Ant Communities on Small Tropical Islands in Indonesia

Journal of Biogeography (J. Biogeogr.) (2010) 37, 229–236

ORIGINAL Ant communities on small tropical ARTICLE islands: effects of island size and isolation are obscured by habitat disturbance and ‘tramp’ ant species Akhmad Rizali1,2*, David J. Lohman3 , Damayanti Buchori1, Lilik Budi Prasetyo4, Hermanu Triwidodo1, Merijn M. Bos5, Seiki Yamane6 and Christian H. Schulze7

1Department of Plant Protection, Faculty of ABSTRACT Agriculture, Bogor Agricultural University, Aim Comparisons among islands offer an opportunity to study the effects of Kampus IPB Darmaga, Bogor, West Java, Indonesia, 2PEKA Indonesia Foundation biotic and abiotic factors on small, replicated biological communities. Smaller (Indonesian Nature for Conservation population sizes on islands accelerate some ecological processes, which may Foundation), Komplek IPB Sindang Barang II, decrease the time needed for perturbations to affect community composition. Bogor, Indonesia, 3Department of Biological We surveyed ants on 18 small tropical islands to determine the effects of island Sciences, National University of Singapore, size, isolation from the mainland, and habitat disturbance on ant community Republic of Singapore, 4Department of Forest composition. Resources Conservation and Ecotourism, Location Thousand Islands Archipelago (Indonesian name: Kepulauan Seribu) Bogor Agricultural University, Kampus IPB off Jakarta, West Java, Indonesia. Darmaga, Bogor, West Java, Indonesia, 5Louis Bolk Institute, Hoofdstraat, Driebergen, Methods Ants were sampled from the soil surface, leaf litter and vegetation in all The Netherlands, 6Department of Earth and habitat types on each island. Island size, isolation from the mainland, and land- Environmental Sciences, Faculty of Science, use patterns were quantiﬁed using GIS software. The presence of settlements and Kagoshima University, Korimoto 1-chome, of boat docks were used as indicators of anthropogenic disturbance. The richness 7 Kagoshima, Japan and Department of of ant communities and non-tramp ant species on each island were analysed Population Ecology, Faculty of Life Sciences, in relation to the islands’ physical characteristics and indicators of human University of Vienna, Austria disturbance. Results Forty-eight ant species from 5 subfamilies and 28 genera were recorded from the archipelago, and approximately 20% of the ant species were well-known human-commensal ‘tramp’ species. Islands with boat docks or human settlements had signiﬁcantly more tramp species than did islands lacking these indicators of anthropogenic disturbance, and the diversity of non-tramp species decreased with habitat disturbance. Main conclusions Human disturbance on islands in the Thousand Islands Archipelago promotes the introduction and/or establishment of tramp species. Tramp species affect the composition of insular ant communities, and expected biogeographical patterns of ant richness are masked. The island with the greatest *Correspondence: Akhmad Rizali, Department estimated species richness and the greatest number of unique ant species, Rambut of Agroecology, University of Go¨ttingen, Island, is a forested bird sanctuary, highlighting the importance of protected areas Waldweg 26, D-37073 Go¨ttingen, Germany. E-mail: [email protected] in preserving the diversity of species-rich invertebrate faunas. Present address: Department of Biology, The Keywords City College of New York, The City University of New York, Convent Avenue at 138th Street, Ant fauna, conservation, Formicidae, Indonesia, invasive species, island New York, NY 10031, USA. biogeography, land use, myrmecology, Southeast Asia.

INTRODUCTION remain inﬂuential more than thirty years after its publica- The Theory of Island Biogeography by MacArthur & Wilson tion (e.g. Hubbell, 2001). Although metapopulation theory (1967) was a milestone in ecology, and tenets of the theory has largely replaced island biogeography as a mechanistic

