Genetic Improvement Strategies in Dairy Cattle Using Genomic Selection

Abstract

Genomic selection (GS) has fundamentally transformed dairy cattle breeding by enabling accurate prediction of breeding values for economically important traits using high-density SNP marker arrays, bypassing the lengthy progeny-testing cycle that characterised conventional pedigree-based selection. This study evaluates genomic estimated breeding value (GEBV) prediction accuracy for six key dairy traits--milk yield, fat percentage, protein percentage, somatic cell score (SCS), fertility index, and longevity--across three breeds (Holstein-Friesian, Jersey, and Brown Swiss) using a reference population of 12,847 genotyped animals from Sweden, France, and Austria. Three prediction models were compared: genomic BLUP (GBLUP), Bayesian Ridge Regression (BRR), and a Gradient Boosting Machine (GBM) integrating genomic and phenotypic covariates. GBLUP achieved the highest cross-validated prediction accuracy for milk yield (r=0.78) and protein percentage (r=0.81), while GBM outperformed both for fertility index (r=0.71) and longevity (r=0.69) where non-additive genetic effects and genotype-by-environment interactions are substantial. Genomic selection reduced the generation interval from 6.2 years (conventional) to 2.1 years, translating to an estimated 47% increase in annual genetic gain for milk protein yield. Multi-trait genomic index optimisation incorporating economic weights for all six traits demonstrated the potential to increase net merit by 18.4% above single-trait selection scenarios.