BIOLOGICAL CONSERVATION

Biological Conservation 121 (2005) 135155 www.elsevier.com/locate/biocon

The eects of extreme forest fragmentation on the bird community of Singapore Island
Marjorie Castelletta a, Jean-Marc Thiollay
a

b,*

, Navjot S. Sodhi

Department of Biological Sciences, National University of Singapore, 14 Science Drive 4, Singapore 117543, Singapore b UMR 5173 CERSP Museum National dHistoire Naturelle, 61 rue de Buon, 75005 Paris, France Received 10 October 2002; received in revised form 15 March 2004; accepted 16 March 2004

Abstract Singapore Island suered one of the highest known deforestation rates in the tropics from the mid-to-late 19th century when over 95% of its native lowland forest was cleared. We compared the current bird community structure and composition among three habitat types, i.e., old (>50 years, 7935 ha) and young ( 6 50 years, 2949 ha) naturally regenerating secondary forests and abandoned wooded plantations (27102 ha) dominated by exotic species. Forest patch area had the strongest inuence on the current species richness. The overall bird richness was not greater in most mature forest patches, but 20 species were only found in the old secondary forests and ve of these were found in <50% of these patches. The rapid decrease in the number of forest species in plantations was oset by an increase in the number of open habitat species. Comparisons with current bird communities in nearby mainland forest sites (Peninsular Malaysia) suggest that the forest avifauna of Singapore is depauperate. The preservation of larger mature and maturing forests is therefore required for conserving the extant forest avifauna in Singapore. Connecting isolated patches can also be envisioned to facilitate movements of forest birds that have low densities and restricted distribution. 2004 Elsevier Ltd. All rights reserved.
Keywords: Tropical forest; Habitat degradation; Conservation; Avifauna; Island; Southeast Asia

1. Introduction Tropical rainforests are unique sources of biodiversity and most of them are disappearing at an alarming rate (Achard et al., 2002). Lowland tropical forests in particular (below 1000 m) are being lost or degraded by logging, hunting, cultivation or urbanisation (Laurance, 1999). Extinctions within the tropical belt primarily as a result of the inability of biotic communities to withstand these environmental changes have been reported (Leck, 1979; Karr, 1982; Brash, 1987; Diamond et al., 1987; Newmark, 1991; Castelletta et al., 2000). Studies on the impact of deforestation and related disturbances on the tropical wildlife are needed to determine the conservation value of remaining disturbed and isolated forests.

Corresponding author. E-mail address: thiollay@mnhn.fr (J.-M. Thiollay).

Such studies are few from Southeast Asia where deforestation has been rapid and extensive (Laurance, 1999; Achard et al., 2002). We studied the landbird fauna of Singapore because it provides one of the most extreme cases of tropical deforestation in the world, with a moderate level of insularisation (Turner et al., 1997). It is a relatively small island (585 km2 ), at the tip of the Malay Peninsula and it was originally covered with dense rainforest. On the island, heavy deforestation started shortly after 1819, pristine forests were cut, then cultivated or left to regenerate when not used for urbanisation and construction of water reservoirs (Fig. 1; Corlett, 1992). Only a tiny fraction of the natural forest (0.04%) remained unexploited and this was reduced to small disturbed fragments now scattered in a highly urbanised matrix. Thus, the local forest bird community, already aected by its insular situation, suered the cumulative eects of habitat reduction, fragmentation and degradation. The

0006-3207/$ - see front matter 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.biocon.2004.03.033

136

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

Fig. 1. A regional map of Southeast Asia with Singapores location. Enlarged view of Singapore showing the location of the study patches (initials, see Table 1). Patches only used for isolation measurements are indicated by a number with 1 Chesnut Peninsula Forest, 2 Lower Pierce Forest, 3 Upper Seletar Peninsula Forest, 4 Seletar Forest, 5 Mount Imbiah Park, 6 Mount Serapong Wood, 7 Ubin Island East Wood. Dotted lines indicate where patches could be connected.

deforestation and concomitant large-scale fragmentation of the forest has had detrimental eects on the forest bird community in Singapore as >90% of the forest avifauna has been lost (Castelletta et al., 2000). The aim of this study was to assess the conservation value of existing remnant and regenerating forests for the forest avifauna of Singapore. We determined the eects of variables such as patch size, forest type, isolation and vegetation characteristics on the bird community. Such variables have been previously found to aect the forest bird diversity (e.g., Pearson, 1975; Opdam et al., 1985; Rotenberry, 1985; Bierregaard et al., 1992; Schmiegelow et al., 1997). We also make two comparisons with neighbouring Peninsular Malaysia. First, to determine the conservation importance of dierent Singaporean patches, we compared their bird richness with that in larger and less disturbed forests of nearby Peninsular Malaysia. Second, we compared bird community similarity of our forested patches with regenerating Malaysian forests to determine whether degraded forests have similar bird richness. Important conservation implications of our case study include identifying priority habitats and sites

for conservation in Singapore, and management directions needed to stop any further impoverishment of the islands forest bird richness. Our results also have conservation relevance for degrading forested landscapes within Southeast Asia.

2. Study area Singapore is a tropical Southeast Asian island that lies between 1090 N1290 N and 103380 E104060 E. It has a maximum diameter of 23 km (NS)42 km (EW) and is only separated from southern Peninsular Malaysia by the 1.4 km Straits of Johor. Most of the country is lowland reaching a maximum elevation of 163 m at the Bukit Timah Nature Reserve. The close proximity of Singapore to the equator accounts for a typical equatorial climate characterised by high mean annual temperatures (30 C) and high average daily humidity of 90% (Corlett, 1992; Jeyarajasingam and Pearson, 1999). The mean annual rainfall is over 2000 mm and exceeds 100 mm even in the driest months. The main wet season

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

137

occurs from November to January with usually another less pronounced rainfall peak in JuneJuly.

3. Methods 3.1. Habitat types Our aim was to study the bird communities in dierent successional stages of the regenerating tropical forest. Based on published literature (Corlett, 1997; Turner et al., 1997) and personal observations, we dened three habitat categories in Singapore: (1) old secondary forest (OSF; >50 years old), (2) young secondary forest (YSF; 6 50 years old), and (3) abandoned plantations (P) mostly of rubber tree (Hevea brasiliensis) with other exotic species. Vegetation details of these forests are in Table 1.

