Geographical variation in and evolutionary history of the Sunda clouded leopard (Neofelis diardi) (Mammalia: Carnivora: Felidae) with the description of a new subspecies from Borneo

Molecular Phylogenetics and Evolution 58 (2011) 317–328

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier .com/ locate /ympev

Geographical variation in and evolutionary history of the Sunda clouded leopard(Neofelis diardi) (Mammalia: Carnivora: Felidae) with the description of a newsubspecies from Borneo

Andreas Wilting a,⇑,1, Per Christiansen b,1, Andrew C. Kitchener c,d,1, Yvonne J.M. Kemp e, Laurentius Ambu f,Jörns Fickel a

a Leibniz Institute for Zoo and Wildlife Research, Alfred-Kowalke-Str. 17, 10315 Berlin, Germanyb University of Aalborg, Department of Biotechnology, Chemistry, and Environmental Engineering, Sohngaardsholmsvej 57, DK-9000 Aalborg, Denmarkc Department of Natural Sciences, National Museums Scotland, Chambers Street, Edinburgh EH1 1JF, UKd Institute of Geography, School of Geosciences, University of Edinburgh, Drummond Street, Edinburgh EH8 9XP, UKe VU University Amsterdam, Institute of Ecological Science, Department of Animal Ecology, De Boelelaan 1085, 1081 HV, Amsterdam, The Netherlandsf Sabah Wildlife Department, Block B, Wisma MUIS, 88000 Kota Kinabalu, Sabah, Malaysia

a r t i c l e i n f o

Article history:Received 3 August 2010Revised 19 October 2010Accepted 3 November 2010Available online 11 November 2010

Keywords:BiogeographyHolotypePleistoceneSunda shelfTaxonomyToba volcanic eruption

1055-7903/$ - see front matter � 2010 Elsevier Inc. Adoi:10.1016/j.ympev.2010.11.007

⇑ Corresponding author.E-mail address: [email protected] (A. Wilting).

1 Contributed equally.

a b s t r a c t

Recent morphological and molecular studies led to the recognition of two extant species of clouded leop-ards; Neofelis nebulosa from mainland southeast Asia and Neofelis diardi from the Sunda Islands of Borneoand Sumatra, including the Batu Islands. In addition to these new species-level distinctions, preliminarymolecular data suggested a genetic substructure that separates Bornean and Sumatran clouded leopards,indicating the possibility of two subspecies of N. diardi. This suggestion was based on an analysis of onlythree Sumatran and seven Bornean individuals. Accordingly, in this study we re-evaluated this proposedsubspecies differentiation using additional molecular (mainly historical) samples of eight Bornean and 13Sumatran clouded leopards; a craniometric analysis of 28 specimens; and examination of pelage mor-phology of 20 museum specimens and of photographs of 12 wild camera-trapped animals. Molecular(mtDNA and microsatellite loci), craniomandibular and dental analyses strongly support the differentia-tion of Bornean and Sumatran clouded leopards, but pelage characteristics fail to separate them com-pletely, most probably owing to small sample sizes, but it may also reflect habitat similarities betweenthe two islands and their recent divergence. However, some provisional discriminating pelage charactersare presented that need further testing. According to our estimates both populations diverged from eachother during the Middle to Late Pleistocene (between 400 and 120 kyr). We present a discussion on theevolutionary history of Neofelis diardi sspp. on the Sunda Shelf, a revised taxonomy for the Sunda cloudedleopard, N. diardi, and formally describe the Bornean subspecies, Neofelis diardi borneensis, including thedesignation of a holotype (BM.3.4.9.2 from Baram, Sarawak) in accordance with the rules of the Interna-tional Code of Zoological Nomenclature.

� 2010 Elsevier Inc. All rights reserved.

1. Introduction

Clouded leopards are the most elusive of pantherine felids andeven today very little is known about their ecology and status inthe wild (e.g. Grassman et al., 2005; Wilting et al., 2006). Recently,analyses of molecular (Buckley-Beason et al., 2006; Wilting et al.,2007a) and morphological data (Christiansen, 2008; Kitcheneret al., 2006) demonstrated that Bornean and Sumatran cloudedleopards are clearly distinct from those on the continental main-land. This led to a taxonomic revision of clouded leopards into

ll rights reserved.

two separate species, N. nebulosa (mainland southeast Asia), andN. diardi (Borneo and Sumatra, including the Batu Islands). In2008 this taxonomic revision was adopted by the IUCN Red Listof Threatened Species and both species are now listed separatelyas Vulnerable in the current assessment (Sanderson et al., 2008for N. nebulosa, Hearn et al., 2008a for N. diardi).

In a previous study mtDNA and microsatellite genotype differ-ences between Bornean and Sumatran clouded leopards suggestedthe possible distinction of two subspecies of N. diardi (Wiltinget al., 2007a,b). This was provisionally supported by craniomandib-ular and dental analyses (Christiansen, 2008), but the small num-ber of individuals, especially in the molecular analysis(NBorneo = 7, NSumatra = 3), was not sufficient to further substantiatethis suggestion. Moreover, the proposed names for the two

318 A. Wilting et al. / Molecular Phylogenetics and Evolution 58 (2011) 317–328

subspecies did not follow the rules of the International Code ofZoological Nomenclature (ICZN), henceforth the Code, and hencea formal description is required to establish a valid scientific namefor the Bornean clouded leopard.

In this paper we present a new, extended phylogenetic analysiswith additional individuals from Borneo and Sumatra, combinedwith analyses of morphological (craniomandibular, dental, andpelage) diversity within populations of Neofelis diardi. The aim ofthis study was to provide a thorough understanding of geographi-cal variation in Neofelis diardi, some insights into its evolutionaryhistory and a systematic revision, including the potential for theformal recognition of subspecies.

2. Materials and methods

2.1. Molecular analysis

In addition to the samples used in Wilting et al. (2007a), we col-lected 22 additional epithelial (from skulls or skins) or maxillo-tur-binal bone tissue samples from Neofelis spp. (18 N. diardi and 4 N.nebulosa) from natural history museums, and three fecal samplesfrom wild-born Bornean clouded leopards kept in the Lokawi Wild-life Park in Sabah, Malaysia (see Appendix 1). We extracted DNA inan isolated ‘ancient DNA laboratory’. Epithelia (30–50 mg) wereminced and turbinates (20–40 mg) were fragmented. For theextraction we followed the protocol of Wisely et al. (2004), butprecipitated the DNA with isopropanol, followed by a washing stepwith 70% ethanol. The dried pellets were dissolved in 200 ll deion-ized sterile water and DNA concentrations were measured spectro-photometrically at 260 nm (ND1000, Peqlab GmbH, Erlangen,Germany).

2.1.1. Mitochondrial DNA (mtDNA) analysisWe amplified segments of control region (426 bp), ATPase-8

(134 bp) and Cyt-b (286 bp), previously used in Wilting et al.(2007a), but modified the ATPase-8 forward primer to [50–ATGCCA-CAGCTAGATACATCC–3]. PCR reactions were performed in 20 ll,containing 4 ll 5� GoTaq PCR Buffer (Promega GmbH, Mannheim,Germany), 2 mM MgCl2, 0.2 mM dNTPs, 1 lM of each primer, 1unit of GoTaq polymerase (Promega) and �50–100 ng of genomicDNA. The reactions were performed with an initial denaturationstep at 95 �C for 3 min, followed by 45 cycles of denaturation at94 �C for 30 s, annealing at 56 �C for 45 s, elongation at 72 �C for45 s, and were completed with a final elongation step at 72 �C for10 min. Excess oligonucleotide primers and dNTPs were removedby incubation with exonuclease I and calf-intestine alkaline phos-phatase (ExoCIAP, Fermentas GmbH, St. Leon-Rot, Germany).Amplicons were directly sequenced bidirectionally using the Big-Dye� Terminator kit (v.1.1) and analysed on an A3130xl automatedsequencer (both Applied Biosystems Deutschland GmbH, Darms-tadt, Germany).

Sequences were assembled, aligned and edited using ClustalX2software (Larkin et al., 2007). Sequences from all three mtDNAfragments were concatenated and trimmed to identical lengths(849 bp) to suit the lengths of additional sequences from Genbank(see Wilting et al., 2007a for accession numbers). The entire dataset consisted of 90 sequences (mainland: N = 58, Borneo: N = 15,Sumatra: N = 16, outgroup: N = 1). To select the best-fitting nucle-otide substitution model for the full data set, we used the hierar-chical likelihood ratio test approach implemented in the softwarejModelTest (v.0.1.1; Posada, 2008). The model selected for the dataset was the K81 model (Kimura, 1981) with an allowance both forinvariant sites (I) and a gamma (G) distribution shape parameter(a) for among-site rate variation (K81 + I + G). Parameter valuesfor the model selected were: -lnL = 1790.7657, I = 0.183, and

a = 0.011 (2 gamma rate categories). The phylogenetic reconstruc-tions based on these parameters were then performed applying themaximum likelihood (ML) (Swofford, 2001) approach under a heu-ristic search scenario with TBR branch swapping and the neighborjoining (NJ) method (Saitou and Nei, 1987) both implemented inPAUP (v. 4.0b10; Swofford, 2001). Support for nodes was assessedby a reliability percentage after 100 (ML), respectively 1000 (NJ)bootstrap iterations. Sequences were also analysed using BayesianInference (BI) as implemented in MrBayes (v.3.1.2; Huelsenbeckand Ronquist, 2001). Posterior probabilities for the BI were deter-mined by running three heated chains (default temperature set-ting: 0.2) and one cold chain for 1 million generations (Ronquistand Huelsenbeck, 2003). The parameters of the optimal model se-lected by jModelTest were specified as priors. Each analysis wasrun twice and trees were sampled every 100 generations. Stabilityof likelihood convergence was determined using the hsumpi com-mand in MrBayes, leading to the exclusion of the first 30,000 sam-ples as burn-in when convergence diagnostics were calculated.Posterior probabilities for nodes were based on the remainingtopologies. The domestic cat (Felis catus) sequence served as out-group (Accession number U20753).

