Metal Oxides in Surface Sediment Control Nickel Bioavailability to

Oct 18, 2017 - (8, 9) The SEM-AVS models provide important information for predicting conditions when sediment metals are present but nontoxic for sev...
0 downloads 15 Views 524KB Size
Subscriber access provided by Universitaetsbibliothek | Johann Christian Senckenberg

Article

Metal oxides in surface sediment control nickel bioavailability to benthic macroinvertebrates Raissa Marques Mendonca, Jennifer M Daley, Michelle L Hudson, Christian Schlekat, G. Allen Burton, and David Costello Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b03718 • Publication Date (Web): 18 Oct 2017 Downloaded from http://pubs.acs.org on October 18, 2017

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

1

Environmental Science & Technology

TOC (Table of Contents Art)

2

ACS Paragon Plus Environment

Environmental Science & Technology

3

Metal oxides in surface sediment control nickel bioavailability to benthic

4

macroinvertebrates

5 6

Raissa Marques Mendonca1*, Jennifer Marie Daley2, Michelle Lynn Hudson2, Christian Eduard

7

Schlekat3, Glenn Allen Burton, Jr.2, David Matthew Costello1

8 9

1

Department of Biological Sciences, Kent State University, 1275 University Esplanade, Kent,

10

Ohio 44242, United States

11

2

12

Arbor, Michigan 48109, United States

13

3

14

Parkway, Suite 240, Durham, North Carolina 27713, United States

School for Environment and Sustainability, University of Michigan, 440 Church St., Ann

Nickel Producers Environmental Research Association (NiPERA, Inc.), 2525 Meridian

15 16 17 18 19 20 21

Corresponding author: Raissa M Mendonca Department of Biological Sciences, Kent State University 1275 University Esplanade, Kent, OH 44242 USA Tel: (330) 554-6244 Email address: [email protected]

ACS Paragon Plus Environment

Page 2 of 32

Page 3 of 32

Environmental Science & Technology

22

Abstract

23

In aquatic ecosystems, the cycling and toxicity of nickel (Ni) are coupled to other elemental

24

cycles that can limit its bioavailability. Current sediment risk assessment approaches consider

25

acid-volatile sulfide (AVS) as the major binding phase for Ni, but have not yet incorporated

26

ligands that are present in oxic sediments. Our study aimed to assess how metal oxides play a

27

role in Ni bioavailability in surficial sediments exposed to effluent from two mine sites. We

28

coupled spatially explicit sediment geochemistry (i.e., separate oxic and suboxic) to the

29

indigenous macroinvertebrate community structure. Effluent-exposed sites contained high

30

concentrations of sediment Ni and AVS, though roughly 80% less AVS was observed in surface

31

sediments. Iron (Fe) oxide mineral concentrations were elevated in surface sediments and bound

32

a substantial proportion of Ni. Redundancy analysis of the invertebrate community showed

33

surface sediment geochemistry significantly explained shifts in community abundances. Relative

34

abundance of the dominant mayfly (Ephemeridae) was reduced in sites with greater bioavailable

35

Ni, but accounting for Fe oxide-bound Ni greatly decreased variation in effect thresholds

36

between the two mine sites. Our results provide field-based evidence that solid-phase ligands in

37

oxic sediment, most notably Fe oxides, may have a critical role in controlling nickel

38

bioavailability.

39 40

Introduction

41

Sediment metals can impair biodiversity and ecological functions in aquatic

42

environments.1-4 However, the cycling and toxicity of metals, such as nickel (Ni), are coupled to

43

other elemental cycles that determine toxic metal speciation and sorption behavior onto aqueous

44

and solid phase ligands. In sediment, metal mobilization is determined by binding (i.e., sorption,

ACS Paragon Plus Environment

Environmental Science & Technology

45

precipitation, chelation) to reduced and oxidized ligands, which controls metal bioavailability

46

and toxicity.5,6 The distribution of solid-phase ligands in natural sediments is complex and often

