Implementation of a Single-Step genomic evaluation system for dairy cattle in Wallonia, Belgium

Abstract

For several years, dairy cattle breeders in the Walloon Region of Belgium have had access to locally estimated breeding values (EBV) for traits of interest. These evaluations enable Wallonia to contribute to the Multiple Across Countries Evaluation (MACE) conducted by Interbull. In the current local genomic evaluation framework, genomic and pedigree data are integrated with local EBV and external information, MACE-derived EBV (MACE EBV), through a pseudo-single-step genomic evaluation system, producing genomically enhanced EBV (GEBV). However, this approach may introduce biases. To address this, the present study aimed to implement a single-step genomic BLUP (ssGBLUP) that simultaneously incorporates all available national data alongside MACE information and validate this method using milk production traits. The proposed strategy first involves defining "pseudo-traits" that represent MACE traits (i.e., 305-day averages for milk, fat, and protein yields over the first three lactations). MACE EBV are then transformed into adjusted pseudo-phenotypes (i.e., deregressed proofs) and effective contributions but avoiding double-counting of Walloon data within MACE EBV. Next, the variance-covariance matrices from the local random regression test-day model were modified to include the three MACE pseudo-traits as correlated traits. Finally, a single-step genomic evaluation was performed, jointly analyzing test-day records and MACE pseudo-phenotypes. Validation of both pedigree-based and single-step genomic evaluations, both integrating MACE information, was carried out using data from the official Walloon genetic evaluations of April 2022. Results show that MACE information is adequately integrated in the local evaluations, because Pearson correlations between MACE EBV and the integrated EBV were higher than 0.97 across traits. The addition of genomic information in single-step evaluations resulted in small changes for all individuals, as illustrated by Pearson correlations ranging from 0.975 and 0.986 for sires with MACE information.