QTL across and between microbial treatments
A total of 32 QTL were identified for 12 traits across all environments (Fig. 3; Table S5, Table S6).  The additive effects of each QTL explained from 2.9 - 22% of trait variation (Table S5). Of these 32 QTL, six QTL occupied unique positions in the genome. The confidence intervals of all other QTL overlapped or colocalized with at least one other QTL. Eight traits (shoot biomass, tiller number, SRL, lateral root length, root diameter, root number, root biomass and total root length) had 16 QTL with overlapping confidence intervals grouped into two hotspots on chromosome three (Fig. 3; Table S5). The hotspot located on 3@4.3 (chromosome number @ centimorgan) showed an additive effect in the direction of parental ecotype divergence, while the other hotspot located on 3@58 showed an additive effect opposite the direction of parental divergence. Pleiotropic genes or linked genes with correlated effects may drive these genomic hotspots of correlated traits. We also found significant epistatic interaction between these two hotspots. Individuals possessing the hallii allele for the QTL on 3@58, masked the effects of their interactive QTL on 3@4.3 (Table S5).
We detected 11 genotype x microbiome QTL for 10 traits that show location specific QTL (Fig. 4a-k; Table S5, Table S6). Each of these QTL were analyzed to directly test in which treatment they were present, and to estimate the direction and magnitude of their effects (Fig. 4a-k; Table S7, Table S8). In the MI treatment, QTL for shoot biomass, root biomass and root number were detected with the hallii allele contributing to a higher trait value (Fig. 4a, i, j). In the CI treatment, QTL for RMR and SRL were detected with the filipesallele contributing to a higher trait value (Fig. 4b, c). In the AI treatment, QTL for tiller number, root number, root diameter, lateral root length, shoot biomass, root biomass, root length and first order root length had allelic effects with the hallii allele contributing to a higher value for all traits except root diameter (Fig. 4d-k). Of these QTL, four have overlapping confidence intervals and are grouped into a hotspot on chromosome seven and three are grouped together on chromosome nine (Fig. 4e-k; Table S7, Table S8). Two of the three QTL present on chromosome nine were also detected in the MI treatment (Fig. 4i, j; Table S7, Table S8). Epistatic interactions between two QTL for root diameter (chr 1@68 and 2@78.8) were present only in AI and CI (Fig. 5a-c, Table S7, Table S8) and individuals possessing genotypes at the two loci from the same parent (recovering the ecotypic configuration) produced traits with the smaller magnitude (Fig. 5b, c).