Principal Investigator: G.I Scott (NOAA National Marine Fisheries Service Charleston and Beaufort Laboratories )

Collaborating scientists: Michael H. Fulton, John Kucklick, John Ramsdell, Peter B.Key, Edward F. Wirth (NOAA National Marine Fisheries Service Southeast Fisheries Science Center Charleston Laboratory)

Dr. Gordon Thayer (NOAA National Marine Fisheries Service Southeast Fisheries Science Center Beaufort Laboratory) and G. Tom Chandler (School of Public Health University of South Carolina)

INTRODUCTION

Both field and laboratory research was conducted in FY97 to assess the potential for contaminant loading of agricultural pesticides via nonpoint source (NPS) runoff into Florida Bay and resulting effects on living marine resources. Field studies were focused on measuring the occurrence of agricultural pesticides, such as endosulfan, in water, sediment, biota (oysters and fish) and Semi-Permeable Membrane Devices (SPMDs) within the environs of Florida Bay so that a potential risk assessment characterization could be made. This risk assessment model would attempt to correlate the prediction of trophically transferred endosulfan and other pesticides from surface waters to estuarine/marine fish and shellfish using a Fugacity Modeling Approach. The use of SPMDs will validate if indeed trophic transfer of pesticides can be predicted based upon water:lipid transfer predicted by Log Octanol Water (KOWs) Coefficients, without having to factor in variables such metabolic and reproductive state of fish and shellfish.

The deteriorating conditions of Florida Bay [increased phytoplankton blooms, declines/death in sea grass beds and coral reef communities, and trophic shifts in fish communities towards more herbivorous fish species] may be related in part to increased contaminant loadings from upland adjacent terrestrial environments including both urban and agricultural areas. Urban contaminants are primarily polycyclic aromatic hydrocarbons (PAHs), some trace metals (e.g. Cu) and pesticides associated with urban landscaping (e.g. chlorpyrifos), termitacide control (formerly chlordane and presently chlorpyrifos), and golf course maintenance (chlorpyrifos and other insecticides as well as numerous herbicides and fungicides). Agricultural pesticides were of greatest concern because of the large concentration of vegetable farming in close proximity to Florida Bay. Vegetable farming requires high intensity pesticide control (>20 pounds of active ingredient pesticide per acre/crop). Often there may be more than one crop per year may further increase the amount of pesticides used. Agricultural pesticides of concern include azinphosmethyl, fenvalerate and endosulfan because of their high toxicity potential to aquatic organisms (Supertoxic = 96h LC50 values of <10 ug/L base U.S. Fish and Wildlife hazard ranking) and high prevalence rate in coastal fishkills (e.g. endosulfan was the primary pesticides involved in coastal fishkills in the U.S. from 1980-89).

Endosulfan was chosen as a contaminant for further toxicological studies because of the high usage on vegetable crops (e.g. >70% of the endosulfan in the southeastern U.S. is applied to vegetable crops in the south Florida), the moderately high KOW (3.50-4.57), high acute toxicity potential (96h LC50 values of < 1.00 ug/L for many for many shellfish and juvenile fish species), the very low marine water quality criteria (0.0085 ug/L) and its known endocrine disrupting capabilities demonstrated in vivo laboratory studies when compared to estrogen and DES. Laboratory studies were focused on assessing the potential for endosulfan to be an endocrine disrupting chemical to crustaceans using the grass shrimp (Palaemonetes pugio) as a water column model and the copepod (Amphiascus tenuiremis) as a sediment model for ecotoxicology risk assessment.

FIELD MONITORING STUDIES

Results of fields studies conducted during 1993-97 indicted the presence of endosulfan and other pesticides in surface waters from Florida Bay and surrounding environments ( Figure 2). Initial sampling during 1993, indicated the presence of endosulfan in both agricultural areas as well as in Florida Bay waters, indicating further study of this issue was warranted. In 1994, detectable levels of endosulfan were measured ranging from 0.017- 0.155 ug/L, but only in surface waters adjacent to agricultural areas. Similarly, detectable concentrations of atrazine ranging from 0.005-0.194 ug/L were measured but only in surface waters adjacent to agricultural areas. During 1995, detectable concentrations of endosulfan were measured in both bay sites and agricultural areas. In Florida Bay, 39% of the sites had detectable endosulfan concentrations ranging from During 1996, sampling of surface waters was conducted with improved sampling (e.g. larger volume water samplers) and extraction methodologies, which allowed for sub parts per trillion (ppt) quantification using negative ion mass spectrometry as well as dual column electron capture gas chromatography analytical procedures. This improved method allows for detection of several additional pesticides which had not been measurable in previous years including trifluralin, chlorthalonil, chlorpyrifos, alpha and gamma chlordane, and t-nonachlor. Results indicated that during one sampling period during the peak of the winter vegetable farming season, detectable levels of pesticides were measured at all sites sampled, including both bay and agricultural sites. All (100%) of the bay sites had detectable concentrations of endosulfan ranging from 0.000329 - 0.002333 ug/L. Similarly, all (100%) of the agricultural sites had detectable concentrations of endosulfan, ranging from 0.001657 - 0.038614 ug/L. None of these measured endosulfan concentrations of endosulfan I and II in both bay and agricultural sites exceeded EPA water quality criteria (0.0085 ug/L for marine waters and 0.0056 ug/L for freshwater). Endosulfan sulfate (generally the predominant isomer) analysis/quantification have not been completed and will be reported in subsequent reports and will ultimately add to these measured concentrations of endosulfan.

