Population structure and breed composition prediction in a multi-breed sheep population using genome-wide single nucleotide polymorphism genotypes

Knowledge of population structure and breed composition of a population can be advantageous for a number of reasons; these include designing optimal (cross)breeding strategies in order to maximise non-additive genetic effects, maintaining flockbook integrity by authenticating animals being registered and as a quality control measure in the genotyping process. The objectives of the present study were to 1) describe the population structure of 24 sheep breeds, 2) quantify the breed composition of both flockbook-recorded and crossbred animals using single nucleotide polymorphism BLUP (SNP-BLUP), and 3) quantify the accuracy of breed composition prediction from low-density genotype panels containing between 2000 and 6000 SNPs. In total, 9334 autosomal SNPs on 11 144 flockbook-recorded animals and 1172 crossbred animals were used. The population structure of all breeds was characterised by principal component analysis (PCA) as well as the pairwise breed fixation index (Fst). The total number of animals, all of which were purebred, included in the calibration population for SNP-BLUP was 2579 with the number of animals per breed ranging from 9 to 500. The remaining 9559 flockbook-recorded animals, composite breeds and crossbred animals represented the test population; three breeds were excluded from breed composition prediction. The breed composition predicted using SNP-BLUP with 9334 SNPs was considered the gold standard prediction. The pairwise breed Fst ranged from 0.040 (between the Irish Blackface and Scottish Blackface) to 0.282 (between the Border Leicester and Suffolk). Principal component analysis revealed that the Suffolk from Ireland and the Suffolk from New Zealand formed distinct, non-overlapping clusters. In contrast, the Texel from Ireland and that from New Zealand formed integrated, overlapping clusters. Composite animals such as the Belclare clustered close to its founder breeds (i.e., Finn, Galway, Lleyn and Texel). When all 9334 SNPs were used to predict breed composition, an animal that had a majority breed proportion predicted to be ≥0.90 was defined as purebred for the present study. As the panel density decreased, the predicted breed proportion threshold, used to identify animals as purebred, also decreased (≥0.85 with 6000 SNPs to ≥0.60 with 2000 SNPs). In all, results from the study suggest that breed composition for purebred and crossbred animals can be determined with SNP-BLUP using ≥5000 SNPs.