BEST LINEAR UNBIASED PREDICTOR FOR SPECIFIC COMBINING ABILITY AND RECIPROCAL EFFECT OF GRIFFIN’S DESIGNS ONE AND THREE

Abstract

The linear model associated to experiments for diallel crosses incorporates fixed an random effects; the estimation of general combining ability and maternal effects have been recently done under the correct mixed effects model. However, until now the Griffing´s designs one and three have been used to make estimations of specific combining ability and reciprocal effects, under the fixed effects model, even when they actually are of random nature. For this reason in the present work, the empirical best linear unbiased predictors for specific combining ability and reciprocal effects are derived, in Griffing ´s designs one and three, under the correct mixed effects model. Furthermore, a computational algoritm in SAS-IML commands is also given to estimate such predictors.