Abstract The principles governing microbial community assembly and the interplay between deterministic selection and stochasticity remain central debates in ecology. We investigated how chemically diverse carbon sources act as ecological filters shaping soil bacterial communities. Using replicated microcosms amended with distinct substrates (glucose, succinate, naphthalene, phenanthrene, or γ-HCH) or under starvation, we tracked community trajectories via high-frequency 16S rRNA gene sequencing (29 timepoints, triplicate) and quantitative ecological modeling. Null model analysis confirmed the carbon source as the primary deterministic filter, enforcing high reproducibility (homogeneous selection governing ~74% of assembly among replicates) and overriding stochastic effects. Crucially, abundant (1%) and rare (0.1%) taxa exhibited decoupled assembly mechanisms. While abundant taxa were driven largely by dispersal limitation (~59%) and variable selection, the rare biosphere displayed a temporal regime shift. Unlike the immediate response of dominant taxa, rare taxa transitioned from stochastic isolation to strong deterministic selection (surging to ~50%) only during later successional stages. This reframes the rare biosphere as a "latent responder" reservoir recruited by metabolic byproducts rather than the primary substrate. Additionally, time-resolved interaction networks revealed that under severe stress from toxicity or starvation, interactions shifted from competitive exclusion to facilitation (e.g., necromass scavenging). These patterns provide strong empirical support for the Stress Gradient Hypothesis in a microbial context. Collectively, our findings demonstrate how deterministic filtering and stress-mediated cooperation jointly structure ecosystems, providing a high-resolution temporal dataset to further interrogate these fundamental ecological principles.
Stari et al. (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: