Feasibility of Stocking Red Clay Creek with Trout

Feasibility of Stocking Red Clay Creek with Trout Delaware DNREC 4/28/2017

Table of Contents Table of Contents... i List of Tables... ii List of Figures... iii List of Appendices... iv Executive Summary...v Principal Investigators/Authors...1 Objective...1 Introduction...1 Field and Laboratory Methods...2 Data Analysis Methods...4 Results...7 Discussion...31 References...33 Appendices...36 i

List of Tables Table 1. General Water Quality Results Red Clay Creek Trout Study...9 Table 2. Lengths and Weights of Individual Trout...11 Table 3. Stomach Contents of Individual Trout...13 Table 4. Concentrations of Contaminants Consistently Detected in Trout Samples...14 Table 5. Concentrations of Contaminants Consistently Detected in Water Samples...20 Table 6. Bioaccumulation Factors for PCB, DDX, Chlordane and Dieldrin in Trout...27 Table 7. Bioaccumulation Factors for PCB, DDX, Chlordane and Dieldrin in Resident White Sucker...28 ii

List of Figures Figure 1. Rainfall and Streamflow during Red Clay Creek Trout Study...7 Figure 2. Figure 3. Mean Lengths and 95 Percent Least Significant Difference (LSD) Intervals for Trout Samples...12 Mean Weights and 95 Percent Least Significant Difference (LSD) Intervals for Trout Samples...12 Figure 4. PCB, Dioxin and Furan TEQ, DDX and Chlordane in Trout Samples...16 Figure 5. Dieldrin and Heptachlor in Trout Samples...17 Figure 6. Percent of Steady-State Concentrations in Trout versus Octanol-Water Partition Coefficient for PCB, DDX and Dieldrin...19 Figure 7. PCB, DDX, Chlordane and Dieldrin in Filtered Water Samples...21 Figure 8. DDX in Water Samples versus Streamflow...21 Figure 9. PCB Homolog Fingerprints in Water Samples and Trout Samples...22 Figure 10. Figure 11. Figure 12. Figure 13. PCB Homolog Fingerprints for Hatchery Trout and 50:50 Mixture of Aroclor 1254 and 1260...23 PCB Homolog Fingerprint in Water and Trout at 12 Weeks versus Aroclor 1248...24 DDT, DDD and DDE in Water and Trout Samples (absolute and percent contribution)...25 Chlordane Components in Water and Trout Samples (absolute and percent contribution)...26 iii

List of Appendices Appendix A. Mean Daily Flow at USGS Gage 01479820 Red Clay Creek Near Kennett Square, PA and Daily Rainfall at UD DEOS Weather Station Hockessin VFL...37 Appendix B. Summary Statistics for Contaminants Consistently Detected in Filtered Water Samples During Red Clay Creek Trout Study...40 Appendix C. Human Health Rick Assessment for Consumption of Rainbow Trout Stocked in the Red Clay Creek...42 iv

EXECUTIVE SUMMARY The Red Clay Creek in Southern Chester County, Pennsylvania and Northern New Castle County, Delaware has a long history of water quality problems. Fish contamination from legacy substances such as polychlorinated biphenyls (PCBs) and the pesticide DDT lead to the issuance of fish consumption advisories in the 1980s. These concerns also lead to the discontinuation of put-and-take trout stocking in this same time period. Conditions have improved to the point where advisories are now less stringent and there s a desire to reintroduce trout stocking to the Delaware portion of the Red Clay. Before that can be done however, the Delaware Division of Fish and Wildlife, in collaboration with the Delaware Division of Watershed Stewardship, studied the uptake of contaminants in stocked trout in 2016 in the area that was historically stocked. The study objective was to assess if contaminants are taken up by the trout after being placed into the Red Clay and if so, how fast, and whether that uptake represents a significant health risk to anglers who may consume their catch. The management objective was to determine the feasibility of reintroduce trout stocking to the Red Clay Creek in Delaware. Rainbow Trout directly from the hatchery had low concentrations of contaminants. When the hatchery trout were placed into the Red Clay Creek, they immediately began to accumulate and continued to accumulate several legacy contaminants including PCB, DDX, chlordane and dieldrin over the course of the study. The degree of accumulation was found to be a function of the concentrations in the water, streamflow, and the hydrophobicity of the contaminant. Despite the accumulation of these contaminants, potential health risks to trout anglers can be kept low by moderating consumption (http://www.eregulations.com/delaware/fishing/fishconsumption-advisories/). This can be done by following the existing advice of no more than twelve (12) eight ounce meals of stocked trout per year which currently applies to other trout stocking waters in Delaware (except for the Christina River where the advice is half as much). Hence, it is feasible to return the Red Clay Creek to its former trout stocking status. More correctly, contaminant uptake alone does not prevent the Red Clay from once again joining the rotation of designated trout stocking waters in Delaware. v

In addition to moderating consumption, another strategy to limit trout anglers from getting elevated exposures to contaminants in recaptured trout is to stock these trout a little later in the season. Because the trout, especially Rainbow Trout, cannot tolerate higher water temperatures, stocking them later shortens the window of time they can survive in the Red Clay and therefore limits the amount of chemical accumulation these fish will experience. This in turn limits potential exposure to trout anglers. vi

