If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Aquatic environments receive inputs of organic pollutants from current and historical industrial sources. Although polychlorinated biphenyls (PCBs) have been banned for several decades, their unique physiochemical properties make them persistent and soluble in fats and, therefore, capable of accumulating in biological tissues 1, 2. Remedial action at contaminated sites is thus necessary to mitigate risks to entire ecosystems and to human health 3, 4. Recommended cleanup activities typically involve the removal of contaminated soils and sediments to reduce or eliminate exposure pathways. Broadly referred to as “source control,” these efforts are aimed at limiting PCB entry into environmental systems and must be completed before additional cleanup steps can be effective. Although PCBs generally accumulate in sediments, it is the dissolved form that is taken up by microbes and plants at the lowest levels of aquatic food webs 5. Therefore, quantifying the dissolved fraction of environmental pollutants is critical for assessing source control and for understanding potential biological impacts.
Passive sampling devices (e.g., low-density polyethylene [PE], semipermeable membrane device, and solid-phase microextraction) have been widely used to assess the bioavailability of hydrophobic contaminants 6–9. Their operation is based on the diffusion of analyte molecules from the sampled medium to a receiving phase in the sampling device 10. The accumulation of analytes in passive samplers continues until equilibrium is established in the system or until the sampling period is completed. Passive samplers are extensively used in environmental assessments because these methods provide more accurate measures of dissolved PCB concentrations than do biota and sediments where biotransformation can occur 8, 11, 12 and because they accumulate a range of hydrophobic organic contaminants, including polycyclic aromatic hydrocarbons and PCBs 6, 7.
Many organic contaminants (e.g., agrochemicals, pharmaceuticals, and organochlorines) are present in the environment as enantiomers that are similar in structure but not superimposable due to restricted rotation 13. Enantioselective analysis of chiral compounds is a proven tool for source apportionment and chemical characterization 14, 15. Chiral signatures can provide insight into the biotransformation of contaminants or the source of contaminants in aquatic food webs because biological processes or a weathered contribution will change enantiomer compositions 16, 17.
Increasingly, PCB congener-specific and chiral analyses in passive samplers have been used to assess the sources of PCB in water. For instance, congener profiles captured by passive samplers were used to distinguish between atmospheric deposition and in situ sediment remobilization as sources of dissolved PCBs 18–20. Recent studies have also employed chiral analysis to provide insight into the sources of PCB in the water column 21, 22. Because their usage ended in 1977, PCBs detected in the study area are not likely from active discharge. Alternatively, current sources of PCB are probably associated with sediment remobilization. In the present study, we applied passive sampling methods and three analytical tools (total concentrations, congener profile, and chiral signatures) to investigate the potential for ongoing PCB sources at a small stream that is part of the Sangamo-Weston Superfund site. The site was contaminated with PCBs released from a former capacitor manufacturing plant in Pickens, South Carolina, USA. Levels of PCB have continued to be high at all sites and trophic levels, and the highest concentrations of PCBs were consistently observed in surficial sediment and biota collected below the discharge point of the plant 23–25. Sediment remobilization alone cannot explain this pattern of high contamination levels near the original release point, and we hypothesize that there were ongoing sources of dissolved PCBs from the plant site entering the stream through other means (e.g., groundwater inputs). If PCBs are still leaking from the site, we expect that (1) PCB concentrations in the dissolved phase near these sources are higher than upstream background levels, (2) the congener pattern of dissolved PCBs at the plant site will resemble the congener distribution in the fresh or historical discharge source, and (3) the chiral signatures of dissolved PCBs should indicate a racemic mixture near the sources and deviate from racemic with distance from the source due to biotransformation processes associated with cycling of PCBs within the stream.
