Effect of PCB Bioavailability Changes in Sediments on

Subscriber access provided by University of Massachusetts Amherst Libraries

Article

The effect of PCB bioavailability changes in sediments on bioaccumulation in fish Hilda Fadaei, Aaron Watson, Allen Place, John P. Connolly, and Upal Ghosh Environ. Sci. Technol., Just Accepted Manuscript • Publication Date (Web): 24 Sep 2015 Downloaded from http://pubs.acs.org on October 4, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 25

1

Environmental Science & Technology

The effect of PCB bioavailability changes in sediments on bioaccumulation in fish

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Hilda Fadaei1, Aaron Watson2, Allen Place3, John Connolly4, and Upal Ghosh1*

17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

In-situ sediment amendment with sorbents such as activated carbon (AC) can effectively reduce the bioavailability of hydrophobic organic chemicals such as polychlorinated biphenyls (PCBs). However, there is limited experimental or modeling assessment of how bioavailability changes in sediments impact bioaccumulation in fish – the primary risk driver for exposure to humans and top predators in the aquatic ecosystem. In the present study we performed laboratory aquarium experiments and modeling to explore how PCB sorption in sediments impacted exposure pathways and bioaccumulation in fish. Results showed that freely dissolved PCBs in porewater and overlying water measured by passive sampling were reduced by more than 95% upon amendment with 4.5% fine granular AC. The amendment also reduced the PCB uptake in fish by 87% after 90 days of exposure. Measured freely dissolved concentrations were incorporated in equilibrium and kinetic models for predicting uptake by fish. Predicted uptake using the kinetic model was generally within a factor of 2 for total PCBs measured in fish. The kinetic model output was most sensitive to overlying water PCBs, lipid fraction, and dissolved oxygen concentration (regulating gill ventilation). Our results indicate that by incorporating changes in freely dissolved PCB concentrations in bioaccumulation models it is possible to predict effectiveness of sediment remediation in reducing PCB uptake in fish.

1

Department of Chemical, Biochemical, and Environmental Engineering, University of Maryland Baltimore County, 1000 Hilltop Circle, Baltimore, Maryland 21250, United States 2

Marine Resources Research Institute, South Carolina Department of Natural Resources, 217 Fort Johnson Road, Charleston, SC 29412 3

Institute of Marine and Environmental Technology, UMCES, Columbus Center, 701 East Pratt St. Baltimore, Maryland 21202, United States 4

Anchor QEA, LLC, 123 Tice Boulevard, Suite 205, Woodcliff Lake, NJ 07677

Abstract

33 34 35

*

Corresponding author: Email: [email protected] Phone: 410-455-8665

36

1

ACS Paragon Plus Environment


37

INTRODUCTION

38

Polychlorinated biphenyls (PCBs) in polluted sediments can be taken up by pelagic or benthic

39

organisms through two major pathways: absorption from water and ingestion of contaminated

40

food and sediment.1 Recent findings indicate that the freely dissolved concentration of PCBs in

41

porewater is the driving force for transport to the water column and bioaccumulation in benthic

42

organisms. Porewater PCBs can be attenuated by the presence of natural or anthropogenic black

43

carbon (BC) and the strong binding of these chemicals to BC can reduce the bioavailability to

44

benthic organisms.2-4 Based on this emerging understanding of PCB bioavailability in sediments,

45

amendment of contaminated sediment with sorbent activated carbon (AC) has gained attention in

46

recent years as a non-removal, in-situ remediation technology. AC amendment has been

47

demonstrated to reduce porewater concentration of PCBs and reduce biouptake by deposit and

48

filter feeders.5-7 Successful laboratory and pilot-scale studies5 has led to an emerging

49

consideration of this new technology for sediment remediation as described in a recent USEPA

50

directive on in-situ amendments 8 and USEPA proposed remedial action for the Superfund site in

51

Housatonic River.9 While there is general agreement that sorbent amendments reduce

52

bioavailability of PCBs to the food chain, there is a lack of quantitative understanding on how

53

reductions in sediment porewater concentrations and reduced uptake at the base of the food chain

54

impact accumulation in fish, the primary driver for PCB exposure to humans and top predatory

55

animals.

56

Food chain models are available that can be used to predict uptake of PCBs in a range of aquatic

57

animals. However, there is limited knowledge of the ability of food chain models to predict PCB

58

levels in fish after AC-treatment of the sediment either in laboratory or field settings. A common

59

practice in remedial investigations at PCB contaminated sediment sites involves using biota-

60

sediment accumulation factors (BSAFs) to calculate contaminant concentrations in benthic

61

invertebrates from bulk sediment and empirical or literature-based partition coefficients to

62

calculate porewater concentrations from bulk sediment concentrations. (e.g., Upper Hudson

63

River10; Housatonic River11; Grasse River12) However, BSAFs and partition coefficients are

64

notoriously unreliable and have a wide range of variability.13, 14 This uncertainty is

65

accommodated by adjusting other model parameters as necessary to calibrate the model to

66

contaminant concentrations measured in the water column and fish. The uncertainty of BSAFs 2


Page 2 of 25

Page 3 of 25


67

and partition coefficients is magnified when sediments are amended with AC or other strong

68

sorbents in an effort to reduce porewater concentrations and bioavailability. If accurate freely

69

dissolved porewater and overlying water concentrations of PCBs can be measured directly, or

70

computed, and used as inputs in bioaccumulation models, changes in biouptake resulting from

71

sorbent amendment may be predicted more reliably. Experimental validation of this approach for

72

predicting changes in PCB uptake by fish can lead to credible models that allow quick

73

assessment of remediation progress by monitoring the critical exposure pathways to fish (benthic

74

organisms, porewater, and overlying water). Passive sampling can be a robust method for

75

measuring freely dissolved concentrations in water, especially for strongly hydrophobic

76

compounds like PCBs.15, 16 Previous work by Werner et al.17 showed that sediment porewater

77

PCB data can be used to predict uptake by freshwater and marine benthic organisms.