We computed genetic diversity within and among the differentgroups (N. nebulosa, N. diardi (Borneo) and N. diardi (Sumatra) byestimating pairwise population FST–values (Cockerham and Weir,1993) and applied different hierarchical analyses of molecular var-iance (AMOVA; Excoffier et al., 1992) to estimate the amount ofpopulation genetic structure. Both tests are implemented in Arle-quin v.3.5 (Excoffier et al., 2005). Because we used concatenateddata sets (see above) with potentially differing selective pressuresamong the various mitochondrial loci (Lopez et al., 1997), we car-ried out Tajima’s D test (Tajima, 1989) to investigate whether theconcatenated data set could be treated as a selectively neutrallyevolving unit or not.

The demographic history of clouded leopards was inferred byseveral approaches. Firstly, we used the mismatch distribution(distribution of numbers of site differences j between each pairof sequences in the populations; Li, 1977, Rogers, 1995, Rogersand Harpending, 1992) to calculate the time t since potentialstep-wise expansion of a relatively small, but constant populationat size h0 to a large population at size h1 over t generations (muta-tional units) in the past with t = s/2l (s: age of expansion, l: muta-tion rate). We applied a generalized non-linear least-squareapproach implemented in Arlequin v.3.5. (Excoffier et al., 2005)and the simplifying assumption of post-expansion population sizebeing infinite (Rogers, 1995). The sum of squared deviations (SSD;Schneider and Excoffier, 1999) and Harpending’s raggedness index(Harpending, 1994) were computed to test goodness-of-fit of theobserved mismatch distribution to that expected under the suddenpopulation expansion model. Parametric bootstrapping (10,000permutations) was applied to obtain confidence intervals aroundall estimated parameters (Excoffier et al., 2005, Schneider andExcoffier, 1999). Displayed graphically, recent sudden populationexpansions or bottlenecks will generate unimodal graphs, whereasstable or slowly declining populations will generate a variety ofmultimodal distributions, reflecting the highly stochastic shapeof gene-trees within populations at demographic equilibrium(Rogers and Harpending, 1992). The mutation rate l, used to calcu-late the age of expansion, was 2.2 � 10�9/site � year (Kumar andSubramanian, 2002). As a second approach to explore the demo-graphic history of clouded leopard populations, we computedFu’s FS statistics (Fu, 1997; 1000 simulations) to detect an excessof low-frequency alleles in a growing population as compared withthe expected number of alleles in a stationary population, wherebysignificantly large negative values indicate population expansion(Fu, 1997, Excoffier and Schneider, 1999). Because assessment ofdemographic history based on mismatch distribution may be

Table 1Pairwise FST (mtDNA) and RST (microsatellite loci) values among the three cloudedleopard populations; N. nebulosa (Mainland), N. diardi (Borneo) and N. diardi(Sumatra).

RST/FST Mainland Borneo Sumatra

Mainland – 0.986 0.986Borneo 0.721 – 0.804Sumatra 1.018 0.202 –

A. Wilting et al. / Molecular Phylogenetics and Evolution 58 (2011) 317–328 319

biased in samples with high Nhaplotypes/Nindividuals � ratios (R) (Fic-kel et al., 2008), we also estimated the divergence time by dividingthe net-between-clade-means genetic distances calculated usingMega v.4.1 (Tamura et al., 2007) following the jModelTest recom-mendations (Table 3). As calibration point we used the split be-tween Panthera and Neofelis at 6.37 million years ago (Johnsonet al., 2006). The estimated net-between-clade-means distance of0.29972 translated into a mutation rate of 29.972% (SE = 9.255%).

2.1.2. Microsatellite analysisTo genotype individuals we used 12 felid dinucleotide microsat-

ellite primers (FCA8, FCA23, FCA43, FCA45, FCA77, FCA82, FCA105,FCA126, FCA132, FCA144, FCA261, FCA310; Menotti-Raymondet al., 1999). One primer of each pair was 50-labeled with a fluores-cent dye (6-FAM or HEX). Based on allele size range and fluorescentdye used we were able to run three multiplex PCRs with primersfor four loci (set1: FCA8, 126, 132 and 261; set2: FCA23, 82, 144,310; set3: FCA43, 45, 77 and 105). Amplification products weresized on an A3130xl automated sequencer, using the ROX500 inter-nal sizing standard (both Applied Biosystems). Only DNA samplesthat produced congruent results in two independent PCR reactionsand successfully amplified at least at nine out of the 12 loci wereincluded in the final microsatellite analysis. Tests for genotypic dis-equilibrium between loci, analysis of molecular variance (AMOVA)and the estimation of RST-values (Slatkin, 1995) were performedusing the software package Arlequin v.3.5 (Excoffier et al., 2005).A Bayesian clustering method, implemented in STRUCTURE(v.2.3, Pritchard et al., 2000), was used to infer population structureand assignment of individuals to populations based on allelic geno-types. A series of tests was conducted assuming different numbersof population clusters (K = 1–9) to guide an empirical estimate ofthe number of identifiable populations, assuming an admixturemodel with correlated allele frequencies. To determine the appro-priate burn-in and run lengths for reliable parameter estimates of Pand Q, we set K = 1 and watched for the likelihoods to converge un-der various burn-in and run lengths. The final burn-in and runlengths were then 100,000 and 200,000 Markov chains, respec-tively. We ran ten independent runs for each K and its associatedparameter set to verify the consistency of estimates across runs.For presentation of the assignments, we used the mean of theten runs and the standard deviation (SD). A factorial correspon-dence analysis (Clausen, 1998; Greenacre and Degos, 1977) wascarried out to examine the relationship between clouded leopardsand allele frequencies using the software package GENETIX(v.4.05.2, Belkhir et al., 1996–2004) under the ‘3D by populations’setting.

2.2. Morphological analysis

2.2.1. Craniomandibular and dental analysisCraniomandibular and dental diversity within Neofelis diardi

was assessed using a sample of 28 adult specimens, of which 18came from Sumatra (10#; 8$) and 10 from Borneo (5#; 5$). Foreach specimen 64 craniomandibular and dental measurementsand three angular variables were analysed (see Christiansen,2008). The samples were analysed with multivariate analyses(Principal Components Analysis [PCA]; and step-wise DiscriminantAnalysis [DA] with subsequent jack-knifed classification analysis)on raw variables, and with bivariate comparisons of craniomandib-ular and dental ratios. Ratios were arcsine transformed prior toanalysis to restore normality (Sokal and Rohlf, 1995), but angularvariables were analysed without transformation.

2.2.2. Pelage analysisSamples of pelages of Sumatran (N = 15) and Bornean (N = 8)

clouded leopards were scored according to eight pelage characters

from Kitchener et al. (2006), although not all pelages yielded scoresfor all characters. Of these, coloration was scored as three subcat-egories for the presence/absence of yellow, grey and tawny. Thescores for each pelage character were summed for each pelage togive a total pelage score (Borneo N = 6; Sumatra N = 14). In addi-tion, similar data were scored, where possible, from camera-trapphotographs of both Sumatran (N = 4) and Bornean (N = 8) cloudedleopards, but owing to limited viewing angles, data were incom-plete for most specimens. Therefore complete data were availablefor totals of 16 Sumatran and seven Bornean clouded leopardsbased on both pelages and camera-trap photographs. Differencesin mean pelage character scores and total pelage scores forBornean and Sumatran samples were assessed using two-samplet-tests with a probability of significance of p = 0.05, using the PASTstatistical package (v.1.99; Hammer et al., 2001).

3. Results

3.1. Mitochondrial DNA analysis

A total of 22 additional individuals were included in the mtDNAanalysis (1 N. nebulosa, 8 N. diardi – Borneo, 13 N. diardi – Sumatra,the latter including the neotype for N. diardi (see Christiansen,2009) and one specimen from the Batu Islands). Besides knownNeofelis haplotypes (Wilting et al., 2007a), we identified sevennew haplotypes for clouded leopards (1 for N. nebulosa, 1 for N. dia-rdi – Borneo and 5 for N. diardi – Sumatra) (Genbank accessionnumbers HM748835 – HM748855).

