Summary: | Inbreeding is a biological phenomenon of special relevance in domestic species, where the overall
inbreeding coefficient can be partitioned in founder-specific partial inbreeding (PI) coefficients.
Taking longevity data of 4,226 Landrace sows as starting point, this research proposed alternative
parameterization for PI effects under Weibull proportional hazard models, and compared their performance
through the deviance information criterion (DIC). Three different a priori distributions were
assumed for PI effects, asymmetric normal (DIC = 55,064.6), symmetric normal (DIC = 55,069.2) and flat
(DIC = 55,077.9). Additionally, the standard model accounting for the overall inbreeding coefficient
was clearly discarded (DIC = 55,078.4). For the model with asymmetric Gaussian prior, the posterior distribution
of PI effects was clearly skewed. An 85.15% of the estimates showed negative effect on sow
longevity whereas the remaining 14.85% ones had null or even positive effect on sow survival. Estimated
heritability was 0.159. |