Implementation of single-step evaluations for fitness traits in the German and Austrian Fleckvieh and Brown Swiss populations

Abstract

The official genomic evaluations for the German-Austrian-Czech Fleckvieh and German-Austrian Brown Swiss populations were implemented in 2011 and since then only bull genotypes were used in the calibration set of the two-step system. In April 2021, genomic breeding values from a routine single-step system for almost all traits in the breeding program were published for the first time. The single-step system includes roughly 300.000 Fleckvieh and 70.000 Brown Swiss genotyped animals, of which for some traits up to 285,000 and 58,000 had direct phenotypes for Fleckvieh and Brown Swiss, respectively. During the process of implementation of single-step for fitness traits, some important points came up that needed to be dealt with. One key aspect is the validation of the estimated breeding values (EBVs). The Linear-Regression-Method published by Legarra and Reverter (2018) shows some considerable advantages, for example to have large validation groups and validation groups of female animals, which are less strictly selected. Other points are scaling and singularity prevention of the genomic relationship matrix (G). Scaling the G to fit to the numerator relationship matrix (NRM) showed positive effects on the validation metrics. The application of singularity prevention methods has shown that there are effects on bias in Mendelian sampling in specific families.