The enlarged sample of N. diardi confirmed 38 out of 39 diag-nostic sites between N. nebulosa and N. diardi (Wilting et al.,2007a) and discarded only one site (position 802 of Wilting et al.,2007a; where the new haplotype DIB6 from Borneo shared anadenosine [A] with N. nebulosa). The four nucleotide differences be-tween Bornean and Sumatran clouded leopards described in Wilt-ing et al. (2007a) at position 204, 234 (Cyt-b), 420 (ATPase-8) and610 (control region) remained fixed.

MtDNA diversity among the three clouded leopard populations(mainland, Borneo and Sumatra) revealed a very low nucleotidediversity for N. nebulosa (p = 0.00057) compared with N. diardi(p = 0.0037) and moderate nucleotide diversities of p = 0.00126and p = 0.00142 for the populations in Borneo and Sumatra,respectively. AMOVA demonstrated that 98.2% of the genetic vari-ability was due to the variance among populations, and only 1.8%was accounted for by differences within the three populations.Pairwise FST comparisons (Table 1) showed that each of the threepopulations was significantly differentiated from each of the othertwo (p < 0.0001).

Phylogenetic analysis of mtDNA haplotypes, using neighborjoining (NJ), maximum likelihood (ML) and Bayesian inference(BI) approaches, generated congruent topologies that strongly sup-ported the monophyletic status of N. diardi with high bootstrapvalues (100% NJ, 99% ML, 0.96 BI) (Fig. 1). Furthermore, the mono-phyletic statuses of Bornean and Sumatran clouded leopards werealso supported with high bootstrap values (Fig. 1). The neotype ofN. diardi shared its haplotype (DIS3) with four other Sumatran indi-viduals and with the specimen from the Batu Islands.

Fig. 1. Phylogenetic relationships among clouded leopards inferred from mtDNA haplotypes from the concatenated 849 bp mitochondrial sequences. Trees for each of thethree methods (NJ/ML/BI) had similar topologies. Numbers above the branches represent bootstrap support, only values >50% are shown. Numbers in parentheses representthe number of individuals sharing the same haplotype. Haplotype codes are shown in Appendix 1. NEB 1 – 6, DIB 1 – 5, and DIS 1 and 2 have been described previously(Wilting et al. 2007a). ⁄Haplotype of the Neotype of N. diardi and the Batu Island specimen; # Haplotype of the specimens from Sarawak.

320 A. Wilting et al. / Molecular Phylogenetics and Evolution 58 (2011) 317–328

Mismatch distributions, generated for all three distinct popula-tions (N. nebulosa, N. diardi – Borneo and N. diardi – Sumatra), re-vealed smooth, unimodal distributions (statistics in Table 2),indicating sudden population expansion (Rogers and Harpending,1992). However, analyses which contained more than one popula-tion, Neofelis spp. or N. diardi, showed multimodal patterns(Fig. 2A). These multimodal patterns would generally be inter-preted as a stable or slowly declining population. In our case it isclearly a result of the inclusion of different distinct populationsin the same analysis. The different peaks can be explained as differ-ences between the populations, as the separate analyses of eachpopulation showed unimodal patterns. These results are a goodexample of where the amalgamation of distinct populations biasesthe result of the mismatch distribution analysis, which may lead toincorrect conclusions about the stability of populations. Therefore,we suggest that prior to applying a mismatch distribution analysis,a check should be made to ensure that all individuals belong to asingle population.

Based on the mismatch distribution (Rogers, 1995), the approx-imate divergence time (t = s/2l) of the Bornean and Sumatran pop-ulations of N. diardi was estimated to be �331 thousand years [kyr]

(CI95%: 60–640 kyr) and 396 kyr (CI95%: 160–700 kyr), respectively(Table 2). Using the calibration date derived from Johnson et al.(2006), we estimated a Bornean and Sumatran populations splitat approximately 117 kyr (Table 3).

3.2. Microsatellite analysis

Composite genotypes from at least nine of the 12 felid-specificmicrosatellite loci could only be obtained for a subset of 21clouded leopard samples (five mainland specimens, seven Borneanspecimens and nine Sumatran specimens, including the neotypefor N. diardi and one individual from the Batu Islands; see Appendix1). There was no significant linkage disequilibrium among loci,indicating their independent inheritance. The microsatellite AMO-VA demonstrated that 37% of the genetic variability was attribut-able to variance among populations, and 63% was accounted forby differences within the three populations. Each of the three pair-wise population comparisons showed highly significant populationgenetic differentiation (p < 0.0001) by pairwise RST’s (Table 1).

The STRUCTURE analysis yielded the following mean values(across 10 runs) of posterior probabilities for the number of

Table 2Analyses performed on clades of Neofelis.

Species Neofelis nebulosa N. diardi N. d. borneensis N. d. diardi

ParameterNumber of individuals N 58 31 15 16Fragment length/ number of usable sites [nc]a 849/849 849/846 849/846 849/849Diversity indicesNumber of haplotypes h 7 13 6 7Number of segregating sites S 6 11 5 4Transition/ transversion ratio 5/1 10/1 4/1 4/0Nucleotide diversity p (SD) 0.00057 (0.00055) 0.0037 (0.00219) 0.00126 (0.00099) 0.00142 (0.00112)Haplotype diversity Hd (SD) 0.227 (0.073) 0.877 (0.032) 0.762 (0.096) 0.833 (0.072)Ratio R = h/N 0.121 0.419 0.4 0.375

Tajima’s test of selective neutralityTajima’s D �1.826 0.438 �1.033 0.18p (Dsimulated < Dobserved) [1000 simulations]c 0.01 0.723 0.158 0.621

Mismatch distributionAverage number of nucleotide differences j 0.339 3.131 1.067 1.275Variance of j 0.658 4.856 0.621 0.672s 3.00 5.795 1.234 1.469

Test of goodness-of-fitSum of squared deviation (SSD) 0.002 0.048 0.024 0.025p (SSDsimulated P SSDobserved)d 0.47 0.2 0.18 0.13Harpending’s raggedness index 0.378 0.094 0.175 0.17p (Ragsimulated P Ragobserved)d 0.62 0.33 0.13 0.07

Age of cladebased on s [kyr] 803 1556 331 396CI95% [kyr] 100 – 1090 27 – 2690 62 – 640 158 – 692

Sudden population expansionFS �4.523b �2.005 �2.61b �3.204b

p (FS) 0.004 0.203 0.016 0.011

nc: nucleotides; SD: standard deviation; s (tau): units of mutational time; t = s/2l; mutation rate l = 2.2 � 10�9 per site and year (Kumar and Subramanian 2002); n.c.: notcalculable, CI95%: confidence interval at a = 0.05; FS: Fu’s statistics (1000 simulations), kyr: 1000 years.

a Uncertain positions (‘‘N’’) were removed prior to analyses.b FS-values are significant (95% level) at p < 0.02 (Fu 1997).c Null-hypothesis H0 is hnon-neutralityi.d H0 is hno fiti.

Fig. 2. Hierarchical model of the clouded leopard populations. (A) Mismatch distributions computed for the concatenated 849 bp of mitochondrial sequences. Solid line:observed distribution of pairwise differences; dashed line: expected distribution under the model of sudden demographic expansion. Numbers in parentheses represent thenumber of individuals analysed. (B) Genotypic assignment of the individual clouded leopards to 2–4 clusters (populations). Posterior probabilities for the number ofpopulations, given as ln Pr(X|K) for K = 2–4, and admixture coefficients (q) are shown. ID codes are shown in Appendix 1, NDD5 = Neotype N. diardi, NDD9 from Batu Islands,NDB5 and NDB7 from Sarawak. � NDD9 q = 0.72. Numbers in parentheses represent the number of individuals analysed in each cluster.

A. Wilting et al. / Molecular Phylogenetics and Evolution 58 (2011) 317–328 321

Table 3Coalescent times based on net genetic distances.

Split Pantheraa/Neofelis N. nebulosa/N. diardi N. d. borneensis/N. d. diardi

ParameterSubstitution model used Tamura–Nei Tamura–Nei Tamura–NeiGamma distribution parameter a 0.011 0.1 0.011d (%) 29.972 8.748 0.550SE (%) 9.255 2.563 0.300Time (kyr) 6370b 1859 117CI95% (kyr) 2436–10,300 769–2948 0–244

a Comprises of Panthera tigris, P. pardus, P. onca; d: net distance between clades, calculated by subtracting the mean intra-clade distances(dX, dY) from the mean inter-clade distances (dXY) using the equation: d = dXY � 1/2(dX + dY) (Nei 1975; Wilson et al. 1985; Nei 1987); CI95%:confidence interval (at 95%) = d ± 2 � SE.

b Calibration date from Johnson et al. (2006); kyr: thousand years. Mutation rates were 29.972% per million years.