Spatial statistical analysis of these data indicated that the mean endosulfan I isomer concentrations in the bay was 0.001168 ug/L (+/- 0.000188) versus mean concentrations of 0.006538 ug/L (+/- 0.004378 ug/L) in agricultural areas. These data suggest that endosulfan I concentrations were reduced by 83% in the bay relative to concentrations measured in areas adjacent to agricultural areas. Similar results have been found in bays in Texas for other pesticides (aldicarb, carbofuran and atrazine) as concentrations declined by 90% in going from drainage ditches adjacent to agricultural areas to shallow Texas bay waters (Scott et al., 1992). Additionally, the endosulfan II isomer was only detected in surface waters in the agricultural areas and in those bay sites in close proximity to the C111 canal (e.g. Barnes Sound, Long Sound and Manatee Bay). These findings for the spatial distributions of endosulfan II suggest that the source of the endosulfan is from agricultural runoff.

In addition to endosulfan, other pesticides were measured with high frequency in Florida Bay (Figure 3) including chlorthalonil (73.6% of sites sampled), chlorpyrifos (94.7%), alpha- chlordane (42%), gamma-chlordane (68.4%),t-nonachlor (100%), alpha-HCH (73.6%), gamma-HCH (47.3%) and trifluralin (5.2%) at concentration which would not be acutely toxic( <0.002600 ug/L).The high prevalence rate of these pesticides as well as endosulfan raises the issues of the potential additive chronic toxicity of these pesticides to living marine resources of Florida Bay, since many of these pesticides are known Endocrine Disrupting Chemical (EDCs), with the potential to disrupt and alter development and reproduction. Recent studies published in Science (McLaughlin et al., 1996) and Environmental Health Perspectives (Sota et al., 1995) in both in vivo and animal studies have demonstrated that endosulfan is clearly an estrogen mimic in vertebrates and that combinations of pesticides in small doses (< than concentrations which cause significant binding with estrogen receptors), including endosulfan, have greater than additive estrogenic activity (e.g. implying the potential for synergistic effects). EDCs effects are different than traditional toxicological concerns in aquatic toxicology in that small doses may be as effective as high doses in altering endocrine function and compounds which are not estrogen agonist may be androgen agonist (e.g. estrogen antagonist), either of which may cause endocrine disruption. Additionally, EDCs may be most effective during sensitive stages of development (e.g. molting stages in crustaceans and during differentiation/development in fish embryos) and effects may be transgenic, manifesting effects in subsequent generations.

Results from 1997 (Figure 4) compared the levels of endosulfan in surface waters at selected sites (A3, A5, A7 and A8) with mean estimated surface water endosulfan concentrations which were derived by lipid normalization of endosulfan body burdens/the log Octanol water coefficient (KOW) for endosulfan. At site A3 adjacent to an intensive agricultural farming area, estimated surface water endosulfan levels derived from fish and Semi Permeable Membrane Devices (SPMDs) yield similar estimates and were comparable indicating that the trophic transfer of endosulfan to biota is highly predictable and can be modeled using a Fugacity (e.g.KOW-Lipid Normalization) Approach. Additionally, both estimated and actual endosulfan surface water concentrations were > the EPA Freshwater Quality Criteria for endosulfan. Similar results were obtained at other stations including both brackish and estuarine/marine stations using SPMDs, fish and oysters. At marine sites, oysters and SPMDs indicated that mean endosulfan surface water concentrations were just below the marine Water Quality Criteria established by EPA. These findings indicate that it my be possible to predict the trophic transfer of endosulfan and other lipophilic pesticides using fugacity modeling and that current monitoring programs of surface waters and sediments in this region (e.g. SFWMD and DERM) may utilize this procedure without having to add additional monitoring of biota requirements.

