The Genetic Impact of Chamois Management in the Dinarides

Transcription

1 The Journal of Wildlife Management; DOI: /jwmg Featured Article The Genetic Impact of Chamois Management in the Dinarides NIKICA SPREM, 1 Faculty of Agriculture, Department of Fisheries, Beekeeping, Game Management and Special Zoology, University of Zagreb, Svetosimunska cesta 25, Zagreb 10000, Croatia ELENA BUZAN, Science and Research Centre, Institute for Biodiversity Studies, University of Primorska, Garibaldijeva 1, Koper 6000, Slovenia ABSTRACT The Dinaric Mountains in Slovenia, Croatia, and Bosnia and Herzegovina provide a unique system to address the effects of past hunting on the genetic structure of northern chamois (Rupicapra rupicapra) and possible hybridization in the contact zone in the Velebit Mountains. The northern Dinaric Mountains should be occupied by alpine chamois (Rupicapra rupicapra rupicapra), whereas the central and southern areas are inhabited by the Balkan chamois (R. rupicapra balcanica). This is the first study to characterize the genetic variation in chamois populations in the area. We used microsatellite and mitochondrial markers to analyze the genetic variation and structure of chamois populations from different geographical areas with different histories. Specifically, we explored the influence of recent human translocations and geographical isolation on the genetic architecture of chamois populations in the assumed contact zone. We successfully genotyped 74 individual samples and the number of alleles/locus ranged from 6 to 20 with a mean of Allelic richness across populations ranged from 2.94 in the Prenj Mountains, Bosnia and Herzegovina to 3.56 in the Biokovo Mountains, Croatia. A similar pattern was also observed for heterozygosity, ranging between and 0.572, and expected heterozygosity, ranging between and in the Prenj and Biokovo mountains, respectively. The global genetic distance (F ST ) for 7 population samples was (range ¼ ). The STRUCTURE tree clusters separated samples from the northern Dinaric Mountains from those of the southern Dinaric Mountains into 2 clusters according to geographic location. The results obtained using a Bayesian clustering methodology was similar. By using mtdna variation in chamois from Slovenia, Croatia, and Bosnia and Herzegovina, the existence of alpine chamois haplotypes in northern areas and Balkan chamois haplotypes in southern areas was confirmed. These results confirm the impact of recent human management (i.e., translocation) into the Velebit Mountains, which established a new contact (hybridization) zone between the subspecies. Therefore, future translocations must be planned carefully to avoid compromising genetic integrity and posing a serious risk to native species, as in this case. Ó 2016 The Wildlife Society. KEY WORDS rupicapra. Balkan Peninsula, Dinaric Mountains, hybridization zone, microsatellites, mtdna, Rupicapra Chamois populations are classified into 2 species based on their morphological and behavioral characteristics (Grubb 2004): Pyrenean chamois (Rupicapra pyrenaica), which is distributed in the Pyrenees, the Cantabrian Mountains, and the Apennines, and northern chamois (R. rupicapra), which extend from the Alps to the Carpathians, the Balkan Mountains, and farther eastward into Asia. Furthermore, on the basis of their morphological traits and geographical distribution, and based on the biological species concept (BSC; Mayr 1942), there are 7 recognized subspecies of northern chamois (Masini and Lovari 1988, Corlatti et al. 2011): chartreuse (R. rupicapra cartusiana), alpine (R. rupicapra Received: 20 October 2014; Accepted: 14 March elena.buzan@famnit.upr.si rupicapra), Tatra (R. rupicapra tatrica), Carpathian (R. rupicapra carpatica), Balkan (R. rupicapra balcanica), Caucasian (R. rupicapra caucasica), and Anatolian chamois (R. rupicapra asiatica). Two of the 7 subspecies of northern chamois recognized in the various mountain systems they occupy within Europe are in the Dinaric region (Corlatti et al. 2011): alpine chamois and Balkan chamois. Both have a possible contact zone and may hybridize in the northern Velebit Mountains (Frkovic 2008). During the early 1900s, chamois populations in the Dinaric region were extirpated because of unsustainable hunting, poaching, livestock grazing, predation, and natural events (Frkovic 2009). The last records of chamois in the Velebit massif date back to 1907, when several animals were observed that later could not be found (Skorup 2005). Historical data and archaeological research confirmed the great abundance of the chamois population in the northwestern Dinaric Sprem and Buzan The Chamois Hybridization Zone 1

2 Mountains in Croatia and western Bosnia and Herzegovina, but as a result of predation, natural events, and unsustainable hunting, the population was extirpated before their taxonomic classification was assessed (Miracle and Sturdy 1991, Miculinic 2012). The main reason for the population s disappearance was attributed to intensive livestock grazing, hunting, and poaching, which forced the chamois out of its range in the early 1900s (Frkovic 2008). Small numbers of chamois were regularly encountered only at sites on Mount Risnjak along the border with Slovenia (Frkovic 2009). This deleterious human impact was counterbalanced by reintroductions after World War II that neglected genetic issues. Chamois were translocated into the Dinaric Mountains between 1964 and 1978 with chamois originating from different areas ( Sprem et al. 2015). Forty-eight animals (36 F, 12 M) were translocated to the Biokovo and northern Velebit Mountains from the Prenj Mountains in Bosnia and Herzegovina. There was also 1 translocation to the northern Velebit Mountains from the Kamnik Alps in Slovenia (3 F, 2 M). Translocations also occurred in the central Velebit Mountains in 1939 and 1956, though the origin of these animals was unknown. However, these attempts failed; the chamois present today spread from the northern Velebit areas ( Sabic and Lalic 2005, Frkovic 2008). Regardless of its current status, the chamois population of the northwestern Dinaric Mountains in Croatia is listed in Red Book of Mammals of Croatia as a regionally extinct species (Tvrtkovic and Grubesic 2006), whereas in Bosnia and Herzegovina, the chamois is listed as vulnerable (Adamic et al. 2006). However, in both countries, the chamois is considered a game species and is hunted in accordance with hunting regulations. The species is listed in Annex V of the European Union Habitat Directive and Appendix III