Best Linear Prediction: An Operational Example

Abstract

Best linear prediction seems well suited to the problem of predicting breeding values and ranking parents in forest tree breeding. In theory, BLP is relatively simple, however, the application of BLP theory to real life situations will generate a number of problems in need of specific solutions. Some of the problems may be somewhat theoretical (for example, obtaining good estimates of the necessary second moments), while others may be more practical (writing a computer program to build the necessary matrices). This chapter will outline our application of BLP for predicting parental breeding values from half-sib progeny test data in a large slash pine breeding program. We discuss a number of specific problems that were encountered, and our attempts to solve these problems. Much of this chapter is adapted from White and Hodge (1988).