The Genetic Improvement of Feed Efficiency in Beef Cattle

Feed efficiency measured as residual feed intake (RFI) and its component traits, including average daily dry matter intake, are highly variable and moderately heritable in the Bos taurus breeds of cattle that are predominantly raised in The United States. The limitation to the genetic improvement of RFI (which is not improved by simply selecting for growth and diluting maintenance requirements) has been the inability to routinely gather intake and growth phenotypes on sufficient numbers of animals to enable meaningful selection differentials to be achieved. Genomic Selection (GS) was initially envisioned to be a solution to this problem via the development of equations to predict the genetic merit of animals for RFI based upon their BovineSNP50 genotypes. GS has been shown to work very effectively within breeds resulting in 2-4 fold increases in selection response. However, prediction equations developed in one breed perform very poorly in other breeds – evenly closely related breeds such as Angus and Red Angus. The reason for this is that the chromosomal architectures of breeds differ. The arrangement of allelic variants on chromosomes relative to the marker loci genotyped on the BovineSNP50 assay differs between breeds, so that the genetic merit of chromosomal segments tagged by marker alleles also differs. Two approaches seem warranted to address this problem. First, if we could identify the breed of origin of chromosomal segments, we could estimate the contribution of these segments (haplotypes) in multi-breed analyses and built GS prediction equations that would first identify the breed of origin of chromosomal segments in tested animals and then estimate genetic merit for RFI based on the genotype x breed of origin of haplotypes. A second approach would be to base prediction equations of those variants that direct cause variation in RFI rather than those that simply tag chromosomal segments containing these variants. The difficulty with this approach is that the majority of causal variants have very small effects and are therefore very difficult to identify. However, using newly developed genotyping assays containing functional variants and the imputation of chip-based genotypes up to whole genome sequence provides an approach to simultaneously test tens of millions of variants for their effects on RFI simultaneously. Meta-analyses performed across traits can then be used to identify loci for which the same alleles improve RFI in all (or the majority) of tested breeds to identify the variants most appropriate to use in cross-breed prediction of RFI.