of the Bern Convention, and Balkan chamois is included in Annexes II and IV of the European Union Habitats Directive and Appendix III of the Bern Convention. A recent study (Buzan et al. 2013) did not confirm the presence of the 2 subspecies (alpine and Balkan) or the hypothesis of a hybridization zone in the northwestern part of the Dinaric Mountains in Slovenia. The authors reported that endemic phylogeographic lineages are frequently confined to small ranges in the central Balkan Peninsula, whereas the northern periphery is occupied by widespread lineages that evolved in more northern refugia (Buzan et al. 2010). Chamois populations in Croatia provide a unique opportunity to address the issues of the effects of past management on genetic structure and possible hybridization where the subspecies (alpine and Balkan) overlap (i.e., contact zone). This is first study on genetic aspects of the population in the 50 years since the successful translocation to the Dinaric Mountains. We make special reference and search for answers to historical data, conjectures, and theory reported by previous authors about the ongoing translocation. Our objective was to use microsatellite and mitochondrial markers to analyze the genetic variation and structure of chamois populations from different geographical areas with different histories. Specifically, we explored areas in the contact zone, assessed the influence on the population structure of recent translocations, and examined how geographical isolation is reflected in the genetic architecture of the chamois populations. Finally, we compared our results with literature to analyze the subspecific status of the chamois in the northwestern Dinaric Mountains. STUDY AREA The Dinaric massif extends over Slovenia and Albania (645 km), and is characterized by small plateaus and meadows. The study area includes elevations between 518 m and 1,831 m, and is composed mainly of fir (Abies alba) and spruce (Picea abies;orsanic et al. 2005). The upper forest limit, mainly >1,400 m above sea level, comprises stands of the Dinaric Mountain pine (Hyperico grisebachii Pinetum mugo) association (Vukelic 2012). The climate is humid-boreal with a Mediterranean influence, a mean annual temperature of 7.78C and a mean annual rainfall of 2,079 mm (Vukelic 2012). Topographically, the area is highly heterogeneous, interrupted by ditches, bays, and rocks, which are developed on limestone and dolomite bedrock. Throughout the study area, lower elevation lands are mostly privately owned and under agricultural production, whereas upper elevation lands are predominantly state-owned timberlands under commercial management. Besides chamois, 5 ungulate species are present: wild boar (Sus scrofa), roe deer (Capreolus capreolus), red deer (Cervus elaphus), fallow deer (Dama dama), and European mouflon (Ovis musimon). Also, 3 large carnivores occur in this region: brown bear (Ursus arctos), gray wolf (Canis lupus), and Eurasian lynx (Lynx lynx). The history of chamois in the study area is different and 3 chamois populations (i.e., northern Velebit, central Velebit, Biokovo) were recently established through translocation. In most sites, chamois populations are stable, whereas chamois in the Prenj Mountains are virtually extinct (Table 1). All but 2 populations originate from native populations in Croatia. The first exception is the Balkan chamois population from the Prenj Mountains (Prenj) in Bosnia and Herzegovina, which participated in the reintroduction at certain sites in Croatia (Frkovic 2008). The second is the population of alpine chamois (Gotenica) from the neighboring Dinaric Mountains, Goteniski Sneznik, in Slovenia. METHODS Analysis of Genetic Markers and Genotyping We obtained tissue and hair samples from 74 chamois in 2010 and 2012 during legal culls by hunters according game management plans (approved by a competent Ministry) and from natural deaths in subpopulations from 7 sampling sites (Fig. 1) across the Dinaric Mountain massif: Gotenica (n ¼ 10), Gorski Kotar (n ¼ 11), northern Velebit (n ¼ 12), central Velebit (n ¼ 7), Dinara (n ¼ 3), Biokovo (n ¼ 16), and Prenj (n ¼ 15). We performed DNA extractions using commercial Qiagen DNeasy 1 kits (Qiagen, Dusseldorf, Germany) following 2 The Journal of Wildlife Management 9999()

3 Table 1. Location, sample size, and brief history of the chamois populations investigated in Slovenia, Croatia, and Bosnia and Herzegovina, 2010 and Area Location Sample size Brief description of populations Gotenica (GOT) N E Stable and constant population of northern chamois, current population size ¼ 250 individuals Gorski Kotar (GKO) N E Stable and constant population of northern chamois in area bordering Slovenia, current population size ¼ 350 individuals North Velebit (NVE) N E Reintroduced population with 10 animals (5 M:5 F) from Prenj (Balkan chamois) and 5 animals (2 M:3 F) from Kamnik Alps (alpine chamois) during , current population size ¼ individuals Central Velebit (CVE) N E Reintroduced population with unknown origin of the animals in 1939 and 1956, current population size ¼ individuals Dinara (DIN) N E Very small native population of Balkan chamois shared between Bosnia and Herzegovina, current population size ¼ 60 individuals Biokovo (BIO) N E Reintroduced population with 48 animals (12 M:36 F) from Prenj (Balkan chamois) during , current population size ¼ individuals Prenj (PRE) N E Virtually extinct native population of Balkan chamois, with only 50 individuals remaining from a few thousand in the 1980s the manufacturer s protocol, in a volume of 200 ml. We sequenced mtdna regions to identify the subspecific provenance of the chamois populations. We amplified the partial cytochrome b gene (cyt b, 349 bp) and its control region (CR, 475 bp) using universal primers and the polymerase chain reaction (PCR) protocol outlined in Rodrıguez et al. (2009, 2010). Each sample was twice isolated and amplified. We sequenced parallels with DNA concentration >5 ng twice in the forward and reverse direction to confirm the fragment sequence (the error rate was <0.6%). We amplified 20 microsatellite loci in 3 multiplex sets, containing 7, 6, and 7 microsatellites, respectively, using the protocol described in Zemanova et al. (2011). For all 3 sets, Figure 1. Distribution of chamois in the Dinaric Mountains in Slovenia, Croatia, and Bosnia and Herzegovina in 2010 and Sampling sites are indicated by their 3-letter code. The pie charts in the map indicate the frequency of the mtdna haplotypes in each locality of the study area within the distribution range of chamois in southeastern Europe: GOT ¼ Gotenica Mountains; GKO ¼ Gorski Kotar; NVE ¼ North Velebit Mountains, CVE ¼ Central Velebit Mountains; DIN ¼ Dinara Mountains; BIO ¼ Biokovo Mountains, PRE ¼ Prenj Mountains. The pie charts indicate the location. The dashed lines represent mountain edge. Sprem and Buzan The Chamois Hybridization Zone 3

