GLM-PCA, a method to detect informative environments and phenotypic stable resistant sources of wheat to yellow rust in multi-environmental trials

Yellow rust (YR), caused by Puccinia striiformis f. sp. tritici, is one of the most important diseases of the wheat crop in all growing areas of the world. Screening of wheat genotypes for resistance in the frame of multi-environmental trials (METs) is a vitally important experiment in the wheat breeding program. The biggest challenge in the analysis of data obtained via scoring plant reaction to the disease is that they often do not satisfy the prerequisites of normality and homogeneity in the conventional parametric analysis. Even in the cases where non-parametric methods fail to distinguish genotype × environment (G × E) interaction, a generalization of PCA to exponential family likelihoods (GLM-PCA) can detect informative locations and phenotypically stable resistant genotypes. In this study, 35 elite wheat genotypes were screened for yellow rust resistance during two successive cropping seasons (2015–2016 and 2016–2017) at three hot spot locations in Iran. Based on the plots developed following GLM-PCA, two highly informative environments were identified exerting relatively huge disease pressure on the wheat genotypes. Moreover, genotypes URBYT(95–98)#4, URBYT(95–98)#7, and URBYT(95–98)#14 were visually detected as general resistant genotypes. The statistical method applied in this research to study G × E interaction could be proposed as a relevant approach to deal with categorical data obtained from a qualitative response in YR–wheat pathosystem under MET conditions.