IN SITU AND LABORATORY TOXICITY TESTS

Most current aquatic toxicity testing models are not designed to adequately evaluate the transgenic effects of EDCs. Chandler (1992) has developed a copepod life cycle assay which allow for multiple generational testing (F0 --->F1---->F2 generational testing) within a 28 day period with both survival and reproduction/development endpoints. Field sediments were collected from a variety of sites (1996: A3=headwaters of the C111 Canal; A8=end of the C111 canal, Joe Bay, Little Madeira Bay and Barnes Sound; 1997: Slagle Ditch, Rankin Key, Snake Bite Canal, East Cape Canal, Crocodile Point, and Manatee Bay) in Florida Bay as well as two Florida Bay control sites and a laboratory reference sediment (pristine North Inlet in South Carolina) and tested for their chronic toxicity potential in a partial life cycle test (F0 ---->F1) using the marine benthic copepod Amphiascus tenuiremis. Twenty five barren females and 25 males are introduced to test sediments for 14 days. Test end points included: Adult survival (male, gravid females and nongravid females), egg production per female, nauplii production, and copepodite production. Results indicated significant (p < 0.05) reductions in the % of surviving males (e.g. selective male toxicity=Joe Bay and the end of the C111 canal), females (A3, A8, Joe Bay, Little Madeira Bay, Snake Bite, and East Cape Canal) and gravid females (A3, A8, Joe Bay, Little Madeira Bay and Manatee Bay) in adults. Reproduction was also affected generally at sites closest to the C111 canal including reduced naupliar production (A3= headwaters of the C111 canal, A8=end of the C111 Canal, Little Madeira Bay and Joe Bay), reduced copepodite production (A3= headwaters of the C111 canal, A8=end of the C111 Canal, Little Madeira Bay, Joe Bay, Snake Bite, Manatee Bay and Slagle Canal), and reduced clutch size per female (A8 and Manatee Bay) using ANOVA as well as multiple comparisons (Dunnett's and Tukeys) statistical analysis. Overall, the % of gravid females ranged from 69-82% (X=74.4%) in Florida Bay reference and control sediments versus 36-58% (X = 51%) at all Florida Bay Sites (an overall 31.5% reduction in the % gravid females). Although this difference for all pooled Florida Bay Sites was not significant (p < 0.09), the reduced clutch size and reduced naupliar production are suggestive of potential alterations in reproduction and development in copepods in certain regions of Florida Bay closest to the C111 canal. Of additional interest is the finding of selective male toxicity at certain Florida Bay sites (Joe Bay and the end of the C111 Canal). Many lipophilic compounds (e.g. oganochlorine pesticides) are more toxic to males than females as females are able to off load these contaminants into eggs/offspring during reproduction. This was observed with Kepone in the James River in blue crabs and other species. Studies are underway to identify any potential toxic compounds in the sediments used in these toxicity tests. Of particular interest are the sediments from Joe Bay since the highest endosulfan concentration detected in oysters by the NOAA NST Program were found there. Reduced benthic copepod production may have significant ecological implications since they are primary grazers on phytoplankton within Florida Bays well as a major food source for many marine/estuarine fish and shellfish species. Given the significant algal blooms and shifts in trophic structure among macropelagic fish populations towards increased herbivorous fish species which have occurred recently in Florida Bay, the role if any that decreased copepod production may play warrants further study. Future collaborations with Dr. Gary Kleppel, who is studying distributions of major water column copepod populations within Florida Bay, are planned which will focus on the toxicity potential of pesticides to pelagic copepod species.

Results of oyster studies involving deployment of oysters at the end of the C111 Canal, a Florida Bay reference site and a SC laboratory reference have indicated significant reductions in condition and gonadal indices in oysters from the end of the C111 canal. These findings are consistent with results of studies of agricultural NPS runoff in South Carolina in which significant reductions in condition and gonadal indices were observed immediately down stream of a major vegetable farming area following significant runoff events in which oysters bioconcentrated endosulfan (30-113 ug/kg) and other insecticides. Ernst (1977) similarly found reduced gonadal production in mussels exposed to endosulfan. Analysis of oyster tissues are underway to determine if deployed oysters bioconcentrated contaminants from surface waters/sediments at each site and if the observed effects can be correlated with chemical contaminant exposure. Results from 1997 confirmed the bioconcentration of endosulfan in deployed oysters but what was most interesting is the fact that our reference site oysters in Long Sound were exposed to lower salinities and higher endosulfan surface water concentrations than oysters at the end of the C111 canal. This is the result of the removal of the western levee of the canal which has allowed more freshwater flow into Long Sound. Studies conducted during the fall (e.g. wet season sampling) further found that salinities were so low (<3 ppt) that deployed oysters ,in Long Sound had 100% mortality due to the low salinities in this area.

In situ toxicity tests conducted in freshwater regions of the C111 Canal (A3 and A5) using mosquito fish (Gambusia affinis) and in marine waters (A7 and A8) using the grass shrimp (Palaemonetes pugio) generally did not measure significant mortality except during the wet season sampling of fall, 1997. During this time period a significant rain event (>4 cm rain/24h) resulted in significant loading of oxygen demanding waste at site A3, which caused dissolved oxygen levels to fall to < 0.05 mg/L dissolved oxygen. This resulted in 100% mortality in fish at site A3. These findings have significant management implications. If highly oxygen demanding waste is discharged into the Everglades as may occur with modifications in the C111 canal flow/management, impacts may be potentially observed in fish communities within the receiving waters. It would be prudent to maintain constant monitoring of surface water dissolved oxygen levels in the C111 canal prior to discharge of waters to the Everglades to assure that high BOD demanding waters are not released into these sensitive aquatic habitats.