Feasibility of Stocking Red Clay Creek with Trout Principal Investigators: Richard Greene Division of Watershed Stewardship Michael Stangl Division of Fish and Wildlife Objective: Determine the feasibility of resuming trout stocking in Red Clay Creek, DE. Introduction: Red Clay Creek (RCC) drains a small (53.3 mi 2 ) and scenic watershed in southeastern Pennsylvania and northern Delaware. The waters of the Red Clay are used for a variety of purposes, including public and industrial water supply, irrigation, and general aquatic life maintenance and propagation. In addition, the Delaware portion of the Red Clay was designated for put-and-take trout fishing and was stocked from at least 1972 through 1986. The stream typically received between 10% - 17% of the total trout stocked in any given year, and the stream received a commensurate amount of effort as determined by streamside creel surveys (Martin and Whitmore 2005). Stocking was discontinued in 1986 based on elevated levels of bioaccumulative contaminants, primarily polychlorinated biphenyls (PCBs), in stocked trout recaptured from the RCC and tested for contaminants. Further, elevated contaminant levels in the resident fish lead the Delaware Department of Natural Resources and Environmental Control and the Delaware Division of Public Health to issue a strict DO NOT EAT fish consumption advisory in 1987. That advisory applied to all resident fish species in the RCC and was due to PCBs and the pesticide DDT and its breakdown products. Following aggressive clean-up of a PCB source located upstream in Pennsylvania s part of the watershed, periodic testing of the resident fish indicated that PCB levels had dropped. This lead to less restrictive advisories for consumption of resident fish from the RCC in 2006 and in 2008. Furthermore, tests performed on brown trout stocked into Red Clay Creek in 2011 indicated declining contaminant levels (Greene and Stangl 2011). Unfortunately, concentrations of PCBs and chlorinated pesticides in resident fish went back up in the fall of 2015 and the advisory needed to be made more stringent. That advisory, issued in 2016 and currently in effect, recommends that people eat no more than three (3) eight ounce meals of fish per year 1

caught from the Delaware portion of the RCC. Despite the recent setback, conditions have improved when viewed over the long run. For this reason and based on a desire to more fully utilize put-and-take trout waters in Delaware, the Delaware Division of Fish and Wildlife commissioned a study to evaluate the feasibility of bringing trout stocking back to the RCC after a 30 year hiatus. To answer that question, we analyzed Rainbow Trout (Oncorhynchus mykiss) delivered directly from the hatchery (to serve as control fish) along with Rainbow Trout stocked into the RCC from the same delivery and recaptured at various time intervals throughout the spring and summer of 2016. The primary objective of the study was to assess the rate of change of contaminant levels in the trout as a result of the fish s exposure to the RCC environment and to assess the associated health risk to anglers who might consume these trout. This report summarizes the methods used in the study along with the key findings and recommendations. Field and Laboratory Methods: Field and laboratory methods for this project are fully described in a project-specific sampling and analysis plan (DNREC 2016). Highlights are covered here. On April 4, 2016, approximately 500 Rainbow Trout, each between 0.75 pound and 1 pound, were stocked into the Red Clay Creek downstream from the DE/PA border and upstream of Yorklyn, DE. This is the traditional stocking area for the Red Clay. Immediately prior to stocking, five (5) trout were randomly collected directly from the hatchery truck to represent a control sample. These fish were never exposed to Red Clay Creek water. They were handled and processed following procedures described below. Spot observations along the stocking area confirmed that no holdover Rainbow Trout were present in the Creek prior to stocking. Low head dams on either end of the area inhibit movement of the trout beyond the stocking area. The stocking was not listed on the schedule supplied to the public, nor was it advertised to the public in hopes of preventing the removal of fish by anglers. On April 18 (i.e., 2 weeks after stocking) personnel from the Division of Fish and Wildlife returned to the creek and collected a sample of five (5) Rainbow Trout using a backpack electroshocker. Each trout was individually wrapped in aluminum foil and stored on ice for transport to the Department s Environmental Laboratory Section (ELS) in Dover for processing. Personnel from the ELS recorded the length and weight of each fish and then prepared a single 2

composite sample by combining equal mass aliquots of skin-on fillets from individual trout. Stomach contents of each fish were also recorded. This sampling and lab procedure was repeated on May 16 (i.e., 6 weeks post-stocking) and June 27, 2016 (i.e., 12 weeks poststocking). The duration of the field sampling was intended to cover the time over which Rainbow Trout could reasonably expect to tolerate water temperatures in the Red Clay Creek and therefore the length of time that trout anglers might catch them. Further, twelve (12) weeks was thought to be sufficient for these fish to reach new steady state contaminant levels based on theoretical bioaccumulation considerations (Chapra 1997). In addition to the trout samples, Department of Natural Resources and Environmental Control (DNREC) personnel also collected surface water samples from the stocking area on April 4, April 18, and May 16 and June 27, 2016 to complement the trout samples. The trout samples were analyzed for PCBs, organochlorine pesticides, dioxins and furans, polyaromatic hydrocarbons (PAHs), methylmercury, moisture content and lipid content. The surface water samples were filtered and analyzed for the same compounds as the fish with the exception of dioxins and furans, methylmercury and lipid. Separate water samples were also analyzed for total suspended solids (TSS) and various forms of carbon (i.e., particulate organic carbon (POC), dissolved organic carbon (DOC), and chlorophyll a (Chl a)). POC and DOC act as chemical sorbents and are needed to carry out chemical partitioning calculations. Finally, dissolved oxygen (DO) measured in mg/l, temperature (T) as C, ph, conductivity (us/cm) and salinity ( ) were recorded at the time of sampling using a hand-held YSI field meter. All field-related activities were recorded in a field notebook. All samples were successfully collected as planned without incident. The organic contaminants were analyzed at a specialty laboratory (SGS AXYS Analytical Ltd.) using high resolution methods under contract with the DNREC. Mercury analyses of the fish were also analyzed by a specialty laboratory (Test America) under contract with the DNREC. The solids and carbon analyses of the water samples were analyzed at the DNREC laboratory using standard procedures. All samples were successfully analyzed and the data were judged to be suitable for their intended use. 3

