Microbial metacommunity ecology: Using null models and networks to understand assembly in a salamander disease system

Animals host diverse microbiomes crucial for their health, which subsequently influence their function in ecosystems. Metacommunity theory offers a useful framework to understand the processes driving host‐associated microbiome assembly across scales, linking regional environmental processes to local host–microbial processes. We formulated hypotheses based on community ecology theory and tested them using 16S ribosomal RNA (rRNA) amplicon sequencing and analytical approaches to analyze the microbiomes of three Appalachian salamander species and in their corresponding environments: red‐backed salamanders ( Plethodon cinereus ) in forest soils, two‐lined salamanders ( Eurycea bislineata ) in streams, and eastern newts ( Notophthalmus viridescens ) in ponds. We hypothesized that salamander skin microbiomes would exhibit niche‐based deterministic processes (e.g., environmental selection) given prior observations that evolutionary history and infection by the fungal pathogen Batrachochytrium dendrobatidis (Bd) influenced microbiome composition. Deterministic processes varied across salamanders, being lowest in N. viridescens , intermediate in P. cinereus , and highest in E. bislineata . Environmental microbiomes in forests and ponds were strongly structured by deterministic processes, whereas stream microbiomes showed moderate deterministic structuring. Given that rare environmental microbes frequently occur on amphibian skin, we hypothesized that patch dynamics, such as competition–colonization trade‐offs, would contribute to metacommunity processes. Network analyses (both unipartite and bipartite) revealed bacterial co‐occurrence patterns that provided correlational support for competition and colonization trade‐offs within salamander skin microbiome networks. Finally, we tested the hypothesis that fungal exposures would cause an increase of Bd‐inhibitory bacteria and found evidence for terrestrial environments linked to higher amounts of putatively Bd‐inhibitory bacteria driven by higher fungal gene copy counts on salamander skin and in their environments. Our findings suggest that niche‐based processes influence microbiome structure at the environmental‐host interface, while patterns consistent with patch dynamics contribute to variation within and among hosts. We provide a framework to measure stochasticity and identify factors that predict microbiome structure based on environment and host biology.