Treatment drives bacterial community composition
We used 16S rRNA gene amplicon sequencing classified into ASVs to characterize both the native microbial communities and communities generated by experimental inoculations as a representation for the overall biotic differences. For parental ecotypes growing under natural habitats, PCoA of bacterial ASV counts revealed strong location/ecotype and compartment effects (i.e. soil, rhizosphere, and root) across axes one and two, respectively (Fig. 1a). Permanova mirrored these results with location/ecotype explaining the most variance (R2= 0.21, P < 0.001) and compartment explaining the second most (R2 = 0.15, P < 0.001; Table 1). Microbiota varied significantly in alpha diversity between compartments, but not between location (Fig. 1b). Phylum level distributions were overall consistent between microbiota of plants growing at the two natural locations with Proteobacteria ,Actinobacteria , and Acidobacteria being dominant members (Fig. 1c, Fig. S1a), which is congruent with results from previous root-associated microbiome studies (Edwards et al. 2014; Lundberg et al. 2012; Singer et al. 2019; Wagner et al. 2016). Only three relatively low abundance phyla displayed significant differences between location-ecotype: WPS-2 and Entotheonellaeota in the rhizosphere and Rokubacteria in the root (Fig. S2). Conversely, microbiota from the two locations were more divergent at the ASV level and we identified 735 unique ASVs which were differentially abundant by compartment (440 in soil, 251 in rhizosphere, and 401 in roots; Fig. 1c).
We next analyzed microbiota acquired under experimental conditions in the greenhouse by sampling roots and rhizosphere from the RIL parents, along with soil from unplanted pots. PCoA revealed that inoculum and compartment significantly impacted microbiota composition (Fig. 1d). Alpha diversity was also impacted by compartment and inoculum: in general, plants inoculated with native soil slurries hosted microbiota with greater Shannon diversity compared to plants with heat-killed, mock microbiota (Fig. 1e). As expected, when comparing the effect of inoculum source within heat killed or native conditions, we found that microbial communities of plants and soil with heat-killed inocula were significantly more similar than if the inoculum was unsterilized and this effect was consistent independent of compartment (Fig. S3). Similar trends were observed at the phylum level where there were many more differentially abundant phyla by soil source with intact inoculum vs. heat treated (Fig. S1b, c). When identifying ASVs whose abundance was impacted by soil inoculation source, many more ASVs were differentially abundant in comparisons between native inoculum compared to heat killed (Fig. 1f). When analyzed together, we found that greenhouse and field microbiomes formed distinct communities, yet were still identifiable by soil source (Fig. S4a, more in Appendix S1). These results indicate that heat sterilization of inoculum dampens the effect of soil source on compositions of the resulting microbiome and that plants inoculated with native microbiota host significantly different communities in the rhizosphere and roots.