PCB Loading from Sediment in the Hudson River - American

the congener signature is consistent with a mixed source consisting of porewater flux and non-scour flux of contaminated sediments. The non-scour sedi...
0 downloads 0 Views 213KB Size
Environ. Sci. Technol. 2004, 38, 3232-3238

PCB Loading from Sediment in the Hudson River: Congener Signature Analysis of Pathways J O N A T H A N B . B U T C H E R * ,† A N D EDWARD A. GARVEY‡ Tetra Tech, Inc., P.O. Box 14409, Research Triangle Park, North Carolina 27709, and Earth Tech, Inc., 300 Broadacres Drive, Bloomfield, New Jersey 07003

The upper Hudson River (NY) was subjected to massive PCB contamination over a period of three decades. A large inventory of PCBs remains in contaminated sediments of the river, most notably in the Thompson Island Pool. During the summer, flow crossing the Thompson Island Pool exhibits a large and consistent PCB load gain. This load gain is not associated with scouring flows and is not accompanied by an increase in suspended solids. A variety of hypotheses have been proposed to explain this load gain, including flux of contaminated porewater and dissolution of unverified reservoirs of pure PCBs. A wealth of congenerspecific PCB data is available for the site throughout the 1990s. Interpretation of the Thompson Island Pool load gain is facilitated by examination of the PCB congener signature of the gain and comparison to the signature of potential sources. This examination suggests that neither the flux of porewater nor the dissolution of unaltered Aroclors are the predominant source of the load gain. Instead, the congener signature is consistent with a mixed source consisting of porewater flux and non-scour flux of contaminated sediments. The non-scour sediment flux, which reaches a maximum in the beginning of the summer growing season, is likely driven by a variety of biological and anthropogenic processes, including bioturbation by benthic organisms, bioturbation by demersal fish, scour by propwash, mechanical scour by boats and floating debris in nearshore areas, and uprooting of macrophytes.

Introduction The history of polychlorinated biphenyl (PCB) contamination in the Hudson River (NY) is well-documented (1-4), and the U.S. Environmental Protection Agency (EPA) has recently issued a Record of Decision calling for remediation of contaminated sediment in the river (5). Large quantities of PCBs were released into the upper river from capacitor manufacturing operations at General Electric (GE) plants at Hudson Falls and Fort Edward, NY, beginning in 1947 and continuing through 1977 (Figure 1). The PCB releases consisted primarily of the commercial PCB mixture Aroclor 1242, although Aroclor 1254 was also used at the plants prior to 1954 (4, 5). Much of the historical PCB releases to the river were stored in sediment behind a hydropower dam at Fort Edward, NY, that was removed for safety reasons in 1973. * Corresponding author phone: (919)485-8278; fax: (919)485-8280; e-mail: [email protected]. † Tetra Tech, Inc. ‡ Earth Tech, Inc. 3232

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 12, 2004

FIGURE 1. Location map, Hudson River PCBs site (NY). Removal of the dam resulted in extensive movement of contaminated sediment downstream, particularly into the pool located behind the Thompson Island Dam, the next impoundment downstream, known as the Thompson Island Pool. High concentrations of sediment PCB contamination remain within the Thompson Island Pool. Significant anaerobic dechlorination of these PCBs has occurred within the buried sediments in the pool, resulting in a shift in the original congener pattern (6). However, additional loads of unaltered Aroclor 1242 have continued from bedrock seeps at Hudson Falls, upstream of the Thompson Island Pool. Since 1991, GE, under a consent agreement, has conducted extensive (approximately biweekly) monitoring of PCBs in the water column of this reach, including monitoring at the Rt. 197 bridge in Fort Edward (upstream end of the Thompson Island Pool) and at Thompson Island Dam (the downstream end of the pool). Analyses were carried to the congener or congener group level, using capillary column gas chromatography by Northeastern Analytical Laboratory (NEA). During summer low flow conditions, a large increase in PCB concentration occurs across the Thompson Island Pool, from Rt. 197 at Fort Edward to Thompson Island Dam (Figure 2). In addition to an increase in concentration, there is a shift in PCB homologue composition across the Thompson Island Pool. The upstream load is dominated by di- through tetrachlorobiphenyls, while the downstream load is dominated by mono- through trichlorobiphenyls. These increases and shifts in PCB load and composition are consistent at flows that do not appear to induce sediment scour within the Thompson Island Pool, and little or no gain in total suspended solids load is typically seen across the pool during the summer months (7). 10.1021/es035453t CCC: $27.50

