Genomic prediction of residual feed intake in the Nordic breeds using data from research herds and 3D cameras in commercial herds

Nordic Cattle Genetic Evaluation NAV (Denmark, Finland, Sweden) has published an index for Saved feed, which comprises maintenance and metabolic efficiency indices. The genetic evaluation for maintenance has been introduced in 2019. Metabolic efficiency, also known as residual feed intake, is more difficult to evaluate genetically because of lack of records for individual feed intake. Breed specific genetic evaluations were developed for residual feed intake in Holstein, Red Dairy Cattle and Jersey. The data for Holstein was repeated feed intake data from the Efficient Dairy Genome Project with data from 2,543 Holstein research farm cows and 420 cows (1st to 6th parity) from a commercial herd with the 3D camera system CFIT. The data for Red Dairy Cattle was repeated feed intake data from 725 1st parity research farm cows and 196 cows (1st to 6th parity) from a commercial herd with the 3D camera system CFIT. The data for Jersey was repeated feed intake data from 1,077 cows (1st to 6th parity) from five commercial herds with the 3D camera system CFIT. The genetic evaluation was constructed as a two-stage model. In the first stage a least squares model was fit to calculate residual feed intake phenotypes. The residual feed intake phenotype was calculated with partial regressions on energy sinks (energy corrected milk, metabolic body weight and change in body weight). The residual feed intake phenotypes were corrected for heterogeneous variance with respect to herd, year and season. In the second stage the repeated residual feed intake phenotypes were modelled by a single-trait (Red Dairy Cattle and Jersey) or multiple-trait (Holstein) single-step genomic prediction model. Common for all breed-specific evaluations a heritability of 15% was modelled. For the Holstein evaluation a genetic correlation of 0.6 was assumed between countries. The developed index for residual feed intake was uncorrelated or has low correlation to the Nordic Total Merit index and Nordic Total Merit sub-traits for all breeds. Use of genomic information increased the reliability of breeding values. However, results from cross validation indicate that reliability of genomic breeding values for metabolic feed efficiency in candidate animals is still low.