4 we performed PCR and fragment analysis using the protocol described in Buzan et al. (2013). We examined microsatellite genotypes using GeneMapper software (Life Technologies, Carlsbad, CA). We performed genotyping on 74 samples using 20 unlinked microsatellite loci, and performed sequencing on 45 samples because the extraction of DNA was limited by the quantity of available tissue. We pre-screened all samples for DNA concentration and used only samples with >5 ng/ml for a study with nuclear markers. To confirm the genotypes, we re-amplified approximately 30% of the individual genotypes that were suspected of allelic dropout (i.e., homozygous at particular loci), and obtained the same genotypes for the replicates. Overall genotyping error rate based on repeat typing was 3.5%. Mitochondrial DNA Phylogenetic Reconstruction We assembled the mitochondrial sequences using Program CodonCode Aligner 1.63 (Ewing et al. 1998). We aligned the resulting consensus sequences using ClustalW 4.0, implemented in the MEGA package 6.0 (Tamura et al. 2013). We combined the pooled dataset (2 regions) of 45 individuals sequenced in this study with previous data on 7 subspecies of chamois (Rodrıguez et al. 2010, Buzan et al. 2013). We designated the mtdna haplotypes using the same combination of fragments as described in Buzan et al. (2013). In the newly sequenced data, we computed the haplotypes in DNASP 5.10 (Librado and Rozas 2009). We analyzed phylogenetic relationships using the same methods and outgroups as in Buzan et al. (2013). We used the neighborjoining and Bayesian approaches under different models of nucleotide substitution. The neighbor-joining tree, based on the number of substitutions/site under the Jukes Cantor model, was constructed from the pooled sequences of 108 chamois haplotypes. We assessed the reliability of the nodes by 1,000 bootstrap replicates (BP). We obtained the Bayesian tree using MrBayes (Huelsenbeck and Ronquist 2001, Ronquist and Huelsenbeck 2003). We used jmodeltest (Guindon and Gascuel 2003, Darriba et al. 2012) to determine the most appropriate model of DNA substitution under the Akaike s Information Criterion (AIC), corrected AIC (AIC c ), and Bayesian Information Criterion (BIC). The jmodeltest selected the Hasegawa, Kishino, and Yano model (HKY) including invariable sites (I) and rate variation among sites (G), HKY þ G þ I(G¼ and I ¼ ) model to be the most appropriate substitution model for the control region haplotypes dataset and the HKY þ G(G¼0.2835) model for cyt b haplotypes dataset. We ran 4 Monte Carlo Markov chains simultaneously for 2 million generations, with the resulting trees sampled every 100 generations. We used Bayesian posterior probabilities (BPP) to assess branch support of the Bayesian tree. We checked convergence by examining the generation plot and visualized it with TRACER 1.4 (Rambaud and Drummond 2007). Genetic Variation of Individual Loci We tested the exact probability test for each locus and population to test the deviation of the observed genotype frequency from the Hardy Weinberg equilibrium (HWE) using the software Genepop (Rousset 2008). We set the basic level of significance to 0.05 and applied a sequential Bonferroni procedure for multiple comparisons. The presence of null alleles may cause a significant heterozygote deficit and deviation from the HWE. We, therefore, estimated the proportion of null alleles at each locus in each population using FreeNA (Chapuis and Estoup 2007). We calculated the mean number of alleles, and observed (H o ) and expected (H e ; Nei 1978) heterozygosities for each locus in all populations using Genetix (Belkhir et al ). We estimated the allelic richness in the population using the rarefaction procedure implemented in FSTAT (Goudet 2001). We did not calculate genetic diversity for population Dinara because of the small sample size (n ¼ 3). Genetic Variability Among Populations We used FreeNa to estimate a global genetic distance (F ST ), with a confidence interval of 95%, assessed with 1,000 permutations. We estimated pairwise F ST in Genetix according to Weir and Cockerham (1984), and tested significant differences of F ST estimators from 0 with 1,000 permutations in the same program. We applied the Mantel test to matrices of F ST /(1 F ST ) estimates based on microsatellite marker data and the natural logarithm of geographic distances between 6 sampling areas to test for the pattern of isolation by distance. Samples from population Dinara were not included in this analysis because of a small sample size. We tested significance of the correlation using 1,000 permutations. We estimated and tested slope of the linear regression of F ST /(1 F ST ) on the natural logarithm of geographic distances between sampling areas. We performed all the calculations using R software (R core team 2013). We used the Genetix package to investigate the genetic relationships among all genotyped individuals by factorial correspondence analysis (FCA). Finally, to analyze population structure, we identified genetic structure using Bayesian algorithms implemented in BAPs 4.1 (Corander et al. 2008) and STRUCTURE (Falush et al. 2003) programs. In Program STRUCTURE, we tested various measures for different genetic populations or clusters (K), from 1 to 10, running the analysis 10 times for each K under a model assuming admixture and correlate allele frequency. We performed 20 runs with a burn-in period of 100,000 replications and a run length of 1,000,000 Markov chain Monte Carlo iterations for a number of clusters ranging from K ¼ 1 to 10. The first mentioned model admits mixed ancestry of individuals caused by admixture and hybridization. We combined the results of replicated runs for each value of K from 2 to 10 using the Greedy algorithm of CLUMPP (Jakobsson and Rosenberg 2007) and displayed the summary outputs graphically using DISTRUCT 1.1 (Rosenberg 2004). In addition, we applied the ad hoc summary statistic DK developed by Evanno et al. (2005), which is based on the rate of change of the estimated likelihood between successive K values. We used Structure Harvester (Earl and vonholdt 2012) 4 The Journal of Wildlife Management 9999()