Because of their secondary origin, most patches were heterogeneous in their vegetation as well as in their regeneration stage (Corlett, 1997). Therefore, each forest patch was classied into one of the three categories above according to its dominant vegetation type (>75% of total area). We collected data from 17 patches of which six were old secondary forests, four were young secondary forests and seven were plantations (Table 1; Fig. 1). Seven additional existing forest patches could not be surveyed because of restricted access. We selected existing trails that were already crossing the patch to avoid creating additional disturbance and to ease transect counts. 3.2. Bird community sampling methods Terborgh et al. (1990) and Thiollay (1994) have pointed out the importance of combining complementary

Table 1 Description of the Singaporean patches surveyed by transect counts between June 1997 and 1999 Patch name (abbreviations) Old secondary forest Bukit Timah Nature Reserve (BT) Nee Soon Forest (NS) MacRitchie Forest (MR) Botanic Garden Forest (BG) Lornie Forest (LF) Labrador Nature Park (LA) Young secondary forest Mount Faber Park (MF) Telok Blangah Hill Park (TB) Kent Ridge Park (KR) Bukit Batok Nature Park (BB) Plantation Poyan Wood (PO) Holland Wood (HO) Bukit Batok West Wood (BW) Ulu Pandan Canal (UP) Loyang Wood (LO) Yishun Park (YI) Khatib Bongsu Wood (KB) Matrix Area (ha) Interpatch distance (m) 8446 8962 8087 8652 8322 10,966 Shrub cover (%) 28.7 (4.3) 63.6 (2.8) 57.7 (2.2) 46.0 (15.0) 27.0 (5.1) 36.3 (10.7) Canopy cover (%) 91.6 (0.9) 92.7 (0.6) 91.0 (0.8) 92.7 (1.4) 93.2 (1.4) 88.5 (0.6) Transects length (m) 2067 4280 5038 966 2012 793 No. of surveys 12 12 12 11 11 12

Urbanized area Water reservoirs and grassland/urbanized areas Water reservoir and urbanized areas Urbanized areas Water reservoir and urbanized areas Urbanized areas and sea front Urbanized areas Urbanized areas Urbanized areas Urbanized areas

145 935 484 7 58 11

49 29 32 38

10,639 10,150 9974 8995

21.1 (8.5) 20.9 (7.3) 12.1 (3.1) 27.0 (9.6)

87.8 (2.4) 72.4 (11.7) 84.1 (0.9) 91.3 (3.1)

1991 1523 1947 863

12 12 12 12

Water reservoir and marshland areas Grassland/urbanized areas Urbanized areas Urbanized areas Urbanized areas Urbanized areas Sea front and grassland areas

35 69 102 45 34 27 100

16,900 8589 10,697 8935 21,064 12,320 13699

43.6 (13.8) 69.6 (10.7) 71.0 (11.0) 16.3 (5.2) 11.3 (11.3) 14.5 (4.5) 86.7 (5.5)

57.5 (16.8) 80.9 (6.0) 40.4 (17.1) 88.8 (2.2) 86.6 (5.9) 92.7 (1.8) 67.6 (13.5)

1321 1531 794 1381 1201 1171 1930

11 9 10 11 12 11 9

For description of habitat types, see Methods. Mean ( SE) of all samples is given for vegetation variables. Interpatch distances are based on all existing forest patches in Singapore.

138

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

methods to provide a more accurate sampling of the tropical bird populations. To determine the adequacy of transect counts, mist-netting (1014 nets of 2.5 12.0 m and 30 mm mesh, opened between 07:00 and 13:00 h) was conducted in the two largest patches of old secondary forest (1442 and 1413 net hours in Nee Soon and MacRitchie Forests, respectively) for a side study (see Sodhi, 2002). Mist-netting conrmed that all species had been detected on the transect count and therefore that our sampling eort was likely to be adequate. This is why we used observed species richness instead of any estimate of actual richness based on a rarefaction or extrapolation procedure. Transect counts were conducted every two months from June 1997 to June 1999 in all patches with some exceptions (Table 1). Transect counts were done from 07:00 (just after sunrise) to 10:30 h, i.e., during the highest bird activity. No survey was conducted during unfavourable weather conditions (rainy or windy days). One experienced observer conducted all the surveys. He slowly walked at a pace of 100 m per 10 min (Ralph et al., 1993) and recorded the number of individuals seen or heard within 50 m on each side of the trail. Birds just ying over the transect were not taken into account, except aerial feeders and diurnal raptors essentially observed ying above the forest. Two consecutive days were needed to survey trails more than 3 km long. Bird locations were mapped in order to avoid double counts. The starting and ending points of consecutive surveys were alternated in order to optimise the bird sampling in the whole patch. Because existing trails were used, the sampling intensity in all the patches could not be strictly proportional to the area. Therefore, initially, we drew the sampling eort curves for each patch using only resident birds because they were the most representative members of the bird community in terms of number and occurrence. For most patches, an asymptote was reached suggesting that an adequate sampling of the avifauna had been achieved (Fig. 2). 3.3. Patch structure and attributes Sixteen vegetation and landscape variables were recorded at each forest site to determine the eects of environmental variables on bird fauna. The vegetation was sampled along the survey trails in all forest patches from April to June 1998 and 1999. The sampling plots were at least 50 m apart and distributed along the trail. In each plot, vegetation was measured within a 5 m radius. Eleven vegetation variables recorded using a modied version of James and Shugart (1970) were: the number of trees higher (NTH) and smaller (NTS) than 10 m, the diameter at breast height (d.b.h.) of trees higher (DTH) and smaller (DTS) than 10 m, the number (NTD) and d.b.h. (DTD) of dead trees, the number (NTP) and d.b.h. (DTP) of palm trees, the litter depth

(LEAF), the understory shrub (SHRUB) and the canopy cover (CAN). Two observers visually estimated tree height and shrub cover and a mean value of the two estimates was then used. The litter depth was measured 12 times per plot by inserting a rule vertically down to soil. The canopy cover was estimated using a densiometer (Lemmon, 1957). We used aerial photographs (1:20,000, from 1995) to dene the boundaries of forest patches separated by >100 m-wide gaps of grassland and urbanised areas. We digitised the forest patches using the Geographic Information System (GIS). We used Arc/Info GIS software on a digitised map of the physical topographic features of Singapore to measure individual area (AREA) and perimeter (PERI) of each patch. Their shape index (SHAPE), dened as Sh P =2pA1=2 , where P and A, respectively, were the perimeter and area of the patch, was then calculated (Patton, 1975). We also used the GIS to calculate two measurements of patch isolation: the interpatch distance, i.e., the mean distance of a patch to all other patches from centre to centre (IDIST), and the percentage of forest cover within a 1 km radius around each patch (ICOV). All existing forest patches in Singapore were used for patch isolation calculations. 3.4. Species status and population density estimates We removed from the analysis all the waterbirds and shorebirds. We recorded raptors separately either from the understory or from openings (road, streams, treefall or clearing) and vantage points (hilltops) outside the forest wherever possible. Birds in Singapore were classied into native resident, migrant and introduced species. We subdivided residents into three categories according to their natural habitat and their association with other habitat types based on personal observations and Lim and Gardner (1997): (1) forest species, i.e., those found in our mature secondary forest patches (both interior and edge), (2) plantation species, i.e., those using other types of wooded areas, mainly old plantations and secondary growth in gaps or clearings of secondary forest, and (3) grassland species using open and little wooded areas, grasslands, wetlands (e.g., ponds) or golf courses. Migrants were temperate/intratropical passage migrants or winter visitors. The common myna (Acridotheres tritis) that naturally colonised Singapore from Southeast Asia was considered as a resident (Ward, 1968). All introduced species presumably resulted from accidental releases of captive birds (e.g., yellow-crested cockatoo Cacatua sulphurea and white-crested laughingthrush Garrulax leucolophus). The velvet-fronted nuthatch (Sitta frontalis) once recorded in Bukit Timah, was considered as a vagrant (rare occurrence). Six species that had both resident and migrant populations