MATERIALS AND METHODS
The study area was in Town Creek (Pickens County, South Carolina, USA), a tributary of the Twelvemile Creek/Lake Hartwell Superfund site. The former Sangamo-Weston company (S-W) operated a capacitor manufacturing plant from 1955 to 1978 in Pickens County, South Carolina 26. During its operation, approximately 200 metric tons of PCBs were discharged directly into Town Creek 26. Passive samplers were deployed from December 15, 2010, to February 15, 2011, at eight sites located about 79 m to 3,000 m downstream from the S-W site (Fig. 1). Site distances and coordinates are provided in Supplemental Data, Table S1. The former discharge point of the S-W site is located approximately 40 m downstream of site D1. Also, a wastewater-treatment plant is located about 1 km upstream of sites D6 through D8. The plant was built to treat nonprocess waste from the S-W plant, in closeout phase under the Bureau of Water of the South Carolina Department of Health and Environmental Control in the early 2000s (C. Zeller, U.S. Environmental Protection Agency Region 4, Atlanta, GA, personal communication). Town Creek is a third-order stream with widths of 3 to 10 m, depths that range from 0.3 to 1 m, and sediment composed primarily of sand, which has low total organic carbon content, ranging from 0.1 to 3.6% 26. The flow and depth were below normal due to several years of below-normal rainfall.
Passive sampler preparation
We used low-density polyethylene (LDPE) membrane as the in situ sampler because it is inexpensive; quickly reaches equilibrium, which reduces the effect of biofouling; and is durable under extreme environmental conditions 8, 27, 28. The LDPE tubing (film thickness of 50 µm) was purchased from Brentwood Plastics. We prepared single-layer sheets (5 cm long, 2.5 cm wide) by cutting along the edges of the tubing and precleaned the PEs by sequentially soaking in dichloromethane for 24 h, in hexane for 24 h, and in methanol for 24 h 29.
Deployment devices for PE were constructed using a stainless steel dipping basket (10 × 6 × 6 in). Sheets of PE were woven on copper wires and then attached to mesh openings of the basket. Two metal posts were driven into the streambed and positioned perpendicular to the current. Each basket containing three PE sheets was suspended approximately 20 cm below the water surface and secured to the metal posts via copper wires. Undeployed PEs served as blanks for analysis of contamination.
Sediment sampling method
Sediment was collected from sites D6 to D8 (Fig. 1) as high concentrations were observed at these sites 23. Approximately 50 g of surficial sediments (5-cm layers) were sampled from multiple depositional areas using a large metal spoon. Samples were composited from these areas, and three replicate samples were prepared. Samples were stored in amber jars and transported to the laboratory in a cooler with ice for PCB extraction and analysis.
To examine the relative equilibration, we conducted a preliminary study by deploying PEs for different time courses and quantifying uptake of different congeners in the samplers. Results indicated that 60 days were enough for uptake of low-weight congeners (but not for heavy-weight congeners) to approach equilibrium (Dang, unpublished data). Therefore, we retrieved PE samplers after 60 days in the present study and gently removed (using tap water) any particulate matter and biofilm on PEs because these could act as sites for accumulation and/or biodegradation of PCBs. Extraction of PCB from PEs was accomplished via dialysis (2 × 24 h) in 20 ml dichloromethane. One hundred microliters of surrogate standards containing two non-Aroclor congeners (PCBs 14 and 169) at 2 mg/L were spiked into the PEs prior to extraction. The combined extracts were then eluted through a drying column made of 10 g baked sodium sulfate. The final extracts were solvent-exchanged in iso-octane and concentrated to 2 ml for gas chromatographic (GC) analysis. Extraction of PCBs from sediment was conducted with an Accelerated Solvent Extractor (Dionex, ASE-200) using 1:1 hexane:acetone solvent 30. Approximately 15 g of air-dried sediment was homogenized with 5 g of baked Na2SO4. The mixture was spiked with surrogate standards and loaded into sample cells for extraction. Extracts were cleaned up on a drying column, followed by an alumina column, and solvent-exchanged in iso-octane.
