Insights on factor analysis applied to multi-trait and multi-environment selection of common bean lines

Abstract

One of the objectives of common bean (Phaseolus vulgaris L.) breeding programs is selection of lines with superior agronomic traits that are adapted to different environmental conditions. Thus, the aim of this study was to verify the applicability of factor analysis as a strategy to assist the selection of common bean lines for multi-traits in multi-environments. Thirty-five carioca common bean lines were assessed in three municipalities of the state of Minas Gerais (Brazil) in the dry, winter, and wet seasons between 2017 and 2019. The lines were assessed for angular leaf spot (ALS) severity, plant architecture (PA), and grain yield (GYD). Lines were selected using two scenarios: (1) selection for one trait in a multi-environment and (2) multi-trait selection in a multi-environment. Scenario 1 compared the following strategies: the overall mean, the confidence index proposed by Annicchiarico, and the genotype-ideotype index associated with factor analysis. The proposal in Scenario 2 was selection based only on factor scores or based on the genotype–ideotype index associated with factor analysis. The strategies of selecting lines for a trait assessed in multi-environments based on the overall mean or on the genotype-ideotype index associated with factor analysis proved to be appropriate for traits with a simple (ALS and PA) or complex (GYD) line by environment interaction. Furthermore, the use of the genotype–ideotype index associated with factor analysis showed balance in selection of lines under the condition of the same trait in multi-environments and under that of multi-traits in multi-environments.