Estimation of Residual Dam-Offspring Correlation for a Maternal Animal Model through a Griddy Gibbs Sampler

MUNILLA LEGUIZAMÓN, S.; CANTET, R. J. C.

Leipzig

Congreso; 9th World Congress on Genetics Applied to Livestock Production; 2010

9th WCGALP Local Organizing Committee

It is generally accepted that the estimates of the direct-maternal genetic correlation are biased by negative environmental associations. While including a covariance between the maternal permanent environmental effects and the residuals theoretically account for this bias, the resulting covariance structure becomes non-linear in the correlation between dam-offspring. In this study we show how to fit such a model, given that the residual covariance matrix is block diagonal due to the presence of maternal families. Further, we describe the estimation of the dam-offspring correlation parameter through a Griddy Gibbs sampler algorithm. The algorithm involves calculating the exponential of the quadratic form in the residuals, and its evaluation usually renders a number so small that has no machine representation. To overcome this problem, we suggest the evaluation of the log function on an adaptative grid.