Measuring and Modeling Organochlorine Pesticide Response to

Apr 4, 2016 - Chevron Energy Technology Company, Richmond, California 94801, ... Department of Integrative Biology, University of California, Berkeley...
1 downloads 0 Views 2MB Size
Subscriber access provided by NORTHERN KENTUCKY UNIV STEELY

Article

Measuring and Modeling Organochlorine Pesticide Response to Activated Carbon Amendment in Tidal Sediment Mesocosms Jay M Thompson, Ching-Hong Hsieh, Thomas P Hoelen, Donald P. Weston, and Richard G Luthy Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.5b05669 • Publication Date (Web): 04 Apr 2016 Downloaded from http://pubs.acs.org on April 5, 2016

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 32

Environmental Science & Technology

1

Measuring and Modeling Organochlorine Pesticide Response to Activated Carbon

2

Amendment in Tidal Sediment Mesocosms

3 4 5

JAY M. THOMPSON†, CHING-HONG HSIEH†, THOMAS P. HOELEN‡, DONALD P.

6

WESTON§, RICHARD G. LUTHY*,†

7 8



9

94305-5080, USA

Department of Civil and Environmental Engineering, Stanford University, Stanford, CA

10



Chevron Energy Technology Company, San Ramon, CA 94583, USA

11

§

Department of Integrative Biology, University of California, Berkeley, CA 94720-3140,

12

USA

13 14

Corresponding author:

15

Richard G. Luthy

16

Stanford University Room 191, Yang & Yamazaki Environment & Energy Building,

17

473 Via Ortega, Stanford, California 94305-4020

18

Email: [email protected]

19

Telephone: 650-721-2615

20

Fax: 650-725-9720

21

ACS Paragon Plus Environment

Environmental Science & Technology

22

ABSTRACT

23

Activated carbon (AC) sediment amendment for hydrophobic organic contaminants

24

(HOCs) is attracting increasing regulatory and industrial interest. However, mechanistic

25

and well-vetted models are needed. Here, we conduct an 18-month field mesocosm trial

26

at a site containing dichlorodiphenyltrichloroethane (DDT) and chlordane. Different AC

27

applications were applied and, for the first time, a recently-published mass transfer model

28

was field tested under varying experimental conditions. AC treatment was effective in

29

reducing DDT and chlordane concentration in polyethylene (PE) samplers, and

30

contaminant extractability by Arenicola brasiliensis digestive fluids. A substantial AC

31

particle size effect was observed. For example, chlordane concentration in PE was

32

reduced by 91% 6-months post treatment in the powdered AC (PAC) mesocosm,

33

compared with 68% in the GAC mesocosm. Extractability of sediment-associated DDT

34

and chlordane by A. brasiliensis digestive fluids was reduced by at least a factor of ten in

35

all AC treatments. The model reproduced the relative effects of varying experimental

36

conditions (particle size, dose, mixing time) on concentrations in polyethylene passive

37

samplers well – in most cases within 25% of experimental observations. While

38

uncertainties such as the effect of long-term AC fouling by organic matter remain, the

39

study findings support the use of the model to assess long-term implications of AC

40

amendment.

41

ACS Paragon Plus Environment

Page 2 of 32

Page 3 of 32

Environmental Science & Technology

42

Introduction

43

As activated carbon (AC) amendment of sediments contaminated by hydrophobic organic

44

contaminants (HOCs) matures as a remedial technology1-6, there has been increasing

45

emphasis7 on collecting field data8-11 and on developing mechanistic models describing

46

mass transfer5, 12-15 in AC-amended sediments. The former allows researchers to assess

47

the feasibility of AC amendment at scale and effectiveness in a natural environment. The

48

latter is necessary to understand long-term AC performance in the field. Such an

49

understanding is needed to create effective remedial designs, as there can be significant

50

differences between field and laboratory performance. For example, Zimmerman et al.

