Gene flow between wolf and shepherd dog populations in Georgia (Caucasus)

Gene flow between wolf and shepherd dog populations in Georgia (Caucasus) NATIA KOPALIANI 1, MAIA SHAKARASHVILI 1, ZURAB GURIELIDZE 1,2, TAMAR QURKHULI 1, and DAVID TARKHNISHVILI 1* 1 Institute of Ecology, Ilia State University, 3/5, K. Cholokashvili Ave., Tbilisi, 0162 Georgia. 2 Tbilisi Zoo, M.Kostava str.#64, Tbilisi, Georgia Running title: Hybridization of wolves and dogs in Georgia * Correspondence: David Tarkhnishvili, Institute of Ecology, Ilia State University, 3/5, K. Cholokashvili Ave., Tbilisi, 0162 Georgia. Tel: (+995) 77553320 E-mail: david.tarkhnishvili@iliauni.edu.ge 1

We studied the distribution of the mitochondrial DNA haplotypes and microsatellite genotypes at eight loci in 102 grey wolves, 57 livestock guarding dogs, and 9 mongrel dogs from Georgia (Caucasus). Most of the studied dogs had mitochondrial haplotypes clustered with presumably East Asian dog lineages, and most of the studied wolves had the haplotypes clustered with European wolves, but 20% of wolves and 37% of dogs shared the same mitochondrial haplotypes. Bayesian inference with STRUCTURE software suggested that over 13% of the studied wolves had detectable dog ancestry and over 10% of the dogs had detectable wolf ancestry. 2-3% of the sampled wolves and dogs were identified, with a high probability, as first generation hybrids. These results were supported by the relatedness analysis which showed that 10% of wolves and 20% of dogs had closest relatives from an opposite group. The results of the study suggest that wolf-dog hybridization is a common event in the areas where large livestock guarding dogs are held in a traditional way, and that gene flow between dogs and grey wolves was an important force influencing gene pool of dogs for millennia since early domestication events. This process may have been terminated (1) in areas outside the natural range of grey wolves and (2) since very recent time, when humans started to more tightly control contacts of purebred dogs. KEYWORDS: Canis lupus, domestic dog, grey wolf, livestock guarding dog, Caucasus, hybridization, gene flow, mitochondrial DNA, microsatellites. 2

INTRODUCTION All existent dog breeds descend from domesticated wolves (Olsen, 1985; Tsudaet al. 1997; Vilàet al. 1997; Leonard et al. 2002; Savolainen et al., 2002; Parker et al., 2008; Hindrikson et al., 2012; Larson et al., 2012). Wolves are thought to have first been domesticated in East Asia (Olsen, 1985; Ding et al., 2011; Sacks et al., 2013), although SW Asian (Clutton-Brock, 1999) and European (Thalmann et al., 2013) origin is also considered. Neither archaeological nor genetic data rule out multiple domestication events (Vonholdt et al., 2012; Larson, 2012; Thalmann et al., 2013). According to most authors, wolves were domesticated in East Asia 14,000 40,000 YA (Vilà et al.,1997; Savolainen et al., 2002; Östrander & Wayne, 2005; Verginelli et al., 2005; Sacks et al., 2013), and the domestication was followed by the expansion of dogs throughout Eurasia. Deguilloux et al. (2009) and Sacks et al. (2013) hypothesized that the expansion led to the replacement of the multiple dog lineages throughout Eurasia, the process that can be traced by the presence of mitochondrial and Y-chromosome DNA haplogroups more typical for the Western than the Eastern Eurasian wolves (Savolainen, 2002; Sacks et al., 2013). On the other hand, western haplogroups could descend from long-lasting gene introgression between dogs and Western Eurasian wolves (Ardalan, 2011; Thalmann et al., 2013). Genetic studies of dogs and wolves from the same geographic area are critical for better understanding of wolf domestication pattern: they help to infer the extent and direction of gene flow between the wild and domestic forms, and to generate ideas how potential gene flow could change gene pools of both dog breeds and wolf populations from different geographic areas since the early Holocene. Multiple researchers explored natural hybridization between wolves and dogs using molecular genetic methods (Vilà & Wayne, 1999; Randi & Lucchini, 2002; Ciucci et al., 2003; Verardi et al., 2006; Randi, 2008; Muñoz-Fuentes et al., 2010; Godinho et al., 2011; Hindrikson et al., 2012; Khosravi, Rezaei & Kaboli, 2013). General findings of these studies and related assumptions are the following: (1) hybridization, at least in Europe, is currently a rare event (Randi & Lucchini, 2002; Hindrikson et al., 2012), although up to 3

5% of wolf population may descend from the admixture events that happened in the last two centuries (Verardi et al., 2006; Godinho et al., 2011); (2) field observations suggest occasional interbreeding between male wolves and female dogs (Vilà & Wayne, 1999; Hindrikson et al., 2012), whereas genetic analyses show that the hybrids may descend matrilineally both from wolves and from dogs (Vilà et al., 2003; Godinho et al., 2011); (3) mitochondrial DNA (mt-dna) studies suggest gene introgression from wolf to dog population long after domestication (Hedrick 2009; Ardalan et al., 2011). Majority of the studies of wolf-dog interactions are geographically limited to Western Europe, where dogs are mostly under tight human control (although some feral populations do exist see Randi, 2008), and wolf populations are small and fragmented (Boitani, 1983; Vilà et al., 2003). Wolf is still a common species throughout most of Russia, Turkey, the Caucasus, Iran, Afghanistan, Mongolia, and Tibet, particularly in continental less human-populated parts of Eurasia (Mech & Boitani, 2012). It is possible that wolf-dog hybridization is currently more common in the areas where (1) feral dogs are common, or domestic dogs are only partly under control such as large livestock guarding dogs in the mountains of the Middle East and Central Asia, or (2) feral or less controlled dogs are sufficiently large to hybridize with wolves. The situation is the most appropriate for hybridization throughout Anatolia, the Caucasus, and mountainous parts of Iran, Iraq, and Turkmenistan, where large livestock guarding dogs are widespread (Rigg, 2001) (Fig. 1) and grey wolves are common. Recent study of Khosravi et al. (2013) suggests that gene flow between wolves and free-ranging dogs does exist in Iran, although the authors refrain from estimating the hybridization rates. During 2008-2012, we collected tissues and feces of wolves and shepherd dogs from different parts of Georgia (the Caucasus; Fig. 2). We studied the distribution of mtdna haplotypes and alleles at eight microsatellite loci in the studied samples. We applied a combination of analytical approaches for identifying specimens of hybrid origin both in dogs and wolves, and to infer current gene flow between 4