Data Analysis Methods: Rainfall, Streamflow, and General Water Quality: Rainfall and runoff can influence surface water quality. This is especially true for legacy contaminants that may be sorbed to surface soils that get eroded off the land and delivered to streams during runoff events. As such, weather conditions over the course of the study, including rainfall, were downloaded from the Delaware Environmental Observing System s web site (http://www.deos.udel.edu/) for the Hockessin-VFC weather station located approximately 1.5 miles WSW from the study area. Streamflow data for the Red Clay Creek were downloaded from the USGS s web site (https://waterdata.usgs.gov/de/nwis/rt) for the Kennett Square, PA gaging station (01479820) located immediately upstream of the study area. Rainfall and streamflow data were reviewed for possible influence over study results. General water quality conditions (e.g., solids and carbon data, plus field measurements such as DO and T) were also reviewed and are discussed in the Results section. Condition of Trout: Hatchery trout are accustomed to being fed and so placing them in a natural stream is expected to create a challenge for them to identify and consume food. Lower feeding rates in the wild can result in overall loss of weight and a drop in stored fat. The lack of significant feeding in the wild can reduce an important pathway for contaminant exposure in the trout; namely, foodchain transfer. Further, loss of fat (measured as % lipid) can also reduce bioaccumulation because lipophilic contaminants preferentially accumulate in fat. For these reasons, part of our preliminary analysis included assessing changes in length, weight, lipid content and stomach contents of the trout. Contaminants: A primary objective of this study was to assess whether and to what extent contaminant concentrations change over the course of the study. Initial assessment therefore included tabulating and plotting the fish and water results as a function of time to observe general trends. Contaminants that increased over the course of the study were singled out for additional examination. This included regression analysis to characterize the rate of increase as well as assessing any possible relationships between hydrologic drivers (rainfall and streamflow) and contaminant concentrations. 4

PCBs were one of the contaminants analyzed that consisted of a complex mixture of 209 possible compounds. Total PCB was calculated as the sum of these 209 congeners. Congeners with the same number of chlorine atoms, regardless of position of the chlorine atoms on the biphenyl base molecule, were binned into one of ten PCB homolog groups. Homolog fingerprints of each sample were then produced by dividing the concentration of each homolog group by the total PCB concentration in the sample. The resulting chemical fingerprints were compared to each other and to commercial Aroclor PCB mixtures using correlation analysis. Dioxins and furans also were examined over the course of the study. Dioxin toxicity equivalents (TEQs) were calculated by multiplying the concentration of each 2,3,7,8-substituted dioxin and furan compound by its associated mammalian toxicity equivalency factor (TEF) and then summing the resulting partial products. TEFs were taken from Van den Berg et al. 2006. Chlorinated pesticides DDT and chlordane levels were evaluated during the study. Total DDT (herein referred to as DDX) was calculated as the sum of the o,p and p,p isomers of DDT, DDD and DDE. The percent of each of these six compounds was calculated to determine how much of the mixture is parent DDT versus the breakdown products DDD and DDE. This is a type of chemical fingerprinting. Total chlordane was calculated as the sum of oxychlordane, the cis- and trans- isomers of chlordane, and the cis- and trans- isomers of nonachlor. The percent contribution of these individual compounds to total chlordane was also calculated to provide a chemical fingerprint. Dieldrin and heptachlor, two additional chlorinated pesticides, also increased over the course of the study. However, there are no isomers of these compounds and so there was no need to add multiple compounds to obtain totals. Even so, it s important to note that the chlorinated pesticide aldrin, which was also measured in this study, rapidly metabolizes to dieldrin in the environment. However, since the concentration of aldrin declined in the trout samples over the course of the study and was not detected in the water samples, no additional attention is given to aldrin herein. Hexachlorocyclohexane (HCH) and polyaromatic hydrocarbons (PAHs) were represented by multiple compounds which were summed to simplify presentation. Total hexachlorocyclohexane (HCH) was calculated as the sum of alpha, beta, gamma and delta HCH. Note that gamma HCH is also known as Lindane. With regard to PAHs, two data reduction techniques were used. First, the sum of thirty-four (34) specific parent PAH compounds and 5

alkylated PAH homologs was calculated for each sample (Burgess et al. 2013). Second, benzo[a]pyrene toxicity equivalents (B[a]P TEQs) was calculated for each sample based on TEFs published by Nisbit and LaGoy (1992). Another key aspect of the study was to assess how the concentration in the trout related to the concentration in the water. This relationship was quantified as a bioaccumulation factor (BAF). BAF was calculated as the lipid normalized concentration of the contaminant in the fish to the freely dissolved concentration of the contaminant in the water. The freely dissolved concentration in the water was estimated based upon equilibrium partitioning. Here, the filtered water result, which was measured and which represents the sum of the truly dissolved contaminant concentration plus contaminant associated with dissolved organic carbon (DOC), is parsed into these two components using a linear free energy relationship (LFER) developed by Burkhard (2000). That LFER depends on the hydrophobicity of the chemical (expressed as the logarithm of the octanol-water partition coefficient) and the measured DOC of the site water. BAF calculations were only carried out using the final concentration in the trout (measured on June 27, 2016) but considered both the final concentration in the water and alternatively the average concentration in the water (which integrates the exposure over a longer time period). The resulting BAFs for the trout are compared to BAFs for resident species at the study site. Human Health Risk Assessment: Finally, the concentrations in the trout were used within a human health risk assessment framework to forecast health risks to trout fishermen who hypothetically catch and consume trout from the Red Clay Creek. In standard human health risk assessment for ingesting contaminants in fish, the concentration of the contaminant in the fish is considered constant and risk is projected into the future. In this case however, the concentration in the fish is not constant over time and so the assessment is more complex. The approach we used involved assessing risk based on the concentrations in the fish at each sampling time following normal Delaware s technical procedures (DNREC and DHSS 2017), fitting the point estimates of risk to an empirical equation, and then numerically integrating that equation over time to determine the number of eight ounce meals a trout fisherman could eat over various times while keeping health risk low (i.e., less than 1-in-100,000 cancer risk). 6