47

characterized by spatial stratification and temporal variation related to dynamic physicochemical

48

conditions. In oxic sediment, oxidized species (i.e., iron, manganese, and aluminum oxide

49

minerals) represent the major binding phases, whereas sulfide minerals are the predominant

50

metal ligand in anoxic sediment layers.5 Lotic sediments are often vertically stratified with thin

51

(mm to cm) aerobic layers overlying anaerobic horizons, and the distinct physicochemical

52

conditions in these sediment layers modify their metal binding capacity. Unlike for sulfide,7

53

binding constants for oxide minerals and the partitioning behavior of nickel to these solid phases

54

in situ have not yet been applied to sediment metal risk assessment efforts.5

55

Bioavailability models based on acid-volatile sulfide (AVS) and simultaneously extracted

56

metals (SEM) have been extensively applied in metal toxicity guidelines and assume sulfide as

57

the main factor controlling metal partitioning.8,9 The SEM-AVS models provide important

58

information for predicting conditions when sediment metals are present but non-toxic for several

59

aquatic species, but in some cases these models fail to predict metal toxicity due to unique

60

sediment redox chemistry in surface sediment and benthic fauna ecology.10-13 In oxic surface

61

sediment, AVS concentrations are generally low and highly variable owing to the rapid oxidation

62

of sulfide in the presence of oxygen.12,14-16 In addition, oxic microenvironments can occur at

63

deeper sediment depths that are oxygenated through diffusion via bioirrigation and bioturbation

64

by benthic organisms.17,18 These variable sediment redox gradients can render SEM-AVS models

65

less suitable for predicting metal bioavailability in naturally stratified sediment, and for deriving

66

toxicity thresholds for taxa with unique bioturbation behavior.19 This limitation generally results

ACS Paragon Plus Environment

Page 4 of 32

Page 5 of 32

Environmental Science & Technology

67

in conservative criteria that do not account for metal binding and co-precipitating with oxide

68

minerals.12,20

69

Incorporating concentrations of oxidized ligands into metal toxicity assessments may

70

provide benchmarks more compatible with environmental conditions, especially for lotic

71

ecosystems.12,14,21,22 In addition to including metal oxide minerals (i.e., Fe, Mn, and Al oxides) as

72

potential binding phases, accounting for the distinct geochemistry in surface sediments should

73

also provide more realistic estimates of geochemical conditions under which sensitive taxa are

74

exposed to sediment metals. While the SEM-AVS models can be effective at distinguishing

75

between non-toxic and potentially toxic conditions in some sediments, the model’s limited

76

applicability to only anoxic conditions constrains its scope in metal bioavailability assessment.

77

To address this limitation, research on the behavior of metal partitioning onto metal oxide

78

minerals in natural sediments is needed to derive information on binding coefficients. Analyses

79

of toxic metals binding to metal oxide minerals have been widely performed on pure synthetic

80

mineral substrates,23-25 but the studies did not examine metal binding within the context of the

81

mixture of competing ligands found in natural sediments. With the potential importance of oxic

82

sediment layers in modifying metal bioavailability, there is a need to explicitly test whether

83

including oxidized species in models of bioavailability more accurately predicts sediment metal

84

toxicity.

85

In this study, we considered whether spatially resolved sediment geochemistry (i.e.,

86

separate oxic and suboxic sediment) and water physicochemistry increased our ability to predict

87

the effects of nickel exposure to macroinvertebrates compared to current methods (i.e.,

88

homogenized sediment and SEM-AVS bioavailability models). Furthermore, we used in situ

89

natural gradient of total Ni from two mine sites to determine if concentration–response

ACS Paragon Plus Environment

Environmental Science & Technology

90

thresholds could be determined from the field using indigenous benthic macroinvertebrates. We

91

hypothesized that sediment geochemistry would be vertically stratified between oxic and anoxic

92

horizons and characterized by a greater proportion of metal oxide minerals and low AVS in the

93

surface layer due to oxidizing conditions. Additionally, we predicted sediment characteristics in