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

139

70

Old secondary forest BG LF BT MR NS LA S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12

Surveys

50

30

10 50
Cumulative number of species

Young secondary forest KR BB

30

TB MF

10 S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 60 Plantations

Surveys

UP KB BW HO 40 LO PO YI

20 S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12

Surveys

Fig. 2. Cumulative number of species recorded during transect surveys (S1-12) in dierent patches. Patches are grouped according to their habitat type (patch names in Table 1).

(dollarbird Eurystomus orientalis, blue-throated beeeater Merops viridis, drongo cuckoo Surniculus lugubris, violet cuckoo Chryosococcyx xanthorhynchus, blacknaped oriole Oriolus chinensis and common koel Eudynamis scolopacea) were included in the resident category. The density of a given species or group was calculated as the total number of individuals recorded per survey divided by the forest area sampled. We referred to total species richness as the total number of species per patch. 3.5. Comparative data from mainland sites To assess the insular and forest degradation eects, we compared the presence/absence data of Singapore to those of study sites on the nearest mainland, i.e., Peninsular Malaysia. Migrants and vagrants were excluded from all analyses. The avifauna of four primary or little

disturbed Malaysian forest sites was derived from Bransburry (1993), including one large (Taman Negara), one medium size (Endau Rompin) and one small (Pasoh) forest patch, as well as an island (Tioman) more than four times smaller than Singapore and 68 km away from Peninsular Malaysia. Four dierent selectively logged sites were in the primary dipterocarp forest of the Sungei Tekam Forestry Concession where Johns (1986) surveyed four compartments (within a continuous forest) one before, during and after logging, and three, respectively, 2, 4 and 6 years after logging. We also used surveys of a 830-ha fragment of primary dipteropcarp forest at the Universiti Kebangsaan Malaysia, 2550 years after it was selectively logged (Ford and Davison, 1995), supplemented by the mist-netting survey of Karim-Dakog et al. (1997). Various studies used dierent methods of data collection (e.g., surveys and surveys as

140

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

well as mist netting). We assumed that despite sampling and individual dierences, each study succeeded in making an adequate bird inventory. Site descriptions are given in the respective references. We classied the bird species of Peninsular Malaysia according to their habitats following: Jeyarajasingam and Pearson (1999), Wells (1999) and personal observations. Species found mostly in the interior of large mature forests and late secondary forest patches were categorised as forest species. 3.6. Statistical analyses We ran stepwise multiple regression analyses to determine which of the environmental variables were related to the total number of species in each patch and to generate speciesarea relationships. Because of small samples sizes, we pooled data from all secondary forests. We are aware that this might have caused some bias in our analysis. Stepwise regression analysis was performed on log-transformed total number of species, and on resident, migrant and introduced species. Environmental data were either counts (NTS, NTH, NTD and NTP), mean percentages (SHRUB, CAN, ICOV), or measurements (DTS, DTH, DTD, DTP, LEAF, AREA, SHAPE, PERI and IDIST) that were, respectively, square root-, arcsin- or log-transformed following Zar (1996). To avoid multicollinearity, we ran Pearson correlations on the 16 environmental variables (Zar, 1996). The most biologically relevant variable among those that were highly correlated (r > 0:7) were retained for further analysis. Thus, 10 independent variables were retained (AREA, ICOV, IDIST, NTH, NTS, DTH, DTP, LEAF, SHRUB and CAN). Because the number of variables tested must be lower than the number of observations (i.e., the number of patch, Sokal and Rohlf, 1981), two of them (DTH and DTP), which had the lowest inuence on the bird community, were deleted. At each step of the stepwise multiple regression, the level to enter or remove a variable was P 0:05. Repeated-measures ANOVA were run to investigate if the density of the species group was signicantly different among the three habitat types. Densities were logtransformed when the residuals were not randomly distributed. TukeyKramer post-hoc test was used to determine which habitats were dierent from the others. To compare the bird diversity of the Singaporean patches with those of primary and logged forests in Peninsular Malaysia, we performed cluster analyses on a presence/absence data matrix of the Singaporean and Malaysian sites using Wards method (MacCune and Meord, 1995). All statistical tests were performed using SAS 6.12 except for cluster analysis performed with PC-ORD 2.0 (MacCune and Meord, 1995) and Repeated measures

ANOVA with SuperANOVA (1989). In text and tables, values are means SE.

4. Results 4.1. Community composition among habitat types in Singapore A total of 167 species were detected from all patches. Eighty-six recorded species were residents (80 species with resident populations and 6 with both resident and migrant populations), 54 were strictly migrant species, 26 were introduced species and one was a vagrant species (Appendix A). The mean species richness of dierent bird categories was the highest in plantations and lowest in old secondary forest (Fig. 3), owing to the large number of less specialized species occurring in young secondary forests and plantations (see Appendix A). There was less distinction among forest types in the densities of each bird category although densities were still signicantly higher in plantations than in old secondary forest (Fig. 4). The density of introduced species was also lower in older forests than in other forest types (Fig. 4). Of the 86 resident species, 20 are here considered to be true forest species while the other 66 occur widely in other habitats. The density of true forest resident forest species was highest in the old secondary forest (F 3:68, P 0:05) but this was masked by the numerous other resident species, which were least common in old forests (Fig. 4). All 20 of the true forest species were found in four old forest patches (BT, NS, MR, and LF; Fig. 5). These true forest species may themselves be tentatively divided into three groups (Fig. 5). (1) Five were restricted to one or two patches of old forest, e.g., the scarlet minivet (Pericrocotus ammeus) and white-bellied woodpecker (Dryocopus javensis) were found in a single patch; (2) a further 12 occurred in 34 patches of old forest; and (3) three species occurred in 35 patches but also in moderate to high densities in one of the young secondary forest patches. This patch BB was close to the old forest BT from which the birds could have originated (see Fig. 1). The occasional presence of some forest birds in young forests and plantations may be a result of accidental release of caged birds, migration from neighbouring Malaysia, or vagrant individuals. However, it is also possible that some forest species are attempting to colonize new patches. 4.2. Fragmentation eects Higher species richness of resident and migrant species and of all species pooled occurred in larger patches (Table 2). Other variables such as shrub cover

