Facilitative interaction networks in experimental microbial community dynamics

Facilitative interactions between microbial species are ubiquitous in various types of ecosystems on the Earth. Therefore, inferring how entangled webs of interspecific interactions shift through time in microbial ecosystems is an essential step for understanding ecological processes driving microbiome dynamics. By compiling shotgun metagenomic sequencing data of an experimental microbial community, we examined how the architectural features of facilitative interaction networks could change through time. A metabolic modeling approach for estimating dependence between microbial genomes (species) allowed us to infer the network structure of potential facilitative interactions at 13 time points through the 110-day monitoring of experimental microbiomes. We then found that positive feedback loops, which were theoretically predicted to promote cascade breakdown of ecological communities, existed within the inferred networks of metabolic interactions prior to the drastic community-compositional shift observed in the microbiome time-series. We further applied directed-graph analyses to pinpoint potential keystone species located at the upper stream positions of such feedback loops. These analyses on facilitative interactions will help us understand key mechanisms causing catastrophic shifts in microbial community structure.

Automated summary

By analyzing the shotgun metagenomic sequencing data of an experimental microbial community, we discovered that the network structure of facilitative interactions in microbial ecosystems can change over time. By using metabolic modeling, we inferred the potential facilitative interaction networks at 13 time points during the 110-day monitoring of the microbiomes. We found that positive feedback loops, which can lead to ecological cascade breakdown, existed in the metabolic interaction networks before the drastic community compositional shift.