78 79

A recent study by Kupryianchyk et al.18 investigated the effect of different AC treatments on

80

reducing bioavailability to fish as well as the potential side effects of the treatment. The results

81

showed reductions in PCB uptake by macroinvertebrates and fish in systems treated with either

82

granular or powdered AC, with greater uptake reduction in the latter case. That study however,

83

did not attempt to explain the observations using a mechanistic biouptake model that can be used

84

to link bioavailability changes in sediment to reduction of uptake in fish.

85 86

In the present study we report results of a laboratory aquarium study where we evaluated the

87

effect of in-situ treatment of sediment with AC on PCB accumulation in fish. The objectives of

88

this study were to evaluate whether previously reported results of PCB bioavailability reduction

89

in benthic invertebrates is also observed in fish, and to test existing approaches of modeling PCB

90

uptake in fish to predict changes observed when the sediment is treated with AC. A passive

91

sampling approach was used to directly measure freely dissolved PCBs in porewater and surface

92

water. Ultimately, a sensitivity analysis was performed to serve as a guide for identifying the

93

crucial parameters of the model that need to be measured accurately to reduce uncertainty in the

94

model predictions.

95 96 97 3



98

MATERIALS AND METHODS

99 100

Sediment source

101

The laboratory exposure study was performed using three types of sediment: clean sediment,

102

PCB-impacted sediment, and PCB impacted sediment treated with AC in the lab. The clean

103

sediment was obtained from the Rhode River (RR) in MD, USA. PCB impacted sediment was

104

obtained from near-shore area outside of the activated carbon treatment areas of Grasse River

105

(GR), NY.19 Coal-based fine granular activated carbon (Carbsorb 75-300 µm; Calgon Carbon)

106

was added to sediments at a target dose of 4.5% by dry weight.

107 108

Test organisms

109

Zebrafish (Danio rerio, 8-12 week old juveniles) was used as model species to understand

110

uptake and was cultivated in-house at the Institute of Marine and Environmental Technology.

111

Fish were fed with fish-meal free, low PCB, plant protein based food flakes at 3.5 percent of

112

their body weight per day which was adequate to maintain the fish at average growth.

113 114

Measurement of aqueous concentration

115

76 µm-thick polyoxymethylene (POM) passive samplers were pre-cleaned via an ultrasonic

116

extraction using 50% acetone in hexane, air-dried under a fume hood for12 hours 20, and cut into

117

strips with mass ~ 0.5 g. Aqueous PCB concentration was calculated from PCB concentration in

118

the sampler based on KPOM values reported by Hawthorne et al.21

119 120

Aquarium set up

121

4 liters of sediment was added to replicate 38-L glass covered fish tanks and allowed to settle

122

and consolidate for a few days in static water. The amount of sediment was estimated based on

123

sediment height in the aquaria and surface area of the tanks. Tanks were arranged in a parallel

124

flow configuration and fed with recirculating, dechlorinated tap water at a recirculation rate of 8

125

gal/d (see Figure 1). Passive samplers were introduced in the water column and in sediment in

126

each aquarium and 10 individuals of zebrafish were added to each tank. The chemical

127

composition of the water (pH, conductivity, and alkalinity), ammonia and nitrite levels were

128

monitored throughout the experiment. pH and conductivity were measured using a Horiba U-22 4


Page 4 of 25

Page 5 of 25


129

multiparameter probe. Alkalinity, ammonia, and nitrate were measured with water quality test

130

kits (Hach Co.). Each aquarium was equipped with a heater to maintain the water temperature

131

uniformly at 28°C, the optimal growth temperature for zebrafish, in all tanks. Tanks were

132

equipped with air stones to maintain adequate oxygen level in the water. A photo period of 14

133

hours of light and 10 hours of darkness was maintained in the laboratory. The water drained from

134

the tanks was collected in two separate sumps. An activated charcoal bag was placed in each

135

sump to adsorb PCBs released from the tanks and prevent cross contamination by recirculating

136

water.

137 138

Sampling and analysis

139

Following 45-day and 90-day exposures, passive samplers were removed from the tanks, rinsed

140

with deionized water and wiped dry. Sediment samples were collected from the tanks at the start

141

of the experiment. Sediment and passive samplers were extracted in hexane/acetone (1:1, v/v),

142

cleaned up, and analyzed for PCB congeners in a gas chromatograph with electron capture

143

detection as described in Beckingham and Ghosh22. Five fish were sampled at the end of each

144

time interval, sacrificed using dry ice, and were lyophilized and frozen until analysis. Fish tissue

145

was ground with anhydrous sodium sulfate and extracted with hexane and acetone mixture (1:1,

146

v/v) following method SW846 3550C. The lipids were removed by treating with concentrated

147

sulfuric acid. Further cleanup was performed by treating the extract with activated copper and

148

passing through a 3% deactivated florisil and acidified silica gel column. The eluate was

149

concentrated by nitrogen evaporation and analyzed for PCB congeners as described above. Lipid

150

content in zebrafish tissue was measured in separate samples in triplicate using the gravimetric

151

method described by Harvey et al.23

152 153

Total Organic Carbon (TOC), Black Carbon (BC) and Activated Carbon (AC) analyses

154

TOC was determined by thermal combustion method on a Shimadzu TOC analyzer with a solids