4 3 2 1 0

(a) All species

1.6 1.2 0.8 0.4 0

(c) Introduced species

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

Log Species richness

Old secondary forest Young secondary forest

F = 3.64, P = 0.05
0.4

Plantation

Old secondary forest Young secondary forest

F = 5.39, P = 0.02
0.4

Plantation

(b) Residents

(d) Migrants

0.3 0.2 0.2

0.1 0

Old secondary forest Young secondary forest

F = 3.40, P = 0.06

Plantation

Old secondary forest Young secondary forest

F = 5.19, P = 0.02

Plantation

Fig. 3. Log mean ( SE) species richness of dierent categories of birds in dierent habitat types of Singapore. Comparisons of means were performed using repeated-measures ANOVA with df 2, 11. Log-transformed values of introduced species were translocated to positive values by adding a value of 2.

141

142

1.5

(a) All species

(c) Introduced species

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

0.5

Log Density (birds/ ha)

0 Old secondary forest Young secondary forest F = 8.16, P = 0.004 Plantation

0 Old secondary forest Young secondary forest F = 9.72, P = 0.002 Plantation

1.5 1

(b) Residents

(d) Migrants

1
0.5 0 Old secondary forest Young secondary forest F = 4.50, P = 0.03 Plantation

0 Old secondary forest Young secondary forest F = 2.96, P = 0.08 Plantation

Fig. 4. Log mean densities ( SE) of dierent categories of birds in dierent habitat types of Singapore. Comparisons of means were made using repeated-measures ANOVA with df 2, 11. Migrant and introduced species densities were logx 2 transformed to have positive values. Log-transformed values of introduced species were changed to positive values by adding a value of 2.

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

143

Fig. 5. Densities of forest species densities in dierent forest types of Singapore. Presumed species status: m? migrant, i? introduced and v? vagrant. See Appendix A for the scientic names of the birds.

Table 2 Relationship between species richness and environmental variables for all species, resident, migrant and introduced species Partial r2 All species AREA (A) SHRUB (Sh) Resident species AREA (A) IDIST (I) Migrant species AREA (A) Introduced species CAN (C) 0.68 0.15 0.98 0.01 0.47 0.55 F 17.24 6.08 296.5 97.1 7.19 9.97
2

P 0.003 0.04 0.0001 0.04 0.03 0.01

Equation S 0:15A 0:25Sh 1:71

S 0:15A 0:24I 2:35

S 0:21A 0:84 S 1:82C 3:02

Stepwise multiple regression with partial coecient of determination (partial r ), ANOVA F and associated probability P . AREA is patch size in ha. CAN is canopy cover. IDIST is a measure of patch isolation. SHRUB is shrub cover. For further description of environmental variables, see Methods.

and inter-patch distance made relatively little contribution to the regression model. More introduced species were found in patches with open canopy (Table 2). On average, eight more resident species were found in patches that were >50 ha than those that were <50 ha (Table 1, Appendix A). The number of true forest resident species peaked at 18 in MR, which was about 500 ha (Fig. 6).

4.3. Comparison with Malaysian forest bird communities The total bird community of all mainland primary and old secondary forest sites (mean species richness 186 35, Fig. 7) were distinct from the 17 Singaporean sites and the small Tioman Island (45 2 species); the similarity between these two groups of communities was <50%. Singaporean sites had more plantation species

144
20

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

MR 18 16 BT NS

Number of species

14 12 10 8 6 4 BB MF HO TB YI BW 200 300 400 500 600 700 800 900 1000 LF

Old secondary forests Young secondary forests Plantations

2 LO BG 0 0 LA

100

Area (ha)
Fig. 6. The number of forest-dependent species in relation to patch area. For patch abbreviations see Table 1.

(1823 versus 718) and grassland species (1522 versus 312) than Malaysian sites, whereas they had fewer forest species (2638 versus 25200). The resident bird species composition of young and old secondary forest patches in Singapore was also completely distinct from that of the Malaysian logged forest patches (Fig. 8). Within the Malaysian group, the two older-regrowth forests (ST and UK, 134161 species) were separated from, and richer than, the four recently logged patches of Sungei Tekam (78119 species). Within the Singaporean group, the four largest patches of old secondary forest clustered independently from the two smaller patches of old forest (BG and LA) and from the four patches of young secondary forests. Resident bird communities in logged sites were richer in species than any of the Singaporean patches. Malaysian logged sites still contained a higher proportion of true forest species (98124) than Singaporean forests (2338). Singaporean patches included more plantation (1821 versus 47) and grassland (1518 versus 13) species than Malaysian logged forests.

5. Discussion 5.1. Forest degradation and bird community The resident species were numerous and more densely populated in young secondary forest and plantations owing to the large number of plantation and grassland species. Yorke (1984) found that the species richness in a Malaysian rubber tree plantation with young secondary growth was very low, even in birds associated with secondary growth. Conversely, in Singapore, a number of forest species were still present in such plantations

(e.g., banded woodpecker Picus miniaceus and greater racquet-tailed drongo Dicrurus paradiseus) suggesting that island birds may be less sensitive to forest degradation probably because of an insular niche expansion (e.g., Blake and Karr, 1984). Old secondary forests still indicated their high quality by having higher densities of 18 of the 20 resident forest species than was found in other habitat types. Furthermore, older forests had fewer introduced species. The number of introduced species was inversely correlated with the percentage of canopy cover suggesting that such factor can limit their distribution. Migrants, on the other hand, had similar overall densities in all three habitats. Many studies have shown that wintering migrants in the tropics mostly used secondary forests or plantations and rarely primary forests (Bell, 1982; Brosset, 1968, 1990; Robbins et al., 1992; Wallace et al., 1996). However, Malaysia is an exception because of the concentration of large numbers of migrants into a narrow peninsula that was mostly covered by primary forest until recently (Wells, 1990). More than a third of all birds recorded were migrants, either wintering in Singapore for up to several months or just stopping en route to and from more southerly wintering quarters. They occurred in all months but with a distinct peak from October to April, and mostly on their southward (OctoberNovember) and northward (March) passages. Thus, their presence scarcely overlapped with the peak breeding period (AprilJuly) of sedentary species. 5.2. Species distribution among forest patches As suggested by other studies (Blake, 1991; Cornelius et al., 2000), patch area was the major variable

