Single-step genomic prediction models for metabolic body weight in Nordic Holstein, Red dairy cattle, and Jersey

Nordic Cattle Genetic Evaluation (NAV) introduced a breed-specific index for Saved Feed in 2020, focusing on the maintenance and metabolic efficiency of cows. Maintenance efficiency is based on genomic breeding values for metabolic body weight (MBW), for which a multi-step (genomic) evaluation was implemented in 2019. The model utilizes body weight and conformation observations from Finland and Denmark, but only conformation observations from Sweden. This study aimed to enhance the MBW evaluation by including carcass weight (CARW) from all three countries and by developing a single-step genomic prediction model. The new model includes three MBW traits and two correlated traits: CARW and stature (STA). The data were collected from Danish, Finnish, and Swedish Red Dairy Cattle (RDC), Holstein (HOL), Jersey (JER) cows born between 1990 to 2020. After data editing, the RDC, HOL, and JER datasets comprised of 2.3 million, 4.3 million, and 0.4 million records, including 0.9 million, 0.5 million, and 11 thousand MBW observations, respectively. The pedigree of RDC, HOL, and JER included 3.9, 7.2 and 0.6 million animals, respectively. Among these, 84 232 RDC, 117 845 HOL, and 39 650 JER animals were genotyped since 2009 onwards. To develop single-step genomic best linear unbiased prediction (ssGBLUP) models, we applied VanRaden method I to construct the genomic relationship matrix, with a residual polygenic proportion of 30%. We utilized the ssGTaBLUP method to solve the models. Separate ssGBLUP models were developed for each breed, and these models were validated through forward prediction cross-validation, linear regression of full data breeding values on reduced data breeding values, and comparison of pedigree-based and ssGBLUP breeding values. The inclusion of carcass weight data substantially increased phenotypic information in all three breeds, resulting in enhanced reliability of MBW breeding values. The new ssGBLUP models showed higher validation reliability and better predictive ability than the pedigree-based BLUP models. Furthermore, the new models corrected the genetic trend of MBW, addressing a previous underestimation in all breeds. Including CARW records as correlated observations and applying ssGBLUP models offers a significant improvement for the Nordic metabolic body weight evaluations, thereby enhancing the Saved Feed index.