Dissolved Organic Carbon Thresholds Affect MercuryBioaccumulation in Arctic LakesTodd D. French,† Adam J. Houben,† Jean-Pierre W. Desforges,†,⊥ Linda E. Kimpe,† Steven V. Kokelj,‡

Alexandre J. Poulain,† John P. Smol,§ Xiaowa Wang,∥ and Jules M. Blais†,*†Department of Biology, University of Ottawa, Ottawa, Ontario, K1N 6N5, Canada‡Northwest Territories Geoscience Office, Government of the Northwest Territories, Yellowknife, Northwest Territories, Canada,X1A 2R3, Canada§Paleoecological Environmental Assessment and Research Lab, Department of Biology, Queen’s University, Kingston, Ontario, K7L3N6, Canada∥Environment Canada, 867 Lakeshore Road, Burlington, Ontario, L7R 4A6, Canada

*S Supporting Information

ABSTRACT: Dissolved organic carbon (DOC) is known to affect the Hg cycle in aquaticenvironments due to its overriding influence on complexation, photochemical, andmicrobial processes, but its role as a mediating factor in the bioaccumulation of Hg inaquatic biota has remained enigmatic. Here, we examined 26 tundra lakes in Canada’swestern Arctic that span a large gradient of DOC concentrations to show that total Hg(HgT) and methyl mercury (MeHg) accumulation by aquatic invertebrates is defined by athreshold response to Hg-DOC binding. Our results showed that DOC promotes HgT andMeHg bioaccumulation in tundra lakes having low DOC (<8.6 − 8.8 mg C L−1; DOCthreshold concentration, TC) whereas DOC inhibits HgT and MeHg bioaccumulation inlakes having high DOC (>DOC TC), consistent with bioaccumulation results in acompanion paper (this issue) using a microbial bioreporter. Chemical equilibriummodeling showed that Hg bioaccumulation factors were elevated when Hg was associatedmainly to fulvic acids, but became dramatically reduced when DOC was >8.5 mg C L−1, atwhich point Hg was associated primarily with strong binding sites on larger, less bioaccessible humic acids. This studydemonstrates that the biological uptake of Hg in lakes is determined by binding thresholds on DOC, a water quality variablepredicted to change markedly with future environmental change.

1. INTRODUCTION

Global anthropogenic mercury (Hg) emissions have increasedsince preindustrial times, with the global emission in 2008being 2000 t and the present-day tropospheric burden (5200 t)being about 3-fold higher than that before the IndustrialRevolution.1,2 Atmospheric Hg transport from industrializedregions to the Arctic is also increasing and this, acting intandem with climate-related changes, has resulted in a netincrease in Hg deposition to land and water surfaces in severalArctic regions.3 Presently, the contribution of anthropogenicHg to the total body burdens of predatory birds and mammals,including humans, ranges from 74% to 94% across the Arctic,4

and evidence of chronic sublethal Hg toxicity has been shownin Arctic wildlife populations and in indigenous peoples whorely primarily on traditional country foods.In addition to the stresses caused by Hg and other pollutants,

Arctic ecosystems are being markedly affected by rapid climatechange. Ground temperatures and thermokarst activity inCanada’s western Arctic have increased with warming temper-atures since the 1970s, such that up to 15% of the water bodiesin the region have shoreline retrogressive thaw slumps (Figure1). Spatial variability in the distribution and magnitude of

permafrost thaw slumps has resulted in extremely wide pH(6.6−8.1) and DOC (6.8−30.0 mg C L−1) gradients amongtundra lakes in the western Arctic.5 Given that these keyvariables regulate Hg bioaccumulation and affect the post-depositional processing of Hg,6−8 we hypothesized thatpermafrost thawing is affecting Hg bioaccumulation in westernArctic lakes. Here, we use the wide gradients in lakewater DOCamong tundra lakes to determine its effect on Hgbioaccumulation patterns in Arctic lakes.

2. MATERIALS AND METHODS

2.1. Water and Amphipod Sampling. Our study wasundertaken in the tundra uplands east of the Mackenzie RiverDelta, Northwest Territories, Canada (Figure 1). Twenty-sixlakes were sampled in summer 2009, which consisted of“Reference Lakes” without shoreline thaw slumps (n = 12) and“Thaw Lakes” with shoreline thaw slumps (n = 14, Figure 1).

