Using the pairwise epistasis model analysis separately for all metabolites, we identified all significant main effect markers, individual genetic loci, and interactions, epistatic terms. For each metabolite, we identified each significant term and estimated the genetic variance controlled by this term by taking the ratio of the Type III sums of squares (SS) to the total sums of squares (total model SS + Residual SS) for that term from Model III. In this figure, we present the distribution of the per marker and per interaction heritability for each marker/interaction for the metabolites that are significant for that marker/interaction. These plots are titled with specific nuclear locus show the distribution of heritability controlled by interactions between the titled marker and all the other loci in the model. Again this is only for interaction terms that were significant in the individual metabolite models.