Founder events, isolation, and inbreeding: Intercontinental genetic structure of the domestic ferret

Abstract Domestication and breeding for human‐desired morphological traits can reduce population genetic diversity via founder events and artificial selection, resulting in inbreeding depression and genetic disorders. The ferret (Mustela putorius furo) was domesticated from European polecats (M. putorius), transported to multiple continents, and has been artificially selected for several traits. The ferret is now a common pet, a laboratory model organism, and feral ferrets can impact native biodiversity. We hypothesized global ferret trade resulted in distinct international genetic clusters and that ferrets transported to other continents would have lower genetic diversity than ferrets from Europe because of extreme founder events and no hybridization with wild polecats or genetically diverse ferrets. To assess these hypotheses, we genotyped 765 ferrets at 31 microsatellites from 11 countries among the continents of North America, Europe, and Australia and estimated population structure and genetic diversity. Fifteen M. putorius were genotyped for comparison. Our study indicated ferrets exhibit geographically distinct clusters and highlights the low genetic variation in certain countries. Australian and North American clusters have the lowest genetic diversities and highest inbreeding metrics whereas the United Kingdom (UK) cluster exhibited intermediate genetic diversity. Non‐UK European ferrets had high genetic diversity, possibly a result of introgression with wild polecats. Notably, Hungarian ferrets had the highest genetic diversity and Hungary is the only country sampled with two wild polecat species. Our research has broad social, economic, and biomedical importance. Ferret owners and veterinarians should be made aware of potential inbreeding depression. Breeders in North America and Australia would benefit by incorporating genetically diverse ferrets from mainland Europe. Laboratories using ferrets as biomedical organisms should consider diversifying their genetic stock and incorporating genetic information into bioassays. These results also have forensic applications for conserving the genetics of wild polecat species and for identifying and managing sources of feral ferrets causing ecosystem damage.

On a global scale, pet owners and veterinarians would be informed of genetic diversity and potential inbreeding depression (Fox & Marini, 2014) and breeders trying to avoid genetic disorders would benefit by applying information about genetic diversity to regional and international breeding programs (Howard, Lynch, Santymire, Marinari, & Wildt, 2015;Willoughby et al., 2015). Biomedical laboratories using ferrets as model organisms would benefit by incorporating genetic diversity information into their bioassays because inbred individuals can be more susceptible to disease exposures (Ball, 2006;Ilmonen et al., 2008). These data could also be used to identify and conserve wild polecat species which could face reductions in genetic diversity when hybridized with feral ferrets (Costa et al., 2013;Rhymer & Simberloff, 1996). Similarly, the conservation genetics approach used here could have forensic and wildlife management applications for identifying and managing sources of feral ferrets causing ecosystem damage (O'Donnell et al., 2017;Wells, 2009).
Although historical artificial selection for hunting ability (Carnegie, 2013;Owen, 1984) and contemporary artificial selection for coat colors and patterns (Blaszczyk et al., 2007;Lewington, 2007b;Piazza et al., 2014) likely had a large impact on ferret translocations and genetic diversity, international differences in trade laws and breeding programs could also have limited or reduced the genetic diversity of certain populations (Lee, 2002;Northern Territory Government, 2016;Queensland Government, 2016;Willoughby et al., 2015).
Our goal was to evaluate the genetic structure and levels of genetic diversity and inbreeding in pet ferrets from multiple countries among the continents of North America, Europe, and Australia. We hypothesized global ferret trade resulted in distinct international genetic clusters. We also hypothesized ferrets transported to other continents would have lower genetic diversity than ferrets from Europe because of extreme founder events and no opportunities to hybridize with wild polecats or genetically diverse ferrets. Our international and intercontinental assessment of ferret population genetics provides broad insights into global patterns of founder events, and the impacts of isolation and inbreeding on the population structure of a domesticated species (Larson & Burger, 2013). Our results will be of broad social, economic, and biomedical importance and have direct applications to the ferret industry.  (Costa et al., 2013). Any individuals known (via breeding programs) or suspected by the owners to be hybrids were removed from analyses and were not included in this sample. Swab samples were stored at −70°C, whereas hair was stored in paper envelopes at stable room temperature. DNA was extracted from buccal swabs or hair follicles from 2010 to 2011 following the exact protocols of Ernest et al. (2012).

