Return to index of bird studies

Return to index of 2000 research reports

The following paper will aim to assess the abundance of 8 different bird species found on Buton Island just off the South East Coast of Sulawesi in Indonesia.

The way in which the birds use different types of habitat will be looked at to see if there is any correlation between the type of habitat and the abundance of the different species.

The birds that are to be studied are from three families, the Flowerpeckers- Dicaediae, Sunbirds- Nectarinidae and the White-eyes- Zosteropidae. There are four species of sunbirds: Olive-backed, Brown-throated, Black and Crimson, two species of flowerpeckers: Yellow-sided and Grey-sided and two species of white-eye, the Pale-bellied and the Lemon-bellied.

This project was chosen for a number of different reasons but the main reason is that the birds of Buton have not been studied in any great detail. There have been previous studies undertaken on the birds of Buton Island by Operation Wallacea but the abundance and habitat usage of the birds proposed in this study have not been investigated in any great depth and so the research would be original.

Another important reason is that three of the study species are endemic to Sulawesi and therefore they are particularly vulnerable to threats such as changes in habitat, land use patterns and from predation. The fact that they are also island birds increases their vulnerability and so if their abundance and habitat requirements were known then it would be easier to implement conservation measures to help conserve them.

The specific aims of the investigation are to estimate the abundance of each of the study species in the area of the study site and from this data to work out which species is the most locally abundant.

Density estimates will be calculated for each species in the different habitat types that the species occurs in and so the habitat preferences of each species can be established.

Overall densities of the species will be calculated to see if there is a correlation between habitat usage and species type on a broader scale.

The study site is on the island of Buton, which is off the South East Coast of Sulawesi in Indonesia, South East Asia. The island of Sulawesi is part of what is known as the area of Wallacea, which refers to the naturalist/explorer Alfred Russel Wallace and his discovery of the Wallacea line in 1850 (Wallace, 1859).

Wallace stumbled across the Wallace line during his travels around the islands of Indonesia where he was collecting specimens to ship back to England. During his travels he noticed that on travelling from Java to Bali the animals and birds were very similar, however this was not the case across the narrow strait to the east on the Island of Lombok, which contained a very different set of species. He found a similar trend between Borneo and Sulawesi with the species present on Borneo being more like the ones found on Java much further to the West.

Wallace suggested that these differences were due to deep-sea trenches that separated Bali and Lombok and Borneo and Sulawesi and that this factor had isolated the islands to such a degree that different species had evolved there.

The Wallace line is the name given to this hypothetical line that follows the deep-water trenches separating the distinctly different regions of biodiversity. (Http://www.opwall.com)

The exact position of the line can be better illustrated on the map below taken from Oosterzee, (1997).

The area of Wallacea has been recognised as important on a global scale due to the unique composition of species that can be found there.

Stattersfield, (1998) identifies the area as a centre of biodiversity, this is due to the high level of endemicity shown by its flora and fauna.

The Island of Sulawesi itself is at the same time the poorest in terms of number of species and the most isolated in terms of the species it contains. Sulawesi has a surface area double that of Java but with half as many species of land birds and mammals (Oosterzee, 1997). The species that Sulawesi does contain are either very closely related to the islands surrounding it or are totally unrelated.

When Wallace first found this the only way that he could explain such an anomaly was to suggest an origin in remote antiquity that provided isolation long enough for evolution to have formed what he described as:

"The most singular island in the world" (Wallace, 1859).

The difference in species between the two sides of the Wallace line is not limited to a few groups and indeed there is evidence from a vast range of different species that supports this ‘biogeographic jump’ between the two sides of the line. One of the first birds that Wallace noticed on the island of Lombok was the Orange-footed Megapode Megapodius reinwardt, which is a member of the Megapodiidae family of which there are 22 members. However no megapodes have been found on the islands of Bali, Java or indeed the greater part of Borneo all of which lie to the west of the Wallace line (Wallace, 1859).

In Borneo Wallace found that the forests were full of monkeys, wild-cats, civets, otters and squirrels but across the narrow Makasser strait in Sulawesi there were few of any of these species and they had instead been replaced with sloth like cuscuses.

The ornithologist G.A. Lincoln studied birds in Java, Bali, Lombok and Sumbawa on either side of the line in the 1970’s. He found that the birds on Java and Bali were predominantly oriental and were dominated by the Oriental Bulbul. On the other side of the line in Lombok the birds were predominantly Australian with the honeyeaters attracting the most attention. The most similar bird faunas were found on the islands of Bali and Java while the least similar bird faunas were found on the adjacent islands of Bali and Lombok, which lie astride the Wallace line (Oosterzee, 1997).

Similar patterns can be found if mammals are considered. R.C. Raven a zoologist sailed back and forth across the Makasser strait between 1912 and 1923 collecting specimens and he noticed that a number of mammalian groups had succeeded in crossing the line but a much larger number had been stopped by it. For example 196 species of squirrel were identified but only 14 of those were found across the line in Sulawesi. Similar reports were found for shrews with only 13 species out of a total of 59 having crossed over the line. Indeed none of the 16 species of the family Tupaiidae, a type of tree shrew were found to have crossed the line even though their range extends right up to the boundary. (Oosterzee, 1997)

The scientist Charles Lyle who was one of the most influential of his day emphasised the need for some form of large natural barrier to separate such distinct regions of biodiversity. This was certainly not expected to be a channel some 25km wide. Also it would not be expected that an archipelago of islands that were very similar physically to be home to such different animals on its extremities let alone across a narrow channel.

The most up to date work on the birds that can be found in the area of Wallacea gives a total number of species of 697 up to the 31^st July 1996 (Coates, 1997). Out of that figure 249 are endemic with virtually every island supporting at least one endemic bird species. The Sulawesi sub-region supports the richest and most highly distinct and endemic avifauna, which includes 16 endemic genera. The Sulawesi mainland supports 224 resident bird species including both land and freshwater species. 41 of these species are endemic to Sulawesi and a further 56 are confined to Sulawesi and/or its satellite islands. The large number of endemics that Sulawesi holds can be favourably compared to its much larger neighbour of Borneo, which has 37 endemic species confined to 5 genres. Further the avifauna of Sulawesi shows a higher degree of endemicity than any other similar land area on earth (Coates, 1997).

Out of the study species three are endemic to the Sulawesi sub-region, these are the Pale-bellied White-eye and both species of Flowerpecker, the Grey-sided and the Yellow-sided.

All species can be under threat to a certain extent from such factors as changes in land use but due to the limited range of endemic species then the threats are all the more profound.

Generally the known inferred change of the extent or the condition of the habitat is one way of measuring the degree of threat that a certain species is under. However trends are impossible to measure unless some baseline data has been previously obtained (Bibby, 1998). For the vast majority of bird species in particular threatened bird species the most basic set of baseline data is not available. The gaining of this data will help to ensure that the more threatened bird species are recognised as exactly that, which will aid in planning for their conservation.

In this particular case the species in question are island species which means they are even more vulnerable.

This can be illustrated as even though only a small percentage of the worlds birds species occur on oceanic islands 93% of the species that have become extinct since 1600 have been island forms (Moors, 1986).