5 to generate graphs for the mean log posterior probability of the data (mean SD). Using allele frequencies and the HWE, the BAPs program estimates the structure of populations by clustering individuals (or groups of individuals [populations]) into genetically distinguishable groups. The incorporation of geographical information into the analysis is also possible, by including the geographic coordinates of sampling localities. We considered individuals from 1 locality as 1 population sample (1 group) and we performed 20 independent runs examining the spatial mixture analysis of groups of individuals (spatial clustering of groups). We used the default values and allowed K to vary from 1 to 10. RESULTS Mitochondrial DNA Variation We amplified and sequenced fragments of cyt b and CR from 45 individuals. Eleven haplotypes of cyt b (deposited in GenBank under Accession Nb. KP KP [HRC1 HRC11]) and 10 haplotypes of the control region were defined (deposited in GenBank under Accession Nb. KP KP [HRCR1 HRCR10]). The combined alignment contained 16 haplotypes defined by 45 variable sites and 76 haplotypes of combined sequences downloaded from GenBank. Both the neighbor-joining and Bayesian trees (Fig. 2) revealed the 3 well-supported major Figure 2. The phylogeny of chamois in Slovenia, Croatia, and Bosnia and Herzegovina from samples collected in 2010 and 2012 constructed by neighborjoining and Bayesian analysis of the 108 haplotypes resulting from the combined sequences of the partial cytochrome b and partial control region (764 bp). Neighbor-joining bootstrap support and Bayesian posterior probability indices are shown below each branch. The GenBank haplotypes of Balkan chamois are indicated with a & symbol and new Balkan haplotypes with a ~symbol. The new chamois haplotypes are indicated with a *symbol and haplotypes from Buzan et al. (2013) with a symbol. Clade mtwest, clade mtcentral, and clade mteast represent the 3 major mitochondrial lineages recognized by Rodrıguez et al. (2010). Sprem and Buzan The Chamois Hybridization Zone 5

6 Table 2. Genetic diversity in chamois populations in Slovenia, Croatia, and Bosnia and Herzegovina, 2010 and H e ¼ expected and H o ¼ observed heterozygosity, F IS ¼ inbreeding coefficient, HWE ¼ Hardy-Weinberg equilibrium, A ¼ number of alleles, and AR ¼ allelic richness (calculated by the rarefaction method for the lowest sample size n ¼ 7). Standard deviations are for average values. Population (n) H e SD H o SD F IS HWE A SD AR SD GOT (10) GKO (11) NVE (12) CVE (7) BIO (16) PRE (15) P < 0.05 for HWE and F IS (before Bonferroni correction). clades previously defined by Rodrıguez et al. (2010): mtwest (mtw), mtcentral (mtc), and mteast (mte). Clade mtw is composed of haplotypes from the Iberian Peninsula (Pyrenean chamois) and the western Alps (alpine chamois), Clade mtc groups haplotypes from the Apennines (Pyrenean chamois) and the Massif of Chartreuse (alpine chamois), and Clade mte groups all haplotypes from the central Alps to the Caucasus. The new haplotypes from northwestern Dinaric Mountains clustered into Clade mte, together with haplotypes from Slovenia and from the central and eastern Alps in a highly supported cluster (BP ¼ 92, BPP ¼ 0.98). The haplotypes from the Biokovo, Dinara, Velebit, and Prenj mountains grouped together with the haplotype of Balkan chamois from the Carpathians (Serbia; ref. nb. Bao017 in Rodrıguez et al. 2010). Six haplotypes from the Velebit Mountains also belong to this clade, which is likely the consequence of past chamois translocations (BP ¼ 90, BPP ¼ 0.97). Although some branching was poorly resolved, both trees showed the same topology and new, well-distinguished haplotypes of alpine chamois from the northern part of the Dinaric massif and Balkan chamois haplotypes from the Central Dinaric Mountains (BP ¼ 81, BPP ¼ 0.83). All haplotypes were highly geographically associated, despite the indication of translocations in the Velebit Mountains (Fig. 1). Seven unique haplotypes were found in 4 locations (Gotenica, northern Velebit, Biokovo, Prenj). Populations from the Dinara Mountains and the Biokovo Mountains shared the Balkan haplotypes with the nearby Prenj Mountains, whereas the Velebit Mountains shared alpine haplotypes with the Gorski Kotar and Gotenica Mountains. Hardy Weinberg Equilibrium and Intra-Population Genetic Diversity We successfully genotyped 74 individual samples and the amplification success varied (91 100%). We estimated 67%, 81%, and 75% null alleles/population at the SY58, SY259, and BM1258 loci, respectively, using FreeNA software. We, therefore, excluded these 3 loci and the monomorphic locus INRA 121. Thus, only 16 microsatellite loci with <7% of null alleles over all samples were finally used in all analyses. To confirm the genotypes, we re-amplified approximately 10% of the individual genotypes suspected of allelic dropout (i.e., homozygous at particular loci), and obtained the same genotypes for the replicates. The populations Prenj, Biokovo, and Gorski Kotar showed significant deviation from HWE on the basis of F IS (significant positive values) We found further evidence of Prenj and Gorski Kotar deviating from HWE based on exact tests in Genepop (P ¼ 0.05). The deviation from HWE did not remain significant after Bonferroni correction (Table 2). The number of alleles/locus ranged from 6 to 20 with a mean of The allelic richness across populations ranged from 2.94 (Biokovo) to 3.24 (Prenj). A similar pattern was also observed for H o (range ¼ ) and H e (range ¼ ; Table 2). Spatial Genetic Structure and Isolation by Distance The global F ST for the 7 population samples was (range ¼ ). The highest F ST value was observed between the Gorski Kotar and the Biokovo populations (Table 3). The FCA plot based on individual genotypes clearly separated Balkan chamois and alpine chamois along the first factorial axis (Fig. 3). The first axis (explaining 38.3% of variation, P ¼ 0.34) separated the 3 Dinaric Mountains populations (Prenj, Biokovo, and Dinara) from all other populations. The second axis (explaining 23.3% of variation, P ¼ 0.20) mainly separated the individuals from the northern Dinaric Mountains from the alpine chamois subspecies. We obtained similar results in the STRUCTURE analysis. The best model indicated 5 clusters (using DK according to Evanno et al. [2005]; Fig. S1, available online in Supporting Information). This result clearly separated samples from the northern Dinaric Mountains (alpine chamois) in 3 clusters and samples from the southern Dinaric Mountains (Balkan chamois) in 2 clusters according to geographic location (Fig. 4). Table 3. Pairwise values of genetic distance (F ST ) among chamois populations in Slovenia, Croatia, and Bosnia and Herzegovina, 2010 and 2012, (GOT ¼ Gotenica, GKO ¼ Gorski Kotar, NVE ¼ North Velebit, CVE ¼ Central Velebit, BIO ¼ Biokovo, PRE ¼ Prenj) based on 20 microsatellite loci calculated as described in Weir and Cockerham (1984). Population GOT GKO NVE CVE BIO GOT GKO NVE CVE BIO PRE The Journal of Wildlife Management 9999()

