Single-step prediction of genomic breeding value in a small dairy cattle population with strong import of foreign genes

Abstract

The breeding value (EBV) of Holstein cattle milk performance from the first lactation was evaluated using a regular Animal Model or by Single-Step Prediction of the genomic breeding value (GEBV). A total of 838 bulls were genotyped using the Illumina BovineSNP50 Beadchip V2. Two overlapping sets of milk performances were evaluated: calving years 1991–2004, with 729 341 lactations and 1 394 487 animals in the pedigree and calving years 1996–2009, with 808 436 lactations and 1 487 608 animals in the pedigree. The older data set included 526 genotyped bulls, in which the daughters’ milk performance was known for 210 individuals. All of the genotyped animals were included in the newer data set. Of the young genotyped bulls from the older set, 279 had more than 50 daughters with performance records in the newer set. Genomic relationship matrices (G) were constructed from the allele frequencies of the current genotyped population or by assuming a constant value of 0.5 for all loci. Using current allele frequencies, the correlation of G with the pedigree relationship (A) was 0.74, while it was 0.77 when the constant value was used. G was blended with A with weights of 80 or 99%. The average EBV of the genotyped bulls exceeded the mean EBV of the entire population by 3 SD. Although the number of reference bulls was small, genotyping resulted in an increase of approximately 0.05 in the correlation of the GEBV of young bulls with their results after progeny testing. Only small differences in correlations were found in dependency on the methods used for the determination of G and in dependency on the weight used in blending G with A. Both EBV and GEBV in the older set showed higher correlations with the GEBV of the newer set than the EBV of the newer set.