Abstract
Within-herd multiple trait evaluations based on information on relatives were obtained from stimulated beef cattle data. Maternal ability, weaning weight, postweaning gain and yearling weight were the traits evaluated. The population consisted of eight herds (50 cows each) through eight calf crops. Selection was practiced for growth and resulted in a measurable genetic trend for all growth traits. A model for individual animals that described completely the variation in the records as simulated was examined for its merit in predicting breeding values in this population with genetic trend. After eight generations of selection the average actual breeding values for weaning weight, postweaning gain and yearling weight were 14, 14 and 28 kg, respectively. The average predicted breeding values for the same three traits were 17, 14 and 31 kg, respectively. The use of yearly deviations in the selection index resulted in under evaluation of individuals for yearling weight in the last generation by 26.2 kilograms. This bias was almost equal to the average (across herds) response to selection. Computational requirements inherent in the use of the complete model led us to investigate approximate models available for consideration under the parameter structure used. Reductions in the complexity of the equations from these models and the loss in accuracy resulting from the approximations in this example are discussed. Copyright © 1981. American Society of Animal Science. Copyright 1981 by American Society of Animal Science.