Repeatable ecological dynamics govern response of experimental community to antibiotic pulse perturbation

In an era of pervasive anthropogenic ecological disturbances, there is a pressing need to understand the factors constituting community response and resilience. A detailed understanding of disturbance response needs to go beyond associations and incorporate features of disturbances, species traits, rapid evolution and dispersal. Multispecies microbial communities experiencing antibiotic perturbation represent a key system with important medical dimensions. However, previous microbiome studies on the theme have relied on high-throughput sequencing data from uncultured species without the ability to explicitly account for the role of species traits and immigration. Here we serially passaged a 34-species defined bacterial community through different levels of pulse antibiotic disturbance, manipulating the presence or absence of species immigration. To understand the ecological community response measured by amplicon sequencing, we combined initial trait data measured for each species separately and metagenome sequencing data revealing adaptive mutations during the experiment. We found that the ecological community response was highly repeatable within the experimental treatments, owing to an increasingly strong yet canalized response at increasing antibiotic levels, which could be partly attributed to key species traits (antibiotic susceptibility and growth rate). Increasing antibiotic levels were also coupled with increasing species extinction probability, making species immigration preventing this critical for community resilience. Moreover, we could detect signals of antibiotic resistance evolution occurring within species at the same time scale, leaving evolutionary changes in communities despite recovery at the species compositional level. Together these observations reveal a disturbance response which appears as classic species sorting but is nevertheless accompanied by rapid within-species evolution.