Mixed models and multivariate analysis for selection of superior maize genotypes

Abstract

Selections via the mixed model and the multivariate analysis approach can be powerful tools for selecting cultivars in plant breeding programs. Therefore, this study aimed to compare the use of mixed models, multivariate analysis and traditional phenotypic selection to identify superior maize (Zea mays L.) genotypes. Seventy-one (71) maize Topcrosses and three commercial cultivars were evaluated using these three methods. Plant height, ear height, ear placement, stalk lodging and breakage, and grain yield were evaluated. There was a difference between selection methods, as the selection with mixed models and the selection based on the average phenotypic afforded the inclusion of genotypes with high productivity, which did not occur for the multivariate analysis. The selection by multivariate analysis allowed the inclusion of genotypes with better agronomic and other desirable traits, not only those with highest productivity, in a maize breeding program.