EV ALUATION OF GENOTYPES IN SERIES OF EXPERIMENTS: DIFFERENCES AND GENETIC PARAMETERS GENERATED FROM TWO MODELS

Abstract

Due to the importance of the unavoidable genotypic evaluation stage when the development of superior cultivars is the goal, plant breeders should be aware of the effects that, in terms of genetic parameters, has in the quality of the analysis the correspondence between reality and model representing it. In this study a preliminary analysis on the effects that show up in the estimation of variance components, hypothesis testing, heritability, etc. due to the use of a wrong model is made. Results indicate that when a model in which years (Y) is nested by locations (L) is used when Y and L are actually crossed factors, biased estimators for the variance components associated to genotypes (G), the GL interaction and heritability are produced. Furthermore, when testing the hypotesis related to GL and G the F values were higher and lower, respectively, relative to what would be under the appropriate model.