Temporal, spatial, and between-host comparisons of patterns of parasitism in lake zooplankton

In nature, multiple parasite species infect multiple host species and are influenced by processes operating across different spatial and temporal scales. Data sets incorporating these complexities offer exciting opportunities to examine factors that shape epidemics. We present a method using generalized linear mixed models in a multilevel modeling framework to analyze patterns of variances and correlations in binomially distributed prevalence data. We then apply it to a multi-lake, multiyear data set involving two Daphnia host species and nine microparasite species. We found that the largest source of variation in parasite prevalence was the species identities of host-parasite pairs, indicating strong host-parasite specificity. Within host-parasite combinations, spatial variation (among lakes) exceeded interannual variation. This suggests that factors promoting differences among lakes (e.g., habitat characteristics and species interactions) better explain variation in peak infection prevalence in our data set than factors driving differences among years (e.g., climate). Prevalences of parasites in D. dentifera were more positively correlated than those for D. pulicaria, suggesting that similar factors influenced epidemic size among parasites in D. dentifera. Overall, this study demonstrates a method for parsing patterns of variation and covariation in infection prevalence data, providing greater insight into the relative importance of different underlying drivers of parasitism.