Yield stability in barley using reaction norms and factor analysis

Abstract

<p>Genotype-by-environment interactions (GxE) occur when the performance of a genotype is dependent on the environment it exists in. Some genotypes are more sensitive to environmental factors, while others show greater stability. This genetic variation could be harnessed in breeding programs to increase the stability of crop performance in variable environments. An appropriate modelling strategy that ranks genotypes on their stability to environmental variation is essential. </p><p> One approach could be to model the reaction norms of genotypes using random regression, where the breeding values of genotypes are regressed on the mean value of a trial. In this model, the slope of the regression line for each genotype is a measure of stability. This is similar to Finlay-Wilkinson regression, although it considers genotypes as random effects rather than fixed, allowing relationships to be accounted for via pedigree or genomic data. This has been a popular method for studying stability in livestock and has recently been applied to plant data. </p><p> Another method proposed by Smith & Cullis (2018) derives a stability score from factor analytic (FA) models. Since FA models differ from reaction norm models in several major ways, it is possible that the estimates of stability might differ between the methods. Therefore, it could be useful to compare solutions to better understand how these models capture variation in stability. </p><p> This study compared genomic estimates of stability in yield derived from reaction norms and factor analysis using a population of 769 barley genotypes grown in 18 environments. Preliminary analysis identified substantial GxE effects, indicating significant genetic variation in stability. Estimates of stability derived from the two models were different, suggesting they detect different elements of GxE. These results may guide breeding programs looking to improve yield stability via genomic selection.</p>