them. The inferred rates of hybridization appeared high enough to suggest that it has strong impact on both wolf and dog gene pools. MATERIALS AND METHODS SAMPLING We collected pieces of wolf pelts available from local hunters, blood of captured wolves, and scat samples from all over Georgia (the Caucasus) (altogether 102 wolf samples; Fig. 2, Table 1). Scats were collected by following wolf snow tracks in winter, as described in Lucchini et al. (2002), to avoid misidentification with dog scats. We collected 57 hair samples from shepherd dogs in Eastern Georgia and 9 samples of mongrel dogs. DNA EXTRACTION AND SEQUENCING DNA was extracted from hair samples of domestic dogs (Canis lupus familiaris) and from skin, blood, hair and scat samples of grey wolves (C. lupus) using the Qiagen DNeasy kit (Blood and Tissue Kit and Stool Kit) according to manufacturer s instructions. 228 bp of the mtdna control region (D-loop) was amplified using the following primers: L 15562 5 - CCATGC ATA TAAGCATGTACA T-3 and H 15790 5 - AGA TGC CAG GTA TAG TTCCA-3 (designed for this study; the primers match positions 15566-15793 of the full mitochondrial DNA sequence of Canis lupus, GenBank accession # NC008092). PCR amplifications were carried out in 20 µl volumes (2-4 µl of template DNA, 0,25 units of Promega Taq polymerase, 10x Promega Buffer, 1 mm of MgCl 2, 0.1 mm of each dntp, and primer concentrations at 0.1 mm). The thermal profile was 3 min at 95 C, followed by 46 cycles of 30 sec at 94 C, 1 min at 56 C, 1 min at 72 C,and a final extension at 72 C for 10 min and 10 C at 10 min. PCR products were checked using 1% by agarose gel with SyberSafe staining. Negative controls were used to check for contamination in each PCR reactions. Single stranded sequencing was performed with the primers of PCR, using the Big-Dye 5

Terminator v.3.1. PCR fragments were sequenced in both directions to assure sequence accuracy. ABI 3130 Genetic Analyzer was used for electrophoresis (Applied Biosystems, Inc.). The sequences were aligned with BioEdit software (Hall, 1999). High-quality sequence fragments containing 204 bp were deposited to GenBank (accession #KF048874-KF048906). A median-joining (MJ) algorithm (Donnelly & Tavaré, 1986; Bandelt, Forster & Rohl, 1999) was applied to reconstruct all possible evolutionary pathways among the inferred haplotypes of Georgian dogs and wolves and dogs and wolves sequences downloaded from GenBank (Table S1). In total, 317 sequences were used in the analyses, including 66 sequences of Georgian wolves and 57 of Georgian dogs. Software NETWORK 4.6.1.1 (Fluxus Technology Ltd.) was used for the network construction. The MJ algorithm was run repeatedly, first for estimating weights for individual substituting positions, and then for using the weighting option, in order to reduce the number of loops in the network. MICROSATELLITE GENOTYPING Eight dinucleotide microsatellites were selected for their polymorphism and reliable scorability of wolves and dogs. The used primers were CXX.103, CXX.109, CXX.121, CXX.172, CXX.173, CXX.20, CXX.200 and CXX.377 (Östrander, Sprague & Rine 1993; Östrander et al., 1995; Kitchen et al., 2005; Stronen et al., 2012). PCR amplifications were carried out in7 µl volumes using the Qiagen master mix (3 µl of Master Mix (2x), 0,6 µl of Q solution (5x0) 1,28 µl of dh2o and 1,4 µl of DNA, and primer concentrations of 0.11-0.14 mms). The thermal profile for PCR reaction was 15 min at 95 C, followed by 15 cycles of 10 sec at 94 C, 1 min and 30 sec at 61 C and 1 min at 95 C, then followed by 29 cycles of 30 sec at 94 C, 1 min and 30 sec at 53 C, 1 min at 72 C, and a final extension at 60 C for 30 min and 10 C for 10 min. Amplified DNA was run on ABI 3130, using deionized Formamade and Genescan size standard Liz 500 (Applied Biosystems Inc., Foster City. CA. USA). Genotypes were screened using Genemapper v 4.0 software package (Perkin Elmer, Waltham, MA, USA). We amplified and screened all studied loci three or more times and calculated genotyping errors (allelic dropouts, false alleles, and genotype reliability), following the guidelines of Waits, Luikart & Taberlet (2001) and Miller, Joyce & Waits (2002)(software used RelioType, GIMLET). 6

