Predictability of bacterial communities across environmental and species complexities

The recent explosion of microbiome studies has uncovered important correlations of bacterial community structures with temporal and environmental gradients.; however, our ability to predict which species complexes can coexist under which environmental conditions remains poor. In this project, we focus on the predictability of bacterial communities across two axes of complexity, environment and species richness. We take advantage of cultivatable bacterial species isolated from Arabidopsis thaliana. These isolates are among the most abundant species found within the leaves of A. thaliana in nature.

We inoculate random subsets of twelve bacterial species into three increasingly complex environments, from liquid media to solid media to gnotobiotic plants, and follow replicated communities over time. To quantify species’ relative abundances, we amplify the V3-V4 region of 16S rRNA gene, sequence the amplicons via the Illumina MiSeq platform, and match reads to the known sequences. We assessed predictability in terms of dissimilarity of species relative abundances across replicates to observe whether replicates converge or diverge over time.

Preliminary results indicate that replicate dissimilarity increases with increasing species richness but not with increasing environmental complexity. Surprisingly, communities from homogenous liquid environments show increased dissimilarity than communities grown on solid media.  By working up from the simplest communities in liquid culture to more complex communities in planta, we provide a framework to make predictions not just for  bacterial communities but also for complex communities in nature.