ª 2009 Blackwell Publishing Ltd www.blackwellpublishing.com/jbi 229 doi:10.1111/j.1365-2699.2009.02194.x A. Rizali et al. framework for describing the dynamics of discrete popula- (forest, settlement or a combination of both; Table 1), and the tions interconnected by occasional migration (Hanski & presence of boat docks were quantiﬁed using GIS software. Gilpin, 1997), island biogeography remains a valuable heuristic tool for conceptualizing the factors that inﬂuence Ant collecting differences among island biotas. However, the impacts of human disturbance on the natural patterns predicted by Ants were collected from March to May 2005 (Table 1). On island biogeography theory remain poorly understood. each island, ants were collected in 5 · 5 m plots. Disturbed Accelerating rates of deforestation and human encroachment habitats, secondary forest and primary forest were sampled on into tropical forests are causing unprecedented rates of biodi- each island, and plots were randomly selected within habitat versity loss, particularly in Southeast Asia (Sodhi & Brook, types. Disturbed habitats included gardens, areas around 2006). Most studies on the decline of animal populations and human habitations, and other areas disturbed by humans. species focus on ‘charismatic megafauna’, such as birds and The number of plots per island was increased until no mammals, but insects and other arthropods, which constitute additional new species were collected, and ranged from 7 to 20 the bulk of animal species, individuals and biomass in tropical plots per island. Ant sampling in each plot lasted a maximum forests (Ellwood & Foster, 2004), have received less attention. of 30 min. Ants were sampled using intensive hand collection Social insects, including ants and termites, are the dominant in three ant-habitat strata within each plot: leaf litter, soil group of terrestrial animals in the tropics (Ho¨lldobler & Wilson, surface, and vegetation up to 2 m high. Ants were stored in 1990), and contribute to healthy ecosystem functioning as predators, scavengers, herbivores, detritivores and granivores, while aiding bioturbation of the soil (Agosti et al., 2002). Ants are ecologically diverse, with different species adapted to particular nest sites, food sources, microhabitats and/or temperature regimes (Andersen, 2000). Habitat disturbance may impact any or all of these factors, thereby decreasing ant species richness and altering the composition of ant communities (Schoereder et al., 2004). Human disturbance and commerce frequently introduce human-commensal species, which, among ants, are frequently called ‘tramp’ species. Tramp ants often have colonizing mechanisms well-adapted to cohabitation with humans (Graham et al., 2004; Schoereder et al., 2004), and they frequently exclude native ant species and decrease species richness (Suarez et al., 1998; Andersen, 2000; Holway et al., 2002; Hill et al., 2003; Bos et al., 2008). The present study examines ant communities on small tropical islands in Indonesia. Each focal island can be viewed as a more or less independent community replicate connected to the others by natural dispersal of sexual forms (alates), and, perhaps more commonly, by transportation on human vessels. By choosing islands that vary in key characteristics related to the diversity of ant species assemblages, we assess the relative impacts of natural and anthropogenic factors on 18 discrete terrestrial communities. We test the hypotheses that human activities facilitate the colonization of islands by tramp ant species, and that human activities disturb the natural faunistic patterns predicted by island biogeography theory.

MATERIALS AND METHODS

Research sites

This research was conducted on 18 islands in the Thousand Islands Archipelago (Indonesian name: Kepulauan Seribu; 10620¢–10650¢ E and 520¢–600¢ S; Fig. 1), a group of several hundred islands near Java; all of the islands have an area of < 1 km2 (Alamsyah, 2003). The island characteristics of Figure 1 Map of study sites in the Thousand Islands Archipelago island size, isolation from the mainland, dominant land cover (Kepulauan Seribu) off Jakarta, Indonesia.

230 Journal of Biogeography 37, 229–236 ª 2009 Blackwell Publishing Ltd Ant communities on small disturbed tropical islands

Table 1 Observed (Obs.) and estimated (Est.) ant species richness and island characteristics of the 18 studied islands of the Thousand Islands archipelago off Jakarta, West Java, Indonesia.

Total non-tramp Total species species

Isolation Area Land No. Dates of Sampled Sampled No. Island (km)* (ha) use Docksà plots sampling subfamilies genera Obs. Est. (%)§ Obs. Est. (%)§

