Sire Selection Using Cow Traits to Improve Quadratic Merit in Progeny

Abstract

Sire selection methods were compared by stochastic simulation for improvement in a three-trait, quadratic merit function. Selection indexes included milk, udder depth, and rear leg set phenotypes recorded on cows and bull progeny. The progeny index predicted progeny total merit from the trait indexes of both sire and mated cow. The sire index predicted progeny total merit from only the trait indexes of the sire. Compared with the sire index, which selects the best sire, the progeny index usually selected a number of different sires. The mean number of sires was 1.2 or 2.0 per 100 cows mated, depending on whether the trait indexes were calculated using normal or unit heritabilities, respectively. Total merit gains were not different using the progeny or sire index methods. Although the gain for improvement in nonlinear merit using the progeny index was expected to exceed that using the sire index, it has errors caused by low accuracy cow trait indexes. The sire index achieved 5% greater milk merit gain than did the progeny index. The progeny index was compared with a two-stage method that first used the sire index for milk merit followed by the progeny index for type merit. The two-stage method produced lower gains: 8% less in total merit and 25% less in milk merit, compared with the progeny index. Methods are needed of constituting portfolios of service sires for mating a cow herd to improve total merit while controlling the risk associated with using one or few sires in a breeding season.