Abstract
A key property of many antibiotics is that they will kill or inhibit a diverse range of microbial species. This broad-spectrum of activity has its evolutionary roots in ecological competition, whereby bacteria and other microbes use antibiotics to suppress other strains and species. However, many bacteria also use narrow-spectrum toxins, such as bacteriocins, that principally target conspecifics. Why has such a diversity in spectrum evolved? Here, we develop an evolutionary model to understand antimicrobial spectrum. Our first model recapitulates the intuition that broad-spectrum is best, because it enables a microbe to kill a wider diversity of competitors. However, this model neglects an important property of antimicrobials: They are commonly bound, sequestered, or degraded by the cells they target. Incorporating this toxin loss reveals a major advantage to narrow-spectrum toxins: They target the strongest ecological competitor and avoid being used up on less important species. Why then would broad-spectrum toxins ever evolve? Our model predicts that broad-spectrum toxins will be favored by natural selection if a strain is highly abundant and can overpower both its key competitor and other species. We test this prediction by compiling and analyzing a database of the regulation and spectrum of toxins used in inter-bacterial competition. This analysis reveals a strong association between broad-spectrum toxins and density-dependent regulation, indicating that they are indeed used when strains are abundant. Our work provides a rationale for why bacteria commonly evolve narrow-spectrum toxins such as bacteriocins and suggests that the evolution of antibiotics proper is a signature of ecological dominance.