| Population genetic structure
We used three approaches to assess population structure, including F statistics, Bayesian population assignment models, and a discriminant analysis of principal components (DAPC). Population divergence (F ST ) was calculated in GenAlEx 6.502 (Peakall & Smouse, 2006, and significance testing was based on 999 permutations. We used spatially explicit hierarchical Bayesian clustering programs GENELAND 4.0  and TESS 2.3 (Durand, Chen, & François, 2009) to assess population assignments, and adegenet 2.0.1 (Jombart, 2008) for the DAPC.
In GENELAND, the number of populations (K) is a parameter optimized by the model. We followed developer recommendations for determining K and individual population assignments (Guillot, Estoup, Mortier, & Cosson, 2005). First, we ran 15 spatial models allowing K to vary from 1 to 10. All models converged on the same K. Thus, we ran five additional models fixing K at the mode and selected the model with the highest log-likelihood to run an admixture model. Each run included 100,000 iterations, a thinning interval of 1,000, and a 25% burn-in period prior to extracting model output.
In TESS, K must be specified and tested over a range of possible values. Model selection must be used to determine the K with the best fit to the data. We followed developer instructions for determining K and population assignments. First, we ran 10 nonadmixture models for each K from 2 to 10. For model comparisons, TESS computes a deviance information criterion (DIC). We ran 10 spatially conditional autoregressive admixture models for each K to the DIC plateau of nonadmixture models. All models included pairwise great circle geographic distances for weighting the Voronoi neighborhood, 100,000 iterations, and a 25% burn-in period. We retained 20% of the models which contained the lowest DIC scores and used CLUMPP 1.1.2 to perform model averaging (Jakobsson & Rosenberg, 2007).
To assess whether spatial priors or sample size were driving population assignments (Puechmaille, 2016), we also used program STRUCTURE 2.3.4 (Pritchard et al., 2000). We randomly subsetted the samples from each country to 15, which was the number of wild polecats sampled. Using 180 subsetted samples, we ran 10 admixture models for each K from 1 to 10 with uniform, nonspatial priors, including 100,000 iterations and a 25% burn-in period.
Because the algorithm for individual assignments in adegenet is not as powerful as Bayesian population assignment algorithms (Jombart, Devillard, & Balloux, 2010), we defined populations in the DAPC using countries as focal groups and treated wild European polecats as a separate group. We retained all principal components for discriminant analyses.

| Microsatellite loci and genetic diversity
Tests for linkage disequilibrium, deviations from Hardy-Weinberg proportions, and null alleles were assessed in GENEPOP 4.5.1 (Rousset, 2008). Given the major differences in sample sizes among countries, we used genetic diversity estimates that are robust to sample size or accounted for sample size. We calculated unbiased expected heterozygosity in GenAlEx (Nei, 1978). To measure the number of alleles, we calculated allelic richness using sample sizecorrecting rarefaction methods in FSTAT 2.9.3.2 (Goudet, 1995;Kalinowski, 2004). To assess inbreeding, we calculated internal relatedness using package Rhh 1.0.2 in Program R 3.3.0 (Alho, Valimaki, & Merila, 2010). Internal relatedness measures a relative outbred-inbred continuum, where negative values are suggestive of outbred individuals and positive scores are suggestive of inbreeding (Amos et al., 2001). Internal relatedness is an extension of standardized heterozygosity which standardizes estimates based on allele frequencies of the entire global sample of ferrets (Amos et al., 2001). We reported both the raw number or private alleles and the percent of private alleles, which was standardized to sample size.

