Models are often employed to integrate knowledge about epidemics across scales and simulate disease dynamics. While these approaches have played a central role in studying the mechanics underlying epidemics, we lack ways to reliably predict how the relationship between virulence (the harm to hosts caused by an infection) and transmission will evolve in certain virus-host contexts. In this study, we invoke evolutionary invasion analysis -- a method used to identify the evolution of uninvadable strategies in dynamical systems -- to examine how the virulence-transmission dichotomy can evolve in models of virus infections defined by different natural histories. We reveal that peculiar ecologies drive different evolved relationships between virulence and transmission. Specifically, we discover patterns of virulence evolution between epidemics of various kinds (SARS-CoV-2 and hepatitis C virus) and that varying definitions of virulence alter our predictions for how viruses will evolve. We discuss the findings in light of contemporary conversations in the public health sector around the possibility of predicting virus evolution and in more extensive theoretical discussions involving virulence evolution in emerging infectious diseases.