1 Onrust 2.2 8.23 S P 7 8.V.2005 3 10 14 20.89 (67.02) 8 13.74 (58.22) 2 Rambut 4.2 45.80 F P 11 9–10.V.2005 5 21 34 49.37 (68.87) 25 37.87 (66.02) 3 Untung Jawa 4.8 39.12 FS P 17 10–12.V.2005 4 21 29 37.42 (77.50) 20 27.22 (73.48) 4 Bokor 7.0 16.34 F P 10 5.V.2005 4 20 33 36.51 (87.65) 24 26.55 (86.63) 5 Lancang Besar 9.4 26.43 S P 15 6–7.V.2005 4 18 28 31.06 (86.93) 19 20.58 (87.46) 6 Pari 16.1 52.87 FS P 20 1,2,4.V.2005 5 24 35 37.18 (91.45) 26 28.59 (87.44) 7 Payung Besar 20.8 22.74 FS P 16 6–8.IV.2005 4 18 25 24.82 (96.70) 17 17.09 (93.62) 8 Tidung kecil 22.8 19.71 FS P 11 15–16.IV.2005 5 21 27 31.89 (81.53) 19 23.79 (75.66) 9 Pramuka 28.6 19.92 FS P 17 26,29.IV.2005 5 20 31 36.86 (81.39) 22 27.78 (75.59) 10 Semak Daun 31.2 1.00 F A 8 12.III.2005 4 13 18 21.43 (79.33) 12 15.67 (70.20) 11 Kotok Besar 34.2 22.65 FS P 15 30.IV.2005 4 19 29 36.20 (80.11) 20 22.97 (87.07) 12 Paniki 35.1 5.80 F A 8 10–11.III.2005 5 20 27 30.85 (87.52) 21 23.45 (89.55) 13 Bira Kecil 43.2 8.62 FS P 8 13–14.IV.2005 5 21 28 29.11 (92.75) 20 20.71 (91.74) 14 Putri Barat 45.9 9.63 FS P 9 12.IV.2005 5 19 28 30.39 (88.85) 19 21.01 (85.67) 15 Bundar 52.6 5.76 FS P 9 28.IV.2005 4 18 26 29.12 (85.85) 19 21.77 (82.68) 16 Nyamplung 54.9 8.96 F A 8 27.IV.2005 5 19 28 34.25 (81.75) 23 27.11 (84.84) 17 Penjaliran Barat 59.6 21.65 F A 12 10.IV.2005 5 14 21 26.07 (80.55) 16 17.28 (92.59) 18 Dua Timur 62.6 21.42 F A 9 9.IV.2005 5 14 19 26.47 (71.78) 15 22.86 (65.62) Totals for all 18 islands 5 28 48 49.55 (96.87) 38 39.80 (95.48)

*Isolation: isolation from mainland Java. F, forest; S, settlement; FS, forest and settlement. àP, dock(s) present; A, dock(s) absent. §Predicted number of species from the incidence-based coverage estimator (ICE); %, sampled species as a percentage of predicted number of species. small vials with 70% alcohol and given a label in the (1994), could also be considered tramp species: Anoplolepis field before being sorted and identified in the laboratory. gracilipes, Monomorium monomorium and Solenopsis geminata Specimens were identified using relevant taxonomic (see Appendix S1). literature (e.g. Bolton, 1994) and the reference collection of Seiki Yamane (Kagoshima University, Japan), regarded as Data analysis one of the most complete collections of identified Asian ants in the world. To assess the completeness of sampling on each island, the incidence-based coverage estimator (ICE; Colwell, 1997) was calculated for each island and for the ant fauna of the Designation of tramp ant species archipelago (i.e. using data from all 18 islands). The ICE uses McGlynn (1999) provides a useful categorization of ants that species presence/absence in successive samples within a have been transferred from their native range by human location to estimate the number of unsampled species and agency. ‘Transferred’ ants refer to any ant found outside its thus the estimated total species richness. Two ICE richness native range, and is analogous to the term ‘non-native’ applied estimates were calculated for each island. First, the full dataset to many other taxa. ‘Invasive’ species are transferred species was used to calculate the total estimated number of ant species that establish long-term populations and expand their range, per island. However, in our experience, tramp species are causing ecological damage. ‘Tramp’ ants are human commen- frequently more abundant, locally widespread and conspicuous sals, some of which can also be categorized as invasive. In this than non-tramp species. Continued sampling on an island paper we use the term ‘tramp’ to refer to species so designated would rarely uncover more tramp species, whereas the by McGlynn (1999): Cardiocondyla nuda, Monomorium discovery of more non-tramp species would be more destructor, Monomorium floricola, Paratrechina longicornis, likely. Therefore, we additionally estimated the number of Tapinoma melanocephalum, Technomyrmex albipes and non-tramp species per island by excluding tramp species from Tetramorium pacificum. We also include as tramps three the dataset before calculating ICE values. ICE and species species categorized as invasive or transferred by McGlynn, but accumulation curves were calculated using EstimateS 5.0 which, in our experience and using the criteria of Passera (Colwell, 1997).

