Strain-level diversity drives alternative community types in millimeter scale granular biofilms
Microbial communities are often highly diverse in their composition, both at the level of coarse-grained taxa such as genera as well as at the level of strains within species. This variability can be driven by both extrinsic factors like temperature, pH, etc., as well as by intrinsic ones, such as demographic fluctuations or ecological interactions. The relative contributions of these factors and the taxonomic level at which they influence community structure remain poorly understood, in part because of the difficulty of identifying true community replicates assembled under the same environmental parameters. Here, we address this problem using an activated granular sludge reactor in which millimeter scale biofilm granules represent true community replicates whose differences in composition are expected to be driven primarily by biotic factors. Using 142 shotgun metagenomes of single biofilm granules we found that, at the commonly used genuslevel resolution, community replicates varied much more in their composition than would be expected from neutral assembly processes. This variation, however, did not translate into any clear partitioning into discrete community types, i.e. the equivalent of enterotypes in the human gut. However, a strong partition into community types did emerge at the strain level for the most abundant organism: strains of Candidatus Accumulibacter that coexisted in the metacommunity--i.e. the reactor--excluded each other within community replicates. Single-granule communities maintained a significant lineage structure, whereby the strain phylogeny of Accumulibacter correlated with the overall species composition of the community, indicating high potential for co-diversification among species and communities. Our results suggest that due to the high functional redundancy and competition between close relatives, alternative community types are most likely observed at the level of recently differentiated genotypes but not higher orders of genetic resolution.