We set the reliable change index (RCI) threshold value as 95%, and verified every locus of each individual twice for heterozygosity and three times for homozygosity. POPULATION GENETIC ANALYSIS: MICROSATELLITES We used microsatellite genotypes at the eight studied loci for inferring population genetic parameters of wolves and the dogs, and estimating gene flow between the two populations. We used four algorithms for evaluating gene flow between wolves and dogs and inferring origin of individual animals: (1) traditional Fst estimates among the populations separated on a priori basis (Dobzhansky & Wright, 1941), (2) Bayesian clustering procedure using STRUCTURE algorithm (Pritchard et al. 2000; Randi, 2008), (3) evaluating relatedness among all individuals using the relatedness measure (Lynch and Ritland, 1999; Belkhir et al., 2002) and (4) Bayesian inference of migrants, using software BayesAss 3.0 (Wilson & Rannala, 2003; Rannala, 2007). We applied ARLEQUINv3.5 (Excoffieret al. 1992, 2005) for exploring genetic structure of the studied samples of wolves and dogs (genetic diversity, linkage disequilibria with 10,000 permutations per locus pair, and Hardy-Weinberg Equilibrium). We inferred gene flow rates (Nm) among the wolf and dog populations using expression Nm = 1/(4*Fst) 0.25 (Dobzhansky& Wright, 1941), where Fst is replaced by Rst for microsatellite data. While using Bayesian structure procedure, we applied software STRUCTURE v2.3 (Pritchard et al. 2000) to separate the entire dataset into two groups with least within-locus and between-locus disequilibria. MCMC parameters were set on burn-in period of 10,000 and 100,000 post-burnin samples. We repeated the procedure ten times for the number of clusters (K) equal to 2 and selected the output of the run with the lowest log-likelihood ratio, following the outline of the software manual. Considering the outlines of Randi (2008), we assumed that wolves and dogs with assignment probabilities to the respective clusters below threshold values 0.9 and 0.8 might have hybrid ancestry. To test the hypothesis on the true 7

hybrid origin of individuals with the intermediate assignment probabilities, we simulated twenty samples of purebred wolves and dogs by shuffling gametes among the conspecific animals with high (>0.9) empirical assignment probabilities. The simulated sample sizes were equal to those of the original dataset (102 wolves vs 66 dogs). For simulation procedure, functions Resample and Monte Carlo Analysis of Microsoft Excel application PopTools (Hood, 2010) were applied. For each simulated sample, the number of individuals with assignment probabilities below the threshold values was identified and compared with that of the original sample. The average difference between the inferred numbers of suspected hybrids in the empirical and generated samples was used as an index of true number of individuals with hybrid ancestry, and the 95% confidence limits were estimated by excluding the lowest and the highest values from the result set. Bayesian algorithm developed by Wilson & Rannala (2003) was used for identifying first-generation hybrids among the studied wolves and dogs (2 nd generation migrants, according to the authors terminology). This algorithm allows estimating the probability of hybrid origin for each genotyped individual. Applying this method helped to rule out the potential presence of mistakenly identified scat samples (those should demonstrate high probability of being 1 st generation migrants ). The software used was BayesAss 3.0 (Rannala, 2007). We ran the program five times, with different random number seed, with the Number of iterations to discard as burn-in equal to 100,000 and the number of iterations for MCMC equal to 10,000,000. We inferred coefficient of genetic relatedness of Lynch and Ritland (1999) r xy among all studied individuals using software IDENTIX (Belkhir et al., 2002). From this analysis, we excluded 5 wolves and 3 dogs which had non-readable genotypes at least at one out of 8 studied microsatellite loci. The coefficient varies between 0 (no relatedness) to 1 (genetic identity). Then, for each individual (wolf or dog) we calculated the average relatedness separately with all wolves and with all dogs. We treated this 8

individual as a wolf-related if the average relatedness with wolves exceeded the average relatedness with the dogs, and as a dog-related in case of the inverse. RESULTS MITOCHONDRIAL DNA HAPLOGROUPS 42 variable positions were found among all 317 studied individuals. 89 haplotypes were identified, individual haplotypes separated by 1-4 substitutions. Both Georgian wolf and Georgian dog haplotypes were deeply nested in the haplotype network consisting European, Asian, Middle Eastern, and North American wolves and dogs studied by Savolainen et al. (2002) (Fig. 3). 45 out of 66 Georgian wolves (68%) had haplotypes associated with some European (and to less extent Asian and American) wolf haplotypes but not with Georgian or other dog haplotypes. Simultaneously, 21 Georgian wolves had haplotypes close or identical to those of clades B and A of Savolainen et al. (2002). Forty shepherd dogs (70% of the studied sample) had haplotypes associated with the the clade A, 10 - with the clade B, 7 with the clade C, and one with D (Fig. 3). Thirteen Georgian wolves and 22 dogs (20% and 37% of the studied samples) had shared haplotypes (present in both Georgian wolves and Georgian dogs). This especially applied to the animals that belonged to the clade B of Savolainen et al. (2002). In summary, there is a high haplotype group sharing between the studied wolves and dogs. 14% of wolves and 17% of dogs belong to a single haplotype of the clade B of Savolainen et al. (2002), almost in equal proportion shared among the two forms. GENETIC DIVERSITY OF THE STUDIED SAMPLES - MICROSATELLITES Analysis of the microsatellite data revealed the presence of 4-16 alleles (10.8 on average) at each studied locus in the studied wolves, and 4-14 alleles (8.8 on average) in dogs. Wolves had higher mean number of alleles and higher Garza-Williamson indices than dogs. No significant differences in observed 9

or expected heterozygozities among the studied groups were recorded, although the observed heterozygozities exceeded the expected values in both groups. The number of pairs of loci significantly (Chi-square test, P<0.05) deviated from linkage disequilibria was 29% for the entire sample, 11% for wolves, and 7% for dogs (Table 2). The analysis showed significant differentiation between the dogs and wolves (exact test, P<0.01). Pairwise Fst value was 0.0637 when comparing wolf and dog populations. The calculated migration rates (Nm = 1/(4*Fst) 0.25 between the dogs and wolves was 3.67. BAYESIAN INFERENCE: WOLVES AND DOGS WITH HYBRID ANCESTRY STRUCTURE simulations showed reasonably good separation of the identified genotypes into two clusters, from here onwards W and D. Most of wolf samples had high assignment probabilities to the cluster W, and most of the dog samples had high assignment probabilities to the cluster D. However, the animals with intermediate assignment probabilities were common: 17.6% of the wolves had assignment probabilities to the cluster W falling below 0.8, and 15.2% of the dogs had assignment probabilities to the cluster D falling below 0.8. 24.5% of the wolves and 24.2% of the dogs had assignment probabilities to the respective cluster below 0.9 (Table 2). Simulation procedure described in the methods section produced some genotypes with intermediate assignment probabilities. Average (for twenty replicates) difference between the empirically obtained and simulated genotypes slightly decreased the number of the individuals with suspected hybrid ancestry: to 15.7-20.5% for wolves and 12.2-19.0 % for dogs (threshold levels 0.8 and 0.9). The respective confidence limits are shown in Table 2. The 95% confidence limits for wolves with suspected hybrid ancestry (considering true threshold level either 0.8 or 0.9) are 13.7-23.5% for wolves and 10.6-24.2% for the dogs. Only three wolves with suspected 10

