Bayesian analysis of heritability of canine hip dysplasia in German Shepherd Dogs

Abstract

SummaryVariance components and heritability were linearly estimated for canine hip dysplasia (CHD) in German Shepherd Dogs using a Bayesian analysis based on the marginal posterior distributions of parameters achieved via Gibbs sampling. Data was provided by the German Shepherd Breeding Association in Augsburg. The material included 21 371 X‐rayed dogs born in the years 1998–2000 in 3755 kennels from 9704 litters. The pedigree file consisted of 38 604 animals and contained ancestors for up to six generations. The full model used considered the fixed effects of sex, age at examination for CHD, the examining veterinary expert, the combined effect of litter size and percentage of examined dogs per litter, and the sire and mother lines as well as the additive genetic random effects of the animal and of the mother, the permanent environmental effect of the litter and the kennel. Submodels were compared with the full model to evaluate the importance of different fixed and random effects, and their influence on other model components. Heritability estimates in the full model wereh2 = 0.26 ± 0.03 (additive genetic animal effect) and = 0.10 ± 0.02 (additive genetic maternal effect). The direct animal and the maternal effects were negatively correlated genetically (ram = −0.65 ± 0.05). The proportion of the variance of the random effect of the common kennel of the total variance was always below 0.015, whereas the corresponding proportion of variance of the random effect of the common litter effect reached a value of 0.045. When submodels were employed, the heritability estimates for the additive genetic animal effect ranged fromh2 = 0.24 toh2 = 0.26. The heritability estimates for the additive genetic maternal effect varied between = 0.09 and = 0.10.