Files
Abstract
Protozoan zoonoses, such Chagas disease and leishmaniasis, remain endemic in large parts of the world, exacerbating social inequity and contributing heavily to the global burden of infectious disease. Novel protozoa species which have emerged from wildlife to humans in the recent decades (e.g., Plasmodium knowlesi, a causal agent of malaria) have proven difficult to control. Our ability to anticipate and prevent future emerging disease threats relies on identifying the characteristics of zoonotic pathogens and targeting surveillance efforts accordingly. While several studies have profiled the traits of zoonotic viruses, protozoa have received limited attention. We compiled a dataset of protozoa species which incorporates both parasite and host traits, including information on community structure and importance within a host-parasite bipartite network. Using a machine learning algorithm, extreme gradient boosting, we distinguished zoonotic from non-zoonotic protozoa with 85% accuracy. Our model found that traits of generalist protozoa (e.g., broad tissue tropism, high network centrality, multiple transmission modes) were most useful for predicting zoonotic status, compared to intrinsic biological traits (e.g., morphology), environmental traits (e.g., temperature), or host-related traits (e.g., life history). Here we report parasitic protozoa species of wild mammals which are most likely to be undiscovered sources of current or future zoonoses, identifying them as priority targets for surveillance.