introduction

1.1 Aims

To contribute to the conservation and management of the endemic Madeira laurel pigeon, Columba trocaz, by:

1. determining the relationships between habitat use, diet and the use of anthropogenic habitats;

2. examining feeding choices and behaviour on agriculture fields where the species is a serious pest.

1.2 Location of the study

The research was carried out on Madeira which is part of the Macaronesian Biogeographic Region, located in the North Atlantic, between 39o 45’ and 14o 49’ N and 31o 17’ and 13o 20 W. Macaronesia comprises the Archipelagos of the Azores, Madeira, Selvagens, the Canaries and Cape Verde (Figure 1.1). These archipelagos share a wide range of biological affinities (Baez and Sanchez-Pinto 1983) and are characterised by high levels of endemism. The Madeiran archipelago, located between 330 10’ and 300 08’ N and 150 50’ and 170 20’W, is formed by the islands of Madeira (737 Km2), Porto Santo (41 Km2), the uninhabited Ilhas Desertas (15 Km2) and the Ilhas Selvagens (3 Km2). Although the latter group of islands are politically a part of the archipelago of Madeira, biogeographically they should be considered a distinct archipelago (Figure 1.1.).

Madeira laurel pigeon Chapter 1.1

Figure 1.1. Macaronesian Biogeographic Region, located in the North Atlantic, between 39o 45’ and 14o 49’ N and 31o 17’ and 13o 20’ W

All the Madeiran islands are of volcanic origin. The first eruption took place in the Oligocene or early Miocene and on the island of Madeira the volcanic activity persisted into the quaternary. This latter activity was responsible for the actual shape of the island which has a mountainous massif with the highest point at Pico Ruivo, 1861m (MitchellThomé 1979). The climate is subtropical, oceanic and variable. A sea current flows past Madeira from the north-east and the north-east trade winds blow on an average 300 days per annum, influencing the islands’ rainfall and vegetation. In Madeira rainfall averages between 500 and 2250 l/m2 and it is greater on the northern side of the island and increases with altitude (Quintal and Vieira 1985). The vegetation differs greatly from island to island within the archipelago (Hansen and Sunding 1985). Laurel forest is found on the main island of Madeira and it is the natural habitat of the endemic Madeira laurel pigeon, Columba trocaz.

Madeira laurel pigeon Chapter 1.2

Madeira has 42 species of breeding birds; the Azores have 32, the Salvages 9, the Canaries 80 and the Cape Verdes 36. Stattersfield et al. (1998) identified two Endemic Bird Areas (EBA) and one Secondary Endemic Bird Area (SA) in Macaronesia. The EBAâ&#x20AC;&#x2122;s are Cape Verde, with 4 endemic birds, and Madeira and the Canary Islands (including the Selvagens) with 11 endemic birds, of which 2 are endemic to Madeira, 5 are endemic to the Canary Islands and 4 are exclusive to the Macaronesian Islands. The SA is the Azores with one endemic species.

1.3 Introduction to the study of island birds

The study of birds on oceanic islands has been crucial to the development of evolutionary and ecological thought (e.g. Darwin 1871, Wallace 1869, Lack 1947, MacArthur and Wilson 1967). One reason why they have attracted the attention of researchers is that islands are isolated systems, with well-defined boundaries and usually with reduced land areas which support simplified communities. Another is that on islands, the same evolutionary forces exist as on mainlands, but isolation results in the emergence of unique organisms and communities. Although island species and communities tend to be unique, they manifest their differences, both in relation to their mainland counterparts and amongst themselves, in many similar and predictable ways (Quammen 1996).

The colonisation of an island depends on the dispersal ability of the animals and plants that exist on the neighbouring portions of land. Not surprisingly, birds are amongst the first candidates to colonise any island, near or remote. Such a fact, together with the selective pressures imposed by the insular environment, turns islands into key locations for the diversity of the worldâ&#x20AC;&#x2122;s bird fauna (Whittaker 1998).

The number of species on islands may be viewed as a function of area, distance from the mainland and a balance between immigration and extinction. Distance from the mainland will determine not only the immigration rate, but also the type of species that colonise the islands (MacArthur and Wilson 1967). Oceanic islands are difficult to reach, and since

Madeira laurel pigeon Chapter 1.3

organisms of different taxa have different dispersal abilities, it is inevitable that these islands will hold a disharmonious sample of species compared to the nearest mainland. Moreover, chance is likely to play a strong role in governing which species arrive at such islands, when, and in what numbers (Grant 1998). Difficulties of establishment, e.g. because of the size of the inoculation (Berry 1996) and that fact that some species are better at establishing themselves than others, will further influence the composition of the island community. As an example of a disharmonious community, prior to the arrival of man, the only terrestrial vertebrates to be found on Madeira were 42 species of birds and 8 species of bats (Baez 1993).

1.4. Evolutionary trends of island birds

Those species that do manage to establish themselves undergo a number of similar evolutionary trends of which the following are amongst the more obvious:

(i) Broader ecological niches - Comparisons of insular bird populations to their mainland counterparts have revealed that the development of a broader ecological niche is a common trend amongst island birds (e.g. MacArthur 1972, Diamond 1975, Williamson 1981, Blondel et al. 1988, Grant 1998). This may be due to the fact that island species, freed from the constraints of mainland competitors, expand the range of their habitats or the dimensions of their feeding niches (Roughgarden 1995). This ecological release can be inferred from the measurement of ecologically significant morphological traits (Grant 1998). An example of one such trait is the fact that island birds from around the world and from a variety of taxonomic groups tend to have larger beaks (Grant 1965, 1998). Essentially, larger beaks would be selectively advantageous under conditions of low food diversity and the need to generalise diets (Williamson 1981).

(ii) Flightlessness and defencelessness - The absence (or scarcity) of predators on islands is the main cause of this evolutionary trend (Steadman 1995, Grant 1998). Another is that with the existence of permanent habitats and a local year-round food supply, there is less

Madeira laurel pigeon Chapter 1.4

need to fly from one suitable habitat to another (Steadman 1995, Grant 1998). Under these conditions the energetic costs of flight are not offset by the selective advantage of finding new food sources (or of avoiding predation). Other indications of an acquired defencelessness include the loss of protective coloration and warning mechanisms, prolonged infancy, and loss of wariness. Quammen (1996) proposes the term ecological naĂŻvetĂŠ to characterise this subset of features. According to Grant (1998), the loss of wariness occurs because, on Islands, the balance between the exploratory behaviour in search for information about the environment and for food, on the one hand, and the avoidance of dangerous organisms, on the other hand, is shifted towards the former.

(iii) Reduced birth rate - This evolutionary trend appears to be general among terrestrial vertebrates (Grant 1998). This may reflect an evolutionary shift away from high reproductive rate and short life span, towards the opposite combination of life history traits. An important determinant in such a shift is intraspecific competition under conditions of sustained high densities in the absence (or scarcity) of predators and heterospecific competitors (Boyce 1984).

(iv) Size changes - Although this is a quite generalised trend, to account for the size changes is not a simple task, as it involves an interaction of food resources, competition, and predation (Blondel 1985). Quammen (1996) states that in birds, though in some cases they are smaller, they are more often larger on Islands than elsewhere. Blondel (1985) suggests the opposite; that island species tend to be smaller - with exceptions for carnivores and flightless species. An extreme example of giantism was seen in the moas of New Zealand, the cassowaries of Australia and New Guinea, and Aepyornis from Madagascar

(v) Speciation - Islands are characterised by high levels of endemism (Quammen 1996). An example from the Atlantic Islands is that 40% of the Chaffinch, Fringila coelebs, subspecies are found there yet these islands represent much less than 1% of the species distributional range (Hagemeijer and Blair 1997).

Madeira laurel pigeon Chapter 1.5

(vi) Adaptive radiation - Island birds provide good examples of radiation, the most famous being the Galapagos finches (Grant and Grant 1997). However, according to Quammen (1996) the Hawaiian honeycreepers are an even better example. These belong to the endemic sub-family Drepanidae, which includes about 30 species. The correct number is somewhat uncertain because of both, taxonomic disagreements and loss by extinction.

1.5. Extinctions and island birds

Island birds are more vulnerable to the threat of extinction than their mainland counterparts. According to Diamond (1984), since 1600, 171 species and subspecies of birds have become extinct, of which 155 (90%) were island taxa. To illustrate this, the same author states that 24 species and subspecies were lost from Hawaii, 8 from the islands around Baja California, 13 from the Mascarene Islands, including Reunion and Rodrigues, as well as Mauritius (besides the well-known Dodo) and 15 from the West Indies. More species and subspecies of birds were lost from the Lord Howe Islands, a small group of Islands located between Australia and New Zealand, than the combined total of those lost in Africa, Asia and Europe.

The Macaronesian Archipelagos have lost 2 species and 2 subspecies from the Canaries (Emmerson et al. 1994) and at least one subspecies from Madeira. In these archipelagos (and particularly in the Cape Verde (Hazevoet 1997)) many species have become restricted to a few islands and face local extinction.

The numbers of species lost from islands become even more significant when we consider that only twenty percent of the worldsâ&#x20AC;&#x2122; bird species are confined to Islands (Quammen 1996). This suggests that an island bird faces about a fifty times greater likelihood of extinction than a mainland bird. Furthermore, three quarters of the insular extinctions have occurred on small islands, which implies that species from smaller islands are much more prone to extinction than those of bigger islands . Also, birds of remote islands are more

Madeira laurel pigeon Chapter 1.6

vulnerable than those of islands close to continents or to other substantial portions of land (Diamond 1984, Whittaker 1998).

The conclusion is that, as a rule, insular isolation and the size of the islands are important correlates of extinction proneness but the causative factors need further analysis. It is well known that all populations fluctuate in size over time under the influence of two kind of factors: stochastic and deterministic. Deterministic factors are those involving straightforward cause-and-effect relations, which can be predicted and controlled to some extent. Essentially we can include in this group all consequences resulting from human induced disturbances such as destruction of habitats, introduction of exotic animals that compete with or prey on native species, introduction of diseases, hunting and trapping, amongst many others. Because they are predictable, rational and subject to control, in theory, such factors can be eliminated (Thomas 1994, Quammen 1996, Whittaker 1998).

Stochastic factors are those that operate beyond human control and that are truly random, or because they are linked to such complex geophysical or biological causes that they seem random. For instance, the weather patterns: a hard winter or droughts are stochastic factors that cannot be predicted nor can they be controlled.

Considering that populations of insular birds can fluctuate as a result of one or both of these two factors, or of both of them acting together, it becomes obvious that the smaller the population, the higher the probability that the fluctuation will decline towards zero, which is to say, towards extinction. Thus, it is easy to understand why small islands, with small areas and smaller populations present higher risks of extinction. Perhaps as an evolutionary response to such a problem, island birds evolved a tendency to form denser populations compensation by density - (MacArthur et al. 1972, Blondel et al. 1988). Nevertheless, this adaptation is apparently not sufficient to balance the threats imposed by the deterministic factors.

Insular isolation gives rise to yet another sort of predicament directly linked to the evolutionary trends shown by island birds. The remoteness of the islands is directly related

Madeira laurel pigeon Chapter 1.7

to the ecological isolation of the species. The remoter the island, the greater the probability that the birds will lose their defences against mainland predators, competitors and diseases; they will thus be particularly susceptible if any of these are introduced by man. Specialisation to exploit resources which might be abundant but low in diversity also renders island species susceptible to rapid and drastic changes in their habitats such as those engineered by man.

Specialisations such as flightlessness and gigantism with the subsequent loss of dispersal ability, are typical examples of how birds become vulnerable and unable to respond to the disruption of their habitats. Quammen (1996) presents a list of bird extinctions as a result of low dispersal ability caused by such trends. The Great auk, now extinct, was an â&#x20AC;&#x2DC;overgrownâ&#x20AC;&#x2122; member of the same sea bird family as puffins and murrelets. Native to Islands of the North Atlantic, the Great auk might have escaped the total eradication by humans if it were not for the fact that, unlike puffins and murrelets, it was a flightless bird. Cuba had its giant flightless owl, Ornimegalonys oteroi; Madagascar the Elephant bird Aepyornis maximus; and New Zealand had at least twelve species of moas.

Besides the introduction of predators and competitors (among which man is included), SoulĂŠ (1983) presents the following classic consequences of human actions upon birds on islands: destruction and degradation of habitats; introduction of diseases, for which the species do not have natural defences; and extinction of mutualistics populations, to mention just a few. Diamond (1984) points out the disruption of trophic chains as another important factor leading to the extinction of many island species.

Most of the time extinction occurs as a consequence of joint action of both, stochastic and deterministic factors. For example, a hard winter will have a greater effect on a small population that is traditionally hunted than on its counterpart that is free from this type of deterministic pressure.

The general trend for island species to become specialised and differentiated and at the same time more extinction-prone was formalised by Wilson (1961) and Ricklefs and Cox

Madeira laurel pigeon Chapter 1.8

(1972, 1978) in the taxon cycle hypothesis. This idea has been examined in relation to some of the Macaronesian islands by Jones et al. (1987). The basis of the idea is that patterns of taxonomic differentiation and geographical distribution of birds suggest that taxa may pass through phases of increasing and decreasing geographical range or taxon cycles. Although decline may often lead to extinction, new phases of expansion, accompanied by increased or shifted habitat distribution are also possible. The taxon cycle is an hypothesis concerned with historical events, but its existence was formerly inferred from contemporary, indirect evidence. Modern molecular studies of land birds now provide relative ages for taxa based on genetic differentiation among island populations. A recent study carried out in the West Indies (Ricklefs and Bermingham 1999), shows that these age estimates confirm that older populations tend to have restricted geographical and ecological distributions. That ecology and geography are strongly correlated with age across taxa also suggests that the time course for evolutionary change through the taxon cycle is relatively consistent among independently evolving populations. Because young taxa have normally more continuous distributions within the islands, gaps in the ranges of older taxa indicate extinctions of island populations. It is obvious that man has very disruptive influence on these cycles.

Recent work by Jones et al. (2001) - although not directly supporting the predictions of the taxon cycle hypothesis - has suggested that the history of a species within an archipelago, rather than ecological traits, has much more influence on local abundance and therefore extinction risk. In this study it was the endemic species and subspecies which were most abundant in the indigenous forest habitats, suggesting that the greater risk of extinction experienced by island birds maybe related to deterministic factors, without which they could survive for a much longer time. This is also backed up by the observation that islands support many relict (and therefore long-lasting) species and even whole ecosystems (e.g. the laurel forest).

Madeira laurel pigeon Chapter 1.9

1.6. Pigeons on Islands

The Columbiformes, with 316 living and recently extinct species (Gibbs et al. 2001), is the 4th largest bird order after Psittaciformes, Apodiformes and Passeriformes. This indicates that biologically the pigeons and doves are a successful group. In terms of body form they have remained somewhat generalised but have still undergone a considerable adaptive evolution enabling them to survive in a wide range of conditions (Murton 1965).

The typical pigeon is a bird with strong wings and a diet of seeds and/or fruits for which they often make seasonal migration and are also predisposed to transoceanic journeys, ranking high on the list of good travellers (Quammen 1996). True pigeons and pigeon descendants are, disproportionately, well represented on some of the most remote islands. They constitute one of the most widespread and conspicuous families of landbirds on the tropical Pacific Islands (Steadman 1997). New Guinea and its surrounding Islands harbour 45 species, roughly one sixth of the world's total (Quammen 1996). Many endemic species and sub-species can be found in Islands all over the world, and examples can be found in the Indian, the Pacific and the Atlantic Ocean: Samoa has the Tooth-Billed Pigeon and the White-throated Pigeon; Palauhas houses the Nicobar pigeon and the Palau ground-dove; SĂŁo TomĂŠ, a small island offshore from West Africa, supports five species of pigeons; Anjouan, in the Indian Ocean, North of Madagascar, also has five different pigeons; the Macaronesian Islands of Madeira and the Canaries have three endemic pigeons. (Bannerman and Bannerman 1963, 1965)

Stattersfield et al. (1998) identified 138 species of pigeons and doves amongst the 2623 bird species with a restricted range (defined by the author as species having a breeding range of 50.000 Km2 or less) throughout historical times. There are more restricted range species in the pigeonâ&#x20AC;&#x2122; family than in any other. As might be expected, most of these species (a total of 120 species, 86 % of the total) live on islands (not considering Australia).

Madeira laurel pigeon Chapter 1.10

According to the same author, following the IUCN Red List Categories as applied by Collar et al. (1994), 40 species (33% of the total), are classified as threatened, 10 as Critical, 5 Endangered and 25 Vulnerable.

In the last 150 years, 6 island pigeons have become extinct in the wild and many more have become relict in one or two restricted places. These extinctions have occurred all over the world: Columba versicolor and Columba jouyi in a group of small Islands near Japan; Zenida graysoni also in the Pacific but near the Mexican Coast; Microgoura meeki in the Solomon islands; Pitilinopus mercierii in the Marquesas Islands (French Polynesia); and Alectroenas nitidisasima in Mauritius. The other Mascarenes, RĂŠunion and Rodrigues, in much earlier times, each harboured a large, flightless species of solitaire (Ornithaptera solitaria on RĂŠunion, Pezophaps solitaria on Rodrigues) which, like the Dodo, Raphus cucullatus, had pigeon affinities and are also now extinct (Quammen 1996).

Steadman (1997) suggests that all the extinct Polynesian pigeons have in common the fact that they were forest birds and that their extinction was related to anthropogenic factors loss and degradation of habitats, hunting, persecution, as well as to introduced competitors and predators. This applies to most, if not all, pigeons that have become extinct worldwide (e.g. Stattersfield et. al. 1998).

The pigeon species listed as Critically Endangered and Endangered in Collar et al. (1994) are all affected by exactly the same types of threat. Eleven species face loss or alteration of habitat, 9 are affected by hunting and persecution and 5 by introduced competitors and predators. Furthermore, in thirteen of the cases, they are species with a small range or small population density, which are extremely vulnerable to stochastic factors.

1.7. Macaronesian pigeons.

On the Macaronesian Islands the genus Columba is represented by five species all of which are considered to be either recent or rather more ancient derivations of C. palumbus

Madeira laurel pigeon Chapter 1.11

(Murton 1965, Goodwin 1985). In the Azores there is an endemic subspecies, Columba palumbus azorica (Bannerman and Bannerman 1963). In Madeira and Canary Islands there are 3 endemic species, Madeira laurel pigeon, Columba trocaz, in the former, and the Dark tailed laurel pigeon, Columba bollii, and White tailed laurel pigeon Columba junoniae in the latter. On the island of Madeira Columba palumbus maderensis became extinct towards the beginning of the last century (Bannerman and Bannerman 1965).

The three endemic species are considered to be ancient elements of the Macaronesian avifauna, which have co-evolved with this unique habitat (VolsĂ¸e 1955). After the precursors of the Madeira laurel pigeon reached Madeira and evolved independently of the mainland stock a further wood pigeon invasion occurred which gave rise to C. p. madeirensis (Murton 1965). This subspecies was darker than mainland forms but is now extinct. A detailed description of the Madeira laurel pigeon will be given in the next chapter but the Canarian counterparts are briefly described here.

The Dark-tailed laurel pigeon, Columba bollii, is endemic to the Canary Islands, and is found in the laurel forest of Tenerife, La Palma, La Gomera and El Hierro (Gonzalez 1996a, 1996b) but may have had a wider distribution throughout the archipelago (Hernandez and Martin 1994, Martin and Lorenzo 2001). In 1994 a census estimated the total population of the archipelago to be about 1700 individuals (Martin et al. 2000). It is a fairly shy species that inhabits the relict laurel forest of Canary Islands. The Dark-tailed laurel pigeon feeds mainly on fruits taken from the ground and trees, but like C. trocaz, it is also known to feed on cultivated areas on cabbages and other crops (Gibbs et al. 2001 and Aurelio and Lorenzo 2001).

Although closely resembling C. trocaz, it is rather smaller and with a finer bill; unlike C. trocaz it has lost the white neck mark and the purple and green iridescent neck plumage is more extensive.

The White-tailed laurel pigeon Columba junoniae is also endemic to the Canary Islands. It was formerly thought to inhabit only La Gomera and La Palma, until it was discovered on

Madeira laurel pigeon Chapter 1.12

Tenerife in the 1970’s (Hernandez and Martin 1994; Gonzalez 1996a). In 1985 a census estimated the total archipelago population to be about 1480 birds (Martin et al. 2000). It is also a fairly shy species associated with the laurel forest. It feeds on fruits of the laurel forest trees. Although recorded as feeding on the ground in association with C. bollii, it seems less inclined to do so (Gibbs et al. 2001) and it also feeds on agricultural fields (Aurelio and Lorenzo 2001).

Both Canary species are threatened by the problems which classically affect island birds, namely habitat loss and degradation, and predation of eggs and nestlings by rats (Hernandez and Martin 1994, Gonzalez 1996a, 1996b; Hernandez et al. 1999).

1.8. The Madeira laurel pigeon.

1.8.1. Introduction

This bird was first described by Heineken (1829) in the Edinburgh Journal of Science and he tentatively proposed the name of Trocaz’, its local name, as a specific designation. Harcourt (1851) pointed out that the middle toe of this pigeon was longer than that of the ‘ring – dove’ and refered to it as the Long – toed pigeon.

Many other ornithologists such as Ogilvie (1890), Hartwig (1891, 1893) Schmitz (1893, 1894, 1895, 1900, 1903a, 1903b, 1909), Meinertzhagen (1925), Bannerman and Bannerman (1965) and Zino (1969) have all described their observations of this bird. Even though there have been more recent studies, (e.g. Zino 1986, Jones 1990, Oliveira and Jones 1995) the species is still poorly known. This is because its habitat is remote, it is often difficult to find and, because of the nature of the terrain, even more difficult to follow and study.

Madeira laurel pigeon Chapter 1.13

1.8.2. General field description

The Madeira laurel pigeon is a fairly recent derivative of the wood pigeon, Columba palumbus, (Goodwin 1985) but it is much darker (Cramp 1985) with dark blue-grey back and wings and its vinous under-parts have deepened in colour and spread as far as the under tail (Murton 1965). The white wing bars have been lost which is typical of isolated island species, which apparently lose such signal marks when reproductive isolation is no longer important (Blondel 1985). However, the white neck mark has extended, forming a complete area of white over the top of the neck. The iridescent plumage of the neck seems to have suffused into the white neck mark which, as a result, has grey and green undertones and appears much duller than in C. palumbus. The bill is red, the eye is pale yellow with a red orbital ring. There are no identification problems as no other large and dark pigeon inhabits Madeira. There is little sexual dimorphism, the males being rather more corpulent than the females (Zino and Zino 1986), but significant differences only found for wing and tarsus (Oliveira and Jones 2001). Juveniles are easily separable (Pers. obs.) as they are duller and of a deeper brown than the adult, lacking glossed plumage and showing pale margins to larger body and inner wing feathers (Cramp 1985).

1.8.3. Conservation Status and Distribution

The Madeira laurel pigeon is endemic to the Island of Madeira although it may have occurred in the past on the neighbouring island of Port Santo (Pieper 1985). It was first listed as Rare in the African Red Data Book (Collar and Stuart 1985) and in the IUCN Red List of Threatened Animals (Groombridge 1993). Although listed by Collar and Andrew (1988) as globally threatened, due to its small range, specialised habitat and small population, the Madeira laurel pigeon has been re-classified as Conservation dependent in Birds to Watch 2 (Collar et al. 1994). This was largely a consequence of the success of the conservation measures recently enacted. In Endemic Bird Areas of the World (Stattersfield et al. 1998) and in Threatened Birds of the World (Stattersfield and Capper 2000) it has

Madeira laurel pigeon Chapter 1.14

also been classified as Conservation dependent. In the first Portuguese Red Data Book it is listed as Vulnerable (Cabral et al. 1990) and in the most recent update of this publication, still in the early stages of preparation, it will most probably keep this status (Almeida J. pers. com.). Recently Oliveira (1999a) in â&#x20AC;&#x153;Management and Conservation status of the Birds of Madeira Archipelagoâ&#x20AC;? has confirmed that it should be considered as Conservation dependent.

Currently the Madeira laurel pigeon is restricted to areas of native forest on the mountainous northern slopes and a few isolated pockets in the south of the island (Figure 1.2. ). It was probably exceptionally plentiful before the first settlement of the island (in 1420) but the laurel forest, is now reduced to ca. 15 % of its original distribution, and overhunting may also have contributed to a severe decline. Population size has been estimated at 3500 to 4900 individuals (Oliveira and Jones 1995) and is apparently increasing but the species may be prone to considerable yearly fluctuations (Zino et al. 1994).

Figure 1.2. Map of Madeira Island showing the present distribution of the Laurel forest which is coincident with the distribution of the Madeira laurel pigeon (adapted from Neves et al. 1996)

Madeira laurel pigeon Chapter 1.15

1.8.4. Ecology of the Madeira laurel pigeon

The Laurel Forest Habitat

The island of Madeira has 16.000 ha of laurel forest (ca. 15% of its total area) which is an indigenous, evergreen and relict forest (Neves et al. 1996). Laurel forest was once common in southern Europe, in particular on the northern edge of the Mediterranean, and the Atlantic islands. Due to the climatic changes that occurred during the last glacial era, this forest disappeared from continental Europe and prevails only in the Atlantic Islands of Madeira, Azores and the Canaries. The difference in the geography of those archipelagos, allied to different climatic conditions and, more recently, to human impact, has produced different laurel forest types in each area.

In the Azores, apart from differences in the composition of the species and relative abundance (e.g. absence of Ocotea foetens and Appolonias barbujana) the laurel forest was dramatically reduced due to the transformation of the land into pastures and timber plantations. At present, this biotope can only be found in remote and small patches on the Island of S達o Miguel (Serra da Tronqueira) and on the island of Terceira (Haggar et al.1989).

In the Canary Islands, only La Palma, Hierro, Tenerife and La Gomera hold reduced areas of laurel forest. The latter island holds the most important area of forest, the Garajonay National Park, which was classified as a Natural World Heritage Site in 1986. The park occupies 3,948 ha. of which not all is occupied by laurel forest (Perez de Paz 1990).

In the Island of Madeira laurel forest is mostly confined to the north-facing slopes of the island, occurring between 300 m and 1300 m, though there are also some isolated pockets to the south (Neves and Valente 1992). A recent study (Neves et al. 1996) showed that the Island of Madeira has the largest and best preserved continuous area of laurel forest in the world. This same study showed that the Til, Ocotea foetens, is the dominant tree species in the forest. The Til is a tree that can reach 30 m in height and prefers deep humid and

Madeira laurel pigeon Chapter 1.16

sheltered valleys (Neves et al. 1996). Its seasonal fruiting patterns have never been studied in detail but flowers and fruits can be found all year round (Press and Short 1994, Oliveira and Jones 1995). The Loureiro, Laurus azorica is smaller (up to 15m), but numerically dominant, occurring from the sea level up to 1700 m high. Although it blooms and fruits all year round, it has a very strong fruiting peak between August and December. Other trees that were found to be relevant in terms of the structure and composition of the laurel forest are the Faia, Myrica faya, the Vinhatico, Persea indica, the Folhado, Clethra arborea and the Barbusano, Apollonias barbujana (Neves et al. 1996).

Besides the Laurel pigeon, the laurel forest supports a very important collection of birds including the Madeiran Firecrest, Regulus ingnicapillus madeirensis, and the Madeiran Chaffinch, Fringilla coelebs maderensis. Whilst currently considered sub-species, because of their marked morphological, ecological and biological differences in relation to their mainland counterparts, it is believed that further studies may support their elevation to specific status (Oliveira 1999a).

The laurel forests support many endemic species of molluscs, millipedes and insects (Baez 1993) and also play a key role in the hydrological balance of the Island of Madeira, providing vital water for agriculture and human consumption.

Habitat Use and Diet

There is strong evidence that the pigeons move from one laurel forest valley to another throughout the year (Oliveira and Jones 1995). A habitat selection study, carried out in summer, showed a strong preference for laurel forest at low altitudes, especially dense canopied forest on steep slopes (Jones 1988) and in places not altered by housing and farming, pigeons may occur very near the coast (Zino and Zino 1986, Jones 1990, pers. obser.). Another more intensive study carried out in three different valleys showed that the biotopes preferred all year round are those dominated by the Til, Ocotea foetens (Oliveira and Jones 1995). Nevertheless those areas with exotic vegetation on the boundaries of the native forest are also used extensively all year round. Pigeons occur within the whole

Madeira laurel pigeon Chapter 1.17

altitudinal range of the laurel forest but they show a definite preference for forest under 850 metres (Oliveira and Heredia 1996).

Since the vegetation of Porto Santo was considerably different from that found in Madeira, the existence of pigeons on the former probably means that before the colonisation of these islands by man, this bird had inhabited several other types of habitat.

During the winter and first half of spring the birds move to agricultural areas to feed. These movements out of the forest may be associated with a lack of natural food (Zino and Zino 1986) or with the existence of predictable food sources in close proximity to the pigeons' natural habitat (Oliveira and Jones 1995).

Most of the literature on the diet of the Madeira laurel pigeon suggests that the main food source is the fruits of the various trees of the laurel forest, namely Loureiro, Laurus azorica, Til, Ocotea foetens, Faia, Myrica faya, and Vinhático, Persea indica (e.g. Harcourt 1851, Godman 1872, Bannerman and Bannerman 1965, Zino 1969, Zino and Zino 1986). The latter authors state that out of these, the Loureiro, is by far the most common and, thus, the most important. Fruits are taken both in tree canopy and on the ground (Cramp 1985). When feeding on the ground they also eat the flowers and leaves of many plants such as Serralha, Sonchus spp., Rabaça, Apium nodiflorum, Agrião, Nasturtium officinale (Zino and Zino 1986) and Cabreira, Phillys nobla, (Oliveira and Jones 1995). In spite of the existence of many references concerning qualitative and descriptive aspects of the diet, nothing is known about the relative role and importance of each food type in the bird’s diet.

When feeding occurs on agricultural land the greatest damage is inflicted on cabbage crops. This does not necessarily represent a preference, since this is the most common and widespread crop. Pigeons also feed on watercress, young plants of beans and peas and have a considerable impact on fruit trees, eating the young shoots and flowers of walnut trees, apple trees, peach tree, cherries and a number of other species (Oliveira 1999b).

Madeira laurel pigeon Chapter 1.18

Reproduction

Little is known about the breeding cycle of the bird. According to an old record from Schmitz (1893) it breeds at all times of the year. On surveying the literature on Madeiran ornithology Bernstrom (1951) concluded that eggs had been found in all months except January, March, June and December. It is very probable the birds also breed in all of these months - for example, the author has found eggs in December. Apparently the Madeira laurel pigeon has an inherently low reproductive rate. The clutch size is usually one (Schmitz 1894, Bannerman and Bannerman 1965, Zino 1969) and although occasionally two eggs can be laid (Zino 1969) no nest with two chicks has ever been recorded in the literature (Oliveira and Heredia 1996). Nothing is known in terms of the capability of raising more than one chick per annum.