51

reported 3.4% (m/m dw) AC reduced polychlorinated biphenyl (PCB) pore water

52

concentration in Hunters Point Shipyard (HPS), CA, USA sediment by 92% after 6

53

months of continuous mixing under laboratory conditions1. However, Cho et al.,

54

observed only a 50% reduction in sediment pore water PCB concentration 6 months post-

55

treatment at the pilot-scale 8, 9. In contrast Beckingham and Ghosh reported reductions of

56

69-99% in Lumbriculus variegatus tissue PCB concentration 3 years post-treatment in a

57

pilot study at the Grasse River, NY, USA10. Beckingham and Ghosh attribute the

58

difference in kinetics between the two pilot studies to differences in sediment

59

characteristics and PCB congener mix, manifested, for example, in the different sediment

60

desorption rates10. These cases underscore the need for models to understand site-specific

61

effects and treatment conditions on AC performance.

62 63

Choi et al., presented a mass transfer model (itself an extension of published AC models12,

64

13, 16

) describing transfer processes governing AC amendment: advection, dispersion, and

ACS Paragon Plus Environment

Environmental Science & Technology

65

intra-particle diffusion14, 17. The model predicted pore water concentrations of PCBs and

66

alkylated PAHs to within a factor of two compared to column tests. However, model

67

field-testing is needed for two reasons18. Firstly, column studies only approximate field

68

conditions and cannot encompass all field variables. Secondly, field validation will

69

encourage the model’s widespread application.

70 71

One challenge with field validation is scale. Effective model validation requires that

72

several experimental conditions be varied. If treatment plots are sized in the tens to

73

hundreds of m2 range that is typical in literature,

74

needed. For example, Cho et al. applied the aforementioned mass transfer model to

75

several 34 m2 field plots, but did not systematically vary AC application conditions15.

76

Here, field mesocosms were used to achieve the study goals without a large treatment

77

area and its concomitant resource requirement.

78

This work has three major outcomes. First, we observed several treatment condition

79

effects on AC performance with respect to organochlorine pesticide concentration in pore

80

water (as estimated by passive sampler observations) and bioavailability. Second, we

81

validated the Choi et al. mass transfer model under field conditions. Third, we assessed

82

the long-term site implications after AC amendment. This study is the first field test of

83

the Choi et al. model14 under varying experimental conditions.

9-11

a large and complex field effort is

84 85

Materials and Methods

86

Materials.

87

Pesticide-grade solvents were purchased from Fischer Scientific (Fair Lawn, NJ). Ten

ACS Paragon Plus Environment

Page 4 of 32

Page 5 of 32

Environmental Science & Technology

88

organochlorine pesticides were examined, cis- and trans-chlordane, trans-nonachlor, 2,4′-

89

and

90

dichlorodiphenyldichloroethylene, and dichlorodiphenyldichloroethane, respectively),

91

and 4,4′-DDMU (dichlorodiphenylmonochloroethylene). Pesticide standards were

92

purchased from Ultra Scientific (North Kingstown, RI). Activated carbon (Type-TOG,

93

Calgon Carbon, Pittsburgh, PA) was used as received (median diameter, D50 = 219 µm,

94

measured by laser diffraction, Particle Technology Labs, Downers Grove, IL) for

95

mesocosms amended with granular activated carbon (GAC). The same material was

96

ground to a powder (D50 = 42 µm) using a stone mill for powdered activated carbon

97

(PAC) application.

4,4′-DDT,

DDE,

and

DDD

(dichlorodiphenyltrichloroethane,

98 99

Field Site and Study Design.

100

The field site for this study was a tidal channel within an industrial property in Northern

101

California, USA, with sediments containing the organochlorine pesticides DDT and

102

chlordane. Partially enclosed mesocosms were used to study the effect of varying AC

103

treatment conditions. Mesocosms were fabricated from 61-cm (24-inch nominal)

104

diameter high-density polyethylene (HDPE) pipe cut to 0.90 m lengths. Twelve 5-cm

105

diameter holes were drilled into each mesocosm as shown in Figure 1 to allow water

106

exchange with channel waters. Photographs are available in the Supporting Information,

107

Figure S1.

