Incorporating phylogenetic metrics to microbial co-occurrence networks based on amplicon sequences to discern community assembly processes.

Co-occurrence network analysis based on amplicon sequences is increasingly used to study microbial communities. Patterns of co-existence or mutual exclusion between pairs of taxa are often interpreted as reflecting positive or negative biological interactions. However, other assembly processes can underlie these patterns, including species failure to reach distant areas (dispersal limitation) and tolerate local environmental conditions (habitat filtering). We provide a tool to quantify the relative contribution of community assembly processes to microbial co-occurrence patterns, which we applied to explore soil bacterial communities in two dry ecosystems. First, we sequenced a bacterial phylogenetic marker in soils collected across multiple plots. Second, we inferred co-occurrence networks to identify pairs of significantly associated taxa, either co-existing more (aggregated) or less often (segregated) than expected at random. Third, we assigned assembly processes to each pair: patterns explained based on spatial or environmental distance were ascribed to dispersal limitation (2%-4%) or habitat filtering (55%-77%), and the remaining to biological interactions. Finally, we calculated the phylogenetic distance between taxon pairs to test theoretical expectations on the linkages between phylogenetic patterns and assembly processes. Aggregated pairs were more closely related than segregated pairs. Furthermore, habitat-filtered aggregated pairs were closer relatives than those assigned to positive interactions, consistent with phylogenetic niche conservatism and cooperativism among distantly related taxa. Negative interactions resulted in equivocal phylogenetic signatures, probably because different competitive processes leave opposing signals. We show that microbial co-occurrence networks mainly reflect environmental tolerances and propose that incorporating measures of phylogenetic relatedness to networks might help elucidate ecologically meaningful patterns.