Estimation of genetic trend in a selected population with and without the use of a control population.

ABSTRACT

Data from a selection experiment conducted with sheep at Massey University, New Zealand, were analyzed to obtain an evaluation of selection response. Selection was for heavy 14-mo greasy fleece weight. Approximately seven generations of selection were represented in the data. Three estimates of genetic superiority of the selected line to the control line were obtained. All three estimates were obtained from a mixed model evaluation using the individual animal model for predicting breeding values from own and relatives' records. The estimators were 1) deviation of selected line predicted yearly phenotypes from control line predicted yearly phenotypes, 2) deviation of the predicted yearly phenotype for the selected line from the year estimate in the control line and 3) the mean yearly breeding value from the analysis of the selected line only. The realized heritability using the first approach was .20. However, the control line was found to have a slight positive drift; hence, this estimate was biased downward. Using Approach 2, accounting for drift, the realized heritability was .23. The same realized heritability, .23, was obtained from an analysis of the selected line ignoring the control (Approach 3), when a prior heritability of .30 was assumed for the mixed model evaluation. The estimate of genetic trend from predicted breeding values in the latter approach is, however, quite dependent on the assumed heritability.