Trait-based approach enables ecologists to find general patterns in species distribution and community assembly. A prerequisite to trait-based approach is the knowledge of functional/physiological traits of each community member. While plant ecology fulfilled this requirement and successfully introduced trait-based approach, the limited availability of microbial trait information, especially those of unculturable microbes, hampers trait-based representation of microbial communities.
Here, we propose that large-scale comparative metagenomics warrants high-throughput inference of microbial traits, especially those related to microbial habitat environments. As a proof-of-concept, we developed a bioinformatic pipeline named ProkAtlas (environmental atlas of prokaryotes) to infer habitat preferences of prokaryotic communities. ProkAtlas includes a database of 361,474 16S rRNA gene sequences from 5,368 metagenomic projects, each labeled with its source environment (e.g. soil, seawater, gut, etc.). Prokaryotic community sequences can be mapped to this database, and ProkAtlas estimates habitat preference of that prokaryotic community in a probabilistic framework, as a proxy of community-wide trait means.
Furthermore, we applied ProkAtlas to soil, human gut, and aquatic microbiomes. We found general patterns in habitat preferences of the community during primary succession and along environmental gradients. Overall, this study suggests that microbial traits inferred by comparative metagenomics works well in trait-based approach.
ProkAtlas is available at https://msk33.github.io/prokatlas.html.