Nests are built with dry twigs in a forest tree, shrub or on sheltered cliff ledges (Schmitz 1894, Bannerman and Bannerman 1965, Zino 1969) and occasionally on the forest floor (Zino 1969). This latter author states that incubation takes from 19 to 20 days followed by a fledging period of up to 28 days. Captive breeding is apparently easily achieved; recently two nestling birds were seen in captivity in two distinct locations of the north of the island (Boaventura e Faj達 da Nogueira) (Pers. Obser.) and this may be a common practice.

In contrast to the Canaries, where pigeon nests can be quite easily found (Aurelio Martin pers. com.), the nests of the Madeira laurel pigeon are very difficult to locate. Schmitz (1887) mentions that he had to search for over one year to find his first egg. Zino and Zino (1986) also gives an account that Schmitz, on other occasion, spent 3 years of unsuccessful searching to obtain eggs of the Madeira pigeon.

Data presented by Zino and Biscoito (1993) suggest that there might be a relationship between availability of fruits and breeding activity. Although this conclusion is based on circumstantial evidence, this would be in accordance with what happens in most fruit pigeons (Clout et al. 1995).

Madeira laurel pigeon Chapter 1.19

1.8.5. Threats and Limiting Factors

Habitat loss and degradation

Madeira was largely covered by laurel forest when discovered in 1419. Since then, much of it has been cleared and its valuable timber used for housing and other human infrastructures (Neves and Valente 1992). This habitat loss presumably caused a major decline in the abundance of the pigeon. Laurel forest now covers only ca. 15% of the Island but since 1982 nearly all areas of forest have been included in ‘Strict’ and ‘Partial Natural Reserves’ under the jurisdiction of the Natural Park of Madeira. In many places outside or still inside the protection zones the forest is recovering through natural regeneration on abandoned agricultural land.

Pest status

The Madeira laurel pigeon is a very unpopular bird amongst the local people because of the damage it causes to crops, especially on the north of the Island where subsistence agriculture is carried out in very small fields. The lack of an organised farming activity makes this type of problem very difficult to solve. The farmers do not cooperate with each other in the search for a solution, and this makes any possible action very costly for the government. Many birds are illegally killed and the “pigeon problem” has a negative influence on the enforcement and management actions undertaken by the Parque Natural da Madeira (Madeira Natural Park) (Pers. Obs.). In 1985 due to the pressure under which the farmers put the politicians, a special legal shooting period was established, covering five consecutive Sundays in January and February. It is probable that over 300 birds were shot. One party of four guns shot 64 birds in four days, in just one place, Chão da Ribeira, and over 140 were shot in this same valley during the five Sundays. More recently, since 1997, the government was again forced to allow the shooting of the pigeons feeding on the agricultural fields. The difference nowadays is that the staff of the Madeira Natural Park control the process and they are the only agents with permits to kill the birds. This is an

Madeira laurel pigeon Chapter 1.20

emergency measure to avoid an even worse consequence, but a total of 500 birds have already been killed.

Illegal hunting and poisoning

Hunting was prohibited in 1989 but still occurs illegally in a few well-defined areas, especially when pigeons leave the forest to feed on agricultural land. Poisoning is also illegal but occurs widely throughout the island. In 1985 it was estimated that between 150 and 200 birds were killed by poison (Zino and Zino 1986).

1.8.6. Conservation measures

Species and Habitat protection

The Madeira laurel pigeon is in Annex I of the European Unionâ&#x20AC;&#x2122;s Wild Birds Directive, which is incorporated into Portuguese law by the Decreto-lei 75/91. Following its inclusion in this European directive, hunting became illegal in 1986. Five of the important bird areas of Madeira in which the Laurel pigeon occurs (Grimmet and Jones 1989) are designated as Special Protection Area under Article 4 of the same European directive. The species is also included in Annex III of the Bern Convention, which is also incorporated into Portuguese law by the Decreto-lei 316/89.

Almost the entire pigeon population is included within areas controlled by the Parque Natural da Madeira (PNM) This Natural Park, fully established in 1993, includes practically all the laurel forest with a legal status of Strict Nature Reserve and Partial Nature Reserve. Moreover, the forest is also protected by other legislation concerning forests in general. Therefore the cutting of tree and other land uses are carefully controlled or even completely forbidden.

Madeira laurel pigeon Chapter 1.21

The areas of laurel forest within the Natural Park of Madeira have been designated as a Biogenetic Reserve by the Council of Europe and in 1999 they were classified as a Natural World Heritage Site by UNESCO.

To try and prevent crop depredation and also the killing of the pigeons, some experiments have been carried out on methods of scaring pigeons away from sensitive areas. In 1989 a bird scarer, a bright red and white doll which is inflated at regular intervals and producing a piercing sound audible at a distance of about 200 meters away, was placed on a property where about 650 walnut tree had been planted. The scarer was immediately successful and no pigeons were seen on the field after it started operating (Biscoito and Zino 1989). In 1993 the Natural Park used five other scarers, also with some success (Oliveira 1999b). The problem with these devices was the relation between price and longevity. The very wet and humid conditions under which they had to operate destroyed the electronics of these scarecrows after little more than one year of use.

Nowadays two different methods are being used: sound scarers and â&#x20AC;&#x153;exclusion netsâ&#x20AC;?. Although both methods have proved to be very effective, the farmers often refuse to use them.

1.9. Background to the study

Despite the long established interest of ornithologists in the Madeira laurel pigeon very few detailed studies of the species ecology, biology and behaviour have been carried out. This is mainly a consequence of the remoteness of the laurel forest and the secretive habits of the bird.

Being an endemic bird with a restricted distribution confined to a relict forest, the Madeira laurel pigeon is an inherently vulnerable and potentially threatened species. Detailed information on many aspects of its ecology, biology and is therefore needed before adequate measures for the management and conservation of the species and its habitat can be undertaken. Madeira laurel pigeon Chapter 1.22

Such task becomes even more relevant and interesting if we consider the fact that the pigeon is something of a paradox: it is a vulnerable endemic species but it is also a significant pest. Again, an adequate strategy to deal with this situation can only be developed from the basic information mentioned above and in combination with detailed information on the pest problem .

This study was thus designed with a bearing on the importance and uniqueness of the species and the need to gather information that would allow the build up of an adequate conservation strategy. Therefore, special attention was paid to the relationship between habitat, agricultural fields use and diet. These data would also hopefully contribute to the management and conservation of island pigeons in general and to the understanding of the extinction process on islands.

Madeira laurel pigeon Chapter 1.23

2. Insular morphological traits and comparisons within Macaronesian pigeons

2.1. Introduction

Islands tend to produce unique species and communities (Grant 1998). Ecologically islands usually have relatively low biodiversity and, since the seminal work of MacArthur and Wilson (1967), the number of species present may be viewed as a function of area, distance from the mainland and a balance between immigration and extinction (for details refer to Chapter 1)

Island animals are often identifiably different from their continental counterparts, whatever taxonomic level is used to describe them. Moreover it has been asserted that a number of evolutionary trends have developed many times in island birds (Hounsome 1993, Quammen 1996, Grant 1998). Collectively Blondel (1985) has described these features as the Island Syndrome which at a species level, includes the following characteristics: shorter and more rounded wings; longer bill and tarsi; smaller size (except for carnivores and flightless species, which may be larger); increased biometric variability; duller, less distinctive plumage; and less sexual dimorphism. With a broader view Quammen (1996) refers to these characteristics as the species attributes of the Insular Menu which includes: loss of dispersal ability; loss of defensive adaptations; size change; endemism; relictualism; archipelago speciation; and adaptive radiation.

Most of the items listed at the Insular Menu are no more than a consequence of the Island Syndrome characteristics as presented by Blondel (1985). It is easy to understand, for example, that the loss of dispersal ability and the loss of defensive adaptations may be viewed in direct relationship to the shorter and more rounded wings presented by the island birds.

The differences, or similarities, amongst populations of island birds provide a very interesting field of investigation. Assuming that the extent of the differences is likely to be proportional to how closely related the populations are, one can both answer

Madeira laurel pigeon Chapter 2. 1

questions about the relationships that they bear to each other (Hounsome 1993), as well as identify some of the changes induced by isolation on islands.

Probably no other group of birds is more interesting, from this point of view, than the Columbidae. They are very mobile and adapt easily to different environments, which is the reason for their success and for them being extremely widespread and present on many islands all over the world (Murton 1965, Goodwin 1985, Cramp 1985, Quammen 1996, Stattersfield 1998).

The aim of this chapter is to identify the insular characteristics of the three endemic Macaronesian pigeons. Biometric data published elsewhere will be compared, using C. palumbus as a reference. Although far from being an extensive analysis this contribution can be useful both to understand their taxonomic place and for the general knowledge of the species zoogeography. Nowadays there are much more appropriate approaches to draw evolutionary conclusions than these classic morphometric methods, thus, the emphasis here is on the identification of Insular Syndrome traits as well as the assessment of inter-specific relationships.

2.2. Relationships between the three Macaronesian endemic pigeons

It is widely accepted that the three Macaronesian endemic pigeons are derivatives of Columba palumbus (Murton 1965, Goodwin 1985). In the past C. trocaz and C. bollii were considered as belonging to the same species (e.g. Murton 1965). Goodwin (1959) was the first to give them a specific rank, which was corroborated later by most modern authors (e.g. Cramp et al. 1985). Recently this was confirmed by DNA analyses (M. Nogales pers. com.). By using classical biometric analyses, Hounsome (1993) had already shown that the three endemic species are indeed separate. Moreover, he also suggested that the two from Canaries are closer to each other than C. bollii to C. trocaz. Figure 2.1. is taken from Hounsome (1993) and shows the relative position of each of these species.

Madeira laurel pigeon Chapter 2. 2

Figure 2.1. Dendogram of the relationship between the macaronesian pigeons and their possible relatives. The dendogram was constructed using the Euclidean distances between the group centroids. It can be interpreted as representing the degree of physical similarity between the groups, where: p=C. palumbus; u=C. uncinata; t=C. trocaz; b=C. bollii; and j=C. junoniae (C. palumbus and C. uncinata are out groups) (Housome 1993)

The relationship between C. trocaz, C. bollii, C. junoniae and the two endemic subspecies from the Azores C.p. azorica and and Madeira C.p. madeirensis (now extinct), was never investigated. Presumably the Madeirean subspecies evolved from a second successful invasion of C. palumbus to this island. This is likely to have happened when C. trocaz had already evolved and could no longer inter - breed with the mainland stock (Murton 1965).

Also nothing is known about the sequence and temporal relationships of the earlier invasions of Madeira, the Azores and the Canaries by C. palumbus. However, in the light of the work of Housome (1993) one might speculate that the birds from the Canaries were the first to arise, followed by the only Madeirean true species. Ongoing investigation based on DNA analyses will provide further information.

2.3. Macaronesian endemics and the insular syndrome.

Biometric data on the three Macaronesian species is not very abundant and is actually too scanty to warrant detailed analysis at this stage. However, Table 2.1. presents a compilation of C. bollii and C. junoniae measurements collected from museum specimens and presented by Cramp (1985), Emmerson (1985) and Martin et al. (2000)

Madeira laurel pigeon Chapter 2. 3

Measurements of recently dead Madeiran pigeon were collected by Zino and Zino (1986) and by the author (Oliveira and Jones 2001 and unpublished data). Biometrics of C. palumbus as presented by Gibbs et al. (2001) was used as reference. Except for the measurements collected by the author original data were not available and a detailed descriptive analysis was not possible. C. bollii

C. junoniae

C. trocaz*

C. trocaz**

C. palumbus

Body length

350 –3 70

370 – 380

454 – 495

445 – 500

400 – 445

Weight

340 – 380

Not available

390 – 440

400 – 530

471 – 568

Wing

194 – 228

200 – 235

230 – 259

236 – 260

240 – 254

Tail

145 – 159

149 – 171

165 – 184

Bill length

17,9 – 22,1

17,7 – 22,0

18,1 – 22,7

18 – 24

19 – 22

Tarsus

25 – 33

30,0 – 38,0

43 – 49

32 – 42

25 – 28

Middle toe

37 – 42,3

38,5 – 45,6

47,0 – 55,1

138 – 149

Table 2.1. Biometric data (in mm. and gr.) from the endemic pigeons of Macaronesia and C. palumbus (from which all are derivates). Data for the Canarian birds are presented by Cramp (1985), Emmerson (1985) and Martin et al. (2000). Data for the Madeirean bird are presented by (*) Zino and Zino (1986), except bill length which is taken from Martin et al. (2000), and (**) Oliveira and Jones (2001). Data for C. palumbus are presented by Gibbs et al. (2001).

It can be seen that the endemic pigeons from the Canaries are considerably smaller than C. palumbus, while C. trocaz is about the same size – higher total length but lighter. Many authors have argued that, with the exception of flightless birds, island species are smaller than elsewhere (e.g. Blondel 1985), while others have stated the opposite – that island species are bigger (e.g. Quammen 1996). According to Grant (1998 and references therein) a general explanation for the size trends is not simple and involves an interaction of food resources, competition and predation.

Bill measurements suggest that the bill of the Canarian birds is proportionally slightly bigger than that of C. palumbus; while the bill of C. trocaz shows a very small tendency to be bigger. This possible increase in size of the bill shown by the Macaronesian species is in agreement with the idea that island species tend to have bigger bills (e.g. Grant 1965, 1998). Essentially, larger beaks would be selectively advantageous under conditions of low food diversity and the need to generalise diets (Williamson 1981). Madeira laurel pigeon Chapter 2. 4

Tarsi of the Canarian birds are bigger in the case of C. junoniae and proportionally bigger in C. bollii. The Madeirean birds measurements have a very wide range â&#x20AC;&#x201C; from 32 mm to 49 mm â&#x20AC;&#x201C; that can be accounted for by the differences in the way measurements were taken. Nevertheless, the Macaronesian birds show a tendency to have a bigger tarsus than that of C. palumbus. According to Blondel (1985) bigger tarsi are also an expected characteristic of island species. Grant (1979) suggests that a greater tarsus length would be selected for by the pattern of distribution of food requiring greater bipedal locomotion on thick branches and ground. However, in this case such explanation is not straightforward because C. palumbus feeds regularly on the ground.

Shorter wings are a reflection of a more sedentary life style (Grant 1979) and should be expected in island species (Blondel 1985). However, none of the Macaronesian pigeons show proportionately shorter wings. This might be related to the ecology of the species, which need to be very mobile to track feeding resources within their habitat. Moreover, in the case of the Canarian species it has been proposed that there might be migrations between the different islands (Gibbs et al. 2001).

Apparently for all measurements the endemic species present wider ranges. Again, this is in accordance with Blondel (1985), who argues that increased biometric variability is one more trait of the island syndrome. Hounsome (1993) suggests that such a fact may be due to character release in the absence of competitors.

From the information presented it can be seen that the island syndrome characteristics are present in the Macaronesian endemic pigeons. Apparently, the Canarian birds exhibit more marked insular traits than the C. trocaz. Whether this is related to the fact that the former species are probably older than the later it is difficult to say. Due to the nature of the data used, conclusions should be treated with caution. Nevertheless, they highlight interesting biogeographical issues, which require further study.

Madeira laurel pigeon Chapter 2. 5

3. Population trends and status.

3.1. Introduction It is crucial for the conservation of any endangered or vulnerable bird species to know its distribution, population size and population trends. Such information is important because it (i) highlights the need to implement specific conservation and management measures; (ii) provides

baseline information against which future changes can be

assessed, and, hence, an important tool to evaluate any ongoing conservation programme; and (iii) helps to assess the relative value of sites of conservation importance as well as their vulnerability â&#x20AC;&#x201C; e.g. by enabling the identification of Important Bird Areas, IBAs, which have proven to be of great value for the protection of many globally threatened species. The value of this information can be greatly enhanced by a

knowledge of such factors such as the effective population size,

minimum viable population and the age structure of the population.

Despite the recognised importance of such data those most basic parameters are still scarcely known for many species. According to Bibby et al. (1998) by 1992 less than a quarter of threatened species in the Neotropics had been subject to any formal count. The inaccessibility of the habitat or/and the secretive behaviour of many birds certainly account for such lack of information. Other important constraints are the lack of economic resources and trained personnel.

In such a scenario it is not unusual for conservationists to have to base their decisions on very scarce information on the species and their actual conservation status. Such is the case with the Madeira laurel pigeon. In 1986 when it was included in Appendix I of the EU Wild Bird Directive it was amongst the list of poorly known species. At that time population estimates and trends where made largely at the basis of guesswork. Based on the numbers seen and shot outside the forest in 1984 and 1985 Zino and Zino (1986) calculated that the population was over 1000 individuals.

Madeira laurel pigeon Chapter 3.1

The first serious attempt to obtain a better understanding of the status of the species was made in 1986 when the population was estimated at 2700+ birds (Jones et al. 1989). One of the aims of that study was to provide a baseline of data on density and distribution of the Madeira laurel pigeon, for future comparisons. This aim was largely achieved mainly because simple and repeatable techniques were used. Moreover, the timing of the survey was very appropriate as it was followed by the reinforcement of the protection of the pigeons, and it allowed an evaluation of the success of the conservation measures introduced. Further fieldwork, carried out in 1991 and 1992, also produced densities and population estimates (Oliveira and Jones 1995). However, because different approaches were followed, a comparison with the numbers obtained on the earlier work was not possible. The next survey, in 1995, employed some of the same methods as in 1986, with the general aim to highlight any population changes. Distance Sampling methods (Buckland et al. 1993) were also used so that total population size could be estimated.

Distance Sampling methods involve estimating distances to bird contacts, and are believed to provide absolute density estimates from which it is possible to derive population sizes (Bibby et al. 1998). Despite the constraints on using such methods in mountainous habitats specific field procedures were develop (refer to methods section) that would ensure the reliability of the data thus obtained. The total population estimate is essential for confirming the conservation status of the species and, in combination with the comparison of encounter rates on transects between years, enables the assessment of any population changes. One other very important reason for using distance sampling methods is that, in theory, they are not subject to bias imposed by habitat structure and configuration. Thus, they allow comparisons not only between different areas of the Madeirean laurel forest, but also to other species of pigeons elsewhere, namely, the closely related Laurel pigeons from the Canaries.

The current study aims were therefore to (i) compare population size with those of 1986 census and (ii) employ Distance Sampling methods (not used in 1986) to obtain estimates of population densities and size.

Madeira laurel pigeon Chapter 3.2

3.2. Methods

The fieldwork took place during the last two weeks of August 1995 using a team of 10 persons under directions of the author. Data were collected by walking line transects through areas of laurel forest and marginal forest habitats. The laurel forest was divided into four areas, each one representing a semi-isolated group of major basins. Figure 3.1. shows the positions of such four forest areas and the approximate location of the routes taken. A short description of these routes is given in Appendix I. Censusing was carried out from sea level up to an altitude of 1200 m. and covered primary and secondary forest and areas subject to different levels of legal protection. At the beginning and/or end of some routes, agricultural land and/or exotic forest were also sampled. It was impossible to site the transects randomly or even to stratify them because of the topography; most censusing was from paths set into the mountain sides.

To monitor changes in population numbers, 13 of the 25 transects used on the 1986 survey were chosen, six of which were proposed as a minimum sample to monitor changes (Jones 1990). In the majority of cases each transect was walked three times by three different teams of two persons (to counter any systematic observer bias). Although the observers walked continuously whilst censusing, the recording period was split into five-minute blocks (as in 1986). All contacts were included regardless of how far away the birds were.

In order to obtain as large sample sizes as possible (for estimating population density â&#x20AC;&#x201C; see later), censusing was carried out during the early daylight hours, coinciding with the first of the birdsâ&#x20AC;&#x2122; daily activity peaks (Jones 1989). In 1986 data were collected at different times of day so, to facilitate a direct comparison, the 1995 results were corrected with reference to the daily activity information provided by Jones (1989).

Madeira laurel pigeon Chapter 3.3

Figure 3.1. Map of the Island of Madeira (320 38’ N and 160 54’ W) where it can be seen the distribution of the Laurel forest. The solid lines represent the limits of the four areas into which the forest was divided. Numbers inside these areas shows the approximate location of the transects (more details are given in Apendix I and Jones 1990)

To estimate ‘absolute’ densities of birds, distances to all contacts were estimated on a total of 18 transect routes (the 13 mentioned above plus five new ones – see Figure 3.1. and Appendix 1). The assumption when using this distance sampling technique is that detection on or near the line walked is certain, and the chance of detection declines with distance from that line. The problem in Madeira is that the probability of detection with distance (the detection function) varies along individual transects. To overcome this, at the end of each five-minute period, detectability was assessed on a scale from one to five. A one denoted that visibility and therefore chance of detection was poor and five that the visibility was excellent throughout most of the five-minute period. All the teams made their own scaling and we choose, for each five minute period, the lowest value found and then only used the periods with a score higher or equal to three. This applies only for the population estimates and not for the 1986 to 1995 comparisons. When a perched bird was sighted, its distance to the nearest location to the path, independently of our position, was estimated to the nearest meter. For birds perched on vegetation above the ground, we estimated the distance to its perpendicular projection

Madeira laurel pigeon Chapter 3.4

on the forest floor. Data from birds in flight were not used in the density analysis unless they were seen to take off within view (often as a result of our presence). Another assumption of the method is that the distances are exact, but if errors in measurement are random and not too large, then reliable density estimates are still possible, especially if the sample size is large (Buckland et al. 1993). To make sure that this assumption was met, following an initial training period, the five principal recorders underwent a trial in which they were required to estimate the distance to 25 different objects (distances ranging between three to 70 m.). Some of biases in their performance are analysed in the results section.

3.2. Results Population changes 1986 to 1995 Figure 3.2. shows the relative and (in the case of 1995) corrected densities (average number of pigeons per five minute period), recorded on transects in 1986 and in 1995. Figure 3.2.A shows the average number of birds per five minute period on all transects. These figures do not necessarily reflect the relative densities of pigeons in different areas as on some transects much more time was spent walking through suitable habitat. Following Jones (1989), a more direct comparison of the density of pigeons within suitable areas is shown in Figure 3.2.B where, for each transect and both years, only the five minute counts which recorded at least one pigeon are included.

The mean number of pigeons found in 1995 was higher for all transects, except for transect 8, where the means are equal. This transect was located in Ribeiro Bonito, one of the best preserved areas of laurel forest. Figure 3.2.B shows a generally similar trend, but now there are three cases where the 1986 counts are higher than the 1995 ones. These transects are numbers 8, 12 and 18 (Ribeiro Bonito, Ch達o da Ribeira and Ribeira da Janela (levada) respectively); all located in areas of very well preserved forest.

Madeira laurel pigeon Chapter 3.5

D ensit ies 9

5,12

14,92

A) All 18,57

19,78

2,07

1,03

37,33

1,32

2,89

1,67

1,46

0,94

7 +-7 ,4 9

6 5

4 +-,9 7

3 2

+-,4 3

+-,3 8

+-.7 5

+-,4 8

+-,9 9

+-,6 7

+-,4 3

+-,4 5

+-0 ,3 7

+-.1 3

0 1

8 10 Transect s

D ensiit ies

B) Good

10 9

1,23

1,83

2,1

1,93

1,53

0,5

1,5

0,93

2,89

1,17

0,45

+-8 ,4 9

8 7 6 5

+-1 ,3 4

4 +-,8 6

3 2

+-0 ,4 8

+-,7 8

+-0 ,4 7 +-1 ,1 4 +-.4 7

86 +-,6

+-,8

+-1 ,4 7

1 0 1

8 10 t ransect s

Figure 3.2. (A and B) Relative densities found for the 1986 and 1995 surveys. A) shows the mean number of pigeons per five-minute walking period and B) the mean number of pigeons per five-minute period in the better habitat (periods where at least one bird was recorded). The numbers above the bars represent the 95% confidence intervals and the numbers on the top of each graph represent the increase rate found for each transect. (see text for explanation and Figure 3.1. and Appendix I for location and description of transects).

Although there does seem to be a general increase in numbers, it is also obvious that there are differences between transects. The rates of increase (mean number per five minute count in 1995 divided by the 1986 figure) are shown in Figures 3.2A and 3.2B and the data from ‘all’ counts plotted in Figure 3.3, against the mean number of pigeons found in 1986. The transects with the lowest densities in 1986 have the highest rates of increase. This trend for the more marginal habitats to exhibit higher increases is confirmed when the ‘all’ and ‘good’ data sets are compared: the former have significantly higher increase rates than the latter (Mann-Whitney U12,12= 36.5; P<0,05).

Madeira laurel pigeon Chapter 3.6

40 35

Increase rate

30 25 20 15 10 5 0 0,00

0,20

0,40

0,60

0,80

1,00

1,20

1,40

1,60

1,80

2,00

2,20

1986 averages per five-minute

Figure 3.3. The relationship between the average number of pigeons by five-minute found in 1986 and the rate of increase between 1986 and 1995 is shown here.

Accuracy of the distance estimates A Runs Test showed that none of the observers had a significant tendency to make any systematic error,i.e, none of the observers systematically under/over estimated the distances (Z=0.74, p=0.45; Z=1.39, p=0.16; Z=1.60, p=0.11; Z=1.21, p=0.22; Z=0.28, p=0.78). The mean absolute deviation of the distance estimates was 0,72m at between 3m and 10m , 2.2m at between 20m and 30m , 8.5m at between 40m and 50m, 11.22m at between 60m and 70m. The majority of contacts with pigeons were at distances of less than 40m so we suggest that the errors were indeed relatively small and random.

Population density and size Table 3.1. shows the density per km2 and the abundance of birds in each of the four areas into which our survey was divided. The highest density is found in Area 3 (S. Vicente Este e Ribeira de S. Jorge) where we have almost 100 birds per km2, the lowest is Area 4 (Faj達 da Nogueira e Funduras) with 31 birds per km2. Adding the number of birds per area we calculate a total population of 10.400 individuals which represents a more than threefold increase over the 1986 estimate.

Area ( cover of forest) 1

Faj達 da Nogueira and

Sample size transects n)

E.S.W.* (95% conf. interval)

7 (93)

63.8 (49.6; 81.9)

Model selected** Hazard/Polynomial

Density per Km2 (95% conf. interval)

Abundance (95% conf. interval)

31 (19; 51)

508 (303; 814)

Madeira laurel pigeon Chapter 3.7

Funduras (16.4 Km2) 2 S.Vicente E and Rib. de S.Jorge (66.6 Km2) 3 Rib. do Seixal and S.Vicente W (31.1 Km2) 4 Rib. da Janela (35.8 Km2)

13 (92)

74.0 (57.5; 95.7)

Half-normal/Cosine

99 (55; 179)

6599 (3661; 11917)

9 (112)

27.6 (17.0; 45.0)

Hazard/Cosine

37 (19; 69)

1149 (615; 2144)

11 (109)

56.2 (42.5; 74.4)

Hazard/Polinomial

59 (38; 92)

2116 (1302; 3096)

Total population:

10359

(5875; 17977) *Effective strip width ** The models for the detection function are selected based on their relative performance in relation to each data sub-set (for details refer to Buckland et al. 1993)

Table 3.1. Densities per Km2 and the abundance on each of the four areas into which our survey was divided

3.3. Discussion

Population numbers

Our first major conclusion is that the Laurel Pigeon population has increased between 1986 and 1995. The increase has occurred all over the island but is proportionately and often numerically greater in the areas which supported lower densities in 1986 (apart from transect 13). The simplest explanation is that as population size increases a density dependent mechanism leads to greater use of more marginal areas. Habitats which are intrinsically less suitable may be quite tolerable if population densities there are lower than in prime habitats (Cody 1985). Since the laurel forest occurs in discrete patches between which the different tree species do not have a homogeneous distribution (Neves et al. 1996), it is likely that they will vary in attractiveness to pigeons. Although at the time of the survey there are differences in densities between transects and also between areas, this may not be a permanent feature. Fruit eating birds generally have to deal with strong spatiotemporal patterning that will affect the way they use their habitat (Herrera 1985) and subtle changes in resources such as fruit density may promote changes in habitat use (Cody 1985). Although the Laurel Pigeon is not an exclusive frugivore, we know that at the time of the year at which our surveys were conducted, the fruits of the bay tree Laurus azorica play an important role on this birdâ&#x20AC;&#x2122;s habitat selection (Oliveira and Jones 1995). The existence of a wide year-to-year variability in fruit production of some Lauraceae has been shown for other forests (Wheelwright 1986) and there is strong evidence that we have the same situation in

Madeira laurel pigeon Chapter 3.8

Madeiran laurel forest (Oliveira and Jones 1995). Thus the pattern of habitat occupancy we have recorded is likely to vary both within and between years. This issue will be further discussed, namely in chapter 4. Seasonal and annual changes in habitat use may also affect comparisons of the four major forest blocks. However, the considerably higher densities found in areas two and four and the greater areas of laurel forest within these areas, suggest that they will always be of critical importance for the preservation of the pigeon.

Our total population estimate of 10.400 individuals suggests a considerable increase over the 2,700+ birds estimated in 1986. The earlier estimate may not be particularly reliable but the comparison of the transects covered in both years confirms that there has been a significant increase. It is possible that natural fluctuations may be responsible for this change but we suggest that the ban on hunting has made an important contribution. In the last legal shooting period - five consecutive Sundays in January and February 1985 - in just one site (Chão da Ribeira), one party of four guns shot 64 birds in four days and over 140 were shot in this valley during the five Sundays (Zino and Zino 1986). If this level of hunting occured in just a few other areas (and according to Zino and Zino (1986) it probably did) then more than 20% of the population could have been lost. The ban on hunting may therefore have removed one of the major constraints on population growth.

Conservation status

Our general conclusion is that with a population of over 10,000 individuals (or even the lower confidence limit of 5,875) and habitat loss and hunting under control, the Madeira laurel pigeon is relatively secure. However, the species’ ‘conservation dependent’ status is still appropriate as other threats remain, namely illegal shooting and poisoning and particularly, predation of eggs and nestlings by rats Rattus rattus (Oliveira and Heredia 1996). Recent studies in the Canary Islands show that this

type of predation is

responsible for the low breeding success of both the Bolle’s Laurel Pigeon Columba bollii and the White tailed Laurel Pigeon Columba junoniae (Hernandez et al. 1999).

Madeira laurel pigeon Chapter 3.9

In Madeira, rats are found at high densities in laurel forest (foraging in the canopy of even the tallest trees) and the steep-sided ravines with sparser vegetation (A. Easby, pers comm). Considering that the Madeira laurel pigeon builds its nest in forest trees and in cavities in cliffs (Bannerman and Bannerman 1965, Zino 1969, Cramp 1985, Oliveira and Jones 2001), it would be surprising if rat predation does not have some effect on breeding output. The effect of such predation on breeding success needs to be clarified. Recent work with dummy eggs (author’s unpublished data) were inconclusive because they failed to identify the causes of “failure” for the majority of the eggs disappeared.