322 A. Wilting et al. / Molecular Phylogenetics and Evolution 58 (2011) 317–328

populations (K): ln Pr(X|K) = �782 (SD = 0.77), �638 (SD = 0.26),�587 (SD = 0.35), �567 (SD = 0.77), �598 (SD = 15.01), �612(SD = 4.32), �635 (SD = 13.66), �651 (SD = 22.75), and �675(SD = 20.03), for K = 1–9, respectively. The probabilities were high-est for K = 3 and K = 4 (underlined). For all K > 4, posterior probabil-ities also varied largely among the ten runs, resulting in largestandard deviations. Based on these results we designed a hierar-chical model with an increasing K from 2 to 4 (Fig. 2B). Weassumed a separation of the two clouded leopard speciesN. nebulosa and N. diardi at K = 2. This assumption was supportedby the analysis and we observed high admixture coefficients ofq > 0.96 and q > 0.99 for N. nebulosa and N. diardi, respectively. AtK = 3, not only N. nebulosa and N. diardi were separated, but alsothe two populations in Borneo and Sumatra, and the high admix-ture coefficients of q > 0.96 (in all but one run) indicated no to verylittle gene flow between groups. At K = 4, Sumatran clouded leop-ards were further separated in two subgroups, indicating a sub-structure below the proposed subspecies level. The loweradmixture coefficients of q > 0.75 indicated the existence (albeitlow) of gene flow between the structural units (populations)(Fig. 2B). The samples from the Batu Islands (NDD9) and the neo-type for N. diardi (NDD6) mingled with the Sumatran individualswith high admixture coefficients.

The results of the factorial correspondence analysis yielded twofactorial components, of which component 1 explained 68% of thevariation among the clouded leopard populations and component2 the remaining 32% (Fig. 3). Neofelis nebulosa and N. diardi wereclearly distinct from each other, as were, to a lesser extent, Borneanand Sumatran clouded leopards. Each of the three populations

Fig. 3. Factorial correspondence analysis of genotype distributions of cloudedleopard populations based on allele frequencies of 12 microsatellite loci. FC = Fac-torial Component.

formed a single common cluster. Both the neotype for N. diardiand the sample from the Batu Islands clustered with the Sumatransamples.

3.3. Craniomandibular and dental analysis

In PCA analysis, N. diardi males from Borneo and Sumatrashowed a distinct degree of separation (Fig. 4a), but for femalesthere was no discernible separation of the populations on eitherPC1 or PC2 (Fig. 4b). In traditional classification analyses, all in-cluded specimens were assigned correctly to their respective taxa.

Fig. 4. Plots of the first two principal components from analyses of 64 cranioman-dibular and dental measurements and three angular variables. (a) Neofelis diardimales; (b) Neofelis diardi females. Symbols: B, specimens of Neofelis diardi fromBorneo; S, specimens of Neofelis diardi from Sumatra. The arrows indicate typespecimens of Neofelis diardi spp.: RMNH1981 (#; National Museum of NaturalHistory, Leiden), neotype of Neofelis diardi and holotype of Neofelis diardi diardi;BM.3.4.9.2 ($; Natural History Museum, London), holotype of Neofelis diardiborneensis.

A. Wilting et al. / Molecular Phylogenetics and Evolution 58 (2011) 317–328 323

However, the more robust jack-knifed procedure implied that onlyone sample was identified with 100% accuracy, owing to the mor-phology of the subspecies being distinct but not markedly so. InDA, N. diardi males proved non-significantly different albeit nearlyso (Wilks’ k = 0.032; F = 7.892; p = 0.0543), but nonetheless, a jack-knifed classification analysis classified Bornean males with 100%accuracy, whereas four males from Sumatra were misclassified(60% correct). Females were also non-significantly different, albeitnearly so (Wilks’ k = 0.014; F = 14.186; p = 0.0576). One of the Bor-nean females was misclassified (80% correct), and three of theSumatran females were also misclassified (63% correct). Accord-ingly, multivariate studies indicate that the Bornean and Sumatranpopulations of N. diardi are craniodentally distinguishable withhigh classification accuracy. Bivariate analyses on ratio variables(see Christiansen, 2008) corroborated the above. Of 125 analysedcraniomandibular and dental ratios, 11 were significantly different(p < 0.05) between the populations of N. diardi from Borneo andSumatra (Table 4).

3.4. Pelage analysis

Only one mean individual pelage character score between Bor-nean and Sumatran clouded leopards was significantly different(Table 5). Spots within clouds showed a higher mean score in Bor-nean (N = 16; mean = 2.75) than Sumatran (N = 19; mean = 2.26)clouded leopard pelages (t = 2.219; p = 0.034) (other, non-signifi-cant results not presented here). These spots do appear more fre-

Table 4Bivariate comparisons of craniomandibular and dental ratios in Neofelis diardi(populations from Borneo and Sumatra); see Christiansen (2008) for a detaileddescription of characters. Underlined values are 0.100 P p P 0.050, and values inbold are p < 0.050. Abbreviations: AP, anteroposterior diameter; CBL, condylobasalskull length.

Character

t p

Nasal width at narial aperture/CBL (0.072)Nasal width at narial aperture/snout width (0.003)Postorbital constriction/postorbital processes (0.085)Palate length/CBL (0.027)Width of pterygoid palate/CBL (0.094)Width of pterygoid palate/width of incisors (0.005)Width of pterygoid palate/width between C1 (0.003)Width of pterygoid palate/ width of palate at P3

(0.087)Width of pterygoid palate/ width of palate at P4 (0.033)Width of incisors/palate length (0.015)Mastoid width/CBL (0.048)P4 paracone length/P4 length (0.032)P4 protocone AP/metastyle length (0.059)P4 protocone AP/P4 length (0.012)Width of P4 at protocone/ P4 length (0.063)P4 protoconid length/P4 length (0.016)P4 protoconid height/P4 length (0.049)

Table 5t-tests on mean scores for pelage characters of Neofelis diardi from Borneo andSumatra. Only cloud spots showed a significant difference, with Bornean animalsdisplaying more cloud spots on average than Sumatran animals.

Character t p N – Borneo N – Sumatra

Cloud spots 2.219 0.034 16 19Lightness 1.245 0.222 16 19Brightness 0.380 0.706 16 19Tawny 0.672 0.506 16 19Grey 0.269 0.789 16 19Nape stripes 0.163 0.872 7 17Shoulder pattern 1.470 0.155 8 17Dorsal stripes 0.699 0.490 13 18

quent and more prominent in the pelages of Bornean animals,but this did not separate all animals. The mean total pelage scorefor Bornean animals was 16.71 (N = 7; standard error (SE) =0.680) compared with 16.19 (N = 16; SE = 0.540) for Sumatranclouded leopards; there was no significant difference in total pel-age scores between the two islands (two-sampled t-test,t = 0.564, N = 23, p = 0.579). Although in strict terms of these char-acters there was no clear separation between Bornean and Suma-tran clouded leopards, careful examination of the sample ofpelages plus the camera-trap photographs suggested that theremay be pelage differences, although these are not 100% diagnosticfor each population.

4. Discussion

The consistent results of molecular (mtDNA and microsatelliteloci), craniomandibular and dental data strongly endorse the previ-ously suggested distinction of Bornean and Sumatran cloudedleopards into two populations with separate evolutionary histo-ries. In the mtDNA tree topology and in the hierarchical model ofthe microsatellite genotypes the degree of separation between Bor-nean and Sumatran clouded leopards should be regarded as the le-vel of different subspecies. Accordingly, the Bornean cloudedleopard should be recognised as subspecifically distinct from theSumatran clouded leopard. The analysis of a single specimen fromthe Batu Islands revealed no genetic traits specific to these islands.This indicates either past continuous gene flow between the BatuIslands and mainland Sumatra, or that clouded leopards had onlyrecently spread to these islands, when land bridges connectedthem with mainland Sumatra until 10 kyr ago (Voris, 2000).

In contrast to the hierarchical models used in the molecularanalysis, the comparative approach used in the morphologicalstudies yielded no systematic predictions as such. Inference of sub-species status to allopatric populations is always going to involvecertain amounts of ambiguity. However, the populations from Bor-neo and Sumatra were clearly distinguishable in bivariate and mul-tivariate studies. The reported overlaps are acceptable within thetraditionally advocated amount of overlapping in subspecies, theso-called 75% rule (Mayr and Ashlock, 1991). Accordingly, basedon craniomandibular and dental analysis alone, the suggestion ofsubspecies status to Bornean and Sumatran populations of N. diardiappears justified.

No significant difference was found between total pelage scoresof Bornean and Sumatran clouded leopards, and only cloud spotsshowed a significant difference among individual pelage charac-ters, with Bornean animals having a higher mean score, but thesample sizes were very small. Given that some overlap in pelagecharacters might be expected between subspecies according tothe 75% rule (Mayr and Ashlock, 1991), the high degree of overlapobserved here could be due to sampling bias alone. There is a sug-gestion that Bornean clouded leopards may have larger, moreangular cloud-like blotches with thicker black borders, and thickerneck and shoulder stripes, as well as more frequent, bolder spots.However, many of the camera-trap photographs were derived fromSabah and these differences may have been local. Therefore, thesepotential differences need to be tested on much larger samplesfrom throughout Borneo in order to provide sufficient data forany potential discriminatory pelage characters.