 2004 American Chemical Society Published on Web 05/12/2004

TABLE 2. Average PCB Concentrations in Surface Sediment and Porewater of the Thompson Island Pool, with Apparent Dissolved Fraction in Water Column Predicted from Equilibrium Partitioninga

FIGURE 2. Typical summer shift in PCB homologue composition across the Thompson Island Pool, 1997.

TABLE 1. NEA Analytical Peaks and Associated PCB Congeners Used in Pattern Analysis NEA peak

homologue group

congeners

peak 2 peak 5 peak 8 peak 14 peak 24 peak 23 peak 37 peak 31 peak 47 peak 48 peak 53 peak 69 peak 82 peak 75

monochlorobiphenyl dichlorobiphenyl dichlorobiphenyl di/trichlorobiphenyl tri/tetrachlorobiphenyl trichlorobiphenyl tetra/pentachlorobiphenyl tetrachlorobiphenyl tetrachlorobiphenyl penta/tetrachlorobiphenyl pentachlorobiphenyl penta/hexachlorobiphenyl hexachlorobiphenyl hexachlorobiphenyl

BZ 1 BZ 4, BZ 10 BZ 8, BZ 5 BZ 15, BZ 18 BZ 28, BZ 50 BZ 31 BZ 44, BZ 104 BZ 52, BZ 73 BZ 70, BZ 76, BZ 61 BZ 95, BZ 66, BZ 93 BZ 101, BZ 90 BZ 118, BZ 149, BZ 106 BZ 138, BZ 163 BZ 153

This paper presents an investigation of the potential sources of the Thompson Island Pool PCB load gain under non-scouring flows. The shift in homologue pattern suggests dechlorinated sediment as a source. The pattern of individual congeners within the load gain does not, however, match that observed in porewater in the Thompson Island Pool, suggesting a process in which contaminated sediment is disturbed and reequilibrated with the water column.

Materials and Methods The general approach used in this study is to examine the congener pattern in the Thompson Island Pool load gain and evaluate the implications for particulate versus porewater sediment sources compared to unaltered Aroclor 1242. Accomplishing this requires information on the following: (i) Congener concentrations and loads in the water column at both the upstream and downstream ends of the Thompson Island Pool. (ii) Congener concentrations within sediment on both particulate matter and in porewater. (iii) Information on partitioning between particulate and dissolved phases in the water column and sediment porewater. PCB load gains across the Thompson Island Pool vary from year to year, depending on hydrologic and other conditions. An informative comparison can be made by examining the relative percent composition of a set of key congeners. For this and subsequent analyses, the GE/NEA capillary column peaks and associated congeners shown in Table 1 were chosen for comparison because (1) they are environmentally significant and (ii) site-specific partition coefficient estimates are available. For each peak the congener of most environmental significance in upper Hudson River sediments is listed first.

BZ 1 BZ 4 + 10 BZ 5 + 8 BZ 15 + 18 BZ 28 BZ 31 BZ 44 BZ 52 BZ 66 + 95 BZ 70 BZ 101 + 90 BZ 118 + 149 BZ 138 BZ 153

surface porewater (ng/L)

surface sediment (µg/kg)

apparent dissolved fraction in water column

4115 4551 119 85 26 35 28 44 33 14 11 15 9 7

4326 8557 2175 1364 667 944 234 871 438 156 137 143 85 43

0.72 0.91 0.80 0.78 0.52 0.56 0.52 0.54 0.34 0.38 0.36 0.22 0.27 0.31

a Sediment and porewater concentrations were collected in 1991 (9); the apparent dissolved fraction in the water column is derived from an equilibrium partitioning analysis of 1991-1997 data (8).