7 Figure 3. A 2-dimensional plot of the factorial correspondence analysis performed using GENETIX for samples of chamois collected in Slovenia, Croatia, and Bosnia and Herzegovina in 2010 and Individuals from different subspecies or locations (identified by mtdna) are indicated by different symbols. GOT ¼ Gotenica, GKO ¼ Gorski Kotar, NVE ¼ North Velebit, CVE ¼ Central Velebit, DIN ¼ Dinara, BIO ¼ Biokovo, PRE ¼ Prenj. The first axis explained 38.3% of the variance, and the second explained 23.3% of the variance. In BAPs, we obtained the highest marginal likelihood indicating maximized posterior probability in spatial clustering of groups (i.e., population samples) for the model of K ¼ 5 (Fig. 5). Geographically, close populations were clustered together. Two alpine chamois populations from the north Dinaric Mountains clustered together. Populations Gotenica and Gorski Kotar formed 2 independent clusters. The Balkan chamois population from the Prenj Mountains was separated from the populations in the Biokovo Mountains and Dinara Mountains. The clustering of populations revealed the same pattern as the STRUCTURE analysis, suggesting robust population structure. Microsatellite-based genetic distances were not significantly correlated with geographical distances among populations, and slope of the linear regression was not significantly different from 0 (slope ¼ , r 2 ¼ 0.16, P ¼ ; Fig. 6), suggesting that there is no underlying isolation by distance between populations. DISCUSSION Phylogeography of Chamois in the Dinaric Mountains The genetic information obtained in this study confirmed that the Dinaric Mountains are inhabited by 2 northern chamois subspecies: alpine and Balkan chamois. The recent study from the northernmost part of Dinaric Mountains failed to find different subspecies (Buzan et al. 2013). According to the present study, the Balkan haplotypes reach their northern boundary within the Velebit Mountain. By using mtdna variation in chamois from Slovenia, Croatia, and Bosnia and Herzegovina, the existence of alpine haplotypes was confirmed at the sites Gotenica, central Velebit, and Gorski Kotar, and Balkan haplotypes at the sites Prenj, northern Velebit, central Velebit, Biokovo, and Dinara. In the phylogenetic tree, the haplotypes C6 C10 and C12 C16 clustered together with the Balkan haplotype from the Carpathian Mountains in Serbia. This Figure 4. Genetic structure (from Program STRUCTURE) of chamois within 7 sampled areas in Slovenia, Croatia, and Bosnia and Herzegovina from samples collected in 2010 and Each individual is represented by a line proportionally partitioned into shaded segments corresponding to its membership in a specific cluster (K). Black lines separate the individuals from different sampling areas. GOT ¼ Gotenica, GKO ¼ Gorski Kotar, NVE ¼ North Velebit, CVE ¼ Central Velebit, DIN ¼ Dinara, BIO ¼ Biokovo, PRE ¼ Prenj. Sprem and Buzan The Chamois Hybridization Zone 7

8 place, hunting began in the Biokovo Mountains ( Sabic and Lalic 2005), whereas hunting in the Velebit Mountains started many years later, with the first harvest in Currently, the populations are stable at both locations. Taking into account the recorded data (documented translocations), the mtdna results presented here show that the signature of recent translocations in the Velebit Mountains (northern Velebit and central Velebit) is visible in the recent genetic composition and has established a new contact (hybridization) zone between the alpine and Balkan subspecies (Fig. 2). We were able to record successful translocations in the central Velebit Mountains, which is contradictory to the conjectures of other authors (Frkovic 1981, Skorup 2005) about the unsuccessful translocation and the theory of its spread from the northern Velebit Mountains (Frkovic 2008). None of the mtdna haplotypes were shared between northern Velebit and central Velebit (Fig. 1). Figure 5. The genetic structure of chamois populations in Slovenia, Croatia, and Bosnia and Herzegovina, 2010 and 2012, based on spatial clustering of groups of individuals using Bayesian analysis in Program BAPs. The best model divided 7 population samples (GOT ¼ Gotenica, GKO ¼ Gorski Kotar, NVE ¼ North Velebit, CVE ¼ Central Velebit, DIN ¼ Dinara, BIO ¼ Biokovo, PRE ¼ Prenj) into 5 clusters (K; top frame). Fixed K ¼ 2 (bottom frame) separated alpine chamois and the Balkan chamois population. phylogenetic relationship confirms the existence of endemic Balkan haplotypes in the Prenj, Dinara, and Biokovo mountains but also reveals the result of recent translocations in the Velebit Mountains (Fig. 2). The Balkan Peninsula houses the main European biodiversity centers for chamois and the distribution pattern of the endemic phylogeographical lineages is evident in the central part of the peninsula, whereas the northern part is occupied by widespread lineages (Previsic et al. 2009, Buzan et al. 2010). Although translocations make sense only when the principal cause of extinction has been eliminated, most translocations were used primarily to increase species populations as a result of strong hunter interest in obtaining an additional game species (Apollonio et al. 2014). The same interest led to chamois translocations in the Biokovo and Velebit mountains from 1964 to The newly founded populations adapted well in a very short time (Frkovic 2008). However, only 10 years after the translocation had taken Genetic Diversity The genetic diversity in chamois populations (Gotenica, Gorski Kotar, northern Velebit, central Velebit) was slightly lower in term of allelic richness and observed heterozygosity than those reported by Buzan et al. (2013) and other authors (Crestanello et al. 2009, Soglia et al. 2010). These differences may be due partly to larger samples used by Buzan et al. (2013), and to the different microsatellites used in Crestanello et al. (2009) and Soglia et al. (2010). The lower genetic diversity in the central Dinaric Mountains could indicate stronger genetic drift within these populations than in the northern Dinaric Mountains described by