4.1. Evolutionary history of N. diardi

The mismatch distribution analysis and the very low FS-valuesrevealed a rapid and recent population expansion for the proposedsubspecies of N. diardi. Although our molecular data did not allowdefinite conclusions regarding both the origin and the direction of

324 A. Wilting et al. / Molecular Phylogenetics and Evolution 58 (2011) 317–328

island colonization, the higher FS-values for Sumatra provided ahint that the expansion in Sumatra may have been more recent.In addition, most of the Bornean samples included in the molecularanalysis originated from the northern, Malaysian part of Borneoand the only specimens from Kalimantan (Indonesian, southernpart of Borneo) included here yielded a previously unrecorded hap-lotype (DIB6). Therefore, we might have underestimated the genet-ic diversity of the Bornean population. In this case the scenariowould be that Bornean clouded leopards populated Sumatra duringperiods of low sea levels in the Pleistocene and were later sepa-rated from their source population by rising sea levels. For�100,000 years of the last 250,000 years (Voris, 2000), the sea levelwas at least 40 m below its present level, thereby exposing a landbridge between Borneo and Sumatra.

Based on the mtDNA data, we estimated that the two cloudedleopard subspecies became genetically separated in the Middle toLate Pleistocene, between 400 and 100 kyr depending on the meth-od applied. The use of mtDNA carries some restrictions in node res-olution (Johnson et al., 2006), depending on locus use. However,Tajima’s D indicated that the concatenated dataset used in ourstudy could be treated as a neutrally evolving unit. The discrepancyin our estimates of divergence time between Bornean and Suma-tran clouded leopards in our two models emphasize that the num-bers should be evaluated carefully. Nonetheless, even the morerecent estimate is similar or longer than estimated subspeciesdivergence times in other pantherine cats (72–108 kyr for tigers:Luo et al., 2004, 170–300 kyr for leopards: Uphyrkina et al., 2001).

Considering the confidence intervals around the divergencetime estimates, the split of the two subspecies of N. diardi corre-sponds roughly with the catastrophic ‘‘super-eruption’’ of the Tobavolcano in Sumatra around 73.5 kyr (e.g. Rampino and Self, 1992).This was the second largest explosive eruption known in the Phan-erozoic history (Ambrose, 1998) and an order of magnitude largerthan any other known Quaternary volcanic eruption (Dawson,1992; Huff et al., 1992; Rose and Chesner, 1990). It is assumed thatthe darkness caused by the dust injected in the stratosphere andthe associated volcanic winter (Rampino and Ambrose, 2000;Rampino and Shelf, 1992, 1993) had a major impact on flora andfauna in the Sunda Shelf, in particular in Sumatra. Large carnivores,occurring in low population densities with large home-ranges, areespecially prone to extinction, even more so if they have restrictedgeographical distributions (e.g., Cardillo et al., 2004; Schmidt et al.,2009; Terborgh, 1974). Although the effects of the Toba eruptionhave been discussed controversially (see Louys, 2007; Oppenheimer,2002), and no large mammal species extinction has so far beenlinked to the eruption (Louys, 2007), post-catastrophe recoloniza-tion events in Sumatra may have obscured local extinctions. Sucha situation is conceivable for the evolutionary history of the Sundaclouded leopard, a species with low population densities and largehome-ranges (Hearn et al., 2008a, Mohamed et al., 2009, Wiltinget al., 2006).

A possible scenario is that Sumatra, emptied by the eruption,was, as the vegetation recovered, recolonized by an expandingclouded leopard population surviving the catastrophic event in arefugium. But what was the center of origin for the recoloniza-tion of Sumatra, Peninsular Malaysia or Borneo? Although Peninsu-lar Malaysia lies much closer to Sumatra than Borneo and arecolonization from this area has been reported for example forthe common palm civet Paradoxurus hermaphroditus (Patou et al.,2010), all recent analyses on clouded leopards show unanimouslythat Sumatran clouded leopards are much more closely related toBornean clouded leopards (both N. diardi) than to their continentalrelatives (N. nebulosa). Several reasons, for instance a more sea-sonal and drier habitat between the Peninsular Malaysia andSumatra, or a large river running through the Straits of Malacca,can be excluded as explanations for the failure of mainland N. neb-

ulosa to colonize Sumatra, because the ‘‘savanna corridor’’ (if at allpresent; see Cannon et al., 2009) had a wider extension (up to150 km in width) between Borneo and Sumatra (Bird et al., 2005;Heaney, 1991), and the contemporary Malacca river running be-tween the land masses of Peninsular Malaysia and Sumatra is sup-posed to be discontinuous through the Straits of Malacca (Birdet al., 2005, Heaney, 1991; Meijaard, 2003; Voris, 2000). Althoughseemingly contradictory, the observed geographical distribution ofclouded leopards may be explained by the close proximity of Pen-insular Malaysia to Sumatra. The consequences of the Toba erup-tion were more severe, indicated by thick ash layers (Rose andChesner, 1987), in Peninsular Malaysia than in Borneo, and it canbe assumed that this region was also emptied by the eruption. Con-sequently, the refugia of N. nebulosa would have been locatedmuch further north from the Toba eruption, potentially some-where in northern Indochina, similar to those of other species(Brandon-Jones, 1996; Luo et al., 2004). This hypothesis is sup-ported by our molecular data (very low FS-value and nucleotidediversity and unimodal mismatch distribution), which indicatedthat the mainland clouded leopard as well had gone through a se-vere population bottleneck and only very recently showed a rapidpopulation expansion. A recent expansion from this refugial retreatwould explain why, in contrast to numerous other species inSoutheast Asia (reviewed in Hughes et al., 2003 for birds, Woodruffand Turner, 2009 for mammals), we could not detect any molecularor morphological differences between clouded leopards north andsouth of the known transition zones, i.e. the Isthmus of Kra or fur-ther north where the peninsula joins the mainland. Thus, the arri-val of N. nebulosa in Peninsular Malaysia might have occurredwithin the last 10 kyr, after the land bridges between the continentand Sumatra had submerged again (Voris, 2000).

In contrast, the prevailing southeasterly winds that blew duringthe time after the Toba eruption would have led to less severe conse-quences in Borneo, and, consistent with this, no Youngest Toba Tuffs(YTT) have been found there (Pattan et al., 2001). Therefore, a largerpopulation of N. diardi could have survived in the Bornean refugium(indicated by the greater nucleotide diversity of N. diardi comparedto N. nebulosa, Table 2). The slightly wetter climate just prior toand after the volcanic eruption, 74–47 kyr (van der Kaars and Dam,1995), likely resulted in an expansion of forests and would have facil-itated the geographical expansion of N. diardi to Sumatra.

This scenario begs the question as to why N. diardi did not con-tinue its expansion to Peninsular Malaysia. A hypothesis of initialcolonization of the peninsula by N. diardi and a subsequentreplacement by N. nebulosa is testable only with fossil remains,which are currently wanting. A larger sample from PeninsularMalaysia (only one specimen was included in this study, haplotypeNEB3) would allow a search for introgression between N. diardi andN. nebulosa, and thus could reveal if the two species ever hybri-dised. So far, pelage (Kitchener et al., 2006), craniomandibularand dental data (Christiansen, 2008) fail to provide support forsuch hybridisation.

The similarity of habitats in Borneo and Sumatra, with mainlyevergreen rainforests, might explain the similarity of pelage patternbetween Sumatran and Bornean clouded leopards (Allen et al.,2010). N. nebulosa, which also lives in more open habitats such asdry and deciduous forests and up to the foothills of the Himalayas(up to 3000 m), shows in contrast a distinct pelage pattern withlighter and larger cloud shape markings (Kitchener et al., 2006).However, the preliminary observation of some pelage differencesbetween animals from Sabah and Sumatra need to be tested on awider geographical sample, but might indicate an example of Glog-er’s Rule (Gloger, 1833), where there is a greater degree of melanismin wetter environments. The craniodental differences between thesubspecies observed in this study might be an indication thatclouded leopards in Sumatra had to specialise in terms of their

A. Wilting et al. / Molecular Phylogenetics and Evolution 58 (2011) 317–328 325

ecological niche (e.g. prey choice), owing to sympatry with potentialcompetitors such as tigers, Asian golden cats (Catopuma temminckii)and dholes (Cuon alpinus), while in Borneo other larger carnivoresare absent (the Bornean bay cat Catopuma badia is smaller).

4.2. Systematics of N. diardi

Description and diagnosis for the species N. diardi follows Kitch-ener et al. (2006) and Christiansen (2008) for pelage and osteoana-tomy, respectively. The vernacular name of a species is inphylogenetics and taxonomy often considered to be irrelevant,but many species, in particular mammals and birds, are discussedmainly by their vernacular names and not by their scientific equiv-alent. As such, a vernacular name for a big cat is not irrelevant, butthe Code has no authority on assignments of vernacular names(Article 1.3.5). For N. diardi different vernacular names have beensuggested, such as Sundaland clouded leopard (Wilting et al.,2007a) and Diard’s clouded leopard (Christiansen, 2008), the latterbased on historical precedence (e.g., Audoin et al., 1823; Jardine,1834; Ripley and Dana, 1858). However, we advocate formal adop-tion of the name Sunda clouded leopard, as also used by the IUCNRed List, because this is the most commonly used name for thisspecies in southeast Asia. Also, it refers accurately to the species’origin in the Sunda region, and it enhances awareness by local peo-ple of the importance of this threatened species, which will in turnhopefully strengthen conservation efforts for this species and otherSundaic endemics.

