Inferring long-term nutrient changes in southeastern Ontario lakes: comparing paleolimnological and mass-balance models

Lake eutrophication continues to be a major concern in many lake regions, but long-term monitoring data are often lacking. Therefore, indirect proxy methods must be used to infer these missing data sets. Two methods were applied to infer pre-industrial and present-day lakewater total phosphorus concentrations (TP) in a suite of 50 hardwater lakes in southern Ontario (Canada). One method inferred TP from the diatom species composition in the tops (present-day inferences) and bottoms (pre-1850 inferences) of sediment cores. The other method applied the Lakeshore Capacity Model (LCM), which is a mass-balance model based on phosphorus export coefficients that relate lakes and their watershed characteristics to epilimnetic nutrient concentrations. Diatom-based estimates of preindustrial to present-day change show that 78% of the lakes increased in TP (29% significantly) and 8% decreased. According to model error, 63% of the lakes have not significantly changed. LCM estimates show that 56% of the lakes have increased in TP, and the remainder (44%) have not changed. The average inferred increase in TP was similar for both models, but a lake-by-lake comparison indicated marked differences in model output. In particular, a paired comparison of diatom-based and LCM-based inferences of preindustrial TP shows no correlation. It is suggested that lake managers be thorough when collecting data for either model, and model selection should be carefully considered. The LCM and diatom-based models perform better in regions that are geologically similar to where the respective models were calibrated. Advantages and disadvantages of each model are further discussed.