Influence of sire by year interactions on the direct-maternal genetic correlation for weaning weight of Western Australian Merino sheep

The relationship between the direct-maternal genetic (co)variance σ am and sire by year (SY) interactions for weaning weight in Merino sheep was examined through simulation and real data analyses. Weaning weight was simulated using models containing interaction and σ am = 0 (S1), interaction and σ am < 0 (S2), interaction and σ am > 0 (S3), and without interaction and σ am < 0 (S4). When S1 data were analysed ignoring interaction, a negative (co)variance was observed and direct and maternal variances were inflated. Analysis of S2 data ignoring σ am resulted in deflated direct and maternal variances, inflated residual and interaction variances, and no change for the permanent environmental component. Ignoring the interaction effect in S3 data resulted again in a negative (co)variance component and highly biased genetic parameters. On application to weaning weight of Merino sheep, the model ignoring SY resulted in a direct-maternal genetic correlation of –0.43. The model using both (co)variance and interaction effects fit the data better ( P < 0.001). The interaction variance represented 9.2% of the phenotypic variance but explained 86% of the (co)variance between direct and maternal genetic effects estimated ignoring SY. A small (–0.096) but still negative estimate of the genetic correlation was obtained. The implication of these findings in the context of Central Test Sire Evaluation and Maternal Sire Central Progeny Test Schemes is discussed. Keywords: (co)variance estimation, simulation, sire by year interaction. Australian Journal of Agricultural Research 54(7) 723 - 729 Full text doi:10.1071/AR02235 © CSIRO 2003