Two subspecies of N. diardi are formally recognised.

4.2.1. Neofelis diardi diardi (Cuvier, 1823)Distribution range: Sumatra and Batu Islands.The original name of Felis diardi is attributable to Cuvier (1823)

based on a specimen allegedly from Java, but this is erroneous

Fig. 5. Holotype of Neofelis diardi borneensis, adult female BM.3.4.9.2 (Natural History Muventral view; (d) mandible in lateral view. Scale bar equals 5 cm.

because clouded leopards are absent from this island, although theywere present in the Neolithic Java (Hemmer and von Koenigswald,1964). In 1827 Coenraad Jacob Temminck published a briefdescription of Stamford Raffles’ Arimau Dahan, which he namedF. macrocelis, and he correctly noted that it was present not onlyon Sumatra, but also on Borneo (Temminck, 1827). Temminckdid in fact discuss the identity of these two forms (1827: p. 103)and was the first to conclude that his Felis macrocelis was not pres-ent on Java, and that the two, accordingly, must be different spe-cies (Temminck, 1827). Cuvier’s specimen was probably fromSumatra instead, as has been suggested by Ellerman and Morri-son-Scott (1951), Corbet and Hill (1992), Kitchener et al. (2006),Christiansen (2009), among others. Thus, Felis macrocelis Tem-minck is a junior synonym of Felis diardi G. Cuvier (ICZN, 2000: arti-cle 23, in particular 23.3). Wilting et al. (2007a) suggested, that iftwo subspecies of N. diardi were formally recognized, the Sumatransubspecies should be called N. d. sumatrensis. Since no holotypewas designated, this name is a nomen nudum and not availableaccording to the Code (Article 15.1). This was amended in Wiltinget al. (2007b). Christiansen (2009) recently designated a neotypefor N. diardi from Palembang, Sumatra (RMNH.1981). The subspe-cies to be given the same trinomen as the species name is fixedby the type specimen (Article 47.1); as such, Neofelis diardi diardiis hereby officially adopted for the Sumatran subspecies of N diardi,as fixed by the species’ neotype specimen. In accordance with theCode (Article 16, specifically 16.4), the type specimen for the sub-species N. d. diardi is hereby fixed as the neotype of the species-taxon, Neofelis diardi, as RMNH1981.

The molecular sequences of the mitochondrial DNA analysis ofthe neotype specimen are available under the Genbank accessionnumbers HM748837 (control region), HM748844 (Cyt-b) andHM748851 (ATPase-8) (mtDNA haplotype DIS3, and NDD6 in themicrosatellite analysis).

seum, London; Sarawak, north-central Borneo); cranium in (a) lateral; (b) dorsal; (c)

Fig. 6. Pelage of the holotype of Neofelis diardi borneensis ssp. nov. (BM.3.4.9.2) (A) dorsal and (B) ventral.

326 A. Wilting et al. / Molecular Phylogenetics and Evolution 58 (2011) 317–328

4.2.2. Neofelis diardi borneensis ssp. novDistribution range: Borneo.Etymology: Described after its origin in Borneo.Diagnosis: Differs from nominal subspecies, N. d. diardi, in the

following craniomandibular and dental characters; greater widthacross the nasal aperture and mastoid processes, and shorter pter-ygoid palate relative to condylobasal skull length; pterygoid palatenarrow; shorter paracone length and narrower across the proto-cone relative to P4 length than in N. d. diardi; and longer and tallerp4 protoconid relative to p4 length than N. d. diardi. Pelage diagno-sis is provisional; more frequent and bolder, cloud spots, larger,more angular cloud-like blotches than in N. d. diardi, which partic-ularly in shoulder region are intermediate in size between those ofN. d. diardi and N. nebulosa. Cloud-like blotches tend to have thickerblack borders, and neck and shoulder stripes tend to be thickerthan in N. d. diardi. Ground color tends towards grey with yellow-ish tinge, whereas Sumatran animals have a tendency towardstawny too. Within the analysed 849 bp of mitochondrial DNA, fourfixed nucleotide differences distinguish it from N. d. diardi. At posi-tions 8788 (ATPase-8), 15241 and 15271 (Cyt-b) and 16,957 (con-trol region) [referring to the domestic cat mitochondrial genome,GenBank accession number U20753] clouded leopards from Bor-neo carry an [C], [T], [T] and [C] respectively, whereas N. d. diardihas [T], [C], [C], and [T] respectively.

Holotype: BM.3.4.9.2 female skin and skull collected 14 May1902 by Charles Hose at Baram, Sarawak, Borneo (see Appendix2 for a description of the holotype).

Referred specimens: ZMB 56310 female skull and skeleton col-lected 1907 by Pagel in Lahad Datu, Sabah, Borneo. The molecularsequences of the mitochondrial DNA analysis of the paratype spec-imen are available under the Genbank accession numbersEF440645 (control region), EF437579 (Cyt-b) and EF437572 (ATP-ase-8) (haplotype DIB1 and NDB6 in the microsatellite analysis).Other referred specimens are listed in Appendix 1 (specimens in-cluded in the molecular analysis), Appendix 3 (craniodental analy-sis) and Appendix 4 (pelage analysis).

Wilting et al. (2007a) conditionally proposed the name Neofelisdiardi borneensis for the Bornean subspecies, but this is a nomen nu-dum, because there was no formal description including designa-tion of a holotype and therefore it was not in accordance with

the ICZN (Article 16.4, specifically 16.4.1, 16.4.2, and Recommen-dation 16C - F). Our findings prompt a holotype designation ofthe Bornean subspecies. Here we designate specimen BM.3.4.9.2,a skull (Fig. 5) and a skin (Fig. 6) of an adult female, collectedMay 14th 1902 by Charles Hose at Baram, Sarawak, Borneo, housedat the Natural History Museum in London. This specimen was se-lected as a holotype, because no specimen in the database ofmolecular data included a skin, skull and mandible. However,two specimens with molecular data also originate from Sarawak(Appendix 1): mtDNA sequences (haplotype DIB1) for one of thesetwo specimens are available under the Genbank accession num-bers EF440645 (control region), EF437579 (Cyt-b) and EF437572(ATPase-8). In the microsatellite analysis the two specimens carry-ing genotypes NDB5 and NDB7 originated from Sarawak. The tri-nominal name of Neofelis diardi borneensis is hereby officiallyrecognised for the Bornean subspecies according to Article 15.1and 16.4 of the ICZN.

5. Conclusion

Following molecular and morphological results presented here,two subspecies of Neofelis diardi are recognised, namely Neofelisdiardi diardi from Sumatra and the Batu Islands, and N. d. borneensisssp. nov. from Borneo. Both subspecies should be protected andmanaged separately to preserve the integrity of their gene poolsand their morphological distinctiveness. The IUCN already recog-nizes these subspecies based on the first suggestions by Wiltinget al. (2007a,b) and lists both as Endangered on the current IUCNRed List (Hearn et al., 2008b – N. d. borneensis, Sunarto et al.,2008 – N. d. diardi). This classification and the recognition of differ-ent subspecies is of utmost importance for conservation and man-agement purposes, as these different conservation units representan important component of the evolutionary legacy of this threa-tened pantherine felid species.

Acknowledgments

AW and JF thank all institutions and persons listed in Appendix1 that supplied the biological specimens this work is based upon.

A. Wilting et al. / Molecular Phylogenetics and Evolution 58 (2011) 317–328 327

We would also like to thank CITES agencies for issuing the specificpermits to the University of Würzburg and the Leibniz Institute forZoo and Wildlife Research. The molecular analysis was funded bythe German Research Foundation (Grant Fi-698/5-1). AW thanksNiko Balkenhol for his assistance to analyse the data, Heike Feld-haar for her help to collect the first data at the University of Würz-burg, and Deike Hesse for valuable comments on earlier drafts ofthis manuscript. PC is indebted to the staffs at the Natural HistoryMuseum in London; Museum national dHistoire Naturelle in Paris;Museum für Naturkunde in Berlin; Naturmuseum Senckenberg,Frankfurt; Staatliches Museum fûr Naturkunde, Stuttgart; Natur-historiska Riksmusset, Stockholm; National Museum of NaturalHistory, Leiden (Naturalis). ACK thanks Paula Jenkins and DaphneHills at the Natural History Museum, London, Chris Smeenk, Natu-ralis, Linda Gordon, National Museum of Natural History, Smithso-nian, Larry Heaney and Dave Willard, Field Museum of NaturalHistory, Chicago, Guy Musser, American Museum of Natural His-tory and Renate Angermann, Zoological Museum Berlin for accessto specimens in their care, and Elizabeth Barrett and Jennifer Wardfor photos of skins from some of these collections, and Sarah Chris-tie, Zoological Society of London, for access to camera-trap photosform Sumatra. YJMK thanks Ronald Vonk, Zoological MuseumAmsterdam, and Dick Roelofs, VU University Amsterdam, for theirsupport. All authors are very grateful to Elsevier for granting a par-tial fee waiver allowing the immediate open access publication ofthis article.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.ympev.2010.11.007.