94

surface sediments (e.g., total nickel concentration, nickel bound to solid-phase ligands) would

95

significantly affect benthic macroinvertebrate composition and explain community variation

96

between sites. Within effluent-exposed tributaries, we predicted the natural nickel gradient

97

would reveal spatial patterns of invertebrate abundance, with sites farther from the effluent

98

discharge more similar to reference conditions. Finally, we hypothesized that nickel-

99

contaminated lotic sediments would be less toxic to the native invertebrate community than

100

expected by current benchmarks due to the partitioning of nickel to metal oxide minerals in the

101

oxic sediment layer.

102 103

Methods

104

Site description

105

Sampling was conducted in tributaries of the Burntwood River near Thompson,

106

Manitoba, Canada in August 2014. Two pairs of reference and effluent-exposed tributaries were

107

selected near the Birchtree (55° 41’ N, 97° 57’ W) and Thompson (55° 49’ N, 97° 41’ W) nickel

108

mines (Fig. S1). The tributary in the Birchtree mining area received effluent from the mine

109

treatment plant, whereas effluent from the Thompson mine passed through a wetland complex

110

(approximately 9 km) prior to reaching the sampling area (Fig. S1). Paired reference tributaries

111

were within 8 km of effluent-exposed tributaries and had similar sediment characteristics but Ni

112

at background concentrations. Within tributaries, ten sampling sites were spatially distributed

ACS Paragon Plus Environment

Page 6 of 32

Page 7 of 32

Environmental Science & Technology

113

from upstream to downstream, which established a gradient of nickel contamination in effluent-

114

exposed sites (Fig. S2-S4).

115 116

Water, sediment and invertebrate sampling and analysis

117

Surface water physicochemical characteristics were assessed in situ with hand-held

118

meters and a logging sonde (YSI 6290). Unfiltered surface water samples were collected for

119

analysis of water hardness (as mg CaCO3 L-1) by EDTA titration. Additional surface water

120

samples were filtered (0.45 µm) and acidified with 1% nitric acid for analysis of Ni and major

121

anions (nitrate, sulfate, and chloride) via inductively-coupled plasma mass spectrometry (ICP-

122

MS) and ion chromatography, respectively. Intact sediment cores were collected using 5-cm

123

diameter plastic tubes and caps. The tubes were manually pressed into the sediment, plastic caps

124

were placed at the ends of the core tube, and the intact core was carefully removed. Sediment

125

cores with visible disturbance of surface sediment were discarded and new cores were collected

126

within a 50-cm radius of the original core. At each reference tributary, sediment cores were

127

collected at four sites (Ref 1–4), whereas sediment samples were collected at 10 sites within both

128

exposure tributaries (n total = 28 sediment cores). The intact sediment cores were processed

129

within 6 h by removing the overlying water, extruding the sediment using a plunger,

130

and sectioning the core into surface sediment samples (0–2 cm) and deep sediment samples (2–4

131

cm). Surface and deep sediment samples were sealed in polyethylene bags and frozen (-18° C)

132

prior to analysis.

133

Prior to geochemical analysis, sediment samples were defrosted and manually

134

homogenized without opening the bags. Analysis of AVS and SEM was performed according to

135

Allen and colleagues,26 which used 1M hydrochloric acid extraction (50 min) to obtain dilute

ACS Paragon Plus Environment

Environmental Science & Technology

136

acid-extractable metal concentrations (i.e., SEM). Total metal concentrations (FeTOT, MnTOT, and

137

NiTOT) were obtained through sediment digestion in concentrated nitric acid at reflux

138

temperatures using a digestion block. For oxidized Fe and Mn and associated metals, selective

139

extractions27 were used to differentiate between the amorphous (MeHFO), crystalline (MeCFO), and

140

total oxidized minerals (MeOXID). Metal concentrations (Fe, Mn, and Ni) for SEM metals, total

141

metals, and selective extractions were determined by inductively coupled plasma optical

142

emission spectrometry (ICP-OES). An elemental analyzer was used to determine carbon content

143

of sediment samples. All analyses were completed with appropriate quality assurance/quality

144

control checks (Table S1), which included blanks, duplicates, and a standard reference material

145

(total metals only, Domestic Sludge, NIST SRM 2781).

