Optimal Designs for Sire Evaluation Schemes

Abstract

Maximizing genetic progress in a breeding program involves using the best method of evaluation and the available information to make optimum breeding decisions. These are based on the current information (y_o) and may include choice and number of individuals culled, choice and number of new animals to include in the evaluation, allocation of progeny testing resources among animals evaluated, non-random mating, or any combination of the above. Let u_i be the merit of the candidates available for breeding and y_i be the additional records obtained, following the i^th decision based on y_o, for i=l,...s. Suppose all the information is used to select a constant number of candidates. This paper considers how breeding decisions based on y_o and selections based on y_o and y_i can be made such that expected genetic progress is maximized. It is shown that choosing the alternative corresponding to the largest conditional mean of merit of candidates selected at the final stage (given y_o), and selecting those with the largest conditional means of merit given y_o and y_i maximizes expected genetic progress.