hybrid ancestry were sampled from feces, hence the misidentification cannot be a potential reason for the intermediate probability assignments for at least 14 individual wolves. Bayesian inference of first generation hybrids with software BayesAss 3.0 showed high convergence among the different run outputs and consistent migration estimates and individual probabilities. At least three out of 102 wolves were shown to be first generation hybrids (=2 nd generation migrants) with probability 0.641, 0.875 and 0.908, respectively. Conversely, two out of 66 dogs were likely first generation hybrids, with probabilities 0.619 and 0.811. These individuals were the same as showing intermediate assignment probabilities with the STRUCTURE algorithm. RELATEDNESS AMONG THE INDIVIDUALS The average Lynch & Ritland sr xy was 0.29 for the entire dataset, 0.26 when comparing individual Georgian wolves with individual dogs, and 0.33 when comparing the wolves with wolves and the dogs with dogs. 9.2% of the wolves had average r xy higher when compared with dogs than when compared with the other wolves. 20.6% of the dogs had average r xy higher when compared with the wolves than when compared with other dogs, suggesting that the dogs with detectable close relations with wolves are more than twice as many as the wolves with the detectable close relations with the dogs. DISCUSSION Gene flow among wolf and dog populations in the Caucasus is substantial and exceeds that recorded for the studied European populations. This includes both gene flow from dog to wolf population and inverse gene flow from wolves to shepherd dogs. The distribution of maternal lineages suggests that the gene introgression both from wolf to dog and vice versa had a remarkable impact on the gene pools of both populations. 11

INTROGRESSION OF MT-DNA BETWEEN WOLF AND DOG IN THE CAUCASUS Domestic dogs have occurred throughout the entire mainland Eurasia since at least 12,000 YA (Larson et al., 2012). Most of the previous molecular genetic studies (Savolainen, 2002; Brown et al. 2011) supported an earlier hypothesis on the East Asian origin of dogs (Olsen, 1985) by showing that maternal haplogroups, which prevail in domestic dogs, are most diverse in East Asian wolves. A recent mt-dna study of dogs from SW Asia suggests that only 2.7% of the studied individuals had mt-dna, presumably associated with Western rather than Eastern Eurasian wolves (Ardalan et al., 2011). The authors hypothesize that low sharing of haplotypes is due to mt-dna introgression rather than to independent wolf domestication in SW Asia. However, the SE Asian domestication hypothesis is challenged by Larson et al. (2012), who suggest that, while testing alternative hypotheses, complexity of the postdomestication evolutionary pattern is underestimated. Prehistorically and in early historical time geographic distribution of dog lineages appears to have undergone substantial changes, as is evident, for instance, from ancient dog mt-dna studies (Deguilloux et al., 2009); dog populations went recently through tight bottlenecks which supposedly impoverished the gene pool of domestic dog breeds 100-150 years ago (Larson et al., 2012). Sacks et al. (2013) revised the existent views on wolf domestication, based on the analysis of Y-chromosome DNA haplogroups of dogs from throughout most of the current range. They showed that multiple domestication centers of wolves did exist throughout Eurasia, but the morphologies best adapted to human needs prevailed in SE Asia, which triggered rapid expansion of their maternal lineages. The most inclusive recent analysis of ancient Canid mitochondrial genomes (Thalmann et al., 2013) suggests that all clades, which present either in modern dogs or in fossilized dog remains from Europe, are shared with modern or ancient European wolves; this study obviously challenges the SE Asian hypothesis. Our data suggest that at least 17% of Georgian shepherd dogs had mt-dna from clade B (Savolainen et al., 2002). This clade dominated in fossil European dog remains (Deguilloux et al., 2009). Savolainen et 12

al. (2002) hypothesized an East Asian origin of this clade, based on the analysis of its internal diversity. However, they simultaneously mentioned that the clade was found in Western and not Eastern Eurasian wolves. Thalmann et al. (2013) suggest that the clade B represents a mitochondrial genome introgressed into dog gene pool from European wolves rather than one established by domestication. Our analysis of multiple published haplotypes suggests that this clade is one of the most common among European wolves, and the haplotype of this clade especially common in Europe is also common in Georgian wolves and shepherd dogs. This finding can be taken as an argument for the local maternal ancestry of a substantial part of the studied shepherd dogs. Most of the Georgian dogs were associated with the presumably East Asian (Savolainen et al., 2002; but see Thalmann et al., 2013) clade A. However, three haplotypes of this clade, found in dogs, were also present in four wolves (all identified as wolves by viewing the pelts), which may reflect gene flow between the populations. 68% of the studied wolves had haplotypes clustered in a group with no dog representatives (neither those from our sample, nor from the bibliography) but shared with Eurasian and North American wolves. Therefore, Georgian shepherd dogs combine mt-dna lineages typical for ancient and East Asian dogs with the lineages that are shared with Caucasian wolves. This can indicate very important contribution of local wolves into the gene pool of Caucasian shepherd dogs. GENE FLOW RATES AND CONSEQUENCES FOR THE WOLF AND DOG GENE POOLS If effective migration rates Nm>1, two populations cannot maintain differences solely by gene flow (Wright, 1951). In the case of domestic vs wild population of the same species, domestic breeds are likely to be maintained by artificial selection against hybrid individuals with potentially undesirable morphology or behavior. At the same time, morpho-ecological characters of the wild forms are maintained by natural selection against those that may decline reproductive success in the wild. 13

