Résumé
In the study of multi-host parasites, it is often found that host species contribute asymmetrically to parasite transmission. Yet in natural populations, identifying which hosts contribute to parasite transmission and maintenance is a recurring challenge. Here, we approach this issue by taking advantage of natural variation in the composition of a host community. We studied the brine shrimps Artemia franciscana and Artemia parthenogenetica and their microsporidian parasites Anostracospora rigaudi and Enterocytospora artemiae. Previous laboratory experiments had shown that each host can transmit both parasites, but could not predict their actual contributions to the parasites' maintenance in the field. To resolve this, we gathered long-term prevalence data from a metacommunity of these species. Metacommunity patches could contain either or both of the Artemia host species, so that the presence of the hosts could be linked directly to the persistence of the parasites. First, we show that the microsporidian A. rigaudi is a spillover parasite: it was unable to persist in the absence of its maintenance host A. parthenogenetica. This result was particularly striking, as A. rigaudi displayed both high prevalence (in the field) and high infectivity (when tested in the laboratory) in both hosts. Moreover, the seasonal presence of A. parthenogenetica imposed seasonality on the rate of spillover, causing cyclical pseudo-endemics in the spillover host A. franciscana. Second, while our prevalence data was sufficient to identify E. artemiae as either a spillover or a facultative multi-host parasite, we could not distinguish between the two possibilities. This study supports the importance of studying the community context of multi-host parasites, and demonstrates that in appropriate multi-host systems, sampling across a range of conditions and host communities can lead to clear conclusions about the drivers of parasite persistence. (C) 2019 Australian Society for Parasitology. Published by Elsevier Ltd. All rights reserved.