108 109

The study employed six mesocosms: one mixed control, one unmixed control, and four

110

mixed AC treatments. The treatment conditions within the four AC-amended mesocosms

ACS Paragon Plus Environment

Environmental Science & Technology

111

were 5% (m/m dw) GAC and PAC mixed for 10 minutes, 10% GAC mixed for 10

112

minutes, and 5% GAC mixed for 20 minutes. An additional 50% of AC over nominal

113

values were used to account for potential overmixing into deeper sediment strata than

114

intended and ensure that the target AC dose was delivered to the mixing zone (i.e., 7.5%

115

AC was added to ensure that a 5% dose was achieved in the target depth interval). Note

116

these levels were selected for mass transport model validation and were not intended to

117

explore the whole range of typical deployment scenarios. For example, AC doses above

118

5% are rarely used in the field and particle size much larger than studied here have been

119

employed in field studies7, 19. A schematic diagram of the experiment is available in the

120

Supporting Information, Figure S2. The mixed control mesocosm was mixed for 10

121

minutes without AC. The unmixed control mesocosm was not mixed or amended.

122 123

Mesocosms were deployed in November 2012. Prior to placing the mesocosms,

124

vegetation lining the creek bank was scraped away with an excavator (Figure S3). The

125

mesocosms were then positioned adjacent to one another and pushed into the sediment by

126

the excavator. AC was then deposited into the appropriate mesocosms and mixed into the

127

upper 15 cm of the sediment for the prescribed duration (10 or 20 minutes), with a dual

128

head electric mortar mixer (Collomix CX-44 DUO, Gaimersheim, Germany).

129

Photographs of the preparation of the area and placement of the mesocosms are available

130

in the Supporting Information, Figure S3.

131 132

Mesocosm Deployment and Field Sampling.

133

Three observations were made: one baseline, one 6-months post-treatment, and one 18-

ACS Paragon Plus Environment

Page 6 of 32

Page 7 of 32

Environmental Science & Technology

134

months post-treatment, commencing February 2013, May 2013, and May 2014,

135

respectively. The duration between AC deployment and sampling events was made on

136

the basis of a short proof-of-concept field study. A timeline of field events is presented in

137

the Supporting Information, Figure S4. The baseline measurements, consisting of six

138

sediment cores and ten low density polyethylene (PE) samplers, were taken immediately

139

outside of the mesocosms. At each post-treatment observation, two PE samplers were

140

deployed inside each mesocosm, at least 15 cm from the mesocosm wall and 30 cm from

141

each other. Mass transport modeling indicated that the distance between the HDPE pipe

142

and PE samplers was sufficient to avoid interference (see Supporting Information). As no

143

attempt was made to convert PE concentrations to pore water concentration, performance

144

reference compounds were not added to the PE prior to deployment. Rather, PE

145

concentrations are reported relative to measurements in untreated control mesocosms and

146

baseline measurements. Further details on polyethylene passive samplers are available

147

elsewhere20, 21. Upon retrieval of each PE sampler, a sediment core sample was collected

148

at the same location.

149 150

The six mesocosms and three sampling events were used to test the effect of AC particle

151

size, AC particle dose, AC mixing time, and deployment duration on amendment

152

performance. This experimental design and allocation of sampling allowed for a number

153

of comparisons to be made: inside the mesocosms with outside the mesocosms, control

154

mesocosms with treatment mesocosm, treated zone with untreated zone within a

155

mesocosm, and temporal trends. However, the spatial density of sampling precluded

156

replicate mesocosms or numerous temporal observations. Instead, spatial variability

ACS Paragon Plus Environment

Environmental Science & Technology

157

within a mesocosm was addressed by placing two samplers (each sectioned to 12

158

segments) in each mesocosm for each sampling event. Spatial variability within the entire

159

study area was addressed with the baseline measurements and modeling (Supporting

160

Information Figure S5-S7). While greater certainty could be obtained with baseline

161

measurements in tandem with duplicate or triplicate mesocosms, such a study design

162

would result in an unmanageable number of samples if the same sample depth resolution

163

was retained.

