Genomic predictions of mastitis-related traits early in the first lactation of dairy heifers using a single-step genomic approach.

Abstract

The objective of this study was to assess the potential benefit of a single-step GBLUP (ssGBLUP) genomic prediction approach to subclinical mastitis (SCM) and SCS traits in the early first lactation of heifers. Subclinical mastitis is highly prevalent during early lactation and poses significant challenges to both animal welfare and farm profitability. Given the low hof SCM, ssGBLUP has emerged as an effective approach for the genomic prediction of such low-htraits. This approach combines phenotypic data and genomic and pedigree information simultaneously through a hybrid relationship matrix to predict GEBV. In this study, accuracy and bias of GEBV for SCM (defined in 6 alternative ways) and SCS were assessed using the ssGBLUP approach in Canadian Holstein heifers early in their first lactation. A reference dataset, consisting of a large random sample of 544,221 heifers from 3,021 herds, containing records up to 2021, was truncated to create another dataset with records up to 2016, which were used for breeding value estimation and validation, while a smaller random sample of 137,518 heifers from 755 herds was used for genetic parameter estimation. Validation reliability and prediction bias of GEBV were estimated using ssGBLUP and were compared with the EBV derived from traditional BLUP. For constructing the hybrid relationship matrix used in the ssGBLUP, various scaling factors were tested for combining genomic and pedigree relationships. The incidence of 6 SCM trait definitions within 5 to 30 DIM ranged from 15.32% to 24.71%. Heritability was 0.047 to 0.069 for 6 SCM traits and 0.102 for SCS. Application of the ssGBLUP model substantially increased validation reliabilities of GEBV of young animals, with average gains of 0.28 (SCM traits) and 0.19 (SCS) points with optimal scaling factors. Furthermore, in comparison to EBVs obtained with a traditional BLUP method, the ssGBLUP model had slightly reduced bias in GEBV (overall with various scaling factors). Moreover, in terms of average theoretical reliabilities, gains of 0.22 and 0.27 and 0.20 and 0.28 were observed, respectively, for reference and truncated datasets of SCM traits and SCS, respectively. We concluded that ssGBLUP produced GEBV with increased reliability and less bias for young animals compared with EBV from a conventional BLUP approach. Hence, implementation of ssGBLUP in routine evaluation of SCM should be further considered within the context of the Canadian dairy industry.