References

Allen, W.A., Cuthill, I.C., Scott-Samuel, N.E., Baddeley, R., 2010. Why the leopards gotits spots: relating pattern development to ecology in felid. Proc. R. Soc. Bdoi:10.1098/rspb.2010.1734.

Ambrose, S.H., 1998. Late Pleistocene human population bottlenecks, volcanicwinter, and differentiation of modern humans. J. Hum. Evol. 34, 623–651.

Audoin, Bardon, I., Brongniart, A., de Candolle, de Férussac, D., Deshaies,Desmoulins, A., Drapiez, Edwards, Flourens, de Saint-Hilaire, G., Guillemin, deJussieu, A., Kunth, de LaFosse, G., Latreille, Lucas fils, Prevost, C., Richard, A., deSaint-Vincent, B., 1823. Dictionnaire Classique d’Histoire Naturelle, Tome 3. Reyet Gravier, Paris. 592 pp.

Belkhir, K., Borsa, P., Chikhi, L., Raufaste, N., Bonhomme, F., 1996–2004. GENETIX4.05, logiciel sous Windows TM pour la genetique des populations. LaboratoireGenome, Populations, Interactions, CNRS UMR 5000, Universite de MontpellierII, Montpellier, France.

Bird, M.I., Taylor, D., Hunt, C., 2005. Environments of insular Southeast Asia duringthe last glacial period: a savanna corridor in Sundaland? Quaternary Sci. Rev.24, 2228–2242.

Brandon-Jones, D., 1996. The Asian Colobinae (Mammalia: Cercopithecidae) asindicators of Quaternary climatic change. Biol. J. Linn. Soc. 59, 327–350.

Buckley-Beason, V.A., Johnson, W.E., Nash, W.G., Stanyon, R., Menninger, J.C., Driscoll,C.A., Howard, J., Bush, M., Page, J.E., Roelke, M.E., Stone, G., Martelli, P., Wen, C.,Ling, L., Duraisingam, R.K., Lam, V.P., O’Brien, S.J., 2006. Molecular evidence forspecies-level distinctions in clouded leopards. Curr. Biol. 16, 2371–2376.

Cannon, C.H., Morley, R.J., Bush, A.B.G., 2009. The current refugial rainforests ofSundaland are unrepresentative of their biogeographic past and highlyvulnerable to disturbance. PNAS 106, 11188–11193.

Cardillo, M., Purvis, A., Sechrest, W., Gittleman, J.L., Bielby, J., Mace, G.M., 2004.Human population density and extinction risk in the world’s carnivores. PLoSBiol. 2 (7), e197.

Christiansen, P., 2008. Species distinction and evolutionary differences in theclouded leopard (Neofelis nebulosa) and Diard’s clouded leopard (Neofelis diardi).J. Mammal. 89, 1435–1446.

Christiansen, P., 2009. Neotype of Neofelis diardi (Mammalia: Carnivora: Felidae).Zootaxa 2110, 58–68.

Clausen, S.E., 1998. Applied Correspondence Analysis: An Introduction. Series:Quantitative Application in the Social Sciences, vol. 12. Sage University,Thousands Oaks, CA, USA.

Cockerham, C.C., Weir, B.S., 1993. Estimation of gene flow from F-statistics.Evolution 47, 855–863.

Corbet, G.E., Hill, J.E., 1992. The Mammals of the Indomalayan Region: A SystematicStudy. Natural History Publ., Oxford Univ. Press, Oxford, UK.

Cuvier, G., 1823, Recherches sur les Ossements Fossiles oú l’on rétablit les Charactéresde plusieurs Animaux d’ont les Revolutions du Globe one détruit les Espéces. IV.Les Ruminans et les Carnassiers Fossiles. Dufour et Dócagne, Paris.

Dawson, A.G., 1992. Ice Age Earth. Late Quaternary Geology and Climate. Routledge,London.

Ellerman, J.R., Morrison-Scott, T.C.S., 1951. Checklist of Palaearctic and IndianMammals 1758–1946. Brit. Mus. Nat. Hist. Lond., 1–810.

Excoffier, L., Laval, G., Schneider, S., 2005. Arlequin (version 3.0): an integratedsoftware package for population genetics data analysis. Evol. Bioinform. 1,47–50.

Excoffier, L., Smouse, P.E., Quattro, J.M., 1992. Analysis of molecular varianceinferred from metric distances among DNA haplotypes: application to humanmitochondrial DNA restriction data. Genetics 131, 479–491.

Fickel, J., Hauffe, H.C., Pecchioli, E., Soriguer, R., Vapa, L., Pitra, C., 2008. Cladogenesisof the European brown hare (Lepus europaeus Pallas, 1778). Eur. J. Wildlife Res.54, 495–510.

Fu, Y.-X., 1997. Statistical tests of neutrality of mutations against populationgrowth, hitchhiking and background selection. Genetics 147, 915–925.

Gloger, C.W.L., 1833. Das Abändern der Vögel durch Einfluss des Klimas (inGerman). Verlag Schulz, Breslau.

Grassman, L.I.Jr., Tewes, M.E., Silvy, N.J., Kreetiyutanont, K., 2005. Ecology of threesympatric felids in a mixed evergreen forest in North-Central Thailand. J.Mammal. 86, 29–38.

Greenacre, M.J., Degos, L., 1977. Correspondence analysis of HLA gene frequencydata from 124 population samples. Am. J. Hum. Gen. 29, 60–75.

Hammer, Ø., Harper, D.A.T., Ryan, P.D., 2001. PAST: paleontological statisticssoftware package for education and data analysis. Palaeontol. Electron. 4 (1),1–9.

Harpending, R.C., 1994. Signature of ancient population growth in a low-resolutionmitochondrial DNA mismatch distribution. Hum. Biol. 66, 591–600.

Heaney, L.R., 1991. A synopsis of climatic and vegetational change in southeast Asia.Climatic Change 19, 53–61.

Hearn, A., Sanderson, J., Ross, J., Wilting, A., Sunarto, S., 2008a. Neofelis diardi. In:IUCN 2010. IUCN Red List of Threatened Species. Version 2010.1.<www.iucnredlist.org> (Downloaded on 21.04.10).

Hearn, A., Sanderson, J., Ross, J., Wilting, A. Sunarto, S., 2008b. Neofelis diardi ssp.borneensis. In: IUCN 2010. IUCN Red List of Threatened Species. Version 2010.1.<www.iucnredlist.org> (Downloaded on 27.04.10).

Hemmer, H., von Koenigswald, G.H.R., 1964. Fossile Nebelparder (Neofelis) aus demPleistozän Südchinas und Javas. Proc. Koninkl. Nederl. Akad. Wetenshappen(Ser. B) 67, 1–16.

Huelsenbeck, J.P., Ronquist, F., 2001. MrBayes: Bayesian inference of phylogeny.Bioinformatics 17, 754–755.

Hughes, J.B., Round, P.D., Woodruff, D.S., 2003. The Indochinese–Sundaic faunaltransition at the Isthmus of Kra: an analysis of resident forest bird speciesdistributions. J. Biogeogr. 30, 569–580.

Huff, W.D., Bergström, S.M., Kolata, D.R., 1992. Gigantic ordovician volcanic ash fallin North America and Europe: biological, tectonomagmatic, and event-stratigraphic significance. Geology 20, 875–878.

International Code of Zoological Nomenclature, 2000. fourth Ed., available ofJanuary 1st, 2000.

Jardine, W., 1834. The Naturalist’s Library, vol. 16: Mammalia. Lions tigers, & c., & c.W.H. Lizars, Edinburgh. 276 pp..

Johnson, W.E., Eizirik, E., Pecon-Slattery, J., Murphy, W.J., Antunes, A., Teeling, E.,O’Brien, S.J., 2006. The Late Miocene radiation of modern felidae: a geneticassessment. Science 311, 73–77.

Kimura, M., 1981. Estimation of evolutionary distances between homologousnucleotide sequences. PNAS 78, 454–458.

Kitchener, A.C., Beaumont, M.A., Richardson, D., 2006. Geographical variation in theclouded leopard, Neofelis nebulosa, reveals two species. Curr. Biol. 16, 2377–2383.

Kumar, S., Subramanian, S., 2002. Mutation rates in mammalian genomes. Proc. Nat.Acad. Sci. 99, 803–808.

Larkin, M.A., Blackshields, G., Brown, N.P., Chenna, R., McGettigan, P.A., McWilliam,H., Valentin, F., Wallace, I.M., Wilm, A., Lopez, R., Thompson, J.D., Gibson, T.J.,Higgins, D.G., 2007. Clustal W and Clustal X version 2.0. Bioinformatics 23,2947–2948.

Li, W.H., 1977. Distribution of nucleotide differences between two randomly chosencistrons in a finite population. Genetics 85, 331–337.

Lopez, J.V., Culver, M., Stephens, J.C., Johnson, W.E., O’Brien, S.J., 1997. Rates ofnuclear and cytoplasmic mitochondrial DNA sequence divergence in mammals.Mol. Biol. Evol. 14, 277–286.

