Variance components and animal rankings for milk yield and fat yield in a multibreed dairy cattle population using genomic-polygenic, genomic and polygenic models

Abstract

The variance ratios were estimated and animal rankings of genomic-polygenic (GP), genomic (G) and polygenic (P) models were compared for milk yield (MY) and fat yield (FY) in a Thai multibreed dairy cattle population. The dataset contained monthly records of MY and FY from 600 first-lactation cows from 56 farms in Central Thailand. The mixed model contained herd-year-season, Holstein fraction, heterozygosity of the cow and age at first calving as fixed effects (all models). Random effects were single nucleotide polymorphisms (SNP; GP and G models), animal polygenic (GP and P models) and the residual. The GP heritability estimates were higher for MY (0.38) and FY (0.41) than for the P model (0.28 for MY and 0.30 for FY). The fractions of the additive genetic variance explained by the SNP markers were 50% for MY and 48% for FY. Rank correlations between GP and G were the highest for both MY and FY (0.99; p < 0.01). Rank correlations between G and P were the lowest for MY (0.89; p < 0.01) and FY (0.73; p < 0.01). SNPs from the GeneSeek Genomic Profiler Low-Density (GGP-LD) 9k BeadChip not only accounted for a sizeable fraction of the additive genomic variance for MY and FY, but also yielded animal genomic estimated breeding values (EBV) whose rankings were highly correlated with the rankings of both genomic-polygenic and polygenic EBV. These results indicated that GGP-LD 9k and likely higher density genotyping chips would help improve the accuracy of prediction and selection in Central Thailand.