Evaluation of model predictions of the persistence and ecological effects of diflubenzuron in a littoral ecosystem

Abstract

The accuracy of the Littoral Ecosystem Risk Assessment Model ( leram) was evaluated using results from a littoral enclosure study of the distribution, persistence, and ecological effects of the insecticide diflubenzuron. leram simulates the community and ecosystem level effects of a chemical stressor by linking single-species toxicity data to a bioenergetic model of a littoral ecosystem. This study evaluated (1) the accuracy of a simple fate and persistence model used in conjunction with leram, (2) leram's ability to predict, a priori, the effects of diflubenzuron on aquatic life, and (3) leram's usefulness for selecting treatment concentrations for an actual littoral enclosure study. The ecological effects of nine initial diflubenzuron concentrations ranging from 0.2 to 35.0 μg/l were simulated. The field study used to evaluate these predictions was performed in 12 littoral enclosures constructed in a 2-ha pond in northeastern Minnesota. There were four untreated controls and two replicates of each of four diflubenzuron concentrations (0.7, 2.5, 7.0, and 30.0 μg/l). The half-life of diflubenzuron predicted by the fate and persistence model was 4 to 5 times lower than the half-life observed in the littoral enclosure water column. This discrepancy was related to the simplistic nature of the two-compartment (water, sediment) model which represented the water column as a single, fully mixed compartment. The smallest model-predicted LOEC (lowest observed effect concentration), where the effect was a decrease in biomass, was 3.2 μg/l. At this concentration, reductions were predicted to occur in biomass of insects, herbivorous copepods, and amphipods. Field-observed LOECs for the same taxa were 2.5, 0.7, and 0.7 μg/l, respectively. Cladocera was the most sensitive taxon in the field study (LOEC = 0.7 μg/l), but leram did not predict direct toxic effects on this taxon until the diflubenzuron concentration was ≥ 6.4 μg/l. This discrepancy, and leram's general ability to predict effects observed in the field, was directly related to the size and quality of the available laboratory single-species toxicity data base used as input for the model. Indirect effects of diflubenzuron on growth of larval bluegill sunfish was predicted by leram, but at a concentration five times higher than observed. Overall, leram correctly predicted that changes in the littoral community would begin at 0.8 μg/l, with concentrations > 35.0 μg/l yielding negligible new information on effects of diflubenzuron on aquatic biota. Predictions at intermediate concentrations varied in accuracy, with some indirect responses being exaggerated by cascading effects through the trophic levels of the ecosystem.