Therefore, the situation can be conceptually similar to a semi-permeable hybrid zone (sensu Arnold, 2006), with neutral genes flowing across the boundary between the hybridizing populations but nonneutral genes remaining separated. The gene flow is substantial and can lead to rethinking both dog and wolf phylogeography in the Caucasus and, possibly, West Asia, since Traditional F-statistic estimation suggests Nm > 3 between the dogs and wolves. STRUCTURE simulations predict the presence of recent wolf ancestry in more than 10% of shepherd dogs, and recent dog ancestry in more than 13% of wolves. Moreover, 2% of the studied wolves and 3% of dogs were, with a high probability, first generation hybrids. According to the Lynch & Ritland s (1999) relatedness index, 21% of dogs had closer average relatedness with wolves than with other dogs, and 9% of wolves had closer average relatedness with dogs than with other wolves. In summary, all analyses applied suggest that the gene flow between dog and wolf populations in Georgia is sufficiently high to cause gene introgression. Gene flow from dogs to wolves remarkably exceeds that identified for some regions of southern Europe (Randi & Lucchini, 2002; Verardi et al., 2006). The majority of publications describing dog-wolf hybridization suggest that mating between male dogs and female wolves is more common than the other way around (Godinho et al., 2011; Hindrikson et al., 2012). However, genetic analyses suggest that hybridization between male wolves and female dogs also occurs in nature (Hindrikson et al., 2012). The presence of dog maternal lineages in wolf populations (Ardalan et al., 2011; this study) or sharing haplotypes between wolves and dogs can only be the result of such hybridization pattern: female dogs can produce offspring both in the wild and in domestic conditions, whereas female wolves can breed mostly in the wild. Our study supports this point of view. Hybridization between male wolves and female dogs might happen both occasionally and deliberately: in mountain parts of Georgia, dogs are occasionally paired with captured wolves, which allegedly improves the breed. In such deliberate hybrid occasions, both male and female wolves can participate. The latter case is the most likely explanation of shared haplotypes between dog and wolf. 14

Throughout Western Europe, where previous wolf-dog hybridization studies were located, the dogs are largely controlled by humans, which prevent their hybridization with wolves (in areas where wild wolf populations still exists), although the proportion of wolves with hybrid ancestry exceeds 4% when feral dogs are relatively common (Randi, 2008; Godinho et al., 2011). However, these studies were not designed to evaluate mixed ancestry in feral dog population. Large livestock guarding dogs, such as Great Pyrenees, are not commonly used any more in a way that they can easily interbreed with wolves, but nobody can say to what extent they interbred with wolves in the past. Larson et al. (2012) suggested that the current gene pool of most dog breeds has been formed very recently, not more than 100-150 years ago. In that time, the breeds were subjected to extensive selection and most likely passed through narrow bottlenecks (Larson et al., 2012). The current distribution of clonally inherited genes in dogs, showing strong dominance of east-asian mt-dna haplogroups, should be viewed in this context. We hypothesize that the situation was much more flexible in earlier times, when most of the dogs used by common people were not subjected to intensive selection, similar to what is now the situation with livestock guarding dogs in the Caucasus and most likely the rest of West Asia. This means that interbreeding with grey wolves was an important part of the dog maintenance, and the situation was much more complicated than the simple pattern including Neolitic domestication with the further expansion of dogs descending from these early domesticated wolves. CONCLUDING REMARKS Gene flow between dog and wolf populations might have substantially influenced the gene pool of both since a very early period of domestication. In Europe, gene flow between dogs and wolves could be remarkable until large livestock-guarding dogs were used for sheep protection and wolves had large continuous range. High rates of gene flow suggest that the gene pool of Western Eurasian wolves could 15

integrate numerous dog maternal lineages. Phylogeographic analysis of both Eurasian wolf and dog should consider this introgression, which could result in biased interpretations. ACKNOWLEDGEMENTS The study has been financed by Georgian National Science Foundation (award no. GNSF/ST07/6-230) and Ilia State University budget. The authors thank Gigi Tevzadze for his moral and administrative support and help during the field work. Lisette Waits and Cort Anderson supervised TQ during her training at the Molecular Genetic Laboratory of the University of Idaho. Mari Murtskhvaladze provided assistance to MS and TK at the molecular genetic laboratory at Ilia State University. Two anonymous referees and R. Wayne provided useful suggestions on the first draft of the manuscript. Alexander Gavashelishvili and one of the two anonymous referees helped to correct English of the MS. REFERENCES Ardalan A, Kluetsch CFC, Zhang A-b, Erdogan M, Uhlén M, Houshmand M, Tepeli C, AshtianiSRM, Savolainen P. 2011. Comprehensive study of mtdna among Southwest Asian dogs contradicts independent domestication of wolf, but implies dog wolf hybridization. Ecology and Evolution 1: 373 385. Arnold ML. 2006. Evolution through genetic exchange. UK: Oxford University Press. Belkhir K, Castric V, Bonhomme, F. 2002. IDENTIX, a software to test for relatedness in a population using permutation methods. Molecular Ecology Notes 2: 611-614. Boitani L. 1983. Wolf and dog competition in Italy. Acta Zoologica Fennica 174: 259-264. Brown SK, Pedersen NC, Jafarishorijeh S, Bannasch DL, Ahrens KD, Wu J-T, Okon M, Sackset BN. 2011. Phylogenetic Distinctiveness of Middle Eastern and Southeast Asian Village Dog Y Chromosomes Illuminates Dog Origins. PLoS ONE 6(12): e28496. doi:10.1371/journal.pone.0028496. 16