164 165

For passive sampler measurements, a polyethylene strip (52 µm, 0.92 g/cm3, Brentwood

166

Plastics, St. Louis, MO) was wrapped around a slotted stainless steel frame. The

167

approximate strip dimensions were 90 × 2 cm, which resulted in sampling over a 45-cm

168

depth. Samplers were embedded in the sediment for 28 days. Upon retrieval, samplers

169

were cleaned with a lint-free tissue, sectioned to 2.5 and 5.0 cm depth intervals (within

170

the 0-15 and 15-45 cm intervals, respectively), wrapped in aluminum foil and stored in a

171

glass jar at 4 ºC. Samples were held for 15 cm) strata, were extracted using gut fluid from the polychaete Arenicola

186

brasiliensis. Details are in the SI, but briefly, the approach is an in vitro assay that

187

quantifies how much of the sediment-associated contaminant may be desorbable from the

188

particles during passage through a deposit-feeder’s gut, and would thus be available for

189

uptake23-27. Specific extraction procedures (e.g., dose, extraction time) are as described by

190

Weston et al. with some minor modifications28. Gut fluid from A. brasiliensis (0.9 ml)

191

was mixed with wet sediment equivalent to 0.3 g dry weight. After a 2.5 h extraction, the

192

gut fluid was recovered and analyzed for the organochlorine pesticides of interest. In

193

parallel with the gut fluid extractions, the sediments were also extracted with a sodium

194

taurocholate solution, a synthetic cocktail designed to mimic the contaminant

195

solubilization properties of actual gut fluid29 and with artificial seawater as a control. The

196

purpose of the sodium taurocholate extractions was to test it as a potential compliment to

197

gut fluid, given the difficulty in collecting the latter. The contaminant extraction

198

efficiency of all fluids was expressed as percent solubilization (mass of contaminant in

199

the extractant as a proportion of that initially in the dry sediment extracted). Initial

200

concentrations in each sediment stratum were determined from the mean of the five to six

201

baseline core samples (Table S1).

202

ACS Paragon Plus Environment

Environmental Science & Technology

Page 10 of 32

203

Model Description.

204

The mass transfer model employed in this study is described elsewhere14. Briefly, HOC

205

transport in AC amended sediment is modeled as the sum of three processes: intraparticle

206

diffusion in sediment and AC, dispersion in sediment pore water, and advection. This is

207

described by Equation 1.

208            =  +  +  − 

        − −

 #   3        +    " $%    ! !

(1)

&'  3 ()  *+ 2-! *+     &' % − . − 0   &' !  *+ /*+

209 210

where CW is the aqueous concentration; Sf, Ss, SAC, and SPE are the volumetric

211

concentrations (g cm-3) in fast- and slow-releasing sediment fractions, in AC, and in PE,

212

respectively; Rf, Rs, and RAC particle diameters (cm) of fast- and slow-releasing sediment

213

particles (collectively, Rsed) and of AC, respectively; VW, Vsed, VAC and VPE are the

214

volumes (cm3) of water, sediment, AC and PE, respectively; Dx, Dy, Dz are the dispersion

215

coefficients (cm2 s-1) in each dimension; Rsed and RAC are the radii (cm) of sediment and

216

AC, respectively; k0 (cm s-1) is the overall mass transfer coefficient for the PE-water

217

boundary layer; xPE is the PE thickness (cm); vsz is the porewater seepage velocity (cm s-

218

1

) and KPE is the PE-water partitioning coefficient (cm3 g-1).

