Short Communication Limited genetic divergence between dog breeds from geographically isolated countries

BREED specifications for pure-bred/pedigree dogs are laid down by the organisations that register and judge dogs, such as the Kennel Club of the UK (KCUK) and the Kennel Union of South Africa (KUSA), as well as the umbrella body the Fédération Cynologique Internationale (FCI). Reproductive isolation between breeds results because animals can only be registered as a specific breed if they have five previous generations of ancestors registered as the same breed, creating a ‘breed barrier ’ which promotes genetic differences among breeds. This genetic isolation in a population of UK dogs (Mellanby and others 2013) has been demonstrated previously. The lowest heterozygosities of around 0.5 were found for breeds such as the German shepherd dog, rottweiler and boxer dog, indicative of a reasonably high level of inbreeding. Labrador retrievers and border collies had heterozygosities of around 0.7, indicating greater genetic diversity. The Jack Russell terrier is not a registered breed with the KCUK. This group had a heterozygosity of close to 0.8 and maintained a high degree of genetic variability. The authors were interested to assess whether dog breeds in another country had similar levels of genetic isolation. Three popular breeds were chosen, which showed differing heterozygosities in the UK: the German shepherd dog, the labrador retriever and the Jack Russell terrier (Mellanby and others 2013). Breed specifications for the labrador retriever were identical for the KCUK and the KUSA, whereas the specification for the German shepherd dog in South Africa is taken from the FCI criteria rather than the KCUK criteria which are very similar. The Jack Russell is not a KCUK-registered breed (Mellanby and others 2013) but is registered with the KUSA, conforming to the breed standard of the Australian National Kennel Council. Therefore, it might be expected to show less genetic diversity in South Africa due to the need to conform to breed specifications. Breed standards are available at https://www.thekennelclub.org. uk/services/public/breed/Default.aspx and http://www.kusa.co. za/index.php/documents/breed-standards. Genotypes using 15 microsatellite markers were established for 14 labrador retrievers, 26 German shepherd dogs and 35 Jack Russell terriers from South Africa, using the previously published protocol (Ogden and others 2011, Mellanby and others 2013). Calculations of heterozygosity and relative genetic distance were performed as described (Mellanby and others 2013). Table 1 shows the calculated heterozygosities and genetic distance for the animals from South Africa, compared with the results for UK dogs. Heterozygosities did not differ between countries, but the pairwise FST values were lower for South African dogs, indicating that the UK dogs have experienced greater genetic isolation (an FSTof 1 indicates complete isolation). Other studies also using microsatellites have achieved comparable heterozygosities. For example, Irion and others (2003) found heterozygosities of 0.64 and 0.78 for labrador retrievers and Jack Russell terriers, respectively, based on 100 microsatellite markers, and Leroy and others (2009) observed heterozygosities of 0.56 and 0.60 for German shepherd dogs and labrador retrievers, respectively. Wade (2011) reported heterozygosity of 0.31 in labrador retrievers using single nucleotide polymorphism (SNP) markers (unknown number). The lower heterozygosity with SNP markers is consistent with the biallelic nature of SNPs which reduces the maximum possible heterozygosity for each SNP to 0.5, compared with the multiallelic microsatellite markers used here with maximum heterozygosity ranging from 0.84 (six alleles) to 0.95 (20 alleles). The software STRUCTURE (Pritchard and others 2000) was then used to identify the mixture of ancestral groups in each dog. This program uses genotype data from many loci to investigate population structure. Results allow the inference of distinct populations within an admixed group. In each STRUCTURE run the number of population subgroups (K) is specified and runs are repeated at increasing values of K until no further substructure is detected. The lowest value of K which detects all