Predicting compositions of microbial communities from stoichiometric models with applications for the biogas process

Microbial communities are ubiquitous in nature and play a major role in ecology, medicine, and various industrial processes. In this study, we used stoichiometric metabolic modeling to investigate a community of three species, Desulfovibrio vulgaris, Methanococcus maripaludis, and Methanosarcina barkeri, which are involved in acetogenesis and methanogenesis in anaerobic digestion for biogas production.

We first constructed and validated stoichiometric models of the core metabolism of the three species which were then assembled to community models. The community was simulated by applying the previously described concept of balanced growth demanding that all organisms of the community grow with equal specific growth rate. For predicting community compositions, we propose a novel hierarchical optimization approach: first, similar to other studies, a maximization of the specific community growth rate is performed which, however, often leads to a wide range of optimal community compositions. In a secondary optimization, we therefore also demand that all organisms must grow with maximum biomass yield (optimal substrate usage) reducing the range of predicted optimal community compositions. Simulating two-species as well as three-species communities of the three representative organisms, we gained several important insights. First, using our new optimization approach we obtained predictions on optimal community compositions for different substrates which agree well with measured data. Second, we found that the ATP maintenance coefficient influences significantly the predicted community composition, especially for small growth rates. Third, we observed that maximum methane production rates are reached under high-specific community growth rates and if at least one of the organisms converts its substrate(s) with suboptimal biomass yield. On the other hand, the maximum methane yield is obtained at low community growth rates and, again, when one of the organisms converts its substrates suboptimally and thus wastes energy. Finally, simulations in the three-species community clarify exchangeability and essentiality of the methanogens in case of alternative substrate usage and competition scenarios.

In summary, our study presents new methods for stoichiometric modeling of microbial communities in general and provides valuable insights in interdependencies of bacterial species involved in the biogas process.