Bioaccumulation in Functionally Different Species: Ongoing Input of PCBs with Sediment Deposition to Activated Carbon Remediated Beds

Abstract

Activated carbon (AC) amended bed sediments reduced total polychlorinated biphenyl (PCB) accumulation in three functionally different marine species, sandworms (Alitta virens), hard clams (Mercenaria mercenaria), and sheepshead minnows (Cyprinodon variegatus), during both clean and contaminated ongoing sediment inputs. Mesocosm experiments were conducted for 90 days to evaluate native, field-aged bed sediment PCBs and ongoing input PCBs added thrice a week. Simulated in situ remediation applied an AC dose equal to the native organic carbon content, which was pre-mixed into the bed sediment for one month. The highest bioaccumulation of native PCBs was in worms, which remained in and directly ingested the sediment, whereas the highest bioaccumulation of the input PCBs was in fish, which were exposed to the water column. When periodic PCB contaminated sediment inputs were introduced to the water column, the AC remedy had minimal effect on the input PCBs, while the native bed PCBs still dominated bioaccumulation in the control (no AC). Therefore, remediation of only the local bedded sediment in environmental systems with ongoing contaminant inputs may have lower efficacy for fish and other pelagic and epibenthic organisms. While ongoing inputs can continue to obscure remedial outcomes at contaminated sediment sites, this study showed clear effectiveness of AC amendment remediation on native PCBs despite these inputs, but no remediation effectiveness for the input-associated PCBs (at least within the study duration).

INTRODUCTION

An enormous challenge at contaminated sediment sites is the occurrence of ongoing inputs of contaminated sediment, and associated dissolved contaminants, to sites either undergoing remediation or that have completed remediation. Ongoing inputs may spread contamination by the slow deposition of contaminated sediment that had resuspended and transported from contaminated areas in the vicinity or from other uncontrolled source zones. Environmental dredging and in situ remedies, like capping and amendment with activated carbon (AC), can be compromised by these ongoing inputs.

Relatively few laboratory studies have investigated in situ sediment amendment with AC in the presence of ongoing inputs, while nearly all field studies had ongoing inputs of some kind. Cho et al. (2009) observed polychlorinated biphenyl (PCB) tissue reductions in field exposed clams in AC amended plots at Hunters Point (San Francisco, CA, USA) after 6 months, but did not observe reductions after 18 months and speculated that the surface deposit-feeding clams were exposed to ongoing inputs. Ongoing inputs were related to wave and tidal induced resuspension and re-deposition (Janssen et al. 2011) and possibly storm sewer outfalls (Zimmerman et al. 2004). Ongoing inputs at Hunters Point were also attributed to reduced remedial efficiency (from 90 to 44%) as measured by bioaccumulation in a nereid worm (Janssen et al. 2011).

In freshwater, Beckingham and Ghosh (2011) observed reduced bioaccumulation in control plots (without AC), attributed it to the deposition of cleaner sediment over time, and considered the deposition of untreated sediment to reduce effectiveness of AC amended field plots. Abel and Akkanen (2018) showed that highly dynamic sediment movements could result in low impact of AC amendment. Ghosh et al. (2011) noted that over time, new deposition of sediment could cover AC treated sediment and that laboratory studies could precisely describe the effect of overlying water and depositional impacts to in situ AC amendment remedies. Lin et al. (2018) examined similar processes in freshwater microcosms.

The present study examined these processes in marine mesocosms, with a focus on bioaccumulation. PCB contaminated New Bedford Harbor (NBH) (MA, USA) sediment, which has been studied by researchers for over 40 years (e.g. Stoffers et al. 1977; Friedman et al. 2009 and 2011; Karim et al. 2015), was used as the bed sediment in this study. However, NBH sediment has only recently been investigated with regard to AC amendment (Bridges et al. 2017; Schmidt et al. 2017). Relative to most AC amendment studies, the NBH sediment used in the present study can be considered “highly contaminated” (about 30 mg/kg (dw), see Schmidt et al. 2017). Tenax extraction of NBH sediment from the same batch as used in this study has also been reported by Sinche et al. (2019), which showed that the PCBs were accessible. Here, bioaccumulation of PCBs was examined in three marine species with different functional traits: sandworms (Alitta virens; formerly Nereis virens), hard clams (Mercenaria mercenaria), and sheepshead minnows (Cyprinodon variegatus). These organisms interact differently with the sediment and therefore allowed for an evaluation of remediation effectiveness across a range of representative functionally distinct species. Janssen et al. (2011b) and Bridges et al. (2017) describe in detail the importance of considering differences in the functional ecology between organisms and the benefits of employing functionally different test species to more comprehensively assess the efficacy of remedial actions.

