Ecological niche modelling as a technique for
Transcription
Ecological niche modelling as a technique for
Diversity and Distributions, (Diversity Distrib.) (2009) 15, 289–298 Blackwell Publishing Ltd BIODIVERSITY RESEARCH Ecological niche modelling as a technique for assessing threats and setting conservation priorities for Asian slow lorises (Primates: Nycticebus) J. S. Thorn*, V. Nijman, D. Smith and K. A. I. Nekaris Nocturnal Primate Research Group, School of Social Sciences and Law, Oxford Brookes University, Oxford OX3 0BP, UK ABSTRACT Aim Data on geographical ranges are essential when defining the conservation status of a species, and in evaluating levels of human disturbance. Where locality data are deficient, presence-only ecological niche modelling (ENM) can provide insights into a species’ potential distribution, and can aid in conservation planning. Presenceonly ENM is especially important for rare, cryptic and nocturnal species, where absence is difficult to define. Here we applied ENM to carry out an anthropogenic risk assessment and set conservation priorities for three threatened species of Asian slow loris (Primates: Nycticebus). Location Borneo, Java and Sumatra, Southeast Asia. Methods Distribution models were built using maximum entropy (MaxEnt) ENM. We input 20 environmental variables comprising temperature, precipitation and altitude, along with species locality data. We clipped predicted distributions to forest cover and altitudinal data to generate remnant distributions. These were then applied to protected area (PA) and human land-use data, using specific criteria to define low-, medium- or high-risk areas. These data were analysed to pinpoint priority study sites, suitable reintroduction zones and protected area extensions. Results A jackknife validation method indicated highly significant models for all three species with small sample sizes (n = 10 to 23 occurrences). The distribution models represented high habitat suitability within each species’ geographical range. High-risk areas were most prevalent for the Javan slow loris (Nycticebus javanicus) on Java, with the highest proportion of low-risk areas for the Bornean slow loris (N. menagensis) on Borneo. Eighteen PA extensions and 23 priority survey sites were identified across the study region. *Correspondence: J. S. Thorn, Nocturnal Primate Research Group, School of Social Sciences and Law, Oxford Brookes University, Oxford OX3 0BP, UK. E-mail: [email protected] Main conclusions Discriminating areas of high habitat suitability lays the foundations for planning field studies and conservation initiatives. This study highlights potential reintroduction zones that will minimize anthropogenic threats to animals that are released. These data reiterate the conclusion of previous research, showing MaxEnt is a viable technique for modelling species distributions with small sample sizes. Keywords Lorisidae, MaxEnt, pet trade, protected areas, risk assessment, Sundaland. Restricted spatial distribution patterns in correlation with anthropogenic disturbance can be used to define species rarity (Rabinowitz, 1981; Kattan, 1992; Gaston, 1997). The geographical extent of a species will continue to shrink as a result of ongoing habitat loss and anthropogenic pressures (Cowlishaw & Dunbar, 2000). Data on geographical ranges are therefore of paramount importance when delineating the conservation status of a species (Anderson & Martinez-Meyer, 2004; Hernandez et al., 2006) and in evaluating current levels of threat and protection (Fuller et al., 2006). Applying ecological niche modelling (ENM) can provide insights into a species’ potential distribution when sufficient survey data do not exist, and can aid in conservation planning © 2008 The Authors Journal compilation © 2008 Blackwell Publishing Ltd DOI: 10.1111/j.1472-4642.2008.00535.x www.blackwellpublishing.com/ddi INTRODUCTION 289 J. S. Thorn et al. (Guisan & Zimmermann, 2000; Loiselle et al., 2003) by highlighting unknown populations (Pearson et al., 2007), suitable sites for reintroduction (Hernandez et al., 2006), key areas for fieldwork (Papes & Gaubert, 2007) and in improving assessment of risk status (Solano & Feria, 2007). Where data are deficient on a species’ geographical distribution, conservationists often rely on its presence within and around protected area (PA) networks in order to assess its conservation status. Species persistence in these areas may, however, be heavily affected by extrinsic factors such as habitat loss and over exploitation (Lande, 1998). Many studies highlight the necessity of incorporating ecological risk assessments into quantifying vulnerability to human disturbance and conservation planning (Williams, 1998; Araujo & Williams, 2000; Harwood, 2000; Villa & Mcleod, 2002; Wilson et al., 2005; Solano & Feria, 2007). Measuring the level of threat posed to the ecological viability of an area by factors such as agricultural encroachment, proximity to human settlements, road building or hunting pressure is necessary for developing conservation strategies for species inhabiting human dominated landscapes (Oates, 1986; Eudey, 1987). We utilized MaxEnt ENM software (Phillips et al., 2006) to carry out such a study on a group of cryptic, little-known primates, the Asian slow lorises (Nycticebus). Although other modelling algorithms have been more widely tested with vertebrate species, recent research has demonstrated