Today island forms account for just over half of the worlds endangered birds. The same factors that occur on land masses also occur on islands but due to their decreased area their affects are more conspicuous. One of the more major affects is deforestation and the clearance of habitat as with a smaller area of available habitat the number of species that can be supported decreases.

Historically the single most significant cause of extinction on islands is from predation from predator’s alien to the islands that have been introduced by man.

On Buton Island the major threat to the bird species is not from predators but from the loss of habitat through deforestation. As yet this is not happening at a massively commercial scale with the majority of timber being taken and used by local people in construction and as a source of fuel. There is however the possibility that this rate of timber consumption could accelerate and then there would be a direct threat to all of the species present.

The causes of endangerment of island birds are not always the same as the causes of extinction. Habitat destruction or its deterioration is the most important cause of endangerment for island birds being the direct cause of 58% of all cases (Moors, 1986). However the same factor has only been the main cause of 19% of extinction’s.

At one extreme of habitat destruction is the complete destruction of forest and at the other end is the most common scenario where the habitat undergoes gradual deterioration. This is the usual situation faced by island birds and can come about via a variety of different sources. Deterioration can be in the form of the encroachment of introduced species of plants into the native habitat or the repression of key species that the bird need in order to survive. Plant suppression often takes place due to the introduction of herbivores such as rabbits. This has been a problem on smaller islands such as Layson Island where rabbits were introduced in 1923. They proceeded to consume the vast majority of the plant life on the island, which led to the extinction of 2 endemic bird species from the island (Ely, 1973).

Therefore endemic island birds are in particular need of conservation priorities in order to maintain their population levels due to the number of factors that can affect them with their limited range making them all the more vulnerable.

Ever since the discovery of agriculture mankind has been destroying the forests of the world. This began with the temperate forests of North America during the seventeenth and nineteenth centuries before this assault expanded to the forests of Europe (Whitmore, 1992).

In the present day with the vast majority of the temperate forests having being destroyed the tropical rain forests are under increasing threats from developers and timber merchants. The people of the lesser-developed countries of the world, which are home to such forests, look over towards the developed world and see the destruction and consequential economic gain that such destruction of the natural world can bring.

So tropical forests are being cleared at an ever-increasing rate to make way for developments, housing and agriculture and also to export the wood for direct economic gain.

The deforestation of tropical lands can have both local and global consequences. On a local scale the hydrological balance of the area, the structure and chemical composition of the soil and local climates can all be affected. On a global scale weather patterns and the chemical composition of the atmosphere could be adversely affected by altering albedo and regional atmospheric water balance.

One of the greatest concerns over the loss of tropical forests is that this is causing a massive loss in the biodiversity that the forests contain. This is of particular concern as our knowledge of the species contained within tropical forests is so poor that the dimensions of this problem are only just being realised (Myers, 1983).

South East Asia is not immune to the processes of deforestation with much of its forests having either been already destroyed or under direct threat of being destroyed. The estimated annual rates of clearance of the forests vary to some extent but the overall conclusion is the same that they are disappearing at an alarming rate. In 1988 the FAO estimated the annual loss was as much as 0.5% of the total original cover. Current figures are unspecific but if the trend of increasing rates continued then this figure will be a significantly larger one.

The Pacific is home to the last remaining ancient tropical forests in the Asia-Pacific region. Sulawesi and Buton have not yet experienced any massive forest clearance but if one looks at a similar example such as Papua New Guinea then the affects can still be illustrated.

Papua New Guinea is covered by the third largest tropical forest on earth and as such is a region of unparalleled biological and cultural diversity. The ancient forests and the people, whose livelihood and culture depend on them, are being threatened by multinational logging companies racing to cut down the forests. In the last five years logging by mainly Malaysian and Korean companies has escalated to over three times the estimated level of sustainability (Http://www.greenpeace.org).

In less than ten years, all the productive lowland forests will have been destroyed leaving the people who lived there with nothing but a degraded environment. Obviously it is not just the people who will suffer with many different species of plants and animals being lost along with the forest.

Actual species are a common value to biologists and ecologists but they are of little use as targets for conservation. A much more practical unit for conservation is that of the protected area.

It is a well-known phenomenon that particular parts of the world are home to concentrations of animal and plant species that are found nowhere else in the world. Consequently these species are more vulnerable particularly if they are not found in great numbers.

Endemic Bird Area is the term given to such pockets of concentration by Birdlife International’s biodiversity project, which began in 1987 with the aim of identifying key areas for biodiversity conservation using restricted range species as indicators.

An EBA can be defined as "an area which encompasses the overlapping breeding ranges of restricted range bird species, such that the complete range of two or more restricted range bird species are entirely included within the boundary of the EBA" (Stattersfield. 1998).

Endemic bird areas are of global conservation value as they are where the greatest concentrations of birds with a restricted range are found. They are also important for avian diversity as a whole as the areas that EBA’s cover at least to some extent encompasses the range of many other bird species that are more widespread and include valuable areas of habitat.

In Wallacea a number of EBA’s have been identified with Sulawesi itself being also identified as a biodiversity hotspot as stated in Letters to nature April 2000 in Cincotta et al’s paper, "Human population and biodiversity hotspots". There are 2 EBA’s on Sulawesi as the island supports some 57 restricted range species. Sulawesi is therefore a globally important site for the conservation of biodiversity and the conservation within such a site is crucial to maintain avian diversity. This is of particular importance for the more threatened bird species such as the endemic species that this study encompasses.

In carrying out research on species that have not been studied before in any great detail then we gain a better understanding of the biodiversity of the planet in which we live. Biodiversity is important for a number of reasons and on a number of different levels.

One of the reasons biodiversity is important particularly in the present age is as a source of income. The actual value of biodiversity is immeasurable but what is certain is that it represents inconceivable wealth and every country in the world reaps some form of economic gain from the exploitation of its biodiversity (Wilson, 1993).

There is also the ethical value of biodiversity to the human species, which varies to some extent with race and beliefs but it seems necessary for our intellectual and spiritual well being. It is therefore important to identify centres of biodiversity so that conservation efforts can be concentrated upon them and the biodiversity of the planet can be maintained.

Birds make up only a very small percentage of the total number of species on earth yet they are important because so much information has been collected on them and they are an ever-popular study species.

Birds have always been used to some extent in the assessment of environmental conditions. In 1962 while researching the effects of DDT and pesticides on the environment Rachel Carson author of Silent Spring also noted the corresponding decline in the numbers of songbirds which could be directly linked to the use of DDT in agriculture. From this relatively early beginning it is now accepted that certain groups of organisms can be used as indicators.

A conference recently organised by Birdlife International (BI) in Malaysia confirmed the importance of birds as environmental indicators. The chairman of BI Dr. Jerry Bertrand warned that,

"Birds are highly sensitive to change, so they give us early warning signs of future environmental crises. Importantly, our findings are showing up a wider problem beyond birds, because places that are rich in birds are generally also rich in other forms of biodiversity" (http://www.swallowsontheline.co.uk

The previous statement presumes that there is congruence amongst taxa in respect that if an area is important for one type of organism then it is also important for others.