Results: Rainfall and Streamflow: Appendix A includes a tabular compilation of daily rainfall and daily streamflow data over the course of the study. Figure 1 displays the data graphically and indicates the dates when the trout and water samples were collected during the time series. Figure 1. Rainfall and Streamflow during Red Clay Creek Trout Study Overall, note that streamflow temporarily increased in response to the larger rainfall events and that streamflow trended downward over the study. It rained on 34 of the 84 days (or 40.5% of the days) with a total rainfall amount of 10.94 inches. The peak rainfall total of 1.19 inches occurred on May 5, 2016, leading to a peak average streamflow on that day of 138 cubic feet per second (cfs). That event occurred a week-and-a- half prior to the third sampling day. Twentysix (or 29.5%) of the study days had daily rainfall amounts greater than 0.1 inch, an amount often 7

assumed to produce at least some runoff from the land. Although no sampling was conducted during rain, sampling did occur within 72 hours following at least 0.1 inches of rain. This happened prior to the first, third and fourth sampling events. Even so, those rainfall amounts were not extreme and they did not result in significant increases in streamflows immediately prior or during sampling. This is not to say that contaminants did not enter the Red Clay Creek during the course of the study as a result of the upland erosion and runoff or that contaminants already in the stream were not mobilized and redistributed during storm events. General Water Quality: Table 1 presents the general water quality measurements obtained over the study period and includes the sample description, results of the lab measurements, and the field measurement results. Quality assurance (QA) data, including a field duplicate and a field blank, are also available and demonstrate reproducibility and clean field and lab methods. Solids concentrations, represented by TSS, ranged between 0.6 mg/l and 2.3 mg/l and were all J qualified, meaning the concentrations were below the range for accurate quantitation based on their low levels. Chlorophyll a, an important natural pigment in aquatic plants such as algae, ranged between 2.4 ug/l and 7.9 ug/l and decreased over the study. This pattern likely reflects initial algal growth in the spring when nutrients are abundant and incident sunlight on the water is not impeded by a full forest canopy. POC ranged between 0.177 mg/l and 0.515 mg/l and showed no pattern over time. DOC ranged narrowly between 2.5 mg/l and 3.5 mg/l and also showed no pattern over time other than sharing its peak concentration with POC on the second sampling date. All of the lab results were within their normal range for the Red Clay Creek. With regard to the field measurements of general water quality, DO decreased over the study from 14 mg/l to 7.58 mg/l. Temperature, on the other hand, increased from 8.25 C to 19.41 C. These ending values, although not lethal to Rainbow Trout, are likewise not optimal. The ph of the water samples, which is a measure of how acidic or basic the water is, ranged from 8.33 to 7.56. All samples were slightly basic and within the expected natural range of 6.5 to 8.5 for free-flowing streams in the area. Finally, conductance, which is related to the ionic strength of the water, ranged between 475 and 515 us/cm and increased slightly over the course of the study. These values are also within the normal range for the area. Further, conductivity often increases as streamflow decreases into the summer months, as was seen here. 8

Table 1. General Water Quality Results Red Clay Creek Trout Study. Sample Description Sample Number 1603035-001 1603035-002 1605016-001 1606036-001 Sample Name RCCWater_Control RCCWater_2Weeks RCCWater_6Weeks RCCWater_12Weeks Sample Matrix Surface Water Surface Water Surface Water Surface Water Sample Date 4/4/2016 4/18/2016 5/16/2016 6/27/2016 Sample Time 10:16 9:06 9:12 9:07 Elapsed Time (days) 0 14 42 84 Elapsed Time (weeks) 0 2 6 12 Sample Type Field Field Field Field Lab Measurements TSS (mg/l) 0.6 J 1.9 J 2.3 J 1.8 J POC (mg/l) 0.306 0.515 0.177 0.289 DOC (mg/l) 2.5 3.5 2.5 2.6 Chl a (ug/l) 7.9 7.11 2.51 2.4 Field Measurements DO (mg/l) 14 11.43 10.76 7.58 Temp ( C) 8.25 12.64 10.66 19.41 ph (SU) 8.33 7.61 7.56 7.83 Cond (us/cm) 475 484 495 515 Salinity (ppt) 0.23 0.23 0.24 0.25 9

Aside from increasing temperature and lower DO toward the end of the study, general water quality was sufficient to at least support survival of the trout. It s important to note that very few of the approximately 500 Rainbow Trout that were originally stocked on April 4, 2016 remained to be recaptured on June 27, 2016. It is doubtful that the reduction in numbers can be attributed to increased water temperature since rainbow trout have been known to tolerate temperature as high as 25 C. At the same time, it s unlikely that this species would have much of a chance of surviving the higher temperatures that typically occur in this stream later in the year, say July, August and September. Condition of the Trout: Recall that each trout sample consisted of fish individual Rainbow Trout that were composited for purposes of contaminants analyses. Prior to compositing, the lengths and weights of the individual fish were recoded, as were the stomach contents (except for the control fish directly from the hatchery). Table 2 contains the length and weight data for the individual fish retained on the four sample dates. The final column of the table also presents the average length and average weight of the fish that went into the composite sample for the four sample dates. Figure 2 compares the mean lengths and 95 percent least significant difference (LSD) intervals between the control fish, the fish at two (2) weeks, the fish at six (6) weeks, and the fish at twelve (12) weeks. Figure 3 is a similar plot for weight. Figure 2 suggests that the mean lengths were similar among the control fish, the fish at 2 weeks and the fish at 6 weeks but that the fish at 12 weeks were not as long. These similarities and differences were confirmed by more formal statistical testing (Fisher s LSD procedure at 95% confidence level). In a similar way, mean weights were not statistically different amongst the control fish, the fish at 2 weeks, and the fish at 6 weeks but the fish at 12 weeks weighed far less (Fisher s LSD procedure at 95% confidence). There was a clear loss of weight starting with the 2 week sample through the 12 weeks sample. Because all of the trout were of similar length and weight upon stocking, the reduction in length at 12 weeks, and more clearly, the reduction in weight from 2 weeks on, is judged to be real and most likely attributable to lower feeding and associated metabolic wasting. Further evidence along this line comes from the examination of stomach contents of the fish and the decline in lipid levels. 10

