Abstract
Development of assessment endpoints and conceptual models aids ecological risk assessors in identifying measurable attributes that will allow quantification and prediction of risk. Measures of exposure and effect are explicitly considered, usually quantitatively, in nearly every ecological risk assessment, while measures of ecosystem characteristics are generally addressed only implicitly, if at all. Yet these characteristics influence both the behavior and location of assessment endpoint entities and the spatial and temporal distribution of stressors. This case study illustrates use of a regression partitioning model to quantify the influence of ecosystem characteristics (e.g., land use patterns, nutrient concentrations) on the concentration of a chemical stressor (atrazine) in surface waters of a large river basin. The model partitioned the basin into five land use groups ranging from High Forested to Very High Agriculture. Literature-derived chronic effects data were used with a joint-probability model to characterize atrazine risk to an aquatic assessment entity in each of these land use subgroups. Atrazine concentrations and risk directly correlated with the intensity of agricultural land use. This permits risk management to focus on agricultural areas within the basin; a focus that would not have been possible without explicitly considering ecosystem characteristics.