Journal of Biogeography 37, 229–236 231 ª 2009 Blackwell Publishing Ltd A. Rizali et al.

To assess the relationship between anthropogenic distur- Occurrence of tramp species bance and tramp ant species richness, the number of tramp species on islands with and without boat docks were Completeness of sampling on individual islands ranged from compared, as were the number of tramp species on islands 67.0 to 96.7% (mean ± SD = 82.6 ± 8.0%) when all ants with and without human settlements, using Wilcoxon rank were considered, and from 58.2 to 93.6% (mean ± SD = sum tests implemented with the statistical software package 80.7 ± 10.1%) when only non-tramps were considered jmp 5.1 (SAS, 2003). To assess the effects of habitat disturbance (Table 1). Tramp species were found on each island, but islands on classical biogeographical patterns of insular species rich- with docks had significantly more tramp species than did those ness, the observed number of tramp species and the estimated without (mean ± SD = 8.4 ± 1.0 vs. 5.2 ± 0.8; P = 0.04), as did ant species richness on each island were regressed against islands with human settlements compared to those without island area and against island distance from the mainland. (S + SF vs. F, mean ± SD = 8.3 ± 1.0 vs. 6.2 ± 2.0; P = 0.04; Regression lines for forested (F) and disturbed (S and FS) Table 1). The concordance between these analyses is not islands (Table 1) were plotted and analysed separately. Anal- surprising: all islands with human settlements had docks, yses of covariance (ANCOVAs) were performed to determine whereas the five islands without docks were exclusively forested the significance of differences in the regression lines between islands without settlements. Two tramp species, Paratrechina forested and disturbed islands. Regressions were performed longicornis and Tapinoma melanocephalum, and one non-tramp with jmp 5.1, and ANCOVAs were performed with statistica species, Pheidole sp. 01, were found on all 18 islands. 6.0 (StatSoft, Inc., 2001). Differences in ant community structure were quantified Effect of human disturbance on patterns of insular ant using Sørensen’s index for similarity based on presence/ species diversity absence of species, which was calculated with the Biodiv97 macro for Microsoft Excel (S. Messner, Universita¨tWu¨rzburg, Consistent with island biogeography theory, we found that ant pers. comm.). The similarity matrix was then reduced to a species richness and island area were positively correlated, two-dimensional ordination using non-metric multidimen- whereas richness and isolation from the mainland were sional scaling (NMDS) performed with statistica 6.0 (Stat- negatively correlated. However, the strength of these patterns Soft, Inc, 2001). Accompanying stress values lower than 0.20 was weaker on disturbed islands (those with settlements) than indicate that the ordination is a good representation of on forested islands. The relationship between island area and the original distance matrix values (Clarke, 1993). Effects of the estimated number of non-tramp species was positive and island characteristics on species composition were assessed marginally non-significant for islands without human settle- by one-way analyses of similarity (ANOSIMs) with 1000 ments (F = 5.63, d.f. = 1, 5, P = 0.06) and significant for random permutations calculated with Primer 5 (Clarke & islands with settlements (F = 6.31, d.f. = 1, 9, P = 0.03; Gorley, 2001). ANOSIMs test for differences in species Fig. 2a). The relationship was steeper for islands that lacked composition between predefined groups of sites (Clarke & human settlements, but the difference in slopes was not Warwick, 2001), or islands (this study). The R statistics of significant (with settlements: slope m = 0.36; without settle- ANOSIM (Legendre & Legendre, 1998) can be used as a ments: m = 0.20; ANCOVA: F = 2.20, d.f. = 1, P = 0.16). The measure of the sensitivity of species composition to changing relationship between estimated non-tramp ant species richness island variables. and island isolation was not significant for islands with (F = 0.02, d.f. = 1, 9, P = 0.88) or without (F = 2.40, d.f. = 1, 5, P = 0.18) human settlements, but was in the expected RESULTS negative direction only for islands that lacked settlements (Fig. 2c). However, the difference in slopes between the two Estimation of ant species richness island groups was again not statistically significant (with Forty-eight ant species from 5 subfamilies and 28 genera were settlements: m = )0.17; without settlements: m = 0.014; found in the Thousand Islands Archipelago (Appendix S1). ANCOVA: F = 1.17, d.f. = 1, 5, P = 0.30). These patterns The ICE estimate for all plots on all 18 sampled islands were identical when estimates of total ant species richness indicated that these sampling methods captured 96.9% of the (including tramp species) were used (Fig. 2b,d). The number archipelago’s ant species richness and 95.5% of the non-tramp of tramp species appeared to be positively related to island area ant species (Table 1). The comprehensiveness of the ant on islands with settlements (F = 4.59, d.f. = 1, 9, P = 0.06), sampling was also demonstrated by the saturation of the but not on islands that lacked settlements (F = 1.24, d.f. = 1, species accumulation curve for all islands (not shown). Ten of 5, P = 0.31). On forested islands, there was a strongly negative the total recorded species (21%) were tramp species; 16 (33%) relationship between the number of tramp species and the were described species not considered to be tramp species; island’s distance from the mainland; that is, forested islands far 11 (23%) were recognized, undescribed species [from the from the mainland had fewer tramp species than forested reference collection of Seiki Yamane (SKY)]; and 11 (23%) islands close to the mainland (F = 93.1, d.f. = 1, 5, remained unidentified and are probably undescribed, and were P = 0.0002). This relationship was not found on islands with assigned a morphospecies number (Appendix S1). settlements (F = 0.0091, d.f. = 1, 9, P = 0.92).