Water Column Data. High-resolution, capillary column GC analyses of PCB congener concentrations in water at both the upstream (Rt. 197) and downstream (Thompson Island Dam) ends of the Thompson Island Pool are available from two sources: GE and the U.S. EPA. Contractors for the General Electric Company have sampled water column concentrations and conducted capillary column analyses for PCB congeners on a regular basis since 1991. These grab samples are generally collected once every 2 weeks. They provide a significant long-term record of PCB concentrations entering and exiting the pool, but do not necessarily always measure the same “parcel” of water entering and exiting the pool. During 1993 only, TAMS Consultants for the U.S. EPA collected intensive data from the Thompson Island Pool and other locations in the Upper and Lower Hudson in conjunction with the reassessment RI/FS for the Hudson River PCBs NPL site (4). Samples at Rt. 197 and Thompson Island Dam include both “transect” grab samples, timed to approximate the same parcel of water at the upstream and downstream ends of the pool, and flow-averaged samples, in which samples were composited on a flow-weighted basis over a 2-week period. Both the U.S. EPA and GE have primarily collected water column data from the lower end of the pool at a station the west shore above the dam, in an area of cohesive sediments with high PCB concentrations. Investigations for GE in 19971998 indicated that concentrations at this station may be biased high relative to flows exiting the Thompson Island Pool and measured at a center channel station (TIP-18C), at least under conditions of low flows and low upstream concentrations. This bias presumably reflects localized loading from contaminated sediments under conditions of low lateral mixing and must be recognized in the analysis. Sediment PCB Data. Sediment PCB data from the Thompson Island Pool are also available from both GE and U.S. EPA. In 1991, GE’s contractors collected sediment samples throughout the Thompson Island Pool. These samples were vertically separated into 5-cm segments and then composited across subareas of the pool. Particulate and porewater fractions were analyzed separately; however, some reequilibration between these fractions may have occurred during handling. In fall 1992, TAMS Consultants for the U.S. EPA collected three cohesive sediment cores from stable VOL. 38, NO. 12, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3233

FIGURE 3. Comparison of congener pattern in 1997 Thompson Island Pool downstream concentrations and gain estimated from center channel station minus upstream station. TID-West: nearshore concentration at Thompson Island Dam, west wing wall. TIP-18C: center channel concentration just above Thompson Island Dam. TIPC-gain: gain in PCB concentration across the Thompson Island Pool estimated from concentration at TIP-18C minus upstream concentration at Rt. 197. depositional areas of the pool. These cores were sectioned in 2-cm layers and cesium dated. Loading to the water column is presumably most closely related to surface sediment. The GE and U.S. EPA results for surface sediment are generally similar (see Figure S-1 in Supporting Information). A sample from the 8-12-cm layer in EPA Core 18 (HR-018-0812) shows strong evidence of dechlorination and a corresponding shift to lighter congeners. All the sediment samples appear to be significantly dechlorinated relative to unweathered Aroclor 1242. PCB Partitioning. The partitioning of PCB congeners between sorbed and dissolved phases plays an important role in constraining potential explanations of the source of PCB load gain. Equilibrium partitioning of PCB congeners in the water column was estimated from U.S. EPA 1993 data. Results are reported by Butcher et al. (8). These partition coefficients represent site-specific conditions for the freshwater Hudson River. The apparent dissolved fraction (truly dissolved plus fraction sorbed to dissolved organic carbon) of PCB congeners predicted from equilibrium partitioning (8) at typical Hudson River conditions of 4.79 mg/L dissolved organic carbon (DOC) and 1.40 mg/L particulate organic carbon, along with the average PCB concentrations in surface sediments and porewater of the Thompson Island Pool is shown in Table 2. Signature Analysis Approach. The main question addressed in this analysis is: Does the gain in PCB concentration and load across the Thompson Island Pool observed during the summer months reflect unaltered Aroclor 1242, flux of porewater, non-scour resuspension of contaminated sediment, or some combination of these sources? The question was evaluated using only in the summer (June-August) monitoring results at non-scouring flows, as this period is when the phenomenon is most evident. The signature of the load source should be evident in the pattern of a selected subset of PCB congeners (Table 2), expressed as relative percents to account for year-to-year variability in uncontrolled exogenous forcings. It was assumed that a porewater source should reflect the typical distribution of congeners found in near-surface porewater (apparent dissolved fraction). On the other hand, a non-scouring sediment resuspension source (without concurrent gain in suspended solids 3234