219

ACS Paragon Plus Environment

Page 11 of 32

Environmental Science & Technology

220

The system is modeled over two periods: an initial well-mixed period and a stagnant

221

period. During the well-mixed phase that simulates brief initial mechanical mixing, mass

222

transfer is a function of intraparticle diffusion12. The results of the well-mixed period

223

become initial conditions to the stagnant period in which advection-dispersion,

224

intraparticle diffusion, and particle distributions are considered.

225 226

Model Parameterization.

227

Site-specific modeling parameters were determined in laboratory experiments as

228

indicated in Table S2, with literature values used when appropriate. Aqueous equilibrium

229

pesticide concentrations in the unamended sediment were obtained by agitating triplicate

230

sediment slurries with PE samplers on an orbital shaker at 80 rpm for 28 days. The

231

aqueous pesticide concentration can be estimated as CW = CPE/KPE, where KPE is the PE-

232

water partition coefficient.

233 234

It was assumed that AC was homogenously distributed and that AC-water partitioning

235

coefficients derived in clean water apply. Although fouling by dissolved organic mater

236

(potentially due to pore blocking at the AC external surface30, 31) has been shown to

237

reduce the activated carbon partitioning coefficient (KAC) of hydrophobic compounds by

238

a factor of 100 or more6, most of the literature on similar systems report attenuation by a

239

factor of approximately ten or less14,

240

attenuation at a similar magnitude would not affect the response in pore water

241

concentration for the duration of the experiment – see Supporting Information.

31-34

. A sensitivity analysis indicated that KAC

242

ACS Paragon Plus Environment

Environmental Science & Technology

243

Sediment desorption rates were determined in triplicate well-mixed sediment slurries

244

exposed to Tenax TA (60/80 mesh, Sigma) as described by Werner et al12. After

245

predetermined periods from 6 h to 160 d, the Tenax was collected and replaced. The

246

Tenax was extracted in hexane and the sediment desorption rate determined by fitting the

247

pesticide mass in Tenax to a known radial diffusion model35. Full details of desorption

248

and aqueous equilibrium procedures are available in the Supporting Information. Pore

249

water velocity was conservatively assumed to be zero. Literature reports suggest that pore

250

water advection-dispersion does not significantly affect HOC mass transport in AC-

251

amended sediments in tidally-influenced environments16, 17.

252 253

Results and Discussion

254

Baseline Assessment.

255

The average DDT concentration in sediment (reported as the sum of 2,4′- and 4,4′-DDT,

256

DDE, and DDD and 4,4′-DDMU) was 296 ± 35 µg/kg (mean ± standard deviation, n =

257

18) in composite samples from a depth interval of 0-15 cm and 403 ± 67 µg/kg (n = 15)

258

in the 15-30 cm interval. DDTs were dominated by DDT metabolites, with DDE and

259

DDD comprising 92% of the overall DDT mass in sediment. Chlordane (reported as the

260

sum of cis- and trans-chlordane and trans-nonachlor) sediment concentrations were 399 ±

261

40 µg/kg (n = 18) and 546 ± 113 µg/kg (n = 15) in the 0-15 cm and 15-30 cm interval,

262

respectively. Cis- and trans-chlordanes made up 77% of the chlordane mass in sediment.

263 264

Baseline PE measurements suggested increasing concentration with depth, consistent

265

with the coring data. At 5 cm depth, the average DDT concentration was 124 ± 17 ng/g (n

ACS Paragon Plus Environment

Page 12 of 32

Page 13 of 32

Environmental Science & Technology

266

= 15) and the average chlordane concentration in PE was 235 ± 28 ng/g. At 35 cm depth,

267

the PE concentration increased to 173 ± 14 ng/g and 397 ± 40 ng/g for DDT and

268

chlordane (n = 15), respectively. The 150 PE observations were used to construct a three-

269

dimensional site model of the relative concentrations within the mesocosm area. This site

270

model was then used to interpret post-treatment observations. A discussion of the

271

baseline assessment and site model is available in the Supporting Information.