Table 2. Lengths and Weights of Individual Trout Controls ELS Sample Number 1603035-005 1603035-006 1603035-007 1603035-008 1603035-009 Sample Name RCCTrout_Control 1 RCCTrout_Control 2 RCCTrout_Control 3 RCCTrout_Control 4 RCCTrout_Control 5 Sample Date 4/4/2016 4/4/2016 4/4/2016 4/4/2016 4/4/2016 Elapsed Time (days) 0 0 0 0 0 Length (mm) 317 317 311 302 331 Weight (g) 364 344 322 300 405 2 Weeks ELS Sample Number 1603035-011 1603035-012 1603035-013 1603035-014 1603035-015 Sample Name RCCTrout_2Weeks 1 RCCTrout_2Weeks 2 RCCTrout_2Weeks 3 RCCTrout_2Weeks 4 RCCTrout_2Weeks 5 Sample Date 4/18/2016 4/18/2016 4/18/2016 4/18/2016 4/18/2016 Elapsed Time (days) 14 14 14 14 14 Length (mm) 337 332 335 301 325 Weight (g) 467 391 417 280 413 6 Weeks ELS Sample Number 1605016-002 1605016-003 1605016-004 1605016-005 1605016-006 Sample Name RCCTrout_6Weeks 1 RCCTrout_6Weeks 2 RCCTrout_6Weeks 3 RCCTrout_6Weeks 4 RCCTrout_6Weeks 5 Sample Date 5/16/2016 5/16/2016 5/16/2016 5/16/2016 5/16/2016 Elapsed Time (days) 42 42 42 42 42 Length (mm) 320 318 312 335 331 Weight (g) 318 319 325 387 349 12 Weeks ELS Sample Number 1606036-002 1606036-003 1606036-004 1606036-005 1606036-006 Sample Name RCCTrout_12Weeks 1 RCCTrout_12Weeks 2 RCCTrout_12Weeks 3 RCCTrout_12Weeks 4 RCCTrout_12Weeks 5 Sample Date 6/27/2016 6/27/2016 6/27/2016 6/27/2016 6/27/2016 Elapsed Time (days) 84 84 84 84 84 Length (mm) 302 317 308 287 304 Weight (g) 253 327 285 224 246 11

340 Mean Length, (mm) 330 320 310 300 290 Control Two Weeks Six Weeks Twelve Weeks Figure 2. Mean Lengths and 95 Percent Least Significant Difference (LSD) Intervals 430 390 Mean Weight, (g) 350 310 270 230 Control Two Weeks Six Weeks Twelve Weeks Figure 3. Mean Weights and 95 Percent Least Significant Difference (LSD) Intervals 12

Table 3 summarizes stomach contents of the individual fish. Although the fish at 2 weeks showed some evidence of feeding, the majority of fish at 6 weeks and 12 weeks showed empty stomachs or low quality food items. Finally, lipid levels, showed a clear decline: specifically, 6.55% in the control; 6.04% after 2 weeks; 4.08% after 6 weeks; and 2.52% after 12 weeks. So, it appears the fish more-or-less maintained their lipid levels for the first 2 weeks and then began to burn those stores for energy when they were unsuccessful in obtaining food. Simple linear regression between mean weight and % lipid using the data at 2 weeks, 6 weeks and 12 weeks has a correlation coefficient (r 2 ) of 0.99. Table 3. Stomach Contents of Individual Trout Sample Fish # Stomach Contents Control 1 Not recorded Control 2 Not recorded Control 3 Not recorded Control 4 Not recorded Control 5 Not recorded 2 Weeks 1 Sticks, dipteran pupae, midge larvae, other invertebrate parts, plant matter that looked like a bunch of miniature orange bananas 2 Weeks 2 Empty 2 Weeks 3 Gravel, a bunch of midge larvae, plant matter and some woody debris 2 Weeks 4 Several midge larvae and pupae, gravel 2 Weeks 5 many midge larvae - some still alive, seeds, filamentous algae, amphipod parts, dipteran pupae, caddisfly larvae, vegetation and a leaf 6 Weeks 1 Empty 6 Weeks 2 Empty 6 Weeks 3 Empty 6 Weeks 4 Wood, gravel, aquatic insect pieces, plant pieces 6 Weeks 5 Woody debris 12 Weeks 1 Empty 12 Weeks 2 Filamentous algae and amphipod parts 12 Weeks 3 Empty 12 Weeks 4 Empty 12 Weeks 5 Empty Contaminant Concentrations in Trout Samples: Contaminants consistently detected in the trout samples are summarized in Table 4. Full laboratory results are available elsewhere (Greene, 2017a). Overall, PCBs, DDX and PAHs exhibited the greatest concentrations in the trout (range: ~4 to 42 ng/g or ppb), followed by chlordane and dieldrin (range: ~2 to 5 ppb), then 13