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 12, 2004

concentration) should reflect the congener distribution sorbed to surface sediment particulate matter and adjusted to reflect repartitioning in the water column and resettling of the solids. The analysis was first conducted for 1997 observations and then extended to the 1991-1999 record. Data from 1997 provides an ideal starting point for investigating load generation from the sediment, as upstream (Rt. 197) concentrations were generally low. In addition, sampling at both the Thompson Island Dam-West (TID-West) and center channel station (TIP-18C) in this year provide an opportunity to examine the impacts of lateral bias in the downstream sampling.

Results Evaluation of Lateral Bias at the Downstream Station. GE proposed that the historical sampling at the TID-West station overestimated the concentration of PCBs exported from the Thompson Island Pool; however, the identification of these sediments as a source of load is not in doubt (5). In 1997, concentrations do tend to be lower at the center channel station (TIP-18C) than at the nearshore TID-West station. When examined on a relative percent basis (fraction of analyzed congeners), the congener patterns at the two stations are nearly identical (Figure 3). The gain pattern from Fort Edward to Thompson Island Dam is also nearly identical in pattern to downstream loads observed at Schuylerville, as upstream load is very low in this periodsbut the pattern is very different from unaltered Aroclor 1242. These observations suggest two conclusions. First, the PCB gain across the Thompson Island Pool is dissimilar to unaltered Aroclor 1242, but instead reflects an extensively dechlorinated source. Second, the evaluation of relative congener percents appears to remove any impact of lateral sampling bias on the analysis of patterns at the downstream end of the pool. Pure Porewater Source. The first hypothesis investigated was whether the summer PCB load gain could be attributed entirely to a porewater flux (advective or diffusive) source from contaminated sediment, as was suggested in early analyses of the site. This was evaluated in two ways: by direct comparison of the load gain to porewater concentrations reported by GE in 1991 and by comparison of the sediment

FIGURE 4. Relative congener percent patterns in water column gain at TIP-18C and surface sediment porewater, 1997. pattern necessary to support the observed gain as a porewater flux to the actual sediment pattern. At first glance, the relative concentration gain measured at TIP-18C near the Thompson Island Dam looks similar to the relative concentrations in surface sediment porewater measured by GE in 1991 (Figure 4). The apparent agreement is, however, largely due to the fact that both patterns are dominated by BZ 4 + 10. For other congeners there is much less agreement, as there is a substantially higher proportion of BZ 1 in porewater than in surface water, while the more highly chlorinated congeners have a relative 21% in the TIP18C gain, but only 5% in porewater. Furthermore, the tetraand higher chlorinated congeners show a pattern that looks more like sediment than porewater. The difference between porewater and surface water concentration patterns does not appear to be attributable to volatilization, as discussed in the Supporting Information. The plausibility of a pure porewater source can also be evaluated by evaluating the sediment concentration that would be necessary to support such a source. Partitioning of PCBs into porewater is best described as a three-phase phenomenon, accounting for partitioning to dissolved organic carbon (DOC) as well as partitioning to solids and a dissolved phase. The truly dissolved phase and the DOCsorbed phase are both subject to porewater flux. Evaluation of this hypothesis is possible because the porewater and particulate concentrations in sediment can be predicted for equilibrium partitioning assumptions, using the equation:

CP )

fOCKOCCPW,a (1 + mDOCKDOC)

where fOC is the fraction of organic carbon in the solid phase; KOC is the partition coefficient to organic carbon; mDOC is the mass of dissolved organic carbon (DOC) per volume of porewater; KDOC is the partition coefficient to DOC; CP is the particulate concentration, and CPW,a is the apparent dissolved concentration (dissolved plus DOC-sorbed). Physical characteristics of the sediment are assumed equal to the average from 0-5-cm core sections within the Thompson Island Pool in the 1991 GE sediment data (9), with fOC ) 0.01788 and mDOC ) 33.68 mg/L. Site-specific three-phase partition coefficients for the freshwater portion of the Hudson River obtained via numeric optimization on observed sediment and porewater concentrations yield the estimates shown in Table 3.