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

145

Fig. 7. Percentage similarity of total forest bird communities among all 17 Singaporean sites (see Table 1 for site details), and comparisons with primary forests on Tioman Island and three Malaysian mainland sites. Woodland types: old secondary forest (OSF), plantation (P), primary forest (PF), primary/secondary forest (PSF), and young secondary forests (YSF). Total bird species are given for each site. NE Taman Negara National Park (primary forest; 430,000 ha), PA Pasoh Forest Reserve (primary forest surrounded by selectively logged forest; 600 ha), RO Endau-Rompin (primary/secondary forest; 80,000 ha), and TI Tioman Island (primary/secondary forest; 13,700 ha).

determining the number of all bird species pooled in patches of secondary forest as well as of resident and migrant species separately. Interpatch distance (isolation) and the percentage of shrub cover (vegetation structure) were less signicant predictors of bird richness. The presence of three forest-dependent species in BB suggests that its nearby old forest patches (e.g.,

BT) might have served as a source area. Three older forests (BT, NS, and MR) are within 2 km of each other and collectively contained all 20 true forest species (Fig. 5). These patches were connected until early 1990s, which suggests that historic connectivity might have played a role in the persistence of some forest species here. However, data are needed to

146

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

Fig. 8. Percentage similarity of resident bird communities of 10 Singaporean sites (see Table 1 for site details and Fig. 7 for abbreviations of woodland types), and comparisons with 6 Malaysian sites. Woodland types: logged forest (LF) and secondary forest (SF). Total bird species are given for each site. ST Sungei Tekam forestry concession (all sites, area unspecied) secondary forest, STL Sungei Tekam forestry concession recently logged forest, ST2 Sungei Tekam forestry concession 2-years old logged forest, ST4 Sungei Tekam forestry concession 4-years old logged forest, ST6 Sungei Tekam forestry concession 6-years old logged forest, and UK Universiti Kebangsaan Malaysia (secondary forest left regenerating for 2550 years; 830 ha).

determine if the forest species are adequately reproducing in these patches. 5.3. Comparisons between Singapore and Malaysia The two largest Malaysian primary forest sites held 198240 species (compared to the 170 initially present in Singapore; Castelletta et al., 2000), quickly dropping to 120 in the smaller isolated patch of Pasoh forest, which is comparable in size to the largest Singaporean forests, and to 45 on Tioman Island, which is nearly four times smaller than the Singapore Island. By comparison, the three largest patches of oldest

forest on Singapore harbour 5164 species. This roughly gives the scale of species loss that may be due rst to island eect (isolation and size of the land mass), then to forest fragmentation (isolation and size of the forest patch) and ultimately to forest degradation (changes in forest structure due to deforestation, disturbance and young age of stand). It is dicult to disentangle the relative importance of each factor because of the relationship between forest area, degree of patch isolation and habitat quality. To do so, one should strictly compare pair of patches that should dier by only one of these three parameters, an unlikely situation in practice.

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

147

Insularisation alone could account for a decrease of >80% in species richness, e.g., between Taman Negara on the mainland and the small Tioman Island. Within Malaysia, the decrease in the primary forest patch area from >100,000 ha (Taman Negara) to <1000 ha (Pasoh) accounts for a 50% decline of species number. Within Singapore, the number of forest species peaked in the second largest site MR (Fig. 6). It is unclear why the largest forest (NS) did not have the highest number of forest species except that this site had no remnants of primary forest, unlike MR. Furthermore, NS had 15 ha area of freshwater swamp forest. Such habitat plus dense and thorny undergrowth of palms and Pandanus may have deterred some forest species. When compared to disturbed or logged forests on the mainland, the Singaporean patches were depauperate. Unlike the Singaporean patches, mainland sites were sometimes surrounded or fringed by primary forests. Their bird community composition, albeit altered, still included a large proportion of forest species according to our denitions. The maintenance of such diverse mainland communities probably depends on the proximity of more pristine forest sources for forest birds, which lessens the impact of forest degradation. The presence of fewer forest species in younger Singaporean secondary forest may be the result of unsuitable habitat for certain species adapted to taller forest and/or of a poor ability to recolonise isolated patches by crossing the urbanised matrix.

5.4. Conservation implications The maintenance of large reserves with mature forests is critical for the conservation of a high avian richness and to curb species loss. In a highly urbanised area such as Singapore, large patches are the only refuges that allow the persistence of sensitive forest species. MR, NS and BT constitute the largest, albeit fragmented, refuge for the remaining, already depauperate Singaporean forest avifauna. Conservation policies should be reinforced: (1) to securely protect the above three patches, (2) to allow forest regeneration with minimal human disturbance to improve habitat quality for forest birds, and (3) to connect the three patches to facilitate movements between recently (early 1990s) isolated sub-populations of some forest species (Fig. 1). Isolation is known to have a negative eect on the forest avifauna (Van Dorp and Opdam, 1987; Bierregaard et al., 1992; Schmiegelow et al., 1997). A distance as short as 100 m may be a barrier that many tropical forest birds are highly reluctant, if not unable, to cross (Stouer and Bierregaard, 1995). Relatively close forest patches may therefore constitute conned habitats for some forest species, and corridors although their applicability is still debated

may be a good option to facilitate the dispersal of such isolated populations and allow them to function as a metapopulation (Simberlo et al., 1992; Beier and Noss, 1998). Corridors can be expensive and may also facilitate the spread of predators and parasites, but isolation of tiny populations of some forest birds may hinder their long-term viability. With this end in view, the park-connector network initiated in 1990s by the National Park Board of Singapore (Sodhi et al., 1999) might well be extended to link isolated fragments (see Fig. 1). To avoid the extinction of already rare and patchily distributed species, and to reinforce their threatened tiny populations, their reintroduction into some of the nonoccupied patches could be envisioned. However, reintroductions should only be attempted if ecological conditions are conducive (e.g., enough habitat is available for the target species) and after careful trials plus monitoring. The secondary forests of Singapore are slowly becoming more mature is an encouraging sign for reintroduction eorts. The fact that escaped individuals of previously extirpated black-headed bulbul (Pycnonotus melanicterus) have become established in the BT, MR and NS (Lim and Gardner, 1997) is an encouraging sign for envisioning reintroductions of some species. Overall, our study shows that in a heavily deforested landscape, larger and older forests serve as better refuges for remnant forest bird richness. Nevertheless, a 58 ha forest patch (LF) contained 12 of 20 forest species. This suggests that all of mature and maturing forest patches in Singapore should be conserved regardless of their size. Further, smaller patches may well be important for the conservation of other taxa (Turner and Corlett, 1996).

Acknowledgements This study was supported by the National University of Singapore (RP960316) and Cheng-Kim Loke Foundation. We thank the National Parks Board for providing the permits and for contributing to the GIS analysis. We sincerely thank Gerard Lacroix for his valuable comments on the statistical analysis. We are grateful to R. Subaraj who censused the bird community, to T. Bird, M. Leong, L. Ong, K. Peh, M. Soh, T. Tan who assisted us in the eld, and to T. R. Shankar Raman, B. N. K. Davis and an anonymous reviewer for valuable comments.