The present study is a sub-set of a larger investigation to determine the effectiveness of not only AC amendment, but also sand caps and AC amended sand caps, on remediating contaminated sediment (Figure S1, Supporting Information, SI). In Schmidt et al. (2017), we previously described the overall experimental treatments, each consisting of six mesocosms, and cross validated two partitioning-based sampling approaches for measuring PCB exposure using ex situ silicone coated jars for equilibrium sampling and in situ low density polyethylene (LDPE) for passive sampling. Mesocosm experiments were conducted with unspiked (“clean”) or PCB-contaminated (“spiked”) sediment inputs dispersed into the overlying water in a pulsed manner thrice a week for the duration of the experiment. Bioaccumulation was determined in two experimental treatments: unamended bed sediment (“control”) and activated carbon amended bed sediment (“AC mix”). The use of laboratory mesocosms allowed for focused examination of ongoing inputs, bioturbation (including bioirrigation) and diffusion, without confounding processes, such as: waves, currents, groundwater intrusion, and ship traffic (prop wash) (Kirtay et al. 2018; Karim et al. 2015). The use of mesocosms also avoided the confounding influence of seasonal (e.g. temperature) and tidal (e.g. salinity) variations and minimized losses of AC during application, which are all common occurrences in the field. Finally, the mesocosms allowed for complete bed mixing with the AC, whereas in the field, there may be a practical limit to the depth of AC mixing.

The specific objectives of the present study were to determine if (1) AC mixed into the bed sediment reduced PCB bioaccumulation relative to corresponding control organisms for NBH sediment; (2) the reduction in bioaccumulation was conserved across three functionally distinct species (worm, clam, fish); (3) continued periodic sediment inputs, contaminated or clean, compromised the remediation technique.

METHODS AND MATERIALS

Summary of experimental apparatus and treatments

A detailed description of the mesocosm experiments with focus on partitioning-based sampling methods was previously described (Schmidt et al. 2017). The AC employed was a high-quality, coal-based product unaltered from Calgon Corp. (Pittsburg, PA, USA) (TOG 80x325 mesh). Using a Beckman Coulter Counter (LS 13 320 Dry Powder System) this AC was measured to have a mean particle size of 111 μm, a d10 value of 32 μm, a d50 (median) value of 112 μm, and a d90 value of 184 μm (Schmidt et al. 2017). Hale et al. (2010) studied an unaltered 80x325 mesh TOG AC for PCB sorption and attenuation caused by sediment. The AC was incorporated into the bed sediment by mixing 4.3% AC by dry weight (dw) (approximately equal to the native organic carbon content in NBH sediment) by a propeller mixer for approximately 1 hour, and then mixing on a drum roller (9 rpm) for one month at ambient temperature (19-38 °C) prior to initiating the 90-day mesocosm experiment. Control sediment was also mixed by the propeller mixer, but not mixed on the drum roller.

A diagram of clean input sediment and spiked input sediment mesocosms for both the control and the AC mix is provided in Figure 1 (panel A). Each of these experimental treatments (control or AC mix) consisted of six separate mesocosms (52 L aquaria); within each treatment, three replicate mesocosms received clean input sediment and three replicate mesocosms received PCB spiked input sediment. Overlying water from each set of three mesocosms (same treatment and input type) was treated for ammonia to maintain organism health by pumping it (13.1 ± 2.9 L/hr per mesocosm) through gravel bed treatment plants, illustrated in Figure 1 (panel B). While each set of three mesocosms was hydraulically connected via a treatment plant, the pumps were turned off prior to addition of sediment inputs, and remained off while the inputs settled; therefore, the influence of the hydraulic connection was minimized between replicates.

Figure 1.

Mesocosm design diagrams. A.) Side view of triplicated (n = 3) aquaria representing clean and spiked sediment input mesocosms for the control and AC mix treatment. B.) Overhead view of an experimental treatment consisting of 6 mesocosms, 2 treatment plants, and a temperature-controlled water bath. A total of 12 experimental mesocosms were included in the present study (6 control mesocosms; 6 AC mix mesocosms).