TABLE 3. Three-Phase Partition Coefficient Estimates for PCBs in Sediment in the Freshwater Portion of the Hudson River PCB congeners (BZ )

log KOC (L/kg)

log KDOC (L/kg)

1 4 + 10 5+8 15 + 18 28 + 50 31 44 + 104 52 + 73 66 + 93 + 95 61 + 70 + 76 101 + 90 118 + 149 + 106 138 + 163 153

4.46 4.69 5.70 5.89 6.53 6.14 6.08 5.95 6.08 6.01 5.85 6.05 6.29 6.13

3.63 3.20 3.61 4.07 4.44 4.27 4.70 4.24 4.52 4.11 4.46 4.85 5.10 5.07

Application of the estimated sediment partition coefficients and average physical characteristics for Thompson Island Pool surface sediment yields a derived estimate of the sediment congener pattern that would be necessary to support the hypothesized porewater flux (see Figure S-2 in Supporting Information). Results are similar whether the analysis is based on nearshore or center channel gain. The computed sediment pattern, however, appears quite different from that seen in the 0-2-cm layer of U.S. EPA cohesive sediment cores (Figure 5), and the difference is even greater when compared to the 0-5-cm layer of GE cores. BZ 28 and 52 are elevated in the calculated source relative to observed surface sediment, while BZ 1, 4, and 10 are depressed. The pattern also does not match raw Aroclor 1242. Non-Scour Sediment Source. During typical summer conditions there is insufficient shear stress at the sedimentwater interface to scour significant quantities of PCBcontaminated sediment, as shown by the lack of gain in total suspended solids concentration across the Thompson Island Pool. The congener pattern of concentration gain in the Thompson Island Pool does not resemble the congener pattern in surface sediments, suggesting that direct resuspension of unaltered sediment particulate material is not a major part of the summer flux. But, there is an alternative to porewater flux for transferring sediment particle-sorbed PCBs into the water column. This involves exchange of PCBs from bulk sediment into the water column or to the VOL. 38, NO. 12, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3235

FIGURE 5. Comparison of sediment congener pattern to support porewater flux derived from summer 1997 gain at TIP-18C to observed sediment concentrations.

FIGURE 6. Concentration gain at TID-West for 1997 predicted as a mixture of porewater and sediment exchange. sediment-water interface, followed by reequilibration in the lower concentration water column environment. In addition to apparent differences in partition coefficients between the water column and sediment, the concentrations of particulate and dissolved organic carbon are much lower in the water column than in the sediment. If sediment is mixed to the sediment surface or suspended into the water column long enough to reequilibrate, the resulting dissolved and DOCsorbed fractions will remain in suspension (on average following the fractions shown in Table 2), while the remaining particle-sorbed component may settle back out. Because sediment particulate matter is much more contaminated than particulate matter in the water column, equilibrium desorption would approximate the fractionation predicted from equilibrium partitioning in the water column. This fractionation process (at equilibrium) would result in 91% of sediment BZ 4 remaining in the water column, but only 22% of BZ 118 at typical summer conditions in the Thompson Island Pool. Possible mechanisms for mixing of bulk sediment to the sediment-water interface or into the water column in the absence of hydrodynamic scour include bioturbation by benthic organisms such as tubificid oligochaetes (10); bioturbation by demersal fish; scour by propwash; mechanical 3236

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 12, 2004

scour by boats and floating debris in nearshore areas; and uprooting of macrophytes by wind, flow, or biological activity. Mixed Sediment-Porewater Source. The congener pattern observed in the water column is consistent with a source partially composed of PCBs on bulk sediment rather than PCBs partitioned from sediment into porewater. To test the reasonableness of this theory, numerical experiments were performed to reproduce observed concentrations by a weighted combination of surface sediment and surface sediment porewater concentrations. Direct combinations which would be consistent with net solids loading from TIP sediments to the water column, coupled with porewater exchangesdoes not yield a close fit to the observed congener pattern. However, a very close fit can be obtained under an assumption of sediment resuspension, exchange with the water column, and settling. When a mixed sediment-porewater source is considered (Figure 6), the congener pattern in the 1997 Thompson Island Pool concentration gain can be predicted quite closely by optimizing the ratio between the sediment and porewater sources. Most notably, the mixed source accurately reproduces the observed ratio between BZ 1 and BZ 4 + 10, whereas the best fit based on a porewater-only source cannot.