There is currently a great deal of debate on this subject but there is certainly evidence to support this theory.

In Myers (2000) biodiversity hotspots for conservation priorities were identified. They were areas where exceptional concentrations of endemic species could be found that were undergoing exceptional loss of habitat.

In this study it was found that 44% of all species of vascular plants and 35% of all species in 4 invertebrate groups are confined to 25 hotspots that make up only 1.4% of the total land surface of the earth. In a number of hotspots there was a high number of endemic plant species found and also a high number of endemic vertebrates but on the other hand there was an equal number of hotspots that had a high number of plant endemics but relatively few vertebrate endemics.

A hotspot study was carried out in the U.K using three groups of species: birds, dragonflies, liverworts and aquatic plants and then the relative diversity of each group was compared across their respective hotspots (Quinn, 2000).

The results of the survey found that there was little to support the idea that areas species rich for well-known groups would also be species rich for less well-known groups. This was tested using birds and dragonflies as the chief examples being the "flagship" species. It was found that only 12% of the hotspots for the two species coincided and when this was carried out for all the groups the highest percentage of coincidence was only 34% between dragonflies and butterflies. However if one taxon was considered and all of the hotspots for that taxon were protected then a large number of species from the other groups would also be represented in them. For example if birds were taken then 100% of butterfly species and 90% of dragonflies, liverworts and aquatic plants occur in the bird hotspots (Quinn, 2000).

There is therefore evidence to support the theory that birds can be used as indicators of the overall biodiversity of an area.

However the evidence is to a certain extent inconclusive but there are numerous examples of where bird populations have been used as bio-indicators and as indicators of environmental change.

In Canada changes in the populations of forest birds are being used as indicators of the biodiversity of the forest as their numbers have been directly linked to the change in forest habitat throughout Canada (http://www.ec.gc.ca).

In Europe in the last 30 years the numbers of what are considered common countryside birds such as sky larks and swallows have decreased by as much as 50%. This can be accounted for by the associated changes in agriculture and land use, which has led to massive reductions in the area of suitable habitat type for many species. As discussed by Wijesekar (2001) birds are closely related to habitat conditions with each species rarely occurring in only one type of habitat. Birds are completely dependent on their habitat for cover, structural and floristic diversity and as a food resource.

This relationship was investigated in the Wet Zone of Sri Lanka and it was found that there was a high variation in the diversity of avian communities among different land use systems. This demonstrates that a change in vegetation structure has a strong influence on the structure of bird communities and so its bird population can monitor a change in habitat structure (Wijesekar, 2001).

The family Nectariniidae is a group of small insectivorous birds that are often brightly coloured and so are often associated with hummingbirds (Trochilidae) and honeyeaters (Meliphagidae). The group is a tropical and subtropical Old World family that can be found from Africa to Australia.

The centre of diversity for the family is in the Oriental region and Africa comprising 123 species in 5 genera (Austin, 1965).

In the area of Wallacea there are 7 species 3 of which are endemic in 3 genera but only the 4 species listed as study species are found on Sulawesi (Coates, 1997).

The males of the species are usually striking compared to the often drab looking females and juveniles and care is required in identification to differentiate between the two. They are very active birds constantly on the move darting between trees and flowering shrubs and plants where they feed on nectar and forage for invertebrates. They have long slender bills that are decurved with the edges of the mandible being serrated and a tubular tongue enables them to feed on both invertebrates and nectar. Similar to hummingbirds, sunbirds often hover at flowers to feed, which interrupts their normal darting flight, which is usually over only short distances.

There are 4 species of Sunbirds on the list of study species which are the Olive backed Sunbird Nectarinia jugularis, Brown throated Sunbird Anthreptes malacenis, Black Sunbird Nectarinia aspasia and the Crimson Sunbird Aethopyga siparaja.

The flowerpeckers are small handsome birds that are confined to the Oriental Region, New Guinea, Australia and North Melanesia. They form a well-defined group that contains 44 species in two genera (Austin 1965).

Their centre of biodiversity is in Sundaland but the family has colonised well and this can be seen in Wallacea where 12 species can be found. Out of these 12, 7 of them are endemic to the Wallacea region (Coates 1997). 3 species of flowerpeckers can be found on Sulawesi, as well as the two already mentioned there is also the Crimson-crowned Flowerpecker Dicaeum maugei. However this species has only been recorded on a handful of occasions on Buton due to its preference of higher elevations and so was not added to the list of study species.

Flowerpeckers are neat highly active birds with small compact bodies and short, pointed slightly decurved conical bills. In general most species are sexually dimorphic with the males exhibiting bright patches of plumage with the females being drab in comparison and often hard to identify. They have a very fast erratic flight, which combined with their small size makes them notoriously hard to identify. They do however have a very distinctive series of calls, which are very loud for such a small bird, and so they are a species that are often heard before they are observed.

Flowerpeckers generally feed on small berries, arthropods and nectar with some species being more specialists in their feeding habits feeding on the fruits of mistletoe.

There are 2 species of Flowerpecker on the list of study species, which are the Yellow-sided Flowerpecker Dicaeum aureolimbatum and the Grey-sided Flowerpecker Dicaeum celebicum.

The family Zosteropidae comprises 96 species in total in 13 genera that are widely distributed mainly in the tropics and subtropics from Africa to Japan, Australia and the islands of the Western pacific (Austin 1965).

The vast majority belong to the single genus Zosterops and it is to this genus that both of the species found on Buton belong. In total Wallacea hosts 22 species of Zosteropidae 16 of which are endemic and they are made up from 5 genera, 5 of which are only found in Wallacea (Coates 1997).

Before DNA studies were carried out by Sibley and Ahlquist it was thought that they were closely related to Sunbirds and flowerpeckers but it is now understood that they are more closely related to the warblers of the old world, Slyviidae.

The genera Zosterops are a uniform group, which are characterised by a white ring around the eye but this is not always present and is sometimes coloured black. In general they have greenish plumage with whitish, yellowish or greyish underparts. They are small birds and have a slim, slender bill encasing a brush tipped tongue, which is an adaptation for nectar feeding although they are mainly insectivorous. The sexes are alike and even immature birds closely resemble adults in their appearance. They are gregarious birds often foraging as part of mixed species feeding flocks and so are often associated with Sunbirds and Flowerpeckers.

There are two species of Zosterops that are included on the list of species to be studied. These are the Pale-bellied White-eye Zosterops citrinellus, which is also referred to as the Sulawesi White-eye as it is endemic to Sulawesi and the Lemon-bellied White-eye Zosterops chloris.

The Olive-backed Sunbird is a widely distributed species ranging from the Andaman and Nicobar Islands of India through Burma, SE China, Thailand, Cambodia, Malaysia, the Philippines, the Greater and lesser Sundas to New Guinea and the Northern tip of Australia (King, 1975).

There are 9 subspecies of N. Jugularis that are grouped into 2 forms the yellow-bellied and the black bellied, it is the yellow bellied sub-species that is found on Buton Island (Coates, 1997).

