Use of an animal model in situations of limited subclass numbers and high degrees of relationships

Breeding value estimation procedures for two traits with moderate and high heritability were evaluated by using a single-trait animal model and computer-simulated data designs. Of interest were the effects of differing numbers of animals and degrees of relationships among animals within and across contemporary groups (tests). Test effects were assumed fixed and animal effects were assumed random. Family size, number of families per contemporary group, and degree of genetic relationships within and across contemporary groups were varied to determine interrelationships among the factors. Results were compared on the basis of accuracy by using both the correlation of true and estimated breeding values and the prediction error variance obtained from the inverse of the coefficient matrix of the mixed-model equations. Small contemporary groups in conjunction with evaluation of closely related families caused average accuracy to decrease relative to that obtained with the same number of unrelated animals because genetically related animals were less accurately evaluated relative to one another. Connecting contemporary groups with a genetic relationship matrix formed a large set of interdependent equations and improved the average accuracy of predicted breeding values. The slight decrease in accuracy for genetically related animals was more than offset by the increase in accuracy of evaluation for their unrelated test mates because the proportion of fixed effects to random effects was smaller. Care must be exercised in designing evaluation schemes involving small populations, and the decision of which fixed effects to include in the model is critical.