Linking stochastic assembly to functional potential, redundancy, and trait patterns in bacterial communities

Stochastic processes shape the taxonomic composition of microbial assemblages. However, their impact on community functioning remains subject to debate, mainly due to functional redundancy. Little is known on the links between stochasticity and functional redundancy. Here, we assessed how stochastic assembly influences redundancy, functional potential, and trait patterns in twenty parallel lake-water bacterial communities enriched under originally identical conditions. Using gene- and genome-resolved metagenomics, we tested whether incomplete dispersal of genes required for cellobiose uptake and processing-"functional dispersal limitation"-explained variation in cellobiose use.

Several communities were composed of genomes that held the required genes but these communities did not utilize cellobiose, rejecting the notion of "functional dispersal limitation." We quantified redundancy across major functional categories such as signaling, regulation, and transport. Functional redundancy reflected the stochastic assembly from the total set of genomes. It was lower within than between communities, likely reflecting limiting similarity vs. habitat-driven functional convergence. Category-resolved patterns of functional dissimilarity were conserved across various diversity scales and even across randomly sampled sets of 28,000 bacterial genomes from the Genome Taxonomy Database. Among these categories, functions mediating environmental and microbe-to-microbe interactions and genetic information processing had highest and lowest dissimilarity, respectively. Aquatic bacteria showed the greatest differentiation across most categories.

Stochastic assembly of bacterial communities shaped the functional trait distribution. Functional redundancy inferred from the metagenomes largely reflected the trait patterns of the total set of MAGs. Functional redundancy and dissimilarity varied according to functional category. Comparison with a null model constructed from genomes of the GTDB allowed us to identify functional selection with various strengths according to the functions. While stochasticity diversified community composition, functional patterns remained conserved, reflecting shared ecological and evolutionary constraints tempered by habitat. Hence, using null models as a reference is important to interpret functional redundancy and may provide a more accurate understanding of how stochastic assembly and ecological constraints shape community-level functional organization. Video Abstract.