greater modelling accuracy with MaxEnt and limited datasets (i.e. < 25 localities) (Hamel et al., 2006; Hernandez et al., 2006; Phillips et al., 2006; Pearson et al., 2007). MaxEnt is a general purpose method for generating predictions or inferences from incomplete information (Phillips et al., 2006). Probabilities of occurrence are generated across the entire study region based on the environmental conditions in areas where the species have been observed. MaxEnt requires presence-only data, using a random selection of background pixels from the study area as pseudo-absences (Phillips et al., 2006). When modelling species distribution, inferring absences with any amount of certainty can be problematic (Sutherland, 2000; Hirzel et al., 2002; Brotons et al., 2004; Elith et al., 2006; Phillips et al., 2006; Gibson et al., 2007), particularly when the taxa are rare or cryptic, or biotic or human-induced factors prevent colonization of all suitable habitat areas (Hirzel et al., 2001; Engler et al., 2004). Absence can more easily be determined if the species is conspicuous and easily identified (Sutherland, 2000). Slow lorises are cryptic and nocturnal primates that, to date, have been little-studied in the wild (Nekaris & Bearder, 2007), including few data on their precise geographical ranges. Low encounter rates of Nycticebus (Nekaris & Nijman, 2007a; Nekaris et al., 2008) confirm the inherent difficulties in determining absence for this genus. Presence-only modelling is therefore highly suitable to generate urgently needed data on their distribution. Already listed as Vulnerable (VU a2cd) or Endangered (EN a2cd) (IUCN, 2006), recent studies show greater taxonomic diversity and variable ecology (Groves, 2001; Roos, 2003; Brandon-Jones et al., 2004; Chen et al., 2006; Nekaris et al., 2006; Nekaris & Bearder, 2007; Nekaris & Jaffe, 290 2007; Groves & Maryanto, 2008), urging a more quantitative analysis of their distribution and threats. In addition to habitat loss plaguing all wildlife in southeast Asia, Nycticebus spp. in particular are threatened by unsustainable harvesting for the pet trade and traditional medicines (Nekaris & Bearder, 2007), even resulting in their elevation to Appendix I of the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES, 2007; Nekaris & Nijman, 2007b). It is essential that organizations involved in the management of Nycticebus are able to identify suitable study areas, extensions of the PA network and sites for reintroduction of individuals confiscated from the illegal wildlife trade (Schulze & Groves, 2004). ENM can play a pivotal role in making these recommendations. Our study focused on modelling the distribution of three threatened species of Nycticebus: on Borneo the Bornean slow loris (N. menagensis), on Java the Javan slow loris (N. javanicus), and on Sumatra the greater slow loris (N. coucang) (~8.00° N– 9.00° S, 94.00° E–120.00° E). We conducted an anthropogenic risk assessment in order to pinpoint distributional regions that are undergoing varying levels of human-induced threat. Finally, we used data from the MaxEnt predictions and anthropogenic risk assessment to set conservation priorities for Nycticebus. METHODS Modelling data We compiled locality data for the three study species using zoological collections, published literature and communication with researchers. Zoological collections were accessed through the online databases of the Mammal Networked Information System (MaNIS) and the Global Biodiversity Information Facility (GBIF). We also visited collections at the Natural History Museum, London, the Natural History Museum, Oxford, Naturalis, Leiden, and the Zoological Museum, Amsterdam. A questionnaire and species photo sheet were emailed to researchers to gather data on the occurrence of and perceived threats to Nycticebus. A total of 147 independent localities for Nycticebus were collected from the above sources, with 48 of these localities selected for use in the final modelling process. Selection was based on exclusion of localities that were now outside of suitable habitat regions, i.e. areas completely denuded of forest. We also aimed to generate spatial independence between localities and to lessen the effects of spatial autocorrelation present at low sample sizes (Hernandez et al., 2006; Phillips et al., 2006; Pearson et al., 2007). Where localities fell within 10 km of each other, only one locality from the cluster was selected to represent species presence in that area. Twenty environmental layers comprising temperature, precipitation and altitude were utilized (see Appendix S1 in Supporting Information). We extracted eleven temperature and eight precipitation variables from the WorldClim database (Hijmans et al., 2005). The set of bioclimatic variables available are derived from monthly temperature and rainfall values to produce more ecologically relevant layers on a grid resolution of ~1 km2. Elevation data were derived from the digital elevation model (DEM) GTOPO30 (US Geological Survey, 1996). All environmental © 2008 The Authors Diversity and Distributions, 15, 289–298, Journal compilation © 2008 Blackwell Publishing Ltd ENM and slow loris conservation variables were resampled to the WGS84 geographical coordinate system at a resolution of ~1 km2 (0.008333 decimal degrees). Ecological niche modelling Potential distributions of N. menagensis, N. javanicus and N. coucang were predicted by generating