Genomic prediction of feed intake using predictor traits

Genomic prediction of feed intake using predictor traits A total of 77,640 weekly records on dry matter intake (DMI), 64,443 on fat and protein corrected milk (FPCM) and 73,415 on live weight (LW) were analysed from 3,188 Dutch dairy cows in 6,820 lactations (first to third lactation) from 1980 to 2015. The objective of this study was to compare the accuracies of the genomic estimated breeding values (GEBV) for DMI, with or without predictor traits included (FPCM and LW) with a single step method (SS-GBLUP). Accuracies of GEBV for DMI was0.36 when FPCM and LW were included as reference traits, 0.37 when DMI was the only reference trait, and 0.38 when all 3 traits (DMI, FPCM and LW) were included as reference traits. When only using predictor traits in the reference population, the accuracies of estimated GEBV for DMI, were lower than in the scenarios using DMI as LW and FPCM can only explain 53% of the variation in DMI. Moreover, there was very little benefit of adding information on predictor traits to the reference population when DMI was already included on the same animals. However, in the absence of DMI records, having records on FPCM and LW from different lactations is a good way to obtain GEBV with a relatively good accuracy.