Addition of continuous inputs

The input sediment was a relatively clean, marine sediment from Bayou Lafourche (Leeville, LA, USA). The full chemical analysis of this sediment is provided in Table S1 and additional information was provided by Schmidt et al. (2017). The inputs were added three days a week (on Mondays, Wednesdays, and Fridays) as 5 mL wet sediment, first pipetted into a large beaker containing approximately 275 mL water from the respective mesocosm, swirled in the beaker, and then poured across each mesocosm to create a well dispersed plume of about 33 mg/L total suspended solids. Contaminated input sediment was spiked with tracer PCB congeners 13, 54, and 173 (Ultra Scientific, North Kingstown, RI, USA), intended as congeners present only in the sediment inputs (not the bedded sediment), so that organism bioaccumulation of PCBs could be traced to bed sediment or input sediment. Spiked input sediment was placed on a jar roller (5.5 rpm, 24 °C) for two weeks before use in the mesocosms to optimize sorption to input sediment (see Schmidt et al. 2017). PCB 13 was later discovered as a native congener in NBH sediment; therefore, only congeners 54 and 173 were considered “tracer” PCBs for the input since they were absent from the NBH bed sediment. Potential exposure of the organisms to the input PCBs was both depositional (with the input sediment) and aqueous (as the input sediment desorbed PCBs into the surface water, which then re-sorbed to various surfaces in the mesocosms). By the end of the 90 days, a theoretical 0.14-cm layer of sediment would have settled on the bed. However, organism activity and fish cage/refuges prevented a uniform layer, and burrows were made through the surficial layers.

The total concentrations (C_total) of the three congeners spiked into the input sediment (approximately 23 mg/kg dw) were selected to approximately equal the total concentrations of PCBs in the NBH sediment; PCB 13, PCB 54, and PCB 173 in the input approximated the native concentrations of (1) di-, and tri-chlorinated congeners, (2) tetra-, and penta-chlorinated congeners, and (3) hexa-, hepta-, and octa-chlorinated congeners, respectively. PCB 54, a tetrachlorobiphenyl, has a log K_ow more similar to di- and tri-chlorinated congeners than other tetra- and penta-chlorinated congeners. PCB 54 was selected because it was one of the few tetra- or penta-chlorinated congeners not present in the bed initially, could be isolated from native congeners in chromatograms, and was commercially available in gram quantities.

Test organisms and operations

Alitta virens, Mercenaria mercenaria, and Cyprinodon variegatus (hereafter denoted: “worm”, “clam”, and “fish”, respectively) were selected as test species to satisfy the following criteria, which are more fully described in Table 1: (1) phylogenetic diversity; (2) distinct functional feeding strategies; (3) differing locations and interactions with the sediment; and (4) common use in ecotoxicological and environmental risk assessment research and management decisions.

Table 1.

Brief description of the diversity in the experimental organisms selected to comprehensively assess sediment remediation efficacy across functionally distinct species.

Organism	Common name	Phylum	Feeding	Habit (behavior, sediment interactions)
Alita virens	Sandworm	Annelida	Deposit-feeders	Infaunal, deep sediment burrowers
Mercenaria mercenaria	Hard clam	Mollusca	Filter-feeders (at water-sediment interface)	Epifaunal, shallow sediment burrowers
Cyprinodon variegatus	Sheepshead minnow	Chordata	Predators	Pelagic, water column and epibenthic swimmers

The worms and clams were field collected and acquired from Aquatic Research Organisms (Hampton, NH, USA), and the fish were laboratory cultured and obtained from Aquatic Biosystems (Fort Collins, CO, USA). The worms and clams were allowed to acclimate to laboratory conditions for at least 24 hours prior to addition to experiments, while the fish were allowed to acclimate for at least one week. The fish (<90 days old) were all males to reduce the variability in lipid content among individuals during maturation. Since preliminary experiments resulted in the fish resuspending large amounts of bedded sediment that would have altered the release of PCBs into the water column beyond what would be realistic in the field, the fish were restricted to the water column using stainless steel cages. Stainless steel dividers and fish refuges (either stainless steel or Teflon) were placed inside the cages to reduce aggressive fish interactions. Fish were periodically moved to the opposite side of the divider to disrupt social dominance hierarchies.

At the start of the experiments, five worms, fifteen clams, and five fish were added to each mesocosm. For clams, this resulted in a density of 118 individuals per m² in the mesocosms, which is on the low end of field density ranges of 100 to 1200 for juvenile clams per m² reported in the literature (Bricelj 1993). For worms, this loading resulted in a density of 39 worms per m² in the mesocosms, which is within field density ranges of 10 to 160 worms per m² (Miron and Kristensen 1993; Caron et al. 1996). These densities are reported as they may relate to relevant levels of bioturbation and contaminant transport (Bosworth and Thibodeaux, 1990). To meet the biological requirements of the three test organisms, all mesocosm experiments were conducted at 20 °C, a salinity of 30‰, and a photoperiod of 16-h light and 8-h dark (USEPA/USACE, 1998). The seawater was prepared by dissolving and equilibrating Instant Ocean^® (Blacksburg, VA, USA) in reverse osmosis water and slowly adding the water to fill the aquaria containing 13-14 kg wet NBH bed sediment (unamended or AC amended). During the experiments, diffused air was continuously supplied to aerate the surface water, and water quality (temperature, pH, dissolved oxygen, and salinity) was monitored daily. Ammonia-N concentrations were monitored at least once a week using either an Orion Dual Star meter (Thermo Orion Electron Corp., Beverly, MA, USA) equipped with an Orion 9512 ammonia-sensitive electrode or LeMotte titration kits (Chestertown, MD, USA). The fish were fed Tetra Marine^® seawater flakes (Blacksburg, VA, USA) ad libitum twice daily (once daily on weekends) to meet nutritional requirements and reduce intra-specific male aggression.