Of all of the species of Sunbird found on Sulawesi it is the Olive-backed that is the most widespread and abundant. It can be found in a wide variety of habitats generally near to the coast and in lowland areas where the suitable habitat can be found. The habitat that the birds frequent include forest edge, secondary forest, coconut plantations, areas of agriculture, mangroves and beach scrub. It is especially common in coconut plantations being less abundant but still common in lowland areas of scrub and degraded forest.

The species often comes into contact with the Brown-throated Sunbird in coastal areas and Black Sunbirds in forest edge environments. On Sulawesi it has been recorded at elevations of up to 800 metres, as Buton does not exceed this figure then the birds should be very widespread in the appropriate habitat (Watling, 1983).

The species often occur as pairs where they are very active and vocal foraging for nectar and invertebrates. The usual call is a wispy seeping series of 2 notes with a song consisting of a mixed jumble of twittering notes. They are very small birds at 11cm and the sexes are very distinctly different. The male has olive underparts with variable yellow facial stripes, a yellow abdomen and a dark purple throat with a yellow belly (Sub species found on Buton). The female is similar to the male but it lacks the distinct brightly coloured throat of the male, this characteristic can also be observed in immature birds.

The Brown-throated Sunbird or Plain-throated Sunbird ranges from Southern Burma through Indochina and the Malaysian Peninsular to the Philippines and the Greater and lesser Sundas in Indonesia (Coates 1997).

Brown throated Sunbirds are common on Buton Island and also on the nearby islands of Muna and Kabaena (Watling 1983).

The species seems to be restricted to the lowlands and does not appear in hill forest or higher ecosystems. It is especially common in areas of mangrove and adjacent coconut plantations, areas of scrub and cultivation. It is also found in areas of forest edge and secondary forest. The birds are regularly seen low down in the understorey frequenting bushes and scrub but they keep mostly to the outer foliage of the mid canopy and the upper canopy of trees.

The species can be found at elevations of up to 1000 metres on the mainland and therefore as can be said for the other species previously mentioned where the suitable habitat exists there should be no limits to where the birds can be found on Buton.

Found singly or in pairs the birds are highly pugnacious not tolerating the presence of other sunbirds in feeding trees or bushes.

The song is usually delivered from a prominent perch being a series of similar notes, swit-swit-sweet or sweet-sweet. The calls can be divided up into three distinctive variations (i) hard swit (ii) a drawn out, thin, high-pitched siiewei and (iii) a rapid nasal chatter of medium pitch, titititi (Watling, 1983).

They are small birds around 13.5 cm in length with a relatively short bill in comparison to other species. The male has a dull brownish throat with iridescent underparts and brownish wings with yellow underparts. The female resembles the female Olive-backed sunbird at first glance with olive above and pale yellow below but on closer inspection it is very mush paler than its olive-backed counterpart.

The Black Sunbird is much more limited in its range than the other species being restricted to New Guinea and the Mollucan Islands of Eastern Indonesia (Coates, 1997).

The species is widely distributed but relatively uncommon being found in a wide number of different habitats but favouring secondary forest and forest edge environments. It can also be found though in lesser numbers in areas of cultivation, gardens and mangrove as well as scrubland. In forested areas the birds rarely venture into the interior of the forest usually keeping to the outer edges of foliage in the canopy and mid storey. In more open habitats the birds regularly frequent the outer edges of shrubs and bushes.

The birds can be found at elevations of up to 800 metres on central Sulawesi, though on other islands such as the Kai Islands it is limited to 200 metres above sea level.

The black sunbird is a very active species often hover-gleaning for nectar and arthropods and it competes pugnaciuously with Olive-backed Sunbirds for feeding sites.

The call is usually a series of shrill notes given singly or in a series, on Sulawesi the usual call is a clear hollow peep with an occasional rapid zi-zi-zi-zi-zi (Watling 1983).

They are very small birds being no more than 11.5cm in length making them the second smallest species found on Buton after the Olive-backed Sunbird.

Males are black with iridescent patches with the female being drab in comparison with grey head, pale grey throat and dull yellow underparts. Immature birds resemble the adult female with young male birds developing a black moustache like streak.

The crimson Sunbird has an extensive range from Nepal through North India, South China, Indochina, Malaysian peninsula, the greater Sundas, western Philippines and Sulawesi. Two forms are found on Sulawesi A. s. flavostriata, which is found in the North of Sulawesi and A. s. beccarii that can be found to the South (Coates, 1997).

Overall the species is rare to uncommon found throughout Buton in forested areas. It is regularly found in primary and secondary forest as well as forest edges and clearings in forested areas. It is very elusive often foraging in the upper canopy where it proves to be inconspicuous. Although most common in forested areas it is not restricted to them also being found in coconut plantations, degraded forest and areas of overgrown cultivation. It is also seen occasionally in more open scrubby habitat, but this habitat does not seem to be favoured.

Crimson Sunbirds do not seem to associate with any other species but they can be seen with Black Sunbirds along the forest edge and males are very aggressive towards each other. The species generally occur singly, in pairs or in small family groups foraging for nectar and small arthropods.

Males usually sing from exposed branches high up in the canopy in a seepy series of notes, siesiep-siepsiep. They are small birds around 12cm in length. The male crimson Sunbirds have red foreparts with a metallic blue forecrown a yellow rump and dusky belly. The female is olive-green with yellowish below and washed red tail and wings.

The Yellow-sided Flowerpecker is endemic to Sulawesi and a few neighbouring islands where two subspecies can be found D. laterale and D. aureolimbatum (Coates, 1997).

Throughout Sulawesi the species is widespread and common inhabiting a wide range of habitats including primary and secondary forest, forest edge, scrub, plantations and lightly wooded cultivation. On Buton it is also common in overgrown plantations and in village gardens but it appears to be most common in forested habitats (Watling, 1983). It can be found singly or in small foraging groups around fruiting bushes and trees. The birds usually frequent the mid-storey of trees and flowering bushes often along the edge of forests and areas of plantation. It is not usually found away from areas of forest but birds can be seen in open areas in order to reach feeding trees. They are also aggressive birds particularly towards other species of flowerpecker around feeding sites.

The call of the Yellow-sided Flowerpecker is a disyllabic, high-pitched s-uit less high pitched than the similar call of the Grey-sided species. The birds are smaller than either the white-eyes or sunbirds at about 8.5cm in length.

Both of the sexes of the Yellow-sided Flowerpecker are similar in appearance unlike the species already mentioned. They are dark yellowish-olive above with blackish ear coverts and deep yellow sides above a white belly.

The Grey-sided Flowerpecker is endemic to the Sulawesi subregion and the Sula Islands where five subspecies are recognised. The subspecies found on Buton is the D. c. celebicium, which is widespread and generally common (Coates 1997).

On the mainland of Sulawesi it inhabits primary and tall secondary forest and forest edge as well as lightly wooded cultivation, gardens and even urban environments. On Buton the species is commonly seen in degraded forest and areas of scrub, shrubs and scattered trees. Its particular favourite fruiting tree is the wild plum and one may often find large flocks of up to 15 birds feeding in such trees. Like the Yellow-sided, the Grey-sided males are very aggressive around fruiting bushes where they will chase each other vigorously. Found singly and in pairs often foraging in the outer canopy and outer foliage gleaning arthropods from bare twigs and branches.

