Analysis of factors affecting length of competitive life of jumping horses

Abstract

Official competition data were used to study the length of competitive life in jumping horses. The trait considered was the number of years of participation in jumping. Data included 42 393 male and gelded horses born after 1968. The competitive data were recorded from 1972 to 1991. Horses still alive in 1991 had a censored record (43% of records). The survival analysis was based on Cox's proportional hazard model. The independent variables were year, age at record, level of performance in competition (these three first variables were time dependent), age at first competition, breed and a random sire effect. The prior density of the sire effect was a log gamma distribution. The maximization of the marginal likelihood of the gamma parameter of the gamma density gave an estimate of the additive genetic variance. The baseline hazard, the fixed effects and the sire effects were then estimated simultaneously by maximizing their marginal posterior likelihood. Jumping horses were culled for either involuntary or voluntary reasons. The involuntary reasons included the management of the horse, for example, the earlier a horse starts competing the longer he lives. The voluntary reasons related to the jumping ability: the better a horse, the longer he lives (at a given time, an average horse is 1.6 times more likely to be culled than a good horse with a performance of one standard deviation above the mean). The heritability of functional stayability was 0.18. The difference in half-lives of the progeny of two extreme stallions exceeded 2 years.