The fish were not fed at the same time as the addition of ongoing input sediment to reduce PCB sorption to feed, thus minimizing dietary exposure. Since the clams were previously reported to close their valves and presumably reduce their filtering rate when suspended food sources were limited (Bridges et al. 2017), they were fed an algae mix (Phytoplex®, Kent Marine, Franklin, WI, USA) ration of approximately 10,000 cells/mL three times per week. Clam food was added immediately after adding ongoing input sediment, which encouraged filter feeding while ongoing input sediment was in suspension. Even though no specific supplemental food was provided, the worms likely ingested bedded sediment organic matter enriched with unconsumed feed and feces. Additional information on the organism feeding is provided in the SI.

At the termination of the experiments, the surface water was siphoned off, sediment core samples were taken for equilibrium sampling, and passive samplers were retrieved as detailed in Schmidt et al. (2017). For each mesocosm, recovered organisms were separated by species, and placed into clean seawater to allow purging of their gut contents for 24 hours. The fish were euthanized with tricaine methanesulfonate (3 g/L MS-222 in clean seawater buffered with 2.1 g/L NaHCO₃). Clams were further dissected to remove remaining sediment from the gut as described by Kennedy et al. (2010). All tissues were frozen until homogenization, extraction, and analysis.

Analytical chemistry

Tissue samples were extracted with hexane (95%, Fisher Scientific, USA) by modified sonication, EPA SW-846 Method 3550. The congeners were extracted either in a sonic bath overnight with hexane for larger tissue amounts, or in a single extraction with hexane using a sonic probe for smaller amounts of fish tissue. Lipids and other interfering compounds were removed from sample extracts either by treatment with concentrated sulfuric acid (modified EPA 3665) for clams and worms, or by silica gel chromatography (EPA 3630) for fish. Modified micro-methods were occasionally required for fish tissue due to lower mass recoveries (Jones et al. 2006). A single value was used for all individual congeners as a reporting limit (RL), with a detection limit (DL) at one third this concentration. Worm and clam DLs were between 0.10 and 0.11 μg/kg, and fish DLs were between 0.12 and 0.28 μg/kg, depending on the tissue mass. Values below the DLs were counted as zero in the summation of PCBs for plots (referred to in the results section). The RL was used as the lowest calibration standard and is supported by method DL studies performed per 40 CFR 136 Appendix B guidelines (USEPA, 2011). Silicone and LDPE extractions were described by Schmidt et al. (2017). Sediments were extracted by pressurized fluid extraction using hexane/acetone (EPA 3545A) and cleaned with H₂SO₄ (modified EPA 3665A). Dual column gas chromatography with two electron capture detectors was used for analysis of approximately 132 PCB congeners as previously described by Schmidt et al. (2017). The sulfo-phospho-vanillin microcolorimetric method of Van Handel (1985) was used to measure the lipid content of tissues, though wet-weight normalized concentrations of PCBs were the focus of this paper.

Statistical Analysis

Statistical comparisons were performed as T-tests using SigmaStat^® v3.1 statistical software (SPSS, Chicago IL). Data normality (Kolmogorov–Smirnov test), homogeneity (Levene’s Test), and treatment differences were determined at the 0.05 α-level. Hence, statistically significant differences have p < 0.05. When normality was not initially achieved, square root or rank transformations were performed and data were analyzed by T-tests, as described above.