The calls of the species include a thin high-pitched seeei repeated at 0.5 to 5.0second intervals, a short repeated tjjt, a single high-pitched tsip and probably the most characteristic of the species a harsh chacking sound often given as a warning when a predator or observer is approaching.

The birds are small at 9cm with the sexes being easily discernible. Males have a purple-black above, with a red throat and breast and white underside. Females have an olive-grey above and are pale below looking quite similar to immature yellow-sided individuals.

The Pale-bellied White-eye is the only Sulawesi endemic to be confined to the SE peninsula of Sulawesi and so it has a tiny global distribution.

Where it is found however it is common, largely confined to forested areas and also found in thick scrubby vegetation that is in close vicinity to more heavily wooded areas. It has been recorded at elevations of up to 300metres.

Found in pairs or small family parties it is often observed in the canopy of tall trees and in the understorey along the forest edge.

It is similar in size to the Yellow-bellied White-eye being between 11.5 and 12cm in length. Its plumage is very distinctive with a golden yellow throat and under tail-coverts contrasting with a very pale off white belly.

Its song is a melodious, characteristically loud warble that is very distinctive which should aid in its identification.

The lemon-bellied white-eye has 5 subspecies, all of which are confined to the Sulawesi sub-region, Mollucas and western Lesser Sundas. It is generally common being locally abundant in certain places inhabiting a range of habitats that include secondary forest, open woodland, scrub, mangroves, cultivation, village and urban gardens.

A survey conducted by Operation Wallacea in 1997 also recorded the bird as being common in non-forested habitats.

In Southern Sulawesi it can be found up to an elevation of 1800metres. This would not be the case on Buton as elevations do not exceed 500 metres but it means that it terms of elevation the species could be found throughout Buton (Coates 1997).

The birds are often found in small flocks foraging at all levels of vegetation in a restless manner and their call can be described as a sparrow like chirp. Its song is a jumbled mixture of short, rapid, high-pitched notes that are given in short bursts of between 1.5 and 3.5 seconds duration (King, 1975). It is a small bird of 11-12cm in length with all yellow underparts and has a characteristic white ring around its eye.

The Sulawesi Sub-region comprises the main island of Sulawesi and its satellite islands from Talaud in the North to Muna, Buton and Tukangbesi in the South. Sulawesi is the largest and most geologically complex island in Wallacea located nearly at the centre of the Indonesian archipelago it covers a total area of some 186,145 square kilometres. It has a very unusual shape, often referred to as a drunken spider, which can be seen on the map given overleaf. It has an exceptionally long coastline of 4,950 Km and consequently there is no site on Sulawesi that is more than 100 km from the coast.

Sulawesi was formed as a result of numerous collisions between two of the earth’s main crustal plates and this accounts for its geological and physical uniqueness. Sulawesi’s lithological and climatic variations are reflected in its rich and varied mosaic of plant and animal communities.

(http://www.geocities.com/TheTropics/Cabana/4600/geography.html)

(http://users.powernet.co.uk/mkmarina/sulawesi.html)

Buton is an island measuring some 150km long by between 10 and 30 km across situated just of the south-east peninsula of Sulawesi, in the Flores Sea. At its highest point Buton reaches just over 1100 metres in elevation.

Buton Island has been divided up into a number of different land use categories by the Forestry Department of Indonesia, which include Conservation areas, Production forest areas, Cultivation/settlement areas, and Settlement with mixed cultivation.

The divisions were made to take into account local environmental conditions and potential economic activities that these areas can support. Only certain economic activities are allowed in different land use categories, but this is often difficult or impossible to implement on the ground.

(http://users.powernet.co.uk/mkmarina/sulawesi.html)

The actual area of study was around a village by the name of Labundobundo situated near the centre of Buton (see above map).

The village is at least partially contained within a conservation area, the Kakanauwe Forest Reserve, which means that no activities other than land improvements are allowed and there is no provision for traditional land use of the area by the indigenous population.

Any form of timber extraction is therefore prohibited but this does not however prevent the locals from removing timber for their own personal use and indeed during the course of the study this was observed on a number of occasions. The chilling sound of chainsaws was also never too far away.

The area encompasses a number of different habitat types with the vast majority being traditionally composed of lowland rainforest, which used to be the primary vegetation type found throughout the Islands of Indonesia (Mabberley, 1992).

In recent times however due to foreign influences and in the pursuit of economic gain a new habitat type has sprung up in ever increasing numbers. This is plantation, usually in the form of Coconut but there are also large areas of Banana plantation and Cocoa.

The primary lowland rainforest has been reduced to poorer quality secondary rainforest through a gradual process of timber extraction and this is now the predominant vegetation type in terms of forest cover.

The main food source of the local people is rice and so there are large areas of paddy field and also there is a significant amount of cleared forest where the locals are removing forest by traditional slash and burn techniques in order to make way for agriculture.

Throughout the whole of Wallacea 18 major habitats have been identified (Coates 1997). These include: Ocean, Inshore waters, Offshore waters, Offshore islets, Seashore, Mangroves, Littoral woodland, Coconut plantations, Swamp forest, Swamps-Marshes and Lakes, Rice fields, Grassland, Scrub, Savannah woodland, Lowland montane forest, Secondary forest and Forest edge, Lowland rainforest, Montane forest and Alpine grassland.

The habitats that were present in the actual area of study could be broken down into the following categories:

1. Secondary Forest. This is a highly variable habitat, which ranges from tall forest to more scrubby woodland. There are some tall trees remaining as remnants of the original rainforest and it also includes the dense regrowth that occurs at forest margins such as along the edges of roads.
2. Poor secondary forest. Due to the diverse nature of secondary forest poor secondary forest is set to include secondary forest that has been particularly degraded by timber extraction or that is in the early stages of regeneration.
3. Forest edge. This type is in essence simply the edge of secondary forest with taller trees giving way to smaller ones with a more open canopy and more shrubs.
4. Paddy fields. Areas of agriculture for rice that are swampy habitat in their early stages and rich grassland in later stages.
5. Coconut plantation. Uniform areas of coconut palms, which may or may not have undergrowth and shrubs depending on how well managed they are.
6. Banana plantations. Banana palms often mixed in with coconut palms and other shrubs.
7. Mangroves. This is a specialist habitat that ranges from pioneering pneumatophores on the seaward side to tall, established forest as one journeys inland.
8. Shrubbery. This is typically the early stage of regenerating secondary forest that will develop over time into secondary forest. It is chiefly composed of shrubs with the occasional mature tree.
9. Scrub. This is a secondary community with varying amounts of low bushes and trees with some areas of patchy grassland.
10. Open forest. This is a low stature low-density forest that is low in diversity with a grassy understorey. It is usually found in drier areas or areas with low seasonal rainfall.
11. Village. This habitat type was included, as there were a number of communities on the sides of the roads. It is generally a mixture of coconut palms with shrubs and the occasional tall tree.

One advantage of carrying out a study on birds is that there has been a vast amount of work already done on the subject.

Therefore there are a number of set methods on how to carry out such work and the methods have to at least some extent been standardised.

