An application of vGWAS to differences in flowering time in maize across mega-environments

Genomic regions containing loci with effect sizes that interact with environmental factors are desirable targets for selection because of increasingly unpredictable growing seasons. Although selecting upon such gene-by-environment (G × E) loci is vital, identifying significantly associated loci is challenging due to the multiple testing correction. Consequently, G × E loci of small- to moderate effect sizes may never be identified via traditional genome-wide association studies (GWAS). Variance GWAS (vGWAS) have been previously shown to identify G × E loci. Combined with its inherent reduction in the severity of multiple testing, we hypothesized that vGWAS could be successfully used to identify genomic regions likely to contain G × E effects. We used publicly available genotypic and phenotypic data in maize (Zea mays L.) to test the ability of two vGWAS approaches to identify G × E loci controlling two flowering traits. We observed high inflation of (Figure presented.) from both approaches. This suggests that these two vGWAS approaches are not suitable to the task of identifying G × E loci. We advocate that similar future applications of vGWAS use more sophisticated models that can adequately control the inflation of (Figure presented.). Otherwise, the application of vGWAS to search for G × E effects that are critical for combating the effects of climate change will not reach its full potential.