Distance sampling and recommendations for future surveys

As long as the appropriate procedures are adopted (eg allowing for the variation in visibility) we suggest that Distance Sampling or particularly the ‘variable distance line transect method’ we used is appropriate for use in this type of terrain. Although the transects routes should not ideally follow existing paths or habitat features in Madeira we have little choice. However the paths that we primarily used have very little impact on the vegetation. Even though the estimates produced have quite large confidence intervals they are certainly more reliable than the previous tentative guesses (Zino and Zino 1986, Jones et al. 1989 and Oliveira and Jones 1995). As referred to above, the total population estimate is essential for confirming the conservation status of the species and, in combination with the comparison of encounter rates on transects between years, assessing population changes. With this in mind we recommend that the monitoring scheme begun in 1986 be continued. The minimum sample could be the one proposed by Jones et al. (1989) but we strongly believe that, due to the conditions under which the work is carried out, the repeat of all the transects presented on this paper would provide more reliable information. Distance sampling techniques and procedures described here should be used again on future surveys.

Madeira laurel pigeon Chapter 3.10

Appendix I

Location and details of the transects that are common to both surveys can be found in Jones (1990). For practical reasons 1986 survey transect numbers were changed as it follows: 1-1; 4-4; 6-16; 7-12; 9-2; 103; 11-6; 15-7; 17-10; 19-8; 20-13; 25-18. The table below shows the location and details of the transects that were carried out only in 1995.

Transect number – Description - Area

Altitude (m)

Length (Km)

(start – end) 1. R. Frio - Lamaceiros (via levada da Serra do Faial) – Area 1

800 – 825

5.7

2. R. Frio – R. Lajes (via levada da Serra do Faial) – Área 1

825 – 900

4.0

3. Central Fajã da Nogueira – EN103– Área 1

650 – 425

3.2

1050 – 1050

3.2

4. R. Seca – R. Lajes (via levada da Serra do Faial) – Área 1(*) 5. Estrada florestal das Funduras – Área 1

450 – 500

4.2

6. Queimadas – Caldeirão Verde– Área 2

900 – 950

4.0

7. Levada S. Jorge – R. Bonito– Área 2

625 – 750

3.4

8. R. Bonito – R. Grande– Área 2

650 – 600

4.3

9. F. do Penedo – R. Fernandes (via Levada dos Tornos) – Área 2(*)

510 – 510

3.6

10. Encumeada – Ginjas (via Levada do Norte) – Área 3

975 – 975

4.1

11. Ginjas – R. Seixal (via Levada do Norte) – Área 3(*)

1010 – 1010

1.8

12. Fanal – C. da Ribeira – Área 3

1200 – 650

13. R. Inferno – Área 3

25 – 25

0.6

14. Montado dos Pessegueiros– Área 3(*)

1200 – 25

5.5

15. Rabaçal – Galhano (via river bed of R. Janela) – Área 4 (*)

850 – 480

4.2

16. Paul da Serra – Galhano– Área 4

1200 – 480

5.7

17. Galhano – Foz da R. da Janela (along the river bed) – Área 4

480 – 25

4.1

18. Cª carga levada R. Janela – Lombo da Eira (to 1st. tunnel) – Área

450 – 450

3.4

(*) Transects not carried out on the original survey

Madeira laurel pigeon Chapter 3.11

4. Fruit abundance and habitat use

4.1. Introduction Birds are extremely mobile and wide ranging and of the many habitats they could potentially pass through or over, only specific ones are occupied. There is thus a considerable scope, and necessity, for habitat selection at the individual level. Hildén (1965) was perhaps the first to distinguish between the ultimate and proximate factors involved in habitat choice. The evolution of habitat preferences is determined by, and determines, the bird’s morphological structure and behavioural functions and its ability to obtain food and shelter successfully in the habitat. The proximate stimuli for the choice of habitat might be structural features of the landscape, foraging, nesting opportunities, or the presence of other species. Such factors might operate independently, hierarchically as a system of sequential decisions or overrides, or synergistically in a complex fashion. Throughout the year, birds’ requirements for specific nutrients may fluctuate (Murphy and Pearcy 1993, Powlesland et al. 1997), so at certain times particular features of the habitat might dictate choice, and at other times yet other features may override them. Knowledge of the proximate factors involved in habitat choice of endangered and conservation dependent species is crucial for their management. Thus habitat selection studies must be conducted against the background of variations of critical resources in time and space. A bird’s response to seasonal variation may reveal which factors are likely to play an important role for the survival of the species. These studies may reveal which factors are limiting population size (Rice et al. 1983, Wiens 1989), inducing local migrations (Crome 1975, Levey and Stiles 1992) or inducing temporal changes in food selection (Havlin 1977, Foster 1977) and habitat shifts (Alatalo 1980, Jordano and Herrera 1981, Carrascal 1984). Many previous studies have focused on the importance of habitat structure in birds’ habitat selection (e.g. Macarthur and Preer 1962, Karr 1971, Willson 1974, Webb et al. 1999). Clearly, habitat structure as we measure it can mean or translate into very different resources for different sorts of birds. Even though we may be able to measure

Madeira laurel pigeon Chapter 4.1

structural features, which correlate well to the density of a certain species, and in that sense we may be able to predict where to find it and how common it will be, this alone does not tell us about the structural features that the bird respond to (Cody 1985). Other variables, such as food availability (and particularly for this species fruit availability) may be related to habitat structure and composition but will undoubtedly have a much more direct bearing on habitat selection. For frugivorous, or partially frugivorous birds, obviously fruit production is a key variable in determining a bird distribution. A seasonal alternation of scarcity and abundance seems to be an outstanding feature of fruit food in both tropical and non tropical habitats (Herrera 1985 and references therein). Thus, fruit and changes in its abundance will constitute a critical resource to which birds might respond very quickly. Patterns of habitat use by frugivores, particularly their movements between distantlyspaced fruit sources, are very poorly known (Herrera 1985), but the causal link between habitat-use and fruit availability has been shown in many situations (Snow and Snow 1988, Pearson and Climo 1993) and gross habitat selection is very often determined by fruit availability alone (Herrera 1985).

This applies for both the continental and island species. But for the latter, especially those living in tropical, sub tropical and temperate islands, this is likely to be the most important factor, since weather is relatively unimportant (Schluter and Repasky 1991), predators and competitors may be few and the vegetation is usually evergreen. Despite the existence of these favourable conditions, fruit production on islands still shows strong temporal and spatial shifts. These variations in fruit abundance and availability are more critical to island birds than to their mainland counterparts, because they normally have restricted food choices and habitat area. A mainland bird might undertake long migrations to search for the same food source or it can shift to other food sources within its home range (Cody 1985).

Although the mutualism between fruiting plants and frugivores seems obvious the existence of co-adaptive relationships remains unclear and uncertain (Snow and Snow 1988, Sallabanks 1993, Herrera 1998). The earlier idea that most interactions between fruiting plants and their avian dispersers are tightly co-evolved (e.g. Mackey 1975) has failed to be fully accepted and been replaced with the belief that such interactions are

Madeira laurel pigeon Chapter 4.2

â&#x20AC;&#x153;diffuseâ&#x20AC;? (Herrera 1985, Snow and Snow 1988). This approach is based on the observation that the seed dispersal involves an array of interacting species and abiotic factors, where fine-tuned and pair-wise co-adaptation is unlikely (Sallabanks 1993).

Despite much research, few general patterns have been identified in fruit â&#x20AC;&#x201C; frugivore systems, even in terms of one-sided adaptation. Some earlier works argued in favour of the existence of an adaptive relationship between fruiting seasons, fruit characteristics and fruit quality and bird dispersers (e.g. Herrera 1982, Jordano 1985 and Snow and Snow 1988). More recently some authors have corroborated the former ideas (Noma and Yumoto 1997 and Baker et al. 1998) and others argued against it (Guitian et al. 1998, Herrera 1998, Eby 1998 and Gervais 1999). These mixed results may simply reflect the complexity of the fruit-frugivore systems and the need far more extensive research.

The Madeira laurel pigeon is not an exclusive frugivore (e.g. Bannerman and Bannerman 1965, Zino and Zino 1986). However, it is the only potential disperser of seeds most of the laurel forest trees. The existence of any adaptive relationship, from either side, has never been studied. All that is known is that fruit abundance of one of the most common trees of the forest, Laurus azorica, has a strong influence on the seasonal distribution of the pigeon (Oliveira and Jones 1995); and obviously this is not direct evidence of mutualism.

As with many pigeons all over the world the Laurel Pigeon undertakes seasonal shifts in habitat use (Gibbs et al. 2001), and during part of the year feeds on cultivated fields (Oliveira 1999b). For many years this was thought to be related to the scarcity of fruits, namely of Laurus azorica, in the forest (Bannerman and Bannerman 1965 Zino 1986), but more recent work has shown that this may not be the case (Oliveira and Jones 1995).

The general aim of this chapter is to examine the links between the distribution, abundance shifts and movements of pigeons and the fruiting patterns in the forest. The specific aims are to analyse the following: (i) seasonal variations in bird numbers between general habitat types, on a broad scale; (ii) the role of vegetation structure and composition in habitat use, at a local scale; (iii) the influence of the phenology of

Madeira laurel pigeon Chapter 4.3

different tree species, together with fruit abundance; (iv) the relation between forest use and the seasonal shift to cultivated fields; and (v) if the existing patterns of use give any clues about adaptive relationships between the fruit-producing trees and the Madeira laurel pigeon.

4.2.

Methods and study area

4.2.1. Study area The study area is located in Ribeira da Janela, a deep and extensive ravine in the northwestern part of Madeira Island. It is an area where rainfall averages above 1700 mm/yr with a relative humidity at about 85% (Sjรถgren 1972). As a result of the variation in its geographic features, from its origin at an altitude of about 1300 m down to sea level, many different forest types can be found in this valley. At higher altitudes heathers Erica spp. and other shrub type species, as the Vaccinium padifolium, dominate while the lowland vegetation is a well-developed closed canopy forest. There are extensive areas of indigenous, and well-preserved forest covering up to 3,969 ha (Neves et al. 1996), and this is characterised by trees of up to 30 m height, mainly of Ocotea foetens. Other common trees are Laurus azorica, Myrica faya, Persea indica, and Clethra arborea. The understorey is dominated by species such as Teline maderensis, Genista tenera, Phyllis nobla and Rumex maderensis. Some exotic species such as Cytisus scoparius, C. striatus and Ageratina adenophora are widely distributed. At the mouth of the valley there are human settlements, agricultural fields and exotic forest. In the Ribeira da Janela valley nine different sites were chosen (eight in the forest and one in the cultivated area at the valley mouth); their location and general description is presented in Table 4.1 and Figure 4.1.

Madeira laurel pigeon Chapter 4.4

Figure 4.1. Coastline of Madeira Island and study area of Ribeira da Janela. Different colours refer to different altitude ranges as presented in the legend. The different study sites are indicated in red..

Code

Location and description

Agri

Cultivated area with subsistence agriculture carried out in small terraces.

250 Exo

Lower part of the valley from sea level up to 250 m, area covered with exotic forest with an increasing number

250 L1

Still on the lower part of the valley up to 250 m, in areas where a low altitude secondary laurel forest is present.

500 Exo

Northerly area of the valley from 250 m up to 500 m, area covered with exotic forest with an increasing number

of laurel forest species occurring as we move towards the inner areas of the valley.

of laurel forest species occurring as we move towards the inner areas of the valley. 500 L1

Contiguous to the previous one also from 250 m up to 500 m, area covered with a â&#x20AC;&#x2DC;transitionâ&#x20AC;&#x2122; forest where the laurel forest gradually becomes the only type of vegetation present.

500 L2

Contiguous to the previous one, with the same altitudinal range but with a dominant laurel forest in good conservation status, defined by Neves et. al (1996) as a forest where all the strata are present, with a high specific diversity and good regeneration.

500 L3

Contiguous to the previous one, with the same altidutional range and climax laurel forest. This type of forest has the same definition as the above, but with larger girth and height (>30 m) trees..

750 L3

Contiguous to the previous one. In the inner areas of the valley from 500 m up to 750 m. Area covered with laurel forest with a good conservation status (as defined above).

1000 LAlt. Contiguous to the previous one, in the upper areas of the valley from 750 m up to 1000 m. Area covered with laurel forest that grades into high altitude scrub.

Table 4. 1. Summary description of the different sites/transects in the main study area. The codes refers to altitude (250, 500, 750 and 1000), to the general type of forest present (Exo, L and Alt) and to the location in the valley (1, 2 and 3). The code Exo. correspond to the areas covered with exotic forest and L1, L2 and L3 relates to laurel forest ranging from their boundaries into their most inner areas. LAlt refers to high altitude laurel forest.

Madeira laurel pigeon Chapter 4.5

4.2.2. Methods

4.2.2.1. Data collection

Habitat description

Vegetation on each of the sites/transects was sampled by the standard nearest neighbour method as proposed by Barbour et al. (1980) and applied by Neves et al. (1996) to characterise the Madeiran forest in terms of its tree density and dominance. The method consisted of walking through the forest and establishing a sampling station every five minutes (about every 400 m). At each station the distance to the two nearest trees and their basal circumference, at breast height, was measured. To better characterise the forest present in the valley, the sample size was enlarged with information from the Natural Park of Madeira data base which was collected in the same area by Neves et al. (1996). Bird data

The fieldwork was conducted between November 1995 and October 1997 and in two distinct time periods: the first ‘year’ between November 1995 and October 1996 and the second ‘year’ between November 1996 and October 1997. During the first year each one of the 8 forest sites was visited once every two weeks. During the second year only five sites were visited - those which provided a good sample of the different forest types found in the valley and which also supported a large number of birds. The 5 sites were visited every two weeks and there was a six week interval before the next visit began. In both years, fieldwork was carried out by a total of 5 Natural Park of Madeira personnel organised by the author. Data on bird abundance were collected on 8 different line-transects on each of the 8 sites. A variant of the simple line-transect method was used, as proposed by Bibby et al. (1992), and successfully used before with this species (Jones 1990, Oliveira and Jones 1995). Whilst walking continuously a team of two observers recorded all contacts with pigeons that were inside a 250 m. belt defined by the altidutional limits of each of our study sites (0-250 m, 250-500 m, 500-750 m, 750-1000 m, >1000 m). Within the

Madeira laurel pigeon Chapter 4.6

constraints of time and personnel available, not all visits were carried out by the same team. However, the author or one other experienced observer was always present. This was a way of addressing the compromise between experience and the attempt to avoid systematic errors that are likely to occur if the same observers carry out all the surveying (Buckland et al. 1993). To avoid disturbance from people and to count the birds during one of their daily activity peak (Jones et al. 1989), the transects started within the first half-hour of light, and had the maximum duration of two hours. Due to the fact that the disturbed birds fly long distances causing strong bias on the counts, when any signs of human presence was detected during a transect (e.g. fresh footprints, voices or the presence of fresh broken spider webs across the path) data was not used and the transect was carried out on the following day.

Data on habitat use at a local scale and on feeding were collected during the same 8 transects. Every time a contact was established with a bird that was not in flight, a number of variables were recorded as shown in Table 4. 2.

Type of strata being used

- Ground

- Identification of the dominant species - Existence of feeding signs / birds seen feeding

- Bushes

- Species identification

(up to 2 m)

- Existence of flowers, fruits and young shoots - Existence of feeding signs /Birds seen feeding

- Trees

- Species identification - Existence of flowers, fruits and young shoots - Existence of feeding signs /Bird seen feeding

Table 4. 2. Information collected for each non-flying sighting.

Fruit abundance data

Vegetation transects were carried out, every two weeks by the same team members mentioned above, to establish the phenology of the five more common fruiting tree species in the study sites (Ocotea foetens, Persea indica, Myrica faya, Laurus azorica, Apollonias barbujana). These were chosen because of their contribution to the overall

Madeira laurel pigeon Chapter 4.7

amount of fruit available in the forest. We followed a method previously used with success in similar conditions (Oliveira and Jones 1995). The method consisted of walking along paths located on each of our study sites, some of these were coincident with our transect and some not. Every 2 to 5 minutes a sampling point was established. The intervals were planned at the beginning of each section of the transects, depending on a number of factors that would affect the observers pace e.g. thee type of terrain, or the ability to scan individual trees, e.g. weather and light conditions. The usual distance between points was of approximately 300 meters. At each point data were collected from the two nearest trees of each of the 5 species. The variables recorded are listed in Table 4. 3. Tree species - Not flowering Flowering - Absence of fruits Presence of fruits

Not mature Mature

Low abundance Average abundance High abundance

Table 4. 3. Variables recorded concerning the phenology and abundance of the most common tree species of the laurel forest (refer to the text for explanation about the abundance categories).

The abundance was then estimated according to three categories (low, average and high) which was based on the previous knowledge of each individual tree species (see below). A low abundance would mean that the tree had less than half of the number of fruit per/m2 that we should normally expect. A high abundance would mean that the tree had more than the double of the mean number of fruit per/m2 that we should expect in a normal situation. Fruits at a very early state of ripening (i.e. still green) were considered mature, because according to our previous knowledge pigeons still feed on these fruits.

Use of cultivated fields.

The use of cultivated areas was estimated indirectly. Due to unpredictable and strong human disturbance it was not possible to carry any transects in the areas. When the birds are disturbed they fly long distances or hide causing important bias on the counts.

Madeira laurel pigeon Chapter 4.8

The study sites, comprising a total of 100 small fields, were visited every two weeks. In each field we just recorded the presence or absence of damage to individual cabbages, since the last visit. The justification for focusing only on this crop is that it is traditionally present in most of the fields in the study area and it is widely used by the pigeons. 4.2.2.2. Data analysis

i) Structure and composition of the forest

The structure and composition of the forest is described in terms of density (nยบ of trees per 10000 m2), relative density (proportion of trees belonging to species i), dominance (area of ground covered by the canopy of all trees combined) and relative dominance (proportional area covered by the canopy of the trees of the species i). These parameters are estimated according to Barbour et al. (1980). We did not attempt to get quantitative data on the understorey vegetation due to the size of our study sites and the time available.

ii) Habitat use

Data were analysed in relation to periods of 6 weeks. Each of these periods accounting to half of each season of the year as can be seen in Table 4.4. During the second period of work (1997) only half of these periods were studied (refer to legend of Table 4.4.).

Code

Season

Late autumn between 1 of November and 15 of December

Early winter between 16 of December and 30 of January *

Late winter between 1 of February and 15 of March

Sp1

Early spring between 16 of March and 30 of April*

Sp2

Late spring between 1 of May and 15 of June

Su1

Early summer between 16 of June and 30 of July*

Su2

Late summer, between 1 of August and 15 of September

Early autumn between 16 of September and 30 of November*

Table 4.4. Periods according to which our data were analysed. Data were available for all periods in 1996 but only for those marked (*) in 1997.

Madeira laurel pigeon Chapter 4.9

Movements and overall habitat choice

Overall habitat selection was analysed in relation to density changes throughout the year from site to site and within each site. Although we walked continuously whilst censusing a relative density was calculated as the number of birds recorded in each 5 minute period.

Habitat choice on a local scale

Habitat use on a local scale was studied in relation to proportion of birds using the different strata of the forest (ground/herbaceous, bushes and trees) and choice of the different tree species. For the first analysis we assumed that the three strata were always present thus the comparison was straightforward, whilst for the second we had to compare the number of birds seen in each tree type relative to their overall dominance.

Influence of fruit abundance and phenology upon habitat use.

For each tree species, two measures of fruits abundance were used: the proportion of trees with fruits and an Index of Fruit Abundance (IFA). This last measure reflects the number of fruit per ha of forest and was estimated as it follows: Index of fruit abundance spi(1) (IFA)= = [ (Proportion of spi trees with low abundance of fruits(2)) * (0,5 * Mean number of fruits per ha spi) ] + + [ (Proportion of spi trees with average abundance of fruits) * Mean number of fruits per ha spi) ] + + [ (Proportion of spi trees with high abundance of fruits) * (2 * Mean number of fruits per ha spi) ] (1)

spi â&#x20AC;&#x201C; Refers to each of the species that were analysed

(2)

The proportion ranges between 0 and 1.

In turn the mean number of fruits per ha spi is estimated as it follows Mean number of fruits per ha spi = Mean number of fruits spi /m2 * 10.000 * Relative dominance of spi.

The total fruit abundance in the forest is given by the sum of each index of fruit abundance for each of the five fruit bearing trees considered.

Madeira laurel pigeon Chapter 4.10

The mean number of fruits per m2 of canopy was obtained for each species by counting the number of fruits inside a quadrat of one metre by one metre. The sample ranged between a minimum of 52 and a maximum of 206 trees, respectively Appolonias barbujana and Ocotea foetens. This was done by walking through variable paths located on each of our study sites, during two consecutive fruiting seasons (1996, 1997). Trees were chosen on the basis of their accessibility. The mean number of fruits is presented on the Table 4.5. Species

Nr. of fruits/m2

Nr. of fruit per ha of forest

Ocotea foetens

1,2

6360

Laurus azorica

3,2

7360

Persea indica

2,4

720

Appolonias barbujana

2,4

Myrica faya*

6800

* A group of drupes = one fruit

Table 4.5. Mean number of fruit by m2 and number of fruits per ha of forest when all the trees assume a mean crop abundance.

It must be noted that these estimates of fruit abundance are based on the proportional area of ground covered by the canopy of each species. Thus they reflect only an approximate number of fruits per area of ground covered by the canopy, rather than the actual fruit production in the habitat. For the present work these estimates of the amount of available fruits are suitable but they should be used with caution for any other purposes.

iii) Use of cultivated fields

To measure the use of the cultivated fields, two simple indices of damage were adopted (i) the proportion of fields that had their crops attacked since the last visit and (ii) the proportion of cabbages attacked in all fields.

Madeira laurel pigeon Chapter 4.11

4.3. RESULTS

4.3.1. Habitat description

Forest structure and composition

The laurel forest in the study area is in its climax state and three forest layers can be found – trees, shrubs and herbaceous plants – with good regeneration in the tree and shrub layers. This does not apply to sites 250Exo and 500Exo, which are located in the boundaries between the indigenous forest and anthropogenic areas.

Figure 4.2. shows the density and dominance of the eleven tree species that were found.

C o m p o s itio n a n d s tr u c tu r e o f th e fo r e s t T o t a l t r e e d e n s it y = 2 9 5 , 7 h a 60

5 3 ,5

D ens.

Dom.

40 %

2 3 ,6

1 7 ,2

20 10

1 ,1

0 ,4

3 ,2

0 ,6

0 ,1

0 ,4

0 Ab

M f

O f

E xo .

s p e c ie s

Figure 4.2. Relative density and dominance of the eleven tree species (8 native and 3 non – native) found in our study area. Where: Ab Apollonias barbujana; Ca Clethra arborea; Ic Ilex canariensis; La Laurus azorica; Mf Myrica faya; Of Ocotea foetens; Pi Persea indica; Sc Salix canariensis; Exo Exotic species (Acacia sp., Eucaliptus globulosus and Pinus sp. )

Laurus azorica, Myrica faya and Ocotea foetens are the most important species both in abundance and dominance. Laurus azorica is the most abundant tree with a relative density of 27,9%, while Ocotea foetens and Myrica faya present relative densities of 21,6% and 21,3% respectively. Ocotea foetens is the dominant species with a relative dominance of 53,3%. This is explained by the fact that Ocotea foetens trees are very large, growing up to 30 m. Laurus azorica and Myrica faya are the other significant species with a relative dominance of 23,6% and 17,2% respectively.

Madeira laurel pigeon Chapter 4.12

Clethra arborea (9,7%) Apollonias barbujana (4,5%), Persea indica (4,7%) and Salix canariensis (5,3%) comprise a second group of trees with lower relative densities. Persea indica (3,2%) has the highest relative dominance of the three. The family that accounts for more than 75% of the tree density and dominance in the forest is the Lauraceae, to which Ocotea foetens, Laurus azorica, Persea indica and Appolonias barbujana belong.

Figure 4.3. shows the total tree density together with the relative densities and dominance for each of the study sites

ii) Habitat use

Movements and general habitat choice Numbers found for each transect can not be compared, since bird detection probability varies a lot amongst them. On the other hand, detection probability does not change throughout the year within each transect, enabling seasonal comparisons. The changes in bird abundance throughout the first year of fieldwork are shown in Figure 4.4.

The mean number of pigeons per five minute periods fluctuated substantially on all transects throughout the year. Significant differences (Kruskal-Wallis test performed for the periods with n â&#x2C6;&#x192;3) were found for two of the eight transects, 250Exo (H=9,988; df=4; p<0,05) and 500Exo (H=13,29; gl=5; p<0,05). These areas are covered by transition forest characterised by low tree densities and relatively high proportions and dominance of exotic species.

The transition forest located below 250m had high densities of birds between February and July (late winter and early summer) and very low densities for the rest of the year. At higher altitude, between 250m and 500m, higher densities were observed between May and November (late spring and early autumn). The abundance were correlated between pairs of transects and Table 4.6. shows the significant relationships.

Madeira laurel pigeon Chapter 4.13

250Exo

250L1

Total tree dens ity = 130 trees /ha

Total tree dens ity = 391,3 trees /ha

% 40

0 Ab

La Mf Spec ies

500Exo

Ex o

La Mf Spec ies

500L1

Total tree dens ity = 145,4 trees /ha

Total tree dens ity = 343 trees /ha 60

% 40

0 Ab

Mf Of Spec ies

Ex o

500L2

La Mf Spec ies

500L3

100

Total tree dens ity = 385,8 trees /ha

Total tree dens ity = 284,6 trees /ha 80

60 %

% 60

40 20 20 0 Ab

Spec ies

750L3

Dens ity

Dominanc e

La Mf Spec ies

dens ity

1000Lalt

dominanc e

Total tree dens ity = 130 trees /ha

Total tree dens ity = 327,6 trees /ha

% 40

Spec ies

La Spec ies

Figure 4.3. Total tree density and relative density and dominance of the most common tree species for each of our study sites (key as presented in Figure 4.2.).

The marginal areas of the forest (500Exo, 250Exo e 1000Lalt) show lower tree densities. Laurus azorica, Myrica faya and Ocotea foetens are the species with the highest densities in all the study sites. Except for 500Exo and 250Exo Ocotea foetens is consistently the dominant species.

Madeira laurel pigeon Chapter 4.14

R.A bundance

R. A bundance

250Exo

250L1

Sp1

Sp2

Su1

Su2

500Exo

Sp2

Su1

Su2

Sp2

Su1

Su2

Sp2

Su1

Su2

Su1

Su2

500L1

Sp1

Sp2

Su1

Su2

500L2

500L3

Sp1

0 A2

Sp1

Sp2

Su1

Su2

750L3

Sp1

1000Lalt

1,5

4 0,5

Sp1 Sp2 Season

Su1

Su2

Sp1 Sp2 Season

Figure 4.4. Relative abundance (mean number of pigeons per five minute periods) on all transects throughout the year where: A1 and A2 = Early and late Autumn respectively; W1 and W2 = Early and late Winter respectively; Sp1 and Sp2 = Early and late Spring respectively; and Su1 and Su2 = Early and late Summer (Note: transect 1000alt y-axis is shown with a different scale).

Transects

(n)

500Exo-500L3

-0,469 (18)

<0,05

500L2-250Exo

-0,609 (14)

<0,05

750L3-1000Lalt

0,540 (20)

<0,05

750L3-250L1

0,570 (13)

<0,05

Table 4.6. Significant Spearman correlation coefficients found between all pairs of transects. Sample size (n) is also shown.

Madeira laurel pigeon Chapter 4.15

There is a negative correlation between 500Exo and 500L3 and between 500L2 and 250Exo. Positive correlations were found between transect 750L3 and 1000Lalt (nearby areas) and between transects 750L3 and 250L1. Again, in each pair, one transect is covered with transition and secondary laurel forest and exotic forest. For all other transects not shown on Table 4.6, correlations were not significant. The relationship between birds and forest structure.

Figure 4.5. Shows the number of pigeons observed using the different forest layers.

nr of individuals

1000 800 600 400 200 0 Tree

Shrub

Herb.

Forest layers

Figure 4.5. Use of the different forest layers (trees, shrubs and herbaceous ground cover)

The use of the trees is much higher than the use of shrubs and forest ground. In turn, a larger number of birds were seen on the ground than on shrubs. The comparison between the amount of contacts obtained on trees (n=669) and those obtained for the other two layers combined (n=234), shows that there is a significant preference for the trees (G=218,6; d.f.=1; p<0,01).

Figure 4.6. shows the seasonal use of the different forest layers of the forest. All year round trees are used more heavily than any of the other strata, except between May and July (late spring and early summer). The forest floor is used throughout the whole year, with a peak in late spring and early summer, while shrubs are only used during spring and summer.

Madeira laurel pigeon Chapter 4.16

253

232

394

284

198

268

100%

% of Individuals

80%

60%

Her.

Shru. 40%

Trees

20%

0% A2

Sp1

Sp2

Su1

Su2

Season of the year

Figure 4.6. Use of the different layers of the forest throughout the year (codes are presented above on the “Methods and study area” section) Numbers above the bars refers to sample size.

Tree Selection Figure 4.7. shows the observed and expected number of pigeons perched on the different trees (calculated on the basis of the dominance of each species present in the study areas) and the results of the associated G test.

Significant differences were found for all transects, except for 500L3, which shows that in most areas trees are not used randomly. In five areas (all except 250Exo and 250L1) Ocotea foetens was used at a higher rate than predicted. The higher numerical and sometimes proportional differences between the observed and predicted values are also shown by Ocotea foetens.

The same is true for Laurus azorica; the pigeons shows a preference for the species although the difference between the observed and expected values is smaller. At lower altitudes the number of pigeons observed using Apollonias barbujana is higher than expected. The other indigenous species are used less than predicted. The number of birds seen using exotic trees on transects carried out below 250m low altitude (250Exo) is remarkably high. In transect 500 Exo there is also a high number of birds using the exotic trees, but actually in lower proportions than expected.

The lack of apparent choice at 500L3 might be explained by the fact that Ocotea foetens and Laurus azorica comprise 85% of the relative dominance of this area – there were very few other trees which the birds could have used.

Madeira laurel pigeon Chapter 4.17

Contacts

Contacts 250Exo (G=14,3; df=4; p<0,05)

250L1 (G=35,7; df=4; p<0,01)

0 Ab

Exo.

500Exo (G=14,3; df=4; p<0,05)

Exo.

500L1 G=73,8; df=4; p<0,001)

0 Ab

500L2 (G=32,5; df=4; p<0,01)

Exo.

500L3 (G=3,2; df=3; N.sig.)