232 Journal of Biogeography 37, 229–236 ª 2009 Blackwell Publishing Ltd Ant communities on small disturbed tropical islands

Figure 2 Human disturbance dampens natural patterns of ant community diversity on tropical islands. The relationship between island size and ant species richness on forested islands (solid lines) appears to be more pronounced than that on islands with human settlements (dashed lines). The pattern applies to estimates of ant diversity that (a) exclude and (b) include tramp ants in calculations of ant species richness estimates (ICE values). The relationship between island distance from the mainland and ant species richness on forested islands (solid lines) appears to be more pronounced than that on islands with human settlements (dashed lines), whether (c) only non-tramp ants or (d) the total number of species is used to estimate ICE values.

A two-dimensional NMDS analysis was used to gain an differences between the island groups indicated in Figure 3c understanding of the factors that might affect species compo- (Global R = 0.22, P = 0.016). However, pairwise comparisons sition among islands (Fig. 3), and one-way ANOSIMs were revealed that ant species composition on islands ‡ 48 km from conducted to examine the effects of the following four variables Java (group D4) differed significantly from the two groups on species composition: island area, island isolation, presence/ of islands nearest the Javan coast (D1 and D2), but no absence of docks, and presence/absence of settlements. We other differences among these groupings could be detected applied a Bonferroni correction to all ANOSIM results by (Table 2). The weak effect of island area on species compo- adjusting the P-values. Islands with human settlements differed sition suggested by the NMDS ordination (Fig. 3b) was not significantly from islands dominated by forest (one-way ANO- confirmed by the corresponding ANOSIM testing for differ- SIM: Global R = 0.31, P = 0.008; Fig. 3a). Boat docks increase ences of species composition between islands larger and the accessibility of an island to humans, and their presence was smaller than 20 ha (Global R = 0.01, P = 0.45). Comparing also related to changes in ant community composition (Global the R values of all ANOSIMs indicates that the presence of boat R = 0.48, P = 0.012; Fig. 3d). Notably, islands without boat docks had the strongest effect on ant species composition, that docks had neither the yellow crazy ant, Anoplolepis gracilipes, island area and the presence of human settlements did not nor the fire ant Solenopsis geminata (Appendix S1). seem to influence community composition, and that island The isolation of an island from the mainland is thought to isolation was a marginally significant factor influencing pose a barrier to colonization by species not already present on community structure. the island. As predicted, distance from the mainland was negatively related to ant species richness on the islands DISCUSSION (Fig. 2d) and, as visualized in the ordination results of the NMDS analysis, also affected species composition (Fig. 3c). Our results show that human disturbance on islands in the This was confirmed by a one-way ANOSIM testing for Thousand Islands Archipelago is related to an increase in the

Journal of Biogeography 37, 229–236 233 ª 2009 Blackwell Publishing Ltd A. Rizali et al.

