Univariate and multivariate parameter estimates for milk production traits using an animal model. II. Efficiency of selection when using simplified covariance structures 英文参考文献.docVIP

Univariate and multivariate parameter estimates for milk production traits using an animal model. 1 University of PM Visscher WG Hill 2 AFRC Edinburgh, Institute Institute of Cell, Animal and Population Biology, West Mains Road, Edinburgh EH9 3JT; of Animal Physiology and Genetics Research, Roslin EH25 9PS, Midlothian, UK . (Received 5 August 1991; accepted 3 June 1992) Summary - The effect of simplifying covariance structures for milk, fat and protein yield in lactations 1-3 on accuracy of selection for lifetime yield was investigated using selection index theory. Previously estimated (co)variances were assumed to be the true A population parameters. modified repeatability model, ie assuming a genetic correlation of unity between performances across lactations but allowing for different heritabilities and variances in different lactations, was used to investigate the efficiency of selecting sire on their progeny performance. For most combinations of heritabilities in first and later lactations, selecting sires using the repeatability model was nearly 100% efficient in terms of accuracy of selection compared to a general multivariate model. Thepredicted response to selection using a modified repeatability model was approximately 10% too high. It was found that applying a transformation to make new traits in lactation 1 uncorrelated at the phenotypic and genetic level to milk, fat and protein yield in later lactations, and assuming that 3 new uncorrelated variates were formed, was highly efficient in terms of accuracy of selection when compared to the accuracy of a general multivariate model. This transformation was recommended for a national BLUP evaluation, since it may take account of selection to a larger extent than when performing separate analyses for milk, fat and protein yield. ’ dairy cattle / milkproduction / selection index / genetic parameters / repeatability / canonical transformation * address: Department of Food and Agriculture