Genetic Parameter Estimation and Breeding Value Ranking for Litter Size Traits in Meat Rabbits: REML and Bayesian Inference Under Limited Data Conditions

Reproductive efficiency is a key factor in meat rabbit production; however, estimating variance components for litter size can be challenging in populations with limited data. Our objective in this study was to estimate genetic parameters for litter size in meat rabbits using frequentist and Bayesian approaches. A total of 956 kindling records were analyzed, with approximately 40–48% of does contributing only a single record. The traits analyzed were the number of kits born alive, stillborn, and total born, and litter size at 7, 35, and 70 days. Variance components were estimated using restricted maximum likelihood (REML) and Gibbs sampling as implemented in the BLUPF90 suite of programs. Univariate and bivariate mixed models were fitted, including additive genetic effects, with and without the permanent environmental effect. Heritability estimates were low for all traits (<0.10), with slightly higher estimates under the Bayesian approach. Including the permanent environmental effect improved model fit but caused convergence problems with REML, whereas the Bayesian approach yielded non-zero estimates and an explicit characterization of uncertainty through posterior distributions. Although overall correlations between predicted breeding values were high, minor changes in animal rankings were detected when comparing full and reduced models under the Bayesian approach. These results highlight the importance of inference methods and model specification in genetic evaluations based on limited-information datasets.