155

sample module (TOC-5000A and SSM-5000A). BC and AC contents of the sediment were

156

measured by wet chemical oxidation pretreatment.24

157 158

Observed sediment-water partitioning constant (Kd) calculation

159

Measured PCB concentration in the sediment (ng/kg) and freely dissolved porewater PCB 5



160

Page 6 of 25

concentration measured with passive sampling (ng/L), were used to calculate Kd from equation 1:

161

1

162 163

Modeling approach

164

PCB concentration in sediment porewater

165

The effect of AC on aqueous partitioning was predicted by modeling sorption to organic carbon

166

(OC), native BC, and applied AC. Sorption to native BC was assumed to be linear17 and sorption

167

to AC was described by a Freundlich isotherm. The first modeling effort ignored the role of the

168

native black carbon pool in partitioning of PCBs (equations 2 and 3):

169

2

170

171

3

172 173

In the second approach, the black carbon pool was included (equations 4 and 5):

174

175

4

176

177

5

178 179

Where CW is the porewater concentration (ng/L), fOC, fBC, and fAC are the fractions of OC, native

180

BC, and AC in sediments, KOC and KBC are water-sorbent distribution coefficients for OC and

181

BC, respectively with unit of (L/kg sorbent). Kd in equations 2 to 5 was defined as the ratio of

182

measured PCB concentration in the sediment to unknown porewater concentration. CW values

183

which satisfy equations 2 to 5 were calculated. Modeled Kd was calculated from the predicted

184

CW (ng/L) and measured CS (ng/kg) values.

185

KOC (L/kg OC) values for PCB congeners were estimated using equation 6.17 To use this

186

empirical correlation the octanol-water partitioning constant (KOW (-)) was obtained from

187

Hawker and Connell.25 log

188

0.74 log

0.15 6

6


Page 7 of 25


189

As reported by Werner et al.17, native black carbon adsorption of PCBs is linear below the

190

microgram/L concentration range. KBC (L/kg BC) was estimated using the following correlation

191

with KOW 17:

192

log

193

0.91 log

1.37 7

194

n and Kf values used in our modeling approach were obtained from Gomez-Eyles et al.26 where

195

the same coal-based fine granular activated carbon was used in the absence of sediment. Since

196

natural organic matter in the sediment environment attenuates adsorption of PCBs to AC, a

197

reduction factor of 16 was applied to the obtained Kf values.27

198

PCB concentration in fish

199

Two approaches were taken in this study to estimate the PCB residue in fish. The first was the

200

steady state approach assuming thermodynamic equilibrium between water and fish lipid. Klipid

201

(L/kg lipid), the lipid-water equilibrium partitioning constant of PCBs, was used as a simplistic

202

model to predict Clipid (µg/kg lipid), using CW,O (µg/L). The overlying water concentrations used

203

as inputs to both equilibrium and kinetic models were measured by passive sampling. KOW (L/kg

204

octanol) can be used as a surrogate to quantify chemical partitioning between the overlying water

205

and the lipid fraction.17

206

,

8

207 208

The second approach was the kinetic approach. The two frequently used bioaccumulation models

209

(Arnot and Gobas1 and Connolly28) were used to generate predictions, and were compared to the

210

measured data. Environmental properties (including measured freely dissolved PCB

211

concentrations in the overlying water), chemical properties of PCBs, and biological

212

characteristics of the fish were incorporated into the kinetic model and input parameters were

213

calculated independently (see Supporting Information for a more detailed discussion of the two

214

models). To ensure that the model is representative of experiment conditions, the following

215

initial assumptions were made to obtain equation 9: (1) sediment porewater contribution to the

216

respiratory exchange of PCBs was assumed to be zero; (2) dietary uptake was zero (PCB-free

217

food); and (3) PCB loss via metabolism was negligible. As described by Arnot and Gobas1, MB 7



Page 8 of 25

218

is the mass (g) of PCB in the fish, dMB/dt is the net flux of PCB being absorbed or depurated by

219

fish at any point in time t (d), WB is the wet weight of the fish (kg) at time t, k1 is the gill uptake

220

rate constant (L/kg.d), mo is the fraction of the respiratory ventilation that involves overlying

221

water (which equals 1 in this case), CW,O is the freely dissolved PCB concentration in the

222

overlying water measured by passive sampling (g/L), k2 is the gill elimination rate constant (d-1),

223

and ke is the fecal egestion rate constant (d-1).

224

,

9

225

Since concentrations in the overlying water increased over time, a linear interpolation between

226

measured values from 0 to 45 and 45 to 90 days was used to define overlying water

227

concentration (Figure S1).

228

229

Integration of equation 10 yields:

230

, @

, @

10

11

231

Where a is the rate of change of aqueous concentration and A is the constant of integration

232

obtained by fitting equation 11 to the initial conditions.

233

PCB concentrations in fish were predicted by solving for the mass of PCB in the fish at 45 and

234

90 days and converted to lipid normalized concentration using equation 12:

235

12

236 237

Where Lf is the lipid content of the fish (kg lipid/kg wet weight). For more information on the

238

model parameters and detailed calculations of the rate constants, see Supporting Information.

239 240

The sensitivity of the model output to changes in each input parameter is described as:

13

241

8


Page 9 of 25


242

Where ΔO is the amount of change in the output (O) due to the amount of change (ΔI) in the

243

input variable (I). The sensitivity of the kinetic model output to dissolved oxygen concentration,

244

overlying water concentration, fish wet weight, and fish lipid fraction was calculated for

245

congeners with log (KOW) values ranging from 5 to 7. Each input parameter was increased and

246

decreased by 10% while maintaining other parameters constant.

