Ecoregional or site-specific lake nutrient criteria? Evidence from ecological fallacy

Ecoregional nutrient criteria are widely used but their validity has rarely been verified competed with site-specific criteria. In this study, we introduced ecological fallacy, which describes phenomenon that site-specific stressor-response relationships cannot be deducted from ecoregional relationship and vice versa, to explore the spatial scale of nutrient criteria. A long-term and nationwide water quality dataset of lakes and reservoirs was used to determine if ecological fallacy existed or not. Ecoregional and site-specific nutrient-Chlorophyll a relationships were built employing Bayesian linear model and hierarchical model, respectively. By comparing ecoregional and site-specific relationships, we found that ecological fallacy existed in each ecoregion. Ecoregional relationship may misidentify limiting nutrient or miscalculate the nutrient effect direction or magnitude. We found huge differences between estimated Chlorophyll a concentrations deduced from regional and site-specific relationships conditioning average TN or TP concentrations. Based on these results, we determined that lake nutrient criteria should be site-specific, primarily to avoid ecological fallacy rather than to improve their accuracy. These findings could guide the future nutrient criteria development. We further recommended partial pooling of data to develop stressor-response relationship facing with intensive environmental and ecological data.