statistically significant models of occurrence based on their ecological niche requirements. MaxEnt software version 3.1.0 (Computer Sciences Department – Princeton University, 2004) was utilized for this purpose. Species occurrence data and environmental variables were entered into MaxEnt and predicted distributions projected using ArcGIS 9.2. The distribution of each study species was modelled separately, with the study region divided into three subregions. Fifteen spatially independent localities on Sumatra (N. coucang), 10 on Java (N. javanicus) and 23 on Borneo (N. menagensis) were utilized. As MaxEnt generates a cumulative probability distribution output between 0 and 100%, we selected a decision threshold to distinguish presence from absence, enabling validation and visual interpretation of model predictions (Liu et al., 2005; Pearson et al., 2007). Binary maps of presence or absence were then created based on this threshold value. A lowest presence threshold (LPT) (Pearson et al., 2007) was used, thus maintaining zero omission error in the model localities, and also representing areas of habitat that are at least as suitable as those where the species has been observed (Hernandez et al., 2006; Pearson et al., 2007). For a detailed explanation of ENM methodology using MaxEnt see Phillips et al. (2006). Model validation We explicitly followed the jackknife validation methodology developed by Pearson et al. (2007), which is shown to be effective for sample sizes of 25 or less. Using this approach, one locality point was removed from the dataset, and the model built using the remaining n–1 localities. Thus, for a species with n localities, n individual models were built for testing. Model accuracy and significance were evaluated based on the ability of each model to predict the one excluded test locality as present (Pearson et al., 2007). Predictions were also made for each species using all n localities to gain best-fit models across the entire study region for visualization and spatial analysis in ArcGIS 9.2. Remnant distributions We used data on remaining habitat and altitudinal limits of Nycticebus to clip the MaxEnt predictions to generate remnant distributions. An altitudinal layer was applied to the model; however, with the tendency for MaxEnt to over-predict, it was necessary to further process the results using known altitudinal limits. We did not incorporate land-cover data in modelling because it is necessary for species localities and environmental layers to correspond temporally (Phillips et al., 2006). Hence, land-cover data are not suitable for use with occurrence data from a number of decades, particularly in regions of rapid deforestation and development. The habitat layer was created using the most up-to-date uniform land-cover map available for Southeast Asia (GLC, 2000). We considered four vegetation types to be suitable habitat for Nycticebus: evergreen montane forest, evergreen lowland forest, mangrove forest and swamp forest. Vegetation patches of less than 10 km2 were considered as unsuitable for maintaining viable populations of Nycticebus and were therefore removed. Based on our locality data we set the altitudinal limit at 1000 m on Borneo and 1500 m on Java and Sumatra. Anthropogenic risk assessment Suitable habitat regions occupied by Nycticebus according to the MaxEnt predictions, and situated outside of PA boundaries were classified using the criteria in Table 1. In our list of PAs, we also included some areas of global conservation importance but with less stringent status, such as the Leuser Ecosystem and the ‘Heart of Borneo’ that are made up of a series of PAs linked by lessprotected forested land. We classified habitat patches as low risk (LR), medium risk (MR) or high risk (HR). Layers of agriculture from the GLC 2000, and populated places, roads and PAs from the World Database on Protected Areas (WDPA) were overlaid with remnant distributions of Nycticebus. We selected a distance of more than 10 km from human access points as a conservative estimate of walking threshold, and therefore anything beyond this distance was considered LR (Peres & Lake, 2002). We then calculated percentages of each species’ suitable habitat area qualifying as LR, MR or HR in order to apply a chi-squared (χ2) test to assess varying risk levels between species. Table 1 Criteria for qualifying habitat patches as low, medium or high risk. The ranking indicates the suitability of habitat patches for supporting viable populations of Nycticebus. Measure Low risk Medium risk High risk Size of forest patch Proximity to protected areas Proximity to populated areas Proximity to roads Proximity to agriculture > 40 km2 Within 20 km* > 10 km > 10 km > 5 km > 20 km2 Within 20–30 km > 5 km > 5 km > 2.5 km > 10 km2 Within 30–40 km Adjacent Adjacent Adjacent *For N. menagensis ‘Proximity to protected area’, the low risk criterion was within 20 km of protected area network or inside the Heart of Borneo. © 2008 The Authors Diversity and Distributions, 15, 289–298, Journal compilation © 2008 Blackwell Publishing Ltd 291 J. S. Thorn et al. Conservation priority areas RESULTS We made recommendations for key PAs to be prioritized as potential survey sites, as well as for extensions of the existing PA Network. Priority survey sites were based on PAs where there is not yet active research on Nycticebus and where MaxEnt predictions identified suitable ecological