FIGURE 7. Summer relative percent gain of PCB congeners across the Thompson Island Pool for 1991-1999.

FIGURE 8. 1991-1999 PCB gain across the Thompson Island Pool predicted as a mix of porewater and surface sediment. Application to 1991-1999 Observations. The previous sections examined 1997 observations in detail and suggested that the concentration gain across the Thompson Island Pool is likely due to a combination of sediment and porewater sources. These results were confirmed through application to monitoring data for summers of 1991-1999. The years 1991 through 1999 have differing flow patterns and also large differences in upstream source strength due to extensive leakage of liquid PCBs from bedrock seeps near Hudson Falls, particularly in 1992-1993. While Total PCB concentration varied greatly, the concentration gain across the Thompson Island Pool shows a lesser degree of variability in absolute magnitude (see Figure S-3 in Supporting Information), while the relative percentage gain of congeners is highly similar from year to year (Figure 7). The remaining variability between years appears to be well within the range of sampling and analytical variability. Therefore, the multiyear series can be fit on a composite basis representing the average of yearly relative percentages. An optimized fit to a mixed porewater-sediment source provides a close match to the 1991-1999 composite estimate, as shown in Figure 8.

The predicted fraction of mass flux contributed by porewater ranges from less than 5% to nearly 50% (BZ 1). For the predominant tri- and tetrachlorobiphenyls in the Hudson River sediments, the predicted average contribution by porewater flux is less than 10% of the total mass flux. Mass Transfer Rates. Assuming concentrations in sediment are much greater than concentrations in the water column, the concentration gain for a given congener may be written in terms of mass transfer rates as

∆C )

AS [k θC + dfkS(1 - θ)FCS] Q PW PW

where ∆C is the concentration gain (M/L3); AS is the sediment source area (L2); Q is the flow (L3/T); kPW is the mass transfer rate for porewater (L/T); θ is the porosity (dimensionless); CPW is the concentration in porewater (M/L3); df is the fraction desorbing in the water column (dimensionless), assumed equal to the equilibrium partitioning estimate of the dissolved and DOC-sorbed fraction of the congener in the water VOL. 38, NO. 12, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3237

column; kS is the mass transfer rate for particulate PCB (L/ T); F is the solid particle density (M/L3); and CS is the) concentration on sediment (M/M). Contractors for both U.S. EPA and GE (11, 12) have estimated a composite mass transfer coefficient (kf) for triand higher chlorinated congeners (tri+) based on porewater concentration only. These are in the range of 10-15 cm/d during summer, which is consistent with the work of Valsaraj et al. (13) on TCDD. To convert to a sediment volume basis, this coefficient must be multiplied by porosity and porewater concentration, implying

kf ) kPW +

dfkSF(1 - θ)CS θCPW

For the application to 1991-1999 data, kPW is estimated at 1.10 cm/d, while the estimate of kS‚F is 0.933 × 10-3 cm/ d‚g/cm3. Insertion of these values into the equation for kf, with θ ) 0.386 and CS/CPWdf ) 1.5 × 103 L/kg for the complete set of tri+ congeners observed in the Thompson Island Pool, yields an estimate of 14.8 cm/d, in reasonable agreement with the total estimates for mass transfer velocities of tri+ PCBs. The sum of kPW and kS is much less than the estimate of kf because much higher concentrations are present on solids than in porewater for the tri+ components of PCBs. Summary. Analysis of congener patterns suggests that the PCB concentration gain observed across the Thompson Island Pool of the Hudson River under summer low flow conditions in the 1990s represents a mixture of porewater flux and direct exchange from bulk sediment into the water column (14). Biological processes are the most likely driver for direct exchange of sediment with the water column. Results are, however, dependent on the analysis of phase distribution of individual congeners in the water column and the sediment, both of which are subject to uncertainty. Incorporating the concepts presented here into a parametric, physically based model of PCB fate and transport would likely improve existing models of PCB cycling in the Thompson Island Pool. Models that rely on diffusive flux of porewater as the only source of PCB load from the sediment under non-scouring conditions likely overestimate the importance of noncohesive, coarser grained sediments relative to cohesive, organic sediments in supplying PCB load to the water column.