Buzan et al. (2013), and obviously the existence of gene flow between the Alps and Dinaric Mountains in the contact area contribute to higher genetic diversity. Wild ungulates have historically been used for hunting in Europe. Therefore, the great numbers of former reintroductions (usually undocumented) were connected with poor management practices by hunters, such as very high hunting bags. During the past 30 years, roe deer have been introduced to the western Italian Alps and the Apennines and as a result, the endemic Italian roe deer (Capreolus c. italicus) was subjected to introgressive hybridization with non-native stock (Focardi et al. 2009). A recent genetic study detected the highest haplotype diversity in reintroduction areas and the lowest in isolated genetically distinct population of Italian roe deer (Mucci et al. 2012). A similar situation was reported with the endemic Iberian roe deer subspecies (Capreolus c. decorus) in Portugal following the introduction of animals from genetically distinct populations from France (Royo et al. 2007). In the early 1960s, alpine chamois were introduced to the Tatra Mountains in Slovakia, and hybridization occurred with the genetically less diverse endemic Tatra chamois (Findo and Skuban 2010). The consequence of introductions from the genetically distinct Balkan chamois (Prenj) population has not left its mark in the genetic richness of the Velebit Mountains and Biokovo Mountains populations. The genetic richness was comparable to those obtained for the Dinaric populations in Buzan et al. (2013). 8 The Journal of Wildlife Management 9999()

9 Figure 6. Isolation-by-distance in Dinaric Mountains chamois populations in Slovenia, Croatia, and Bosnia and Herzegovina, 2010 and We calculated genetic distance (F ST ) and plotted F ST /(1 F ST ) against the natural log of geographical distance (km). The genetic and geographical distances are not significantly correlated. Slope of the linear regression ¼ , r 2 ¼ 0.16, P ¼ Historical data beginning from 1966 state that the population from the Prenj Mountains numbered roughly 4,000 individuals (Gafic and Dzeko 2009), a level maintained until the early 1990s. In the recent war, the population was decimated by illegal hunting with up to 95% of the stock lost, and today it numbers about 50 individuals (Frkovic 2008). Despite this recent event that swept through in the course of only 2 3 generations, significant differences in the genetic variation were not found in comparison to other sites. The recent genetic variation could be also a consequence of a past, large effective population size and its ancient inherent diversity or, perhaps, the existence of restored connectivity from just the few neighboring remaining patches in the area ( Cvrsnica Mountains). Genetic Differentiation and Population Relationship The global F ST value was 0.103, and this value is easily explained because our data includes a wider range of the Dinaric Mountains, with 2 geographical populations recognized as subspecies. The geographical isolation of populations is clearly evident from mtdna; 8 of the 16 haplotypes encountered were unique (e.g., c1 in Gotenica, c4 in Gorski Kotar, c8 and c9 in northern Velebit, c10 and c11 in central Velebit, c12, c13, and c14 in Biokovo, c14 in Prenj). The presence of unique haplotypes in populations from the Velebit Mountains (C8, C9, and C10) perhaps suggests that not all the chamois disappeared from these sites in the early 1900s, as was previously believed (Skorup 2005). Levels of genetic differentiation vary among groups, and we did not confirm a significant isolation by distance pattern over the range of the studied area (Fig. 6). This could suggest existing gene flow between populations, despite the different history of each population or clusters of populations and recent human-mediated reintroduction. The signature of the translocation is visible in the mitochondrial variation; we found the Balkan haplotype (from Prenj) in northern Velebit (c6) and Biokovo populations (c16). Lessons from past translocations highlight the risk of genetic admixture with introduced conspecifics and loss of endemic diversity from the native taxon, which today has become a major conservation issue (Focardi et al. 2009). Although the overall picture poses several puzzles, the results were consistent across all tests, supporting the strong differentiation in these chamois populations. The results of FCA clearly separate northern alpine chamois (Gotenica, Gorski Kotar, northern Velebit, and central Velebit) from southern Balkan chamois populations (Dinara, Biokovo, and Prenj). The results support 2 different geographical populations and confirm the historical data (Frkovic 2009) and Bayesian clustering. The STRUCTURE analysis with K ¼ 2 gave a clear split between alpine chamois (Gotenica, northern Velebit, central Velebit, and Gorski Kotar) and Balkan chamois (Prenj, Biokovo, and Dinara) populations. In addition, even in the best model with K ¼ 5 populations, we failed to find a signal of Balkan genotypes in the Velebit Mountains populations. Only a weak signal of different genotypes could be noted. It is likely that the genetic signal of the Prenj Mountains population was lost in the nuclear data (but is still evident in mtdna; Fig. 1). It is now well established that STRUCTURE can sometimes fail to detect population genetic structure at lower levels of genetic differentiation, especially when using a smaller number of individuals or loci (Kalinowski 2011, Frantz et al. 2012, Sprem et al. 2013). The clustering of chamois populations Sprem and Buzan The Chamois Hybridization Zone 9

