Genomic evaluation for juvenile mortality including causal variants in French Holstein and Montbéliarde cattle breeds.

Abstract

Juvenile mortality is a major concern for the dairy cattle industry because of its wide-ranging economic, environmental, and ethical effects. Over the last few years, an increased number of genetic defects associated with juvenile mortality has been identified, creating new opportunities and challenges for their management in selection. The implementation of a routine genomic evaluation may therefore contribute to reducing juvenile mortality. However, no existing method currently includes causal variants and incorporates their additive and dominance effects in a single-step GBLUP (ssGBLUP) model. This study was conducted on 2 French dairy breeds: Holstein (HOL) and Montbéliarde (MON), both of which are affected by known recessive genetic defects (bovine lymphocyte intestinal retention defect and cholesterol deficiency in HOL, and mitochondropathy in MON). In this study, we first estimated genetic parameters for juvenile mortality. Second, we developed ssGBLUP models that included causal variants, either by assigning higher proportions of genetic variance explained a priori by the causal variants or by estimating and incorporating causal variants' additive and dominance effects as covariates. Third, we evaluated the models' performance to select the most accurate one for genomic evaluation. To this end, 2 independent training and validation datasets were constructed. The training dataset included juvenile mortality records for 1,275,746 HOL and 476,361 MON females, of which 84,328 HOL and 29,944 MON were genotyped. Models were validated using mortality records of 689,502 HOL and 149,801 MON daughters of 262 and 254 genotyped bulls, respectively. No difference in model performance was seen at the population level. However, at the individual level, heterozygous bulls for the causal variants were more accurately distinguished from wild type homozygous when using EBVs obtained from models that included causal variants compared with a conventional ssGBLUP. We therefore advocate the routine implementation of a genomic evaluation that accounts for causal variants to help reduce juvenile mortality. A genotyping effort for dead animals is also suggested because it would enhance model performance. Additionally, further research is needed to determine the most effective method for incorporating causal variants into ssGBLUP models and to assess whether their inclusion improves the model's prediction accuracy.