Methods of achiral analysis were adapted and modified from previous work 22 (V.D. Dang, 2007, Master's thesis, Clemson University, Clemson, SC, USA). Briefly, congener-specific analysis was conducted with an HP 6890-GC equiped with an RTX-5 column (Restek; 60 m length, 0.25 mm diameter, and 0.25 µm film thickness) and a 63Ni electron capture detector (ECD). Helium (99.99% high purity) and nitrogen (99.95% high purity) were employed as carrier and makeup gas, respectively. The GC conditions and quantification methods are described in Supplemental Data.
Chiral signatures or enantiomeric fractions (EFs) were analyzed on an Agilent 6850 GC-ECD and a 30-m Chirasil-Dex column according to the methods described by Wong et al. 31 and A.A. Hall (2004, Master's thesis, Clemson University, Clemson, SC USA) and modified by Dang et al. 22. For EF determination, two separate standard solutions were collectively prepared using neat congeners and a 1:1:1 mixture of Aroclors 1016, 1254, and 1260 dissolved in iso-octane, respectively. Both solutions were analyzed on the GC-ECD using a 30-m Chirasil-Dex column, and no interference with three chiral PCBs (91, 95, and 149) was present. Interferences were defined as a coelution of the enantiomers with another homologous congener. The elution order of enantiomers was obtained from Wong et al. 32. A check standard was included in every batch of six samples. If its optical rotation was known, the EF value was calculated by the peak area of the (+) enantiomer divided by the sum of the peak area of the (+) and (−) enantiomers. Otherwise, the EF was defined by the peak area of the first eluting enantiomer divided by the total peak areas of the first and second eluting enantiomers. Therefore, the EF for standards ranged between 0.5 and 5% relative standard deviation for all target chiral congeners.
The dialysis extraction method achieved a recovery of 80.57% (±8.60% standard deviation) and 87.02% (±18.31% standard deviation) for PCBs 14 and 169, respectively. Meanwhile, recovery of surrogate standards averaged 75% (±5.20% standard deviation) in sediment samples. Concentrations of PCB measured in the PEs were weight-normalized and not corrected by the recovery results. Because passive samplers are used to measure the activity or fugacity of hydrophobic organic pollutants in the environment based on partitioning between the samplers and water, PE-normalized concentrations are presumably related to the dissolved concentrations. The limit of quantification for total PCBs was 72.2 ± 31.1 ng/g PE, and the concentrations in the deployed samplers were always greater than detection limits.
Among-site differences in total PCB concentrations and EF values were compared by one-way analysis of variance, with Tukey's honestly significant difference post hoc test (p < 0.05). Congener composition patterns in the PEs, the mixture of Aroclors 1254 and 1016 (4:1 v/v), and sediment were evaluated using nonmetric multidimensional scaling ordination (NMS, PC-ORD 6.0; MjM Software) 33. This is an ordination method that is applied when data are not normally distributed or have discontinuous and unequal scales. The NMS analysis was run in autopilot mode, which allowed the program to choose the best solution at each dimensionality 34. We used the Sørenson (Bray-Curtis) coefficient as the distance measure in the analysis. We ran the NMS analysis using the mean relative abundances of PCB homologues (proportion of total PCBs) for the PEs and sediment at each site. Relative abundance data were arcsine-transformed prior to analysis.
RESULTS AND DISCUSSION
Deployment and retrieval of PEs were successful except for sites D5 and D7. The samplers at site D5 were lost, and the basket containing the samplers at site D7 settled to the streambed and mixed with sediment. Results from these grounded samplers were excluded as they may not represent water column conditions. Stream flows were stable in Twelvemile Creek, the recipient stream of Town Creek, during the study, showing an increase only in early February (Supplemental Data, Fig. S1). This indicates that PE samplers remained submerged and suspended in the water column during deployment.