conditions. Recommendations for PA extensions were based on LR areas for Borneo, and LR/MR areas for Java and Sumatra due to higher densities of human land-use features. In order to minimize the potential level of human–wildlife conflict, we selected PA extensions only in areas where habitat was classified as LR or MR. The jackknife method for model testing (Pearson et al., 2007) demonstrated high predictive success rates for all three species, with highly significant models for N. coucang and N. javanicus (P < 0.01) and significant models for N. menagensis (P < 0.05) (Table 2). The predicted distributions in Fig. 1 reflect the output generated by MaxEnt before further application of altitudinal limits and ‘current’ forest cover to create the remnant distributions. Nycticebus javanicus is significantly more vulnerable to anthropogenic activity (χ2 = 2689, d.f. = 4, P < 0.0001). Intensive human land use can be found across much of the study region, Figure 1 Species predicted and remnant distributions for (a) Nycticebus coucang on Sumatra, (b) N. menagensis on Borneo and (c) N. javanicus on Java. Predicted distributions (grey) represent the MaxEnt output before clipping to the altitude and vegetation layers. Remnant distributions (black) represent the predicted distribution clipped to the altitude and vegetation layers. 292 © 2008 The Authors Diversity and Distributions, 15, 289–298, Journal compilation © 2008 Blackwell Publishing Ltd ENM and slow loris conservation Table 2 Results of the jackknife validation method of model testing for Nycticebus coucang, N. javanicus and N. menagensis, showing the sample sizes included for modelling, and the data used to calculate the P-values. Species Locality sample size Number of successes Mean fractional predicted area Lowest presence threshold (LPT) P-value N. coucang N. javanicus N. menagensis 15 10 23 13 9 21 0.54 0.49 0.72 18.664 12.523 4.886 0.006 0.003 0.027 Figure 2 Percentage of suitable habitat classified as low risk, medium risk or high risk for Nycticebus menagensis, N. javanicus and N. coucang according to the risk assessment criteria. and is most prevalent on Java. The lowest vulnerability levels are for N. menagensis, which has both the highest percentage of LR and the lowest percentage of HR areas across all species (Fig. 2). Percentage of MR areas are most evenly spread across species. Conservation priority areas The risk assessment on Sumatra identified potential for six PA extensions (Fig. 3a). Six areas were identified for PA extensions on Borneo, providing connectivity among ten PAs (Fig. 3b). Due to the fragmented nature of the habitat on Java, only one PA extension has potential to provide connectivity between two PAs, on the west side between Gunung Tilu and Gunung Simpang. A total of five further PA extensions were identified (Fig. 3c). Six PAs have been recommended as priority survey sites on Sumatra, seven on Borneo, and ten on Java (see Table 3). DISCUSSION MaxEnt as a conservation technique Application of the jackknife validation method confirms that MaxEnt is effective at predicting distributions that are representative of species’ actual distribution, using limited occurrence data. The models were significantly validated for all taxa based on the calculated P-values. It is important to treat model predictions using low sample sizes as representing areas of high habitat suitability as opposed to presence of the species (Pearson et al., 2007). Based on the jackknife tests, significant predictive ability can be achieved with as few as ten localities. Both Pearson et al. (2007) and Hernandez et al. (2006) even suggest that useful predictions using MaxEnt can be attained with sample sizes below ten, and this presents positive implications for the scope of applying MaxEnt when setting conservation priorities. Reducing the minimum number of occurrences required to produce useful predictions of species distribution will greatly increase the number of species that can be studied (Pearson et al., 2007). This is particularly important for rare and cryptic species, where inclusion of known distributions is vital for effective conservation planning. As with other studies of this kind (e.g. Raxworthy et al., 2003; Papes & Gaubert, 2007; Pearson et al., 2007; Sergio et al., 2007; Ward, 2007), the resulting MaxEnt predictions demonstrate how ENM uses environmental variables to generate outputs not possible purely from observed localities. The model indicates these areas as having similar environmental conditions to areas where the species is known to occur. The distribution maps clearly show how the models project into areas of unknown occurrence when the training localities are overlaid with their respective predicted distributions. This highlights the value of MaxEnt in discriminating unrecorded populations and areas of high habitat suitability without relying on confirmed species presence. As more field studies are undertaken for Nycticebus, further records of occurrence will influence the shape of predictions, particularly if the new locality data are sourced from an area with unique environmental conditions not yet represented by the current data set (Pearson et al., 2007). Model limitations based on taxonomy Taxonomic uncertainty can contribute to decreased modelling accuracy, with models performing most poorly for species that have high taxonomic confusion (Hernandez et al., 2006). Misidentification is