146

Benthic macroinvertebrate community samples were taken using a 152 x 152 mm petite

147

ponar grab (surface area 231 cm2, depth 5–7 cm). Samples were collected at each of the ten

148

sites within the reference and exposure tributaries (total n = 40 invertebrate samples). The

149

sediment collected with the ponar was sieved through a 500-µm bucket sieve, washed, and

150

manually transferred into polystyrene specimen vials filled with 90% ethanol. All invertebrates

151

within each sample were sorted and identified to family28 using a dissecting microscope. All

152

pelagic taxa identified were excluded from statistical analyses.

153 154

Statistical analysis

155

Sediment geochemical characteristics were analyzed with three-way analysis of variance

156

(ANOVA) that used location (Birtchtree or Thompson), tributary (Ref or Exp), and sediment

157

depth (surface or deep) as predictor variables. When necessary, data were log transformed to

158

meet assumptions of ANOVA. Analyses for NiTOT required an additional statistical block for site

ACS Paragon Plus Environment

Page 8 of 32

Page 9 of 32

Environmental Science & Technology

159

to account for the variation related to the effluent dilution gradient. Concentrations of select

160

extractions of metal (i.e., SEM, oxidized metal extractions) were expressed as a percentage of the

161

total concentration of metal. When significant interactions were identified, post-hoc Tukey’s

162

multiple comparison tests were used to detect significant pairwise differences. Water

163

physicochemical properties were analyzed by analysis of covariance (ANCOVA) with location

164

(Birchtree or Thompson) as a main effect and sediment total nickel (mean of surface and deep)

165

as a covariate to account for the gradient in effluent exposure.

166

Invertebrate community composition was analyzed via transformation-based redundancy

167

analysis (RDA) to assess the proportion of variance in the invertebrate community explained by

168

the environmental variables.29 The objective is to obtain components of the invertebrate

169

community matrix that can be linearly explained by the predictor matrix (i.e., environmental

170

variables) and that represent most of the variance in the community. In reference tributaries,

171

invertebrate community samples that lacked paired sediment geochemistry measures were

172

assigned mean values from sampled reference sites in the respective tributaries (n = 4).

173

Invertebrate taxa abundances were transformed using Hellinger distance, which avoids the use of

174

dual absences as an indication of similarity between sites. All benthic taxa identified (n = 38

175

families) were included in the statistical analysis (Table S2). Four RDA models comprising

176

different sets of explanatory variables were considered: water only, surface sediment only, deep

177

sediment only, and all environmental variables. The statistical significance of each RDA model

178

was tested by permutation with a type I error rate (α) of 0.05. Forward selection on variables

179

within significant models was used to determine the most parsimonious model based on adjusted

180

R2 and variable significance (α = 0.05) as selection criteria.30 Final models are depicted in

ACS Paragon Plus Environment

Environmental Science & Technology

181

ordination plots, and taxa with a significant proportion of variation explained by any RDA axis

182

(coefficient of determination > 0.20), referred to as best-fit families, are identified.

Page 10 of 32

183

The most abundant and responsive mayfly taxon (i.e., Ephemeridae) was further analyzed

184

using log-linear regression to assess responses to Ni exposure. Ephemeridae relative abundance

185

(% of total invertebrate abundance) was regressed against three potential measures of Ni

186

bioavailability: simultaneously extracted nickel (NiSEM), SEM Ni in excess of AVS corrected for

187

organic carbon ((NiSEM-AVS)/ƒOC), and SEM Ni in excess of oxide-bound Ni corrected for

188

organic carbon ((NiSEM-NiHFO)/ƒOC). The latter fraction represents the concentration of NiSEM that