80 60

40 40 20 20 0

0 Ab

Exo

Mf Of Species

750L3 (G=6,8; df=2; p<0,05)

40 30 20 10 0 Ab

Mf Species

Pi N.Obs.

Exo. Exp.

Ab Apollonias barbujana Ca Clethra arborea Ic Ilex canariensis La Laurus azorica Mf Myrica faya Of Ocotea foetens Pi Persea indica Sc Salix canariensis Exo. Exotic species

Figure 4.7. Observed and expected number of pigeons perched on the different trees (calculated on the basis of the dominance of each species present in the study areas) and the results of the associated G test.

Madeira laurel pigeon Chapter 4.18

Use of shrubs and herbs

Shrubs were only used between May and July. All the birds (70) were observed on flowering plants of three species: Teline madeirensis, Cytisus scoparius and Genista tenera (which due to identification problems are considered as a group of leguminosae).

The use of the herbaceous ground cover species is very difficult to access. However, also between May and July, 10 birds were seen feeding on patches consisting only of Phyllis nobla. This is a semi arbustive plant that can reach one-metre high, but normally it is much shorter. Table 4.7. shows a list of herbaceous plants on which pigeons were seen feeding and/or on which feeding signs were found during the transects.

Species

Apium nodiflorum

Late

Early

Late

Early

Late

Early

Late

Early

autumn

winter

Spring

spring

summer

autumn

√

Aspalthium bituminosum

√

Bidens pilosa √

Bystropogon maderensis √

Cedronella canariensis Chenopodium sp.

√

√ √

Erysimum bicolor Nasturtium officinale

√

Phyllis nobla

√

Plantago major

√

Solanum nigrum √

Sonchus fruticosus Trifolium campestre

√

Rumex pulcher Rumex maderensis

√

Table 4.7. List of plants on which birds were seen feeding and/or on which feeding signs were found.

Since these observations are not related to the availability of the species they have minor quantitative value. However, it is important to note that most plants were used consistently all year round.

Madeira laurel pigeon Chapter 4.19

ii) The relation between the use of the trees and fruit abundance

Phenology and fruit abundance

Two measures of abundance of fruits were used: the percentage of trees with fruits and the Index of Fruit Abundance (IFA). Due to a relative lack of variation in the abundance of fruit in the different fruiting trees of each species, these two measures were closely related throughout our study period. Figure 4.8. shows, for each species, the IFA together with the percentage of trees with fruits and the year round abundance of fruits in the forest.

Two trees produced fruits all year round: Ocotea foetens and Persea indica. Ocotea foetens, the larger tree in the forest, has higher numbers of fruits between late autumn and early spring. Minimum Ocotea foetens fruit abundance occurs in late spring. Persea indica has its peak of fruits from autumn to early winter and its lowest during spring.

A third species, Apollonias barbujana, also produced fruits almost the whole year round, with a peak in late summer. Myrica faya and Laurus azorica are more seasonal, having fruits for part of the year only. Laurus azorica bears ripe fruits between late summer and autumn, while Myrica faya has ripe fruit between early summer and late autumn. Ocotea foetens and Myrica faya showed high abundance of fruits while all the other trees presented a very low abundance.

For most of the year the total abundance of fruits in the forest clearly follows the fruiting patterns of Ocotea foetens. The structure and composition of the laurel forest where O. foetens is the dominant and the second more dominant tree provides the explanation for this. Only in late summer, when the contribution of the other fruit bearing trees is proportionally higher, does the fruiting pattern of the forest as a whole deviates from that of O. foetens.

Madeira laurel pigeon Chapter 4.20

Laurus az oric a

O c otea foetens

350

3500

8,3

3000

250

3,3

200 150 100

22,6

1500

20,3

1000 15

500

0 A2

W2 Sp1 Sp2 Su1 Su2 A 1

Pers ea indic a

Sp1 Sp2 Su1 Su2

Appolonias b arb ujana

58,3

37,5

350

46,6

60,2

300 250

32,7 36,6

200 14,2

150

35 nr. of f ruits

400

nr of f ruits

2000

8,8

100

30 25

11,5

30,4

22,38

15 9

5,6

5 0

0 A2

W2 Sp1 Sp2 Su1 Su2 A 1

Myric a faya

Sp1 Sp2 Su1 Su2

All s pe cie s com bine d

24,5

2000

4500

1800

4000

1600

3500 18,1

1200

3000

nr. of f ruits

1400 nr. of f ruits

45,9 46,6

49,5

2500

3,8

nr. of f ruits

300

2500

1000

2000

800

1500

600 3,6

400 200

1000 500

0 A2

W2 Sp1 Sp2 Su1 Su2 A 1 Seas ons

Sp1 Sp2 Su1 Su2

Seas ons

Figure 4.8 Index of Fruit Abundance obtained for each season. Above the bar is shown the percentage of trees with fruit (for details and codes refer to the “Methods and study area” section)

Madeira laurel pigeon Chapter 4.21

Use of fruiting trees

Birds were seen eating fruits of Ocotea foetens, Laurus azorica, Persea indica, Myrica faya, Appolonias barbujana and Ilex canariensis (not in our sample set). However, due to the nature of the laurel forest, namely tree density and height, and to the secretive nature of the Madeira laurel pigeon, feeding observations are actually very difficult to make whilst censusing. Thus, such observations are very scarce and this hinders any quantitative approach. The use of fruiting trees was analysed by indirect evidence and to guarantee the independence between observation all the analyses were performed in relation to the number of contacts â&#x20AC;&#x201C; 1 bird = 1 group = 1 contact. Table 4.8. shows the number of contacts observed in trees with and without fruit, the expected frequencies in trees without fruit (if the trees were used randomly) and the results of the associated G tests.

Tree species

With fruit*

Without fruit

Exp. Without fruits

Ocotea foetens

59,3

39,42

<0,01

Persea indica

6,6

9,07

<0,01

Myrica faya

7,0

--------

Laurus azorica

23,8

--------

Appolonias barbujana

5,1

--------

*Only considering the places and months when fruits where present **Expected frequencies are too small to guarantee the validity of the test.

Table 4.8. Number of contacts with pigeons perched in trees with/without fruits. The expected frequencies are calculated based on the dominance of trees with fruits at the time of the contact.

For all species the observed frequencies in trees without fruit is lower than predicted, which provides strong evidence that trees with fruits, in general, are favoured over the others. Use of fruiting trees is significant for Ocotea foetens and Persea indica, while all sightings in Myrica faya occurred on trees with fruits. There were not enough observations perform the G test for Laurus azorica (expected frequencies < 3) but it is clear that the pattern is similar. Table 4.9. shows the proportion of contacts in trees with fruits and the number of available fruit - species in the forest.

Madeira laurel pigeon Chapter 4.22

Seasons

Percentage of contacts in trees with fruits A. barbujana

Available fruiting species

L. azorica

O. foetens

P. indica

M. faya

100 (1)

100 (5)

91,7 (11)

8,3 (1)

Sp1

87,5 (14)

12,5 (2)

Sp2

92,3 (12)

7,7 (1)

Su1

40 (6)

20 (3)

Su2

38,4 (5)

30,7 (4)

26,6 (4)

46,7 (7)

6,6 (1)

13,3 (2)

Table 4.9. Proportion of contacts in trees with fruits and the number of available fruit - species in the forest. In brackets is shown the number of contacts and (*) indicates that fruits were actually available at that time.

For most of the year a higher proportion of birds was seen in fruiting Ocotea foetens. However, as the number of choices increases the use of other fruiting trees also increases. An inverse significant correlation was found between the proportion of birds in Ocotea foetens and the number of available species (rs=-0,767; df=7; p<0,05).

Table 4.10. shows the observed and expected frequencies of contacts in the different fruiting tree species in the forest throughout the year. When considering the overall number of contacts obtained throughout the year a significant selection of some fruits over the others is found. L. azorica and A. barbujana fruits are favoured over those of Myrica faya, while fruits of O. foetens and P. indica are used as predicted. Since there were not enough observations, most of the expected frequencies had to be aggregated to perform a G â&#x20AC;&#x201C; test for each season. This obviously hides the differences between the predicted and observed values.

Madeira laurel pigeon Chapter 4.23

Seasons

Observed and expected frequencies in the different fruiting trees L. azorica

O. foetens

P. indica

A. barbujana

M. faya

Obs.

Exp.

Obs.

Exp.

Obs.

Exp.

Obs.

Exp.

Obs.

Exp.

0,0

0,8

0,1

0,0

4,5

0,5

11,5

0,5

0,0

Sp1

15,8

0,2

Sp2

0,0

Su1

4,4

4,6

0,5

5,5

1,99

N.sig

Su2

0,7

5,0

0,7

0,1

6,5

0,33

N.sig

1,1

8,6

1,2

0,1

4,1

0,68

N.sig

TOTAL

1,8

62,6

8,8

0,7

16.1

28.86

>0,01

Table 4.10. Observed and expected frequencies of contacts throughout the year in the different fruiting tree species and associated G test.

iii) Use of cultivated fields in relation to fruit abundance Cultivated fields were used from the beginning of February to the end of July. Figure 4.9. shows the proportion of available fields that were used during this period together with the proportion of individual cabbages used.

A peak in the presence of pigeons is clearly identified by both measures from late winter (first week of March) to earlier spring (first week of April). The day with the highest proportion of cabbages used was the 25th of March, while the 04th of April accounts for the highest proportion of fields used. Both measures are also consistent in that they both show the existence of a second smaller peak from late spring (second week of June) to early summer (second week of July).

When comparing the use of agriculture fields with the phenology of the forest (Figure 4.8.) it can be seen that there is no direct relation between them. The use of the crops and the peak of pigeonâ&#x20AC;&#x2122;s activity in fields happened when fruit abundance in the forest was relatively high. The sharp decrease of fruit abundance, which happened in late

Madeira laurel pigeon Chapter 4.24

spring, was not followed by an increase on activity on the fields. Actually, during this period the use of crops decreased notably.

0,4

0,6

0,5

0,4

% fields

% cabbages

Fields use

0,3 0,2

0,3

0,2

cabbages use

0,3 0,4

0,1 0,1

0,1 0,0

28 .1 1 A2 7. 12 26 A 2 .0 1 21 A 2 .0 2 05 w 2 .0 3 25 w 2 .0 3 04 s p1 .0 4 15 s p1 .0 4 s 1. p1 05 15 s p1 .0 5 30 s p2 .0 5 12 s p2 .0 6 26 s p2 .0 6 06 Su1 .0 7 17 Su 1 .0 7 01 Su 1 .0 8 19 Su 2 .0 8 Su 18 2 .0 9 A1 8. 10 30 A 1 .1 0 A1

Dates

Figure 4.9. Proportion of available fields and available cabbages that were used. (codes as presented above in the â&#x20AC;&#x153;Methods and study areaâ&#x20AC;? section).

iv) Comparison between years

Movements and general habitat choice

Changes in the relative abundance of birds, throughout the 2nd year of work are shown in Figure 4.10. In this second year, the changes in pigeon abundance did not follow the same patterns as in the first year. This applies to most transects but especially to those located in low altitude forest where the numbers of birds were much more stable between seasons than in 1997.

Madeira laurel pigeon Chapter 4.25

R .d e n s .

R .d e n s . 10

2 5 0 E xo

10 8

250L1

Sp1

Su1

Sp1

5 0 0 e xo

Su1

500L1

S p1

S u1

Sp1

Seasons

500L2

8 6 4 2 0 W1

Sp1

Su1

Seas ons

Figure 4.10. Relative densities (mean number of pigeons per five minute periods) on all transects throughout the 2nd year of field â&#x20AC;&#x201C; work (1997)(codes as presented above in the Methods and study area sectionâ&#x20AC;?)

The relationship between birds and forest structure

Figure 4.11. shows the number of pigeons observed using the different forest layers. There is a broadly similar pattern to the first year; the use of trees is much higher than the use of shrub and ground levels and again, many more birds were seen on the ground than on shrubs.

The general pattern observed for the first year is repeated with the trees showing a higher use throughout the year. It is noticeable that the use of the ground during early summer and shrubs in early autumn decreased markedly. Figure 4.12. shows how these three layers of the forest are used throughout the year.

Madeira laurel pigeon Chapter 4.26

350 300

Indiv iduals

250 200 150 100 50 0 Tr e e

Shrub

He rb .

F o r e s t la y e r s

Figure 4.11. Annual use for 1997 of the different forest layers (trees, shrubs and forest covered with herbs).

116

100% 90% 80%

Herb.

70%

% of use

Shru. 60%

Trees

50% 40% 30% 20% 10% 0% W1

Sp1

Seasons

Su1

Figure 4.12. Use of the different layers of the forest throughout the year during the 2nd period of field work (codes are presented above on the â&#x20AC;&#x153;Methods and study area sectionâ&#x20AC;?). Numbers above each column refers to sample size.

Tree selection

Figure 4.13. shows the observed and expected number of contacts with pigeons using the different trees and the results of the associated G test. Significant differences were found for all transects, except for the low altitude transect carried out in an area covered by exotic forest (500Exo). This agrees with the 1996 findings and confirms that trees are not used randomly as they occur in the forest. At all study sites the number of contacts in Ocotea foetens is higher than expected. The observed values for Laurus

Madeira laurel pigeon Chapter 4.27

azorica are higher than expected in 4 transects. All the other species are used less than expected except for Persea indica in one situation (500Exo). The only area where a significant selection of trees was not found in 1996 was not surveyed in 1997.

The use of these species followed the same patterns as in the first year and no additional plants were seen being used by the pigeons.

Contacts

Contacts 250Exo (G=55,9; df=3; p<0,01)

250L1 (G=7,6; df=2; p<0,05)

0 Ab

Exo.

500Exo (G=8,4; df=4; NS)

Exo.

500L1 G=32,7; df=4; p<0,01)

0 Ab

Exo.

Mf Of Species

500L2 (G=13,9; df=3; p<0,01)

0 Ab

Mf Species

Exo

N.Obs.

Exp.

Ab Apollonias barbujana Ca Clethra arborea Ic Ilex canariensis La Laurus azorica Mf Myrica faya Of Ocotea foetens Pi Persea indica Sc Salix canariensis Exo. Exotic species

Figure 4.13. Observed and expected number of contacts with pigeons using the different trees during 1997 and the results of the G test performed.

Madeira laurel pigeon Chapter 4.28

iii) The relation between the use of the trees and fruit abundance.

Phenology and fruit abundance

Figure 4.14. shows for each species the IFA together with the percentage of trees with fruits and the year round abundance in the forest. Myrica faya, Ocotea foetens and Persea indica produced fruits at the same time of the year for both periods of fieldwork. The fruiting period of Laurus azorica started earlier and finished later, while Appolonias barbujana had a shorter fruiting season (for comparison refer to Figure 4.8).

Comparing the number of trees with fruits in 1996 and 1997 it can be seen that a significantly higher proportion of Laurus azorica trees produced fruits in the summer (X2=8,3; d.f.=1; P<0,01) and autumn (X2=16,24; d.f.=1; P<0,01) of the 2nd year. Also in autumn of the 2nd year a significantly higher number of Myrica faya trees produced fruits (X2=4,29; d.f.=1; P<0,05). The opposite happened with Appolonias barbujana, that showed a significant lower production of fruiting trees in the autumn (X2=6,94; d.f.=1; P<0,01) of the 2nd year. Ocotea foetens and Persea indica showed no significant differences between years.

The total fruit abundance in the forest is fairly consistent between both years except in early summer where in 1997 the production of fruits was much higher. As a consequence, the abundance of fruits in the forest in 1999 was more constant throughout the year.

Madeira laurel pigeon Chapter 4.29

O c otea foetens

Laurus az oric a

2500

1200

20,8

41,5

13,4

1000

1500

13,7

1000

nr. of f ruits

2000

500

800

600

5,5

400

200

0 W1

S p1

S u1

S p1

Pers ea indic a

S u1

Appolonias b arb ujana 3

350

4,5 44,1

2,5

300

4,7

60,7

2 nr. of f ruits

nr. of f ruits

250 40,3 200 150

1,5

1 13,3

100

0,5

0 W1

Sp1

Su1

All species com bined

Myric a faya

1600

4500

21,3

4000

1400

3500 nr. of f ruits

1200 nr. of f ruits

Sp1

1000 800

10,5

600

3000 2500 2000 1500

400

1000

200

500

0 W1

Sp1

Su1 Seas ons

Sp1

Su1

Seas ons

Figure 4.14. Index of Fruit Abundance obtained for each season during 1997. Above each column is shown the percentage of trees with fruits (codes presented above in the â&#x20AC;&#x153;Methods and study areaâ&#x20AC;? section)

Use of fruiting trees

Table 4.11. shows the number of contacts observed in trees with and without fruit, the expected frequencies in trees without fruits (if the trees were used randomly) and the results of G associated G test. As in 1996 all observed frequencies in trees without fruits are significantly lower than expected.

Madeira laurel pigeon Chapter 4.30

Tree species

With fruit*

Without fruit

Exp. Without fruits

Ocotea foetens

34,6

22,92

<0,01

Persea indica

6,5

11,61

<0,01

Myrica faya

10,4

18,32

<0,01

Laurus azorica

19,8

15,63

<0,01

Appolonias barbujana

3,8

-------

*Only considering the periods when fruit was present **Expected frequencies too low. Table 4.11. number of contacts with pigeons perched in trees with/without fruits. The expected frequencies are calculated based on the number of trees with fruits at the time of the contacts

iii) The use of cultivated fields and fruit abundance

During this second period of work no birds were recorded on agricultural fields in the study area

4.4. Discussion

General movements The finding that the Madeira laurel pigeon uses many different areas of the forest throughout the year agrees with what has already been proposed in previous work (Jones 1990; Oliveira and Jones 1995). Those earlier studies also hypothesised that birds probably move from valley to valley all year round. However, results of our present study do not fully support that idea. The correlation found between the number of birds using different areas of the same valley, and observational data on birdsâ&#x20AC;&#x2122; movements suggests that most of the movements occurred within the study area. This is presumably a consequence of the geographic features of this specific valley. It is the longest valley on Madeira Island (ca. 10 Km long) with very steep and high ridges up to 1400 m. Those ridges are covered by marginal habitats, which, although widely used, are less suitable for the pigeons (Jones 1990, Oliveira and Jones 1995). Birds may therefore prefer to move up and down the valley, rather than from valley to valley. Such

Madeira laurel pigeon Chapter 4.31

movement is likely to be more frequent in smaller areas which may be unable to support the pigeons all year round. Considering that transects were carried out in contiguous areas and that the Madeira laurel pigeon is very mobile, as pigeons in general are (e.g. Pearson and Climo1993, Bancroft and Bowman 1994, Powlesland et al. 1997, Small et al. 1995), the marked seasonal fluctuations in abundance are striking. They suggest that birds are relatively sedentary for long periods and give some clues about the foraging behaviour of the Madeira laurel pigeon. The birds might take short-range migrations while searching for food, but when a suitable patch is found they tend to use it heavily, probably until a certain level of depletion is attained, before moving on to another one. This is consistent with other work on the foraging behaviour of birds that are dependent on spatially and temporally variable resources as fruit (e.g. Pyke et al. 1977, Herrera 1985, Levey and Stiles 1992, Powlesland et al. 1997). Hence, one factor that would be expected to influence habitat use and, consequently, those movements is fruit abundance and this will be discussed later.

On the other hand, the presence of large numbers of birds using the same area simultaneously can be explained by the dynamic spatial distribution of the species. Although mainly solitary, the Madeira laurel pigeon has an aggregated spatial distribution (Oliveira 1992), resulting from the formation of temporary feeding aggregations, whose numbers can go up to 35 individuals. Apparently the birds tend to search for areas where other individuals are active and foraging. In many species of pigeons that feed on the ground or on the canopy of fruit trees, foraging individuals are strongly attracted by the sight of others of their own species (Goodwin 1983, 1985).

Habitat use and fruit abundance

It has been widely documented that the Madeira laurel pigeon is a tree as well as a ground feeder (e.g. Bannerman and Bannerman 1965; Zino 1986). However, the relative importance and role of each of these foraging and feeding habits has not been evaluated in detail to date. To fully understand a birdâ&#x20AC;&#x2122;s habitat use, movements and spatial distribution such level of information is extremely valuable. The total number of feeding observations made during the present work, on trees and on the ground,

Madeira laurel pigeon Chapter 4.32

provides a sample too scanty to lead to any definitive conclusions. Our discussion is, therefore, based on indirect evidence provided by the interaction between birds and fruiting trees.

Whether foraging or not, during both periods of work significantly higher numbers of birds were seen using trees, in opposition to the ground and shrub layers. Trees were not used randomly and Ocotea foetens proved to be the preferred species, almost all year round. On the other hand, for all species, trees with ripe fruits were significantly favoured over the others. The fact that O. foetens is the tallest and dominant tree in the forest (Neves et al. 1996), producing fruits all year round explains such a pattern of use. It also agrees with the findings of Jones (1990), who detected a positive correlation between the presence of birds and the height of the tree layer.

The inverse correlation found between the proportion of birds using O. foetens with fruits and the number of fruit species available, suggests that the use of O. foetens may be seen as rather opportunistic. In fact for most of the year the Madeira laurel pigeon use the fruits of the most common tree species in the forest, but when other fruits are available they are used in considerable proportions. However, not all fruits are used in such an opportunistic way and L. azorica and A. barbujana seem to be favoured over M. faya.

This is in accordance with the results of a previous work on habitat use by the Madeira laurel pigeon (Oliveira and Jones 1995). It was then suggested that the areas dominated by L. azorica were mostly used only in years when this species bore ripe fruits abundantly. In other years, when fruits of L. azorica were scarce, the areas dominated by O. foetens continued to be heavily used all year round.

On the other hand, previous studies suggests that foraging habits of fruit pigeons in subâ&#x20AC;&#x201C; tropical forests in Australia seem to be largely opportunistic and pigeons will use whatever fruits are available at particular times (e.g. Wheelwright 1986, Innis 1989). The findings of the present work only agree with this to a certain extent, since the Madeira laurel pigeon favours same fruiting species over others (e.g. L. azorica over M. faya)

Madeira laurel pigeon Chapter 4.33

Given that the pigeons apparently forage successfully on Ocotea foetens for most of the year, why do they use other species, exactly when this tree reaches (or is very near) its highest values of yearly fruit abundance? The answer to this question is speculative and theoretical rather than a conclusion from the present data set. Laurus azorica and Myrica faya have clumped and more conspicuous fruits than O. foetens. It has been proposed by many authors that plants producing many fruits are more conspicuous and attract more birds (e.g. Snow 1971, Baird 1980, Herrera 1982, Sallabanks 1993). This might explain why trees that are present with low relative dominance are used even when their contribution to the total fruit abundance in the forest is relatively low (e.g. Laurus azorica during the first year). Pigeons may therefore make such choice based on individual fruit abundance since that strategy may maximise foraging efficiency by minimising search and travel time. On the other hand, this choice can also be explained by the fruit characteristics (fruit length, diameter, and fresh and dry mass) and nutritional composition of its pulp (water, protein and lipid amongst others) as proposed by many authors for frugivores birds (e.g. Snow 1971, Berthold 1976, Stilles 1980, Herrera 1981, 1982, Sorensen 1981, Johnson et al. 1985, Brooke and Jones 1995 and Herrera 1998).

When the total abundance of fruits in the forest decreases, birds shift from the trees to the ground and shrub layers. During this period a higher number of birds are seen foraging and feeding on herb foliage, buds and flowers. It occurs in late spring and early summer, exactly when O. foetens fruits reach their lowest abundance. Once more this is accounted for by the influence of that species on the total abundance of fruits in the forest. Data from the second year of work corroborates such an idea, since a smaller proportion of birds using the ground followed a higher total abundance of fruits in late spring and early summer. This suggests that the herbs and shrubs are alternative food sources that are used in direct relation to the total abundance of fruits in the forest, which, in turn, depends heavily on the fruit abundance of O. foetens. However, at least some birds were seen feeding in the lower layers all year round and irrespective of fruit abundance. This implies that some herbs and shrubs, mainly their leaves, flowers and buds, may provide important nutrients that are not present in the fruits, making up important supplementary food sources. This will be further discussed in chapter 6.

Madeira laurel pigeon Chapter 4.34

The nutrient requirements of the Madeira laurel pigeon are not known, but as for most species of birds (Murphy and Pearcy 1993) they are presumed to fluctuate throughout the course of the year (e.g. protein requirement increase during egg development and chick growth and during moult). Thus, habitat use is likely to also be influenced by the annual cycle of the pigeons.

Due to the composition and phenology of the laurel forest, O. foetens is the most important amongst the fruit species for the Madeira laurel pigeon. The failure of such species to produce fruit crops, especially during winter and spring, would probably have a very negative effect on the pigeonâ&#x20AC;&#x2122;s population. Fortunately, such downfall seems unlikely to happen because different periods of study (1989 to1991, 1995 and the present one) have shown that O. foetens shows quite consistent fruiting patterns across the years. Such inference is not consistent with the general fruiting patterns identified by Wheelwright (1986) for the Lauraceae in general and for this genus in particular. Probably individual trees will follow patterns of almost erratic fruiting periodicity, as defined by Wheelwright (1986), however, such a trend is buffered by the high density and dominance of this species in the Madeirean forest. An accurate assessment could only be achieved after a long-term study of the phenology of individually marked trees. The relationship between forest composition and the Madeira laurel pigeon feeding habits will be further discussed on chapter 6.

Movements and fruit abundance

As with most fruit eating birds, subtle changes in resources such as fruit density will promote changes in habitat use. Birdsâ&#x20AC;&#x2122; movements within the forest will also change yearly and most certainly in direct relation with the local abundance and availability of fruits. This work was unable to substantiate such a direct relationship. However, the data presented provide strong evidence that bird numbers and movements are related to fruit usage and abundance. Maybe the relationship between bird abundance and fruit production patterns are not clear, because the different tree species are mixed together. Even if trees may fruit at different times and the pigeons may prefer some species over the others, they are able to use them differentially without obvious movements and changes in the distribution and abundance within good areas of laurel forest.

Madeira laurel pigeon Chapter 4.35

This relationship between fruit and bird movements has been shown for other frugivore pigeons as the Chatham Islands Pigeons (Pearson and Climo 1993) and several other species of New Zealand (Crome 1975; Innis 1989) and New Guinea (Firth et al. 1976). All these studies also suggested that pigeon abundance, foraging and movements are governed by forest phenology.

Our data set provides additional circumstantial evidence that such trend actually applies to the Madeira laurel pigeon. First, the number of birds that moved to the boundaries of the laurel forest reached a peak when birds were using the ground, which agrees with the fact that the tree density in this area is much lower and floristic diversity much higher. If a bird is foraging on the ground, it makes sense that it will search for areas with less arboreal vegetation and higher diversity and abundance of herbs and shrubs. Secondly, during the second year of work the total abundance of fruit, on one hand, and the number of birds, on the other, were more stable all year round, indicating that birds moved less. Thirdly, the between years comparison of bird numbers present in different areas of the valley shows that apparently there are no predictable seasonal movements, which concurs with the general patterns of movements of frugivores.

This also corroborates the idea proposed at the start of this section, that is, when a suitable area is found, e.g. areas with ripen fruits of any of the laurel forest trees, pigeons tend to stay there for long periods. Presumably they will only leave the area when the total abundance of fruits falls to some threshold value.

Use of crops and fruit abundance

The logic of the argument presented above suggests that it should now be possible to understand the relationship between the use of agricultural fields, habitat use and forest phenology. The fields were heavily used, in 1996, between late winter and late summer. During this period the level of damage neither followed the phenology of the forest nor the phenology of Ocotea foetens, which has been identified as the most important tree for the pigeons. It is also notable that the damage inflicted on the fields started and had its peak when birds were heavily using the fruiting trees in the forest. When birds started feeding more heavily on the ground due to a marked decrease of fruit abundance the damage on the fields also decreased. Therefore, it can be concluded that intra-annual

Madeira laurel pigeon Chapter 4.36

fluctuations in fruit abundance alone may not be sufficient to account for the fluctuations on the level of use of the agricultural fields.

In 1997 the birds did not cause any damage on the study fields. During the potentially most critical period, late winter, the total abundance of fruits in the forest, was very similar to that of 1996 when birds were feeding heavily on crops. This suggests that inter-annual variations in the use of agricultural fields are also not explained by fluctuations fruit in abundance.

Most of the pigeon damage to crops occurs consistently between January and July (Oliveira 1999a, Oliveira and Jones 2001), however, there is a great inter-annual fluctuations in the total amount of damage and it is seldom evenly spread throughout the island (authors unpublished data from 1990 to 2000). These patterns, together with the present work, suggests that the use of agricultural fields is not associated with lack of natural foods, as customarily proposed by many authors (e.g. Zino and Zino 1986). It is more likely connected to the existence of predictable food sources in close proximity to the forest (Oliveira and Jones 1995); the use of the agricultural fields may be mostly opportunistic and is governed by the birdsâ&#x20AC;&#x2122; movements in the forest. Thus, ultimately, fruit phenology will influence the use of agricultural areas only to the extent that it governs such movements.

It has been widely proposed that the understanding of the ecology and behaviour of frugivorous birds can only be achieved with the establishment of long term studies. This work has been able to outline a few general tendencies that need to be confirmed through such a long-term study.

In terms of the management implications of the work, the most important aspect is the understanding of the importance of large areas dominated by Ocotea foetens for the survival of this species and that stability is provided by a mixture of species with an overlapping fruiting patterns. Given the high level of protection rendered to this forest, this should not constitute a problem in the long run.

Madeira laurel pigeon Chapter 4.37

5. Agricultural fields use

5.1. Introduction

Agricultural habitats support important population of many birds all over the world (e.g. Dhindsa and Saini 1994, Tucker and Evans 1997, Boutin et al. 1999, Manikowski 1984). Some of these species may be beneficial for agriculture but others, specially granivorous, frugivorous and omnivorous birds, having adapted to the agricultural habitat, inflict economic losses to crops, fruits, stored grains and livestock food (Dhindsa and Saini 1994, Dolbeer et al. 1994). The use of this habitat may be continuous throughout the year or seasonal (Gill 1996, Bourne 1997). The severity and seasonality of the damage depends on a wide range of complex variables and factors such as a shortage of natural food in winter (Gremmel et al. 1988, Peach et al 1999), climate and/or weather conditions such as summer droughts (Ghidiu and Buttler 1992), need for prĂŠ migratory diet supplementation (Bourne 1997) and opportunistic feeding on available crops (Dolman et al. 1996).

Avian pests are found in many different bird groups and in most countries they are the most significant vertebrate pests (Moran 1993). In the passerines the Bullfinch Pyrrhula pyrrhula, for example, is a common pest of orchards, whilst thrushes such as Turdus merula are pests of soft fruit (Tucker and Evans 1997). Wildfowl commonly damage vegetable crops (Lane and Higuchi 1998) and all over the world many Columbidae, are considered as generalist pests (Giamoustaris and Mithen 1995). The woodpigeon, Columba palumbus, is the major bird pest in the UK, feeding on a wide range of arable crops (Inglis et al.1994) such as brassicae, clover, wheat, barley, peas, beans, turnips and oilseed rape (e.g. Murton 1965, Murton and Jones 1973, Jimenez et al. 1994, Gill et al.1998b).

