Abstract
Abstract Spatial network structure of biological systems drives ecology and evolution by distributing organisms and their genes. The ubiquitous host-parasite systems are no exception. However, past work has largely ignored relevant spatial complexity, hampering the translation of theoretical predictions to real ecosystems. Thus, we develop an eco-evolutionary metapopulation model of host-parasite dynamics where hosts and parasites disperse through realistically complex spatial networks representing major biomes: riverine aquatic and terrestrial. We generate the testable prediction that parasite virulence, or how parasites harm their hosts, is unimodal with dispersal and can reach greater values in aquatic landscapes but saturates to lower values in terrestrial systems. Moreover, we show that kin selection drives virulence evolution. Spatial networks generate characteristic patterns of parasite relatedness which drive differential virulence evolution. Finally, we show that accounting for virulence evolution allows us to predict the distribution of key epidemiological variables (e.g., parasite extinction risks) within spatial networks.