189

is not sorbed to amorphous metal oxides, which is similar to how NiSEM-AVS represents the Ni

190

that is not precipitated as NiS. These measures of sediment Ni were selected for comparison due

191

to their use in current sediment risk assessment (i.e., (SEM-AVS)/ƒOC)7 or their identification as

192

important variables in this study (i.e., FeHFO explained significant variation in the community

193

matrix). All measures of Ni and binding ligands were from measurement in surface sediments. A

194

sigmoid log-logistic model, which is the conventional model for laboratory concentration–

195

response, could not be fit for each mine location due to greater inherent variability and a limited

196

sample size. Thus, for each mine location, a simpler log-linear regression was used to derive

197

concentration–response relationships. Log-linear regression requires positive concentrations,

198

therefore sites with negative values for Ni bioavailability (e.g., (SEM-AVS)/fOC < 0) were

199

substituted with the minimum positive value measured in the corresponding mine location and

200

tributary. We chose this approach because we wanted to use all the invertebrate data available,

201

and if negative values were excluded from the regression, the dataset would be biased towards

202

only those sites with greater concentrations of bioavailable Ni. From each concentration–

203

response curve, we determined the concentration that reduced the abundance of Ephemeridae by

ACS Paragon Plus Environment

Page 11 of 32

Environmental Science & Technology

204

20% (EC20) relative to the abundance of Ephemeridae in downstream effluent-exposed sites

205

with minimal measurable exposure to the effluent (i.e., NiTOT ≤ reference mean NiTOT + 1

206

standard deviation, n = 2 sites). The downstream Ephemeridae relative abundance was used as a

207

baseline in lieu of the mean abundance at reference sites because in general the reference

208

tributaries had greater abundance of taxa not found at effluent-exposed sites (e.g., Gammaridae,

209

Polycentropodidae; Table S2), which lead to lower mean relative abundance of Ephemeridae in

210

reference sites. For each bioavailability measure we calculated a relative difference in EC20

211

between each mine site (RD = |EC20BIRCHTREE-EC20THOMPSON|/mean EC20) to assess how well

212

the bioavailability model normalized the concentration–response relationship (i.e., lower RD

213

indicates better agreement between locations). For two mine sites AVS was in excess of NiSEM;

214

for comparative purposes, we derived thresholds for concentration-response relationships that

215

both included corrected concentrations (i.e., set negative (SEM-AVS)/fOC to lowest positive

216

value) and excluded these negative values. Concentration-response relationships were attempted

217

for additional sensitive taxa (e.g., Talitridae, Caenidae) but the absence of organisms in over

218

25% of the sampled sites prevented the detection of nickel-related effects.

219

Reported values are either means ± standard deviation or for log-normally distributed

220

variables geometric means ± geometric standard deviation. All statistical analyses were

221

performed using RStudio31 with appropriate packages for generalized linear regressions (lme4

222

version 1.1-11)32 and community ordinations (vegan version 2.3-2).33

223

ACS Paragon Plus Environment

Environmental Science & Technology

224

Results

225

Sediment geochemistry

Page 12 of 32

226

Acid volatile sulfide concentrations did not differ between Thompson and Birchtree

227

locations (p = 0.30) but varied significantly with sediment depth (p < 0.001) and between

228

reference and effluent-exposed tributaries (p = 0.017; Fig. 1A). All surface sediment (geometric

229

mean = 0.08 µmol g-1 dw, geometric SD = 5.4) had on average 82% less AVS than deep

230

sediment (geometric mean = 0.48 µmol g-1 dw, geometric SD = 4.9). Effluent-exposed sites had

231

greater AVS concentrations (maximum = 7.6 µmol g-1 dw) than reference tributaries. In effluent-

232

exposed tributaries, elevated sulfate concentrations in mine effluent provided abundant substrate

233

for microbial reduction of sulfur and precipitation of metal sulfide minerals. Sediment carbon

234

content was consistent across mine locations and between surface and deep horizons (1.5–9.2%),

235

but differed marginally between reference and effluent-exposed tributaries in Birchtree sediment

236

(2% difference, p = 0.03).