The methodology employed in this study is based on similar studies that have been carried out in previous years to estimate abundances of birds and their habitat preferences. One example of such a study was carried out in the USA to assess the bird communities of the shrub-steppes in different habitat types.

In this example a line transect method was used in which transects were marked out and then as they were walked the birds that were observed along the transect were recorded along with their distance from the transect. Individual routes were walked four times when bird detectability was at its highest during the middle of June. Records were accumulated in a number of bands up to a range of 244 metres either side of the transect. Habitat data was recorded by laying ten perpendicular transects at fixed intervals across each route and recording at one randomly located point within each 10 metre length (Wiens, 1985).

Similar transect methods have also been employed in Finland to monitor the population levels of both wintering and breeding birds and in the North Sea to assess the distribution and abundance of seabirds.

Another popular method the point count method has been widely used for example the Danish Ornithological society used such a method to investigate the habitat preference of Danish breeding birds (NÃ hr, 1983).

The methodology for the study is based on a traditional line transects survey as described by Bibby (1992) and (1998) and Taylor (1983).

The other methods that could have been used include point counts, mark recapture, territory mapping and distribution studies but it was felt that line transects best suited the requirements of this study.

Line transects were chosen primarily because of the problem of access within the study site so the existing lines of communication such as roads and rough tracks through the forest could be used as transects.

Line transects also do have certain features that made them particularly attractive to this form of study. Due to the limiting factor of time then it was especially important to generate large amounts of data in as short a time span as was possible. By keeping moving it is possible to cover a lot more ground than if a fixed point method such as point counts were used and so a lot of data can be generated.

The ideal method to be utilised when conducting a study using line transects is to position the transects through a stratified random technique. This involves dividing up the study site with a grid, either on a map or actually on the ground and then using random co-ordinates to position the sampling site within each grid square.

However due the lack of access to many parts of the reserve this method could not be used and so instead existing means of communication were used to set out the transects. Therefore the greater number of transects were set up on roads with the rest utilising man made or animal tracks through the forest. As this form of sampling is prone to bias as bird communities that are found along forest edges would be over represented in the data collected an attempt to mitigate this was to ensure that each transect went through as many different habitat types as possible.

In total 9 transects were set up to try and represent the chief habitat types that were found in the immediate vicinity of Labundobundo. A brief scouting survey showed that the main types appeared to be secondary forest both good and poor, forest edge, coconut plantations and shrubbery and scrub. Therefore the transects were positioned so as to represent these main habitat types. In general there was little habitat homogeneity with each transect passing through a wide number of different habitat types. This meant that a great deal of data could be collected on the preference of the different species to habitat types without having to travel too far from base.

The transects are very different both in terms of the habitat that the transect contains and there length. The length of the transects was chiefly governed by the length of the communication network that they were following and so they vary in length from 1km to 1.9km. The total length of habitat that the transects cover is 27km which is sufficient to allow for reasonably accurate estimates according to Bibby, (1998).

The transects were set up by cutting a length of string 100metres long and then one person held the string while somebody else set off along the transect and marked the point along the transect when 100 metres was reached. This was marked using coloured tape with the distance also being written on. To ensure that the same birds were not recorded on more than one transect, the transects were well spaced out with no one transect overlapping any other.

Once the actual transects had been measured out then each one was walked and the habitat that was contained within each transect was measured. This was achieved by constructing maps of each transect and then recording the habitat type as far as the eye could see up to a maximum of 50 metres either side of the transect (An example of the maps used can be found in the Appendices). The transects were divided up into 10 metre blocks for this purpose and the habitat was recorded using symbols into 1 of the 11 categories previously described above.

Due to the limiting factor of time specific habitat surveys were not carried out with each section being assigned to a habitat class qualitatively rather than quantitatively. This does pose the problem of subjectivity as what one person sees as shrubbery another may see as open forest. In order to try and limit this factor at least to some extent not just one persons views were used and it is hoped by using more than one person to classify the habitat then this problem has to some extent been overcome. Also previous habitat studies by Coates, (1997) were referred to so as to form a basis for the habitat class each was assigned to.

Once the transects had been set up and the habitat within each transect had been mapped the data could start to be collected. The basic idea behind line transects is to walk along the transect and then map where an individual is seen and how far it is from the transect. This immediately poses 2 problems:

1. The correct identification of the bird species.
2. The correct measurement of the distance of the bird from the transect.

In order to overcome these problems the first week of the survey was spent getting to know the calls of each species of bird along with their likely occurrence and plumage. Pre recordings of the study species’ calls were studied and numerous "bird walks" helped all of the team to quickly become adapt at recognising the key species. In order to ensure that distances were measured correctly this skill was practised by one person walking a set distance away into the forest and then whistling. This was practised until each person could accurately estimate the distance from where the call originated.

Once the team was competent in these areas then data collection could begin.

Each transect was walked with a map of the transect and so when a bird was spotted its exact position could be marked upon the map. This would therefore show the habitat type that it was observed in and the distance that it was from the transect.

As well as this basic data the following information was also recorded where possible:

• Time of day bird observed
• Sex of bird
• Behaviour
• Weather conditions in the form of percentage cloud cover

The behaviour of the bird was recorded as a class with the following forms of behaviour recorded:

1. Singing/calling
2. Feeding
3. Preening
4. Perching
5. Flying
6. Playing/Displaying
7. Foraging

Where it was not possible to record the form of behaviour that the bird was exhibiting then a value of 999 was recorded and this was replicated for any other variable that was not obtained.

In order to try and minimise any form of bias that may distort the results the conditions when each transect was walked were kept as similar as was possible. It is a well-known phenomenon that birds are at their most active in the early morning and late afternoon (Bibby 1997). Therefore to ensure similarity between transects each one was walked at the same time each morning starting at 6am and also in the afternoon at 15.30pm.

Also the same amount of time should be spent looking for birds on each route and so the speed that the transects are walked should also be standardised. Bibby (1992) recommends a speed of about 2km per hour for open habitats and about 1km per hour for more dense habitats. Due to the fairly dense nature of the habitat on the routes then a speed of about 1km per hour was employed.

The one thing that could not be standardised to such a degree was the weather as due to limited time available every opportunity to get out into the field was taken advantage of. This therefore meant that even on poor days data was recorded but this will be taken into consideration in the analysis of the results as fewer individuals are generally recorded during adverse weather conditions. Adverse weather conditions include rainfall, either on the morning when the count is taking place or on the previous night as this can still affect the birds behaviour. The temperature, as on cooler mornings birds tend become active later on in the morning and also strong winds can make birds less conspicuous so fewer are recorded.

Each transect was visited on more than one occasion so as to replicate the results and give more viable data to be analysed. Also if one transect was visited on a bad day for example if there was a predator in the vicinity then if it was only visited once it would not be a true representation of the species composition. So by replicating the data there would be less chance of external factors influencing the results. Another reason would be to ensure that there is enough data in order to calculate abundance estimates. Sample sizes for line transect distance sampling data have to be quite large as small sample sizes contain little information about density and their precision is poor, regardless of the analytical method used. An ideal minimum should be approximately 60-80 records according to Bibby (1997).

