Endosulfan application to a stream mesocosm: studies on fate, uptake into passive samplers and caged toxicity test with the fish M. ambigua.

A model mesocosm system was used to simulate the effect of endosulfan entering a waterway from episodic events such as accidental overspray or in runoff water containing contaminated sediment following storm events. The fate of technical endosulfan applied to a 24-stream mesocosm was compared in experiments where the pesticide was applied either directly as water contamination or after being pre-bound to sediment. The flow of water through the streams was discontinued for a 10-12 h period following the pesticide application. Following the water application, only approximately 3% of endosulfan remained in the streams after 4 days and then was not detectable after 7 days. In contrast, after application pre-bound to sediment, approximately 33% of the endosulfan remained in the streams after 4 days and 14% after 7 days. Additionally, with the sediment-bound application, the proportion of endosulfan was higher in the substrate (11%) after 7 days than in the overlying water (3%), and approximately 1% was oxidised to the sulphate form. The dissipation of endosulfan in the water column of both experiments followed a two-parameter exponential decay model characterised by a relatively fast first-order single-phase process. In sediment of both experiments and the gravel of the sediment-dosing experiment, the dissipation of endosulfan followed more closely a four-parameter bi-exponential decay model characterised by first-order kinetics of two fractions: the first fraction dissipates quickly, and in the longer term the second fraction dominates the overall dissipation with a slower rate. In the gravel section of the water-dosing experiment, endosulfan dissipation was characterised by relatively very slow two-parameter exponential decay. The overall dissipation rates of the alpha- and beta-endosulfan isomers were significantly higher in the water-dosing than in the sediment-dosing experiment, except in the gravel section of the mesocosm. The uptake of the endosulfan into passive samplers constructed from polyethylene membrane bags containing trimethylpentane solvent (TRIMPS) placed in the overlying water was linear. In contrast, TRIMPS buried in sediment failed to uptake endosulfan from the sediment substrate, indicating that short-term deployment of passive samplers can only be used to determine time-weighted average concentrations of bioavailable chemicals in the water column. A 34-h LC50 of 2.8 microg/l [95% confidence interval (CI) 1.5-4.2 microg/l] for juveniles of the native fish Macquaria ambigua was obtained when exposed during the water-dosing experiment. This study demonstrated that the pulse entry of sediment contaminated with endosulfan into a receiving waterway was more persistent compared with direct aqueous contamination and the endosulfan would be bioavailable to pelagic organisms following a gradual partitioning to the water column.