Ciucci P, Lucchini V, Boitani L, Randi E. 2003. Dewclaws in wolves as evidence of admixed ancestry with dogs. Canadian Journal of Zoology 81: 2077 2081. Clutton-Brock J. 1995. Origins of the dog: domestication and early history. In: Serpell J (ed). The Domestic Dog, its Evolution, Behavior and Interactions With People. Cambridge University Press: Cambridge, 7 20. Deguilloux MF, Moquel J, Pemonge MH et al. 2009. Ancient DNA supports lineage replacement in European dog gene pool: insight into Neolithic South-East France. Journal of Archaeological Science 36: 513-19. Ding Z-L, Zhang Y-P, Oskarsson M, Ardalan A, Angleby H, Dahlgren L-G, Tepeli C, Kirkness E, Savolainen P. 2012. Origins of domestic dog in Southern East Asia is supported by analysis of Y-chromosome DNA. Heredity 108: 507 514. Dobzhansky T, Wright S. 1941. Genetics of natural populations.v.relation between mutation rate and accumulation of lethals in populations of Drosophila pseudoobscura. Genetics 26: 23-51. Excoffier L, Laval G, Schneider S. 2005. Arlequin (version 3.0): An integrated software package for population genetics data analysis. Evolutionary Bioinformatics Online 2005(1): 47-50. Godinho R, Llanez L, Blanco JC, Lopes S, Alvares F, Cortea EJGY, Talego J, Ferrand N. 2011. Genetic evidence for multiple events of hybridization between wolves and domestic dogs in the Iberian Peninsula. Molecular Ecology 20: 5154 5166 Hall TA. 1999. BioEdit, a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucl. Acid Symp. Ser., 41: 95-98. Hindrikson M, Männil P, Ozolins J, Krzywinski A, Saarma U. 2012. Bucking the Trend in Wolf-Dog Hybridization: First Evidence from Europe of Hybridization between Female Dogs and Male Wolves. PLoS ONE 7 (10): e46465. doi:10.1371/journal.pone.0046465 17

Hood GM. 2010. PopTools version 3.2.3. Available on the internet. Available at: http://www.poptools.org (accessed 15 September 2013). Khosravi R, Rezaei HR, Kaboli M. 2013. Detecting hybridization between Iranian wild wolf (Canis lupus pallipes) and free-ranging domestic dog (Canis familiaris) by analysis of microsatellite markers. Zool. Sci. 30(1):27-34. Kitchen AM, Gese EM, Waits LP, Karki SM, Schauster ER. 2005. Genetic and spatial structure within a swift fox population. USDA National Wildlife Research Center - Staff Publications. Journal of Animal Ecology 74: 1173 1181. Larson G, Karlsson EK, Perri A, Webster MT, Ho SYW, Stahl PW, Piper PJ, Lingaas F, Fredholm M, Comstock KE, Peters J, Modiano JF, Schelling C, Agoulnik A, Leegwater PA, Dobney K, Vigne J-D, Andersson L, Lindblad-Toh K. 2012. Rethinking dog domestication by integrating genetics, archeology, and biogeography. PNAS 109/23: 238878 8883 Lynch M, Ritland K. 1999. Estimation of pairwise relatedness with molecular markers. Genetics 152: 1753-1766. Leonard JA, Wayne RK, Wheeler J, Valadez R, Guillén S, Vilà C. 2002. Ancient DNA Evidence for Old World Origin of New World Dogs. Science 5598: 1613-1616. Lucchini V, Fabbri E, Marucco F, Ricci S, Boitani L, Randi E. 2002. Noninvasive molecular tracking of colonizing wolf (Canis lupus) packs in the western Italian Alps. Molecular Ecology 11: 857-868. Mech LD, Boitani L. (IUCNSSC Wolf Specialist Group) 2010. Canis lupus. In: IUCN 2012. IUCN Red List of Threatened Species. Version 2012.2.<www.iucnredlist.org>. Downloaded on 06 June 2013. Miller CR, Joyce P, Waits LP. 2002. Assessing allelic dropout and genotype reliability using maximum likelihood. Genetics 160: 357 366. Munoz-Fuentes V, Darimont CT, Paquet PC, Leonard JA. 2010. The genetic legacy of extirpation and recolonization in Vancouver Island wolves. Conservation Genetics 11: 547 556. Olsen SJ. 1985. Origins of the Domestic Dog: The Fossil Record. Tucson: University of Arizona Press. 18

Östrander EA, Sprague GF, Rine J. 1993. Identificationand characterization of dinucleotide repeat (CA) n markers for genetic mapping in dog. Genomics 19: 207 213. Östrander EA, Mapa FA, Yee M, Rine J. 1995. One hundred and one new simple sequence repeat-based markers for the canine genome. Mammalian Genome 6: 192 195. Östrander EA, Wayne RK. 2005. The canine genome. Genome Res. 15(12): 1706-1716. Parker HG, Kim LV, Sutter NB, Carlson S, Lorentzen TD, Malek TB, Johnson GS, De France HB, Östrander EA, Kruglyak L. 2004. Genetic structure of the purebred domestic dog. Science 304: 1160-1164. Randi E, Lucchini V. 2002. Detecting rare introgression of domestic dog genes into wild wolf (Canis lupus) populations by Bayesian admixture analyses of microsatellite variation. Conservation Genetics 3: 31 45. Randi E. 2008. Detecting hybridization between wild species and their domesticated relatives. Molecular Ecology 17: 285 293. Rannala B. 2007. BayesAss Edition 3.0 User s Manual. University of California, Davis. Last updated 12 April 2013. Pritchard JK, Stephens M, Donnelly P. 2000. Inference of population structure using multilocus genotype data. Genetics 155: 945 959. Rigg R. 2001. Livestock guarding dogs: their current use worldwide. IUCN/SSC Canid Specialist Group Occasional Paper No 1 [online]. URL: http://www.canids.org/occasionalpapers/. Downloaded on 06 June 2013. Sacks BN, Brown SK, Stephens D, Pedersen NC, Wu JT, Berry O. 2013. Y chromosome analysis of dingoes and southeast Asian village dogs suggests a Neolithic continental expansion from Southeast Asia followed by multiple Austronesian dispersals. Molecular Biology and Evolution 30: 1103-1118. Savolainen P, Zhang YP, Luo J, Lundeberg J, Leitner T. 2002. Genetic evidence for an East Asian origin of domestic dogs. Science 298: 1610-1613. 19

