Abstract

Microbes form diverse communities whose activities play key roles in numerous application areas, including biotechnology, agriculture, and medicine. The structure and function of these communities are shaped by interactions between the species that comprise them. However, measuring the interaction network of such communities remains challenging. To address this knowledge gap, we used a novel high-throughput microfluidics device to measure thousands of microbial interactions across multiple environments. Our data revealed that while interactions typically change qualitatively across environments, they can be predicted based on the growth abilities of individual species. Furthermore, we showed that interactions between multiple species combine in a highly non-additive manner, and are dominated by the strongest pairwise interaction. These findings provide the first step toward a predictive understanding of microbial interactions and their collective properties.