by toxaphene (range: ~0.6 to 1.7 ppb). Dioxin and furan TEQ concentrations ranged between ~0.01 and 0.1 pg/g or pptr. Methylmercury concentrations in the trout ranged from 18 to 27 ppb. Table 4. Concentrations of Contaminants Consistently Detected in Trout Samples Sample ID RCCTrout_Control RCCTrout_2Weeks RCCTrout_6Weeks RCCTrout_12Weeks Axys ID L24975-1 (A) L24975-2 L24975-3 L24975-4 Site Description Hatchery Control Red Clay Creek Red Clay Creek Red Clay Creek Latitude 38.808074 38.808074 38.808074 38.808074 Longitude -75.681388-75.681388-75.681388-75.681388 Sample Type Fish Tissue Fish Tissue Fish Tissue Fish Tissue Species Rainbow Trout Rainbow Trout Rainbow Trout Rainbow Trout Sample Date 4/4/2016 4/18/2016 5/16/2016 6/27/2016 Time Past Stocking (weeks) 0 2 6 12 Time Past Stocking (days) 0 14 42 84 UNITS ng/g (wet ) ng/g (wet ) ng/g (wet ) ng/g (wet ) Total PCB 8.12 11.97 19.51 42.11 Dioxin & Furan TEQs 0.0000142 0.0000183 0.00003 0.000101 Total PAH (sum of 34) 9.36 18.877 14.816 13.011 Benzo[a]Pyrene TEQs 0.006935 0.018962 0.011097 0.013665 Total DDT & Metabolites 4.041 7.164 15.118 42.442 Total Chlordane + Nonachlor 3.045 3.39 3.643 5.001 Total Hexachlorocyclohexane 0.167 0.147 0.062 0.023 Aldrin 0.021 0.02 0.016 0.011 Dieldrin 2.36 3.57 5.03 5.76 Endrin 0.063 0.073 0.074 0.047 Endrin Ketone <0.011 0.035 0.038 0.018 alpha-endosulphan 0.019 0.024 0.028 0.023 beta-endosulphan 0.029 0.036 0.033 0.038 Endosulphan Sulphate 0.015 0.035 0.037 0.027 Heptachlor 0.007 0.009 0.014 0.016 Heptachlor Epoxide 0.673 1.08 1.44 1.39 Hexachlorobenzene 0.343 0.348 0.322 0.302 Methoxychlor <0.021 0.182 0.271 0.174 Mirex 0.014 0.015 0.014 0.015 Technical Toxaphene 1.21 1.59 1.66 0.634 Methyl Mercury 18 27 22 24 % Lipid 6.55 6.04 4.08 2.52 % Moisture 73.9 75.2 77 78.7 Contaminant concentrations which exceeded Delaware s fish tissue screening values (SVs) are highlighted in Table 4. Importantly, all contaminant levels in the control fish (directly from the hatchery) were less than Delaware s SVs (DNREC and DHSS 2017). Few other results 14

exceeded the SVs. The concentration of PCB in the trout recaptured after 12 weeks exceeded Delaware s fish tissue screening value of 27 ng/g (ppb). The concentration of dieldrin in the trout recaptured after 2 weeks, 6 weeks and 12 weeks exceeded Delaware s fish tissue screening value of 3 ng/g (ppb). As an aside, the methylmercury concentrations in the trout were all well below Delaware s standard of 300 ng/g (ppb). It s important to note that Delaware s fish tissue screening values assume regular, long-term exposure to contaminants in fish, not shorter-term exposures that might occur as a result of occasionally eating stocked trout over a few months per year. The trout stocked are part of a put and take fishery, and harvesting the fish is encouraged. Catch and release rates determined for Delaware trout streams were as low as 3% (Stangl 1994) indicating most people keep what they catch. An assessment of exposure and health risk associated with consuming trout from the Red Clay Creek will be addressed later in this report. It is nevertheless useful to compare the results for the trout to Delaware s SVs to provide some context. PCBs, dioxin and furan TEQs, DDX, chlordane, dieldrin, and heptachlor concentrations in the trout increased over the course of the study. PCBs, dioxin and furan TEQs, DDX, and chlordane in the trout increased exponentially over the 12 week study with r 2 values ranging between 0.97 and 0.99. When the final data point (at week 12) is removed from the regression, the increase in concentration in PCBs, dioxin and furan TEQs, and DDX in the trout is essentially linear with r 2 values between 0.99 and 1. The exponential regressions over the full study versus the linear regressions over the first half of the study (i.e., time zero to 6 weeks) are shown side-by-side in Figure 4 for PCBs, dioxin and furan TEQs, DDX, and chlordane. The nature of the increase for dieldrin and heptachlor in the trout exhibits a different pattern. For these compounds, there s a gradual initial increase followed by a leveling out over time. These data were fit to a more complex bioaccumulation model which accounts for initial uptake followed by approach to a steady-state concentration (Greene 2011). Although this nonlinear model is well suited for dieldrin and heptachlor when viewed over the full 12 week study (with r 2 values of 1 and 0.98, respectively), simple linear regression over the first half of the study is nearly as compelling (with r 2 of 0.97 and 0.99, respectively). The nonlinear model over the full study versus the linear model over the first half of the study is shown side-by-side in Figure 5 for dieldrin and heptachlor. 15

Figure 4. PCB, Dioxin and Furan TEQ, DDX and Chlordane in Trout Samples. Left column exponential increase over full study; right column linear increase over 42 days. 16

The fact that dieldrin and heptachlor in the trout showed signs of leveling off suggests that these contaminants were approaching a steady-state concentration. This idea is supported by similar concentrations of dieldrin in resident fish and the trout. White sucker collected from the Red Clay Creek on 10/14/2015 at the same location as the trout study had a dieldrin concentration of 4.12 ppb and a lipid content of 1.19%. The Rainbow Trout recaptured after 12 weeks (on 6/27/2016) had a dieldrin concentration of 5.76 ppb and a lipid content of 2.52%. These concentrations are similar with a relative percent difference of 33.2%. The fact that the trout had a slightly higher concentration of dieldrin than the resident species may be due to the higher lipid content of hatchery fed trout relative to the wild resident fish. Figure 5. Dieldrin and Heptachlor in Trout Samples. Left column gradual approach to steadystate over full study; right column linear increase over 42 days. 17