(a) (b)

Figure 3 Two-dimensional non-metric multidimensional scaling ordination (NMDS) of ant species composition by island in the Thousand Islands Archipelago based on Sørensen’s similarity index (stress = 0.17). Islands are grouped by (a) the presence or absence of human settlements, (b) area, (c) isolation and (d) the presence of boat docks. Numbers refer to islands listed in Table 1. F, forest; S, settlement; FS, forest and settlement. A1, island area 0–20 ha; A2, island area > 20 ha. D1, 0–15 km from mainland; D2, 16–31 km from mainland; D3, 32–47 km from mainland; D4, 48–63 km from mainland. P, dock(s) present; A, docks absent. number of tramp ant species. Boat docks were present on all common tramp ant species (Anoplolepis gracilipes and Solen- islands with human settlements, and were also found on two opsis geminata) on islands without boat docks. This suggests forested islands near Jakarta that have no apparent human that human commerce and habitat modiﬁcation can increase settlements but are frequented by tourists. Docks facilitate the the likelihood of a tramp species colonizing and/or establishing access of humans to the islands by boat, and tramp species a population on an island. might therefore be more easily introduced to islands with As expected from classical island biogeography theory, we docks, for example in the soil of ornamental or agricultural found that species richness was positively related to island plants (Deyrup et al., 2000). Islands with docks had greater size and negatively related to island isolation from the numbers of tramp species, as did islands with human mainland. However, there was a tendency for islands with settlements, and this was emphasized by the absence of two human settlements to dampen the magnitude of these natural patterns, although our island sample size limits the statistical power of our conclusions. Inhabited islands have equivalent Table 2 One-way analyses of similarity testing for differences in numbers of tramp species regardless of their distance from species composition between islands grouped according to their the mainland, whereas forested islands are strongly affected isolation from the coast of Java: D1, 0–15 km; D2, 16–31 km; D3, by isolation. This pattern suggests that tramp species disperse 32–47 km; D4, 48–63 km. to inhabited islands by means of human agency, whereas Pairwise tests RPforested islands are colonized by natural dispersal and establishment. D1 vs. D2 0.15 0.095 The observed and estimated insular ant species richness did D1 vs. D3 0.06 0.341 not approach local richness recorded from mainland Java, D1 vs. D4 0.41 0.016 and all ant species found in the Thousand Islands Archipel- D2 vs. D3 0.19 0.095 ago represent a subset of the Javanese myrmecofauna. Over D2 vs. D4 0.41 0.016 200 ant species were found in the semi-natural Bogor D3 vs. D4 0.16 0.200 Botanical Garden (Ito et al., 2001) on mainland Java, and

234 Journal of Biogeography 37, 229–236 ª 2009 Blackwell Publishing Ltd Ant communities on small disturbed tropical islands