Total PCB concentrations
Average ∑PCB concentrations ranged from 748.8 ng/g PE at site D1 (40 m upstream of the discharge ditch) to 3,547.6 ng/g PE at site D8 (1.7 km downstream of the ditch) (Fig. 2). Total PCB concentrations increased fivefold in the downstream direction and were highest at sites D6 and D8. Sites D2, D3, and D4 had similar concentrations in the samplers, while the concentrations were not statistically different between sites D6 and D8 (p < 0.05). We observed two step increases in ∑PCBs with downstream distance. Concentrations doubled between sites D1 and D2, which fall on either side of the S-W discharge ditch. Concentrations approximately doubled again at sites D6 and D8, which are downstream of the former municipal wastewater-treatment plant. These data indicate multiple inputs into the system from the original S-W site and from the wastewater-treatment plant. Elevated PCB concentrations at site D1 (40 m upstream of the ditch) suggest there is an ongoing source migrating from the S-W to the system as they are well above the background level of 0.05 ppm measured in Corbicula fluminea deployed approximately 1 km upstream of the ditch 24. Likewise, the steep increase in concentrations at D2 indicates inputs in the vicinity of the ditch, which was historically identified as the PCB point source to Town Creek. In addition, Walters et al. 23 conducted a food-web study in Town and Twelvemile Creeks and measured the highest total PCBs near sites D6 and D8. These findings led them to hypothesize that the abandoned municipal wastewater treatment facility was a potential undocumented source of PCBs to the stream. It is possible that the facility treated PCB-contaminated waste that could reach the stream via runoff or groundwater inputs.
PCB congener patterns
Among-site differences in congener or homolog distributions also supported the hypothesis of ongoing sources to the stream system. For example, site D1 had an overwhelming fraction of di- and tri-chlorinated biphenyls (CBs), which is indicative of a recent discharge because these more volatile, lower chlorinated PCBs have not yet achieved equilibrium with the atmosphere (Supplemental Data, Fig. S2a). Although site D2 is located 79 m downstream of the discharge ditch, we did not observe di-CBs. When moving downstream of site D1, the data appeared to reflect volatilization of group 1 (e.g., di- and tri-CBs) and sorption of group 2 (e.g., hexa- and hepta-CBs) (Supplemental Data, Fig. S2b–f). A rapid loss of di-CBs from the water column as reflected in the PEs appeared to occur within a 100-m distance (site D1 to D2). This trend is inconsistent with steady-state models, which documented a gradual loss of low-weight congeners in sediment over a 10-km distance 35. The lack of di-CBs downstream of and near the discharge ditch in the present study is thus difficult to explain via the mechanism of volatilization. Alternatively, di-CBs could be susceptible to biodegradation to a greater extent than volatilization. Conversely, some di-CBs were again measured in PEs at sites D6 and D8 (Supplemental Data, Fig. S2e and f). Because Town Creek is shallow and well oxygenated, anaerobic reductive dechlorination of higher chlorinated congeners to di-CBs is expected to be negligible. It is more likely that the presence of di-CBs in the PEs at sites D6 and D8 is indicative of a potential fresh source (e.g., groundwater input via seepage in fractured rock or soil erosion from land) in the vicinity of the wastewater-treatment plant.
Homolog distribution at site D1 resembles the 4:1 mixture of Aroclors 1016 and 1254 released from the S-W site rather than the sediment profile, which suggests that desorption from the sediment is not the source (Fig. 3). Walters et al. 23 also observed greater accumulation of lower chlorinated congeners in different types of organic matter (e.g., periphyton) near the S-W site compared with other sites farther downstream. The NMS analysis indicated an increased proportion of higher chlorinated CBs along axes 1 and 2 for PEs from sites D2 to D8 and for sediment, respectively, but not for PEs at site D1 (Fig. 4). The congener pattern in the samplers at site D1 and in the mixture of Aroclors 1016 and 1254 (4:1 v/v) consisted of predominantly low-weight congeners (di- and tri-CBs), while the PEs from site D2 to D8 represented a consistent pattern of increasing tetra- and penta-CBs. Ordination of higher chlorinated congeners (hexa- and hepta-CBs) showed correspondence with the homolog distribution in the sediment. However, the presence of di-CBs at D6 and D8 suggests an ongoing source because the homolog distributions did not match the sediment distribution (Figs. 3 and 4). Therefore, the similarity of PCB homologues between the PEs at site D1 and Aroclors implies that a source of relatively nonweathered PCBs from the S-W site is still being released into Town Creek, while the di-CBs at sites D6 and D8 suggest inputs from the former wastewater-treatment plant.