237

The concentration of total Fe (FeTOT) ranged from 2.1–3.9% in all sediment samples and

238

was consistent among sediment depths and mine locations, only differing slightly between

239

Thompson reference and effluent-exposed tributaries. The proportion of dilute acid-extracted

240

iron (FeSEM) was significantly greater in Birchtree sediments and reference tributaries (p
0.20), and aligned near reference sites, which suggests

317

that these taxa were most abundant in sites with lower total NiSURF concentrations and/or greater

318

concentrations of FeHFO. The RDA identified a single tolerant invertebrate family (i.e.,

319

Sphaeriidae (SPH)), whose abundance was significantly explained by the RDA but aligned

320

closer to effluent-exposed sites with higher total NiSURF. Both the simplified models for

321

RDAWATER and RDASURF explained a significant proportion of the variation in the

322

macroinvertebrate community (p < 0.001).

323 324

Ephemeridae concentration–response

325

The relative abundance of Ephemeridae (Ephemeridae abundance/total invertebrate

326

abundance at each site) declined with increasing concentrations of NiSEM in the surface sediment,

327

but the concentration–response relationships differed among mine locations (Figure 3A).

328

Ephemeridae relative abundance in Birchtree declined at a slightly lower NiSEM concentration

329

(0.6 µmol g-1) than those in Thompson (0.8 µmol g-1, EC20 relative difference (RD) = 0.28). Ni

330

concentrations in excess of AVS corrected for organic carbon also demonstrated concentration–

331

response relationships with Ephemeridae relative abundance, with increasing concentrations of

332

(NiSEM-AVS)/ƒOC reducing the abundance of mayflies (Figure 3B). However, a large difference

333

in concentration–response thresholds was observed between Birchtree (EC20 = 4.9 µmol g-1) and

334

Thompson (EC20 = 0.4 µmol g-1) sites, and the difference between EC20s expressed as (NiSEM-

335

AVS)/ƒOC (RD = 1.70) was much greater than the difference between locations for NiSEM. The

336

increased relative difference in the (NiSEM-AVS)/ƒOC model is partly due to the fact that sites

337

with high concentrations of AVS relative to NiSEM, which should be indicative of no toxicity, had

ACS Paragon Plus Environment

Page 17 of 32

Environmental Science & Technology

338

much lower Ephemeridae relative abundances (14% and 3%) than the baseline relative

339

abundance (43%).

340

Considering the critical role of Fe identified in the multivariate community analysis and

341

the differences in NiHFO between sites, we related Ephemeridae relative abundance to an index of

342

sediment Ni that accounts for binding by metal oxide minerals and that can be directly compared

343

to the AVS-based model. We found that the relative abundance of Ephemeridae declined with

344

increasing concentrations of Ni not bound to amorphous Fe (Figure 3C). Considering the relative

345

difference between concentration–response thresholds, the Fe-corrected Ni model did not

346

decrease inter-site variation substantially more than the already low NiSEM model. However,

347

while (NiSEM-NiHFO)/ƒOC is a coarse measure of Fe binding, the difference in EC20 thresholds

348

between Thompson (EC20 = 6.3 µmol g-1) and Birchtree (EC20 = 4.8 µmol g-1) locations

349

expressed as (NiSEM-NiHFO)/ƒOC was much smaller (RD = 0.27) when compared to (NiSEM-

350

AVS)/ƒOC (RD = 1.70).

351 352

Discussion

353

In this study, we assessed if the spatially stratified geochemistry of field-contaminated

354

sediments improved our ability to predict the effects of nickel exposure to benthic

355

macroinvertebrates compared to current methods that partly overlook the distinct

356

physicochemistry of oxic and suboxic sediments. Moreover, we used the natural contamination

357

gradient of nickel to determine field-based concentration–response thresholds using indigenous

358

benthic macroinvertebrates. Though our field-based assessment involved organism exposure to

359

nickel in both overlying water and sediment, the metal gradient at the study sites along with our

ACS Paragon Plus Environment

Environmental Science & Technology

Page 18 of 32

360

multivariate analysis allowed us to distinguish whether macroinvertebrate taxa were responding

361

to either sediment or overlying water impairment.