94 ant species have been recorded from urban habitats in The island with the greatest estimated species richness and Bogor (Rizali et al., 2008). the greatest number of unique ant species was a forested bird The net result of insular anthropogenic disturbance appears sanctuary, which underscores the destructive role of human to be biological homogenization of the ant communities on disturbance and associated tramp ant species on the natural each island, which obscures natural biogeographical patterns ant communities of the other islands, and highlights the and masks the effects of natural processes, as suggested by the importance of protected areas. ‘flattening’ of the regression lines relating species richness in disturbed habitats to island area and isolation from the ACKNOWLEDGEMENTS mainland. Moreover, there were ant species that did not seem to tolerate any form of human disturbance. For example, We are very grateful to Peter Hartmann and Anna Spengler for Pachycondyla SKY sp. 42 and Ponera sp. 01 were found only on support and cooperation during field research. We thank the large, forested island of Rambut, which is just 4.2 km from Navjot S. Sodhi, Phil Lester and an anonymous referee for Java, and Pheidole sp. 02 was found on each forested island and helpful comments on earlier versions of this paper. This on none of the islands with human settlements. research project was funded by the University of Bayreuth, Despite being the closest island to mainland Java, Onrust Germany; PEKA Indonesia Foundation (Indonesian Nature for Island had the lowest ant species richness compared with any Conservation Foundation), Indonesia; and a grant from the of the other islands (14 species) and the highest percentage Directorate of Higher Education (Hibah Tim Pascasarjana – of tramp species (6 tramps, 43% of species total). Historically, DIKTI), Indonesia. In addition, D.J.L. was supported by grant Onrust Island harboured ships for the Dutch East Indies R-154-000-270-112 from the Singapore Ministry of Education Company and served as a fortress for Dutch colonists. Today, and M.M.B. by grant SFB-552 (STORMA) from the German the island is a popular tourist destination (PEMDA DKI, 2003). Research Foundation. The more than 300 years of anthropogenic disturbance on this island is likely to have caused the comparatively low ant REFERENCES species richness. The highest number of observed and estimated ant species Agosti, D., Majer, J.D., Alonso, L.E. & Schultz, T.B. (2002) was recorded on Rambut Island, and this high ant richness Ants: standard methods for measuring and monitoring bio- even included two apparently undescribed species, Hypoponera diversity. Smithsonian Institution Press, Washington, DC. sp. 04 and Ponera sp. 01. Despite its proximity to Java and to Alamsyah, A.T. (2003) Sealand regionalism: new paradigm for the heavily disturbed Onrust Island, Rambut Island has water-based settlement development of the Jakarta metro- remained forested and comparably undisturbed owing to its politan area. Seventh International Congress of APSA, Hanoi designation as a bird sanctuary. The high ant diversity on this 2003: Planning theory and history. Asian Planning Schools island may be maintained by the undisturbed primary forest Association (APSA), Hanoi, Vietnam. and high nutrient input from bird and flying fox excrement. Andersen, A.N. (2000) Global ecology of rainforest ants: In conclusion, human disturbance on small tropical islands functional groups in relation to environmental stress may promote the introduction and/or establishment of tramp and disturbance. Ants: standard methods for measuring and species with well-known invasive properties and documented monitoring biodiversity (ed. by D. Agosti, J.D. Majer, L.E. negative effects on native biota. Although exotic species are Alonso and T.R. Schultz), pp. 25–34. Smithsonian Institu- frequently limited to mainland areas that are climatically tion Press, Washington, DC. similar to their native ranges, they are more likely to invade Bolton, B. (1994) Identification guide to the ant genera of the islands at lower latitudes, perhaps because of reduced compe- world. Harvard University Press, Cambridge, MA. tition (Sax, 2001). This suggests that tropical islands are Bos, M.M., Tylianakis, J.M., Steffan-Dewenter, I. & Tscharntke, particularly susceptible to invasion. Our results strongly T. (2008) The invasive yellow crazy ant and the decline of suggest that anthropogenic disturbance and transportation forest ant diversity in Indonesian cacao agroforests. Biological hasten the colonization of islands by exotic species, which then Invasions, 10, 1399–1409. alters natural communities, thereby obscuring the patterns Clarke, K.R. (1993) Non-parametric multivariate analyses of predicted by biogeographical theory. changes in community structure. Australian Journal of This retrospective view masks the processes that produce Ecology, 18, 117–143. these patterns. Indirect (interference) competition between Clarke, K.R. & Gorley, R.N. (2001) PRIMER v. 5: user manual/ species and potentially complex species-specific interactions tutorial. Primer-E, Plymouth, UK. between tramp and non-tramp ants may occur over a Clarke, K.R. & Warwick, R.M. (2001) A further biodiversity considerable time span and differ with island area and many index applicable to species lists: variation in taxonomic other factors. In the fortuitous circumstances in which an ant distinctness. Marine Ecology Progress Series, 216, 256– community was studied before, during, and after invasion by 278. the Argentine ant, Linepithema humile, Sanders et al. (2003) Colwell, R.K. (1997) EstimateS 5: statistical estimation of species found that community assembly rules were rapidly altered richness and shared species from samples. Available at: after invasion. viceroy.eeb.uconn.edu/estimates (last accessed 7 June 2009).

Journal of Biogeography 37, 229–236 235 ª 2009 Blackwell Publishing Ltd A. Rizali et al.