Acknowledgments The work described in this paper was supported by US EPA and USACOE as part of the Reassessment RI/FS of the Hudson

3238

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 12, 2004

River PCBs NPL site. The opinions and findings expressed in this paper are, however, those of the authors, and do not necessarily represent official Agency positions. We thank the General Electric Company and its contractors for providing access to their data and thank the anonymous reviewers for their helpful comments. Preliminary results of this work were presented at the SETAC Annual Meeting, Philadelphia, November 14-18, 1999.

Supporting Information Available Three supplementary figures and a discussion of the potential effects of volatilization on observed congener patterns. This material is available free of charge via the Internet at http:// pubs.acs.org.

Literature Cited (1) Brown, J. F.; Wagner, R. E.; Bedard, D. L.; Brennan, M. J.; Carnahan, J. C.; May, R. J. Northeastern Environ. Sci. 1984, 3, 166-178. (2) Sanders, J. E. Northeastern Environ. Sci. 1989, 8, 1-86. (3) Bopp, R. F.; Simpson, H. J. Contamination of the Hudson River, The Sediment Record. In Contaminated Marine Sediments: Assessment and Remediation; National Academy Press: Washington, DC, 1989. (4) TAMS/Cadmus/Gradient. Phase 2 Report, Further Site Characterization and Analysis, Volume 2C: Data Evaluation and Interpretation Report (DEIR), Hudson River PCBs RI/FS; Prepared for U.S. EPA Region 2 and United States Army Corps of Engineers (USACE); TAMS Consultants, Inc.: Bloomfield, NJ, 1997; available at http://www.epa.gov/hudson/reports.htm. (5) U.S. EPA. Hudson River PCBs Site, New York; Record of Decision; U.S. Environmental Protection Agency: Washington, DC, 2002; available at http://www.epa.gov/hudson/d_rod.htm#record. (6) Rhee, G. Y.; Sokol, R. C.; Bethoney, C. M.; Bush, B. Environ. Sci. Technol. 1993, 27, 1190-1192. (7) Garvey, E. A.; Atmadja, J.; Butcher, J. B. Clearwaters 2002, 32, 21-29. (8) Butcher, J. B.; Garvey, E. A.; Bierman, V. J. Chemosphere 1998, 36, 3149-3166. (9) O’Brien & Gere. Data Summary Report, Hudson River Sampling and Analysis Program, 1991 Sediment Sampling and Analysis Program; Prepared for General Electric Company Corporate Environmental Programs; O’Brien and Gere Engineers, Inc.: Syracuse, NY, 1993. (10) Reible, D. D.; Popov, V.; Valsaraj, K. T.; Thibodeaux, L. J.; Lin, F.; Dikshit, N.; Todaro, M. A.; Fleegler, J. W. Water Res. 1996, 30, 704-714. (11) Connolly, J. P.; Zahakos, H. A.; Benaman, J.; Ziegler, C. K., Rhea, J. R.; Russell, K. Environ. Sci. Technol. 2000, 34, 4076-4087. (12) TAMS Consultants, Limno-Tech, Inc., Menzie-Cura & Associates, Inc., and Tetra Tech, Inc. Further Site Characterization and Analysis, Vol. 2D: Revised Baseline Modeling Report, Hudson River PCBs Reassessment RI/FS; Prepared for U.S. EPA Region 2 and U.S. Army Corps of Engineers, Kansas City District; 2000; available at http://www.epa.gov/hudson/reports.htm. (13) Valsaraj, K. T.; Thibodeaux, L. J.; Reible, D. D. Environ. Toxicol. Chem. 1997, 16, 391-396. (14) Butcher, J. B.; Garvey, E. A. Congener Pattern Matching to Evaluate Sediment PCB Source in the Upper Hudson River. Presented at SETAC Annual Meeting, Philadelphia, PA, November 14-18, 1999.

Received for review December 23, 2003. Revised manuscript received April 13, 2004. Accepted April 14, 2004. ES035453T