No abstract

No abstract

1. The aim of the present study was to compare different models to estimate variance components for egg weight (EW) in laying hens. 2. The data set included 67 542 EW records of 18 245 Mazandaran hens at 24, 28, 30, 32 and 84 weeks of age, during 19 consecutive generations. Variance components were estimated using multi-trait, repeatability, fixed regression and random regression models (MTM, RM, FRM and RRM, respectively) by Average Information-Restricted Maximum Likelihood algorithm (AI-REML). The models were compared based on Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC). 3. The MTM was the best model followed by the Legendre RRMs. A RRM with 2nd degree of fit for fixed regression and 3(rd) and 2(nd) degrees of fit for random regressions of direct additive genetic and permanent environmental effects, respectively, was the best RRM. The FRM and RM were not proper models to fit the data. However, nesting curves within contemporary groups improved the fit of FRM. 4. Heritability estimates for EW by MTM (0.06-0.41) were close to the estimates obtained by the best RRM (0.09-0.45). In both MTM and RRM, positive genetic correlations were estimated for EW records at different ages, with higher correlations for adjacent records. 5. The results suggest that MTM is the best model for EW data, at least when the records are taken at relatively few age points. Though selection based on EW at higher ages might be more precise, 30 or 32 weeks of age could be considered as the most appropriate time points for selection on EW to maximise genetic improvement per time unit.

No abstract