Deyrup, M., Davis, L. & Cover, S. (2000) Exotic ants in Florida. Proceedings of the National Academy of Sciences USA, 100, Transactions of the American Entomological Society, 126, 2474–2477. 293–326. SAS (2003) JMP: the statistical discovery software. SAS Institute, Ellwood, M.D.F. & Foster, W.A. (2004) Doubling the estimate Inc., Cary, NC. of invertebrate biomass in a rainforest canopy. Nature, 429, Sax, D.F. (2001) Latitudinal gradients and geographic ranges of 549–551. exotic species: implications for biogeography. Journal of Graham, J.H., Hughie, H.H., Jones, S., Wrinn, K., Krzysik, A.J., Biogeography, 28, 139–150. Duda, J.J., Freeman, D.C., Emlen, J.M., Zak, J.C., Kovacic, Schoereder, J.H., Sobrinho, T.G., Ribas, C.R. & Campos, R.B.F. D.A., Chamberlin-Graham, C. & Balbach, H. (2004) Habitat (2004) Colonization and extinction of ant communities in a disturbance and the diversity and abundance of ants fragmented landscape. Austral Ecology, 29, 391–398. (Formicidae) in southeastern fall-line sandhills. Journal of Sodhi, N.S. & Brook, B.W. (2006) Southeast Asian biodiversity Insect Science, 4, 1–15. in crisis. Cambridge University Press, Cambridge. Hanski, I.A. & Gilpin, M.E. (1997) Metapopulation biology: StatSoft, Inc. (2001) STATISTICA (data analysis software ecology, genetics, and evolution. Academic Press, San Diego, system), version 6. StatSoft, Inc. Available at: http://www. CA. statsoft.com (accessed 7 June 2009). Hill, M., Holm, K., Vel, T., Shah, N.J. & Matyhot, P. (2003) Suarez, A.V., Bolger, D. & Case, T.J. (1998) Effects of frag- Impact of the introduced yellow crazy ant Anoplolepis mentation and invasion on native ant communities in gracilipes on Bird Island, Seychelles. Biodiversity and coastal southern California. Ecology, 79, 2041–2056. Conservation, 12, 1969–1984. Ho¨lldobler, B. & Wilson, E.O. (1990) The ants. Harvard SUPPORTING INFORMATION University Press, Cambridge, MA. Holway, D.A., Lach, L., Suarez, A.V., Tsutsui, N.D. & Case, T.J. Additional Supporting Information may be found in the (2002) The causes and consequences of ant invasions. online version of this article: Annual Review of Ecology and Systematics, 33, 181–233. Hubbell, S.P. (2001) The uniﬁed neutral theory of biodiversity Appendix S1 Ant species recorded from individual islands and biogeography. Princeton University Press, Princeton, and habitats in the Thousand Islands Archipelago. NJ. Please note: Wiley-Blackwell is not responsible for the Ito, F., Yamane, S., Eguchi, K., Noerdjito, W.A., Kahono, S., content or functionality of any supporting materials supplied Tsuji, K., Ohkawara, K., Yamauchi, K., Nishida, T. & by the authors. Any queries (other than missing material) Nakamura, K. (2001) Ant species diversity in the Bogor should be directed to the corresponding author for the Botanical Garden, West Java, Indonesia, with descriptions of article. two new species of the genus Leptanilla (Hymenoptera: Formicidae). Tropics, 10, 379–404. Legendre, P. & Legendre, L. (1998) Numerical ecology, 2nd edn. Elsevier, Amsterdam. MacArthur, R.H. & Wilson, E.O. (1967) The theory of island BIOSKETCH biogeography. Princeton University Press, Princeton, NJ. McGlynn, T.P. (1999) The worldwide transfer of ants: geo- Akhmad Rizali is a staff scientist at the Department of Plant graphical distribution and ecological invasions. Journal of Protection, Faculty of Agriculture, Bogor Agricultural Uni- Biogeography, 26, 535–548. versity, Indonesia, is also a staff scientist at PEKA Indonesia Passera, L. (1994) Characteristics of tramp ants. Exotic ants Foundation (Indonesian Nature for Conservation Founda- (ed. by D. Williams), pp. 23–43. Westview Press, Boulder, tion) in Bogor, Indonesia, and is currently a PhD student in CO. the Department of Agroecology, University of Go¨ttingen, PEMDA DKI (2003) Onrust Island. Department of Culture and Germany. His main interests include myrmecology, ecology Museums, Pemerintah Daerah, Daerah Khusus Ibukota and biogeography. (PEMDA DKI), Jakarta. Author contributions: A.R., D.B., L.B.P., H.T. and C.H.S. Rizali, A., Bos, M.M., Buchori, D., Yamane, S. & Schulze, C.H. conceived the ideas; A.R. collected the data; A.R. and S.Y. (2008) Ants in tropical urban habitats: the myrmecofauna in identiﬁed the ant specimens; A.R., D.J.L. and C.H.S. analysed a densely populated area of Bogor, West Java, Indonesia. the data; and A.R., D.J.L. and M.M.B. led the writing. HAYATI Journal of Biosciences, 15, 77–84. Sanders, N.J., Gotelli, N.J., Heller, N.E. & Gordon, D.M. (2003) Community disassembly by an invasive species. Editor: Kate Parr

236 Journal of Biogeography 37, 229–236 ª 2009 Blackwell Publishing Ltd