AN ALGORITHM TO ANALYSE, ESTIMATE AND PREDICT ON BALANCED DIALLEL EXPERIMENTS

Abstract

Diallel crosses designs are used on investigations in plant and animal breeding. The aim is to generate experimental information allowing to evaluate different genetical aspects in relationship with a set of parents. Researchers have selected some particular cases of the general linear model, with fixed or random effects. In classical theory, the methodologies for analyzing this kind of experiments are described under models I and II, this last model called a variance components model. A third kind of inference, the prediction of linear combinations of fixed and random components (or only among random components), has important advantages in comparison with the estimators obtained with the fixed effects model. Most of the available algorithms, analyze these experiments considering either the first two models, or the third possibility in different cases, under complete randomized blocks design. However, these analyses are usually complicated and the completely randomized design is frequently used. In this paper an integrated computational algorithm is presented in SASIML, in order to analyze balanced experiments of diallel crosses, with or withount the inclusion of reciprocal crosses, and laid out either in completely randomized design or in randomized blocks design. The algorithm considers the estimation or parameters prediction of in cases where the genetical and environmental components in the model, are considered either as fixed or as random effects.