In contrast to dieldrin, the contaminants PCBs, DDX, and chlordane did not approach or attain steady-state concentrations in the stocked trout. Two lines of evidence demonstrate this point. First, the uptake curves for these compounds continue to increase, showing no sign of leveling off. Second, the final peak concentrations of these compounds (at 12 weeks) did not get close to attaining concentrations seen in resident fish at the study site. Resident fish at the site occupy the location over much longer durations and contaminant levels in those fish provide a direct measure of expected steady-state. For instance, the PCB concentration in the white sucker sample discussed in the previous paragraph was 149 ppb, while the peak PCB concentration in the trout after 12 weeks was only 42.1 ppb. So the peak concentration of PCB attained in the trout was only 28.2% of that in the resident white sucker. Similarly, the DDX concentration in the resident white sucker collected from the site in the fall of 2015 was 258 ppb, while the maximum in the trout after 12 weeks (on 6/27/2016) was only 42.2 ppb. Hence, the trout only acquired 16.5% of the DDX concentration observed in the resident white sucker. The observation that dieldrin in the trout approached a concentration similar to that in resident fish but that PCBs and DDX did not is interesting and important. Here we postulate that differences in hydrophobicity (indicated by the octanol-water partition coefficient) affect the relative ease or efficiency at which a chemical can be taken up by the fish. Although all of these compounds are considered hydrophobic, dieldrin is less hydrophobic than typical PCB mixtures. Furthermore, typical PCB mixtures are less hydrophobic than typical DDX mixtures. Using the ratio of concentration in the trout at 12 weeks to the concentration in resident fish as an approximate measure of percent of steady-state in the trout, and plotting those percentages versus octanol-water partition coefficient (log K ow ) for dieldrin, PCB and DDX, we get the a straight line with an r 2 of essentially 1 (see Figure 6). Here we assumed that dieldrin had essentially reached steady-state (100%). The log K ow for dieldrin (5.20) was taken from EPA s EPI Suite (EPA 2017) and is based on the experimental value reported by De Bruijn et al. 1989. Log K ow for PCB (6.21) is a weighted average based on the homolog distribution in the trout at 12 weeks and the mean log K ow values reported by Mackay et al. 1992 for the various PCB homologs. Log K ow for DDX (6.4) is a weighted average based on component contributions in the trout at 12 weeks and the log K ow values appearing in EPA s EPI Suite (EPA 2017). Although we are admittedly dealing with a small sample size and have made several assumptions, it would appear 18

that the rate of uptake in the trout is related to the hydrophobicity of the chemical. This isn t altogether unexpected (Arnot and Gobas 2004), but at the same time, having field data which appears to demonstrates the point is rare. 100 % of Steady-State 80 60 40 20 y = -70.1x + 464.4 r² = 0.9997 0 5 5.5 6 6.5 7 log Kow Figure 6. Percent of Steady-State Concentration in Trout versus Octanol-Water Partition Coefficient for PCB, DDX and Dieldrin. Upper left to lower right order is Dieldrin, PCB then DDX. Contaminant Concentrations in Filtered Water Samples: Contaminants consistently detected in the filtered water samples or otherwise of interest based on their detection in the trout are presented in Table 5. Full laboratory results for the water samples, including quality assurance results, are available elsewhere (Greene 2017b). Of the contaminants which increased over the study, concentrations were greatest for DDX and PCBs (which were similar), followed by dieldrin, then chlordane. This was the same general ordering as observed in the trout. While the trout exhibited exponential increases for several contaminants over the study, this was not the case for the water samples. Increases in the water, when they occurred, were linear and generally ranged within a factor of 2 (see Figure 7: r 2 = 0.98 for DDX and 0.87 for PCBs). Interestingly, the increase in DDX was strongly correlated with a decrease in streamflow (Figure 8: r 2 = 0.99). An increase in DDX concentration accompanied by a decrease in 19

streamflow is of interest and provides an important clue regarding possible source(s), transport pathways and fate of DDX in this watershed. Table 5. Concentrations of Contaminants Consistently Detected in Water Samples Sample ID RCCWater_Control RCCWater_2Weeks RCCWater_6Weeks RCCWater_12Weeks Axys ID L24853-1 L24935-1 L25105-1 L25350-1 Site Description Control Red Clay Creek Red Clay Creek Red Clay Creek Latitude 38.808074 38.808074 38.808074 38.808074 Longitude -75.681388-75.681388-75.681388-75.681388 Sample Type Filtered Water Filtered Water Filtered Water Filtered Water Sample Date 4/4/2016 4/18/2016 5/16/2016 6/27/2016 Time Past Stocking (weeks) 0 2 6 12 Time Past Stocking (days) 0 14 42 84 Daily Ave. Flow (cfs) 38 32 26 15 UNITS pg/l pg/l pg/l pg/l Total PCB 904 842 1130 1290 Total DDT & Metabolites 879 1081 1248 1742 Total Chlordane + Nonachlor 100 151 130 236 Total Hexachlorocyclohexane 62 64 49 71 Aldrin <14 <15 <15 <15 Dieldrin 417 428 458 597 Endrin <35 <37 56 <37 Endrin Ketone 58 38 <47 55 alpha-endosulphan 77 54 111 86 beta-endosulphan 119 117 171 62 Endosulphan Sulphate <35 <37 45 98 Heptachlor <14 <15 <15 <15 Heptachlor Epoxide 139 192 174 168 Hexachlorobenzene 13 24 15 51 Methoxychlor 152 108 117 116 Mirex <14 <15 <15 <15 Technical Toxaphene <260 <510 <360 <340 For completeness, Appendix B presents summary statistics for contaminants consistently detected in the water samples, first for those which trended upward during the study, then for those with no apparent trend over the study. Note that heptachlor in the water samples was consistently below the laboratory detection limit, despite being detected in all of the trout samples. Again, dioxins and furans were not tested in the water samples. 20

Figure 7. PCB, DDX, Chlordane and Dieldrin in Filtered Water Samples 2.4 2.0 Total DDX, (ng/l) 1.6 1.2 0.8 0.4 0.0 y = -0.037x + 2.275 r² = 0.99 0 10 20 30 40 50 Daily Ave Flow, (cfs) Figure 8. DDX in Water Samples versus Streamflow 21