RESULTS AND DISCUSSION

Native PCBs in all species

Mixing the bed sediment with AC greatly reduced the total bioaccumulation of native bed congeners for all three marine organisms. Relative to the control, the total native PCB bioaccumulation in worms exposed to the AC mix in mesocosms receiving clean and spiked inputs was significantly reduced by 89% (p = 0.001) and 91% (p < 0.001); similarly, bioaccumulation in clams was significantly reduced by 94% (p < 0.001) and 95% (p < 0.001), respectively (Figure 2). There was a 94% reduction of total native PCBs accumulated in fish by the AC mix relative to the control in both the clean input (p = 0.002) and spiked input mesocosms (p = 0.1) (Figure 2). Tissue masses (Table S2), lipid content (Table 2), and total lipid-normalized tissue concentrations (Figure S2) are provided. Sum homolog bioaccumulation reductions by AC mix occurred for both clean and spiked input mesocosms, except for the octa and nona-chlorinated congeners in worms, and the hepta-, octa-, and nona-chlorinated congeners in clams and fish (Figure 3). Data from Figure 3 are provided on a linear scale in Figure S3 (tabulated data provided in Table S3). Our results are consistent with others on the lower effectiveness of AC in reducing bioaccumulation of higher chlorinated PCBs at least in the short term (Millward et al. 2005; Cho et al. 2009; Beckingham and Ghosh, 2011; Cornelissen et al. 2012 and 2015; Choi et al. 2016; Kirtay et al. 2018). The native congeners for which no reduction in bioaccumulation was detected made up less than 1% of the total native PCB bioaccumulation in the control and less than 10-20% of the total native PCB bioaccumulation in the AC mix. Tissue concentrations in organisms exposed to the AC mix were plotted against tissue concentrations in organisms exposed to the control (Figure 4). A reduction in bioaccumulation occurred for most congeners, as indicated by data that were well below the diagonal 1:1 line. To indicate the large number of congeners that were reduced below DLs by the AC mix, Figures S5 through S8 show the fraction reduction (AC mix versus control) in tissue, all versus log K_ow. In all species, approximately 30 PCB congeners were reduced to non-detect in the tissues of organisms exposed to the AC mix. Between four and eleven congeners “appeared” in tissues exposed to AC mix that were not detected in the controls (data not shown, but details are provided in the SI). Further research should be conducted to determine if these increases and “appearances” of PCB congeners with slow sediment-to-AC mass transfer were due to an experimental artifact (e.g. bed sediment heterogeneity) or real phenomenon (e.g. increased feeding rates). For the control, maximum total native homolog bioaccumulation occurs in the tri- to tetra-chlorinated congener range (Figure 3). In the AC mix treatment, this maximum shifts to the hexa-chlorinated congeners mostly due to greater reduction in bioaccumulation of lighter (fast) congeners (Figure 3). This shift in maximum PCB bioaccumulation was observed for all species. Abel and Akkanen (2018) also observed a shift with AC application and low sediment deposition.

Figure 2.

Summary of sum congeners sourced from the bedded sediment and from the sediment inputs. Total native congeners sourced from the bedded sediment (excluding input congeners) in wet tissue (mg/kg) for clean and spiked input mesocosms (left side plots). Total input congeners (Sum PCB 13, 54, and 173) in tissue (right side plots). Both the control and AC mix are shown in each plot as brown and black bars, respectively. The top plots show worm (Alitta virens) results, the middle plots show clam (Mercenaria mercenaria) results, and the bottom plots show fish (Cyprinodon variegatus) results. These results are shown on a lipid basis in the SI (Figure S2).

Table 2.

Mean lipid contents for Alitta virens (worm), Mercenaria mercenaria (clam), and Cyprinondon variegatus (fish) by AC mix and control treatments. Within each treatment, the clean and spiked input mesocosms are also shown. Organism weights are provided in Table S2.

%Lipids in Wet Tissue
Control		AC mix
Clean	Spiked	Clean	Spiked
Alitta virens (worm)
1.76 ± 0.19	1.91 ± 0.34	1.83 ± 0.23	1.73 ± 0.21
Mercenaria mercenaria (clam)
0.87 ± 0.06	0.89 ± 0.08	0.97 ± 0.06	1.13 ± 0.06
Cyprinodon variegatus (fish)
5.16 ± 0.72	3.39 ± 0.70	8.63 ± 1.11	8.45 ± 1.09

Figure 3.

Summary of congener homolog groups sourced from the bedded sediment. Total native bed congeners (excluding input congeners) in wet tissues (mg/kg) for clean (left side plots) and spiked (right side plots) input mesocosms. Both control (brown bars) and AC mix (black bars) are shown in each plot. The top plots show worm (Alitta virens) results, the middle plots show clam (Mercenaria mercenaria) results, and the bottom plots show fish (Cyprinodon variegatus) results. The results are shown on a linear scale (Figure S3), and also provided in tabulated format (Table S3) in the SI.

Figure 4.

Concentrations of individual PCB congeners in wet tissues exposed to the AC mix (Y axis) versus the control (X axis) in both clean input mesocosms (circles, top plots) and spiked input mesocosms (squares, bottom plots). Congener symbols are color coded by homolog. Native congeners are hollow symbols, while input congeners (PCB 13, 54, and 173) are larger solid symbols. Congener 13 (a dichlorobiphenyl) was an input and also present as a native in the bed initially, and is represented by a large half-filled blue square. Values below the DLs were not included in these plots. An analogous presentation of PCBs in silicone equilibrium samplers is provided in Figure S4.