272 273

Activated Carbon Distribution in Mesocosms.

274

Measurements of TOC indicated that the AC dose delivered to the nominal treatment

275

depth interval of 0-15 cm was equal to or greater than the design dose (Figure S8).

276

Sediment TOC was related to AC dose with Equation 29: 123&' = 45 − 45! /86.1 − 45100%

(2)

277 278

where TOC (%) is for an amended core section, 86.1 is the TOC of the type TOG AC (as

279

measured by Cho et al.8), DoseAC is the AC dose (% dw) and TOC0 is the average TOC at

280

the corresponding position within the mixed control mesocosm. Measured AC doses were

281

in line with the nominal applied dose (Figure S8). Overall, there was little spatial

282

variability in AC dose within the mixed zone in a given mesocosm, with the standard

283

deviation of the observed AC dose on the order of 10-25%. This homogeneity can be

284

attributed to the intense mixing imparted to the sediment by the impeller.

285 286

AC recoveries of 118%, 98%, 110%, and 117% were observed for the 5% GAC, 5%

287

PAC, 10% GAC, and 5% GAC 2× mix mesocosms, respectively. The recovery

ACS Paragon Plus Environment

Environmental Science & Technology

288

calculations here account for the entire core, not just the nominal treatment zone.

289

Significant AC was recovered below the nominal treatment zone, indicating overmixing

290

with respect to depth.

291 292

BC measurements closely agreed with AC dose estimates by TOC (see Figure S9),

293

confirming the suitability of TOC measurements for AC dose estimation. To confirm

294

visual observations that untreated sediments were deposited into the mesocosms, the

295

black carbon content (BC) of the thin (~2-5 mm), topmost layer of soft sediment was

296

measured and compared to the BC content of the overall AC amendment zone (0-15 cm).

297

If sediment from outside the mesocosms settled into the mesocosms, the BC content of

298

the surficial sediment would be less than the overall AC amendment zone, as the settled

299

sediment would be relatively free of black carbon while the mixed zone would have

300

elevated BC content due to AC addition. BC, rather than TOC, was used in this

301

comparison to limit any bias due to settlement of organic-rich material. The surficial

302

sediment BC content was found to be significantly (four-way ANOVA, p < 0.05) lower

303

than the composite mixed zone– up to a factor of 7. Along with visual observations, this

304

provides an additional line of evidence that untreated sediment settled into the

305

mesocosms.

306 307

Response in Pesticide PE Concentration.

308

For all AC treatments, the PE pesticide concentrations within the treatment zone were

309

significantly reduced compared to the mixed control and to the deeper PE samples. PE

310

pesticide concentration depth profiles were compiled 6 and 18 months after treatment for

ACS Paragon Plus Environment

Page 14 of 32

Page 15 of 32

Environmental Science & Technology

311

chlordane and DDT (Figure 2 and Figure 3). An example depth profile with the

312

corresponding TOC profile is presented in Figure 2 for the 10% GAC mesocosm, 18-

313

month observation. Omitting the most surficial sample, the variability of the duplicate PE

314

samples was low, typically ? C>?,IJKELJC,! C>?,ABCDEFGB = H M H M C>?,IJKELJC C>?,!

(3)

336 337

where CPE,Relative is the relative pesticide PE concentration normalized to the mixed

338

control mesocosm, CPE is the pesticide PE concentration within a treatment mesocosm,

339

CPE, Control is the PE pesticide concentration in the mixed control mesocosm at the depth

340

corresponding to the original sample, and CPE,

341

baseline concentrations. Data from the surficial zone, as shown in Figure 2, are neglected

342

from these composite measurements. Six months post-treatment, PE samplers contained

343

71 and 76 % less chlordane and DDT, respectively, in the 5% GAC mesocosm compared

344

to the mixed control. Similarly, chlordane and DDT PE concentrations were both 71%

345

lower in the 10% GAC mesocosms compared to the mixed control. Chlordane and DDT

346

concentrations were 65 and 75% lower, respectively, in the 5% GAC mesocosm with 2×

347

mixing. Reductions were more significant in the 5% PAC mesocosm, with chlordane and

348

DDT concentrations 93 and 94% lower, respectively, in the 5% PAC mesocosms

349

compared to the mixed control.

