Prediction ability of an alternative multi-trait genomic evaluation for residual feed intake

Area of Science:

• Animal Genetics
• Quantitative Genetics
• Beef Cattle Breeding

Background:

• Feed efficiency is a key selection goal in beef cattle breeding programs.
• Residual feed intake (RFI) is used to reduce feed intake without impacting performance, but phenotyping is costly.
• Genomic selection and multi-trait prediction offer cost-effective strategies to improve genetic gain.

Purpose of the Study:

• To compare the prediction ability of univariate and multi-trait models for genomic estimated breeding values (GEBVs) of RFI in Hereford bulls.
• To evaluate the effectiveness of including weaning weight (WW) as a predictor trait in a multi-trait model.

Main Methods:

• The study utilized a population of genotyped Hereford bulls in Uruguay.
• Two models were compared: a univariate model for RFI and a multi-trait model including RFI traits and weaning weight.
• Genomic Best Linear Unbiased Prediction (ssGBLUP) was used, with prediction ability assessed via random validation groups and across feed intake tests.

Main Results:

• The univariate model consistently outperformed the multi-trait model across all tested validation strategies.
• Including weaning weight as a predictor trait did not enhance prediction ability for RFI GEBVs.
• Model performance was evaluated based on bias, dispersion, accuracy ratios, and relative accuracy increases.

Conclusions:

• For genotyped beef cattle, a univariate model is superior for predicting RFI GEBVs compared to a multi-trait model incorporating weaning weight.
• Adding weaning weight as a proxy trait does not improve prediction accuracy when genomic information is available for all animals.
• These findings suggest focusing on direct RFI measurements or alternative predictor traits for more effective genetic evaluations.