A simulation of bacterial communities

Abstract

This study constructs and tests an agent-based model of bacterial communities with the goal of modeling the observation that the majority of bacteria in nature cannot be cultured. The new field of metagenomics, the direct, mass sequencing of DNA recovered from the environment, is the source of this observation. The hypothesis tested is that bacteria form interdependent communities so that viable levels of energy production are rare in bacteria when they are grown in monoculture. A new game, the metabolism game is introduced. Agents produce energy by playing this game with one another. Studies are run with different number of bacterial species in the simulation. The energy level for viability is set by running simulations with a single bacterial species and then the hypothesis is tested in simulations with multiple bacterial species. Multiple bacterial species are evolved in a novel type of multi-population evolutionary algorithm called a multiple worlds algorithm. The fraction of culturable bacterial agents recovered from the simulation is larger than that found in nature but still quite low, supporting the hypothesis that bacteria may not be culturable because they require the presence of partner species.