Appendix A See Table 3 overleaf.

Table 3 Composition of the bird communities of the study patches in Singapore Species St Old secondary forest BT Red junglefowl (Gallus gallus) Sunda pygmy woodpecker (Dendrocopos moluccensis) Rufous woodpecker (Micrapternus brachyurus) White-bellied woodpecker (D. javensis) Banded woodpecker (P. miniaceus) Laced woodpecker (Picus vittatus) Common ameback (Dinopium javanense) Lineated barbet (Megalaima lineata) Red-crowned barbet (Megalaima raesii) Coppersmith barbet (Megalaima haemacephala) Oriental pied hornbill (Anthracoceros albirostris) Great hornbill (Buceros bicornis) Dollarbird (Eurystomus orientalis) Common kingsher (Alcedo atthis) Blue-eared kingsher (Alcedo meninting) Oriental dwarf kingsher (Ceyx erithacus) Stork-billed kingsher (Halcyon capensis) White-throated kingsher (Halcyon smyrnensis) Black-capped kingsher (Halcyon pileata) Collared kingsher (Todiramphus chloris) Blue-throated bee-eater (M. viridis) Blue-tailed bee-eater (Merops philippinus) Chestnut-winged cuckoo (Clamator coromandus) I R R R R R R I R R M I R M R M R R M R R M M X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X NS MR BG Young secondary forest LF LA X X X X MF TB KR X X X BB Plantation PO X X X HO UP X X BW X X KB LO X X X YI X X

148 M. Castelletta et al. / Biological Conservation 121 (2005) 135155

X X

X X X

Indian cuckoo (Cuculus micropterus) Banded bay cuckoo (Cacomantis sonneratii) Plaintive cuckoo (Cacomantis merulinus) Rusty-breasted cuckoo (Cacomantis sepulcralis) Little bronze cuckoo (Chrysococcyx minutillus) Violet cuckoo (C. xanthorhynchus) Drongo cuckoo (S. lugubris) Asian koel (E. scolopacea) Chestnut-bellied malkoha (Phaenicophaeus sumatranus) Greater coucal (Centropus sinensis) Lesser coucal (Centropus bengalensis) Red lory (Eos bornea) Rainbow lorikeet (Trichoglossus haematodus) Yellow-crested cockatoo (C. sulphurea) White cockatoo (Cacatua alba) Tanimbar corella (Cacatua goni) Blue-rumped parrot (Psittinus cyanurus) Blue-crowned hanging parrot (Loriculus galgulus) Rose-ringed parakeet (Psittacula krameri) Blossom-headed parakeet (Psittacula roseata) Red-breasted parakeet (Psittacula alexandri) Long-tailed parakeet (Psittacula longicauda) Himalayan swiet (Collocalia brevirostris) Silver-backed needletail (Hirundapus cochinchinensis) Brown-backed needletail (Hirundapus giganteus) Asian palm swift (Cypsiurus balasiensis)

M R R R R R R R R R R I I I I I R R I I I R M M M R X X X X X X X X X X X X X X X X X X X X X X X

X X X

X X

X X X

X X

X X X X X X X X X X X X X X X X X X X X

X X X

X X X

X X X X X

X M. Castelletta et al. / Biological Conservation 121 (2005) 135155 X X X X X X X X X X

X X

X X

X X

X X

X X

X X

X X X X

X X X X X X X

X X

X X

X X X X X X

X X X X

X X X X

X X X X X X

X X X X

X X

X X X

X X

X X X

X X X X

X X X X 149

150

Table 3 (continued) Species St Old secondary forest BT Fork-tailed swift (Apus pacicus) House swift (Apus anis) Grey-rumped treeswift (Hemiprocne longipennis) Rock pigeon (Columba livia) Spotted dove (Steptotelia chinensis) Red collared dove (Streptopelia tranquebarica) Emerald dove (Chacophaps indica) Peaceful dove (Geopelia striata) Pink-necked green pigeon (Treron vernans) Thick-billed green pigeon (Treron curvirostra) Jambu fruit-dove (Ptilinopus jambu) Pied imperial pigeon (Ducala bicolor) Osprey (Pandion haliaetus) Black baza (Aviceda leuphotes) Oriental honey-buzzard (Pernis ptilorhyncus) Black-shouldered kite (Elanus caeruleus) Brahminy kite (Haliastur indus) White-bellied sea eagle (Haliaeetus leucogaster) Grey-headed sh eagle (Ichthyophaga ichthyaetus) Crested serpent eagle (Spilornis cheela) Shikra (Accipiter badius) Chinese sparrowhawk (Accipiter soloensis) Japanese sparrowhawk (Accipiter gularis) Besra (Accipiter virgatus) Common buzzard (Buteo buteo) M R R I R I R R R R M M M M M R R R R R M M M M M X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X NS X X X X X X X X X X X X X X X X MR BG Young secondary forest LF LA MF TB KR X X X X X X X X X BB Plantation PO X X X X X X X X X HO UP BW KB X X LO X X X YI

X X

X X

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

X X

X X

X X

X X

X X

X X

X X

X X X X

X X X

X X

X X

Changeable hawk eagle (Spizaetus cirrhatus) Common kestrel (Falco tinnunculus) Peregrine falcon (Falco peregrinus) Hooded pitta (Pitta sordida) Blue-winged pitta (Pitta moluccensis) Golden-bellied gerygone (Gerygone sulphurea) Asian fairy bluebird (Irena puella) Greater green leafbird (Chloropsis sonnerati) Lesser green leafbird (Chloropsis cyanopogon) Blue-winged leafbird (Chloropsis cochinchinensis) Golden-fronted leafbird (Chloropsis aurifrons) Tiger shrike (Lanius tigrinus) Brown shrike (Lanius cristatus) Red-billed blue magpie (Urocissa erythrorhyncha) House crow (Corvus splendens) Large-billed crow (Corvus macrorhynchos) Black-naped oriole (O. chinensis) Pied triller (Lalage nigra) Ashy minivet (Pericrocotus divaricatus) Scarlet minivet (P. ammeus) Pied fantail (Rhipidura javanica) Black drongo (Dicrurus macrocercus) Ashy drongo (Dicrurus leucophaeus) Crow-billed drongo (Dicrurus annectans) Greater racket-tailed drongo (D. paradiseus) Asian paradise-ycatcher (Terpsiphone paradisi)

R M M M M R R R R R I M M I I R R R M R R M M M R M

X X X

X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X 151 X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X M. Castelletta et al. / Biological Conservation 121 (2005) 135155 X X X X X X