Louys, J., 2007. Limited effect of the Quaternary’s largest super-eruption (Toba) onland mammals from Southeast Asia. Quaternary Sci. Rev. 26, 3108–3117.

Luo, S.-J., Kim, J.-H., Johnson, W.E., van der Walt, J., Martenson, J., Yuhki, N.,Miquelle, D.G., Uphyrkina, O., Goodrich, J.M., Quigley, H.B., Tilson, R., Brady, G.,Martelli, P., Subramaniam, V., McDougal, C., Hean, S., Huang, S.-Q., Pan, W.,Karanth, K.U., Sunquist, M., Smith, J.L.D., O’Brien, S.J., 2004. Phylogeography andgenetic ancestry of tigers (Panthera tigris). PLoS Biol. 2, 2275–2293.

Mayr, E., Ashlock, P.D., 1991. Principles of Systematic Zoology, second ed. McGrawHill, New York.

Meijaard, E., 2003. Mammals of south-east Asian islands and their Late Pleistoceneenvironments. J. Biogeogr. 30, 1245–1257.

Menotti-Raymond, M., David, V.A., Lyons, L.A., Schäffer, A.A., Tomlin, J.F., Hutton,M.K., O’Brien, S.J., 1999. A genetic linkage map of microsatellites in the domesticcat (Felis catus). Genomics 57, 9–23.

328 A. Wilting et al. / Molecular Phylogenetics and Evolution 58 (2011) 317–328

Mohamed, A., Samejima, H., Wilting, A., 2009. Records of five Bornean cat speciesfrom Deramakot Forest Reserve in Sabah, Malaysia. Cat News 51, 12–15.

Nei, M., 1975. Molecular Population Genetics and Evolution. North HollandPublishing Company, Amsterdam, New York.

Nei, M., 1987. Molecular Evolutionary Genetics. Columbia University Press, NewYork.

Oppenheimer, C., 2002. Limited global change due to the largest known Quaternaryeruption, Toba 74 kyr BP? Quaternary Sci. Rev. 21, 1593–1609.

Patou, M.L., Wilting, A., Gaubert, P., Esselstyn, J.A., Cruaud, C., Jennings, A.P., Fickel, J.,Veron, G., 2010. Evolutionary history of the Paradoxurus palm civets – a newmodel for Asian biogeography. J. Biogeogr. 37, 2077–2097.

Pattan, J.N., Shane, P., Pearce, N.J.G., Banakar, V.K., Parthiban, G., 2001. Anoccurrence of 74 ka Youngest Toba Tephra from the Western ContinentalMargin of India. Curr. Sci. 80, 1322–1326.

Pritchard, J.K., Stephens, M., Donnelly, P., 2000. Inference of population structureusing multilocus genotype data. Genetics 155, 945–959.

Posada, D., 2008. JModelTest: phylogenetic model averaging. Mol. Biol. Evol. 25,1253–1256.

Rampino, M.R., Ambrose, S.H., 2000. Volcanic winter in the Garden of Eden: theToba super-eruption and the Late Pleistocene human population crash. In:McCoy, F.W., Heiken, G. (Eds.), Volcanic Hazards and Disasters in HumanAntiquity. Geological Society of America Special Paper 345, pp. 71–82.

Rampino, M.R., Self, S., 1992. Volcanic winter and accelerated glaciation followingthe Toba super-eruption. Nature 359, 50–52.

Rampino, M.R., Self, S., 1993a. Bottleneck in human evolution and the Tobaeruption. Science 262, 1955.

Rampino, M.R., Self, S., 1993b. Climate-volcanism feedback and the Toba eruption ofB74, 000 years ago. Quaternary Res. 40, 269–280.

Ripley, G., Dana, C.A., 1858. The New American Cyclopaedia: A popular Dictionary ofGeneral Knowledge. D. Appleton and Company, New York. 775 pp.

Rogers, A.R., 1995. Genetic evidence for a Pleistocene population explosion.Evolution 49, 608–615.

Rogers, A.R., Harpending, H., 1992. Population growth makes waves in thedistribution of pairwise genetic differences. Mol. Biol. Evol. 9, 552–569.

Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylogenetic inferenceunder mixed models. Bioinformatics 19, 1572–1574.

Rose, W.I., Chesner, C.A., 1987. Dispersal of ash in the great Toba eruption, 75 ka.Geology 15, 913–917.

Rose, W.I., Chesner, C.A., 1990. Worldwide dispersal of ash and gases from earth’slargest known eruption: Toba, Sumatra, 75 ka. Palaeogeogr. Palaeocl.Palaeoecol. 89, 269–275.

Saitou, N., Nei, M., 1987. The neighbor-joining method: a new method forreconstructing phylogenetic trees. Mol. Biol. Evol. 4, 406–425.

Sanderson, J., Khan, J., Grassman, L. Mallon, D.P., 2008. Neofelis nebulosa. In: IUCN2010. IUCN Red List of Threatened Species. Version 2010.1. <www.iucnredlist.org> (Downloaded on 21.04.10).

Schmidt, K., Kowalczyk, R., Ozolins, J., Männil, P., Fickel, J., 2009. Genetic structure ofthe Eurasian lynx population in north-eastern Poland and the Baltic states.Conserv. Genet. 10, 497–501.

Schneider, S., Excoffier, L., 1999. Estimation of past demographic parameters fromthe distribution of pairwise differences when the mutation rates vary amongsites: application to human mitochondrial DNA. Genetics 152, 1079–1089.

Slatkin, M., 1995. A measure of population subdivision based on microsatellitefrequencies. Genetics 139, 457–462.

Sokal, R.R., Rohlf, F.J., 1995. Biometry. W.H. Freeman & Co., New York. 887 pp.Sunarto, Sanderson, J., Wilting, A., 2008. Neofelis diardi ssp. diardi. In: IUCN 2010.

IUCN Red List of Threatened Species. Version 2010.1. <www.iucnredlist.org>(Downloaded on 27.04.10).

Swofford, D.L., 2001. PAUP⁄ phylogenetic analysis using parsimony and othermethods. Sinauer Associates, Sunderland, MA.

Tajima, F., 1989. Statistical method for testing the neutral mutation hypothesis byDNA polymorphism. Genetics 123, 585–595.

Tamura, K., Dudley, J., Nei, M., Kumar, S., 2007. MEGA4: Molecular EvolutionaryGenetics Analysis (MEGA) software version 4.0. Mol. Biol. Evol. 24, 1596–1599.

Temminck, C.J., 1827, Monographies de Mammalogie ou Description de quelquesGenres de Mammiféres, don’t les Espéces ont été observées dans les DifférensMusées de l’Europe, vol. I. G. Dufour et D’Ocagne, Paris.

Terborgh, J., 1974. Preservation of natural diversity: the problem of extinction-prone species. BioScience 24, 715–722.

Uphyrkina, O., Johnson, W.E., Quigley, H.B., Miquelle, D.G., Marker, L., Bush, M.E.,O’Brien, S.J., 2001. Phylogenetics, genome diversity and origin of modernleopard, Panthera pardus. Mol. Ecol. 10, 2617–2633.

Voris, H.K., 2000. Maps of Pleistocene sea levels in Southeast Asia: shorelines, riversystems and time durations. J. Biogeogr. 27, 1153–1167.

Van der Kaars, W.A., Dam, M.A.C., 1995. A 135.000-year record of vegetational andclimatic change from Bandung area, West-Java, Indonesia. Palaeogeogr.Palaeocl. Palaeoecol 117, 55–72.

Wilson, A.C., Cann, R.L., Carr, S.M., George, M., Gyllensten, G.B., Helm-Bychowski,K.M., Higuchi, R.G., Palumbi, S., Prager, E.M., Sage, R.D., Stoneking, M., 1985.Mitochondrial DNA and two perspectives on evolutionary genetics. Biol. J. Linn.Soc. 26, 375–400.

Wilting, A., Fischer, F., Abu Bakar, S., Linsenmair, K.E., 2006. Clouded leopards, thesecretive top-carnivore of South-East Asian rainforests: their distribution,status and conservation needs in Sabah, Malaysia. BMC Ecol. 6, 16.

Wilting, A., Buckley-Beason, V.A., Feldhaar, H., Gadau, J., O’Brien, S.J., Linsenmair,K.E., 2007a. Clouded leopard phylogeny revisited: support for speciesrecognition and population division between Borneo and Sumatra. Front.Zool. 4, 15.

Wilting, A., Feldhaar, H., Buckley-Beason, V.A., Linsenmair, K.E., O’Brien, S.J., 2007b.Two modern species of clouded leopards: a molecular perspective. Cat News/IUCN SSC 47, 10–11.

Wisely, S.M., Maldonado, J.E., Fleischer, R.C., 2004. A technique for sampling ancientDNA that minimizes damage to museum specimens. Conserv. Genet. 5, 105–107.

Woodruff, D.S., Turner, L.M., 2009. The Indochinese–Sundaic zoogeographictransition: a description and analysis of terrestrial mammal speciesdistributions. J. Biogeogr. 36, 803–821.