247 248

RESULTS AND DISCUSSION

249 250

Sediment characteristics

251

TOC content in Rhode River was measured as 3.9±0.06% by dry weight and PCB concentration

252

was determined to be below the level of detection (0.01 μg/g dry wt.). Total PCB concentration

253

in the Grasse River sediment was measured as 0.67 μg/g dry wt. and averaged 2.2±0.06% for

254

TOC content. Sediment PCB data at a congener level is shown in Table S1.

255 256

Aqueous PCB concentrations

257

Total porewater dissolved PCB concentration in GR sediment was reduced from 631 (±23) to 1.3

258

(±0.2) ng/L, 99% reduction, in 90 days after amendment with AC (Figure 2, Table S4). The

259

reduction in porewater concentration of AC-amended field sediments was previously reported to

260

be in the range of 70-99% for hydrophobic organic chemicals.5 Total PCB concentration in

261

overlying water 90 days after AC amendment was reduced from 184 (±15) to 7.6 (±2.5) ng/L (a

262

96% reduction) and was close to that seen in the clean RR sediment tanks (7.8 (±0.5) ng/L). In

263

the untreated GR sediment tanks, porewater PCB concentrations were 3 to 7 fold higher than the

264

overlying water concentrations, indicating sediment as the PCB source to the water column

265

during the course of the experiment. Dichloro, and trichlorobiphenyls showed the highest flux

266

from untreated GR sediment into the overlying water (Table S3). This concentration gradient

267

was reversed in the AC-treated GR sediment tanks (Table S4) leading to a flux back into the

268

sediment, indicating that AC-treated sediment acts as a sink (Table S3) as also reported in field

269

observations by Beckingham and Ghosh.29

270 271

Comparison of 45-day and 90-day results (588 (±16) and 631(±23)) shows that porewater

272

concentrations in the untreated GR tanks were close to equilibrium after 45 days (t test, α=0.05). 9



273

The overlying water concentration however, increased from 96 ±(12) to 184 (±15) from 45 to 90

274

days likely due to continued sorption and gradual saturation of the sorptive surfaces with time

275

(pipes, sump, and charcoal bag) (Table S4). Porewater PCB concentration in control RR

276

sediments was low and similar to that in the AC-treated GR sediment. As illustrated in Figure 1,

277

control RR tanks received the same overlying water recirculation as the GR sediment tanks they

278

were adjacent to. PCB data at a congener level is shown in Tables S1 and S2 for porewater and

279

overlying water.

280 281

Although effort was made to avoid cross contamination between the GR and RR tanks by

282

placing an activated charcoal bag in each sump, PCBs in the effluent recycle were not totally

283

removed by bagged AC and therefore the recycled water carried some of the overlying water

284

PCBs between tanks. This was mainly an issue between the untreated GR tanks and the RR

285

control tanks adjacent to them (Table S4). However, the resulting elevated PCB concentration in

286

the recirculated overlying water does not impact the study because overlying water

287

concentrations were measured directly by passive sampling and in most contaminated field sites,

288

ongoing inputs to overlying water is a reality. In the treated GR tanks, the porewater PCB

289

concentration also remained similar in 45 and 90 days (t test, α=0.05) and was in the range of 1-2

290

ng/L. However, the overlying water PCB concentrations in the treated GR tanks increased from

291

3.5 ±0.7 to 7.6 ±2.5 ng/L (mostly contributed by dichlorobiphenyls see SI) and were similar to

292

the adjacent control tanks.

293 294

Partitioning coefficients (Kd) for untreated and treated GR sediments were predicted with

295

sorption models for 21 dominant congeners in water, representing di, tri, tetra, and

296

pentachlorobiphenyls (Figure 3). The observed Kd values for untreated GR sediment (based on

297

90-day porewater values) fell within the range predicted by Kd models with and without

298

including BC. Observed Kd values for the treated GR sediment were nearly 2 orders of

299

magnitude higher than untreated GR sediment, demonstrating the efficacy of AC amendment in

300

enhancing sorption capacity of the sediment and reducing aqueous PCB concentrations. There

301

was some scatter in the observed Kd values for the treated GR sediment partially due to the

302

uncertainty of measurements approaching analytical detection limits. For the untreated sediment,

303

the model based on sorption to natural organic matter (equation 2), underestimated sorption 10


Page 10 of 25

Page 11 of 25


304

(solid red line) while equation 4, which accounts for additional sorption to BC, overestimated

305

sorption to untreated GR sediment (dashed red line). Depending on the model used (equation 2

306

vs. 4) the predicted porewater concentrations can vary by one order of magnitude, highlighting

307

the previously discussed uncertainty associated with standard partition coefficients for estimating

308

porewater concentrations. PCB sorption in the treated sediment is dominated by sorption to AC.

309

Therefore, including or not including the BC pool in the sorption model (equations 3 and 5) did

310

not make a difference in the results for the treated sediment. The modeled results are shown in

311

discrete gray symbols because the relationship between log Kd and log KOW is no longer linear

312

after incorporation of Freundlich isotherm for AC (becomes concentration dependent). The

313

sorption models overestimated Kd for treated GR sediment likely due to uncertainty in estimating

314

actual sorption capacity of the AC in the sediment matrix.

