(PH251) Estimating organochlorine pesticide concentrations in fish using watershed characteristics and measured concentrations in bed sediment.
Nowell, L1, Crawford, C2, Thelin, G1, Nakagaki, N1, 1 U.S. Geological Survey, Sacramento, CA, USA2 U.S. Geological Survey, Indianapolis, IN, USA
ABSTRACT- Regression models were developed for estimating organochlorine pesticide concentrations in whole fish from U.S. streams using measured concentrations in bed sediment and selected watershed characteristics, including historical use intensity. Concentrations in whole fish and bed sediment were measured in samples collected at over 650 stream sites nationwide during 1992–2001, as part of the U.S. Geological Survey's National Water Quality Assessment (NAWQA) Program. Explanatory variables included estimates of past agricultural use of organochlorine pesticides, which were calculated from regional application rates developed using U.S. Department of Agriculture data on pesticide use by farmers in 1966 and county-level harvested acreage of crops from the 1964 and 1978 Census of Agriculture. Other explanatory variables included surrogates for urban use (such as population density and percent urban land use and land cover in the basin), soil characteristics, hydrologic parameters (such as mean annual runoff), climate variables, and agricultural management practices. Tobit regression methods were used for developing the models because many of the pesticide concentration values used as response variables were reported as less than a detection threshold, resulting in censored data. The models explained between 50 and 70 percent of the variability in concentrations of individual organochlorine analytes, such as p,p'-DDE, trans-nonachlor, and dieldrin, in whole fish.
Key words: regression model, organochlorine pesticide, contaminants in fish, contaminants in sediment