Control,0

and CPE,

0

are the pesticide PE

350 351

PE pesticide concentration within the treatment zone significantly (4-way ANOVA, p
0.50), as

ACS Paragon Plus Environment

Page 18 of 32

Page 19 of 32

Environmental Science & Technology

400

indicated in Figure 3. For example, 6 months post-treatment, PE chlordane concentration

401

was reduced by 71% in the 5% GAC mesocosm and 71% in the 10% GAC mesocosm.

402

Chlordane PE concentration was reduced by 88% and 91% in the 5% and 10% GAC

403

mesocosm at 18 months post-treatment, respectively.

404 405

There was no significant mixing duration effect on PE chlordane concentration within the

406

nominal treatment zone (4-way ANOVA, p > .10), as shown in Figure 3. PE chlordane

407

concentration within the 5% GAC mesocosm with 20 minutes mixing time was reduced

408

by 65 and 85% 6 months and 18 months post-treatment, respectively, compared with

409

71% and 88% in the 10-minute mixed GAC mesocosm. The response with respect to

410

DDT was similar, with DDT reductions of 75 and 84% 6 months and 18 moths post-

411

treatment, compared with 76 and 83% in the 10-minute mixed GAC mesocosm. Mixing

412

can enhance the AC performance by uniformly distributing AC particles which results in

413

shorter diffusion distances between particles14, but these results imply that 10-minute

414

mixing was sufficient to homogenize the sediment. Neglecting the effect of mixing on

415

sediment homogenization, the long-term mass-transport effect of mixing is minimal at

416

these mixing durations (see Figure S15). However, note that field deployment at scale

417

will likely not achieve the degree of homogenization observed here9.

418 419

Gut Fluid Extraction.

420

In all treatment mesocosms, AC amendment reduced the pesticide mass solubilized by

421

the A. brasiliensis digestive fluids or the taurocholate synthetic analog. In control

422

treatments lacking AC, quantifiable gut fluid extraction efficiencies typically ranged from

ACS Paragon Plus Environment

Environmental Science & Technology

Page 20 of 32

423

0.2-2% for most analytes, and never exceeded 3.1% (Figures S16, S17). Such values are

424

typical for other hydrophobic substances in field-collected sediments, for example, high

425

molecular weight PAH extraction efficiencies are typically 0.1-10%24-26,

426

organochlorines of the present study are insecticides that have not been used locally in at

427

least 30 years, and the gut fluid results indicate approximately 99% of the sediment-

428

bound residues cannot be extracted under digestive conditions even without AC addition.

27

. The

429 430

Within the 0-15 cm composite sediment sample, addition of AC reduced gut fluid

431

extractability even further. At the 6-month sampling point, extractability for DDT and

432

chlordane declined from approximately 0.3-0.7% in the unmixed control treatment to

433

near 0 in all AC treatments. Note that while there is some difference in extraction

434

efficiency between the mixed and unmixed controls, these differences also appear in PE

435

measurements from outside the mesocosm (Figure S7), suggesting the difference stems

436

from a localized area of high concentration near the unmixed control is not an effect of

437

the mixing. At the 18-month time point, extractability declined from approximately 1.0-

438

1.5% in the unmixed control to approximately 0.1% in all AC treatments. It appeared to

439

make little difference whether 5 or 10% AC was used, whether PAC or GAC, or whether

440

mixed for 10 or 20 min. There was a high degree of correlation in the amount of

441

contaminant solubilization by gut fluid and the taurocholate mimic (Figure S18). At both

442

the 6-month and 18-month sampling point, the extraction efficiency of the two fluids was

443

highly correlated (p