A field test of the dilution effect hypothesis in four avian multi-host pathogens

The Dilution Effect Hypothesis (DEH) argues that greater biodiversity lowers the risk of disease and reduces the rates of pathogen transmission since more diverse communities harbour fewer competent hosts for any given pathogen, thereby reducing host exposure to the pathogen. DEH is expected to operate most intensely in vector-borne pathogens and when species-rich communities are not associated with increased host density. Overall, dilution will occur if greater species diversity leads to a lower contact rate between infected vectors and susceptible hosts, and between infected hosts and susceptible vectors. Field-based tests simultaneously analysing the prevalence of several multi-host pathogens in relation to host and vector diversity are required to validate DEH. We tested the relationship between the prevalence in house sparrows (Passer domesticus) of four vector-borne pathogens-three avian haemosporidians (including the avian malaria parasite Plasmodium and the malaria-like parasites Haemoproteus and Leucocytozoon) and West Nile virus (WNV)-and vertebrate diversity. Birds were sampled at 45 localities in SW Spain for which extensive data on vector (mosquitoes) and vertebrate communities exist. Vertebrate censuses were conducted to quantify avian and mammal density, species richness and evenness. Contrary to the predictions of DEH, WNV seroprevalence and haemosporidian prevalence were not negatively associated with either vertebrate species richness or evenness. Indeed, the opposite pattern was found, with positive relationships between avian species richness and WNV seroprevalence, and Leucocytozoon prevalence being detected. When vector (mosquito) richness and evenness were incorporated into the models, all the previous associations between WNV prevalence and the vertebrate community variables remained unchanged. No significant association was found for Plasmodium prevalence and vertebrate community variables in any of the models tested. Despite the studied system having several characteristics that should favour the dilution effect (i.e., vector-borne pathogens, an area where vector and host densities are unrelated, and where host richness is not associated with an increase in host density), none of the relationships between host species diversity and species richness, and pathogen prevalence supported DEH and, in fact, amplification was found for three of the four pathogens tested. Consequently, the range of pathogens and communities studied needs to be broadened if we are to understand the ecological factors that favour dilution and how often these conditions occur in nature.