Genomic–polygenic evaluation of multibreed Angus–Brahman cattle for postweaning feed efficiency and growth using actual and imputed Illumina50k SNP genotypes

Abstract

The objectives were to estimate the fractions of additive genetic variances for four postweaning feed efficiency and growth traits explained by 46,909 actual and imputed SNP genotypes, to compare rankings of calf additive genetic predictions from genomic–polygenic (GP), genomic (G), and polygenic (P) models, and to determine trends of predicted additive genetic values from calves ranging from 100% Angus to 100% Brahman in a multibreed population. Traits were postweaning residual feed intake (RFI), daily feed intake (DFI), feed conversion ratio (FCR), and weight gain (PWG). Phenotypes were from 807 bull, heifer, and steer calves housed at the Feed Efficiency Facility of the University of Florida from 2006 to 2010. Imputation from 2899 SNP (Illumina3k) to 46,909 SNP (Illumina50k) was done with program findhap2 using a reference population of 828 Brangus heifers. Fixed effects for all models were contemporary group (year–pen), age of dam, sex of calf, age of calf, Brahman fraction of calf, and heterozygosity of calf. Random effects were additive SNP (GP and G models), additive polygenic (GP and P models), and residual. Software GS3 was used to compute variance components and heritabilities, and additive genetic predictions. Heritabilities were 0.30 for RFI, 0.37 for DFI, 0.25 for FCR, and 0.33 for PWG. Fractions of additive variances explained by the 46,909 actual and imputed SNP were 0.48 for RFI, 0.36 for DFI, 0.50 for FCR, and 0.28 for PWG. These fractions were 3.2, 3.2, 2.0, and 1.8 times larger than those previously obtained for these four traits using the 2899 SNP from the Illumina3k chip. Rank correlations between calf additive genetic predictions were high between GP and P and between GP and G models (0.89–0.98; P<0.0001), and somewhat lower between G and P models (0.69–0.81; P<0.0001). Regressions of additive genetic predictions on Brahman fraction of calf were negative with the G model for DFI (P<0.0344) and with all models for PWG (P<0.0171 to P<0.0001). Imputation from the Illumina3k to 50k substantially increased the explained fraction of additive SNP and total genetic variances resulting in considerably higher rank correlations between calf additive genetic predictions from G and GP, and from G and P models in this Angus–Brahman multibreed population.