Habitat, hosts, and fungus in the field: Regulators of Metschnikowia epidemics in natural zooplankton communities

Background/Question/Methods

Explaining variation in epidemic size remains a major challenge in disease ecology.  Variation in habitat can directly regulate disease, especially when pathogens have key environmental stages.  However, habitat can also select resident community composition, including abundance of predators and host species diversity.  In turn, these community changes can regulate disease.  For instance, the ‘Healthy Herds Hypothesis’ predicts that epidemics are smaller when predators consume infected hosts.  Similarly, the ‘Dilution Effect Hypothesis’ predicts that epidemics are smaller when species diversity (specifically, a higher frequency of ‘diluter’ hosts) interferes with disease transmission.  However, it is unclear if these hypotheses can operate together, and if they do, which is more important in regulating disease.  In our study system, a virulent ascomycetous yeast (Metschnikowia bicuspidata) causes yearly epidemics in the dominant zooplankton grazer (Daphnia dentifera) of many Midwestern lakes.  We sampled 28 lakes throughout the 2014 epidemic season.  We quantified deep-water refuge size (a critical habitat component), the intensity of fish predation, zooplankton community composition (including frequency of two potential diluter hosts), and epidemic size.  Then, we used Structural Equation Modeling to infer causality and quantify the relative contributions of habitat, predation, and the Dilution Effect to variation in epidemic size.  

Results/Conclusions

The dilution effect strongly reduced the size of epidemics in lakes with a high frequency of diluters (specifically, Ceriodaphnia sp.).  Lakes with more intense fish predation also had smaller epidemics; however, this effect was indirect and mediated entirely via the zooplankton community.  Thus, we found no direct support of the ‘Healthy Herds Hypothesis’.  Simultaneously, lakes with less intense fish predation had a higher frequency of a second potential diluter species (D. pulicaria).  However, frequency of this species was unrelated to epidemic size.  Thus, our results support the Dilution Effect Hypothesis when diluters are predominantly Ceriodaphnia, but not when they are D. pulicaria.  Finally, causality of all links in our final model can be traced back to variation in habitat.  Small deep-water refuges caused high predation intensity, which in turn caused a high frequency of Ceriodaphnia diluters, which ultimately lowered disease.  In contrast, large deep-water refuges caused low predation intensity, which in turn caused a high frequency of D. pulicaria, but which had no effect on disease.  In summary, 1) the dilution effect mediated by Ceriodaphnia was the strongest and only direct regulator of disease; and 2) this community-level regulator of disease can be traced back to variation in habitat.