Fig. 5: GbM variation is associated with geoclimatic variables.

a,b, Correlation (R2) matrices of epiGWA P values (a) and environmental variables (b) for 57 genes identified in at least three epiGWA analyses. Associations between epiallelic states (UM and gbM) of genes and environmental variables were examined using a mixed linear model. Supplementary Tables 20 and 21 list individual environment labels in order. c–e, Associations between gbM and environmental data for CCS (c), CHY1 (d) and FLC (e). f, Pearson’s correlation between springtime atmospheric NO2 (billion molecules (MOL) per mm2) and flowering time (FT_16 °C) of individual accessions. g, Average (± s.e.m.) FT_16 °C and NO2 concentrations in Sweden (SWE, number of accessions (N) = 187), Russia (RUS, N = 47), Italy (ITA, N = 48), Spain (ESP, N = 170), USA (N = 41), France (FRA, N = 37), UK (N = 56) and Germany (GER, N = 102). h, Prevalence of FLC UM epiallele as a function of NO2. R2 and P values indicated in g and h are derived from Pearson’s correlation test.