Variance component estimations for the South African Merino sheep breed for application in genetic and genomic evaluations

Abstract

The estimation of genetic parameters allows for the accurate prediction of estimated breeding values (EBVs), which plays a crucial role in developing selection indices that can be used to generate genetic progress for economically important traits. This study aimed to estimate genetic parameters for the estimation of EBVs and genomically enhanced breeding values (GEBVs) for the South African Merino sheep breed for application in genetic and genomic evaluations. A population of 864 754 South African Merino sheep were included in this study. Variance components were estimated for all sheep born after 2009, consisting of 62 460 animals. A multi-trait animal model was used to estimate the variance components, using VCE6 statistical software (Groeneveld et al., 2010). The multi-trait animal model estimated the heritabilities and standard errors for seven traits: Direct weaning weight (0.25 ± 0.01), maternal weaning weight (0.20 ± 0.00), body weight at wool test (0.26 ± 0.01), clean fleece weight (0.33 ± 0.01), fibre diameter (0.63 ± 0.00), staple length (0.33 ± 0.00) and coefficient of variation of fibre diameter (0.43 ± 0.00). Conventional EBVs were obtained using PEST2 (Groeneveld et al., 2019), employing the same genetic models as used for variance component estimation. The genetic trends were determined by averaging the breeding values of measured animals per year of birth per trait and were plotted against the national trends provided by SA Stud Book. SNP data was incorporated for the estimation of GEBVs with Single-Step GBLUP methodology, using MIX99 (Lidauer et al., 2013), for a highly heritable (fibre diameter) and a lowly heritable (wean maternal) trait. The genetic trends for the GEBVs were determined by averaging the GEBVs per year of birth for measured animals per trait. The EBVs and GEBVs for all active, as well as for the genotyped animals were correlated to assess the effect of the inclusion of genomic information in EBV estimation on the ranking of the animals, as well as on the accuracy of estimation of the breeding values. The results indicated that including genomic information did not change the ranking of animals or the accuracy in which EBVs were estimated for fibre diameter. In contrast, for the lowly heritable trait, wean maternal, the inclusion of genomic information improved the accuracies of the EBVs, increasing from a range of 18% and 97% to a range of 37% and 97%.