Local environment and sampling bias drive parasite prevalence estimates in freshwater fish communities

Parasite occurrence and infection estimates vary through time and space, making understanding the underlying drivers highly complex. Comparative studies based on empirical data must consider the factors of variation involved in estimating infection metrics in natural populations to make appropriate and reliable comparisons. Using a multi‐scale approach, we explored the sources of variation in the estimation of infection prevalence, focusing on black spot disease in littoral freshwater fish communities sampled across 15 lakes in Québec, Canada. Method‐related sampling biases led to significant variations in prevalence estimates and spatial patterns of disease occurrence. Our results also indicated that low sampling efforts tend to overestimate the prevalence of infection in the landscape, with minimum sampling effort required to estimate an accurate infection prevalence depending on the sampling method employed. Our results showed that infection prevalence is spatially heterogeneous across the landscape with evidence of infection hotspots and coldspots. Physico‐chemical characteristics of the sites and local fish community structure were found to be the best drivers of infection at smaller spatial scales. Furthermore, our results suggest dilution effects, due to physical obstruction and compatibility barriers, limit the survival of the free‐living cercaria parasite life stage. Several relationships between infection prevalence and environmental drivers revealed non‐linearity, suggesting complex interactions. Examining infection prevalence data at various spatial scales revealed method‐induced biases, sampling effort effect, and environment‐driven relationships underscoring the importance of context‐dependencies and scale‐dependencies in empirical studies on host–parasite interactions.