Direct and diffuse cross-kingdom interactions in plant microbiome assembly

Abstract Studies of plant-associated microbial communities consistently indicate a role for classic assembly mechanisms, such as environmental and host filters, but often leave substantial unexplained variation. Biotic interactions within microbial communities may help to fill this gap, specifically cross-kingdom interactions between fungi and bacteria, as these are increasingly found to be important to both assembly and function. We hypothesized that direct interactions between bacteria and fungi are an important driver of composition in low-diversity leaf habitats, where pairwise interactions are more likely. In high-diversity root habitats, we expected diffuse, indirect interactions to be more relevant to composition. To test these hypotheses, we characterized bacterial and fungal communities of switchgrass (Panicum virgatum L.) leaves and roots at 14 sites spanning mountain to coastal ecoregions of North Carolina, USA. We analyzed putative direct and diffuse interactions using ecological network inference and partitioned variance explained in microbial community composition by spatial, environmental, and biotic interactions. We found that cross-kingdom biotic interactions contributed to microbial community structure. The largest improvements to variance explained (5-11%) were from direct interactions, except for root fungal communities where diffuse interactions (7.5%) explained more than double that of direct interactions (2.8%). These contributions were comparable to those from environmental and spatial factors. The joint effects of putative biotic interactions and environmental conditions also contributed to the explained variation, highlighting the importance of environmental tracking in microbes. These findings suggest that using network inference for identifying cross-kingdom ecological interactions can improve our fundamental understanding of how plant-associated microbiomes assemble, which is also directly relevant to applied efforts such as the effective development of synthetic communities. Competing Interest Statement The authors have declared no competing interest. Footnotes Open Research Statement: Raw sequence data are available on the NCBI Short Read Archive under BioProject PRJNA648664. Data and R code are available on Github (https://github.com/HawkesLab/Hammer_etal_DOE_Networks) and will be transferred to Dryad and Zenodo prior to publication for permanent archiving. Section on environmental gradient analyses updated to clarify community-level patterns; Figure 3 revised; Supplemental files updated.