Fig. 1

Flowchart of experimental analyses. To investigate how putatively selected regions influence variation for G × E (a), 30 temperate and 30 tropical inbreds were used to calculate F_ST in 20 base pair sliding windows across the genome. Windows with mean F_ST > 0.5 were categorized as “high” F_ST, and windows with mean F_ST < 0.15 were categorized as “low” F_ST. SNPs from the Maize G × E project hybrids that were within high or low F_ST windows were categorized as high or low F_ST SNPs, and used to estimate G × E variances attributable to high and low F_ST regions of the genome. To investigate location of variants associated with G × E (b), hybrid phenotypes were regressed on the means of common hybrids at each location. The slope and mean squared errors from each hybrid’s regression were used as response variables in GWAS, and the 50 most significant SNPs from each GWAS were evaluated for their position relative to the nearest gene