The impact of deregressed foreign breeding values on national beef cattle single-step genomic evaluation

In recent years, genetic evaluations in cattle breeding have shifted from purely national evaluations to multinational evaluations considering relatives from other countries. Integrating international estimated breeding values (EBVs) into national genomic evaluations presents challenges due to differences in evaluation methodologies and data sources. This study focused on the impact of blending internationally derived EBVs with national EBVs in the single-step genomic evaluation of German beef cattle using three approaches to deregressing EBVs. The national phenotypic data for four breeds (Angus, Charolais, Limousin, and Simmental) were obtained from the routine German beef cattle evaluation of December 2022, and the international EBVs were obtained from the routine Interbeef evaluation of January 2023. Scalar (Garrick (GA), Van Raden (VR)) and matrix deregression approaches were compared for reversibility of EBVs. A forward validation study was used to evaluate the accuracy, dispersion and level bias obtained in a purely national single-step evaluation, and single-step genomic evaluations blended with DRPs obtained from the three deregression approaches.

A validation study based on forward prediction showed improved accuracy, and reduced dispersion bias in the EBVs blended with international EBVs compared to purely national EBVs, particularly for the direct and maternal effects of 200-day weight. As expected, Pearson correlation analysis revealed that the matrix deregression (> 0.99) approach outperformed the scalar deregression approaches (0.75-0.99), exhibiting a greater correlation between the EBVs obtained from DRPs and the EBVs obtained from phenotypes across the various breeds and traits in our study. A forward validation study with and without integrating foreign data across the three deregression methods showed improvement in reducing dispersion bias, as indicated by the regression coefficient. The GEBVs from an evaluation incorporating foreign information with national data showed a higher correlation to the GEBVs from a truncated evaluation than those from an evaluation without foreign information.

These findings underscore the importance of accurately integrating foreign EBVs to enhance national genomic evaluations and genetic progress in livestock populations. Our results show that the matrix approach to deregressing EBVs performs optimally across traits and breeds. However, the VR deregression approach can serve as an alternative in situations where the matrix deregression approach might be too technical to implement.