EFFECTS OF DATA STRUCTURE ON VARIANCE PREDICTION ERROR AND ACCURACY OF GENETIC EVALUATION

Several features of data structure were studied to determine their eff ects on variance of prediction error and accuracy of evaluation. Assig ning 50 sires with progeny to a portion of 10, 25, or 50 contemporary groups according to a sire model with and without additive genetic rel ationships, or assigning 50 individuals with their own record to one o f 2, 5, or 10 contemporary groups according to an animal model, establ ished the designs. Additive genetic relationships were based on simula ted pedigree files. Low, medium, and high heritabilities (.10, .25, an d .40, respectively) were considered. The inverse of coefficient matri ces gave variances of prediction error. Populations derived from the s ire model (n = 8,100) consisted solely of progeny-tested individuals. For them, number of progeny had a quadratic (P < .001) association wit h variance of prediction error (R(2) = 56 to 82%), which selection ind ex theory underestimated when there were < 100 progeny. Number of dire ct connections (sires of contemporaries of progeny) together with prog eny numbers explained variance of prediction error (R(2) = 76 to 90%) better than either variable alone. With no direct connections, varianc e of prediction error was maximum unless a relative with at least one direct connection itself existed. Populations derived from the animal model (n = 900) consisted of animals with designs representing a proge ny test, performance test, or a combination of both (34, 41, and 25% o f the total, respectively). For performance-tested animals (without pr ogeny), number of genetic connections was not highly correlated with v ariance of prediction error (r = -.10, across h(2)), but relatives pre vented zero accuracies when contemporary groups consisted of one anima l. Even when animals had no relatives, more than five members per cont emporary group gave little additional increase in accuracy. For other than a progeny test, designs were complex, being described by many var iables that were confounded.