Disentangling competitive and cooperative components of the interactions between microbial species

Interactions between microbial species have been characterized by the net influences, positive or negative, that each species in a pair exerts on the other. This conventional view of interactions being either positive or negative proves restrictive in predicting the behaviour of microbial communities and, more importantly, influencing them towards desired community structures. Here, we propose a more fundamental characterization of the interactions. The net interactions typically comprise positive and negative underlying components. Yet, the conventional view prevails because the components have been difficult to disentangle. We have developed a methodology to disentangle them when metabolic interactions predominate. We conceived a theoretical resource partitioning between species that helps estimate the positive components. The negative components then follow from knowledge of the net interactions. The interactions between two species are then characterized by the quartet of these components. We applied the methodology in silico to 28 species pairs from the human oral microbiome, yielding 56 net interactions and their 112 quartet components. We found that on average the net interactions comprised positive and negative components of comparable strengths. Interestingly, weak net interactions often arose from the cancellation of strong underlying components. Furthermore, we found species pairs with similar net interactions but vastly different underlying components. Extant community ecology theories, based on net interactions, cannot distinguish between such pairs. The quartet explained several confounding experimental observations and offered new insights into microbial community ecology. We envision its implications in the construction of more refined ecological theories and the engineering of synthetic microbial communities.