Random regression analyses to model the longitudinal measurements of yolk proportions in the laying chicken

Cubic spline function was used in a genetic evaluation to model the change of yolk proportion over the lay life. A total of 19,862 yolk proportion records of 2,324 hens was used. The evaluated submodels consisted of 3 to 6 knot models. The same knots were fitted for genetic and permanent environmental splines. The residual effects were specified to be independently and normally distributed, but with heterogeneous variance for each test week. (Co)variance components were estimated by the average information restricted maximum likelihood (AIREML) method. The best fitting random regression model (RRM) was a submodel with 4 knots at 32, 36, 52, and 72 wk of age for genetic and permanent environmental effects. The estimate of genetic variance was larger than that of permanent environmental variance at the same time point. The heritability of yolk proportion ranged from 0.32 to 0.55, and the repeatability ranged between 0.45 and 0.73. The genetic correlations between test wk were from moderate to unity. To the best of our knowledge, this was the first report on the use of a RRM to evaluate yolk proportion. The results of this study showed that random regression models with the spline function could be used for improvement of yolk proportion.