315 316

Bioaccumulation in fish

317

The lipid content of the zebrafish based on wet weight was measured as 5.5±0.5%. PCB

318

concentration in zebrafish exposed to untreated and treated GR sediments as well as control RR

319

sediments are compared in Figure 4. The total PCB concentration in fish lipids was 27 ±1.3 µg/g

320

in untreated GR sediment tanks and 3.5 ±0.3 µg/g in treated GR sediment tanks after 90 days

321

exposure (87% reduction). Kupryianchyk et al.18 reported a 95% reduction in PCB uptake by

322

Golden Orfe fish after 6 months of exposure to sediment treated with powdered AC and only

323

45% reduction in the same fish with granular AC (0.425-1.7 mm). Observed reduction in fish

324

lipid PCBs in our study is consistent with our use of an intermediate particle size AC (75-300

325

µm) and shorter exposure period of 3 months. Fish PCB concentrations in control RR sediment

326

tanks were in the range of what was measured for the fish in treated GR sediment tanks (1.8 ±0.6

327

to 7.4 ±2.8). As discussed before, the elevated PCB levels in the fish from the control tanks is

328

likely due to a combination of trace PCB levels in the RR sediment and a result of ongoing

329

inputs from the overlying water of the adjacent tanks containing GR sediment. PCB results in the

330

fish at a congener level are shown in Tables S1 and S2. The change in fish concentrations over

331

time is shown in Figure S5. The variation between PCB levels in the fish at day 45 and day 90

332

shows that PCB concentrations may not have reached equilibrium in the fish. Plots of log (BCF)

333

vs. log (KOW) for fish exposed to untreated sediment are shown in Figure S6. The slopes of the

334

line fitted to the observed data from 45 and 90 days were close to unity and an intercept close to 11



335

but less than zero indicating near octanol-like partitioning in the lipids. The observed BCF values

336

for untreated GR sediment fell within the range predicted by modeled values based on Di Toro et

337

al.30 and Gobas.31

338 339

Equilibrium model

340

Overlying water concentrations measured by passive samplers were incorporated in the

341

equilibrium model (assuming Klipid = KOW) to predict PCB residues in fish for 34 dominant

342

congeners (equation 8; Figure 5a). Since fish were mainly in contact with the water column,

343

predictions were based on the overlying water concentrations. The big reduction in uptake by

344

fish after treatment with AC was captured well by the model. There was a reasonable agreement

345

between observed and predicted PCB concentrations in zebrafish exposed to the treated GR

346

sediment. However, prediction of uptake for fish exposed to untreated GR sediment was higher

347

than observed. This could be either due to overestimation of Klipid values (assumed to be KOW) or

348

non-equilibrium conditions.

349 350

Kinetic model

351

Overlying water concentrations were used to predict uptake of the same 34 congeners used for

352

the equilibrium model. Treatment trends were predicted well by the Arnot and Gobas1 kinetic

353

model (equation 11; Figure 5b) and the root mean squared error was smaller than the values for

354

the equilibrium model (Table S5). The predicted total PCBs for the fish exposed to untreated

355

GR sediment exceeded the observed values by a factor of 2 (ranged from 0.1-8 for individual

356

congeners). However, for the fish exposed to treated GR sediment the predicted total PCBs were

357

lower than the observed values by a factor of 2 (ranged from 0.3-15 for individual congeners). It

358

is important to recognize that the model parameters were not fitted to the experimental results

359

but obtained from existing literature. The use of empirical correlations for estimating the model

360

parameters such as uptake efficiency and filtration rate (equations 2 and 3 in the Supporting

361

Information) is one possible reason for the discrepancy between the observed and predicted

362

values. These empirical correlations in the literature were approximated based on observations

363

over a range of fish species, which leads to variations for individual species. Further, a generic

364

fish fecal egestion rate constant was used (equation 5 in the Supporting Information) which

365

likely does not capture species-specific differences. Overall, the kinetic model resulted in better 12


Page 12 of 25

Page 13 of 25


366

predictions than the equilibrium model because it accounted for the time variable nature of

367

uptake and loss processes (Figure S7).

368 369

The sensitivity of the kinetic model output (Clipid) with respect to each parameter was tested for

370

conditions in the untreated GR tanks after 90 days (Figure S8). A positive sensitivity value

371

indicates a direct relationship between the input parameter and the output and a negative value

372

indicates an inverse relationship. Sensitivities to the positive and negative changes were identical

373

for overlying water concentration (CW,O) and similar for all congeners. However, the trends

374

deviated for dissolved oxygen concentration (DO), fish wet weight (WB) and lipid fraction (Lf)

375

due to the non-linear change of model output with respect to these parameters. For the lower

376

chlorinated PCBs (log (KOW) less than 6), the model was most sensitive to CW,O. For log (KOW)

377

larger than 6, the model was most sensitive to Lf, and DO and WB. For all congeners DO

378

concentration was inversely related to the uptake as high DO reduced ventilation rate. However,

379

this effect was less pronounced for the lower chlorinated PCBs, likely due to the faster exchange

380

kinetics of these compounds. In contrast, for the higher chlorinated PCBs with slower mass

381

transfer, oxygen concentration was a rate limiting factor in fish uptake. Overall, the sensitivity of

382

the model to WB was lower compared to other three parameters over a large range of KOW. The

383

low sensitivity to changes in WB was also reported for Gobas 29 and Thomann et al.31, 32 models.