Even if taking the steps outlined above reduces the effects of bias there are still some assumptions that have to be made if the data is to be used in order to produce abundance estimates.

1. The most important of these is that all of the birds exactly on the route are detected as even the most sophisticated distance model will not work if birds that are directly on the line are overlooked. This can be overcome by walking at such a pace that allows adequate time for detection of species but not too slow as to make the study nothing more than an elongated point count.
2. Birds do not move before detection as all variants of density estimation assume that the birds are randomly distributed with respect to the distance from the route Bibby (1992). If the bird moves in response to the observer disturbing them then this will cease to be the case. It is possible that some birds may be attracted to the observer but the most usual reaction would be to flee.
3. In order to use any form of abundance model then it is also assumed that all the distances to the bird are measured accurately. Even if the bird does flee then it is not only important to measure to the point where it was first of all detected but also that one measures the distance to the bird perpendicular to the transect.
4. Individual birds are counted only once to give an accurate estimation of abundance.
5. Individual birds are detected independently. This can be a problem if the bird is more detectable at higher densities because the action of one bird singing may stimulate others to do so and so mean they are not detected independently. Conversely at lower densities birds may be less likely to reveal themselves, as they are not stimulated by others.
6. Bias is understood. The bias can be in a number of different forms as described in the previous section including observer skill, weather conditions, time of day and season. All of the possible forms of bias should be understood and mitigated against if such factors are to be removed as a form of error.

The results that are obtained will be able to be used in a number of ways. The first of these will be to produce relative estimates of abundance for each of the bird species as a representation of the whole area studied. Relative abundance’s rather than absolute estimates will be calculated, as the study area is not a full representation of the whole area and so the accuracy of the estimate is of secondary importance. The actual density of the birds can therefore only be calculated on a local scale and cannot be calculated to a high degree of accuracy on a large scale.

How the density of one species relates to another can then be compared using the abundance data and how common each species is in relation to the others can be determined.

The most basic method of estimating the abundance of the species is to divide the number of birds by the area of the transect to give a value of the number of birds per unit area. This can then be taken one step further by incorporating how far away from the transect the bird was observed and so a more accurate figure can be produced. In order to perform such calculations large sample sizes are needed and so the transect data shall be combined to in effect give one set of data for one transect.

In order to analyse the habitat preferences of the study species then the number of birds that were observed within each of the different habitat classes will be divided by the area that the particular habitat type covers. If one merely looked at the number of birds that were observed within a certain habitat type within a particular transect then this would give biased results. This would be primarily due to the fact that the amount and composition of habitat type varies between the transects and so more birds may be found in a certain transect because it has got a greater area of the habitat type that the bird prefers. Therefore combining the transect data will give a much more curate representation.

Unfortunately the sample size for each species within each habitat type is not great enough to allow for more detailed/accurate estimates to be made for each species within each habitat. However a value can still be produced that will be able to compare the numbers of each species within each habitat class and therefore the habitat preference of each species can be determined.

The transects pass through a wide ranging array of different habitats and the amount of each habitat is not uniform as the graph below summarises.

The most abundant habitat class in terms of the percentage of the total area is clearly poor secondary forest with 20.9% closely followed by shrubbery with 17.9%.

There are very similar amounts of forest edge and coconut plantations both around 15% with the least abundant habitat type being paddy field and banana plantation with under 2% each of the total area.

The habitat type that used to dominate such tropical regions, good secondary forest now only makes up 9% of the total area of the study site with such introduced habitat types as coconut plantation and banana plantation now making up a much greater percentage.

As a result of the deterioration of the natural forests there is also much greater expanses of habitat types in varying stages of succession. This varies from the very early stages of scrub and shrubbery to the later stages of open forest.

In total 9 transects were sampled and from these transects 1016 individual birds were recorded covering the 8 study species.

From table 1 and graph 2 we can see that the most frequently recorded bird was the Grey-sided Flowerpecker followed by the Brown-throated Sunbird. The Lemon-bellied White-eye was the bird with the fewest number of recordings and the Crimson Sunbird also proved to be uncommon. If all of the birds occurred in similar concentrations then we might expect to find similar numbers of them. A null hypothesis that the birds are all found in similar concentrations can be tested statistically using a Chi-squared test. A value of 362 is obtained with 7 degrees of freedom, which is not significant, as it is greater than the critical value for P>0.01 of 18.48.

Therefore the null hypothesis is rejected as we can see that the birds are not found in similar concentrations throughout the study area.

If this is then split up into the different habitat classes then as Table 2 summarises very different counts of individuals were recorded in each habitat class.

By examining this data we can see that there is certainly a clear difference in the total number of birds that were observed in the different habitat classes if all of the transect data is combined.

The greatest numbers of species were recorded within shrubbery with the least number of individuals being found in banana plantations and mangrove. If the percentage of the total number of birds per class is looked at then shrubbery accounts for nearly 1/3 of all sightings with the next largest classes being poor secondary forest and coconut plantation each accounting for 15% of the sightings. This information can be represented clearer on graph 3.

This data can be tested statistically to see if the birds are distributed randomly throughout the different types of habitat.

The following null hypothesis could be tested using a Chi-Squared test: that the same frequency of birds could be expected to be found in each type of habitat and they are therefore evenly distributed throughout the different habitat types.

A value of 1004 is obtained with 10 degrees of freedom and the critical values for P>0.05 and P>0.01 are 18.31 and 23.21.

Therefore as the value of Chi-squared is greater than both of these values there is a significant difference between the observed and expected frequencies.

We can reject the null hypothesis that the birds are found in even numbers throughout the different habitat types and therefore they are not evenly distributed.

Estimates of the abundance of each species can be made from the number of birds observed divided by the total area of the transect. This however does not take into account the distance that the bird was from the transect when the observation was made.

The distance to k curve first of all needs to be calculated for each species showing how the probability of detecting a bird decreases with distance from the observer. There are a number of models that can be used to estimate the cut off distance for a species but as this study employed full distance modelling then it was possible to construct individual models for the different species.

From the graphs above it is then possible to work out a cut off distance for each species. This is the distance away from the transect where the likelihood of detecting an individual becomes improbable.

The cut off distance is important as it is used in the calculations of the abundances of the birds.

All of the species of Sunbirds showed a relatively steady decline in detection away from the observer. This implies that more birds are detected nearer to the transect and the decline in observations is due to the effect of increasing distance and more vegetation between the birds and the observer henceforth making them harder to observe.

The cut off distance for all of the Sunbird species was therefore taken as 30 metres.

A similar trend was also found for the Grey-sided Flowerpecker and so it was assigned with the same cut off value.

The Yellow-sided Flowerpecker was recorded to a much greater extent close to the observer with numbers of observations decreasing rapidly with increasing distance. This could be related to the fact that it is a very vocal species and so as it was only disturbed in close proximity to the transect that is where it was most conspicuous. The cut off distance was therefore shorter at 20 metres.

Both of the species of White-eyes were generally more conspicuous at distances further away from the observer perhaps keeping a lower profile when an observer was near by and so they were given a greater cut off distance of 40 metres.

