Genomic prediction accuracy in Korean Holstein population: A comprehensive study of milk production traits

Abstract

This study evaluated methods to improve genomic prediction in Korean Holstein cows by comparing pedigree-based BLUP (PBLUP), genome-based BLUP (GBLUP), and single-step genomic BLUP (ssGBLUP). Phenotypic data were obtained from 14,377 Korean Holstein cows, provided by the National Agricultural Cooperative Federation and Korean Livestock Improvement Association. These records included first and second lactation data for 305-day total milk yield (Milk305), milk fat yield (Fat305), and milk protein yield (Prot305). Genotypic data for 2016 cows were obtained using the Illumina Bovine 50 K SNP BeadChip. Heritability estimates for Milk305, Fat305, and Prot305 were consistently the highest when using PBLUP for both lactations, whereas GBLUP yielded lower estimates and ssGBLUP provided intermediate values. The average accuracies for Milk305 were 0.424, 0.411, and 0.528 for PBLUP, GBLUP, and ssGBLUP, respectively, during the first lactation and 0.385, 0.375, and 0.486 for the second lactation. For Fat305, the accuracies were 0.403, 0.457, and 0.529 for the first lactation and 0.381, 0.304, and 0.487 for the second lactation. For Prot305, the accuracies were 0.409, 0.374, and 0.523 for the first lactation period, and 0.388, 0.371, and 0.496 for the second lactation period. The results show that ssGBLUP consistently outperforms traditional PBLUP and GBLUP in prediction accuracy. This study highlights the potential of advanced breeding techniques and strategic genomic integration to improve the efficiency and sustainability of dairy production in South Korea and beyond.