Stronen AV, Tessier N, Jolicoeur H, Paquet PC, Henault M, Villemure M, Patterson BR, Sallows T, Goulet G, Lapointe F-J.2012. Canid hybridization: contemporary evolution in human-modified landscapes. Ecology and Evolution doi: 10.1002/ece3.335 Tamura K, Peterson D, Peterson N, Stecher G, Nei M, and Kumar S. 2011. MEGA5: Molecular Evolutionary Genetics Analysis using Maximum Likelihood, Evolutionary Distance, and Maximum Parsimony Methods. Molecular Biology and Evolution 28: 2731-2739. Thalmann O, Shapiro B, Cui, P, Schuenemann VJ, Sawyer SK, Greenfield DL, Germonpré MB, Sablin MV, López-Giráldez F, Domingo-Roura X, Napierala H, Uerpmann H-P, Loponte DM, Acosta AA, Giemsch L, Schmitz RW, Worthington B, Buikstra JE, Druzhkova A, Graphodatsky AS, Ovodov ND, Wahlberg N, Freedman AH, Schweizer RM, Koepfli K-P, Leonard JA, Meyer M, Krause J, Pääbo S, Green RE, Wayne RK. 2013. Ancient Canids Suggest a European Origin of Domestic Dogs. Science 342:871-874. Tsuda K, Kikkawa Y, Yonekawa H, Tanabe Y. 1997. Extensive interbreeding occurred among multiple matriarchal ancestors during the domestication of dogs: evidence from inter- and intraspecies polymorphisms in the D-loop region of mitochondrial DNA between dogs and wolves. Genes and Genetic Systems 72: 229-38. Verardi A, Luccini V, Randi E.2006. Detecting introgressive hybridization between free-ranging domestic dogs and wild wolves (Canis lupus) by admixture linkage disequilibrium analysis. Molecular Ecology 15: 2845 2855. Verginelli F, Capelli C, Coia V, Musiani M, Falchetti M, Ottini L, Palmirotta R, Tagliacozzo A, De Grossi M, Mariani-Costantini R. 2005. Mitochondrial DNA from Prehistoric Canids Highlights Relationships Between Dogs and South-East European Wolves. Molecular Biology and Evolution 22: 2541 2551. Vilà C, Savolainen P, Maldonado JE, AmorimIR, Rice JE, Honeycutt R L,Crandall KA, Lundeberg J, Wayne RK. 1997. Multiple and ancient origins of the domestic dog. Science 276: 1687 1689. Vilà C, Wayne RK. 1999. Hybridization between wolves and dogs. Conservation Biology 13:195 198. 20

Vilà C, Walker C, Sundqvist AK, Flagstad O, Andersone Z, Casulli A, Kojola I, Valdmann H, Halverson J, Ellegren H. 2003. Combined use of maternal, paternal and bi-parental genetic markers for the identification of wolf-dog hybrids. Heredity 90: 17 24. Vonholdt BM, Pollinger JP, LohmuellerKE, Han E, Parker HG, Quignon P, Degenhardt JD, Boyko AR, Earl DA, Auton A, Reynolds A, Bryc K, Brisbin A, Knowles JC, Mosher DS, Spady TC, Elkahloun A, Geffen E, Pilot M, Jedrzejewski W, Greco C, Randi E, Bannasch D, Wilton A, Shearman J, Musiani M, Cargill M, Jones PG, Qian Z, Huang W, Ding ZL, Zhang YP, Bustamante CD, Östrander EA, Novembre J, Wayne RK. 2010. Genome-wide SNP and haplotype analyses reveal a rich history underlying dog domestication. Nature 464: 898 902. Waits LP, Luikart G, Taberlet P. 2001. Estimating the probability of identity among genotypes in natural populations, cautions and guidelines. Molecular Ecology 10: 249 256. Wilson GA,Rannala B. 2003. Bayesian inference of recent migration rates using multilocus genotypes. Genetics 163: 1177-1191. Wright S. 1951. The genetical structure of populations. Annals of Eugenics 15: 323 354. 21

Table 1. Studied samples of wolves and dogs from Georgia Sample Scats Skin pieces/ hair/ blood STR mt-dna Grey wolves (eastern Georgia) 53 20 73 39 Grey wolves (western Georgia) 2 27 29 26 Livestock guarding dogs 0 57 57 53 Mongrel dogs 0 9 9 9 22

Table 2.Genetic diversity of wolves and that of dogs from Georgia. Ho observed heterozygozity, He expected heterozygozity; NA allele number; GW Garza-Williamson Index; GWM- modified Garza-Williamson Index; PLD proportion of alleles in linkage disequilibria (from 28 total allele pairs). Variable wolves dogs all samples Ho Mean ± SD 0.780 ± 0.112 0.826 ± 0.219 0.760 ± 0.124 He Mean ± SD 0.738 ± 0.094 0.729 ± 0.188 0.760 ± 0.124 NA Mean 10.75-3.81 8.75-3.62 12.00-4.38 GW Mean ± SD 0.936 ± 0.077 0.789 ± 0.148 0. 0.833 ± 0.117 GWM Mean ± SD 0.840 ± 0.125 0.668 ± 0.119 0.715 ± 0.116 PLD proportion 0.107 0.071 0.285 23