In the Canaries Islands the Dark tailed laurel pigeon Columba bollii and the White tailed laurel pigeon C. junoniae cause losses to crops and fruit trees (Nogales and Martin pers. com.). However they are not as serious a pest as the Madeira laurel pigeon C. trocaz , which causes more damage to agriculture than any other bird in the Archipelago of Madeira. Madeira laurel pigeon Chapter5. 1

Pigeons are likely to become pests because they are generalists birds, which are able to use many different sorts of food types and different types of habitat (Gibbs et al. 2001). Some of these species have become well adapted to cultivated areas and feed continuously on crops or their associated habitats, e.g. the woodpigeon (Murton 1965) Others undertake occasional or seasonal habitat or diet shifts (e.g. Columba sjostedti from the equatorial West Africa and the C. flavirostris in Peru and Ecuator (Gibbs et al. 2001)) and the importance and/or use of these habitats varies temporally. Most authors (e.g. Gibbs et al. 2001 and references therein) have proposed a direct relationship between crop feeding and the scarcity of natural food sources. However, very few studies have even tried to demonstrate a direct cause â&#x20AC;&#x201C; effect relationship, and most of the authors base their conclusions on circumstantial data. Goodwin (1985) argues that in some situations artificial sources of food may share certain features with natural ones and in such cases innate food stimulus recognition may play a part - the feeding habits of the stock dove being a good example of this (Goodwin 1985).

Conventional approaches to manage bird pests are often reactive rather than preventive. Historically, density reduction of bird populations by shooting has been the preferred method. However, besides the ethical aspects of this solution, most of the times it is not as effective as might be supposed. This is due to the stimulation of density dependent mechanisms, which result in a remarkable resilience of some species, in terms of the way they recover from density reduction. This method is also inappropriate for those species that are considered both as pests of agricultural land and as species of conservation interest. More acceptable methods, depend mainly on the use of scaring and physical exclusionary devices (e.g. Lane and Higuchi 1998). Both are relatively effective, though this effectiveness may be temporary because the birds have a great ability to habituate to audio and visual stimuli (Schmidt and Johnson 1982). These strategies may also be very costly to implement (Vanvuren and Smalwood 1996).

The use of simulated predators or chemical repellents, either synthetic or naturallyoccuring, (e.g. ciannamide or sucrose), have sometimes been effective in modifying the feeding behaviour of the target species (Brugger et al.1993, Giamoustaris and Mithen 1995, Gill et al. 1998a, Gill et al. 1998b); other strategies include the appropriate

Madeira laurel pigeon Chapter5. 2

management of grassland to reduce attractiveness to geese (Vickery and Gill 1999) or the cultivation of hybrids which show higher resistance to damage (Dolbeer et al. 1995).

An alternative is to reduce bird damage through relatively modest changes in crop phenology (VanVuren and Smallwood 1996) and adoption of certain crop management practices (Gill 1996, Odderskaer et al 1997, Linz et al. 1997). These relatively simple preventative solutions are very attractive because of their generally low implementation costs. However, they are not widely used because they depend on a detailed knowledge of the temporal and spatial dynamics of pest species and their specific food preferences and behaviour. Such information is not available for most species.

The Madeira laurel pigeon problem

There are two particular problems associated with pest control for the Madeira laurel pigeon. The first is that as a vulnerable endemic species, control measures which significantly reduce population size would not be appropriate and the second is that the type of agriculture practiced in Madeira exacerbates the pest problem.

Agriculture in Madeira is an ancient and traditional activity and for many years has been one of the most important economic activities on the Island. Nowadays its importance is more limited and in most areas it had become a subsistence activity. This applies especially to the North of the Island where the geography dictates that the farmers use small fields, usually not bigger than 750 m2, located in narrow terraces. An important feature is that most of these fields are surrounded by native and exotic forest, so there is never a great distance between most of the fields and the forest.

At the end of last century, Godman (1872) already mentioned the use of cultivated fields by the Madeira laurel pigeon, so this is undoubtedly a very old problem. For many years the most effective way of protecting the fields was to control the pigeon population at very low numbers. However, for an endemic and vulnerable species, this wasnâ&#x20AC;&#x2122;t an appropriate and sustainable practice, and in 1989 hunting was banned. Since then different methods of crop protection have been tried, the main types being scaring devices and exclusive nets (Oliveira and Heredia 1996). Although they are reasonably effective their widespread implementation is not feasible, mainly because the farmers do

Madeira laurel pigeon Chapter5. 3

not like to use them. The main reasons for this attitude are the unpopularity of the pigeon and the extra expense and work that the use of these devices might represent. Since 1993 the Natural Park of Madeira has had a long-term programme to reduce these costs by providing free scarers and exclusion nets. The success of these actions is limited due to a lack of personnel and co-operation from the farmers (Oliveira 1999a). Thus the management of the Madeira laurel pigeon, as a pest of agricultural fields, has to be seen from an alternative perspective and the solution must involve the development of a low cost and maintenance free method, essentially not dependent on manpower.

When a pigeon, or any bird, arrives in a agricultural area, it faces a hierarchy of decisions: (1) which field to feed in; (2) which crop to pick and (3) which part of the crop/field to visit. The knowledge of the way these decisions are taken, together with an understanding of the temporal and spatial dynamics of the pigeons may represent the solution for the problem, through the adoption of new crop management practices. Therefore, the aims of this study were 1) to record the patterns of use of specific fields and crops, and to describe and predict the vulnerability of the different fields 2) to test the predictive value of the findings by applying the â&#x20AC;&#x153;modelâ&#x20AC;? to a different study area and 3) with an understanding of when and why pigeons attack certain fields, suggest appropriate strategies to reduce pigeon damage.

5.2. Methods and study area

5.2.1. Study area

The fieldwork was carried out on the lower areas of the Ribeira da Janela valley. This part of the valley has a significant human presence, and an absence of mature laurel forest. Some indigenous trees may be present but the exotic forest is always dominant. As in many other places in Madeira the agricultural activity here is mainly a subsistence one; characteristically there are small terraces with different crops grown simultaneously. Five distinct sites, accounting for a total of 100 fields, were chosen. All these sites are traditionally visited by the pigeons and their general characteristics are shown in table 5.1.

Madeira laurel pigeon Chapter5. 4

Code 01

Nr. of fields 9

Altitud 350 mts

300 mts

350 mts

300 mts

250 mts

General location and characteristics Near a secondary road, on the upper limit of the village houses Only accessible by foot, on the boundaries between forest and human settlements Near a secondary road, on the upper limit of the village houses Only accessible by foot on the boundaries of the forest away from housing Placed in the most densely populated area of the village very near housing.

Table 5.1. General location and description of our five study sites.

5.2.2. Methods

5.2.2.1. Data collection

The fieldwork was conducted between November 1995 and October 1997. This period was split into two distinct sets of time: the first year (between November 1995 and October 1996) and the second year (between November 1996 and October 1997). During the first year the fields were visited every two weeks. During the 2nd year the visits were reduced to one each month.

A detailed map of each area was prepared based on aerial photographs (taken at low altitude from a helicopter) and verified on the ground. The fields were mapped together with the surrounding vegetation and habitations. From these maps a total of 24 variables where measured. These variables were divided into four major types; field characteristics, general characteristics of the field boundaries, location and “neighbourhood”. Some variables were “unchangeable” (e.g. distance to the nearest house) and others were “continuously changing” (e.g. crops present in the neighbouring field). The specific variables are described in table 5.2.

Madeira laurel pigeon Chapter5. 5

Variable type

Variable code

Variable description

Field characteristics

Cabden

Cabbage density

Cabnr

Cabbage number

Location

Field boundaries

Size

Field size

Dforest

Distance to the forest

Dtree

Distance to the nearest tree

Dgrtree

Distance to the nearest group of trees

Dhouse

Distance to the nearest house

Dgrhouse

Distance to the nearest group of houses

Droad1

Distance to a secondary road with low levels of disturbance

Droad2

Distance to a road with medium levels of disturbance

Dpath1

Distance to a path with low use

Dhuman

Distance to any human infrastructure (except houses and roads)

Swallsp

Proportion of the length of the field with an obstacle (wall, scrub, etc) higher than 20 cm

Swallh

Maximum height of the obstacle on the side of the field

Sdrop

Proportion of the length of the field with a “drop” higher than 50 cm

Sdroph

Maximum height of the drop on the side of the field

Sno

Proportion of the length of side with no defined boundaries (inexistence of an

Sfruit

Existence of fruit trees on the side of the field

obstacle or a drop)

Neighbourhood

Oscab1

Neighbouring field planted and with cabbages

Oscul1

Neighbouring field planted but without cabbages

Oshuman

Existence of a human infrastructure (including houses) on one of the sides of the field.

Osroad

Existence of a road on the boundaries of the field

Osshrub1

Neighbour area not planted and with vegetation up to 50 cm.

Osshrub2

Neighbour area not planted and with vegetation up to 100 cm.

Osshrub3

Neighbour area not planted and with vegetation higher than 1 m.

Table 5.2. Variables recorded to describe field characteristics, field boundaries, location, and “neighbourhood”. Some of the characteristics did not changed throughout the study period while others were continuously changing.

On each visit, together with the presence/absence of eaten crops, any activities and changes in the fields, which might have affected bird behaviour were recorded Table 5.3. summarises the information that was collected every two weeks.

Madeira laurel pigeon Chapter5. 6

• Type of crops present and their relative proportions • Average height of each crop (measured from a sample of 50 items) • Ground cover or density of each crop * • Which crops had been used by the pigeons • Quantification of the relative amount taken to each crop. * Depending on crop type.

Table 5.3. Information collected every two weeks from the fields.

For each field a diagram was prepared with the location of the different crops and individuals plants (fruit trees or cabbages). When a predated item was detected the variables shown in Table 5.4. were assessed. Variable code

Variable description

Side

Defines if the predated item is on the margins or on the middle of the field.

Swalldroh

Defines if a predated item is near an obstacle higher than 25 cm (e.g. fence, wall etc) (1), near a drop higher than 25 cm (2) or on a flat area without obstacles or drops (3).

Ostype

Defines the type of ‘boundaries’ around the field containing the predated item: field with cabbages (1); field without cabbages (2); a field not in use, shrubs or bushes (3); open field without vegetation or very low and sparse vegetation (4); others (5).

Height

Defines if a predated item is lower (1) the same height (2) or higehr (3) than the surrounding items.

Cabdens

Defines the density of cabbages on the area around the used item (radius of 3m).

Table 5.4. Variables used to describe the predated items and their surroundings.

5.2.2.2. Data analysis i) Critical period and critical crops To identify (i) the period(s) when there is a higher use of the fields and (ii) the most vulnerable crops, three measures of field usage were calculated: usage rate, field predation rate and corrected field predation rate. These are defined as follows:

Madeira laurel pigeon Chapter5. 7

Usage rate (US) = Total number of fields ‘attacked’ / Total number of fields available.

Field predation rate (FPR) = Pi / Total availability of the crop Where Pi is the sum of the percentages eaten of each item predated or the proportion of the available area used and Total availability of the crop is the sum of the number of items or area planted available

Field predation rate (corrected) = FPR * Correction factor

The correction factor was used to standardise the amount of material eaten. The problem, particularly with a quick growing crop is that, for example, a 20% damage to young plant represents less material eaten that 20% damage to a plant only 2 weeks older. The correction factor was estimated on the basis of information gathered on a control field, where the relationship between size and age of “normally” growing crops was followed.

Usage rate reflects the distribution of the attacks in relation to the number of fields available, it does not necessarily relates to intensity of damage. This is given by the Field predation rate. G tests or Fisher Exact tests were used to assess the relationship between the timing of damage and number of available fields and average size of other crops. ii) Field selection and within field choices A principal components analysis (PCA) followed by logistic regression was used to identify the field characteristics that might influence pigeon choice. The PCA was used to reduce the large number of inter-correlated field habitat variables. The components extracted were used in logistic regressions with predated/not predated as the dependent variable. Two distinct analyses were carried out; in the first, the predated category only included the fields which were first attacked in the peak planting season (between the 5th of March and the 4th of April). In the second, all fields that were used at least once throughout the year were included in the predated category.

Madeira laurel pigeon Chapter5. 8

These two levels of analysis will allow distinction between the choice when many fields are available (peak planting season), and the overall choice. The rationale is that birds may be more selective at the beginning of the season, when many fields are available. After depletion of the “best” fields birds may start using less suitable areas.

From the diagram generated for each field it was possible to assess the choice of food items within fields. A t-test to compare the average predation rate of the crops located on the middle of the fields with those located on the margins (inside and outside a belt of 1,5 m respectively). Again, to distinguish ‘first choice’ from ‘general’ choice two separate analyses were performed, the first using data from the first day of predation and the second from the day with highest levels of predation. In order to combine different fields for this analysis, the predation rates of the side and the middle of each field were standardised by the overall level of predation found in each field (sum of percentage of predation found in each cabbage divided by the number of cabbage in the field - calculated in the same way as the FPR – Field predation rate). To analyse the influence of the “within field” characteristics on the choice made by the feeding pigeons a multiple regression analysis was performed (backward elimination). As before, data for the ‘first visit’ (first day of predation) and ‘later in the year’ (peak damage day) were analysed separately. The dependent variable is the standardised predation rate of each item (percentage of predation of item i/ predation rate for the field).

5.3. Results

5.3.1. General use patterns

i. Field phenology

The study area is characterised by subsistence agriculture and this is typical of much of Madeira. Most of the fields are small and harbour many different crops. Figure 5.1. presents the general phenology of the fields in the area.

Madeira laurel pigeon Chapter5. 9

Cab95

nr. of fields

Cab96

Pot.

S.Pot.

F.beans

Corn

Bean

Onion

Peas

Cereal

Date

Figure 5.1. Crop phenology of the study area, where: Cab95 = cabbage planted in the year prior to the study; Cab96 = cabbages planted in 96; Pot. = Potatoes; S. Pot. = Sweet potatoes; and F.beans. = Fava beans. For most of the crops, like potatoes or sweet potatoes, the day of emergence and not the day of planting is indicated; for others, like cabbages, the first entry on the graph shows the day they were planted.

There are mainly two planting seasons: one from the end of February to the end of April and another throughout June. During the first season most of the fields are planted with cabbages and potatoes (the lines are not coincident because the potatoes were only accounted for after emergence). At the same time most of the old cabbages are removed from the fields. After the end of May, beginnings of June, crops such as, sweet potatoes, beans and peas replace the potatoes. Cabbages are left in the fields throughout the year as the farmers only use the older and bigger leaves. Other crops are present all year round in small amounts. ii. Critical periods and critical crops From the initial sample of 100 fields, 64 were followed all year round (most of the remaining 36 fields where abandoned or deeply disturbed, e.g. road building, after the study started). Twenty-nine fields (45 % of the total) were used by pigeons between the 21st of February and the 1st of August. They were evenly distributed throughout all the study areas. This is an expected pattern due to the fact that these areas were chosen on

Madeira laurel pigeon Chapter5. 10

the basis of their potential to be used by the pigeons. From the many crops available only cabbages, recently planted (hereafter refer to as “96 cabbages”) or planted on the previous season (hereafter refer to as “95 cabbages”), were eaten. The use of “96 cabbages” Figure 5.2. shows the number of fields with “96 cabbages” that were used throughout the study period together with their availability and proportional usage (number of fields used / number of fields available).

0,60

50 45

0,50 40 35 0,40

0,30

P.U.

Fields

20 0,20 15 10 0,10 5 0

0,00 21.02

05.03

25.03

04.04

15.04

30.04

15.05

30.05

day.m onth

12.06

26.06

06.07

17.07

01.08 A U PU

Figure 5.2. Available fields with recently planted cabbages (A), used fields (U) and proportional usage (PU), where PU = A/U.

The number of fields that were used fluctuated very much. Two peaks of activity can be identified: the first between 05.03 and 04.04 and the second between 06.07 and 17.07. The days with highest number of fields being used and the day with a high proportion of fields used are the 25.03 and 05.03 respectively. Pigeons start using the fields soon after they are planted and when the young cabbages are available. Apparently there is a relationship between the increase of availability and field use during the first critical period (between 05.03 and 04.04). The high PU that it is evident for 05.03 is due to the fact that a small number of fields were available on that date.

Madeira laurel pigeon Chapter5. 11

Figure 5.3. shows the corrected field predation rate. This is the measure that best describes the amount of cabbages eaten and consequently the impact of the pigeons on the fields. The maximum damage occurred between 05.03 and 04.04 with a peak on the latter date. Between 06.07 and 17.07 the amount of cabbages eaten increased again but not to the level of the first period.

500 450 400 350

Predation rate corr. (*100)

300 250 200 150 100 50 0 21.02

05.03

25.03

04.04

15.04

30.04

15.05 day.month

30.05

12.06

26.06

06.07

17.07

01.08

Figure 5.3. Total predation rate (corrected) of â&#x20AC;&#x2DC;96â&#x20AC;&#x2122; cabbages throughout the year.

Putting together both measures of field use it is clear that for the newly planted cabbages the critical periods are between 05.03 and 04.04 and 06.07 and 17.07. Since total depletion only occurred in six fields, scarcity or non-availability of cabbages cannot account for the fluctuating patterns of field use. In fact there is a positive correlation between the number of fields used and intensity of use (predation rate) (Rs=0.751, n=13, p<0.01). This suggests that the increase in predation rate is due to the use of a higher number of fields and not due to a more intense use of the same fields.

Madeira laurel pigeon Chapter5. 12

Use of “95 cabbages”

Figure 5.4. shows the number of fields where ‘95 cabbages’ were used throughout the study period, together with their availability and proportional use (number of fields used / number of fields available).

1,20

1,00

P.U.

fields

0,60

20 15

P.U.

0,80

0,40

10 0,20 5 0

0,00 21.02

05.03

25.03

04.04

15.04

30.04

15.05

30.05

12.06

26.06

day.month

Figure 5.4. Available fields with “old” cabbages (A), used fields (U) and proportional use (PU).

The availability of ‘95 cabbages’ decreases markedly after the 21.02. because most of the farmers take out these cabbages when they plant new ones. A higher proportion of fields was used between the 25.03 and 04.04; although these cabbages were available before this period, pigeons only started using them after the fields were planted with potatoes and young cabbages.

At this time of the year it is not possible to estimate accurately the proportion eaten by the pigeons. This is due to the fact that the farmers cut a large number of leaves prior to harvesting or cutting the whole plant.

Madeira laurel pigeon Chapter5. 13

Relative preferences between ’95’ cabbages and ‘96’ cabbages The year round relative availability of ‘95’ cabbages” and ‘96’ cabbages is shown in figure 5.5. It is obvious that during the first planting season the older cabbages (‘95’ cabbages) are replaced by young ones (‘96’ cabbages).

cab. 95

cab. 96

nr. fields

35 30 25 20 15 10 5 0 26,11

7,12

26,01

21,02

5,03

25,03

4,04

15,04

30,04

15,05

30,05

12,06

26,06

6,07

17,07

1,08

19,08

18,09

8,10

30,10

day .month

Figure 55. Trends on the numbers of fields planted with ‘95’ and ‘96’ cabbages through the year.

Table 5.5. shows, for both types of cabbages, the proportion of available fields that were used by the pigeons.

Dates 26.01 21.02 05.03 25.03 04.04

Cabbages ‘95’ A D % 39 0 0 36 0 0 29 3 10 21 6 29 14 3 21

Cabbages‘96’ A D % 0 0 0 7 1 14 15 8 53 33 13 39 35 10 29

Test G G --------9.467 0.101 0.575

d.f. ------1 1 1

P --------<0.01 N.S. N.S.

Table 5.5 - Available fields (A) and damaged fields (D). Results of the statistical test (G test) performed.

Pigeons started to use the new cabbages soon after these were available. Initially they either used only the newly planted cabbages (21.02) or these were significantly preferred (05.03). As the use of the fields increased in number and intensity the new cabbages were still used in higher proportions, although not in a significantly different quantity.

Madeira laurel pigeon Chapter5. 14

To test if these preferences are biased by the fact that some of the “better” fields might have experienced total depletion prior to the beginning of the work, the use of both types of crops was compared but only considering the fields where they were available simultaneously. A significant preference was found for the "96 cabbages" (G=5.455; df=1; p<0.05), showing that total depletion is not a source of bias on this analysis.

Although the use of the ‘95’ cabbages only started after the ‘96’ cabbages were planted there is not a significant relation between the proximity of ‘96’ cabbages and the use of ‘95’ cabbages (Fischer Exact Test P=0.646). This is to say that although pigeons only used the ‘95’ cabbages after the ‘96’ cabbages were available, the proximity of the latter apparently does not influence the use of the former 5.3.2. Factors influencing field selection and field use. i. Field Selection The number of variables that describe the characteristics of the fields in the study area were reduced using a principal components analysis (PCA). Table 5.6. (a to e) shows the rotated factor matrix obtained (the varimax method was used to rotate the factor matrix) and only loadings greater than 0.5 were used (Norusis/SPSS inc. 1994). Table 5.6f the new variables are shown and described.

To simplify the analysis, and to generate more easily interpretable factors, the variables were entered in four distinct groups (refer to table 5.2. in the methods section). The variables belonging to one group (field characteristics) change with the season and so two separate PCAs were carried out. The first considered the conditions of the fields the first day they were used during the most critical period. The second considered the conditions of the fields the first day they were used at any time of the year

The PCA reduced the number of variables from 25 to 11. Most of the new variables are easily interpretable, for example factor 1 in Table 5.6a has high positive loading for distances to the forest, the nearest group of trees and the nearest individual tree, and negative loadings for distance to the nearest group of houses and secondary roads. This factor obviously relates to the distance from human settlements and to the forest.

Madeira laurel pigeon Chapter5. 15

Factor 1 DFOREST

0,85129

DGRHOUSE

-0,74994

DGRTREE

0,82305

Factor 2

DHOUSE

0,68702

DHUMAN

0,92253

DPATH1

Factor 3

0,94594

DROAD1

0,97214

DROAD2

-0,51979

DTREE

0,73898

0,72652

Table 5.6a: Factor loadings for the’ field location’ variables. See Table 5.2. for description of variables. All variables remained constant through the study period

Factor 1

Factor 2

OSCAB1

0,95997

OSCUL1

0,96496

OSHUMAN

Factor 3

0,92066

OSROAD

-0,66334

OSSHRUB1

0,77247

OSSHRUB2

0,86495

OSSHRUB3

0,62519

Table 5.6b: Factor loadings for the ‘field neighbourhood’ variables. All variables remained constant through the study period except that two fields changed from OSSHRUB1 to OSSHRUB2 and one changed in the opposite direction. Since these changes did not produce any significant alterations in the factor extraction and factor loadings, they were ignored.

Factor 1

Factor 2

SNO SWALLH

0,77540

SWALLSP

0,76339

SDROP SDROPH

Factor 3 -0,82736

0,80840 0,83939

SFRUIT

0,64684

Table 5.6c: Factor loadings for the ‘field boundary’ variables. All variables remained constant though the study period.

Madeira laurel pigeon Chapter5. 16

Factor 1

Factor 2

CABDEN1

0,77990

-0,58028

CABNR1

0,96317

SIZE

0,98332

Table 5.6d: Factor loadings for the ‘field characteristics’ variables for period 1 (for explanation refer to ‘Methods’ section). Factor 1 CABDEN2

0,97077

CABNR2

0,72205

SIZE

Factor 2

0,66343 0,86719

Table 5.6e: Factor loadings for the ‘field characteristics’ variables for period 2 (for explanation refer to ‘Methods’ section).

Variable type

Variable code

Variable description

Field characteristics

Cabdenr

Density and number of cabbages

Size2

Size of the field

Dhdf

Distance to forest and proximity to human settlements

Dhuminf

Distance to human infrastructure

Dpat

Distance to path

Swdh

Boundaries of the field with a obstacle and a drop. This variable

Location

Field boundaries

relates to the topography of the field Swpsf

Existence of an obstacle, including fruit trees, on the side of the field

Neighbourhood

Snosdrop

Boundaries of the field with drop and no flat boundaries

Osnrs

Neighbour area not planted and with shrub vegetation (from up to 50 cm to higher than 1 m) and absence of road

Oscc

Neighbour area cultivated with or without cabbages

Oshinf

Human infrastructure present on the neighbouring area

Table 5.6f. New set of variables obtained after data reduction by a factor analysis by groups

Table 5.7. (a and b) shows the results of a logistic regression analysis performed to analyse the influence of each of the new variables on field selection by the pigeons. The dependent variable was a dichotomous one with 1=used; 0=not used. The idependent variables were the factor scores provided by the PCA were used (Norusis/SPSS inc. 1994). Separate analyses were carried out for period 1 which was soon after planting when most of the fields were available, and period 2 when all the fields had been used at least once.

Madeira laurel pigeon Chapter5. 17

Variable

S.E,

Wald

Sig.*

Exp (B)

Oshinf

-2,9639

1,2812

5,3515

0,0207

-0,2433

0,0516

Dhdf

-0,5351

0,6456

5,6543

0,0174

-0,2541

0,2154

Dhuminf

2,0833

1,2008

3,0099

0,0828

0,1336

0,0313

Constant

-0,0961

0,7723

0,0155

0,9010

* Variables with P<0,1 are included in the model (Norusis/SPSS inc. 1994)

Table 5.7a. Results of the logistic regression of field use by the pigeons. This analysis refers to the day with higher predation levels, early during the critical period.

Variable

S.E,

Wald

Sig.*

Exp (B) 4,2419

Osnrs

1,4450

0,5162

7,8376

0,0051

0,2904

Oshinf

-0,3854

1,1104

9,2957

0,0023

-0,3246

0,0339

Dhdf

-1,7301

0,6267

7,6218

0,0058

-0,2850

0,1773

Dpat

1,5275

0,8518

3,2155

0,0729

0,1325

4,6065

Constant

-0,4620

0,4826

0,9166

0,3384

* Variables with P<0,1 are included in the model (Norusis/SPSS inc. 1994)

Table 5.7b. Results of the logistic regression of field use by the pigeons. This analysis refers to the day with higher predation levels found for individual fields.

For the first period the use of the fields is negatively influenced by the distance to forest and proximity to human settlements and existence of human infrastructures on the neighbouring field. It is positively influenced by distance to human infrastructures.

For the second analysis the use of the fields is negatively influenced by the distance to forest and proximity to human settlements and existence of human infrastructures on the neighbouring field. It is positively influenced by the existence of neighbouring areas which are not cultivated and which have shrub vegetation (from up to 50 cm to higher than 1 m) and by the absence of a road and distance to a path. For both analyses a total of five variables were selected, two of which are present in both in period 1 and period 2. In general all these variables clearly relate the use of the fields to the absence of human presence and the consequent proximity to the forest.

Madeira laurel pigeon Chapter5. 18

ii. Within field choices Pigeons used the cabbages both on the margins and on the inside of the fields. However, on their first feeding visit they significantly favour the cabbages located on the margins of the fields (t = 2.019; d.f. = 100; p<0.05). As the level of use of individual fields increases, ‘reaching the peak day of damage’, this preference is no longer significant (t= 0,589; d.f. = 95; p>0.05). Table 5.8. (a and b) shows the coefficients of the predictors included in the model by the multiple regression analysis (backward elimination) performed to analyse the influence of the “within field” characteristics on the choice made by the pigeons.

Unstandardised coefficients

Standardised

Sig.

11,714

0,000

5,749

0,000

coefficients B

Std. Error

(constant)

0,788

0,067

Height

0,208

0,036

Beta

0,272

Table 5.8a. Results of the multiple regression analysis performed to analyse the influence of the “within field” characteristics on the choice made by the pigeons. This analysis refers to the “first feeding visit”. The dependent variable is the standardised predation rate of each item (for explanation and codes refer to methods section and table 5.4.)

Unstandardised coefficients

Standardised

Sig.

8,194

0,000

8,298

0,000

coefficients B

Std. Error

(constant)

0,616

0,075

Height

0,296

0,036

Beta

0,382

Table 5.8b. Results of the multiple regression analysis performed to analyse the influence of the “within field” characteristics on the choice made by the pigeons. This analysis refers to “latter in the year” damage (the peak damage day). The dependent variable is the standardised predation rate of each item (for explanation and codes refer to methods section and table 5.4.).

Individual cabbage use was significantly positively influenced by the height of the cabbages in relation to the rest of the crops, i.e., cabbages were more likely to be eaten if they were taller than the surrounding crops. None of the other variables (Cabdens,

Madeira laurel pigeon Chapter5. 19

Swalldroh and Ostype) alone or interacting, significantly influenced the way cabbages were used. iii. The influence of other crops on cabbage predation.

Since cabbage height, in relation to other crops, significantly influences the use of the fields, it is important to examine the relationship between height, field phenology and cabbage vulnerability. For this analysis, two distinct periods were analysed, fields available until 04.04 and fields available after 04.04. Figure 5.6. shows the number of fields used in relation to the number of weeks after planting. The use of the cabbages was highest between the 2nd and the 4th week and between the 16th and 20th week after planting.

used f ield s

16 14 12 10 8 6 4 2 0 0

weeks

Figure 5.6. Number of fields used in relation to the number of weeks after planting.

Since only potatoes and cabbages were present in these fields, the interaction described refers only to these crops. Figure 5.7. shows the growth curves for these crops and also the percentage ground cover for the potatoes.

Madeira laurel pigeon Chapter5. 20

Pot .height Cab.height Pot .cov.

Height (cm) 70

%cover

50 39

30 19

20 10

54 77

63 40

90 80

70 60 50

28 11

100

40 30 20 10

0 0

0 2

weeks

Figure 5.7. Height increases of the cabbages and increase in height and cover of the potatoes.

Between the 6th and the 10th week the potatoes are higher than the cabbages and the ground cover is greater than 50 % between the 4th and the 14th week. The number of fields used/not used until the 10th week (first predation period), in relationship to the height and percentage ground cover is shown in Table 5.9. During this period, when the cabbages are above the potatoes, the number of fields used is significantly higher (G=4.779; df=1; p<0, 05). No significant differences were found when analysing predation in relation to percentage ground cover alone (G=0.981; df=1; p>0.05).

Height of potatoes

Potatoes ground cover

Used

Not used

<25cm

>25cm

<50%

>50%

Table 5.9. Relationship between height and percentage ground cover and the number of used/not used fields during the first period of damage.

