Estimation of Genetic Trend in Racehorse Breeding

Publisher Website
Google Scholar
Abstract
In Norway 40–50% of trotters start to race before the age of five years. This clearly represents a preselection process which may affect the ability to estimate the true genetic trend on the basis of mixed linear model methods. This problem was conceptually examined by stochastic simulation and the animal model by assuming positive genetic and environmental correlations between a preselection criterion and transformed earnings. Inclusion of data on started horses, only, in the analysis (a selected data approach) resulted in a severe underestimation of the true genetic trend, while addition of zero unrankable earnings for horses culled prior to races (a dummy data approach) reduced the bias from preselection. The dummy data approach appeared robust to changes in preselection intensity over time, but sensitive to increasing phenotypic variance in transformed earnings due to inflation. The latter problem was taken into account by expressing the phenotypes as standard normal deviates within birth year.