Evaluation of association mapping results by ANOVA.

GLM and MLM analyses were used to evaluate the 21 candidate Al tolerance genes, using the net root growth trait collected from Al treated plants. These models incorporated the population structure (Q) of the Diversity Panel, the relative kinship (K) of the Diversity Panel and net root growth of the Diversity Panel grown without Al stress as a fixed effect covariate (C). The GLM model used only factor Q, while the MLM models used factors Q+K and Q+K+C. Six candidate genes gave significant results and are shown, with the number of significant sites (p<0.01) identified per locus for each model. The maximum value for variance explained by a marker within a gene in any model is reported. False Discovery Rates were empirically calculated for each model based on 1,095 random SNPs throughout the genome and are expressed as percentages. Narrow (h2) and broad sense (H2) heritability estimates were generated for each trait based upon variance estimates from the MLM.