Fig. 2

Actual number of infected Biomphalaria glabrata as a function of live Bi. glabrata at the end of the experiment. The response shown is restricted to mesocosm tanks in which infected Bi. glabrata were present (n = 15 mesocosms), effectively depicting the count portion of the zero-inflated model used to analyze effects on infected Bi. glabrata density. However, all 60 replicate mesocosms were used in the analysis (see Methods and Supplementary Table 2 for full model results). Live Bi. glabrata density was the only significant predictor of the count of infected Bi. glabrata in the model (other than spatial block) and explained 89% of the variation in the density of S. mansoni-infected Bi. glabrata in replicates in which infected snails were present. The solid line represents the predictions from a generalized linear model with a Poisson distribution including first- and second-degree polynomial terms for live Bi. glabrata density as predictors. Dashed lines indicate the 95% confidence band