particularly significant for cryptic species such as Nycticebus, both in the wild and in captive environments (Nekaris & Jaffe, 2007). Many researchers, museum curators and rescue centres still utilize the 1971 taxonomy of Groves (Molur et al., 2003; Schulze & Groves, 2004), which will lead to modelling inaccuracies if using occurrence data from these sources. Nycticebus is undergoing wide taxonomic revision, and preliminary evidence has been put forward to suggest the existence of two forms on Borneo, two on Java and two on Sumatra (Nekaris & Jaffe, 2007; Nekaris & Nijman, 2007a). A number of major © 2008 The Authors Diversity and Distributions, 15, 289–298, Journal compilation © 2008 Blackwell Publishing Ltd 293 J. S. Thorn et al. Figure 3 Recommendations for protected area extensions and priority survey areas based on species remnant distributions and results of the risk assessment for (a) Nycticebus coucang on Sumatra, (b) N. menagensis on Borneo, and (c) N. javanicus on Java. Protected area extensions are shown in dark grey and priority survey areas are shown in black. river systems originated from the highlands of Borneo, Java and Sumatra, and transected the lowlands during the Pleistocene period (Voris, 2000). Combined with more recent forest fragmentation events, these waterways may have represented effective zoogeographical barriers and agents in population isolation and speciation (Harrison et al., 2006). Distinct forms on the outlying islands of Banka (Lyon, 1906) and Natuna (Chasen, 1935) are also a possibility. It would be interesting to test whether further models of Nycticebus distributions can be improved by partitioning occurrence data into the geographical 294 boundaries of the new proposed forms. Such data are still forthcoming, however, and a study of this kind would be founded on speculation of geographical extents. Genetic analyses and field-based morphometrics are urgently needed to solve these taxonomic discrepancies. Implications for the conservation status of Nycticebus Many forest patches highlighted to have suitable ecological conditions for Nycticebus may actually be devoid of populations © 2008 The Authors Diversity and Distributions, 15, 289–298, Journal compilation © 2008 Blackwell Publishing Ltd ENM and slow loris conservation Table 3 Table of recommendations for priority survey areas and protected area (PA) extensions for Borneo, Java and Sumatra, based on the distribution of Nycticebus and results of the risk assessments. Region Recommendations for priority survey areas Recommendations for PA extensions Sumatra • • • • • Bukit Barisan Selatan Kerinci Sebelat Bukit Tigapuluh Berbak Way Kambas Borneo • • • • • • • Crocker Range Kayan-Mentarang Betung Kerihun Bukit Batutenobang Bukit Baka Bukit Raya Bukit Sepat Haung Taman Negara Banjaran Java • • • • • • • • • • Ujung Kulon Halimun Gunung Gede Pangrango Gunung Masigit Kareumbi Gunung Tilu Gunung Simpang Gunung Sawal Bromo Tengger Semeru Yang Highlands Meru Betiri • • • • • • • • • • • • • • • • • • • of Nycticebus, due to unsustainable harvesting for the wildlife trade, already shown to be true for some areas (Nekaris & Nijman, 2007a; Nekaris et al., 2008). Nycticebus has been documented as the most prevalent legally protected primate taxon in live animal markets (Malone et al., 2002; Webber & Nekaris, 2004; Mcgreal, 2007). This is clearly an area of research that needs to be prioritized with a systematic approach to quantify the effects of the wildlife trade on wild populations of Nycticebus. The situation for N. menagensis is optimistic, with a high percentage of LR areas. For N. javanicus and N. coucang, however, due to higher human land-use densities, it may be necessary to prioritize LR and MR areas to create sites that are viable in terms of size as well as levels of anthropogenic risk. These areas will require extensive surveys to establish current population densities and habitat viability (Streicher, 2004b). Ideal sites for reintroduction will incorporate high arboreal connectivity with low population density (Nekaris & Nijman, 2007a). Assessing threats and setting conservation priorities The human land-use features assessed in this study contribute strongly to the intensity of harvesting for the pet trade. Virtually all forest areas in the study region are intersected by roads or within the vicinity of populated places and much of the suitable habitat for Nycticebus has already been converted to agriculture. Ulu Masen Bukit Barisan Barumun Berbak Siranggas Wildlife Reserve to Leuser Ecosystem Bukit Hitan Protection Forest to Bukit Balai Rejang Protection Forest Ulu Temburong Gunung Mulu Labi Hills Ladan Hills Maliau Basin Danum Valley Kayan Mentarang to Betung Kerihun Bukit Sepat Haung to Betung Kerihun and Bukit Batutenobang Bukit Batutenobang to Bukit Baka Bukit Raya Bukit Perai to Bukit Rongga Halimun Gunung Gede Pangrango Gunung Simpang to Gunung Tilu Effective conservation planning for human-dominated landscapes needs to address issues of PA connectivity (Fuller et al., 2006) alongside the requirements of the local human population (Gillingham & Lee, 1999), possibly through multicriteria analysis (Moffett & Sarkar, 2006). When time, money and resources are limited, it is more efficient to build upon existing protected areas than to create new ones (Noss, 1987; Johns, 1991; Tutin et al., 1997; Marsh, 2003; Fuller et al., 2006; Strier, 2007). Establishing connectivity