Predictive models of benthic invertebrate methylmercury in Ontario and Quebec lakes

Multivariate analyses on benthic invertebrate methylmercury concentrations ([MeHg]) and water chemistry from 12 Quebec water bodies were used to guide the construction of simple, predictive models of benthic invertebrate [MeHg] in 23 lakes in Ontario and Quebec. Separate predictive models for collector–shredder and predatory benthic invertebrates were constructed using multiple regression and were assessed for their predictive utility by cross-validation. Predatory benthic invertebrate [MeHg] was negatively related to pH and positively related to dissolved organic carbon (DOC) concentration (cross-validation r2 = 0.31). Collector–shredder [MeHg] was positively related to [DOC] only (cross-validation r2 = 0.13). Predictive utility of our models is similar to or surpasses that observed in previously published zooplankton MeHg models tested against independently collected data. Significant environmental variables and their contribution to the overall explanatory power of benthic invertebrate MeHg models are similar to those found in zooplankton models, suggesting that in both pelagic and benthic food webs, pH and DOC are important indicators of MeHg bioavailability. Although seasonal patterns in invertebrate [MeHg] were examined, none was detected. These models represent an effective means of identifying water bodies of interest for researchers and for reconstructing past benthic invertebrate [MeHg] patterns using archived water chemistry data.