Table 3. Assignment probabilities of the eastern and western wolves and dogs to the two clusters (W and D) identified using Bayesian inference. The proportion of the individuals is shown which assignment probability to a cluster n equal or exceeding: 0.9, 0.8, 0.5, 0.2, and means of posterior distribution. Lower panel (corrected average) shows respective figures corrected using simulated purebred genotypes (see text for details), arithmetic mean ± standard error for 20 simulations. Cluster wolves (w) dogs (d) (empirical) 0.95 0.598 0.606 0.9 0.755 0.758 0.8 0.824 0.848 0.7 0.873 0.909 0.6 0.882 0.955 0.5 0.902 0.970 means 0.859 0.901 (corrected average) 0.95 0.678± 0.005 0.695± 0.008 0.9 0.795± 0.004 0.810± 0.006 0.8 0.843± 0.003 0.878± 0.005 0.7 0.833± 0.002 0.928± 0.004 0.6 0.891± 0.002 0.970± 0.003 0.5 0.908± 0.002 0.981± 0.003 24

Figure Legends Fig. 1. Upper panel - Livestock guarding dog; middle panel livestock guarding dog with the inferred wolf ancestry (1 st generation hybrid); lower panel wolf (all from Kazbegi, Georgia) Fig. 2. Sampling locations. Open circles wolves from Western Georgia; closed circles wolves from Eastern Georgia; asterisks dogs. Fig. 3. Median-joining network showing genetic links among the wolf and dog haplotypes inferred during this study and ones downloaded from the GenBank. Light green Georgian dogs; dark green Georgian wolves; yellow European wolves; red East Asian wolves; pink West Asian wolves; blue North American wolves; red forward diagonal lines dogs, clade A (after Savolainen et al., 2002); black forward diagonal lines dogs, clade B; vertical lines clade C; horizontal lines clade D. Individual haplotypes separated by 1-4 substitutions (only shown if the number of substitutions > 1). Fig. 4. The distribution of the assignment probabilities to the clusters (K=2) identified using Bayesian inference, in Georgian wolves and in livestock guarding dogs. 25

26

Fig. 2 27

Fig. 3 28

Fig. 4. 29

Table S1. GenBank accession numbers of wolf and dog sequences downloaded from the GenBank and used in the mt-dna analysis and the respective references. Sequence Number FJ978005 - FJ978035 Source Area Publication Mitochondrion Canis lupus (gray wolf) Europe Pilot M, Branicki W, Jedrzejewski W, Goszczynski J, Jedrzejewska B, Dykyy I, Shkvyrya M, Tsingarska E. 2010. Phylogeographic history of grey wolves in Europe. BMC Evol. Biol. 10. 104 GQ376509 JF311414 JN182077- JN182088 JX415351- JX415353 JX508633- JX508635 AF530561- AF530566 AF531654- AF531741 AF115687- AF115692 AF115693 AF115694 Mitochondrion Canis lupus (gray wolf) Mitochondrion Canis lupus (gray wolf) Mitochondrion Canis lupus (gray wolf) Mitochondrion Canis lupus chanco (Mongolian wolf) Mitochondrion Canis lupus (gray wolf) Mitochondrion Canis lupus (gray wolf) Mitochondrion Canis lupus (gray wolf) North America North America East Europe Central Asia East Europe Eurasia Weckworth BV, Talbot SL, Cook JA. 2010. Phylogeography of wolves (Canis lupus) in the Pacific Northwest. J. Mammal. 91 (2): 363-375 Weckworth BV, Dawson NG, Talbot SL, Flamme MJ, Cook JA. 2011. Going Coastal: Shared Evolutionary History between Coastal British Columbia and Southeast Alaska Wolves (Canis lupus). PLoS ONE 6 (5) Hindrikson M, Mannil P, Ozolins J, Krzywinski A, Saarma U. 2012. Bucking the trend in wolf-dog hybridization: first evidence from Europe of hybridization between female dogs and male wolves. PLoS One 7(10) Dou HL. 2012. Genetic research of wolf population by mtdna D-loop region (unpublished) Baltrunaite L, Balciauskas L, Akesson M. 2013. The genetic structure of the Lithuanian wolf population. Cent Eur J Biol 8 (5),:440-447 Savolainen P, Zhang YP, Luo J, Lundeberg J. and LeitnerT. 2002. Genetic evidence for an East Asian origin of domestic dogs. Science 298 (5598): 1610-1613 Europe Randi E, Lucchini V, Christensen MF, Mucci N, Funk SM, Dolf G, Loeschcke V. 2000. Mitochondrial DNA variability in Italian and East European wolves: detecting the consequences of small population size and hybridization. Conservation Biology. Volume 14: 464 473 30

AF115695 AF115696 AF115697 AF115698 - AF115704 AF115707 AF115714 AY172677 AY172678- AY172681 AY570181 AY812736- AY812741 AB007374- AB007379 Mitochondrion Canis lupus (gray wolf) Mitochondrion Canis lupus (gray wolf) Mitochondrion Canis lupus (gray wolf) Mitochondrion Canis lupus pallipes (Indian wolf) Asia Asia North America Asia Yang YH, Kim NY and Kim KI. 2002.Phylogenetic relationships among Asian and other dog breedsdetermined by using mitochondrial DNA D-loop sequence polymorphism (unpublished) Ardalan A, Savolainen P, Houshmand M, Miraei-Ashtiani SR. 2004. Partial D-loop nucleotide sequence of wolf mitochondrion (unpublished) Leonard JA, Vila C, Wayne RK. 2005. Legacy lost: genetic variability and population size of extirpatedus grey wolves (Canis lupus). Mol. Ecol. 14 (1): 9-17 Tsuda K, Kikkawa Y, Yonekawa H, Tanabe Y. 1997. Extensive interbreeding occurred among multiple matriarchal ancestors during the domestication of dogs: evidence from inter-and intraspecies polymorphisms in the D-loop region of mitochondrial DNA between dogs and wolves. Genes Genet. Syst. 72 (4): 229-238 31