384

Generally, the kinetic model over-predicted the uptake of the low KOW compounds and under-

385

predicted the uptake of the high KOW compounds (see Figure S9). There are a few possible

386

explanations to this observation: 1) uncertainties in measuring aqueous concentrations and

387

estimating fish ventilation rates, 2) model parameters have dependencies on KOW that may not be

388

accurate, 3) some higher chlorinated PCBs may be coming from ingestion of sediment and/or

389

food that picked up PCBs in the aquaria. To address the issue of ingestion exposure, equation 9

390

was modified by the addition of incidental ingestion of food/sediment. Exposure through

391

ingestion was calibrated (separately for the untreated and treated GR sediment) by accounting for

392

the observed uptake of the dominant heptachlorobiphenyl PCB-180 that could not be explained

393

by uptake solely from water (Details in the Supporting Information). Overall, including the

394

ingestion exposure increased predicted uptake in fish, especially for the high KOW compounds

395

(Figure 5c). Including ingestion pathway affected the predictions for treated sediment more due

396

to the fact that water concentrations were lower in the treated GR tanks, causing ingestion of 13



397

PCBs to play a significant role in uptake. The relative contribution from the ingestion pathway to

398

overall PCB uptake increased with increasing PCB hydrophobicity due to lower solubility and

399

thus lower exposure of the heavier PCBs through water (Figure S10). Gill uptake followed the

400

opposite trend as respiration was the dominant exposure pathway for the more water soluble

401

PCBs. Although net uptake was greatly reduced after sediment treatment, the percent

402

contribution from the ingestion pathway to the total PCB uptake in fish was 17% and 69% for

403

untreated and treated sediment, respectively. This is due to lower aqueous PCB concentrations

404

and thus gill uptake becoming a less significant pathway (congener-specific contributions are

405

shown in Figure S10). Assuming that the fish exposed to untreated and treated GR sediment had

406

the same ingestion rates, the ratio of assimilation efficiencies of untreated to treated sediment-

407

bound PCBs was calculated to be 2, indicating a reduction in assimilation efficiency of PCBs in

408

the sediment upon amendment with AC.

409 410

Arnot and Gobas1 and Connolly28 models were similar in terms of performance with root mean

411

squared error values being close to each other. It should be emphasized that the predicted values

412

generated by both models were in good agreement with the observed values (Figures 5b and

413

S11) despite the fact that these models were not calibrated to the data. This highlights the broad

414

applicability of such bioaccumulation models to a wide variety of environmental and biological

415

conditions. The kinetics of uptake were modeled for congeners from tri to hexa groups, assuming

416

constant water concentrations (Figure S12). Both models predict faster equilibrium times for

417

lower chlorinated congeners than the higher ones due to higher mass transfer rates. Overall, the

418

Arnot and Gobas model predicts shorter equilibration times due to faster exchange kinetics

419

estimated by this model.

420 421

Percent reduction in aqueous PCBs and bioaccumulation

422

Observed percent reductions in porewater, overlying water, and fish concentrations 90 days after

423

amendment with AC are shown in Figure 6. For porewater, overlying water, and fish the effect

424

of the treatment after 90 days was most pronounced on congeners with log (KOW) less than 7.

425

This is explained by faster mass transfer kinetics from sediment to AC for lower chlorinated

426

compared to the higher chlorinated PCBs and greater bioavailability of these compounds.

427

Porewater and overlying water showed similar reductions over the log (KOW) range because 14


Page 14 of 25

Page 15 of 25


428

overlying water concentration is controlled by the flux from sediment and therefore reductions in

429

porewater concentrations translate to reductions in overlying water. It is noteworthy that the

430

observed reduction in fish was less than reductions observed in the overlying water, likely due to

431

the concentrations in fish having not reached equilibrium in 90 days. To confirm this hypothesis,

432

long-term simulations were conducted with the kinetic model for 120,150,180 and 210 days of

433

exposure. The predictions for the 180 and 210-day exposures were not statistically different.

434

Therefore, the 210-day results are shown in Figure 6. The predicted reduction in fish at

435

equilibrium (green triangles) matched with reductions in porewater and overlying water. Thus,

436

in the long-term in the field, the percent reductions observed in the porewater and water column

437

is expected to be reflected in reductions in fish (in the absence of other ongoing inputs to the

438

system).

439 440 441 442

IMPLICATIONS OF THIS RESEARCH

443

organisms can be reduced by amending with strong sorbents that attenuate sediment porewater

444

concentrations, there is a lack of data on how fish responds to bioavailability changes in

445

sediment. Results presented here confirm that indeed bioavailability changes in sediment are

446

reflected in uptake in fish, primarily through reductions in PCB flux from sediments. In this

447

work we explain those observations also through direct measurement of freely dissolved

448

porewater and overlying water concentrations and modeling uptake pathways to fish to

449

mechanistically explain the experimental results. The research also addresses a key challenge in

450

monitoring effectiveness of in-situ remedies by demonstrating that by targeting assessment of

451

exposure pathways to fish and measuring freely dissolved concentrations in overlying water

452

using passive sampling, we can make reasonable assessments of long-term recovery of PCB

453

residues in fish. However, it is noteworthy that these direct measurements will not replace freely

454

dissolved concentrations computed from fate and transport models when long-term predictions

455

are needed. Bioaccumulation models can be linked to fate and transport models which rely on

456

robust partitioning estimates and include contributions from ongoing inputs in the field. Once

457

linked, this combination can capture the effect of ongoing external PCB loads 33, 34 and predict

458

PCB concentrations in fish tissue over time. After validation, such models can also be used as a

While numerous studies in the past have demonstrated that PCB accumulation in benthic

15



459

reliable means of back-calculating the AC amendment dose required to achieve a desired

460

endpoint in recovery of fish tissue PCB levels.