10 supports their differentiation into 5 geographically associated clusters and reveals a strong substructure within mountain ranges with suboptimal chamois habitat. Population from the Velebit Mountains clustered together, as did the populations from the Biokovo and Dinara mountains. All other populations are associated with genetically independent clusters. Evidently, small groups of chamois may remain isolated in restricted habitat patches, separated by extensive forest with no chamois habitat between, with occasional but very restricted gene flow between patches. The similar population structure revealed by the spatial model in BAPs and the non-spatial model in STRUCTURE indicates that the differentiation between populations is so strong that it was even detected by the less sensitive model (Chen et al. 2007, Safner et al. 2011). The small population size and low genetic variability of the chamois populations in the Dinaric Mountains make them vulnerable to many factors. Illegal poaching and translocation have given rise to a distinct substructuring of the existing gene pool. In a review paper, Corlatti et al. (2011) pointed to the same threats to the populations (i.e., hybridization, translocations), which may increase the risk of losing differentiated gene pools and cause genetic extinction of taxa (e.g., chartreuse chamois, Tatra chamois, and Balkan chamois). Balkan chamois, a charismatic ungulate with endangered status, could serve as an excellent flagship species for the conservation of mountain ecosystems in the Balkan Peninsula. In Croatia and Bosnia and Herzegovina, the species is protected by the national and European legislation. The translocation to the Dinaric Mountains (especially Biokovo Mountains) has twice the benefit because this is one of the main reasons for the reappearance of the 3 large predators in the area (i.e., brown bear, gray wolf, and European lynx), which has ultimately had a positive consequence in restocking the natural community of large mammals and increasing the biodiversity of the area (Apollonio et al. 2014). Translocations and subsequent hybridization between individuals from different populations or taxa have had a significant genetic impact on alpine and Balkan chamois in the Dinaric Mountains. Unfortunately, the results of the present study show that the translocation of individuals from geographically (and therefore genetically) distant populations have resulted in a hybrid zone in the Velebit Mountains, which can be detrimental to the conservation of the species. MANAGEMENT IMPLICATIONS Conservation units for the alpine and Balkan chamois need to be defined for future management strategies, which may or may not include translocations with individuals carrying the appropriate genetic background. In the meantime, the interspecific hybridization occurring in the Velebit Mountains should be regarded as an ongoing and unplanned experiment and the evolutionary consequences of this process should be carefully monitored. Because the current range of the Balkan chamois in the Prenj Mountains does not correspond to the actual habitat suitability and it is on the verge of extinction, the feasibility could be evaluated to use the Biokovo Mountain population as a source for translocations to the former habitats. The 5 clusters, re-evaluated with a spatial analysis of genetic variation, have undergone independent demographic histories and should be regarded as independent units for management purposes. We reject the presumption of the homogeneity of the chamois populations of the Dinaric Mountains. Considering the population structuring of the Velebit, Dinara, and Biokovo mountain populations, which seem likely able to maintain viable populations, there should be restricted gene flow from the west (introduced northern Velebit and Biokovo population), having lost the signal (nuclear markers) of their founder population and only exhibiting their own genetic makeup. Further research on the genetic structure of these populations has the potential to deepen the insight into any gene flow and to define the contact zone where hybridization of the subspecies is taking place. For the effective future management of the Velebit Mountain population, it is crucial to establish whether chamois really did disappear from the area prior to the implementation of translocations. ACKNOWLEDGMENTS We thank the associate editor, 2 anonymous reviewers, and T. Safner for helpful comments on earlier drafts of the manuscript, and L. Zanella for proofreading the manuscript. LITERATURE CITED Adamic, M., Z. Rapaic, Z. Popovic, S. Kunovac, M. Koprivica, V. Soldo, B. Markovic, R. Maunaga, M. Micevic, and V. Ilic Endangered game species in Bosnia and Herzegovina Final report. Megaprojekt/Waldprojekt, Banja Luka, Bosnia and Herzegovina. [In Bosnian.] Apollonio, M., M. Scandura, and N. Sprem Reintroductions as a management tool for European ungulates. Pages in R. Putman and M. Apollonio, editors. Behaviour and management of European ungulates. Whittles Publishing, Dunbeath, Caithness, United Kingdom. Belkhir, K., P. Borsa, L. Chikhi, N. Raufaste, and F. Bonhomme GENETIX 4.05, logiciel sous Windows TM pour la geńe tique des populations. Laboratoire Geńome, Populations, Interactions, CNRS UMR 5000, Universite de Montpellier II, Montpellier, France. Buzan, E. V., D. W. Forster, J. B. Searle, and B. Krystufek A new cytochrome b phylogroup of the common vole (Microtus arvalis) endemic to the Balkans and its implications for the evolutionary history of the species. Biological Journal of the Linnean Society 100: Buzan, V. E., J. Bryja, B. Zemanova, and B. Krystufek Population genetics of chamois in the contact zone between the Alps and the Dinaric Mountains: uncovering the role of habitat fragmentation and past management. Conservation Genetics 14: Chapuis, M.-P., and A. Estoup Microsatellite null alleles and estimation of population differentiation. Molecular Biology and Evolution 24(3): Chen, C., E. Durand, F. Forbes, and O. Francois Bayesian clustering algorithms ascertaining spatial population structure: a new computer program and a comparison study. Molecular Ecology Notes 7: Corander, J., J. Sireń, and E. Arjas Bayesian spatial modelling of genetic population structure. Computational Statistics 23: Corlatti, L., R. Lorenzini, and S. Lovari The conservation of the chamois Rupicapra spp. Mammal Review 41(2): Crestanello, B., E. Pecchioli, C. Vernesi, C. Mona, N. Martınkova, M. Janiga, H. C. Hauffe, and G. Bertorelle The genetic impact of translocations and habitat fragmentation in chamois (Rupicapra spp). Journal of Heredity 100: Darriba, D., G. L. Taboada, R. Doallo, and D. Posada jmodeltest 2: more models, new heuristics and parallel computing. Nature Methods 9: The Journal of Wildlife Management 9999()

