Genomic prediction of residual feed intake in US Holstein dairy cattle

摘要

Genomic selection is an important tool to introducefeed efficiency into dairy cattle breeding. The goals ofthe current research are to estimate genomic breedingvalues of residual feed intake (RFI) and to assessthe prediction reliability for RFI in the US Holsteinpopulation. The RFI data were collected from 4,823lactations of 3,947 Holstein cows in 9 research herdsin the United States, and were pre-adjusted to removephenotypic correlations with milk energy, metabolicbody weight, body weight change, and for several environmentaleffects. In the current analyses, genomic predictedtransmitting abilities of milk energy and of bodyweight composite were included into the RFI model tofurther remove the genetic correlations that remainedbetween RFI and these energy sinks. In the first partof the analyses, a national genomic evaluation for RFIwas conducted for all the Holsteins in the nationaldatabase using a standard multi-step genomic evaluationmethod and 60,671 SNP list. In the second partof the study, a single-step genomic prediction methodwas applied to estimate genomic breeding values of RFIfor all cows with phenotypes, 5,252 elite young bulls,4,029 young heifers, as well as their ancestors in thepedigree, using a high-density genotype chip. Theoreticalprediction reliabilities were calculated for all thestudied animals in the single-step genomic predictionby direct inversion of the mixed model equations. Inthe results, breeding values were estimated for 1.6 milliongenotyped Holsteins and 60 million ungenotypedHolsteins, The genomic predicted transmitting abilitycorrelations between RFI and other traits in the index(e.g., fertility) are generally low, indicating minor correlatedresponses on other index traits when selecting forRFI. Genomic prediction reliabilities for RFI averaged34% for all phenotyped animals and 13% for all 1.6million genotyped animals. Including genomic informationincreased the prediction reliabilities for RFI comparedwith using only pedigree information. All bullshad low reliabilities, and averaged to only 16% for thetop 100 net merit progeny-tested bulls. Analyses usingsingle-step genomic prediction and high-density genotypesgave similar results to those obtained from thenational evaluation. The average theoretical reliabilityfor RFI was 18% among the elite young bulls under 5yr old, being lower in the younger generations of elitebulls compared with older bulls. To conclude, the sizeof the reference population and its relationship to thepredicted population remain as the limiting factors inthe genomic prediction for RFI. Continued collectionof feed intake data is necessary so that reliabilities canbe maintained due to close relationships of phenotypedanimals with breeding stock. Considering the currentlylow prediction reliability and high cost of data collection,focusing RFI data collection on relatives of elitebulls that will have the greatest genetic contribution tothe next generation will give more gains and profit.