461 462 463

ACKNOWLEDGEMENT

464

We would like to thank the National Institute of Environment and Health Sciences, Superfund

465

Research Program for financial support (Grant # R01ES020941). We thank Larry McShea from

466

Alcoa for providing the PCB-impacted sediments used in this research. UG is a co-inventor of

467

two patents related to the technology described in this paper for which he is entitled to receive

468

royalties. One invention was issued to Stanford University (US Patent # 7,101,115 B2), and the

469

other to the University of Maryland Baltimore County (UMBC) (U.S. Patent No. 7,824,129). In

470

addition, UG is a partner in a startup company (Sediment Solutions) that has licensed the

471

technology from Stanford and UMBC and is transitioning the technology in the field.

472

This is contribution # 15-166 from the Institute of Marine and Environmental Technology and

473

contribution # 5019 from the University of Maryland Center for Environmental Sciences.

474

16


Page 16 of 25

Page 17 of 25


475

References

476 477 478

1. Arnot, J. A.; Gobas, F. A. P. C., A food web bioaccumulation model for organic chemicals in aquatic ecosystems. Environmental Toxicology and Chemistry 2004, 23, (10), 23432355.

479 480 481

2. Reichenberg, F.; Mayer, P., Two complementary sides of bioavailability: Accessibility and chemical activity of organic contaminants in sediments and soils. Environmental Toxicology and Chemistry 2006, 25, (5), 1239-1245.

482 483

3. NRC, Bioavailability of Contaminants in Soils and Sediments:Processes, Tools, and Applications. The National Academies Press: Washington, DC, 2003.

484 485 486 487

4. Luthy, R. G.; Aiken, G. R.; Brusseau, M. L.; Cunningham, S. D.; Gschwend, P. M.; Pignatello, J. J.; Reinhard, M.; Traina, S. J.; Weber, W. J.; Westall, J. C., Sequestration of Hydrophobic Organic Contaminants by Geosorbents. Environmental Science & Technology 1997, 31, (12), 3341-3347.

488 489 490

5. Ghosh, U.; Luthy, R. G.; Cornelissen, G.; Werner, D.; Menzie, C. A., In-situ Sorbent Amendments: A New Direction in Contaminated Sediment Management. Environmental Science & Technology 2011, 45, (4), 1163-1168.

491 492 493

6. Sun, X.; Ghosh, U., PCB Bioavailability Control in Lumbriculus Variegatus through Different Modes of Activated Carbon Addition to Sediments. Environmental Science & Technology 2007, 41, (13), 4774-4780.

494 495 496 497

7. Cho, Y.-M.; Ghosh, U.; Kennedy, A. J.; Grossman, A.; Ray, G.; Tomaszewski, J. E.; Smithenry, D. W.; Bridges, T. S.; Luthy, R. G., Field Application of Activated Carbon Amendment for In-Situ Stabilization of Polychlorinated Biphenyls in Marine Sediment. Environmental Science & Technology 2009, 43, (10), 3815-3823.

498 499

8. EPA, Use of Amendments for In Situ Remediation at Superfund Sediment Sites; Office of Superfund Remediation and Technology Innovation: 2013.

500 501

9. EPA, Statement of Basis for EPA’s Proposed Remedial Action for the Housatonic River “Rest of River”; Region 1 Report: 2014.

502 503

10. Anchor QEA, The Upper Hudson River PCB Modeling System, Hudson River PCBs Superfund Site. Prepared for General Electric Corporation: 2010.

504 505

11. EPA, Model Calibration: Modeling Study of PCB Contamination in the Housatonic River; Region 1 Report: 2004.

506 507

12. Alcoa, Modeling PCB Fate in the Lower Grasse River, Appendix A of the Lower Grasse River Analysis of Alternatives Report; submitted to USEPA Region 2: 2010.

508 509

13. Cornelissen, G.; Gustafsson, Ö.; Bucheli, T. D.; Jonker, M. T. O.; Koelmans, A. A.; van Noort, P. C. M., Extensive Sorption of Organic Compounds to Black Carbon, Coal, and Kerogen 17



Page 18 of 25

510 511

in Sediments and Soils:  Mechanisms and Consequences for Distribution, Bioaccumulation, and Biodegradation. Environmental Science & Technology 2005, 39, (18), 6881-6895.

512 513 514

14. Lohmann, R.; MacFarlane, J. K.; Gschwend, P. M., Importance of Black Carbon to Sorption of Native PAHs, PCBs, and PCDDs in Boston and New York Harbor Sediments. Environmental Science & Technology 2005, 39, (1), 141-148.

515 516

15. Mayer, P.; Tolls, J.; Hermens, J. L.; Mackay, D., Equilibrium sampling devices. Environmental Science & Technology 2003, 37, (9), 184A-191A.

517 518 519

16. Lohmann, R., Critical Review of Low-Density Polyethylene’s Partitioning and Diffusion Coefficients for Trace Organic Contaminants and Implications for Its Use As a Passive Sampler. Environmental Science & Technology 2011, 46, (2), 606-618.

520 521 522

17. Werner, D.; Hale, S. E.; Ghosh, U.; Luthy, R. G., Polychlorinated Biphenyl Sorption and Availability in Field-Contaminated Sediments. Environmental Science & Technology 2010, 44, (8), 2809-2815.

523 524 525

18. Kupryianchyk, D.; Rakowska, M. I.; Roessink, I.; Reichman, E. P.; Grotenhuis, J. T. C.; Koelmans, A. A., In situ Treatment with Activated Carbon Reduces Bioaccumulation in Aquatic Food Chains. Environmental Science & Technology 2013, 47, (9), 4563-4571.

526 527 528

19. Beckingham, B.; Buys, D.; Vandewalker, H.; Ghosh, U., Observations of limited secondary effects to benthic invertebrates and macrophytes with activated carbon amendment in river sediments. Environmental Toxicology and Chemistry 2013, 32, (7), 1504-1515.