The EF values, which are near racemic in the vicinity of fresh sources and then depart from racemic with distance from the source, can provide evidence for recent fresh sources to the systems 17. The only two chiral congeners detected in the passive samplers were PCBs 91 (2,2′,3,4′,6-pentaCB) and 95 (2,2′,3,5′,6-pentaCB). Both PCBs 91 and 95 were below the detection limit in the samplers at site D1, which is closest to the former S-W plant. For all other locations, EF values for PCB 91 were significantly nonracemic (>0.5) (Fig. 5a), while the values for PCB 95 were racemic or close to racemic (Fig. 5b). The EF values varied little with distance among sites, in contrast with concentrations that varied over fivefold among sites.
The lack of detectable chiral pentachlorobiphenyls (PCBs 91 and 95) in the samplers at site D1 suggests that the ongoing inputs enter the stream attached to particles. Perhaps, equilibrium has not been established between sorbed PCB 91 and PCB 95 and the water column at D1. Farther downstream (from D2 to D8), these congeners have desorbed enough to be detected in the samplers. The spatial consistency of nonracemic EF values for PCB 91 suggests that its biotransformation occurs rapidly in the stream or before it enters the system. It is likely that any biotransformation in Town Creek occurs under aerobic conditions because it is a shallow stream. PCB 95 does not appear to be susceptible to aerobic transformation under the conditions found in the creek given its racemic and near racemic EF values. Singer et al. 36 found that PCB 95 was less likely to undergo aerobic biotransformation than PCB 91 with pure cultures under ideal laboratory conditions. Previous work with microcosms incubated with sediment from Lake Hartwell indicated that both chiral congeners could undergo anaerobic biotransformation 37. On the other hand, similarity in EF values in the PEs among the sites does not provide any strong evidence to indicate additional inputs of PCB to the system.
Three lines of evidence including total PCBs, congener patterns (in the PEs and sediment), and chiral signatures were used to assess ongoing inputs of PCB to the water column of Town Creek. Overall, we observed that (1) the total PCB concentrations increased within a 2- to 3-km distance from the S-W plant site to near the mouth of Town Creek, (2) the pattern in congener composition in the PEs upstream of the discharge ditch was similar to the congener composition of the mixture of Aroclors 1016 and 1254 (4:1 v/v) released from the S-W plant, and (3) the EF values for chiral PCBs 91 and 95 detected downstream of the ditch were nonracemic and racemic, respectively. Thus, results from total PCBs and congener-specific analysis provide evidence to conclude that there is an ongoing source of PCBs to the system. The evidence from the chiral PCB congeners suggests that dissolved PCBs are subject to aerobic biotransformation. However, chiral evidence does not shed any light on possible ongoing inputs.
The present study is crucial because it draws attention to the question of source control, which is often neglected in ongoing monitoring efforts. The study suggests that the former wastewater-treatment plant is also a potential ongoing source of PCBs to the system. Therefore, a fine-scale survey (over tens of meters) using passive samplers to measure dissolved PCBs (total PCBs and congeners) is recommended to assess this potential input.
Figs. S1. and 2. (180 KB DOC).
Support for this work was provided by the National Science Foundation (CBET-0828699). We thank C. Zellar for reviewing this manuscript. Any use of trade, firm or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.