152

Table 3 (continued) Species St Old secondary forest BT Common iora (Aegithina tiphia) Orange-headed thrush (Zoothera citrina) Eyebrowed thrush (Turdus obscurus) Brown-chested jungle ycatcher (Rhinomyias brunneata) Dark-sided ycatcher (Muscicapa sibirica) Asian brown ycatcher (Muscicapa dauurica) Yellow-rumped ycatcher (Ficedula zanthopygia) Mugimaki ycatcher (Ficedula mugimaki) Siberian blue robin (Luscinia cyane) Oriental magpie robin (Copsychus saularis) White-rumped shama (Copsychus malabaricus) Asian glossy starling (Aplonis panayensis) Purple-backed starling (Sturnus sturninus) White-shouldered starling (Sturnus sinensis) Common myna (A. tritis) White-vented myna (Acridotheres cinereus) Hill myna (Gracula religiosa) Velvet-fronted nuthatch (S. frontalis) Barn swallow (Hirundo rustica) Pacic swallow (Hirundo tahitica) Red-rumped swallow (Hirundo daurica) Straw-headed bulbul (Pycnonotus zeylanicus) Black-crested bulbul (Pycnonotus melanicterus) R M M M X X X X X X X X X X X X X NS X MR X BG X Young secondary forest LF X LA X MF X TB X KR X BB X Plantation PO X HO X UP X BW X KB X LO X YI X

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

M M M M M R R R M M R I R V M R M R I X X X

X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X

X X X X X X X X X X X X X X

X X X X X X X X

X X

X X

X X

X X

X X

X X

X X X

X X

X X

X X X

X X

X X

X X

X X

X X

X X

X X

X X X

X X

X X X

X X

X X

X X X

X X

X X X

Red-whiskered bulbul (Pycnonotus jocosus) Yellow-vented bulbul (Pycnonotus goiavier) Olive-winged bulbul (Pycnonotus plumosus) Cream-vented bulbul (Pycnonotus simplex) Red-eyed bulbul (Pycnonotus brunneus) Ashy bulbul (Hemixos avala) Yellow-bellied prinia (Prinia aviventris) Oriental white-eye (Zosterops palpebrosus) Japanese white-eye (Zosterops japonicus) Lanceolated warbler (Locustella lanceolata) Rusty-rumped warbler (Locustella certhiola) Black-browed reed warbler (Acrocephalus bistrigiceps) Oriental reed warbler (Acrocephalus orientalis) Common tailorbird (Orthotomus sutorius) Dark-necked tailorbird (Orthotomus atrogularis) Rufous-tailed tailorbird (Orthotomus sericeus) Ashy tailorbird (Orthotomus ruceps) Arctic warbler (Phyllooscopus borealis) Eastern crowned warbler (Phylloscopus coronatus) White-crested laughingthrush (G. leucolophus) Greater necklaced laughingthrush (Garrulax pectoralis) Black-throated laughingthrush (Garrulax chinensis) Hwamei (Garrulax canorus) White-chested babbler (Trichastoma rostratum) Abbotts babbler (Malacocincla abbotti)

I R R R R M R I I M M M M R R R R M M I I I I R R X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X

X X X

X X X

X X X

X X X X X

X X X X X X X

X X X X X X X

X X X

X X X X X

X X X X X X X M. Castelletta et al. / Biological Conservation 121 (2005) 135155

X X X X

X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X

X 153

154

Table 3 (continued) Species St Old secondary forest BT Short-tailed babbler (Malacocincla malaccensis) Chestnut-winged babbler (Stachyris erythroptera) Striped tit-babbler (Macronous gularis) Orange-bellied owerpecker (Dicaeum trigonostigma) Scarlet-backed owerpecker (Dicaeum cruentatum) Brown-throated sunbird (Anthreptes malacensis) Purple-throated sunbird (Nectarinia sperata) Olive-backed sunbird (Nectarinia jugularis) Crimson sunbird (Aethopyga siparaja) Little spiderhunter (Arachnothera longirostra) Eurasian tree sparrow (Passer montanus) Forest wagtail (Dendronanthus indicus) Yellow wagtail (Motacilla ava) Grey wagtail (Motacilla cinerea) Paddyeld pipit (Anthus rufulus) Baya weaver (Ploceus philippinus) Javan munia (Lonchura leucogastroides) Scaly-breasted munia (Lonchura punctulata) White-headed munia (Lonchura maja) Total species number R R R R R R R R R R R M M M R R I R R 84 83 98 49 62 49 80 X X X X X X X 68 79 63 X X X X X X X 97 81 78 67 63 78 56 X X X X X X X X X X X X X X X X X X NS X X X X X X X X X X X X X X X X X X X X X X X X MR X X X X X X X X X X X BG Young secondary forest LF X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X M. Castelletta et al. / Biological Conservation 121 (2005) 135155 X X X X X X X X X X X X X X X X X X X X LA MF TB KR BB Plantation PO HO UP BW KB LO YI

Species status (St), with resident species (R), migrant species (M), intoduced species (I), and vagrant species (V). For description of habitat type, see Methods, and for patch codings, see Table 1. Nomenclature follows Inskipp et al. (1996).