529 530 531

20. Hale, S. E.; Kwon, S.; Ghosh, U.; Werner, D., Polychlorinated Biphenyl Sorption to Activated Carbon and the Attenuation Caused by Sediment. Global NEST Journal 2010, 12, 318326.

532 533 534

21. Hawthorne, S. B.; Miller, D. J.; Grabanski, C. B., Measuring Low Picogram Per Liter Concentrations of Freely Dissolved Polychlorinated Biphenyls in Sediment Pore Water Using Passive Sampling with Polyoxymethylene. Analytical Chemistry 2009, 81, (22), 9472-9480.

535 536 537

22. Beckingham, B.; Ghosh, U., Field-Scale Reduction of PCB Bioavailability with Activated Carbon Amendment to River Sediments. Environmental Science & Technology 2011, 45, (24), 10567-10574.

538 539 540

23. Harvey, H. R.; Eglinton, G.; O'Hara, S. C. M.; Corner, E. D. S., Biotransformation and assimilation of dietary lipids by Calanus feeding on a dinoflagellate. Geochimica et Cosmochimica Acta 1987, 51, (11), 3031-3040.

541 542

24. Grossman, A.; Ghosh, U., Measurement of activated carbon and other black carbons in sediments. Chemosphere 2009, 75, (4), 469-475.

543 544

25. Hawker, D. W.; Connell, D. W., Octanol-water partition coefficients of polychlorinated biphenyl congeners. Environmental Science & Technology 1988, 22, (4), 382-387.

18


Page 19 of 25


545 546 547 548

26. Gomez-Eyles, J. L.; Yupanqui, C.; Beckingham, B.; Riedel, G.; Gilmour, C.; Ghosh, U., Evaluation of Biochars and Activated Carbons for In Situ Remediation Of Sediments Impacted With Organics, Mercury, And Methylmercury. Environmental Science & Technology 2013, 47, (23), 13721-13729.

549 550 551

27. Werner, D.; Ghosh, U.; Luthy, R. G., Modeling Polychlorinated Biphenyl Mass Transfer after Amendment of Contaminated Sediment with Activated Carbon. Environmental Science & Technology 2006, 40, (13), 4211-4218.

552 553 554

28. Connolly, J. P., Application of a food chain model to polychlorinated biphenyl contamination of the lobster and winter flounder food chains in New Bedford Harbor. Environmental Science & Technology 1991, 25, (4), 760-770.

555 556 557

29. Beckingham, B.; Ghosh, U., Polyoxymethylene passive samplers to monitor changes in bioavailability and flux of PCBs after activated carbon amendment to sediment in the field. Chemosphere 2013, 91, (10), 1401-1407.

558 559 560

30. Di Toro, D. M.; McGrath, J. A.; Hansen, D. J., Technical basis for narcotic chemicals and polycyclic aromatic hydrocarbon criteria. I. Water and tissue. Environmental Toxicology and Chemistry 2000, 19, (8), 1951-1970.

561 562

31. Gobas, F. A. P. C., A model for predicting the bioaccumulation of hydrophobic organic chemicals in aquatic food-webs: application to Lake Ontario. 1993, 69, (Issues 1–2), 1–17.

563 564 565

32. Thomann, R. V.; Connolly, J. P.; Parkerton, T. F., An equilibrium model of organic chemical accumulation in aquatic food webs with sediment interaction. Environmental Toxicology and Chemistry 1992, 11, (5), 615-629.

566 567

33. Connolly, J.; Rhea, J.; Benaman, J. Effective decision-making models for evaluating sediment management options; Quantitative Environmental Analysis, LLC: 1999.

568 569 570

34. SERDP, Improved Understanding of the Impact of Ongoing, Low Level Contaminant Influx to Aquatic Sediment Site Restoration; Strategic Environmental Research and Development Program: 2012.

571

19



Figure 1. Aquaria setup to study the uptake of PCBs in fish. RR: clean Rhode River sediment; GR untreated: PCB impacted sediment from Grasse River; GR treated: PCB impacted sediment from Grasse River mixed with 4.5% AC in the laboratory.

Figure 2. Freely dissolved PCB concentration in (a) sediment porewater and (b) overlying water of untreated and treated Grasse River sediments after 90 days. Error bars represent standard error.

1


Page 20 of 25

Page 21 of 25


Figure 3. Observed and predicted Kd values versus KOW for untreated (U) and treated (T) Grasse River sediments after 90 days. The red lines and the gray closed symbols represent the predicted values obtained from the partitioning models.

2



Figure 4. PCB concentration in zebrafish exposed to (a) untreated and treated Grasse River sediments and (b) Rhode River sediment for 90 days. Rhode River-A represents data from tanks adjacent to untreated Grasse River sediments and Rhode River-B represents data from tanks adjacent to the treated Grasse River sediments (see Figure 1). Error bars represent standard error.

3


Page 22 of 25

Page 23 of 25


Figure 5. Observed and predicted PCB concentrations in zebrafish using the (a) equilibrium model (b) Arnot and Gobas bioaccumulation model without ingestion, and (c) Arnot and Gobas bioaccumulation model with ingestion. Closed symbols refer to 45 days results and open symbols refer to 90 days. 4



Figure 6. Percent reduction in water and fish concentrations after AC amendment.

5


Page 24 of 25

Page 25 of 25


PCB in fish (µg/g lipid)

30

20

87% reduction

10

0 Before AC treatment

After AC treatment


Effect of PCB Bioavailability Changes in Sediments on

Recommend Documents