Worms, Alitta virens

The large 90% reductions of native PCBs by the AC mix were consistent with Janssen et al. (2010), who found large 95% reductions in a smaller nereid worm, Neanthes arenaceodentata. In the present study, lipid fractions (% lipids) in worms (Table 2) between the AC mix and the control were similar, suggesting no apparent treatment related impacts on worm health, consistent with findings by Millward et al. (2005) and Janssen et al. (2012) for N. arenaceodentata. Thomas et al. (2014) observed slightly higher lipid contents in N. arenaceodentata from treated sediments at Hunters Point, perhaps indicating improved health due to a reduction in contaminant exposure.

The worms reside in the sediment and their feeding behavior exposes them to the surficial deposition layers and some surface water (Freidman et al. 2009). The tissue concentrations of the three individual input congeners were at mid-to-high levels compared to individual native bed congeners (Figure 4), but the total input congeners made up a small fraction of the total PCBs in worms (Figure 2).

There was minor inconsistency in the concentrations of input PCBs 13, 54, and 173 spiked in batches of input sediment between the control and AC mix (see Schmidt et al. 2017), but generally the total input congener bioaccumulation in worms was similar in the control (0.14 ± 0.03 mg/kg tissue) and the AC mix (0.13 ± 0.02 mg/kg tissue) (spiked input mesocosms, Figure 2). Therefore, the large reductions achieved for the native congeners by AC mix were not observed for the input PCBs. This was primarily due to the lack of adequate mixing of the freshly deposited sediment with AC and short sediment-to-AC mass transfer time compared to the 1-month pre-mix for native bed congeners. Over time in the field, if the input PCBs continue to be integrated into the AC amended bed, reductions in worm bioaccumulation of input PCBs may be possible if input sources are eventually controlled. Likewise, Cornelissen et al. (2011) maintained that thin-layer AC amended caps could sorb contaminants in newly deposited or redeposited material better than caps made of less sorptive inert materials.

In the clean input mesocosms, PCB 13 was detected (at 0.03 μg/kg worm tissue) in the control, which could be attributed to native PCB 13 in the bed sediment (Figure 2). In the spiked input mesocosms, PCB 54 was significantly higher (p < 0.001) in the AC mix (Figure 4), even though PCB 54 was spiked into the input sediment at (unintentionally) slightly lower levels than the control (10.4 mg/kg dw for the AC mix versus 12.5 mg/kg dw for the control). The higher PCB 54 worm bioaccumulation was not expected based on reductions in the concentration of freely dissolved (C_free) porewater PCB 54 measured using passive samplers in the surficial bed sediment. Passive samplers also indicated a diffusive gradient of PCB 54 with high concentrations in the surface water to low concentrations in the sediment bed. A large portion of the input congeners ended up in the bed, either by deposition or diffusion (see calculations in the SI, and Table S4 and S5). Input PCB 173 had low reductions in bioaccumulation (AC mix versus control), similar to many slow (sediment-to-AC mass transfer) native PCBs (Figure 4).

Clams, Mercenaria mercenaria

The total native PCB concentrations in the clams were substantially less than in the worms (Figure 2). Differences in the bioaccumulation levels between the worms and clams were due to the variations in feeding strategies of the two species; that is, worms feed directly from sediments while clams are suspension feeders from the overlying water (Bridges et al. 2017; Kirtay et al. 2018). Significantly higher (p < 0.001) lipid content of the worms was also a factor (Table 2). Reduction patterns for native congeners were similar for both the clean and spiked input mesocosms (Figure 3). While both worms and clams are thought to reach steady-state with their environment within 90 days (Rubinstein et al. 1983; Kennedy et al. 2010; Bridges et al. 2017), the environment in the mesocosms was dynamic due to the ongoing inputs. Furthermore, the clams would be in a greater state of disequilibrium because the surface water and benthic boundary layer were more dynamic than the bed.

