Pedigree and genomic-based relationships in a dog population

In many dog breeds, small population sizes, associated with intense selection schemes, have led to considerable losses of genetic diversity. This complicates the production of accurate genomic estimates of parentage. The Lupo Italiano, with ~300 living dogs, is an Italian breed created in 1966 by crossing the Apennine grey wolf (Canis lupus italicus) with German Shepherd dogs (GSD). The aim of this work was to compare calculated relationships from genomic and pedigree data, using the Lupo Italiano as an example of a small population. The entire pedigree of the Lupo Italiano is known, consisting of up to 12 generations and dating to the founder animals. The pedigrees of 28 Lupo Italiano dogs (provided by AAALI) were used to build an additive relationships matrix (A) (CFC software). These 28 dogs were genotyped on the Illumina CanineHD 170K SNP chip (University of Milan and National Institutes of Health in Bethesda, MD), and the resultant genotypes were used for the estimation of the within-breed Genomic Relationships Matrix (GRMa) (GCTA64 software). The mean parentage values for GRMa (-0.02±0.05) and A (0.80±0.05) were not equivalent, however, they did display a significant positive correlation (R=0.75; P<0.001). Four additional populations, genotyped on the same panel, consisted of 20 Apennine grey wolves (ISPRA), 30 GSDs, 14 grey wolves, and 31 village dogs (publicly available in Dryad, Shannon et al. 2015). The GRM produced with the combined set (GRMb) led to a higher correlation with A (R=0.80; P<0.001) and to higher estimates of parentage between Lupo Italiano individuals (0.53±0.05). The calculation was expanded a final time to include an additional 250+ GSDs (GRMc). This matrix showed a decrease in the correlation with A (R=0.76; P<0.001) balanced by a strong increase in the parentage values (0.82±0.08), making it the closest to the A matrix.These results show that estimation of genomic relationships from populations with greater allelic diversity can improve the correlation and accuracy with pedigree-derived estimates. Consideration of these implementations can allow for better management of mating schemes and conservation of genetic variation in dog breeds with small population sizes.