Search

The aim of this study was the multivariate prediction of breeding values for canine hip dysplasia (CHD), elbow dysplasia (ED) and humeral osteochondrosis (OCD) for Bernese mountain dogs of the Schweizer Sennenhund-Verein für Deutschland e. V. (SSV). For the analysis the pedigrees of eight generations and radiographic screening results of the birth cohorts from 1995-2008 were used. The number of dogs with scores for CHD was 5513, for ED 5175 and OCD 1240. Breeding values were multivariately predicted using a mixed linear model for CHD, ED and OCD as well as for the occurrence of a fragmented coronoid process of the medial ulna (FCP) and the ED-score without FCP. The pedigree breeding value (eRZWp) which is used as the selection criterion reached a reliability to predict the phenotype of the offspring at 2.8-2.9% for CHD, 2.9% for ED, 1.1% for ED without FCP, 1.8% for FCP and 0.8-1.3% for OCD. The reason for the low predictive value of the eRZW(P) is caused by the very high influence of the own performance of the animal and the very uniform distribution of contributions of the breeding values of the relatives.These results indicate that even a multivariate prediction of breeding values does not lead to a faster progress in breeding against CHD and ED, however, does allow breeding against OCD in the Bernese mountain dog. In comparison to phenotypic selection, there is some improvement in the selection response when using breeding values. Due to the general low predictive power of breeding values better approaches for selection of future breeding animals are urgently warranted to achieve improvements in breeding Bernese mountain dogs.

In the present study genetic parameters for canine hip dysplasia (CHD), canine elbow dysplasia (CED) and osteochondrosis dissecans of the humeral head (OCD) were analysed in Bernese mountain dogs. The data set included the official scores for CHD and CED from 5534 Bernese mountain dogs born in the years 1995-2008. A multivariate linear animal model was employed to estimate heritabilities, additive genetic and residual correlations using residual maximum likelihood (REML). Heritability estimates were h2 = 0.26 for CHD, h2 = 0.22 for CED and h2 = 0.40 for OCD. The additive genetic correlation between CHD and CED was 0.31, between CHD and OCD 0.25, and between CED und OCD -0.49. A further multivariate analysis of the prevalence of the FCP (fragmented coronoid processus), CHD and OCD revealed a heritability of h2 = 0.59 for FCP. The additive genetic correlations among FCP and CHD as well CED except FCP were positive, but negative with OCD. Multivariately estimated breeding values will lead to higher genetic progress because the correlation structure of the traits can be taken into account and possible genetic antagonisms among traits are better reflected in the breeding values.

Phenotypic and genetic trends were examined in 157,455 German shepherd dogs of the German population born between 1985 and 2003 which were Xrayed for canine hip dysplasia (CHD). A subset of 40,270 dogs born between 1998 and 2003 was examined for which information was available on age at examination and X-raying veterinary practitioner. Relative breeding values were predicted using best linear unbiased prediction methods (BLUP) and Gibbs Sampling. There was an almost linear and significant decline in both phenotypic and genetic trends for CHD scores (scale 1 to 5 for CHD free to severe CHD) over a period of 19 years, with b = -0.021 +/- 0.002, and b = -1.69 +/- 0.05, respectively. After introduction in 1999 of a selection scheme based on relative breeding values (RBVs, mean 100, standard deviation 20) of a linear animal model, phenotypic and genetic trends declined less than before 1999, with b = -0.019 +/- 0.011 and b = -1.43 +/- 0.29. Analysis of the effects of RBVs of parental combinations among sires and dams on the percentage of non-affected progeny showed that lower RBVs were associated with a higher percentage of non-affected progeny than were higher RBVs. In addition, it was also shown that selection procedures based on RBVs are more efficient than selection schemes based on phenotypic records of parents. When the maternal RBV was above 100, the proportion of non-affected progeny was between 10% and 20% lower than when the maternal RBV was below or equal to 100. There was a significant relationship between higher inbreeding coefficients and higher RBVs of the animal.