Accuracy of predicting genetic merit from pedigree information for bulls entering stud sampling programs

Four sources of pedigree information were used to determine the most accurate prediction of daughter yield deviation and PTA milk, fat, and protein for AI-sampled bulls (birth years from 1984 to 1986). Pedigree sources compared were parent average, pedigree index, PTA of sire, and PTA of sire and dam. Bulls were included only if they did not have a USDA evaluation in July 1989, so that a parent average was nearly uninfluenced by daughters of the bull. When pedigree estimates from immediately prior to first evaluation were used, parent average was a better predictor than pedigree index. Inclusion of PTA of dam with PTA of sire was more accurate than PTA of sire alone for predicting each trait and resulted in R2 comparable with parent average. Predictability of daughter yield deviation from parent average for each trait improved over time as the amount of daughter information increased. Regression coefficients and R2 increased considerably for parent average and PTA of dam but not for pedigree index or PTA of sire when evaluations of young bulls began influencing estimates for their ancestors. Parent average was always a better predictor than pedigree index. Although animal model evaluations of cows have improved predictive value, regression coefficients for all pedigree sources were less than theoretically expected.