SURFACE WATER SAMPLING AND ANALYSIS TO PRIORITIZE POTENTIALLY PESTICIDE “IMPAIRED” WATERBODIES IN MISSISSIPPI'S UPPER PEARL RIVER BASIN

Nationwide litigation over the past decade has forced the Environmental Protection Agency (EPA) and state primacy agencies to address the Clean Water Act's Total Maximum Daily Load (TMDL) provisions. Briefly, the TMDL process is tied to state primacy agencies' submission of a biannual CWA 303(d) list of impaired waterbodies to EPA. The primacy agency must then prepare and submit to EPA a TMDL for each listed impaired waterbody. The TMDL must address both point and nonpoint source (NPS) contaminants. Mississippi's 1998 303(d) list, which was submitted by the Mississippi Department of Environmental Quality (MDEQ) and approved by the EPA in September 1999, is currently the most recent list from Mississippi that has been approved by the EPA (MDEQ 1999). A more recent list has been drafted by the MDEQ and was submitted to the EPA in 2002, but is pending approval from the EPA (MDEQ 2002). On the approved 1998 303(d) list, Mississippi has 732 stream segments listed as impaired from a variety of 22 different sources of impairment. The Upper Pearl River Basin (UPRB) comprises a total of 25 impaired stream segments with 11 types of impairments (MDEQ, 1999). In Mississippi, “pesticides” are frequently listed as an “evaluated” NPS of contamination based on past land use patterns, with no regular monitoring data to support the listing. The UPRB was selected as a pilot project to validate correlatio ns between remotely sensed land use/land cover data and chemical specific pesticide levels in nearby surface waters because of its unique water quality and public health issues. Specifically, the Upper Pearl River has approximately 18 stream segments listed as impaired for pesticides and ultimately feeds into the Ross Barnett Reservoir, which is the drinking water supply for Jackson, Mississippi (MDEQ 1999). However, due to changes in land use/land cover (LULC) in the Upper Pearl River Watershed, waters that were once impaired by pesticides may not currently be impaired.To assess the current level of impairment by pesticides in this watershed, a sampling regime was implemented to collect grab samples at seven United States Geological Sur vey (USGS) gauged locations within the watershed. Samples were collected weekly from May through August 2002, and monthly thereafter through May 2003. Samples were extracted via Solid Phase Extraction (SPE), and the extracts were then analyzed using gas chromatography – mass selective detection (GC-MSD) or high performance liquid chromatography – photodiode array detection (HPLCPDA). Each extracted sample set includes two- liter samples from the seven selected sites, two lab spikes in deionized (DI) water, a field spike, a DI water blank, and a glassware wash. A multi-residue method was then used to analyze the surface water samples for fifteen pesticides: triclopyr, 2,4-D, tebuthiuron, simazine, atrazine, metribuzin, alachlor, metolachlor, cyanazine, norflurazon, hexazinone, pendimethalin, DDT insecticide degradation product p,p'-DDE, diuron, and fluometuron.Mean percent recoveries for spiked samples ranged from 31.4% for the metribuzin field spike to 128.0% for the norflurazon low spike. However, most average spike recoveries fell within an acceptable range (i.e., 70 to 120% recovery). The level of quantification (LOQ) for all compounds was 0.1 ppb. Results include only detections that were at or above the LOQ. Out of 690 total detections, Burnside was the site with the highest percentage of total detections at 18.7%. Walnut Grove had the lowest percentage of detections at 11.4%. Metolachlor was the most frequently detected compound at 76%, and atrazine was also detected 51% of the time. Fluometuron was the least detected compound at 9% out of a possible 167 detections. Subsequent phases of this study include comparing the aforementioned pesticide results with pesticide loads predicted by a runoff model based on remotely sensed land use patterns. Pesticide detections and loads will also being compared to types of land use/land cover near the sampling sites