11 Earl, D. A., and B. M. vonholdt STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conservation Genetics Resources 4: Evanno, G., S. Regnaut, and J. Goudet Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology 14: Ewing, B., L. Hillier, M. C. Wendl, and P. Green Base-calling of automated sequencer traces using Phred. I. Accuracy assessment. Genetics Research 8: Falush, D., M. Stephens, and J. K. Pritchard Inference of population structure using multilocus genotype data: linked loci and correlated allele frequencies. Genetics 164: Findo, S., and M. Skuban Ungulates and their management in Slovakia. Pages in M. Apollonio, R. Andersen, and R. Putman, editors. European ungulates and their management in the 21st century. Cambridge University Press, Cambridge, United Kingdom. Focardi, S., P. Montanaro, V. La Morgia, and F. Riga, editors Piano d azione nazionale per il capriolo italico (Capreolus capreolus italicus). Quaderni Conservazione della Natura 31, Ministero Ambiente ISPRA, Italy. Frantz, A. C., S. Bertouille, M. C. Eloy, A. Licoppe, F. Chaumont, and M. C. Flamand Comparative landscape genetic analyses show a Belgian motorway to be a gene flow barrier for red deer (Cervus elaphus), but not wild boars (Sus scrofa). Molecular Ecology 14: Frkovic, A Naseljavanje divokoze u loviste Senjski Sjeverni Velebit. Goransko-Primorsko sumsko gospodarstvo, Delnice, Croatia. [In Croatian.] Frkovic, A Reintroduction of chamois in Northern Velebit. Sumarski list 11 12: [In Croatian with English summary.] Frkovic, A Chamois in Gorski Kotar. Risnjak National Park, Crni Lug, Croatia. [In Croatian.] Gafic, M., and S. Dzeko Prenj. Turisticka zajednica Kantona Sarajevo, Sarajevo, Bosnia and Herzegovina. [In Bosnian.] Goudet, J FSTAT, a programme to estimate and test gene diversities and fixation indices (version 2.9.3). fstat.html. Accessed 3 Mar Grubb, P Order Artiodactyla. Pages in D. E. Wilson, and D. M. Reeder, editors. Mammal species of the world. A taxonomic and geographic reference, third edition, volume 1. John Hopkins University Press, Baltimore, Maryland, USA. Guindon, S., and O. Gascuel A simple, fast and accurate method to estimate large phylogenies by maximum-likelihood. Systematic Biology 52: Huelsenbeck, J. P., and F. Ronquist MrBayes: Bayesian inference of phylogeny. Bioinformatics 17: Jakobsson, M., and N. A. Rosenberg CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics 23: Kalinowski, S. T The computer program STRUCTURE does not reliably identify the main genetic clusters within species: simulations and implications for human population structure. Heredity 106: Librado, P., and J. Rozas DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25: Masini, F., and S. Lovari Systematics, phylogenetic relationships, and dispersal of the chamois (Rupicapra spp.). Quaternary Research 30: Mayr, E Systematics and the Origin of Species. Columbia University Press, New York, New York, USA. Miculinic, K Archeozoological study of the former presence of larger mammals. Rewilding Europe-Velebit, Zagreb, Croatia. Miracle, P., and D. Sturdy Chamois and the karst of Herzegovina. Journal of Archaeological Science 18: Mucci, N., F. Mattuci, and E. Randi Conservation of threatened local gene pools: landscape genetics of the Italian roe deer (Capreolus c. italicus) populations. Evolutionary Ecology Research 14: Nei, M Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89: Orsanic, M, D. Tomljanovic, and J. Tomljanovic Managing forests of Northern Velebit. Pages in J. Vukelic, editors. Forests and Forestry of Northern Velebit, Hrvatske sume d.o.o., Zagreb, Croatia. [In Croatian.] Previsic, A, C. Walton, M. Kucinic, P. T. Mitrikeski, and M. Kerovec Pleistocene divergence of Dinaric Drusus endemics (Trichoptera, Limnephilidae) in multiple microrefugia within the Balkan Peninsula. Molecular Ecology 18: Rambaud, A., and A. J. Drummond Tracer, version bio.ed.ac.uk/tracer. Accessed 3 Mar Rodrıguez, F., S. Hammer, T. Perez, F. Suchentrunk, R. Lorenzini, J. Michallet, N. Mantinkova, J. Albornoz, and A. Dominguez Cytochrome b phylogeography of chamois (Rupicapra spp.). Population contractions, expansions and hybridizations governed the diversification of the genus. Journal of Heredity 100: Rodrıguez, F., T. Perez, S. Hammer, J. Albornoz, and A. Domınguez Integrating phylogeographic patterns of microsatellite and mtdna divergence to infer the evolutionary history of chamois. BMC Evolutionary Biology 10: Ronquist, F., and J. P. Huelsenbeck MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: Rosenberg, N. A DISTRUCT: a program for the graphical display of population structure. Molecular Ecology Notes 4: Rousset, F Genepop 007: a complete reimplementation of the Genepop software for Windows and Linux. Molecular Ecology Resources 8: Royo, L., G. Pajares, I. Alvares, I. Fernandez, and F. Goyache Genetic variability and differentiation in Spanish roe deer (Capreolus capreolus): a phylogeographic reassessment within the European framework. Molecular Phylogenetics and Evolution 42: Sabic, B., and N. Lalic Game management plan for state hunting ground no. XVII/1 Biokovo" for the period from 01 April 2005 to 31 March Hrvatske sume d.o.o. Zagreb, Makarska, Croatia. [In Croatian.] Safner, T., M. P. Miller, B. H. McRae, M. J. Fortin, and S. Manel Comparison of Bayesian clustering and edge detection methods for inferring boundaries in landscape genetics. International Journal of Molecular Sciences 12: Skorup, V Big game and hunting. Pages in J. Vukelic, editors. Forest and forestry of Northern Velebit. Hrvatske sume d.o.o. Zagreb, Zagreb-Senj, Croatia. [In Croatian.] Soglia, D., L. Rossi, E. Cauvin, C. Citterio, E. Ferroglio, S. Maione, P. G. Meneguz, V. Spalenza, R. Rasero, and P. Sacchi Population genetic structure of alpine chamois (Rupicapra r. rupicapra) in the Italian Alps. European Journal of Wildlife Research 56: Sprem, N., C. A. Frantz, V. Cubric-Curik, T. Safner, and I. Curik Influence of habitat fragmentation on population structure of red deer in Croatia. Mammalian Biology 78: Sprem, N., D. Zanella, D. Ugarkovic, I. Prebanic, P. Gancevic, and L. Corlatti Unimodal activity pattern in forest-dwelling chamois: typical behaviour or interspecific avoidance? European Journal of Wildlife Research 61: Tamura, K., G. Stecher, D. Peterson, A. Filipski, and S. Kumar MEGA6: Molecular Evolutionary Genetics Analysis Version 6.0. Molecular Biology and Evolution 30: Tvrtkovic, N., and M. Grubesic Chamois. Pages in N. Tvrtkovic, editor. Red Book of Mammals of Croatia. Ministry of Culture and State Institute for Nature Protection, Zagreb, Croatia. Vukelic, J Forest vegetation in Croatia. University of Zagreb Faculty of Forestry, Zagreb, Croatia. [In Croatian.] Weir, B. S., and C. C. Cockerham Estimating F-Sstatistics for the analysis of population structure. Evolution 38: Zemanova, B., P. Hajkova, J. Bryja, J. Zima Jr., A. Hajkova, and J. Zima Development of multiplex microsatellite sets for noninvasive population genetic study of the endangered Tatra chamois. Folia Zoologica 60: Associate Editor: Emily Latch. SUPPORTING INFORMATION Additional supporting information may be found in the online version of this article at the publisher s website. Sprem and Buzan The Chamois Hybridization Zone 11