Abstract
Bacteriophages ("phages") are hypothesized to be key drivers of bacterial population dynamics, driving microbial community composition, but empirical support for this is mixed. One reason why phages may have a less-than-expected impact on community composition is that many different phages and other mobile genetic elements (MGEs) interact with each bacterium. For instance, the same phage may have higher or lower costs to different bacterial strains or species. Assuming that resistance or susceptibility to MGE infection is not consistent across all MGEs, a simple prediction is that the net effect of MGEs on each bacterial taxon may converge with an increasing number of interactions with different MGEs. We formalized this prediction using in silico population dynamics simulations and then carried out experiments using three bacterial species, one generalist conjugative plasmid, and three species-specific phages. While the presence of only phages or only the plasmid altered community structure, these differential effects on community structure canceled out when both were together. The effects of MGEs were largely indirect and could not be explained by simple pairwise bipartite interactions (i.e., between each MGE and each bacterial species). Our results suggest that the effects of MGEs may be overestimated by studies that focus on a single MGE and not on interactions among multiple MGEs. IMPORTANCE While bacteriophages ("phages") are often cited as some of the key drivers of microbial diversity, evidence for this is greatly mixed. We demonstrate, in silico and experimentally, that the impact of phages, an example of a mobile genetic element (MGE), on community structure can diminish with increasing MGE diversity. This is because MGEs can have diverse effects on host fitness, and therefore as diversity increases, their individual effects cancel out, returning communities back to an MGE-free state. In addition, interactions in mixed-species and MGE communities could not be predicted from simple pairwise interactions, highlighting the difficulty in generalizing a MGE's effect from pairwise studies.