Multiple infections are known to affect virulence evolution. Some studies even show that coinfections may decrease the overall virulence (the disease-induced mortality of a coinfected host). Yet, epidemiological studies tend to overlook the overall virulence, and within-host models tend to ignore epidemiological processes. Here, I develop an epidemiological model where overall virulence is an explicit function of the virulence of the coinfecting strains. I show that in most cases, a unique

more »

... n is evolutionarily stable (in accordance with the model I use here). However, when the overall virulence is lower than the virulence of each of the coinfecting strains (i.e., when coinfections decrease virulence), the evolutionary equilibrium may be invaded by highly virulent strains, leading to the coexistence of two strains on an evolutionary timescale. This model has theoretical and experimental implications: it underlines the importance of overall virulence and of epidemiological feedbacks on virulence evolution.