A mixed model approach for selecting Merino ewes
Delport, Gideon Jacobus
MetadataShow full item record
1. The possibility of using Mixed Model Methodology in the selection of Merino sheep for ewe productivity was investigated. Ewe productivity, defined as a function of both the mass of lamb weaned and the mass of wool produced by the ewe, presented a situation ideally suited to the use of canonical variates in a multiple-trait model. 2. The methodology and computer algorithms for the application of the mathematical model were validated by, using a data set, (1907 ewes with two successive records) obtained from the Klerefontein selection experiment of the Department of Agriculture and Water Supply. The experiment consisted of three flocks, one selected for high clean fleece mass, another for overall visual excellence and an unselected control flock. 3. Computer programmes developed for the purpose of this study proved to be useful for other situations of determining breeding values, separating genetic and environmental effects and inbreeding calculation. 4. The fact that no (co)variances between mass of lamb weaned and wool production could be obtained from literature led to the determination of the necessary parameters from the available data (al though not ideally suited to this purpose). 5. The effect of using different (co)variance structures was investigated. Results indicate that adjustment to these structures does not lead to selection of vastly different ewes. The lowest correlation between predicted breeding values of any of four different structures was 0.811. The best one of these structures was selected on the basis of the distribution pattern in the base population as well as the accuracy of predicting breeding values. 6. It is possible to make a reasonably accurate preliminary selection of ewes on their dams' breeding value predictions. The dam-daughter, correlations (between breeding values) were as high as 0.66 in the case of the chosen (co)variance structure. The possibility of selecting sires on the basis of their daughters' predicted breeding value in ewe productivity does, however, not seem to be promising. 7. The evidence presented points to no significant adverse effects on lS-month performance when selecting for ewe productivity. Thcorrelation with BLUP on ewe productivity with BLUP breeding values for body mass (r=+0.301), and clean fleece mass (r=+0.136) seems to fit in with current selection practices. The correlation of ewe productivity values with breeding value predictions for fibre diameter (r=+0.l120) may present a problem in the sense that negative selection for fibre diameter is currently a high priority in the industry. The fact that the correlation is very low indicates that slight selection pressure on fibre diameter will take care of this relationship. 8. Genetic trends in ewe productivity (EPI) and the components thereof are presented. Being correlated responses to both artificial and natural selection for body mass, genetic changes were slight. The highest regression coefficient for ewe productivity (EPI) was 0.256 (P(0<05). 9. The general conclusion is that the concept of ewe productivity poses a practical possibility for increasing returns from the ewe flock. Applying the technique of multiple-trait analysis on canonical variates, may be a viable proposition in the solution of many other problems.