Abundance estimates could be made using the formula given by Bibby (1992) for densities from full distance recording.

The following equation is used to give values for a₁ and a₂:

a_k = (2/nw) Sum cos (Pikx/w), for k = 1, 2, 3 etc

n = the number of birds in the sample

x = the perpendicular distance from the transect

w = the cut off distance

L = the length of the transect

One a value for a₁ and a₂ has been determined a critical value can be calculated,

1/w(2/n + 1)^1/2>abs(a_m+1)

If this value exceeds the value of a₂ then a value of f can be obtained from the formula:

f = a₁ + 1/w and from there a density estimate can be produced:

Density (D) = nf/2L to give a value in number of birds per metres² which can be multiplied by 10000 to give a figure per hectare.

The results of carrying out the analysis, which can be found in full in the appendices, are summarised below.

All of the density estimates are very low as the results give the abundance of the birds throughout the entire study area.

Even from this data one can see that the most densely abundant species is the Grey-sided Flowerpecker followed closely by the Yellow-sided Flowerpecker and the Brown-throated Sunbird. The most sparsely abundant species is clearly the Lemon-bellied White-eye with low figures also found for the Crimson Sunbird and Pale-bellied White-eye.

As already established the study species were found in varying numbers in the different habitats around the study site, which can be summarised in graph 12 below.

In order to determine the habitat preferences of the species each one shall be looked at in turn.

To get a better representation of the habitat preference of the birds the numbers found in a particular habitat class were divided by the actual amount of that habitat that was contained within the transects, to give a frequency per hectare of habitat.

Also in order to look at specific habitat preferences one must first prove that the birds are not distributed evenly throughout the habitats that they were found in and so a Chi-squared test will first of all be carried out for each species.

Table 4 and graph 13 clearly shows that there is a difference in the number of Olive-backed Sunbirds over the different types of habitat that they were observed in.

This can also be proved statistically with a Chi-squared value of 105 with 7 degrees of freedom being higher than the critical value of P>0.01 which is 18.48.

Therefore the result is not significant and the observed results vary from the expected results by a margin that cannot be put down to chance.

The highest counts were observed in shrubbery with relatively high counts also found in scrub, coconut plantation and open forest.

If the frequency per hectare is examined then the highest figure was also obtained for shrubbery with the next highest being scrub and then open forest.

The most preferred habitat type of the Olive-backed Sunbird appears to be shrubbery, which had by far the highest count per hectare and the highest overall count, followed by scrub and open forest.

Existing literature states that coconut plantations are its preferred habitat type and so these results do not correspond. Shrubbery was also stated as a habitat where the bird is usually found and so this does compare favourably. One possible reason for this discrepancy could be that there is a higher percentage of shrubbery than coconut plantation in the study area but not so great as to account for such a large variation.

Table 5 and graph 14 shows that there is an uneven distribution between occurrence in different types of habitat and this is also reflected statistically using Chi-squared.

A value of 338 with 7 degrees of freedom proved to be not significant as it was greater than the critical value of P>0.01, 18.48.

The highest counts were observed in coconut plantations and shrubbery with the lowest counts being found in banana plantations and village environments.

The frequency of birds per hectare also followed this pattern with by far the greatest concentration found in coconut plantation and shrubbery with the lowest found in banana plantation.

The preferred habitat type of the Brown-throated Sunbird would appear to be coconut plantations and areas of shrubbery where the highest count and greatest concentration per area were found.

This ties in well with existing literature, which gives coconut plantation as its preferred habitat.

Table 6 and graph 15 again show an uneven distribution between habitat types, which can be reflected statistically.

A value of 162 was obtained for Chi-squared with 9 degrees of freedom, proving to be not significant as this is much larger than the critical value of P>0.01 of 21.67.

The greatest number of individuals was found in coconut plantations with high numbers also found in poor secondary forest and shrubbery. Relatively high values were also found for forest edge and good secondary forest with the remaining habitat types containing few individuals.

The greatest frequency per area of habitat was also found in coconut plantations followed by poor secondary forest, forest edge and shrubbery. A relatively high figure was found within mangroves but due to the small number of birds actually recorded there this is not a true representation.

The Black Sunbirds preferred habitat would therefore appear to be coconut plantations and areas of forest and shrubbery, as it was found in high concentrations and numbers in the whole range of forested habitats.

Previous work showed that the birds were found to be most common in areas of secondary forest and forest edge so this gives a positive correlation.

Table 7 and graph 16 illustrate that the Crimson Sunbird is not distributed as unevenly as the species so far mentioned. This is statistically significant with a Chi-square value of 17.09 with 7 degrees of freedom being below the critical value of P>0.01, 18.48 and so the results are significant to a level of 0.017.

The Crimson Sunbird would therefore appear to be distributed fairly evenly throughout the different habitats. If the count data is looked at then there was a much higher count found in shrubbery and poor secondary forest but this is not backed up by the frequency per area data. There the highest values were found in open forest and village environments but this is from single observations and so should be discounted.

There is therefore not sufficient difference in the numbers of observations to accurately state the preferred habitat type of the species but one could loosely say that they prefer shrubbery and poor secondary forest from the high counts found there.

Overall the species was not found in any great numbers being one of the most sparsely encountered birds on the study and so further study is needed to clarify its preference. On the other hand the species may not have specific habitat requirements, which would explain why it was found in similar numbers throughout the different habitats.

This does tie in to a certain extent with existing literature on the species, which states it is a rare to uncommon species found in a wide range of different habitats.

At first glance at table 8 and graph 17 the Yellow-sided Flowerpecker seems to be fairly evenly distributed but this is not backed up statistically.

A Chi-squared value of 47.9 with 5 degrees of freedom is higher than the critical value of P>0.01, 15.09 and so it is not significant.

High counts were found in all of the habitat types that the birds were found to occupy apart from open forest and to a lesser extent scrub.

However the birds were only found in a fairly small number of habitat types, with the highest number found in shrubbery and poor secondary forest.

If the frequency per hectare is examined then high values were obtained for all of the forest type habitats with the lower values for shrubbery and scrub still being relatively high.

Therefore the preferred habitat type of the Yellow-sided Flowerpecker would appear to be areas of forest especially forest edge and also areas of shrubbery and scrubland.

This is backed up by previous work, which states that the species was common throughout Buton in virtually all forested areas (Previous Op Wall study).

From table 9 and graph 18 one can see that the Grey-sided Flowerpecker is not distributed evenly throughout the different habitats. This can also be backed up statistically as a value of 234 was obtained for Chi-squared and with 7 degrees of freedom this is greater than the critical value of 18.48 and so is not significant.

By far the highest number of birds was counted in shrubbery with high numbers also found in poor secondary forest, forest edge and scrub.

Examining the frequency per area data showed that the highest concentration was found in forest edge with shrubbery, poor secondary forest and areas of scrub also containing high concentrations.

This corresponds to the count data and so the preferred habitat type of the Grey-sided Flowerpecker is shrubbery but it also frequents forest edge and secondary forest and scrub.

Previous studies noted that this species was widespread and not particularly fussy in its choice of habitat and this certainly relates reasonably well with the results obtained.