between PAs is especially significant for arboreal primate species such as Nycticebus which, due to behavioural adaptations, will not cover large distances terrestrially (Nekaris & Bearder, 2007). Nycticebus spp. have been encountered dead on power lines, suggesting failed attempts at dispersal between forest patches, when the natural habitat has been destroyed (Streicher, 2004a, Jan Beck, personal communication, July 2007). Animals have also been observed as victims of road kill, where roads intersect forest patches (Nekaris & Schulze, 2004). These two factors highlight the difficulty animals face in trying to disperse in a humandominated landscape, and further emphasize the importance of establishing arboreal connectivity between PAs. A challenging issue facing organizations involved in the conservation of Nycticebus is locating suitable sites for reintroduction of individuals confiscated from the wildlife trade (Schulze & Groves, 2004). The MaxEnt predictions and risk © 2008 The Authors Diversity and Distributions, 15, 289–298, Journal compilation © 2008 Blackwell Publishing Ltd 295 J. S. Thorn et al. assessment are a preliminary step towards highlighting areas of high habitat suitability, which may also be effective reintroduction zones that minimize anthropogenic threats once the animals are released. Furthermore, these data reiterate the conclusion of previous research, showing MaxEnt is a viable technique for modelling species distributions with small sample sizes. ACKNOWLEDGEMENTS We would like to thank all those who responded to email requests for slow loris data. The following individuals provided localities for use in the ENM: Imam Basuki Jan Beck, Susan Cheyne, Wawan Djum, Colin Groves, Kristofer Helgen, Agung Nugroho, Matt Struebig, Reed Wadley and Serge Wich. Thanks also go to Richard Pearson for advice on using the Jackknife Validation Method. We thank the following individuals for access to and help with their zoological collections: Hein van Grouw (Naturalis Leiden), Daphne Hill and Paula Jenkins (Natural History Museum London), Adri Rol (Zoological Museum Amsterdam), and Malgosia Nowak-Kemp (Natural History Museum Oxford). Finally, we would like to thank Sarah Jaffe for her invaluable comments and suggestions through various versions of this manuscript. Funding was provided by the Systematics Research Fund of the Systematics Association and Linnaean Society of London, Oxford Brookes University Research Strategy Fund and SYNTHESYS (NL-TAF-3491). This research formed a part of an MSc in Primate Conservation by J S Thorn at Oxford Brookes University. REFERENCES Anderson, R.P. & Martinez-Meyer, E. (2004) Modelling species’ geographic distributions for preliminary conservation assessments: an implementation with the spiny pocket mice (Heteromys) of Ecuador. Biological Conservation, 116, 167– 179. Araujo, M.B. & Williams, P.H. (2000) Selecting areas for species persistence using occurrence data. Biological Conservation, 96, 331–345. Brandon-Jones, D., Eudey, A.A., Geissmann, T., Groves, C.P., Melnick, D.J., Morales, J.C., Shekelle, M. & Stewart, C.B. (2004) Asian primate classification. International Journal of Primatology, 25, 97–164. Brotons, L., Thuiller, W., Araújo, M.B. & Hirzel, A.H. (2004) Presence-absence versus presence-only modelling methods for predicting bird habitat suitability. Ecography, 27, 437–448. Chasen, F.N. (1935) On a collection of mammals from the Natuna Islands, South China Sea. Bulletin of the Raffles Museum of Singapore, 10, 5–42. Chen, J.H., Pan, D., Groves, C., Wang, Y.X., Narushima, E., Fitch-Snyder, H., Crow, P., Thanh, V.N., Ryder, O., Zhang, H.W., Fu, Y.X. & Zhang, Y.P. (2006) Molecular phylogeny of Nycticebus inferred from mitochondrial genes. International Journal of Primatology, 27, 1187–1200. CITES (2007) UNEP–WCMC Species Database: CITES-Listed Species. www.cites.org. Accessed 30 June 2007. 296 Cowlishaw, G. & Dunbar, R. (2000) Primate conservation biology. The University of Chicago Press, Chicago, Illinois. Elith, J., Graham, C.H., Anderson, R.P., Miroslav, D., Ferrier, S., Guisan, A., Hijmans, R.J., Huettmann, F., Leathwick, J.R., Lehmann, A., Li, J., Lohmann, L.G., Loiselle, B.A., Manion, G., Moritz, C., Nakamura, M., Nakazawa, Y., Overton, J.McC., Townsend Peterson, A., Phillips, S.J., Richardson, K., Scachetti-Pereira, R., Schapire, R.E., Soberon, J., Williams, S., Wisz, M.S. & Zimmermann, N.E. (2006) Novel methods improve prediction of species’ distributions from occurrence data. Ecography, 29, 129–151. Engler, R., Guisan, A. & Rechsteiner, L. (2004) An improved approach for predicting the distribution of rare and endangered species from occurrence and pseudo-absence data. Journal of Applied Ecology, 41, 263–274. Eudey, A.A. (1987) Action plan for Asian primate conservation, pp. 1987–91. IUCN, Gland, Switzerland. Fuller, T., Munguia, M., Mayfield, M., Sanchez-Cordero, V. & Sarkar, S. (2006) Incorporating connectivity into conservation planning: a multi-criteria case study from Central Mexico. Biological Conservation, 133, 131–142. Gaston, K.J. (1997) What is rarity? The biology of rarity: causes and consequences of rare-common differences (ed. by W.E. Kunin and K.J. Gaston), pp. 30–47. Chapman and Hall, London. Gibson, L., Barrett, B. & Burbidge, A. (2007) Dealing with uncertain absences in habitat modelling: a case study of a rare ground-dwelling parrot. Diversity and Distributions, 13, 704– 