M. Castelletta et al. / Biological Conservation 121 (2005) 135155

155

References
Achard, F., Eva, H.D., Stibig, H.-J., Mayaux, P., Richards, T., Malingreau, J.-P., 2002. Determination of deforestation rates of the Worlds humid tropical forests. Science 297, 9991002. Beier, P., Noss, R.F., 1998. Do habitat corridors provide connectivity? Conservation Biology 6, 12411252. Bell, H.L., 1982. A bird community in lowland rain forest in New Guinea. 1. Composition and density of the avifauna. Emu 82, 2441. Bierregaard Jr., R.O., Lovejoy, T.E., Kapos, V., Dos Santos, A.A., Hutchings, R.W., 1992. The biological dynamics of tropical rainforest fragments. BioScience 42, 859866. Blake, J.G., 1991. Nested subsets and the distribution of birds on isolated woodlots. Conservation Biology 5, 5866. Blake, J.G., Karr, J.R., 1984. Species composition of bird communities and the conservation benet of large versus small forests. Biological Conservation 30, 173187. Bransburry, J., 1993. A Birdwatchers Guide to Malaysia. Waymark Publishing, UK. Brash, A.R., 1987. The history of avian extinction and forest conversion on Puerto Rico. Biological Conservation 39, 97111. Brosset, A., 1968. Localisation cologique des oiseaux migrateurs dans e la for^t quatoriale du Gabon. Biologia Gabonica 4, 211216. e e Brosset, A., 1990. A long-term study of the rain forest birds in MPassa, Gabon. In: Keast, A. (Ed.), Biogeography and Ecology of Forest Bird Communities. SPB Academic Publisher, The Hague, pp. 259274. Castelletta, M., Sodhi, N.S., Subaraj, R., 2000. Heavy extinctions of forest avifauna in Singapore: lessons for biodiversity conservation in Southeast Asia. Conservation Biology 14, 18701880. Corlett, R.T., 1992. The ecological transformation of Singapore, 1819 1990. Journal of Biogeography 19, 411420. Corlett, R.T., 1997. The vegetation in the nature reserves of Singapore. Gardens Bulletin, Singapore 49, 147159. Cornelius, C., Cofr, H., Marquet, P.A., 2000. Eects of habitat e fragmentation on bird species in a relic temperate forest in semiarid Chile. Conservation Biology 14, 534543. Diamond, J.M., Bishop, K.D., Van Balen, S., 1987. Bird survival in an isolated Javan woodland: island or mirror? Conservation Biology 1, 132142. Ford, H.A., Davison, G.W.H., 1995. Forest avifauna of universiti Kebangsaan Malaysia and some other forest remnants in Selangor, Peninsular Malaysia. Malayan Nature Journal 49, 117138. James, F.C., Shugart Jr., H.H., 1970. A quantitative method of habitat description. Audubon Field Notes 24, 727737. Jeyarajasingam, A., Pearson, A., 1999. A Field Guide to the Birds of West Malaysia and Singapore. Oxford University Press, Oxford. Johns, A.D., 1986. Eects of selective logging on the ecological organization of a peninsular Malaysian rainforest avifauna. Forktail 1, 6579. Karim-Dakog, A., Shahrim-Senik, M., Zubaid, A., Davison, G.W.H., 1997. Understorey avifauna of the UKM forest reserve, a fragmented and disturbed lowland rainforest. Malayan Nature Journal 50, 337346. Karr, J.R., 1982. Avian extinction on Barro Colorado island, Panama: a reassessment. American Naturalist 119, 220239. Laurance, W.F., 1999. Reections on the tropical deforestation crisis. Biological Conservation 91, 109117. Leck, C.F., 1979. Avian extinctions in an isolated tropical wet-forest preserve, Ecuador. Auk 96, 343352. Lemmon, F.R., 1957. A new instrument for measuring forest overstorey density. Journal of Forestry 55, 667668. Lim, K.S., Gardner, D., 1997. Birds: An Illustrated Field Guide to the Birds of Singapore. Sun Tree Publishing Limited, Singapore. MacCune, B., Meord, M.J., 1995. PC-ORD. Multivariate Analysis of Ecological Data, version 2.0. MjM Software design, Gleneden, US. Newmark, W.D., 1991. Tropical forest fragmentation and the local extinction of understory birds in the eastern Usambara mountains, Tanzania. Conservation Biology 5, 6778.

Opdam, P., Rijsdijk, G., Hustings, F., 1985. Bird communities in small woods in an agricultural landscape: eects of area and isolation. Biological Conservation 34, 333352. Patton, D.R., 1975. A diversity index for quantifying habitat edges. Wildlife Society Bulletin 3, 171173. Pearson, D.L., 1975. The relation of foliage complexity to ecological diversity of three Amazonian bird communities. Condor 77, 453466. Ralph, C.J., Geupel, G.R., Pyle, P., Martin, T.E., DeSante, D.F., 1993. Handbook of Field Methods for Monitoring Landbirds. Pacic Southwest research station, US. Robbins, C.S., Dowell, B.A., Dawson, D.K., Colon, J.A., Estrada, R., Sutton, A., Sutton, R., Weyer, D., 1992. Comparison of Neotropical migrant landbird populations in tropical forest, isolated fragments and agricultural habitats. In: Hagan III, J.M., Johnson, D.W. (Eds.), Ecology and Conservation of Neotropical Migrant Landbirds. Smithsonian Institution Press, Washington and London, pp. 207220. Rotenberry, J.T., 1985. The role of habitat in avian community composition: physiognomy or oristics? Oecologia 67, 213217. Schmiegelow, F.K.A., Machtan, C.S., Hannon, S.J., 1997. Are boreal birds resilient to forest fragmentation? An experimental study of short-term community responses. Ecology 78, 19141932. Simberlo, D., Farr, J.A., Cox, J., Mehlman, D.W., 1992. Movement corridors: conservation bargains or poor investments? Conservation Biology 6, 493504. Sodhi, N.S., 2002. A comparison of bird communities of two fragmented and two continuous Southeast Asian rainforests. Biodiversity and Conservation 11, 11051119. Sodhi, N.S., Briett, C., Kong, L., Yeun, B., 1999. Bird use of linear areas of a tropical city: implications for park connector design and management. Landscape and Urban Planning 45, 123130. Sokal, R.R., Rohlf, F.J., 1981. Biometry. W.H. Freeman and Company, New York. Stouer, P.C., Bierregaard Jr., R.O., 1995. Use of Amazonian forest fragments by understory insectivorous birds. Ecology 76, 24292445. SuperANOVA., 1989. Abacus Concepts. Inc., Berkeley, CA. Terborgh, J., Robinson, S.K., Parker III, T.A., Munn, C.A., Pierpont, N., 1990. Structure and organization of an Amazonian forest bird community. Ecological Monographs 60, 213238. Thiollay, J.-M., 1994. Structure, density and rarity in an Amazonian rainforest bird community. Journal of Tropical Ecology 10, 449 481. Turner, I.M., Corlett, R.T., 1996. The conservation value of small, isolated fragments of lowland tropical rainforest. Trends in Ecology and Evolution 11, 330333. Turner, I.M., Wong, Y.K., Chew, P.T., Ibrahim, A.B., 1997. Tree species richness in primary and old secondary tropical forest in Singapore. Biodiversity and Conservation 6, 537543. Van Dorp, D., Opdam, P.F.M., 1987. Eects of patch size, isolation and regional abundance on forest bird communities. Landscape Ecology 1, 5973. Wallace, G.E., Alonso, H.G., McNicholl, M.K., Batista, D.R., Prieto, R.O., Sosa, A.L., Oria, B.S., Wallace, E.A.H., 1996. Winter surveys of forest-dwelling neotropical migrant and resident birds in three regions of Cuba. Condor 98, 745768. Ward, P., 1968. Origin of the avifauna of urban and suburban Singapore. Ibis 110, 239255. Wells, D.R., 1990. Migratory birds and tropical forests in the Sunda region. In: Keast, A. (Ed.), Biogeography and Ecology of Forest Bird Communities. SPB Academic Publisher, The Hague, pp. 357370. Wells, D.R., 1999. The Birds of the Thai-Malay Peninsula. Academic Press, London, UK. Yorke, C.D., 1984. Avian community structure in two modied Malaysian habitats. Biological Conservation 29, 345362. Zar, J.H., 1996. Biostatistical Analysis. Prentice-Hall,, Englewood Clis, NJ.