Quantifying the intra- and inter-species community interaction in a microbiome by dynamic covariance mapping

Abstract A microbiome’s composition, stability, and response to perturbations is dictated by the community interaction matrix 1-10 that is commonly assayed by pair-wise species competition. In their natural environment however, microbes concurrently experience multiple species, face conditions that may be difficult to mimic in vitro, and have members that are impractical to isolate. Additionally, due to overlapping of evolutionary and ecological timescales, the community interaction matrix is also influenced by intra-species diversity, but how and to what extent remains poorly understood 11-14. Here, we develop a general approach called Dynamic Covariance Mapping (DCM) to estimate the interaction matrix of multispecies microbiome community in its natural environment from abundance time-series data. Together with intra-species high-resolution lineage tracking via chromosomal barcoding, we quantify the inter- and intra-species community interaction matrix during E. coli colonization of mice gut microbiome with increasing complexity: germ-free, antibiotic-perturbed, and innate microbiota. With DCM, we differentiate three temporal phases of invasion in the susceptible communities: 1) initial loss of community stability as E. coli enters; 2) recolonization of some gut bacteria; and 3) recovery of stability with E. coli clones coexisting with resident bacteria in a quasi-steady state. These phases are influenced by specific interactions between E. coli sub-lineages with other species in the community. These results highlight the transient nature and time-dependence of community interaction networks in microbiomes driven by the persistent coupling of ecological and evolutionary dynamics. Our theoretical and experimental approach can be applied to characterize coupled ecological-evolutionary dynamics of bacterial communities in vitro and in situ.