Conditions under which selection for mean productivity, tolerance to environmental stress, or stability should be used to improve yield across a range of contrasting environments

Abstract

Computer simulations were conducted to determine whether conclusions obtained for the special case of a single stress and a single non-stress environment apply to the more general situation where a population of testing environments includes a range of stress and non-stress environments. Mean productivity, tolerance to environmental stress, and a regression coefficient stability parameter of genotypes across environments were compared to determine conditions under which these selection criteria should be used to improve yield across a range of contrasting environments. The results obtained from a worked example based on the single crosses from a 7 × 7 diallel cross in maize and the simulation experiment showed that the conclusions of Rosielle & Hamblin (1981) cannot be directly applied to a population of stress and non-stress environments. Selection for mean productivity should increase yield in both stress and non-stress environments unless the genetic variance in stress environments is more than double that in non-stress environments, and the genetic correlation between yields in contrasting environments is highly negative. Mean productivity and tolerance were shown to be positively correlated even if the genetic variance in stress environments is half that in non-stress environments. Genotypes with a high tolerance to stress were found to have low regression coefficient stability parameters, even when a range of stress and non-stress environments was used.