The number of fields used/not used after the 16th week (second predation period), in relationship to percentage ground cover is shown in Table 5.10. During this second predation period (after the 16th week) the potatoes decrease in percentage ground cover (the leaves start to die) and they are no longer present in many of the fields. This is the beginning of the second planting season. It is a period throughout which the cabbages

Madeira laurel pigeon Chapter5. 21

are very conspicuous because many fields either have no other crops or only recently planted crops. During this period the use of the fields is significantly higher when the fields have no other crop or when the ground cover is low (G=34,565; df=1; p<0,001).

Ground cover

Used

Not used

<50%

>50%

Table 5.10. Relationship between ground cover and number of fields used/not used in the second period of damage

iiii. The influence of general phenology and changes in the area

Apparently there is a relation between the intensity of bird use of fields and the periods at which more changes happened on the study area. On the 21.02 the first cultivated fields and the first predated crops were recorded simultaneously and this was the beginning of the first critical period. The second critical period happens when the crops are replaced, between 12.06 and 1.08. During this time on the majority of the fields the potatoes are harvested and the fields are either cultivated with other crops or abandoned. (see figures 5.1., 5.2. and 5.3.). It is important to understand if bird use is just a consequence of the phenology of a particular field or whether the general phenology of the area has an influence over what is happening in each field.

To further analyse this, the first planting season was divided into two sub – periods. Number of fields used during these periods is shown in table 5.11. Before 04.04 the majority of the fields are prepared and planted in a short period of time, the whole area is changed in a very conspicuous way and the number of available fields increases rapidly. After this period (between 04.04 and 15.05) the increase is much slower and a small number of available fields are added to the area. The important issue here is to see how pigeons respond to the evolving environment and in particular if the ‘early’ and ‘late’ planted fields are used in the same way in the two critical weeks after cultivation.

Madeira laurel pigeon Chapter5. 22

Until 04.04 After. 04.04

A 18 6

D. until 2nd week(%) 15 (83,33) 1 (1,67)

D after 2nd week(%) 3 (16,67) 5 (83,33)

Table 5.11. Number of fields available (A) and damaged (D)(before and after the 2nd week) for both periods. The percentage used is shown in parentheses.

Table 5.11. shows that 83% of the fields available before 04.04 were used on the first two weeks following cultivation. There was a significantly different pattern (G=8.926; df=1; p<0,01) for those fields that were only available after 04.04 with only 16% of these used during the first two weeks. This means that the cabbages planted before 04.04 are more likely to be used soon after being planted than those planted later on the season. The likelihood and speed of damage may therefore be related to how many other fields are being planted at the same time.

5.4. Damage risk assessment.

The patterns of use of specific fields and crops described during this work, can be used to create a model which will assess the risk of attack for each field. To be effective and more likely to be used in a practical management situation by non-trained personnel, it has to be simple. Therefore this method should be based on a small number of variables which explain the majority of the damage occurred.

With this purpose the locational variables that have been identified to influence pigeon damage on fields (positively or negatively) and the descriptive analysis of the data set, were used to assess the damage risk for each field. The â&#x20AC;&#x2DC;locationâ&#x20AC;&#x2122; variables Dhouse, Dhuman and Dforest were examined individually to establish the distance beyond which the probability of being attacked/not attacked would fall below 0.8 (this value was chosen because a level of damage of 20% is thought to be acceptable). Figure 5.8 shows for each of the three variables the relationship between distance and frequency. It can be seen that the distances at which the frequency falls below 0.8 are 30m to Dforest and Dhuman and 60 m to Dhouse.

The other significant predictor of damage was the neighbourhood variable Oshinf, the presence/absence of human infrastructures in the neighbouring fields correctly predicted

Madeira laurel pigeon Chapter5. 23

67% of the events (attacked or not attacked). This variable was combined with the location variables to generate the risk assessment chart shown in table 5.12. In this table location variables which negatively influence the likelihood of attack (Dhouse and Dhuman) are set against the variable which positively influences the attack of the fields (Dforest). The variable Oshinf has the same effect irrespective of location variables. The level of risk is then identified as very high, high medium or low depending on the particular combination of factors. Freq. of fields 1

used

Dforest

not used

0,8 0,6 0,4 0,2 0 0-15

15-30

30-45

45-60

60-75

75-100

100-125

Dhouse 1 0,8 0,6 0,4 0,2 0 0-20

20 - 40

40 - 60

60 - 80

80 - 100

100 - 120

120 - 140

>140

DHuman 1 0,8 0,6 0,4 0,2 0 0 - 15

15 - 30

30 - 45

45 - 60 60 - 75 distance (meters)

75 - 100

100 - 125

>125

Figure 5.8. Relationship between distance of the fields to the three chosen variables and frequency of used/non used fields (for explanation and codes refer to methods section and table 5.2.)

Madeira laurel pigeon Chapter5. 24

Dhouse > 60 Dhuman > 30

Dhouse < 60

Dhuman < 30

Dhuman > 30

Dhuman < 30

Oshinf

(No)

(Yes)

(No)

(Yes)

(No)

(Yes)

(No)

(Yes)

Dforest

V. high

High

Medium

<30

(1-0.8)

(0,8-0,6)

(0,6-0,4)

High

Medium

Low

V. low

(0,8-0,6)

(0,8-0,6) (0,6-0,4)

(0,6-0,4)

(0,4-0,2)

(0,2-0.0)

Dforest >30

Table 5.12. Criteria used to classify the fields according to their bird damage risk. The probability of attack estimated for each combination is shown in parentheses. A description of each factor is given in Table 5.2

This risk assessment chart was used to predict damage in two additional study areas comprising a total of 79 fields, located in Lombo Galego and Faial. Risk assessment was carried out in March, at the beginning of the season, when the majority of the fields were still abandoned. The areas were visited twice more; soon after planting (April) and when the crops changed (July). Table 5.13. shows the predictions that were made and the number of fields attacked. Dhouse > 60

Dhouse < 60

Dhuman > 30

Dhuman < 30

Dhuman > 30

Dhuman < 30

Oshinf

(No)

(Yes)

(No)

(Yes)

(No)

(Yes)

(No)

(Yes)

Dforest <30

8 (8)

7 (7)

5 (5)

2(1)

5(4)

4 (3)

3 (3)

6 (2)

Dforest >30

5 (4)

6 (4)

4 (3)

3 (1)

2 (0)

7(0)

6 (0)

Legend: V.high

High

Medium

Low

V. low

Table 5.13. Number of fields that were included in each risk of damage category and number of fields that were attacked (inside brackets).

A total of 45 fields received damage and all of them in both periods. Damage was successfully predicted in 96% (21) and 77% (21) of the fields included in categories V. high and High respectively. The other fields included in these categories (one V. high and six High) were not attacked as predicted. Three fields (27%) rated as Medium were attacked, while none of the fields in the Low and V. low categories were attacked. There

Madeira laurel pigeon Chapter5. 25

were no fields damaged which were predicted not to be damaged. Table 5.14. shows the observed number of fields in each category that were attacked and the predicted range according to the probability scales shown in Table 5.12.

Category

Observed

Predicted

V. high (22)

22 – 17,6

High (27)

21,6 – 16,2

Medium (11)

6,6 – 4,4

Low (13)

5,2 – 2,6

V. low (6)

1,2 - 0

Table 5.14. Observed and expected numbers of fields in each risk category. The total numbers of fields are shown in parentheses.

The number of fields in the V. high and High categories that were attacked fall inside the expected range. The fields in the lower risk categories were attacked less than predicted. As expected most of the damage (77%) occurred on the two weeks following plantation.

5.4. Discussion

The phenology of the fields in the study area is characterised by the existence of two major planting and sowing seasons per year; February – March when all fields are planted, and June – July when, after harvesting, new crops are sown or planted. The two critical periods for bird use are, for a number of reasons, closely related to the existence of these two seasons. There is a first period of intense bird use of the fields, between the first week of March and the first week of April and a second, later in the year, throughout July.

The data provide strong evidence that there are two kinds of factors that influence this pattern. The first are ‘global’ factors” that are related to general field phenology and the second are specific factors related to the interactions between the existing crops within each field. These are concurrent but to a certain degree independent factors.

Madeira laurel pigeon Chapter5. 26

The global factors are mainly related to the high level of synchronisation found in the agricultural activities performed in the area. For climatic reasons, most of the farmers start preparing and planting the fields at the same time of the year, using exactly the same crops (cabbages and potatoes). As a consequence the fields become very conspicuous and the whole area can be easily distinguished from the rest of the landscape. This conspicuousness may be the proximate factor that attracts pigeons. Many bird species use simple cues to determine food choice (Alcock 2001), and also pigeons have been shown to use a range stimulus to trigger their food choice (Troje 1999).

The birds that came to these areas may be attracted because they are experienced and have learnt to recognise signals linked to predictable, nutritious and relatively easily accessible food sources. It has been shown that within each species individuals that feed on agricultural land, either on crops or not, can be placed in distinct groups: experienced birds that have become specialised on the sources provided by these habitats and young sub-adult birds that have little foraging experience (Blanco et al. 1998). In the case of this study, experienced birds may respond to the stimuli provided by the newplanted/recently harvested fields. This is supported by the fact that old cabbages are not used until the new ones are planted. It has been proposed that such experienced pigeons may recognise sources of food through learning and memory, where there can be no possibility of innate recognition (Goodwin 1985). In general cognitive processes such as perception, learning and memory can play an important role in foraging decisions and many other animal behaviours (Shettleworth 2001). Furthermore, field studies with rock doves Columba livia have demonstated their capacity to remember the location of food rewards (Spetch and Edward 1986).

The high rate of depletion found in the fields can be explained by the fact that, as with other species of pigeons, individuals are attracted by the sight of others already feeding (Goodwin 1985). Social learning, which refers to learning that occurs as the result of observation of or interaction with another animal, has been described in many species of birds (e.g. Campbell et al. 1999, Johannes and Kotrschal 1998), and it is also known that young pigeons follow their parents to their feeding grounds (Goodwin 1985).

Madeira laurel pigeon Chapter5. 27

When analysing the different factors that influence pigeon choices when they arrive at field (the specific factors), only cabbage height in relation to other crop was seen to be significant (except cabbages on the edges of fields on first visits). Birds choose to feed upon those cabbages that are above other crops and, preferentially, are in areas with low ground cover. Birds may choose these cabbages because they are obviously easier to locate or as an anti predator mechanism (Alcock 2001) â&#x20AC;&#x201C; these cabbages provide the possibility of a continuous scanning of the surroundings and may enable a quick escape.

Subramanya (1994) has suggested that selective feeding on crops is governed by the predator vigilance pattern such that foraging by bird pests in agro-ecosystems is nonrandom and dependent on factors favouring predator avoidance behaviour and not on resource maximisation. The data presented here, and other anecdotal observations on the behaviour of the Madeira laurel pigeon when feeding on the crops agrees with this idea - birds usually avoid feeding in places where they can not scan a wide area.

When feeding in agricultural areas pigeons tend to gather in groups, and very seldom feed solitarily. These flocks are much larger than those found in the forest, where solitary feeding birds are common (pers. obser.). This is typical anti predator behaviour of birds that are using alternative or sub-optimal habitats (e.g. Krebs 2001). In their natural habitat adult pigeons face a very low risk of predation, but when feeding in open fields they became very highly exposed to man and other predators. It is easier to shoot a solitary bird than one in flock, since the single bird allows closer approach.

The impact of potential predation on feeding behaviour is further underlined by the fact that pigeons chose to feed on those fields which are far from human disturbance. In general they chose the fields that are closer to the forest and, consequently far from human settlements and infrastructures. The Madeira laurel pigeon also prefer cultivated fields with adjacent shrub vegetation (from up to 50 cm to higher than 1 m) rather than cultivated fields near other cultivated areas. This is also indirectly related to human disturbance because, nowadays the further away the fields are from houses and from human infrastructures the more abandoned they are, presenting shrubby vegetation surrounding them. The preference to feed close to forest or, in general, to vegetation cover has been shown in other birds (e.g. Subramanya 1994) and can be explained by the search of a positive trade off between feeding and danger (Krebs 2001).

Madeira laurel pigeon Chapter5. 28

A final specifc factor relating to feeding site choice is the preference for newly-planted cabbages; the birds only start using heavily the fields when they become available and during their first visits birds they markedly prefer the new cabbages rather than those planted in the previous year. With time this preference is attenuated and they start using older cabbages. The preference towards the young cabbages may be related to their nutritional value and probably higher digestibility (Marques pers. comm.).

Management implications

The results obtained in this study can be used to improve crop protection measures that have been used over the past 10 years and also to propose alternative methods to minimise pigeon damage in fields. The methods of protection that have been used, such as scaring devices and exclusions nets, are effective only to a certain degree. They are too expensive for most of the small farmers on Madeira and have to be supported by the local authorities, hence, any way of increasing their effectiveness would be very valuable. This study has identified the most critical periods for damage and the fields at most risk. On the basis of the data it is therefore now possible to predict when and where a crop is potentially in danger of being attacked and where methods of protection can achieve their greatest impact. This level of knowledge has been used elsewhere to overcome the problems caused to same crops by other avian pest species (e.g. geese on winter cereal seedlings (McKay et al. 1996), Dickcissels in rice and soghurm (Basili and Temple 1999) and crows and parakeets in sunflower crops (Mahli 2000)).

Since the general timings of planting and sowing are dictated by the seasons, it is not feasible to act upon the ‘global’ factors. Farmers increase the productivity of their fields by planting them during the most suitable period of the year, therefore it will be difficult to overcome the problems caused by the local synchronisation of crop planting.

It is however feasible to act upon the ‘specific’ factors, using the risk of damage assessment chart created. The testing of the predictive value of this chart has shown that the risk of certain fields being attacked is, to an extent, over than under-estimated. Although this might result in same unnecessary effort on those fields that may not be attacked, it leaves a safe margin of confidence and wrong decisions by default are very

Madeira laurel pigeon Chapter5. 29

unlikely to happen. On the other hand a good reason to have a safety margin is that birds may be deflected from the very high-risk fields to the lower risk ones.

On a first approach. attention should be paid to all the fields rated as V. high and High risk. Secondly, and depending on the resources available, the fields rated as Medium risk can be the aim of the protective measures. It is important to note that with this 2nd level of approach the managers will actually be avoiding damage in only 30% of the fields on which they are acting upon. The other fields included on the Low and V. Low categories can be safely ignored. In practice managers should pay special attention to fields that are in close proximity to the forest and far from any source of human disturbance. In addition the phenology of the area and of the specific field should also be the subject of detailed analysis. Farmers should inspect their fields for damage regularly until the end of the fourth week and particularly between the 10th and 16th day after planting. There should be a response to any damage but considering that the efficiency of the scaring devices decreases rapidly with their use, farmers should be advised to reduce their use after the fourth week following planting. With a proper management of human resources and scaring devices, it is possible to reduce the costs of this operation.

Alongside these measures, manipulation of crops within the fields could result in a very effective decrease in crop damage, especially to cabbages. To reduce the effects of pigeonâ&#x20AC;&#x2122; activity during the first planting season, from late February to late March, farmers should only plant the cabbages after potatoes have emerged (10 to 15 days later). This simple measure would prevent the use of fields during the first critical period without a relevant increase in costs.

To prevent the second critical period, farmers should plan their planting in such a way that cabbages would never be left alone in the fields. Prior to removal of potatoes, for example, corn could be planted in order to act as a barrier between the cabbages and the pigeons. Although these measures are unlikely to eliminate crop damage entirely they would reduce it to an acceptable level.

Additional control may be achieved by the use of repellents. These products (preferentially naturally-based rather than completely artifical), may provide another

Madeira laurel pigeon Chapter5. 30

effective measure acting alone or synergistically. This has been achieved with many species like greenfinches, Carduelis chloris, and great tits, Parus major (Gill et al. 1998a). Obviously this would require a detailed knowledge of the Madeira laurel pigeon’s response to these compounds, because the repellency/tolerance effect varies from species to species according to the repellent and concentrations used (refer to Luttick 1998). However the information included here on risk assessment for cultivated fields would allow targeted repellent use and an economically efficient use.

The specificity’s of Madeira agriculture and the socio – economic background of the people involved in this activity probably will be the main constraint on the implementation of these methods. Although the measures may be inexpensive, previous experience has shown that it is very difficult to convince local farmers to increase their work load or even to co – operate in allowing managers to put scaring or protective devices on their fields. From the farmer’s point of view the solution to this problem lies in shooting the birds and in economic compensation for their losses. As referred to above, shooting a protected species should not be considered to be an appropriate measure. Also, a programme of compensation involves economic and logistical actions that are not available to Madeirean wildlife managers. Since the farmer’s represent a very strong political lobby, the solution can only came from increased awareness of the importance of the Madeira laurel pigeon and the legal obligations that Madeira has to protect the pigeon and its habitat. This could be helped by a proper acknowledgement and accurate assessment of the crop losses caused by birds, something that has not yet been achieved. Only an integration of approaches can provide the framework for an Integrated Management Programme for the Madeira laurel pigeon as a protected pest species.

Madeira laurel pigeon Chapter5. 31

6. Diet and Fruit availability: a Study using microhistological analysis

6.1. Introduction

Knowledge of animal nutrition, as a component of both wildlife ecology and management, is central to an understanding of the life history of any population (Robbins 1993). Considerable theoretical and experimental research has been devoted to the optimal choice of foods because of the wide range of potential practical and academic applications of this information (e.g. MacArthur and Pianka 1966, Pulliam 1974, Pyke et. al. 1977, Stephen and Krebs 1986, Szaro et al. 1990). There has also been a considerable interest in the ways in which birds and plants have interacted and have become adapted to one another and particularly in the relationships between frugivores and fruit-producing trees (e.g. Stiles 1980, Herrera 1982, Herrera 1985, Snow and Snow 1988, Cotton 1998, Hamback 1998, Bleiweiss 1998). Such fruit – frugivore system are complex (Sallabanks 1993, Brooke and Jones 1995) and it has seldom proved possible to describe all the dimensions of this relationships.

A bird, namely a frugivore, will select a particular food because of its energy content (Krebs 1978, Sutherland 1982, Parrish 1997), nutrients (Murphy and King 1987), secondary compounds (Buchsbaum et. al. 1984, Jakubas et al. 1989, Bairlein 1996), physical characteristics (e.g. size, shape and colour of the fruit) (Snow and Snow 1988, Noma and Yumoto 1997) or a mixture of these variables (Greig-Smith and Wilson 1985, Diaz 1990). Since most fruits are very conspicuous it might be expected that foraging on fruit would be easier and more rewarding than foraging for seeds or insects. Yet the number of exclusive frugivores, comparing to the number of birds exclusively adapted to other food types, is quite low (Howe and Smalwood 1982, Murray 1987, Izhaki and Safriel 1989). The suggested reason for this is the fact that fruit supply is patchy, unpredictable and subject to strong spatio-temporal patterning (Herrera 1985). Bird dispersed-fruit have also been considered nutritionally limited because of dilution of nutrients with indigestible seed bulk (Levey and Grajal 1991). The low protein content of fruits has been identified as another reason for the

“inefficiency” of

Madeira laurel pigeon Chapter 6.1

frugivory (Moermond and Denslow 1985). Many studies have focused on this but they have yielded few clear determinants of fruit choice by birds, and the nutritional limitations of fruits for avian frugivores remain poorly understood (e.g., Willson 1994, Witmer and Vansoest 1998).

Whatever the proximate causes of choice, most frugivores react to the spatio â&#x20AC;&#x201C; temporal variation in fruit by making occasional or seasonal changes in food selection (e.g. Witmer 1996, Bairlein 1996, Wood 1997, Franklin 1999) and habitat shifts (Herrera 1985, Robbins 1993). Among temperate birds there is an almost general absence of year-round frugivory. This is due to the fact that these regions have long periods of extreme fruit scarcity occurring simultaneously over vast areas, what precludes the possibility of short-range habitat shifts (Herrera 1985).

Changes in food selection, to exploit different fruit producing trees or distinct food sources such as flowers, leaves or even insects, might also occur as a strategy to complement the diet. Ingesting nutritionally complementary foods may be the most effective way used by many free-living animals to match nutrient ingestion with nutrient needs (Rapport 1980, Jordano 1988, Murphy and Pearcy 1992).

Little is know about fruit selection and related habitat shifts in pigeons and doves. The few studies which have been made (Crome 1975a and 1975b, Powlesland et al. 1997) are far from providing a good understanding about this important aspect of the life of pigeons. This certainly applies to species of high conservation value such as the three Macaronesian endemic pigeons. Despite the strong interest in the study of feeding habits of these three pigeons, from the evolutionary and ecological point of view (Snow & Snow 1988), there as been no systematic research.

In the case of the Madeira laurel pigeon, most of the information is based on descriptive and observational data, listing mainly the consumption of fruits belonging to trees characteristic of the laurel forest (e.g. Laurus azorica, Ocotea foetens, Myrica faya) and vegetative parts of phanerogam plants (Phyllis nobla, Apium nodiflorum, Nasturtium officinale) (e.g. Harcourt 1851, Godman 1872, Sarmento 1948, Zino 1969, Zino & Zino 1986, Oliveira & Jones 1995, Oliveira 1999a). They also feed on a wide range of

Madeira laurel pigeon Chapter 6.2

cultivated crops (Oliveira and Heredia 1996); this aspect of the diet is dealt with in chapter 7.

Direct observation of feeding is a technique widely used to assess pigeon diets and this method is useful for the examination of foraging behaviour but it is subject to observational bias (Snow & Snow 1988, Rosenberg & Cooper 1990). Like many other pigeons, C. trocaz is very shy and wary and feeding observations have to be made from a distance, usually through twigs and foliage. Therefore, Ralph et al. (1985) suggested that faecal samples are the best source for assessing the diet, and droppings can be obtained without disturbing the birds.

The macroscopic analysis of faecal samples has been used to describe the diet of many species, including pigeons (e.g. Innis 1989, Powlesland et al. 1994). However it is likely that this approach produces a strong bias towards those items that, due to their greater digestive resistance (e.g. seeds and hard leaves, amongst others), are present in the macroscopic fraction of the sample in higher proportions.

The procedure followed in this study is a microhistological analysis of faecal samples, which allows, potentially, a less biased and more detailed analysis of the diet. The absence of these kinds of studies on pigeon diets could probably be explained by the fact that this is a very complex and time-consuming technique that requires the preparation of a reference collection of all the potential food items. Microhistological techniques have been widely and satisfactorily used on studies of herbivorous mammal (see Sherlock & Fairley 1993, Mohammad et al. 1995) but only occasionally on frugivorous birds (Jordano & Herrera 1981, Herrera 1998).

The aims of this study are to determine (1) the diet composition of Madeira laurel pigeon in the laurel forest, (2) the seasonal variation in diet, (3) the relationship between the abundance of fruits and their consumption, (4) the seed digestive treatment of each plant species, and (5) the reliability of the microhistological method in this type of study. The situation of an endemic island pigeon perhaps feeding on a limited number of laurel forest plants has important evolutionary and ecological implications; it is hoped, therefore, that this information will be found useful for the conservation of the Madeira laurel pigeon as well of other forest pigeons.

Madeira laurel pigeon Chapter 6.3

6.2. Methods and study area

6.2.1. Study area.

The study was carried out in the laurel forest of the Ribeira da Janela. We followed the same paths described in chapter 4 (refer to Figure 4.1.), which ranged from the sea level up to an altitude of 1200 m.

6.2.2. Methods

6.2.2.1. Data collection

Pigeon diet.

Fieldwork was carried out through the year from spring 1996 to the winter of 1997. A total of 224 fresh faecal samples were collected along 3 previously selected paths that crossed the most representative areas of the forest at different altitudes (250, 500 and 750 m a.s.l.). In order to avoid seasonal sampling bias, faecal samples were collected along the three paths during one visit in each of four seasons (March-May: n = 58 faecal samples, June-August: n = 39, September-November: n = 64 and December-February: n = 63). Due to the high abundance and mobility of birds in this, pseudo-replication (many samples originated from the same bird at a specific place) within the faecal sample data should not be significant.

Although the paths were at different altitudes, because of the steepness of the valley the sampling areas were quite close. Considering that birds were seen to move over several km within the valley, the faecal samples could not be used to study spatial variation of the diet. We avoided collecting faecal samples of other species because pigeon faeces have larger size (no other vegetarian bird species of a smaller size inhabits in the study area) and they often show a characteristic morphology (spiral on an edge) when they are recent. Samples were frozen and later analyzed by microhistological methods (Chapuis 1979).

Madeira laurel pigeon Chapter 6.4

Faecal analysis.

Due to the fact that plant food items were often unidentifiable macroscopically, it was necessary to use a binocular microscope with 40 x magnification during the plant item identification. Every faecal sample was dispersed in 100 ml of water and cleared by the addition of two drops (+-2 ml) of commercial leach (Sodium Hypochlorite in 4,5%). From this mixture, approximately 50 µl were taken and put under the binocular microscope and 50 optical fields were randomly selected, using 10 x magnification, from each faecal sample. In each optical field (thereafter OF) the presence or absence of each plant item was scored. Epidermal tissues were identified by comparison with a reference collection of leaves, stems, flowers, fruits and seeds of 139 species from the Ribeira da Janela laurel forest, including trees, shrubs, herbs and ferns.

Faecal samples were present in three forms or types: 1) plant material unidentifiable macroscopically, 2) plant material identifiable macroscopically, and 3) plant material partially identifiable macroscopically. To quantify the relative proportion of each plant item present, different methods had to be applied for each distinct type: Type 1 - we studied 50 randomly selected OF in each sample (previously explained), Type 2 – we transformed the percentage of volume of the macroscopical fraction of each item to OF (considering a total of 50 OF per faecal sample) by a simple rule of three, and Type 3 – we estimated the percentage of the macroscopical fraction to transform it in OF (by the same way used previously in Type 2) and the rest of optical fields to 50 were studied in the micoscopical fraction of the faecal sample (for example, in a Type 2 sample, if 30% is spp A and 70% spp this is transformed into 15 OF of sp. A and 35 OF of sp. B ). This procedure produces a common unit (number of optical fields, OF) for analysis. At the end of the analyses (n = 224 faecal samples), a total of 11 200 Of had been viewed, including those obtained from the transformation of the macroscopical fraction.

Due to the impossibility of quantifying the number of damaged seeds present in the faecal samples (many of them frequently appear destroyed), the effect of the passage of the different seed species through the digestive tract was preliminarily evaluated by the frequency of occurrence in the faecal samples of damagd and undamaged seeds (one sample = one seed).

Madeira laurel pigeon Chapter 6.5

Food resource availability

Phenology of fruits from the five main tree species of the forest (Neves et al. 1996) was described by the use of two measures of fruit abundance: the proportion of trees with fruits and an Index of Fruit Abundance (IFA), as described in chapter 4.

Feeding trials

Feeding trials were carried out to establish the reliability of the microhistological method and particularly to assess the bias introduced by the possible differential digestibility of the food items. Two Madeira laurel pigeons were kept in outdoor cages and fed (the complete samples were very carefully pushed down their throat) with different pair wise combinations of food species (fruits, leaves and flowers) normally present in the diet. The composition of each pair is only important when differential digestibility amongst the components of pair is found. In this cases an extra feeding trial has to be carried out including a food item, chosen amongst those pairs that have not shown differential digestibility. Birds were also fed with plants that, although not present in the faecal samples, were known to be part of the pigeons diet. This was done to confirm that their absence in the samples was due to a sampling effect, rather than total destruction of the epidermis during digestion.

Based on previous knowledge of the stomach contents of several dead pigeons (authors unpublished data), birds were fed with three grams of each item every four hours of daylight. Each combination of food was given to the birds three consecutive times and water and grit were continuously available. After the three consecutive applications of each combination of foods we collected at least one faecal sample per individual for analysis. Sample size of OF for this group of faecal samples was doubled (n = 100/faecal sample) in order to increase the precision of the results.

4.2.2.2. Data analysis

Throughout the analysis two measures are used to quantify the different food items in the diets, percentage of OF and percentage of occurrence in faecal samples. The former was used to describe diet quantification while the latter was employed for statistical Madeira laurel pigeon Chapter 6.6

analysis. Chi-square contingency analysis was used to test for seasonal differences in the frequency of occurrence of particular plant items in the faecal samples. Furthermore, general seasonal variation of the OF (arc-sin transformed) between species and seasons was analyzed by two-way ANOVA. Similarity or overlap in Madeira laurel pigeon diets between the dry (spring and summer) and rainy (autumn and winter) seasons was evaluated using Morisita index of similarity on percentage of OF (Krebs 1999).

Niche breadth (using percentages of OF) was evaluated using the standardized Levinsâ&#x20AC;&#x2122; and Hurlbertâ&#x20AC;&#x2122;s niche-breadth indices, where a value close to 0 indicates dietary specialization, and a value close to 1 indicates a broad diet (Krebs 1999). Hulbertâ&#x20AC;&#x2122;s index incorporates a measure of the proportional abundance of the main food items. Seasonal relationship between availability of fruits (number of fruits/ha) and percentage of fruits in OF was also studied by a simple regression analysis for each of the five main fruit species (L. azorica, O. foetens, P. indica, A. barbujana and M. faya).

Feeding trials were analyzed by the use of Chi-square tests (goodness of fit). After confirming a close similarity between the faecal samples from the two individual pigeons in the different treatments (ANOVA two-way, P > 0.05), the results from the two animals were pooled.

6.3. Results

Diet composition

The diet of the Madeira laurel pigeon included over 33 plant species (Table 6.1.) and at least 80% of the faecal samples contained only one or two plant species. Fruits (pulps and seeds) represented about 57% of OF and leaves (39%) and flowers appearing in a lower percentage (< 1%). Fruits of Ocotea foetens, Ilex canariensis and Laurus azorica were the most important sources of food for the pigeons, found in more than 54% of the OF. However the consumption of Persea indica fruits must be higher according to the underestimation of 41% recorded in the feeding trials (see section below). Therefore, the percentage of OF could oscillate from 8% observed in the faecal samples analysis to

Madeira laurel pigeon Chapter 6.7

11%. After these species, Aspalthium bituminosum, Teline maderensis, Phyllis nobla, and an unidentified herbaceous species were found in more than 8% of the samples.

From the leaves eaten by the pigeons, 59% of the OF belonged to herbaceous and shrubs species, 26% to native trees (mainly I. canariensis: 9.63% of OF) and 9% came from introduced trees (mainly Malus domestica and Prunus persica).

The importance of the food items, measured as OF, is supported by their frequency of occurrence in faecal samples through the year. There was a strong correlation between percentages of OF and occurrence in the faecal samples (Pearson correlation coefficient, rp = 0.956, P < 0.001).

Seasonal variation of the diet

The diet showed a significant variation among the four seasons (two-way ANOVA, F1,32 = 0.002, P < 0.01) (Table 6.1.). However, if we consider the dry (spring and summer) and rainy (autumn and winter) seasons, the pigeons diet showed a subtle variation, with medium overlap or similarity in food items between seasons (Morisita Index, CÎť = 0.59).

