Sub-community-level network analysis reveals distinct microbial dynamics in a lake littoral sediment

Lake littoral sediments are dynamic environments supporting diverse ecological functions, yet the temporal dynamics of their microbial communities remain understudied. This study investigated the temporal changes in microbial communities within the surficial sediments of Lake Bourget, with a high temporal resolution (2.3 ± 2 days over five months). We hypothesized that microbial temporal dynamics are shaped by different ecological processes (deterministic vs. stochastic), depending on the organizational scales considered. To test this, we examined patterns at both the whole-community and sub-community (module) scales, combining β-diversity metrics with co-occurrence network analysis.

At the community scale, temporal changes in microbial composition were relatively constrained. Null model analysis indicated that stochastic processes, such as ecological drift and homogenizing dispersal, predominated, while deterministic processes from environmental variables played a minimal role. This was supported by variance partitioning, which showed that environmental parameters explained only 17% of the temporal variation in community structure. By reducing the community into smaller co-varying groups of taxa (modules) using network-based clustering, we identified six microbial modules, each with distinct temporal patterns. The two largest modules showed contrasting responses to environmental variables, highlighting distinct ecological niches succeeding over time. At the module scale, environmental variables explained a larger fraction of temporal variation (up to 51% for Module M3), indicating that deterministic processes, particularly environmental filtering, played a more dominant role than at the whole-community scale.

Our study demonstrates that considering multiple organizational scales within microbial communities provides deeper insights into the ecological processes driving microbial temporal dynamics. Finer-scale resolution via network modules reveals patterns and drivers that may be masked at the broader community level, allowing for a better understanding of the balance between stochastic and deterministic forces in sediment microbiomes.