362

As expected, we observed a substantial effect of mine effluent discharge on water quality,

363

largely related to the input of ions in effluent-exposed tributaries. We observed a significantly

364

higher dissolved nickel concentration in Thompson effluent-exposed tributaries, which was

365

unexpected considering that the effluent-exposed tributary at the Thompson mine had slightly

366

lower NiTOT than at the Birchtree mine. The difference in dissolved nickel in the tributaries can

367

likely be attributed to differences in the effluent discharge at the two mines. First, the Birchtree

368

mine effluent was inactive (i.e., summer shutdown) during the sampling period. Second, the

369

Thompson mine effluent flows through a wetland complex (approximately 9 km) before reaching

370

the sampled tributary, whereas the Birchtree effluent discharges directly upstream of the sampled

371

tributary. The long flow path between outflow and sampling location at the Thompson tributary

372

likely results in more consistent dissolved Ni concentrations, while the Birchtree site would be

373

more responsive to mine operating conditions. Finally, the Birchtree mine effluent treatment

374

plant was updated in 2008 which substantially decreased metal loading due to 87% reduction in

375

dissolved nickel concentration and 40% increase in discharge capacity.34 Previous environmental

376

monitoring of the Birchtree effluent-exposed tributary during regular mine activity reported

377

dissolved nickel concentrations on average 2.5× higher than the 9 ± 8 µg L-1 observed in this

378

study.34 This difference in overlying water Ni concentration between mine sites likely explains

379

why dissolved Ni did not improve RDAWATER model fit, and also indicates that water–sediment

380

partitioning of Ni may be more stable in Thompson sediment over both long and short

381

timescales.

ACS Paragon Plus Environment

Page 19 of 32

Environmental Science & Technology

382

As hypothesized, we observed distinct vertical variation in sediment chemistry and metal

383

speciation in both Birchtree and Thompson sediment. AVS concentrations were substantially

384

lower in surface sediment when compared to deep sediment. This distinct difference is despite

385

the fact that surface samples (i.e. top 2 cm) were likely not discrete samples containing only

386

oxidized sediment. The oxic layer of fine-grained sediments is often 0.2).

ACS Paragon Plus Environment

Environmental Science & Technology

Ephemeridae abundance (%)

50

A

Page 32 of 32

C

B

Birchtree Thompson

40 30 20 10 0 0.1

736

1.0

10.0

NiSEM (µmol g-1)

10

100

1000

1

10

100

-1 (NiSEM-AVS)/ƒOC (µmol g-1) (NiSEM-NiHFO)/ƒOC (µmol g )

737

Figure 2. Concentration–response relationships between Ephemeridae relative abundance and

738

select sediment nickel variables: bioavailable sediment nickel (A), bioavailable nickel as a

739

function of AVS corrected for organic carbon (B), bioavailable Ni not associated with

740

amorphous metal oxides corrected for organic carbon (C). Solid (Thompson mine, open

741

symbols) and dashed (Birchtree mine, closed symbols) lines are best-fit lines from least-squares

742

regression for effluent-exposed sites at each mine location. Crosses (+) on panel B indicate sites

743

with AVS concentration in excess of NiSEM (negative NiSEM-AVS) for which the minimum

744

positive value measured in the corresponding mine and tributary was assigned to meet

745

requirements of log-linear regression. The latter sites are also identified by crosses (+) on panels

746

A and C. Concentration–response relationships and relative differences in EC20s show that the

747

amorphous iron-corrected bioavailability model was more effective in normalizing the

748

concentration–response relationship between Birchtree and Thompson mines than the AVS-

749

based model. Note the log scale on the x-axis of each panel.

ACS Paragon Plus Environment