713. Gillingham, S. & Lee, P.C. (1999) The impact of wildlife-related benefits on the conservation attitudes of local people around the Selous Game Reserve, Tanzania. Environmental Conservation, 26, 218–228. GLC (2000) European Commission Joint Research Centre, 2003. www.gvm.jrc.it/GLC200. Accessed 15 February 2007. Groves, C.P. (2001) Primate taxonomy. Smithsonian Institution Press, Washington, DC. Groves, C. & Maryanto, I. (2008) Craniometry of slow lorises (genus Nycticebus) of insular Southeast Asia. Primates of the oriental night (ed. by M. Shekelle, C. Groves, I. Maryanto, H. Schulze and H. Fitch-Snyder), pp. 115–122. LIPI Press, Jakarta. Guisan, A. & Zimmermann, N.E. (2000) Predictive habitat distribution models in ecology. Ecological Modelling, 135, 147– 186. Hamel, P., Barker, S., Benítez, S., Baldy, J., Cisneros Heredia, D., Colorado Zuluaga, G., Cuesta, F., Davidson, I., Díaz, D., Ganzenmueller, A., García, S., Girvan, M.K., Guevara, E., Hamel, P., Hennessey, A.B., Hernández, O.L., Herzog, S., Mehlman, D., Moreno, M.I., Ozdenerol, E., Ramoni-Perazzi, P., Romero, M., Romo, D., Salaman, P., Santander, T., Tovar, C., Welton, M., Will, T., Pedraza, C. & Galindo, G. (2006) Modelling the south American range of the cerulean warbler. Proceedings of the Twentysixth ESRI International User Conference, San Diego, California, 6–11 August 2006. [Online]. http://www.srs.fs.usda.gov/pubs/ 27262. Harrison, T., Krigbaum, J. & Manser, J. (2006) Primate biogeography and ecology on the Sunda Shelf Islands: a paleontological © 2008 The Authors Diversity and Distributions, 15, 289–298, Journal compilation © 2008 Blackwell Publishing Ltd ENM and slow loris conservation and zooarchaeological perspective. Primate Biogeography: Progress and Prospects (ed. by S.M. Lebman and J.G. Fleagle), pp. 331–374. Springer, New York. Harwood, J. (2000) Risk assessment and decision analysis in conservation. Biological Conservation, 95, 219–226. Hernandez, P.A., Graham, C.H., Master, L.L. & Albert, D.L. (2006) The effect of sample size and species characteristics on performance of different species distribution modeling methods. Ecography, 29, 773–785. Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P. & Jarvis, A. (2005) Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology, 25, 1965–78. Hirzel, A.H., Helfer, V. & Métral, F. (2001) Assessing habitat suitability models with a virtual species. Ecological Modelling, 145, 111–121. Hirzel, A.H., Hausser, J., Chessel, D. & Perrin, N. (2002) Ecological-niche factor analysis: how to compute habitatsuitability maps without absence data. Ecology, 83, 2027–36. IUCN (2006) IUCN Red List of Threatened Species. http:// www.iucnredlist.org. Accessed 1 May 2007. Johns, A.D. (1991) Forest disturbance and Amazonian primates. Primate responses to environmental change (ed. by H.O. Box), pp. 115–135. Chapman & Hall, London. Kattan, G.H. (1992) Rarity and vulnerability: the birds of the Cordillera Central Colombia. Conservation Biology, 6, 64–70. Lande, R. (1998) Anthropogenic, ecological, and genetic factors in extinction. Conservation in a changing world: integrating process into priorities for action (ed. by G. Mace, A. Balmford and J.R. Ginsberg), pp. 29–51. Cambridge University Press, Cambridge, UK. Liu, C., Berry, P.M., Dawson, T.P. & Pearson, R.G. (2005) Selecting thresholds of occurrence in the prediction of species distributions. Ecography, 28, 385 –393. Loiselle, B.A., Howell, C.A., Graham, C.H., Goerck, J.M., Brooks, T., Smith, K.G. & Williams, P.H. (2003) Avoiding pitfalls of using species distribution models in conservation planning. Conservation Biology, 17, 1591–1600. Lyon, M.W. (1906) Mammals of Banka, Mendanau, and Billiton Islands, Between Sumatra and Borneo. Proceedings of the US National Museum, 31, 575–612. Malone, N., Purnama, A.R., Wedana, M. & Fuentes, A. (2002) Assessment of the sale of primates at Indonesian bird markets. Asian Primates, 8, 7–11. Marsh, L.K. (2003) The nature of fragmentation. Primates in fragments: ecology and conservation (ed. by L.K. Marsh), pp. 1–8. Kluwer Academic/Plenum, New York. Mcgreal, S. (2007) Loris confiscations highlight need for protection. IPPL News, 34, 3. Moffett, A. & Sarkar, S. (2006) Incorporating multiple criteria into the design of conservation area networks: a minireview with recommendations. Diversity and Distributions, 12, 125– 137. Molur, S., Brandon-Jones, D., Dittus, W., Eudey, A., Kumar, A., Singh, M., Feeroz, M.M., Chalise, M., Priya, P. & Feeroz, S.E. (2003) Status of south Asian primates: conservation assessment and management plan (C.A.M.P.) workshop report 2003. Zoo Outreach Organisation/CBSG-South Asia, Coimbatore, India. Nekaris, K.A.I. & Bearder, S.K. (2007) The lorisiform primates of Asia and mainland Africa: diversity shrouded in darkness. Primates in perspective (ed. by C.J. Campbell, A. Fuentes, K.C. Mackinnon, M. Panger and S.K. Bearder), pp. 24 –45. Oxford University Press, Oxford, UK. Nekaris, K.A.I. & Jaffe, S. (2007) Unexpected diversity within the Javan slow loris trade: implications for slow loris taxonomy. Contributions to Zoology, 76, 187–196. Nekaris, K.A.I. & Nijman, V. (2007a) Survey on the abundance and conservation of Sumatran slow lorises (Nycticebus coucang hilleri) in Aceh, Northern Sumatra. 