| Population genetic structure
All 31 loci were polymorphic among pooled samples. Only 3% of individuals had any missing data. Most individuals with missing data were missing allelic information at a single locus; however, two individuals had two missing loci, and a single individual had four missing loci. There was no evidence for null alleles or deviations from Hardy-Weinberg proportions in each assigned population after Bonferroni corrections. European polecats had the lowest F ST values with ferrets from England and Scotland and the highest differentiation with ferrets from Australia, Canada, and the United States (Table 1). Australia had high pairwise F ST values with every country except New Zealand and the Netherlands. European countries were not strongly differentiated but exhibited moderate differentiation with the United States and Canada. When population differentiation was assessed based on population assignments in GENELAND, the United Kingdom (UK: England & Scotland ferrets) cluster was not strongly differentiated from non-UK European cluster (Table 1). The North American cluster was strongly differentiated from the Australian cluster and moderately differentiated from both European clusters.
Program GENELAND identified four genetic clusters (Figure 1a

| Microsatellite loci and genetic diversity
Based on genetic diversity estimates that are robust to sample size or account for sample size, ferrets from Australia, Canada, and the United States had the lowest genetic diversity (   European polecats all exhibited variation at the microsatellite loci developed from the domestic ferret (Ernest et al., 2012), implying common ancestry (Blandford, 1987;Hosoda et al., 2000;Kurose et al., 2000Kurose et al., , 2008. However, we did not sample other polecat species and therefore cannot determine whether the domestic ferret was domesticated from other possible species. Identifying the source location for each domestic cluster remains difficult because we do not have samples from each European country where ferrets are present and ferrets in Europe could potentially be hybridized with wild polecats or potentially other species (Costa et al., 2013;Lodé et al., 2005;Pitt, 1921;Poole, 1972), reducing any signal of relationships (Davison et al., 1999;Marmi et al., 2004). Although spatial population assignment models did not differentiate polecats from ferrets, a subsampled dataset with equal sample sizes and uniform priors indicated a genetic distinction, which is consistent with the DAPC.