The large reductions in bioaccumulation of total native PCBs in clams by the AC mix were not observed for the input congeners. The sum of input congeners was 0.15 ± 0.01 mg/kg tissue in the control and 0.13 ± 0.01 mg/kg tissue in the AC mix (spiked input mesocosms, Figure 2). In the clean input mesocosms, some input congeners were detected at very low levels (<0.01 mg/kg clam tissue) in the control (Figure 2). This was due to the presence of native PCB 13, and possibly some cross-contamination associated with the treatment plant wet wells for PCB 54. All input PCBs (including PCB 173) had higher clam tissue concentrations than the individual native PCBs (Figure 4). Congener 173 showed small increases in bioaccumulation by the AC mix versus the control, similar to individual slow native PCBs. The individual input PCBs were at higher concentrations in clam tissues than the individual native congeners for a combination of reasons, including: (1) freshly spiked input PCBs versus field-aged PCBs in the bed sediment, (2) shorter sediment-to-AC mass transfer time for PCBs (the native congeners had one month of mass transfer time during the pre-mixing), and (3) clams are suspension feeders and would be most actively feeding near the benthic boundary layer, which was more highly influenced by the input sediment relative to the deeper, bedded sediment. For this last reason, the difference between the inputs and native congeners was greater for the clams than the worms (Figure 4). Similarly, Samuelsson et al. (2015) attributed newly deposited particles to reduced bioaccumulation in worms relative to clams, and Kirtay et al. (2018) also noted this.

Fish, Cyprinodon variegatus

Overall, there was greater variability in the fish data relative to the other species. Reductions in the total native PCBs of 94% on a wet wt. basis were consistent with Fadaei et al. (2015), who found 87% reductions in lipid-normalized accumulation in the zebra fish by AC amendment to sediment. The mesocosms showed that fish can be a good indicator of remedy effectiveness when they have a close association with the sediment (e.g. small home ranges or non-migratory). This supports the findings of field studies (Oziolor et al. 2018; Schäfer et al. 2015). However, similar to the observation for worms and clams, AC amendment provided no overall reduction in fish bioaccumulation of input PCBs.

Van Geest et al. (2011) estimated 209 days would be required for fathead minnows to reach steady-state PCB concentrations in tissues, and Fadaei et al. (2015) predicted 180 days to steady-state in zebra fish. Here, the fish may not have reached steady-state with their environment, which was further confounded by the dynamic environment in the mesocosms. Fadaei et al. (2015) demonstrated that the majority of the PCB uptake in pelagic zebra fish exposed to untreated sediment was through gills rather than ingestion of sediment or feed that picked up PCBs in the aquaria. In the AC amended aquaria, the ingestion pathway then contributed to most of the total PCB uptake in zebra fish (Fadaei et al. 2015). Similar pathways may be occurring in the mesocosms, though the ongoing inputs and bioturbation added more complexity.

Unlike the worms and clams, the fish collected from the AC mix treatment had significantly higher lipid content relative to the fish collected from the control (p = 0.01 for clean input mesocosms, p = 0.003 for spiked input mesocosms). Furthermore, within the control, fish from the spiked input mesocosms had lower (p = 0.05) lipid content relative to fish collected from the clean input mesocosms (Table 2). Fish survival ranged from 27% (control, spiked inputs) to 92-93% (AC mix or control, clean inputs). Critical body burdens for baseline toxicity are reported to range between 40 and 160 mmol/kg lipid (van Wezel and Opperhuizen, 1995), while the fish in the control were only at 0.64 and 0.14 mmol PCBs/kg lipid, respectively, for the spiked and clean input mesocosms. Other factors likely explain the fish stress and mortality, such as aggressive interactions between fish. Fish dividers and refuges were only introduced toward the end of the control experiment (after aggressive interactions were observed) and at the beginning of the AC mix experiment (which started a month after the control experiment) and was therefore a likely cause for survival differences. Another factor was high copper and other heavy metals in NBH sediment (Stoffers et al. 1977), although Cu concentrations were not measured in this study. In any case, organism health likely contributed to the variability in the bioaccumulation of PCBs in fish, which was not observed for the other species (Figure 4). Since only one to two individual fish were recovered for the tissue composite in chemical analysis, individual variability was more pronounced and thus impacted intra-replicate variability more between treatments, relative to the other species. Finally, variability in flow rates through the treatment plants between mesocosms might be a third factor contributing to variability in the fish PCB bioaccumulation. Since the fish were confined to the surface water, they were expected to be the species most affected by surface water PCB concentrations. The treatment plants influenced surface water PCB concentrations (see Tables S4 and S5).

Though it was less noticeable than in other species, total native bedded PCB concentrations in fish were higher, yet not significantly (p = 0.3) in the spiked input mesocosms than in the clean input mesocosms (Figure 2, control experiment). This may be due to the clean input sediment providing mild treatment by sorbing native bedded PCBs, both as suspended particles and as surficial deposition, relative to the spiked input sediment. Supporting this, the passive sampler data showed higher (p = 0.002 or less) surface water native congener C_free for spiked inputs than for clean inputs (Figure S9).