2nd Congress of the European Federation for Primatology. Charles University, Prague, Czech Republic. Nekaris, K.A.I. & Nijman, V. (2007b) Cites proposal highlights rarity of Asian nocturnal primates (lorisidae: Nycticebus). Folia Primatologica, 78, 211–214. Nekaris, K.A.I. & Schulze, H. (2004) Historical and recent developments of human–loris relations in South and Southeast Asia [abstract]. Primate Eye, 84, 17–18. Nekaris, K.A., Roos, C., Pimley, E.R. & Schulze, H. (2006) Diversity slowly coming to light: reconsidering the taxonomy of pottos and lorises. International Journal of Primatology, Abst #441. Nekaris, K.A.I., Blackham, G. & Nijman, V. (2008) Conservation implications of low encounter rates of five nocturnal primate species (Nycticebus sp.) in Southeast Asia. Biodiversity and Conservation, 17, 733–747. Noss, R.F. (1987) Corridors in real landscapes: a reply to Simberloff and Cox. Conservation Biology, 1, 159–164. Oates, J.F. (1986) Action plan for African primate conservation, 1986–90. IUCN/SSC Primate Specialist Group, Stony Brook, New York. Papes, M. & Gaubert, P. (2007) Modelling ecological niches from low numbers of occurrences: assessment of the conservation status of poorly known viverrids (Mammalia, Carnivora) across two continents. Diversity and Distributions, 13, 890–902. Pearson, R.G., Raxworthy, C.J., Nakamura, M. & Townsend Peterson, A. (2007) Predicting species distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. Journal of Biogeography, 34, 102–117. Peres, C.A. & Lake, I.R. (2002) Extent of nontimber resource extraction in tropical forests: accessibility to game vertebrates by hunters in the Amazon Basin. Conservation Biology, 17, 521–535. Phillips, S.J., Anderson, R.P. & Schapire, R.E. (2006) Maximum entropy modelling of species geographic distributions. Ecological Modelling, 190, 231–259. Rabinowitz, D. (1981) Seven forms of rarity. The biological aspect of rare plant conservation (ed. by H. Synge), pp. 205–217. Wiley, New York. Raxworthy, C.J., Martı´Nez-Meyer, E., Horning, N., Nussbaum, R.A., Schneider, G.E., Ortega-Huerta, M.A. & Peterson, A.T. (2003) Predicting distributions of known and unknown reptile species in Madagascar. Nature, 426, 837–841. © 2008 The Authors Diversity and Distributions, 15, 289–298, Journal compilation © 2008 Blackwell Publishing Ltd 297 J. S. Thorn et al. Roos, C. (2003) Molekulare Phylogenie der Halbaffen, Schlankaffen, und Gibbons [diss.]. University of Munich, Munich, Germany. Schulze, H. & Groves, C.P. (2004) Asian lorises: taxonomic problems caused by illegal trade. Conservation of primates in Vietnam, pp. 33 –36. Frankfurt Zoological Society, Frankfurt, Germany. Sergio, C., Figueira, R., Draper, D., Menezes, R. & Sousa, A.J. (2007) Modelling bryophyte distribution based on ecological information for extent of occurrence assessment. Biological Conservation, 135, 341–351. Solano, E. & Feria, T.P. (2007) Ecological niche modelling and geographic distribution of the genus Polianthes 1. (Agavaceae) in Mexico: using niche modelling to improve assessments of risk status. Biodiversity and Conservation, 16, 1885–1900. Streicher, U. (2004a) Seasonal changes in colouration and fur patterns in the pygmy slow loris (Nycticebus pygmaeus). Conservation of primates in Vietnam (ed. by T. Nadler, U. Streicher and U. Ha Thang Long), pp. 29 –32. Frankfurt Zoological Society, Frankfurt, Germany. Streicher, U. (2004b) Confiscated primates – health aspects and long-term placement options. Conservation of primates in Vietnam (ed. by T. Nadler, U. Streicher and H. Thang Long), pp. 154–160. Frankfurt Zoological Society, Frankfurt, Germany. Strier, K.B. (2007) Conservation. Primates in perspective (ed. by C.J. Campbell, A. Fuentes, K.C. Mackinnon, M. Panger and S.K. Bearder), pp. 496–509. Oxford University Press, Oxford, UK. Sutherland, W.J. (2000) The conservation handbook: research, management and policy. Blackwell Science, Oxford, UK. Tutin, C.E.G., Ham, R.M., White, L.J.T. & Harrison, M.J.S. (1997) The primate community of the Lope Reserve, Gabon: diets, responses to fruit scarcity, and effects on biomass. American Journal of Primatology, 42, 1–24. Villa, F. & Mcleod, H. (2002) Environmental vulnerability indicators for environmental planning and decision making: 298 guidelines and applications. Environmental Management, 29, 335–348. Voris, H.K. (2000) Maps of Pleistocene sea levels in Southeast Asia: shorelines, river systems and time durations. Journal of Biogeography, 27, 1153–1167. Ward, D.F. (2007) Modelling the potential geographic distribution of invasive ant species in New Zealand. Biological Invasions, 9, 723–735. Webber, C. & Nekaris, K.A.I. (2004) Survey of primates and other mammals in markets of Indonesia, with an analysis of conditions and health of the animals. Folia Primatologica, 75, 60. Williams, P.H. (1998) Key sites for conservation: area-selection methods for biodiversity. Conservation in a changing world: integrating process into priorities for action (ed. by G. Mace, A. Balmford and J.R. Ginsberg), pp. 211–249. Cambridge University Press, Cambridge, UK. Wilson, K., Pressey, R.L., Newton, A., Burgman, M., Possingham, H. & Weston, C. (2005) Measuring and incorporating vulnerability into conservation planning. Environmental Management, 35, 527–543. Editor: Mathieu Rouget SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Appendix S1 Environmental layers used in distribution modelling. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article. © 2008 The Authors Diversity and Distributions, 15, 289–298, Journal compilation © 2008 Blackwell Publishing Ltd