In spring, leaves of I. canariensis, A. bituminosum and an unidentified herbaceous plant were the principal foods in the diet, found in more than 55% of OF. The diet in summer was characterized by a larger number of different plant species (at least, n = 22) than in the other seasons (at least, n = 16 in each respectively), with L. azorica (basically fruits) and A. bituminosum (> 12% of OF) being the highest contributors. Fruits of L. azorica, O. foetens and P. indica dominated in the faeces in autumn (â&#x2030;&#x2C6; 70% of OF). Finally in winter, fruits of O. foetens and I. canariensis were the principal species, present in more than 80% of OF. Fruits (pulps and seeds), leaves and flowers were used differentially between the seasons (Ď&#x2021;26 = 64, P < 0.001). Fruits (all from trees) were the major component in the diet during autumn (77% of OF) and winter (85%), while in spring (74%) and summer (55%) vegetative parts of plants (leaves and flowers) were the main component (Figure 6.1.).

Madeira laurel pigeon Chapter 6.8

Table 6.1. Diet and seasonal variation pattern of the Madeira laurel pigeon in Ribeira da Janela. Values are expressed as percentage of optical fields observed in the faecal samples analyzed and as percentage of occurrence observed in all the faecal samples. Bold type indicates values above 5%. * Component apparently underestimated; theoretical percentage is shown in brackets. The part of the plant consumed is indicated in parenthesis : (f) fruit, (l) leaves and (fw) flowers. % of optical fields Species

Spring Summer Autumn Winter

% of occurrence Total

Spring Summer Autumn Winter Total

Lauraceae/Apollonias barbujana (f)

1,03

5,13

3,13

0,16

2,19

1,72

10,26

3,13

1,59

3,57

Lauraceae/Laurus azorica (f)

5,07

20,67

38,35

1,33

16,24

6,80

25,64

46,88

1,59

20,09

Lauraceae/Ocotea foetens (f)

5,80

0,62

18,09

44,63

19,33

10,34

2,56

28,13

57,14 27,23

LauraceaePersea indica* (f)

5,26

15,20

7,78

7,45 (10,5)

5,17

15,63

7,94

Aquifoliaceae/Ilex canariensis (f, l)

25,10

3,74

6,78

35,80

19,16

41,38

7,70

9,38

49,21 28,57

Theaceae/Visnea mocanera (f)

1,38

7,18

1,61

1,72

12,82

2,68

Myricaceae/Myrica faya (f)

0,72

2,56

1,64

0,11

1,13

3,45

5,13

3,13

3,17

3,57

Oleaceae/Picconia excelsa (f)

1,28

0,22

2,56

0,45

Myrtaceae/Eucalyptus globulus (fw)

1,43

0,40

1,59

0,45

Fabaceae/Aspalthium bituminosum (l)

13,69

12,30

0,02

0,78

5,91

34,48

25,64

1,56

3,17

14,73

0,90

0,16

5,13

0,89

Fabaceae/Teline maderensis (l)

3,67

3,20

3,03

1,35

2,79

13,80

5,13

12,5

4,76

9,38

Gen. Sp. Indet. (l)

16,59

1,90

4,83

27,59

Rubiaceae/Phyllis nobla (l)

3,72

6,59

4,29

2,22

3,96

8,62

12,82

12,5

3,17

8,93

Rosaceae/Malus domestica (l)

3,45

8,77

2,42

3,45

10,26

2,68

5,13

0,89

5,13

0,89

Vitaceae/Vitis vinifera (l)

6,93

1,79

12,07

3,13

Ericaceae/Erica scoparia (l)

0,09

0,02

0,25

0,10

1,72

1,56

6,35

2,68

5,13

0,89

5,13

0,89

Fabaceae/Cytisus sp, (l)

Rosaceae/Prunus persica (l)

Chenopodiaceae/Chenopodium ambrosioides (l)

8,04

22,22 13,39

(Continues other side of page)

Madeira laurel pigeon Chapter 6.9

(Continuation of table 6.1.)

% of optical fields Spring Summer Autumn Winter

% of occurrence Total

Spring Summer Autumn Winter Total

Chenopodiaceae/Chenopodium sp, (l)

0,51

0,09

2,56

0,45

Lamiaceae/Bystropogon maderensis (l)

1,49

0,25

2,56

0,45

Lamiaceae/Micromeria varia (l)

2,41

0,42

2,56

0,45

Lamiaceae/Prunella vulgaris (l)

0,10

0,02

1,56

0,45

Lamiaceae (l)

1,13

5,14

0,13

1,70

5,13

15,63

1,59

5,80

Asteraceae/Bidens pilosa (l)

0,50

0,14

1,56

0,45

Asteraceae/Erigeron karvinskianus (l)

0,59

0,17

3,17

0,89

Asteraceae/Sonchus sp. (l)

0,13

0,02

2,56

0,45

0,10

0,05

0,08

0,06

1,72

2,56

1,59

1,34

Caryophyllaceae/Cerastium vagans (l)

0,16

0,04

1,56

0,45

Cruciferae/Arabis caucรกsica (l)

0,02

0,01

1,56

0,45

Cruciferae/Erysimum bicolor (l)

0,03

0,04

1,72

0,45

Pteridophyta

1,88

1,20

0,89

5,17

5,13

2,23

0,86

0,32

0,33

14,06

3,17

4,91

Poaceae (l)

Bryophyte Total number of species

Madeira laurel pigeon Chapter 6.10

100

% of optic fields

80 60

Fruits Veg. Parts

40 20 0 Spring

Summer

Autumn

Winter

Figure 6.1. Variation of the presence of fruits and vegetative parts of plants, leaves and flowers (all expressed as percentage of optical fields), throughout the different seasons in the diet of the Madeira laurel pigeon.

Resource use and fruit availability

Levin’s niche breadth, B, was clearly narrower in the rainy seasons (autumn and winter, B = 0.22 and 0.13, respectively) than in the dry ones (spring and summer, B = 0.36 and 0.44, respectively). This suggests that the Madeira laurel pigeon has a tendency to specialize when fruit resources are abundant (in autumn and winter). Considering only fruit availability, Hurlbert’s niche breadth (B’), which weights resource use by an estimate of abundance, gave a broader niche value for the rainy seasons (B’ = 0.66) compared to the dry seasons (B’ = 0.10). These results are logical since there is a much greater availability of fruit in autumn and winter (Figure. 6.2.). This figure, also shows the close relationship between the abundance of the fruits and their consumption by the pigeons in the different seasons for all studied species (L. azorica: r2 = 0,78, O. foetens: r2 = 0,75, P. indica: r2 = 0,72, A. barbujana: r2 = 0,86 and M. faya: r2 = 0,67; P < 0,05 and n=4 for all species).

Madeira laurel pigeon Chapter 6.11

Ocotea foetens

Laruus azorica

% of optic f ields

f ruits /ha 1000

4000

800

3000

600 400

30 20

2000

200

1000

Persea indica

20 10 0

A pollonias barbujana 20

250 200

100

6 5 4 3 2 1 0

150

M yrica faya

All spe cie s

800

600

400 1

200

S pr.

S um .

A ut.

Seas ons

40 30

W in.

5000

100

4000

3000

2000

1000

S pr.

S um .

A ut.

W in.

Seas ons

Figure 6.2. Relationship between fruit abundance, expressed as fruits/ha (bars), and consumption, expressed as percentage of optical fields (line), throughout the different seasons of the year.

Seed digestion

Seeds eaten by the Madeira laurel pigeon were found both destroyed and externally unharmed depending on the fruiting tree to which they belong (Table 6.2.). Seeds from I. canariensis, M. faya, V. mocanera, O. foetens and A. barbujana were rather resistant to mechanical digestive treatment, while those of P. indica and especially L. azorica were frequently damaged.

Madeira laurel pigeon Chapter 6.12

Total number of faecal samples with seeds

% faecal samples containing damaged seeds

Ilex canariensis

Species

Myrica faya

Visnea mocanera

Ocotea foetens

Apollonias barbujana

28.6

Persea indica

42.1

Laurus azorica

82.1

117

33.3

TOTAL

Table 6.2. Frequency of seed occurrence in the faecal samples and effect of their passage through the Madeira laurel pigeon digestive tract.

Feeding trials

Seven of eight combinations of items given to the captive pigeons yielded no significant differences (Table 6.3.). Therefore, important differential digestibility amongst the main components of the birds diet was not observed, and these results strongly validate the data obtained in the food analyses. The only exception, in a total of 16 food items given, was P. indica where there was a 41% underestimation of its occurence.

Feeding trials Sp1

vs.

Number of OF

Sp2

Sp1

Statistical parameters Sp2

Ď&#x2021;2

Ocotea foetens vs. Ilex canariensis (fruits)

0.30

0.58

Persea indica vs. Laurus azorica (fruits)

4.90

0.03

Laurus azorica vs. Apollonias barbujana (fruits)

0.09

0.75

Visnea mocanera vs. Myrica faya (fruits)

2.10

0.14

Aspalthium bituminosum vs. Phyllis nobla (leaves)

1.18

0.27

Aspalthium bituminosum vs. Cytisus striatus (leaves)

0.035

0.85

Rumex maderensis vs. Phyllis nobla (leaves)

1.02

0.31

Brassica oleracea (leaves) vs. Cytisus striatus (flowers)

0.69

0.40

Table 6.3. Feeding trials carried out using the most frequent combinations of the main components of the diet of the Madeira laurel pigeon and also other species that are known to be used. Under the columns P1 and P2are presented the results of each individual bird involved in the experiment and T is the sum of both results.

Madeira laurel pigeon Chapter 6.13

6.4. Discussion

Diet composition

Our findings indicate that a wide range of different plant species are found in the diet of Madeira laurel pigeon. It is basically a frugivorous pigeon although leaves and flowers are also well represented. Fruits of most species present in the laurel forest are used but those of Ocotea foetens, Laurus azorica, Persea indica and Ilex canariensis proved to be the most important ones. If the use of O. foetens, L. azorica and P. indica could be explained by their importance, dominance and density in the forest (Neves et al. 1996 and data presented on chapter 4), the former producing fruits all year round, the same explanation cannot be applied to I. canariensis. This species is somewhat rare in the laurel forest of Madeira, especially in our study area, where it has a clumped and patchy distribution.

Corroboration of the diet analysis findings comes from the study of habitat use (chapter 4), where birds were found using trees in preference to the round and shrub layers, trees were not used randomly and O. foetens proved to be the preferred species, almost all year round. Furthermore, for all species, trees with ripe fruits were significantly favored over the others. Field observation also showed that pigeons feed on other plants such as Rumex maderensis, Plantago major and Nasturtium officinale, which did not appear in the microhistological analysis.

The data also agree with most of the descriptive literature on the diet of the Madeira laurel pigeon, which suggests that the main food source is the fruits of the various trees of the laurel forest, namely fruits of Laurus azorica, Ocotea foetens and Myrica faya (Harcourt 1851, Godman 1872, Bannerman and Bannerman 1965, Zino 1969, Zino and Zino 1986). Previous work has also stated that when feeding on the ground pigeons also eat the flowers and leaves of many plants like Sonchus spp., Apium nodiflorum, Nasturtium officinale (Zino and Zino 1986) and Phyllis nobla (Oliveira and Jones 1995).

Madeira laurel pigeon Chapter 6.14

Although the Madeira laurel pigeon has a fairly wide diet, there is an obvious preference for fruits over leaves and flowers. Nutrient analyses of several fruits of different families (e.g. Powlesland et al. 1997) suggest that, as a consequence of their relatively high levels of carbohydrates, fruits provide birds with a more concentrated source of energy than do other foods.

The diet of the other two endemic pigeons from the macaronesian islands, Canaries (C. bollii and C. junoniae) is also basically frugivorous, including a wide range of fruits and some vegetative parts of plants (MartĂn et al. 2000). Other forest fruit pigeons such as the Chatham Island Pigeon, Hemiphaga novaeseelandiae chathamensis (Pearson and Climo 1993; Powlesland et al. 1997) and White crowned pigeon, Columba leucocephala (Bancroft and Bowman 1994) show a similar varied vegetarian diet. The data presented also demonstrate the importance of lauraceous fruits in the diet of some species of pigeons (Crome 1975).

Diet seasonal variation and availability of fruits

The diet of the Madeira laurel pigeon showed a clear seasonal variation. Fruits were mostly used in autumn and winter exactly when the fruit production of the forest was at its highest. During the former season L. azorica, O. foetens and P. indica were very important in the diet; throughout this period not only was the overall fruit production of the forest at its maximum, but particularly that of L. azorica and P. indica. In winter fruits of O. foetens and I. canariensis were also heavily consumed in concordance with the fruit abundance patterns identified for both species. It is interesting to note that the consumption of the fruits and leaves of I. canariensis is much higher than one should expect on the basis of their clumped distribution and low abundance in the forest. Despite the absence of any systematic data, it is known that fruit production in this species is seasonal, with ripe fruits present in early winter to late spring (Pers. Obs.). The habitat use study carried simultaneously already suggested the existence of some kind of selection and preference (refer to chapter 4). Factors influencing fruit choice were also already discussed on the mentioned previous chapter. In summary they will depend on fruit conspicuousness, characteristics related to size and nutritional composition, among other fruit traits (for references and more detailed discussion refer

Madeira laurel pigeon Chapter 6.15

to chapter 4). Probably leaves choice will follow similar patterns, although we didn’t find any studies on this issue.

In spring the use of leaves (I. canariensis, A. bituminosum and an unidentified herbaceous species) becomes more important. This is presumably a result of the low fruit production in the forest at that time. The high diversity of species of leaf consumed reaches its maximum during the summer, when at least 22 species were identified in the faecal samples. Yearly fruit production reached minimum values during this season, even lower than those of spring.

These patterns of resource use suggest that insular pigeons, that live in a limited area and habitat, need to be highly flexible in order to exploit each food resource as it becomes available. This flexibility in diet was demonstrated by the strong seasonality in pigeons diet, incorporating dietary switching, and niche-breadth expansion and contraction. Hurlbert’s index and the positive correlation between percentage of OF and availability of fruits indicate that niche-breadth is determined by food availability. Madeira laurel pigeons exhibit a narrower diet in response to a high availability of fruits but acquire a wider food spectrum when this resource is scarce. These strategies may enable the Pigeon to closely track a fruit resource base which exhibits high temporal variability in abundance, and may suggest an evolutionary relationship between this pigeon and plant fruit resources.

The study of habitat use carried out simultaneously at the same site (refer to chapter 4 for details) corroborates the seasonal patterns observed in the diet. When the total abundance of fruits in the forest decreased, birds shifted from the trees to the ground and shrub layers. This occurs in late spring and early summer, exactly when O. foetens fruits reach their lowest abundance. Furthermore, on a wider scale, Oliveira and Jones (1995) were able to show a marked relation between fruit production and habitat use.

In the Canaries, in the particular case of C. bollii, most of the consumption of vegetarian material occurred during the spring and summer (Martín et al. 2000), probably coinciding with a period of low fruit availability (Hernández et al. 1999). Unfortunately, there is not much information on the temporal variation in diet of C. junoniae.

Madeira laurel pigeon Chapter 6.16

In a wider geographic context, Innis (1989) showed that foraging habits of fruit pigeons in sub-tropical forests in Australia were largely opportunistic; the birds used whatever fruits were available at particular times. Furthermore, Frith et al. (1976) and Crome (1975) identified the same pattern for several Australian Columbidae species. In a seven year study carried out in Costa Rica Wheelwright (1986) showed that birds shifted their use of different lauraceous fruit species as same became scarce and others abundant,

The use of other parts of plants by fruit pigeons, when fruit is in short supply, has been documented for other species such as in Hemiphaga novaeseelandiae chathamensis (Powlesland et al. 1997). Furthermore the consumption of some leaves and herbs all year round, even when fruit production is high, could imply that these foods could supply important diet components. This strategy of using supplementary food sources, is present amongst frugivores generally (Jordano 1988, Izhaki and Safriel 1989, Sallabanks and Courtney 1992) and has already been described for other fruit pigeons, namely the White crowned pigeon, Columba leucocephala (Bancroft and Bowman 1994) and the Chatham Island Pigeon, Hemiphaga novaeseelandiae chathamensis (Powlesland et al. 1997). The latter species seems to be able to supplement their protein intake by including herb foliage (available all year round) and flower buds (available seasonally) in their diet.

Seed digestion

Most of the seeds of O. foetens and A. barbujana present in the faecal samples passed unharmed through the digestive tract of the birds. However, almost half of the P. indica seeds and most of L. azorica present in the faecal samples appeared damaged. These data suggest that pigeons potentially act as legitimate seed dispersers for the first two mentioned species. Given that O. foetens is the only fruit consistently present throughout the year (Oliveira and Jones 1995, unpublished data and present study), it is probable that, in the long run, their seeds will have a greater probability of dispersal by the pigeons than the other trees. The way this might influence forest composition was discussed in chapter 4.

The other tree species present in the faecal pellets (I. canariensis, M. faya and V. mocanera), being smaller in size and more compact showed a high resistance to the

Madeira laurel pigeon Chapter 6.17

effects of the digestive treatment. Despite the low frequency of the last two these data suggest that the Madeira laurel pigeon is a potential dispersal agent. One should note that the relationship between birds and the dispersal of seeds of the species upon which they feed is not always straightforward and depends on many factors (Pratt and Stiles 1983, Sallabanks and Courtney 1992). However the fact that the seeds pass intact the digestive tract of the pigeons is just one of the many conditions that is needed for an effective dispersal of the seeds.

Validation of the microhistological method

Differential digestibility and rates of passage through the digestive tract may introduce some bias in the analysis (Rosenberg and Cooper 1990). However, Rosati and Bucher (1992) have suggested that the impact on the estimated botanical composition will rarely cause significant alterations in diet estimates. Furthermore, some authors have previously suggested microhistological methods as a suitable alternative technique to study avian diets (Jordano and Herrera 1981, Herrera 1998).

In order to validate the results obtained in this study, we performed parallel feeding trials. When the birds were fed with known amounts of different components of the diet (in different combinations), these were mainly present on the faecal samples in about the same proportions as expected. Plants that were known to be eaten by the Madeira laurel pigeon, but were absent or in a very low frequency in our faecal samples, were also found in the feeding trial faecal samples in the expected proportions (leaves from Rumex maderensis and Cytisus striatus). Moreover, despite the greater digestibility of the flower soft tissues, feeding trials carried out proved that they are likely to occur in the same proportion as the other vegetative parts of the plants. This strongly validates our conclusions on the composition and seasonal changes in the diet.

In the light of these findings we suggest that the microhistological techniques and procedures described here provide an acceptable level of accuracy and are appropriate for the study of herbivorous pigeon diet and probably extensible to other bird species that consume abundant plant food.

Madeira laurel pigeon Chapter 6.18

7. Winter diet on agricultural fields: a complementary study

7.1. Introduction The Madeira laurel pigeon is the major bird pest of the archipelago. More than 100 years ago, Goodman (1872) had already recorded the use of cultivated fields by the pigeons, so this is undoubtedly a very old habit. The pigeons feed on a wide variety of crops and fruit trees throughout the island, mainly, between late February and late July. Damage has been recorded on cabbages, peas, beans, watercress, walnut trees, cherries trees and peach trees amongst others (Zino and Zino 1986, Oliveira 1999a). Before the present study, feeding on agricultural fields had never been studied systematically. Consequently, the proximate factors leading to such behaviour and feeding habit have never been fully explained. Patterns of use of agricultural fields and the relationship between feeding on crops and fruit abundance on the forest was studied in chapter 5 and 4 respectively and the relationship between diet and fruit availability was analysed in chapter 6. The data presented there referred to samples collected in the forest and obviously did not provide information on the individuals feeding on crops. Although it is known that the birds use the crops mainly from late winter to late summer, the present study took place only in the winter when fruit production was very high. Consequently this is a good time to look at food choice in the fields, i.e., birds are not attacking the fields because of starvation. The specific aim of this chapter is, therefore, to provide information on the diet of the pigeons when they are feeding on agricultural fields. The results will hopefully help in the interpretation of the data on the birdsâ&#x20AC;&#x2122; diet and movements elsewhere and perhaps also help with the development of management strategies.

Madeira laurel pigeon Chapter 7. 1

7.2. Methods and study area 7.2.1. Study area Samples were collected from 3 areas of cultivated fields, located between 500 and 600 metres a.s.l. on the boundaries of the laurel forest â&#x20AC;&#x201C; ChĂŁo da Ribeira, Lombo Galego and Cedro Gordo (Figure 7.1). The 2 latter sites will be considered together due to their proximity and similarities, henceforth referred to as Faial. Distance from primary laurel forest varied between the sites. At ChĂŁo da Ribeira fields were located close to patches of well preserved forest (between 50 to 200 m), whilst at Faial they were more than 300 m away. The fields are surrounded by a mix of exotic and indigenous trees. Both areas are heavily cultivated

and the land is divided into small terraces for subsistence

agriculture. There are some exceptions in the Lombo Galego area of Faial where some fields are relatively bigger.

Figure 7.1. Agricultural areas where faecal samples were collected. The grey areas show the distribution of the laurel forest. Ribeira da Janela valley where the other study of the diet was carried out (refer to Chapter 6 ) is presented for reference purposes.

Madeira laurel pigeon Chapter 7. 2

7.2.2. Methods 7.2.2.1. Data collection Pigeon diet Fieldwork was carried out in January and February 1997. A total of 94 fresh faecal samples were collected in 21 fields located inside the 2 study sites. Within each area the maximum distance between fields was less than 250 m. In order to avoid sampling bias, only one visit was made to each field (ChĂŁo da Ribeira: n=50 faecal samples, Faial: n=44, L. Galego n=22 and C. Gordo n=22). When more than one dropping was found near an eaten item, in order to avoid pseudo-replication of faecal samples, only the fresher sample was collected. Pigeon faeces are easily identifiable when they are recent because of their large size (no other vegetarian bird species of medium size as this pigeon â&#x20AC;&#x201C; 462.5 gr., n=36 - inhabits in the study area) and characteristic morphology (spiral on an edge). Samples were frozen until they were analyzed by microhistological methods. Analysis of faecal samples and feeding trials were carried out following the microhistological methods presented in chapter 6.

7.2.2.1. Data Analysis Two measures are used to quantify the different food items in the diet, percentage of OF that an item occurred in and percentage occurrence in faecal samples. As in chapter 6 the former was used to describe diet quantification while the latter was employed for statistical analysis. Chi-square analysis was used to test for spatial (site vs. site) differences in the frequency of occurrence of particular plant items in the faecal samples (tests were based upon comparisons of presence and absence in optical fields). Niche breadth (using percentages of OF) was evaluated using the standardized Levins Index, where a value close to 0 indicates dietary specialization, and a value close to 1 indicates a broad diet (Krebs 1999). Feeding trials data were analyzed as presented in chapter 6.

Madeira laurel pigeon Chapter 7. 3

7.3. Results

Diet composition When feeding on agricultural fields the diet of the Madeira laurel pigeon included 12 plant species (Table 7.1) and about 60% of the faecal samples contained only one or two plant species. Just one crop was present in our samples â&#x20AC;&#x201C; Brassica sp. â&#x20AC;&#x201C; which represented about 54% of the OF; the mean number of droppings with Brassica alone was of about 36 %. Leaves of various plants and fruits of Ocotea foetens accounted respectively for 34 and 2 % of the OF.

% of optic fields

% of occurrence

(n = 4700 optic fields)

(n = 94 faecal samples)

Brassica cf. oleraceae (l, fw)

54,2

84,0

Erysimum bicolor (l)

1,2

9,6

species Cruciferae

Fabaceae Aspalthium bituminosum (l)

0,08

2,1

Teline maderensis (l)

13,5

44,7

15,9

51,1

1,57

5,31

Laurus azorica (l)

0,1

1,06

Ocotea foetens (f)

2,3

5,3

0,02

1,06

Gen. spp. indet. (l)

Tr.

1,06

Prunus persica (l)

1,02

1,1

0,5

4,2

Rubiaceae Phyllis nobla (l) Lamiaceae Gen. spp. indet. (l) Lauraceae

Aquifoliaceae Ilex canariensis (f, l) Poaceae

Table 7.1. Plants species identified in the faecal samples in the two study areas. Values are expressed as percentage of microscope optic fields (OF) and as percentage of occurrence observed in the droppings. Values higher than 5% are indicated in bold type and plant part used presented as: f = fruit, l = leaves and fw = flowers; tr = trace amount (< 0.05%).

From the leaves eaten by the pigeons, 96,6 % of the OF belonged to herbaceous and shrubs species, while the remaining 3,3 % came from one introduced tree (Prunus

Madeira laurel pigeon Chapter 7. 4

persica) and two native trees (Laurus azorica and Ilex canariensis). From the herbaceous and shrubs species, Phyllis nobla and Teline maderensis accounted for the majority of the OF; 48 and 41% respectively. The importance of the food items, measured as OF, is supported by their frequency of occurrence in faecal samples. There was a strong correlation between percentages of OF and occurrence in the faecal samples (Pearson correlation coefficient, rp = 0.942, P < 0.001).

Spatial variation of the diet Spatial variation of the diet between the two sites is shown in Table 7.2. Qualitatively there is a considerable similarity between both areas. Fifty percent of the species were found simultaneously in samples from both places and the food spectrum was also very similar. At Chão da Ribeira birds used one more species (9) but Faial was the only place where fruits of Ocotea foetens were found. However, there was spatial variation in the occurrence frequency of particular plants species in samples and optical fields. Although cabbage was the highest food component consumed both in Chão da Ribeira (χ21 = 29.8, P < 0.001) and Faial (χ21 = 867.9, P < 0.001), it showed a significant higher presence in those samples from Faial respect to Chão da Ribeira (χ21 = 160.8, P < 0.001). Phyllis nobla and Teline maderensis, typical shrubs from the edges of the laurel forest, were mainly consumed in Chão da Ribeira than in the other locality (χ21 = 872.6, P < 0.001 and χ21 = 215.0, P < 0.001, respectively). More than 60% of fecal samples from Faial contained plant remains of only one or two species, while they were more diverse in Chão da Ribeira (χ21 = 35.1, P < 0.001). Furthermore, Levins’ niche breadth was clearly narrower in Faial (B = 0.07) than in Chão da Ribeira (B = 0.24). This suggests that the Madeira laurel pigeon showed a tendency to dietary specialization in sites far from forest areas, while this pattern changed when the distance to laurel forest was shorter

Madeira laurel pigeon Chapter 7. 5

species

Ch達o da Ribeira % of optic fields

% of occurrence

Faial % of optic fields

% of occurrence

Cruciferae Brassica cf. oleraceae (l, fw)

45,6

89,0

64,0

88,6

Erysimum bicolor (l)

2,2

16,0

0,05

2,3

Aspalthium bituminosum (l)

0,2

4,0

Teline maderensis (l)

20,1

76,0

6,0

9,1

29,0

84,0

6,0

13,6

1,5

8,0

1,7

2,3

Laurus azorica (l)

0,2

2,0

Ocotea foetens (f)

5,0

11,4

0,05

2,3

Gen. spp. indet. (l)

Tr.

2,0

Prunus persica (l)

2,2

2,3

0,1

4,0

1,0

6,8

Fabaceae

Rubiaceae Phyllis nobla (l) Lamiaceae Gen. spp. indet. (l) Lauraceae

Aquifoliaceae Ilex canariensis (f, l) Poaceae

Table 7.2. Spatial variation of the diet between the two sites under study. Values are expressed as percentage of optical fields observed in the faecal samples and as percentage of occurrence in all faecal samples. Values higher than 5% are indicated in bold type and plant part used presented as: f=fruit, l=leaves and f =flowers; tr. = trace amount (< 0.05%).

7.4. Discussion The results indicate that a number of plant species are included in the diet of the Madeira laurel pigeon when feeding outside the forest in winter but, only a relatively small proportion of species are used abundantly. In the previous diet study carried out at Ribeira da Janela (refer to chapter 6), it was demonstrated that the feeding niche breadth of pigeons is determined by food availability and that it is narrower in winter when there is high fruit production. Since the present study was carried out in winter, the narrow feeding niche breadth found may not come as a surprise. However, a very low proportion of fruits was found in the diet of the birds feeding on agricultural fields. The differences between the use of fruits and vegetative part of plants, including leaves and Madeira laurel pigeon Chapter 7. 6

flowers of Brassica sp, inside the forest and in agricultural fields in winter can be appreciated in Figure 7.2.

% of occurrence 100 80 Fruits

Non fruits

40 20 0 forest

agricultural

Figure 7.2. Use of fruits and vegetative parts of plants inside the forest (data from section 6) and in the agricultural fields during winter.

It is interesting to notice that fruits are almost absent from the diet on agricultural fields while at the same time they comprise more than 80 % of the food consumed inside the forest. Assuming that in ChĂŁo da Ribeira and Faial fruit production follows the same patterns as that of Ribeira da Janela (chapter 6), it is difficult not only to explain this difference in the diet but also the decision to feed on crops. Thus, the question remains: why do birds leave their natural habitat to feed on crops when nutritious food sources are abundant? Even assuming that fruit availability in ChĂŁo da Ribeira and Faial is low, the explanation is still not straightforward. During periods of fruit scarcity pigeons are able to expand their food spectrum inside the forest. Under these circumstances one should expect a wider variety of species in the pigeonâ&#x20AC;&#x2122;s diet when they are using agricultural fields. The only possible explanation is that crops, in this case Brassica sp, are quite advantageous food sources for the birds; when faced with the decision of moving unknown distances in search for fruits or feeding on very predictable and visible food sources, some birds obviously choose the latter. Further studies are needed to confirm

Madeira laurel pigeon Chapter 7. 7

such hypothesis; it is important to know the physiological needs of the Madeira laurel pigeon as well as the proportion of the population that use crops throughout the year. It has been suggested (Chapter 5) that individual pigeons will normally use each field or nearby group of fields intensively until full depletion occurs. This is in accordance with the high proportion of Brassica sp. found in our faecal samples. It is possible that after feeding for a certain period of time on agricultural fields individual birds will move on to search for other food sources, namely fruits. The difference in diet between ChĂŁo da Ribeira and Faial may indicate a higher dietary specialisation in sites far from the forest. This may be related to a decrease in the availability of food and to the fact that the birds chose to fly there to use the crops. There are two potentially important sources of bias in our data. Firstly, the above mentioned â&#x20AC;&#x153;fidelityâ&#x20AC;? to fields and, secondly, the fact that during one feeding session birds normally fly from one cabbage to another. Putting these two aspects together, it is impossible to be sure that droppings from the same bird were not repeatedly collected. However, the high number of birds that normally feed together in the fields suggests that there may not be too much pseudoreplication. In order to validate the results obtained with the microhistological method, in the previous diet study we performed parallel feeding trials (for details refer to chapter 6). When the birds were fed with known amounts of different components of the diet (in different combinations), these were present on the faecal samples in about the same proportion. Teline maderensis was the only species well represented in the diet to which any feeding trial was carry out. However, at the light of the results obtained with other species, namely Cytisus striatus, It is believed that the probability of T. maderensis being under or sub estimated is relatively small. In summary, birds that fed on crops behaved similarly in different areas and although their diet differs considerably from the diet of those birds that feed inside the forest, natural and crop plants are still taken together.