Total input congeners were 0.83 ± 0.20 mg/kg tissue in the control and 0.88 ± 0.54 mg/kg tissue in the AC mix (spiked input mesocosms, Figure 2). Total input congeners were also plotted on a lipid basis (Figure S2). In the clean input mesocosms, some input congeners were detected at very low levels (≤ 4 μg/kg), due to native PCB 13 and possibly some cross-contamination (as previously mentioned) in the case of PCB 54 (labeled with an arrow in Figure 4). PCB 54 was detected in the clean input mesocosms at only 4 μg/kg versus 680 μg/kg in the spiked input mesocosms. The total input PCBs in the spiked input mesocosms were 5 to 7 times higher in the fish compared to the other species (Figure 2). High uptake was the combined effect of the fish ventilating in the surface water, fish ingesting input sediment and feed contaminated within the mesocosm (though unlike the clams, fish were not fed during the addition of inputs), and the high lipid content of the fish.

Input versus native PCBs in all species

The percentages that input congeners contributed to total bioaccumulation were plotted for all species in Figure 5 (also Figure S10). This provides a measure of the relative importance of bed sediment versus surface water and surface depositional layers as exposure routes across species in these mesocosms. The worms had the most prominent association with the sediment bed, which supports the conclusions of Bridges et al. (2017). Relative to the worms, the clams were more influenced by the surface water and surface deposition than the sediment bed. This emphasizes the importance of ongoing inputs as a source of PCBs to both filter-feeding and surface deposit feeding clams, and the limitations of AC for reducing bioaccumulation in a surface feeding clam exposed to ongoing inputs, as hypothesized by Cho et al. (2009).

Figure 5.

The percent of the total PCBs that were input PCBs across species (for spiked input mesocosms). Note: native PCB 13 was counted as an input in these plots. Figure S11 is similar, but excluding PCB 13 and only comparing the sum of PCB 54 and 173 versus the native PCBs. The fractions did not change considerably after excluding PCB 13.

In the control, the fish tissue was still dominated by the bed congeners (83%) despite being exposed in the surface water (Figure 5). The large contribution of native PCBs to total fish bioaccumulation may be partly because the input sediments were pulsed and only in suspension with the treatment plants turned off for up to about 21 hours per week (i.e.,13% of the time), while the contribution from the bed sediment was more constant. Bioturbation (see Bosworth and Thibodeaux, 1990) was likely a major transport process from the bed to the surface in the mesocosms.

The large relative contributions of inputs to clams and fish in the AC mix was not due to an increase in the mass of input congeners in tissues, but simply due to a reduction of the native PCBs in these organisms by the AC mix (Figure S11). Relative to the clams, and especially the worms, bioaccumulation in fish from the congeners sourced from the input was more important for determining tissue burdens, especially for the AC mix treatment.

Bridges et al. (2010) called for more research on the impact to bioaccumulation of thin-layers of contaminated sediment overlying clean sediment. With respect to congeners 54 and 173, the bed sediment was clean. Ongoing inputs, resulting in thin-layers ranging from 0 cm (at day 0) to the estimated theoretical thickness of about 0.14 cm (at day 90), led to measurable bioaccumulation in the short term. Compared to accumulation of bedded congeners, the influence of the inputs was at most 11 to 17% for clams and fish (Figure 5). However, with an AC remediated bed, the input contribution was a substantial part of the overall bioaccumulation. Within the context of these 3 month mesocosms, a thin-layer producing input could be acceptable if <83% reduction in bioaccumulation is sufficient. However, if >83% reduction in bioaccumulation is required, these mesocosms showed that ongoing inputs (resulting in thin-layer residuals) will prevent remediation goals from being achieved.

CONCLUSIONS

In mesocosms where bedded sediment was a major source of PCBs, amendment with AC reduced overall (native plus input) PCB bioaccumulation in aquatic species belonging to three vastly different functional trait groups (Figure S12). Worms feeding in/on sediments, clams at/above the sediment-water interface, and fish in the water column showed substantial reductions in total PCB bioaccumulation of 89%, 90%, and 95%, respectively in clean input mesocosms and 91%, 86%, and 78%, respectively in spiked input mesocosms. When periodic sediment inputs with spiked non-native PCBs were introduced to the water column, the AC mix treatment had minimal effect on these tracer congeners for the duration of the study. Recovery after any remedy implementation will thus be influenced by ongoing inputs. If ongoing inputs persist in the field similar to conditions in these mesocosms, the AC may not provide sufficient benefit to fish or clams and the AC may never “catch-up” to sequester and remediate inputs. This will be a concern with any kind of remedy (including environmental dredging and capping) and emphasizes the need to control ongoing sources at contaminated sediment sites.