Comparing alternative definitions of the contemporary group effect in Avileña Negra Ibérica cattle in a Bayesian analysis

Data on weaning weight from 12,740 animals were used to compare different contemporary group (CG) definitions to be applied in the current genetic evaluation in the Avileña Negra Ibérica beef cattle breed. Several classical ‘ad-hoc’ statistics and criteria for statistical model comparison within a Bayesian analysis were used. Six alternative definitions for the CG effect were compared: Herdyearseason of calving (HYS), with seasons defined according to the four natural seasons; herd-year-month of calving (HYM); herd clusters with length of 30 days (HC30-30) or 90 days (HC90-90), and, adaptive clusters with two time limits, 30 and 90 days (HC30-90), and 30 and 180 days (HC30-180). Classical statistics pointed at HYM and HC30-30 as the models showing smaller within CG variation and smaller residual variance, with slightly worse performance in terms of accuracy of prediction for breeding values for direct genetic effects. Bayes factors and cross-validation predictive densities, allowed for a better discrimination among models. Models including CG spanning 30 days were more plausible and showed better predicting ability than models spanning 90 days. Adaptive CG showed intermediate results. Model definition had a relatively large impact on the variance components estimates. Overall, HYM showed best results but implied the largest loss of data (14%). HC30-90 might represent a compromising solution for this population. Abbreviation Key: BF = Bayes Factor; CG = Contemporary groups; HC30-30: Herd cluster spanning 30 d; HC30-90: Herd cluster spanning 30 d minimum and 90 d maximum; HC30-180: Herd cluster spanning 30 d minimum and 90 d maximum; HC90-90: Herd cluster spanning 90 d; HYM = HerdYear-Month; HYS: Herd-Year-Season; MD = Marginal density of the data