Madeira laurel pigeon Chapter 7. 8

Further research is needed to fully understand the relationship between crop use and food availability. Such research is particularly true for the spring and summer periods when the use of agricultural fields increases significantly.

Madeira laurel pigeon Chapter 7. 9

8. General discussion

This study was aimed at increasing our understanding of the conservation and management of the endemic Madeira laurel pigeon by (1) determining the relationship between habitat use, diet and the use of anthropogenic habitats and (2) examining feeding choices on agricultural fields. It has been addressed by a hierarchy of scale levels from general movements and broad habitat use, down to feeding choices in the forest and feeding behaviour at agricultural fields. The results suggest that the correct management of this island species can only be achieved by integrating all these scale levels in a multidimensional approach. This chapter will deal with a general overview on the issues discussed previously, which will provide the framework for a future management plan for the Madeira laurel pigeon and guidelines for the update of the species Action Plan (Refer to Chapter 9).

Madeira laurel pigeon: an island species

Being an island species the Madeira laurel pigeon raises very specific management problems that rarely occur with mainland threatened birds. In summary, it has a restricted habitat and inherently small population numbers. As discussed previously pigeons and doves are a very successful group occupying most of the globe and are present on many islands worldwide. Paradoxically, their ability to travel long distances and to adapt to new environments is a source of vulnerability. In many remote islands, where they differentiated in isolation, pigeons became endangered, or even extinct, soon after the arrival of man. In the last 150 years, 6 pigeon species have become extinct and Collar et al. (1994) suggest that 40 species are currently threatened with extinction. Furthermore the IUCN 2002 Red List of Threatened animals listed 12 pigeons as critically endangered, 14 as endangered and 34 as vulnerable. Of the large bird families in the world few contain such a high proportion of threatened species.

In much earlier historical times flightless species with pigeons affinities have become extinct on islands (for details and references refer to Chapter 1). Perhaps the most famous of these was that of the Dodo, Raphus cucullatus, a giant and flightless species of a pigeon-like bird, which lived on Mauritius. The species is symbolic of the Madeira laurel pigeonChapter 8. 1

combination of insular specialisation, evolutionary simplification but vulnerability to extinction (Grant 1998). Most likely, the Dodo fed mainly on fallen fruit. Since food was widely available and there were no predators, there was no need to fly and subsequently it grew large in size and became flightless. The big beak allowed the species to use a wider range of food sources, and the fat storage enabled it to survive throughout the seasons of scarcity (Quammen 1996). Flightlessness also dictated that the dodo put its nest on the ground. Such practice was not a drawback, since ground nesting in this ecosystem was safe and economical. On the whole, these were, apparently, all well-balanced compromises that yielded more advantages than disadvantages. Nevertheless, it was a one way road to human-mediated extinction. Mauritius, in its primordial state contained no terrestrial mammals and the Dodo was not prepared, literally and ecologically, for mankind and his pigs, monkeys, goats, cats and dogs. Some of these animals, like the goats, started competing for food, others like the pigs and monkeys were deadly predators of the flightless Dodo, its juveniles and its nests.

Having arrived in small numbers and being omnivorous, these introduced mammals were not limited by a single food supply. This was one side of the problem; the other was man himself. The Dodos were very tame and fearless of men, thus not surprisingly they were a very easy catch. The combined action of mankind and animals resulted in the inevitable extinction of the Dodo. The last credible eyewitness's account of living Dodos dates back to1662 (Quammen 1996)

After all these years the Dodo stands as the best example of insular evolution involving the transformation of an ‘adventurous’, strong flying ancestral species into a grounded descendent no longer capable of going anywhere but towards complete extinction when faced with man’s destructive power. The Dodo, probably an ancient species which survived for million of years far from human interference, stands as a remainder that insular evolution, for all its wonderousness, tends to be a one-way ticket to extinction in the modern world.

Soon after the extinction of the Dodo an endemic tree species of the Sapotaceae family – Calvaria tree or Tambalacoque Sideroxylon grandiflorum – reduced drastically their rate of germination. According to many authors there is a cause/effect link between

Madeira laurel pigeonChapter 8. 2

these two incidents. It has been widely proposed that the germination of Calvaria seed was greatly enhanced (if not obligatory) by the passage through the digestive tract of the Dodo (Quammen 1996 and references therein). Like the Dodo, and many other pigeons and doves, the Madeira laurel pigeon was a victim of island evolution and specialisation. It is an island species adapted to a restricted habitat, with a narrow diet and single egg clutch, as examples of adaptations to island environment. Same of the key differences between the Dodo and the Madeira laurel pigeon are that it is has not developed flightlessness and it is still adaptable enough to use foods provided by humans.

Population numbers, habitat use and diet

Nevertheless, the impact of man has been severe, the pigeonâ&#x20AC;&#x2122;s present distribution range is largely a reflection of the available area of native forest and its margins; after the arrival of man, the laurel forest was reduced to 15% of its previous likely coverage. Moreover, the pigeon became extinct on the nearby island of Porto Santo, where most certainly the bird had managed to adapt to a quite different habitat.

After this historical severe decline in abundance and distribution, the population was probably kept at very low numbers by both legal and illegal hunting. The increase in numbers between 1986 and 1995 is likely to be a response to the reinforcement of the protection to the pigeons, namely the ban on hunting and the establishment of the Natural Park of Madeira (for details refer to chapter 3).

Although little is know about the population dynamics of Island pigeons, in particular from those of the Macaronesian archipelagos, the present Madeira laurel pigeon population may well be large enough to ensure long-term survival (as long as there are no significant changes in habitat legislation). In the literature, there are several examples of species, belonging to different bird families, restored from extremely low numbers whose capacity to recover does not seem to have been affected, including the Chatham Island Black Robin Petroica traversi, Hawaiian goose Branta sandvicensis and the Cook Island kakerori Pomarea dimidiata (Groombridge et al. 2001 and references therein). These species recovered from numbers much smaller than those of the Madeira Madeira laurel pigeonChapter 8. 3

laurel pigeon and so short term survival may not be a problem for the pigeon. Rather more problematic is medium to long term survival which may be affected by the loss of genetic diversity experienced by small populations. On the basis of the populationviability theory a species chances of survival will be severly affected by extended periods of small population size (Nunney and Campbell 1993). One could speculate that the population on Madeira has not been low enough for long enough for there to be a significant loss.

As population size increased a density dependent mechanism may have been responsible for the greater use of theoretically less suitable patches of the forest. This is possible due to the adaptability of the Madeira laurel pigeon, which is able to use many resources of the forest distributed amongst the different forest strata â&#x20AC;&#x201C; tree canopy, bushes or young trees and herbaceous vegetation. Whether this is an acquired island characteristic or not will be dealt with below.

Like most frugivorous pigeons, the Madeira laurel pigeon undertakes short â&#x20AC;&#x201C; range erratic migrations while searching for food. In large continuous areas of the forest (e.g. Ribeira da Janela), birds tend to move up and down the valley rather than from valley to valley. In more fragmented patches that are unable to maintain the pigeons all year round (e.g. Funduras), movements are likely to follow different patterns.

The data presented in this study consistently show that one factor which influences habitat use and diet, and consequently movements, is fruit abundance (for details refer to chapters 4 and 6). In turn the overall availability of fruits in the forest is largely dictated by the phenology of Ocotea foetens, which it is the largest and most dominant tree in the Madeira laurel forest. For this reason these fruits are the most important quantitative, and probably qualitative, food source for the Madeira laurel pigeon. It should be stressed, however, that the importance of this fruiting tree for the pigeons might be merely circumstantial, i.e. it is used a lot by the pigeons just because it is common and not because of active discrimination. This type of relationship has been found elsewhere by Firth et al. (1976), who compared his results with those of Crome (1975a) and found that the differences between the relative importance of a few plant species to Pitinolopus magnificus, P. superbus and Ducula spilorrhoa are mainly because availability differs between the forests where the studies were carried out.

Madeira laurel pigeonChapter 8. 4

Webb et al. (1999) also corroborates this type of relationship for the Pacific pigeon Ducula pacifica from the Samoan archipelago.

Although there may be no evidence of an active choice of O. foetens, pigeons are the only native dispersal agents for the trees of the indigenous forest of Madeira and may be indirectly responsible for the dominance of O. foetens â&#x20AC;&#x201C; this species bears fruits consistently throughout the years, the seeds are not destroyed by the passage throw the pigeons digestive tract and so in the long run they will have a much greater probability of dispersal than other seeds. This possible relationship between pigeon feeding habits and forest composition and structure does not imply any kind of mutual adaptation, it only relies on the use/availability patterns that were identified.

More evidence from longer-term studies is obviously required to establish the ecological and evolutionary relationships between the pigeon and its diet species but the findings so far suggest a relationship between endemic bird and its habitat which is rather more diffuse than the Dodo â&#x20AC;&#x201C; Calvaria example refereed to earlier in this section.

During periods of fruit scarcity (spring and summer) pigeons shifted from dense canopied forested patches to forest openings where they relied mainly on flowers and leaves of various plants. This suggests that the species requires a mosaic of habitat types to be successful. Other studies have shown that seasonal changes in resource levels and habitat conditions mean that resident frugivores and herbivores require extensive areas of different habitats throughout the year (e.g. Ramos 1993 and references therein). When considering the management of the Madeira laurel pigeon these data are very relevant, as they highlight the importance of marginal and secondary laurel forest areas. Until now it was thought that pigeons survival was exclusively linked to the protection of the best preserved areas of laurel forest.

In spring and summer there is a wider structural and functional use of the forest which is directly linked to diet shifts. The seasonal variation in the diet showed that the Madeira laurel pigeonâ&#x20AC;&#x2122;s niche breadth is determined by food availability (for details refer to chapter 6). They exhibit narrower diets in response to a high availability of fruits and they acquire a wider food spectrum when this resource is scarce. Such a

Madeira laurel pigeonChapter 8. 5

strategy may enable this species to closely track a fruit resource base that exhibits high temporal variability in abundance.

The role of food supply hypothesis in explaining patterns of island bird abundance has been invoked in several studies e.g. Galapagos finches (Schluter 1986) and the Hawaiian honeycreeper Loxioides bailleui (Scott et al. 1984). With food resources being less diverse on islands together with other factors such as predation and weather bearing relatively little importance (Schluter and Repasky 1991) this seems a logical explanation. The flexibility in diet and habitat use of the Madeira laurel pigeon is an evolutionary feature that enhances the survival probability of the species. It can be seen as a successful strategy to overcoming the constraints of food supply as a limiting factor.

Niche expansion (broader spectrum of occupied habitats and change in feeding habits) of insular populations have been documented by many authors (MacArthur et al. 1972, Lack 1976, Blondel 1985 amongst others). The classical explanation for niche expansion is that extra resources become available because of a lack of competitors and other features of the insular communities. Although the Madeira laurel pigeon could be said to have a broad niche (wide structural use of the habitat, a wide food spectrum and uses the fruit from all the trees present on the Madeira laurel forest) this is not necessarily an example of niche expansion; pigeons as a group inherently have great adaptability and can use many different habitats and food sources. Moreover, a similar diet composition â&#x20AC;&#x201C; combining the consumption of fruits with vegetative parts of plants, has been recorded for other forest pigeons both in continental areas (e.g. C. fasciata from Western part of North America - Neff 1947; C. palumbus from England - Snow & Snow 1988; C. elphinstonii from India â&#x20AC;&#x201C; Gibbs et al. 2001), and islands (e.g. Ptilinopus insularis from Henderson Island, Polynesia - Brooke & Jones 1995; Hemyphaga novaeseelandiae from Chatham Island, New Zealand - Powlesland et al. 1997; C. bollii and C. junoniae from the Canary Islands - MartĂn et al. 2000).

The large number of plant species consumed by the Madeira laurel Pigeon is also not unusual: in a 5-year study of six species of fruit pigeons in Australia, Innis (1989) found that 89 species of plants were used. The White crowned pigeon C. leucocephala in Florida feeds on fruiting trees and shrubs, of which 35 species have been recorded

Madeira laurel pigeonChapter 8. 6

(Gibbs et al. 2001). The diet of the Torres Strait pigeon Ducula bicolor in mainland Australia rainforest comprises at least 30 different species (Crome 1975b).

Proper tests of niche expansion would require comparisons between closely related mainland and island forms; Columba trocaz is supposed to be a derivative of the largely mainland C. palumbus but this species itself has a wide niche - it feeds mainly on the ground but it is quite capable of reaching fruits and seeds on trees (Gibbs et al. 2001). Such ecological and biological plasticity of the columbiformes in particular, and frugivores in general, confounds or hides the existence of potential niche breadth adaptation of the Madeira laurel pigeon. This issue can only be addressed by a much more extensive and detailed study of habitat use and feeding habits within the genus. Another line of investigation could follow potential physiological modifications of the Madeira laurel pigeon so as to deal with foods with relatively high levels of toxins and/or low level of nutrients.

One last point that should be present during the discussion of the above issues is that food preference in the Madeira laurel pigeon is not determined by resource availability alone. Birds consistently used fruits from the most common trees (e.g. O. foetens), but species with a very modest contribution to the composition of the forest were used quite heavily (e.g. I. canariensis). Furthermore, the number of pigeons foraging in some fruiting tree species (e.g. L. azorica in 1996) increased disproportionately to their contribution to the overall fruit abundance in the forest. This is not surprising and traditionally it has been proposed that choice can be explained by fruit characteristics related to size (fruit length, diameter, and fresh and dry mass) and nutritional composition of its pulp (water, protein and lipid amongst others) (e.g. Snow 1971, Berthold 1976, Stilles 1980, Herrera 1981, 1982, Sorensen 1981, Johnson et al. 1985, Brooke and Jones 1995; Herrera 1998). Within this line of reasoning, Wheelwright (1988) demonstrated that even when fruit availability was held constant, the American robin showed seasonal variation in fruit use, which indicates that physiological needs, and not always fruit availability influences seasonal fruit preferences. It has been proposed by other authors that fruit conspicuousness and individual tree abundance are also important factors (e.g. Snow 1971, Baird 1980, McPherson 1987, Sallabanks 1993) and so pigeons may use food sources strategically to maximise foraging efficiency by minimising search and travel time. Although the data presented here suggest that active Madeira laurel pigeonChapter 8. 7

choice is taking place, it is not yet possible to identify the specific factors influencing that choice.

Relationship between fruit availability and use of crops One of the most important aims of this work was to understand the relationship between habitat use, fruit phenology and the use of crops. Inter â&#x20AC;&#x201C; annual variation in the use of agricultural fields is not explained solely by fluctuations in fruit abundance. It is proposed that the use of the agricultural fields is mostly opportunistic and is related to the changes of dispersion and movement of birds in the forest. Thus, ultimately, fruit phenology will influence the use of agricultural areas only to the extent that it influences such movements. Since no regular movements were identified, the results of the study fail to fully explain the consistency of the temporal patterns of use of agricultural fields; this means that other factors must be involved. One possibility is that more severe weather at higher altitudes may cause pigeons to come to lower areas at certain times, or there may be more subtle altitudinally-based ripening patterns of the fruit trees.

Very little is known about the patterns of fruit production within the Madeiran laurel forest and research on this topic would be crucial to the understanding and prediction of bird movements. In the past there has only been one short â&#x20AC;&#x201C; term (2 year) study carried out (Oliveira and Jones 1995) and, as in the present study, strong fluctuations on fruit abundance were identified between years. Studies elsewhere (Wheelwright 1986 and references therein) have also shown quite unpredictable annual fluctuations of fruit production in the Lauraceae. To fully understand fruiting patterns at the community level, and the relationship between frugivores and the Lauraceae long â&#x20AC;&#x201C; term research should be performed using individually- marked trees. The results would also help in the prediction of the spatio and temporal vulnerability of crops.

Although the study was designed only to explain the relationship between the use of agricultural fields, habitat use and forest phenology, it is possible to speculate upon the behavioural mechanisms which might relate to the pattern of use of the fields.

Firstly, I suggest that experienced birds using the forest in the vicinity of the agricultural areas may be able to recognise when the fields are most suitable. The ability to

Madeira laurel pigeonChapter 8. 8

recognise sources of food through learning and memory has been proposed for other pigeons (e.g. Goodwin 1985); most of the fields here were used immediately after being cleaned and cultivated, activities which may provide proximate signals of food availability.

Secondly, pigeonsâ&#x20AC;&#x2122; social behaviour when feeding may also impact upon crop predation. Pigeons normally fly to fields singly or in pairs and join birds which are already feeding. When disturbed, the birds scatter in small groups but will then recongregate in larger groups on a nearby field if left undisturbed. Perhaps it is the experienced birds which provide the nuclei around which the other birds congregate.

Thirdly, it has also been suggested that the Madeira laurel pigeons tend their feeding within a suitable patch in the forest before moving on to another (for details refer to chapter 4). Such behaviour might be mirrored in the way that pigeons concentrate their feeding within certain fields for long periods and can completely strip all of the crop plants. All these behavioural mechanisms are in accordance with the hypothesis that the use of the fields is opportunistic.

The number of birds and the proportion of the population involved in the use of crops are unknown. Are the cultivated lands used intensively by a small number of pigeons or are they used less intensively by a large proportion of the population? The answer to these questions could help to understand the use of anthropogenic areas and, consequently, the design of appropriate management action. At the beginning of this study, an attempt was made to catch and tag individuals feeding on crops, but this was unsuccessful due to the cost in terms of personnel and time. This is obviously an important gap on our knowledge, and further research on this topic would also be of great benefit. This will be discussed in more detail on the Action Plan update presented below.

Patterns of use of agricultural fields

If the relationships between fruit phenology and use of agricultural fields remain rather unexplained, it was possible, however, to identify the hierarchy of decisions made by Madeira laurel pigeonChapter 8. 9

the pigeons when they feed on agricultural fields. In general (i) they chose those fields nearest to the forest and far from human settlements or infrastructures; (ii) they prefer to feed on open fields and on high standing items; and (iii) they do not discriminate between the centre and the margin of the fields.

These patterns are presumably a reaction to the risk of predation. In most animals predation risk is a determinant factor behind habitat and micro – habitat choice, since it can be minimised or reduced by foraging in less risky places. On the other hand, alternative, but potentially more dangerous habitats can be very advantageous if they provide good food resources with little energy expenditure (Krebs 2001). Thus, the trade – off between risk and the advantages of feeding on alternative habitats must be one of the factors ruling the use of agricultural fields.

It has been widely documented that species that do not encounter predators for many generations lose their ability to respond to them (e.g. Whittaker 1998; Grant 1998). Nevertheless, anti – predator behaviour is not inevitably lost after long periods of isolation (Blumstein et al. 2000). The degree to which anti predator behaviour persists or is lost under relaxed selection is of considerable theoretical interest (Blumstein et al. 2000 and references therein) and it also has practical implications for the conservation and management of geographically isolated populations (Quammen 1996).

It is known that some island bird species have become very tame and are easily approached by man; the classical examples of the Dodo (Quammen 1996) or the Galapagos finches (Grant 19998) can be invoked here. According to historical records, at the time of the colonisation of the Island, like the Dodo, pigeons were fearless of man and thus very easily caught (Zino and Zino 1986). Nowadays the pigeon’s behaviour on the fields can be explained by a fear of man and, from an evolutionary and behavioural point of view, it is interesting to speculates how this change has occurred - how did the Madeira laurel pigeon turn from an apparently fearless into a very secretive and shy bird? There are perhaps two possibilities: artificial selection driven by man and animal learning behaviour. The former, or what can be called the genetic hypothesis, would suggest that the increased probability of being killed by man promoted the disappearance of tamer birds. To support this hypothesis there is the fact that pigeons were heavily persecuted and population numbers were kept artificially low for centuries.

Madeira laurel pigeonChapter 8. 10

The learning hypothesis suggests that throughout their lives pigeons are able to learn to recognise man as a threat and to convey that notion to inexperienced birds. It has been shown that some birds are able to distinguish between different threatening human behaviours (Knight and Temple 1986a). The general ability and advantages of learning under many different circumstances have been discussed in various recent works (e.g. Irwin and Price 1999, Seferta et al. 2001).

In terms of the conservation and management of the species it is of course relatively unimportant to understand how the current-predator behaviours have evolved. What is more important is the understanding of how the use of crops is ruled by the trade off between fear of man and feeding in a profitable habitat. In chapter 5 it was proposed that it is possible to successfully predict which fields are attacked and when, and also to manage the crops so that damage can be reduced. Moreover it has been argued that use of crops is opportunistic and this supports the idea that simple management of the fields could reduce the problem - crops are used opportunistically rather than being an essential food when fruit production is poor so birds could easily be persuaded to go elsewhere.

Management considerations

The findings of this study contribute to the understanding of the relationship between a vulnerable endemic bird species and its relict forest habitat. Conservation of the Madeira laurel pigeon depends on preserving all habitat types required to meet its resource needs throughout the year. Nowadays, with the Madeira laurel forest well protected, habitat degradation and loss is no longer a threat for the species. However, the knowledge that pigeons use many different plants, which occur more abundantly on the edges of the forest, can be used to redefine the conservation value of these areas.

Illegal hunting, poisoning and unpopularity due to the damage it causes to crops, is the present major threat for the species, and the pigeonsâ&#x20AC;&#x2122; activity on the agricultural fields clearly remains the most pertinent issue for its conservation. The information presented here can be crucial to understand the proximate factors which lead to the seasonal use of these areas and provides an ecological framework for the problem. It is this level of Madeira laurel pigeonChapter 8. 11

knowledge that enables us to create a set of predictive criteria and to propose different preventive actions. If the suggested actions are follow then the survival of Madeira laurel pigeon, an ancient species in a relict ecosystem, can be assured for some considerable time.

Specific recommendations for management and proposals for future research will be made on the â&#x20AC;&#x153;Action plan updateâ&#x20AC;? presented in chapter 9 of the present work.

Madeira laurel pigeonChapter 8. 12

9. Management recommendations. Up date of the Madeira laurel pigeon Action Plan.

The aim of this project would not be fully achieved without up dating the existing â&#x20AC;&#x153;Action Plan for the Madeira laurel pigeonâ&#x20AC;? prepared by the author in 1996 and published by the Council of Europe. This is a good opportunity to present specific management recommendations based on the findings of the present work. The whole action plan, including those issues not so far addressed here, will also be reâ&#x20AC;&#x201C;evaluated. This scheme will provide an integrated and consistent management proposal based on a very broad, practical and scientifically sustainable approach.

Aims and objectives

Aims

In 1996 the aims were set at 3 levels: short term (maintain population at no fewer than 3500 individuals); medium term (ensure its continued increase towards occupying all suitable habitat); and long term (to enable the recolonisation of areas of its former range through habitat restoration).

The short-term aim was fully achieved and efforts should be made to maintain numbers much higher, at least at no fewer than 10,000 individuals. In the medium term keep the aim of ensuring the increase towards occupying all suitable habitat areas. Experience has shown that the long term aim proposed in 1996 is rather unpractical and unrealistic. Therefore the medium term aim should expanded into a long term one.

Madeira laurel pigeon Chapter 9.1

Objectives

1. Policy and legislation (as in the 1996 edition) 1.1.

To ensure an adequate legal and financial framework for the conservation

of the Madeira laurel pigeon 1.1.1. To establish the management plan for the Natural Park of Madeira

The objectives proposed in 1996 were not fully achieved. The financial framework was partial accomplished mainly through EC life projects, but such efforts should be maintained. The management plan for the Natural Park of Madeira, where this action plan would be included, is still lacking.

Priority: High Time scale: ongoing

1.1.2. To attract funding from international organisations

This was done successfully in the past but it is still a very pertinent issue

Priority: High Time scale: ongoing

1.1.4. To recognise the global importance of the Natural Park of Madeira

An important step towards achieving this objective was the recognition of the laurel forest of Madeira as a World Heritage Site by UNESCO. Efforts should be made to keep the present international recognition.

Priority: Medium Time scale: Ongoing

Madeira laurel pigeon Chapter 9.2

1.1.5. To upgrade the protection status under the Bern Convention.

Although the Madeira laurel pigeon is listed in Annex I of the EC Wild Birds Directive, it will be important to upgrade their status under the Bern Convention. This would provide additional legal protection for the species and their habitat.

Priority: Low Time scale: long.

2. Species and habitat protection. 2.1.

To reduce human predation.

2.1.1. To prevent hunting and poisoning.

Hunting and poisoning are mostly carried out by farmers as a response to damage caused by the pigeons. Measures, such as an increase in the number of wardens, should be taken to minimise the crop problem and to promote the awareness of the importance of the Madeira laurel pigeon (see below).

Priority: High Time scale: ongoing

2.1.2. To take preventive appropriate measures to minimise crop problem.

It is proposed that managers should use the criteria presented in this work to predict field damage, and act preventively either by providing (i) bird scarers and advice on their use, (ii) exclusion nets, and (iii) advice on field manipulation to minimise their utilisation by the pigeons. In practice they should pay special attention to fields that are in close proximity to the forest and far from any source of human disturbance. Managers should also plan their activities in order to be able to respond very quickly and adequately during the most critical periods.

(i)

Bird scarers devices and methods

Farmers should visit their fields regularly until the end of the fourth week and between the 10th and 16th week after planting. Since the efficiency of the Madeira laurel pigeon Chapter 9.3

scaring devices decreases rapidly with their use, as birds become accustomed to them, farmers should be advised to reduce their use after the fourth week following planting. With a proper management of human resources and scaring devices, it is possible to reduce the costs of this operation. (ii)

Exclusion nets

On specific situations like fields very far from human settlements or whose topography decreases the effectiveness of scaring devices the use of exclusion nets should be favoured. (iii)

Manipulation of the fields

Concurrently to the measures already mentioned, manipulation of the fields could result on a very effective and reasonable decrease of the damage inflicted on crops, especially on cabbages. To reduce the effects of pigeonsâ&#x20AC;&#x2122; activity during the first planting season, from late February to late March, farmers should only plant the cabbages after the potatoes have emerged (10 to 15 days later). This simple measure would prevent the use of fields by the birds during the first critical period without a relevant increase of costs.

To reduce damage in the second critical period, farmers should plan their planting in such a way that cabbages would never be left alone in the fields. Prior to harvesting the potatoes, for example, corn could be planted to act has a barrier between the cabbages and the pigeons.

Probably these measures would not end the damage inflicted on crops but they would definitely reduce it down to an acceptable level. Further efforts could be made on this area; a few experimental fields should be chosen to test other methods such as the use of repellents (refer to â&#x20AC;&#x153;monitoring and researchâ&#x20AC;?).

Priority: Very high Time scale: ongoing

2.1.3. To prepare a contingency reactionary plan.

An evaluation of the cost of crop damage should be undertaken yearly and a compensation plan prepared for emergency cases. Compensation should take the form Madeira laurel pigeon Chapter 9.4

of payments in kind (e.g. seed and fertilisers for the next season). However, these measures should only be implemented if the preventive measures failed.

Priority: Medium Time scale: long

2.2.

To enforce current habitat protection

2.2.1. To increase the protection status of laurel forest areas.

The whole laurel forest is included in the Natural Park of Madeira. Nevertheless, at least half of the forest area is still classified as Partial Nature Reserve only. This protection status should be increased and all dense, high canopy laurel forest below 1000 metres, should receive the maximum status of protection; this would apply to Ribeira da Janela, Ribeira Grande, Ribeira do Inferno and FajĂŁ da Nogueira.

Priority: high Time scale: ongoing

2.2.2. To increase the protection status of the forest boundaries.

It was shown that forest openings, namely those occurring on their lower boundaries, are very important habitats during spring and summer. Some of these areas should be included within the Natural Park of Madeira. Wardening should be implemented during spring and summer since birdsâ&#x20AC;&#x2122; vulnerability to man increases dramatically when they are using that habitat.

Priority: medium Time scale: short

Note: It is considered that other topics referred to in the original Action Plan, as (i) to improve breeding success, (ii) to prevent disturbance from tourism, (iii) to encourage the spread of the Madeira laurel pigeon into suitable habitat and (iv) to prevent fire damage to regenerating laurel forest, are of lesser importance and relatively irrelevant Madeira laurel pigeon Chapter 9.5

when compared to the issues addressed here. Their reference in the “Action plan” could confound and obscure the relevance of these much more important issues under scrutiny here.

3. Monitoring and research

3.1. To obtain regular information on numbers, range and trends of the population.

The monitoring scheme implemented in 1986 and repeated during the present work should be performed on a regular basis (e.g. every 3 years). Additional information could be provided by yearly partial census (e.g. only transects of one of the study areas). The use of distance sampling, following the methods proposed in this work should be favoured when attempting to obtain absolute densities of birds. This monitoring scheme is most valuable to assess the ongoing management measures and to decide upon the need to implement additional measures.

Priority: High Time scale: Ongoing

3.2. Monitor reproductive parameters

Little is known about the breeding biology and breeding performance of the Madeira laurel pigeon. Efforts should concentrate on breeding success, chronology, seasonality and number of broods per year. The effect of rat predation on breeding success needs to be clarified. Recent work with dummy eggs (author’s unpublished data) were inconclusive because they failed to identify the causes of “failure” for the majority of the eggs missing.

Priority: Medium Time scale: short

Madeira laurel pigeon Chapter 9.6

3.3. To test the use of repellents to protect crops.

The best approach to pigeon damage problems is to use several complementary techniques together; the relative success of each method depends on field location and crop type, amongst many other factors. Repellents have been widely used with success with many other species. Experiments should be performed to test whether it is a feasible method to be used in Madeira; it is important to test to which repellent the Madeira laurel pigeon reacts and to make a cost assessment for the implementation of such technique. The Natural Park of Madeira is seeking funds to carry out this project.

Priority: Medium Time scale: ongoing.

3.4. To provide a better understanding of the use of cultivated areas.

During the present work important steps towards the explanation of the cause/ effects of the use of cultivated lands by the Madeira laurel pigeon were given. However, there are important gaps in our knowledge on this issue. The next approach should try to understand (i) the number of birds and/or proportion of the population that actually uses crops, and (ii) what type of individuals (e.g. specialists, juveniles, weaker birds etc) are involved.

This level of knowledge would be of the utmost importance both to

understand the factors behind that use and to design better management plans.

4. Public awareness and training

4.1.

To increase public awareness of the Madeira laurel pigeon

The most serious threat for the pigeon is unpopularity amongst the farmers and the rural population in the north of the island. Specific campaigns should be prepared with the objective of (i) explaining the importance and special value of the bird and (ii) providing information about the techniques available to prevent damage to crops.

Priority: High Time scale: ongoing Madeira laurel pigeon Chapter 9.7