Chemical Fingerprints: Chemical fingerprints were prepared for PCB homologs, DDX isomers, and chlordane components. Figure 9 includes PCB homolog fingerprints for the water samples and the trout samples. Note that the fingerprints for the water samples are uniform across the duration of the study with a dominant H4 (tetrachlorobiphenyl) contribution. Correlation analysis confirms that the fingerprints in the water samples are nearly identical to each with correlation coefficients (r values) ranging from 0.998 to 0.999. Further, the fingerprints in the water are highly similar to Aroclor 1248 with correlation coefficients ranging between 0.991 and 0.996. Homolog fingerprints for Aroclors were taken from Greene (2012). Aroclor 1248 was used as hydraulic fluid and in synthetic resins, among other uses (ATSDR 2000). Figure 9. PCB Homolog Fingerprints in Water Samples (left) and Trout Samples (right). H in the x-axis refers to Homolog and the number after H refers to the number of chlorines associated with each homolog group. For instance, H4 refers to the tetrachlorobiphenyl homolog group. The PCB homolog fingerprints in the trout were more variable over the study. Close examination shows that there was a gradual shift from a higher chlorinated homolog mixture to a lower chlorinated homolog mixture as the study progressed. Figure 10 shows a comparison between the PCB homolog fingerprint in hatchery trout before being placed in the Red Clay and the PCB homolog fingerprint in a hypothetical 50:50 mixture of Aroclor 1254 and Aroclor 1260. The correlation coefficient between the two fingerprints in Figure 10 is 0.974. Aroclor 1254 and Aroclor 1260 are both highly chlorinated PCB mixtures that were widely used in the past in electrical equipment, hydraulic fluids and other applications (ATSDR 2000). 22

0.6 0.5 Hatchery Trout 50:50 Aroclor 1254/1260 Decimal Fraction 0.4 0.3 0.2 0.1 0.0 H1 H2 H3 H4 H5 H6 H7 H8 H9 H10 Figure 10. PCB Homolog Fingerprint of Hatchery Trout versus 50:50 Mixture of Aroclor 1254 and 1260 Importantly, by the end of the study (t = 12 weeks), the homolog pattern in the trout had transformed to resemble the homolog pattern in the water (r = 0.973). Again, the pattern in the water remained essentially constant over the study, while the pattern in the trout gradually shifted to acquire the pattern in the surrounding water. Since the pattern in the water resembled Aroclor 1248, this suggests that the pattern in the trout by the end of the study also resembled Aroclor 1248. This is born out based on a correlation coefficient of 0.957 between the homolog pattern in the trout at 12 weeks and Aroclor 1248. Figure 11 includes a comparison between the homolog pattern in the water, the trout at 12 weeks and Aroclor 1248. Even though the PCB homolog pattern in the trout fully acquired the pattern in the water over the course of the study, recall from the previous discussion that the total PCB concentration accumulated in the trout was only a fraction of the projected steady-state concentration inferred from PCB levels in resident fish at the location. Hence, the rate of uptake is as important as what gets taken up. 23

0.6 Decimal Fraction 0.5 0.4 0.3 0.2 0.1 Water @ 12 weeks Trout @ 12 weeks Aroclor 1248 0.0 H1 H2 H3 H4 H5 H6 H7 H8 H9 H10 Figure 11. PCB Homolog Fingerprints in the Water and Trout Samples at Twelve (12) Weeks versus Aroclor 1248 With regard to DDX, the percent contribution of DDE, DDD and DDT remained similar in the water samples over the course of the study, with 4,4 -DDD contributing the most (34.8% - 43.7%), followed by 4,4 -DDE (25.7% - 28.6%), then 4,4 -DDT (20.9% - 28.1%). For the trout, 4,4 -DDE dominated (38.4% to 59.1%, with the peak in the control sample), followed by 4,4 - DDD (21.6% - 32.2%), and then 4,4 -DDT (8.6% - 17.3%, with the minimum value in the control). See Figure 12 for stacked bar charts including DDT, DDD and DDE on an absolute and percent contribution basis in both the water and trout samples. Correlation coefficients for DDX fingerprints in the water ranged between 0.952 and 0.999. Correlation coefficients for DDX fingerprints in the trout ranged from 0.879 and 0.997. Relative to the hatchery control, there was a nominal increase in the percentage of DDT and DDD in the trout with an attendant decrease in the percentage of DDE. Based on these findings, when stocked trout are placed in RCC, they are exposed to higher overall levels of DDX and those higher levels included a greater percentage of DDT and DDD than what they experienced at the hatchery. 24

Figure 12. DDT, DDD and DDE in Water and Trout Samples. Water samples on left; trout samples on right. Absolute concentration on top; percent contribution on bottom. Although the DDX fingerprints in the water samples were similar to each other, and the DDX fingerprints in the trout samples were similar to each other, there were only modest similarities between the DDX fingerprints in the water and trout samples. The greatest similarity between the water and trout was for the samples collected at 6 weeks, which had a correlation coefficient of 0.940. The lowest correlation across between the water and trout (0.560) was between the trout control sample and the water sample at 12 weeks. None of the fingerprints in the water or trout were similar to technical grade DDT (WHO 1989). For most fish and water samples in Delaware, DDE predominates with a lesser contribution from DDD and still less or no measurable contribution from parent DDT. This is consistent with several other datasets which indicates advanced weathering of parent DDT after roughly 40 years of being taken off the market. The higher contribution of DDT and DDD in the RCC is atypical and suggests slower than normal breakdown of parent DDT in the watershed. 25

Alternatively or in addition, this observation could indicate daylighting of deeper soils (including mushroom compost) that may be more contaminated with DDT than current surface soils (USFWS 1993), followed by contemporary remobilization during earth works. Finally with regard to chemical fingerprints, we considered chlordane. The primary components of chlordane detected in both the water and trout samples were cis-chlordane, transchlordane, and trans-nonachlor. See Figure 13 for stacked bar charts on an absolute and percent contribution basis in both the water and trout samples. Although the water data were more variable than the trout samples, the contribution of the 3 components just mentioned typically comprised between two-thirds (66.6%) to three-quarters (75%) of total chlordane in both the water and trout. Figure 13. Chlordane Components in Water and Trout Samples. Water samples on left; trout samples on right. Absolute concentration on top; percent contribution on bottom. Correlation coefficients between the water sample chlordane fingerprints ranged between 0.929 and 0.995. Correlation coefficients between the trout sample chlordane fingerprints 26