| DISCUSSION
A previous study observed high expected heterozygosities (mean ± SE: 0.58 ± 0.12) and allelic richness (3.93 ± 0.13) among populations of European polecats from mainland Europe (Pertoldi et al., 2006). Cross-breeding between ferrets and genetically diverse polecats could explain the high genetic diversity observed in non-UK European countries. In contrast, polecats in the UK experienced a dramatic reduction in population size during the 19th century (Langley & Yalden, 1977), leading to a genetic bottleneck (Costa et al., 2013). Our observations of genetic diversity for polecats from UK are consistent with previous reports (Costa et al., 2013), which are higher than those of domestic ferrets but lower than polecats from mainland Europe (Moller et al., 2004;Pertoldi et al., 2006). Ferret hybridization is common in the UK but does not seem to be increasing genetic diversity in pet ferrets, possibly because ferret hybridization is occurring with a genetically recovering polecat population that recently went through a bottleneck (Costa et al., 2013). Additionally, three of the 15 polecats assigned as UK ferrets using STRUCTURE, which could indicate we sampled 12 wild polecats and three feral ferrets. Combined, this could explain the high internal relatedness observed in polecats, despite the high genetic diversity. Notably, Hungarian then Danish ferrets had the most private alleles and the highest measures of genetic diversity.
Hungary is the only country we sampled with both European polecats (M. putorius) and Steppe polecats (M. eversmanii Lesson, 1872) (Lanszki & Heltai, 2007;Šálek et al., 2013). Similarly, Denmark has both European polecats and American mink (Neovision vison Schreber, 1777) (Hammershøj et al., 2006). Although we do not have direct evidence for hybridization, the mechanism leading to unique alleles in these countries could be a result of hybridization or large effective population sizes.
None of the countries in which we sampled ferrets had an expected heterozygosity (mean among all domestic ferrets in our dataset: 0.41 ± 0.10) or allelic richness (mean among all domestic ferrets in our dataset: 2.80 ± 0.10) approaching those of their wild counterparts (Moller et al., 2004;Pertoldi et al., 2006), indicating founder events, genetic drift, and/or inbreeding have affected ferret genetic diversity (Diamond, 2002;Petersson et al., 1996;Teletchea & Fontaine, 2014).
One of the primary ways to reverse inbreeding depression is through genetic restoration (Frankham, 2015;Whiteley, Fitzpatrick, Funk, & Tallmon, 2015). In domestic animal populations, genetic restoration requires human intervention and breeding programs (Ralls & Ballou, 1986;Rollinson et al., 2014). However, international trade laws could limit the feasibility of introducing new genetic material to genetically depauperate pet ferret populations, especially on continents without potential for interbreeding with wild polecats. For example, ferrets are common pets in Australia (Talbot, Freire, & Wassens, 2014) and our study supports the assertion that inbreeding has been a national problem (Lewington, 2007b). However, ferret importation is currently illegal in Australia (Department of the Environment and Energy, 2017) and ferrets are completely prohibited in Queensland (Queensland Government, 2016) and the Northern Territory (Northern Territory Government, 2016). Despite New Zealand allowing ferret exportation for permitted breeders, the purchase and importation of ferrets were banned because of damage to native species (Lee, 2002;O'Donnell et al., 2017;Wells, 2009). Thus, Australia and New Zealand breeders should actively minimize inbreeding among currently available ferrets.
We do not see any particular utility for maintaining genetically distinct ferret clusters in most cases. Thus, for countries with inbred ferrets, we recommend that pet and laboratory breeding programs should incorporate ferrets from other countries. Although outbreeding depression could result from such crosses, the potential for inbreeding depression is much more likely (Rollinson et al., 2014). Additionally, it could also be genetically beneficial for breeding programs to introduce new genes into their domestic lines through ethical and legal crossbreeding with wild polecats. However, this could potentially result in ferrets with different behaviors than those that are considered desirable for pets (Hernádi et al., 2012) or the introduction of unwanted diseases. If this action is taken, breeders should ensure ferrets do not breed into the wild population, which could put the wild population at risk of genetic disorders (Costa et al., 2013;Rhymer & Simberloff, 1996). To reduce potential risk of inbreeding depression, ferret breeding programs designed for specific coat colors should consider mating unrelated individuals and occasionally "diluting" specific coat-color lines with individuals of different varieties.  (Williams et al., 1996). The opportunity for increasing domestic ferret genetic diversity in North America is currently available and should be considered given the application of ferrets as model organisms in North American biomedical laboratories (Ball, 2006;Jones et al., 2017;Peng et al., 2014;Porter, 2016;Vanchieri, 2001). Researchers should consider the effect low genetic diversity and inbreeding can have on the results and inferences from disease exposure experiments (Spielman, Brook, Briscoe, & Frankham, 2004;Whiteman, Matson, Bollmer, & Parker, 2006). Although causal links between neutral genetic markers and neoplasms cannot be made, our research highlights the lack of genetic variation present in certain ferret clusters. With the sequencing of the ferret genome being recently completed (Peng et al., 2014), researchers can now search the genome for specific genes which may be linked to cancer acquisition in this developing cancer model system (Aizawa et al., 2013(Aizawa et al., , 2016. Overall, our study identified international and intercontinental domestic ferret population structure and indicates Australian and North American ferrets have the lowest genetic diversities and are most highly diverged from the European polecat. Given current importation bans, New Zealand ferrets are essentially isolated and, in the future, may also exhibit patterns of low genetic diversity and inbreeding. Prevention or mitigation of inbreeding depression will require international cooperation among breeding programs and should include genetically diverse ferrets from mainland Europe. In some cases, political factors limit the ability of potential international breeding programs. This research highlights the need for politicians, coat-oriented breeders, and laboratory-stock breeders to reassess their policies. Given the close relatedness among domestic ferrets, European polecats, Steppe polecats, black-footed ferrets, American mink, and European mink (Cabria et al., 2011;Kurose et al., 2008;Lodé et al., 2005;Williams et al., 1996), the conservation genetics approach used in this study has practical applications for conserving the genetic diversity of related wild species that could face reduction in genetic diversity when hybridized with feral ferrets (Bonesi & Palazon, 2007;Šálek et al., 2013;Wisely, Buskirk, Fleming, McDonald, & Ostrander, 2002) and for the identification and management of feral ferret source populations causing ecosystem damage (Bodey, Bearhop, & McDonald, 2011;Byrom, Caley, Paterson, & Nugent, 2015;O'Donnell et al., 2017;Wells, 2009).