Received: August 29, 2013Revised: February 13, 2014Accepted: February 13, 2014Published: February 13, 2014

Article

pubs.acs.org/est

© 2014 American Chemical Society 3162 dx.doi.org/10.1021/es403849d | Environ. Sci. Technol. 2014, 48, 3162−3168

Water for the analysis of total mercury (HgT), totalmethylmercury (MeHgT), DOC, total dissolved Fe (FeD) andAl (AlD), major ions (DIC [∑CO3

2−, HCO3−, CO2], Mg2+,

Ca2+, Na+, K+, SO42−, Cl−), NH3 and NO3

− concentrations, andpH was collected from the littoral zone (0.5-m depth) of eachlake in precleaned/acid-washed polyethylene bottles. Bottlesand caps were triple rinsed with lake water before samplecollection, and bottles were filled to rim to void the sample ofair. Bottles and samples for Hg analyses were maintained insealed bags to prevent contamination. Water samples for HgTand MeHgT analysis were preserved in the field with 0.5% HCl,and those for FeD and AlD analyses were preserved,immediately after filtration, with conc. HNO3 to 2% of samplevolume. Littoral amphipods were collected with a D-net (63-μm). Amphipods were separated into large (>2000-μm) andsmall (250−2000 μm) size classes, and whole-body HgT andMeHgT dry-weight (dw) concentrations in each class weredetermined. We used a three-factor Lorentzian model toexamine Hg bioaccumulation factor (BAF) to amphipodsacross the broad water HgT and DOC gradients. VisualMINTEQ Version 3.0 (MINTEQ) was used to model Hgcomplexation to DOC across the water chemistry gradient.2.2. Chemical Analyses. Water pH was determined with a

Thermo Orion Model 106 m. Water HgT and MeHgTconcentrations were determined by cold-vapor atomic fluo-rescence spectrometry (CV-AFS, Tekran Instrument Corp.Series 2600 spectrophotometer) and capillary gas chromatog-raphy-atom fluorescence spectrometry (GC-AFS, Ai CambridgeModel 94 GC with a CTC Autosampler and PSA MerlinDetector), respectively, according to EPA method 1631

revision E, with a detection limit of 0.1 ng/L. Instrumentblanks averaged 0.01 ± 0.13 ng/L (1 standard deviation), whilefield blanks averaged 0.12 ± 0.17 ng/L. Hg recovery ofstandards was 99% ± 4% based on concurrent analyses ofNIST-certified stock solutions, with a r2 = 0.9996 for thecalibration curve. MercuryT concentrations in amphipodcomposites were determined at the University of Ottawausing a Nippon Hg Analyzer SP-3D (Nippon InstrumentsCorporation, Osaka, Japan) with a detection limit of 0.01 ng/sample. Each sample was analyzed with 0.1 M NaOH buffer,calcium hydroxide [Ca(OH)2], activated alumina (Al2O3), and1:1 sodium carbonate (Na2CO3) as additive agents. Approx-imately 5 to 10 mg of amphipod tissues were analyzed. Fivereplicates of DORM-3 certified reference material (0.382 ±0.06 mg/kg; National Research Council of Canada, Ottawa,ON) were run with samples, producing a mean recovery of102% with a CV of 2.2%. MeHgT concentrations in amphipodswas analyzed using capillary GC coupled with atomicfluorescence spectrometry (Hewlett-Packard, GC modelnumber HP 6890 series with HP 7683 injector) at a methoddetection limit of 0.02 ng/L. The average precision was 9.5%(range, 1.4−30.1%), and the average MeHg and standarddeviation of DORM4 was 326 ± 13, 92.1% of the expected 354± 31 ng/g.Water samples for DIC, DOC, FeD and AlD determinations

were hand pumped, within hours of collection, through 47-mmdiam./0.45-μm Sartorius filters (supported AcetatePlus, plain)and the filtrate analyzed. Dissolved inorganic carbon and DOCconcentrations were determined by infrared CO2 detection(Phoenix 8000 TOC analyzer with gas/liquid separator, mist

Figure 1. Study lakes in the tundra uplands east of the Mackenzie River Delta (Middle and East channels shown), Northwest Territories, Canada.Reference Lakes (without shoreline permafrost thaw slumps) and Thaw Lakes (with shoreline permafrost thaw slumps) are denoted “a” and “b”,respectively. Inset shows the study area (red box) in relation to Inuvik (population 3000) and the Beaufort Sea (Arctic Ocean). Plate shows a 140-mwide thaw slump on the western shoreline of Lake 14b. Table provides the general morphometry of the study lakes and the thaw slump area aspercentage of catchment area.

Environmental Science & Technology Article

dx.doi.org/10.1021/es403849d | Environ. Sci. Technol. 2014, 48, 3162−31683163

trap, Cl− and CO2-air scrubbers, and NDIR CO2 detector)following sequential conversions, by acidification, of DIC andDOC to CO2. Water FeD and AlD concentrations weredetermined by inductively coupled plasma-sector field massspectrometry (ICP-SFMS, Thermo-Finnigan Mass Spectrom-eter Element II) with direct sample aspiration. Chloride andSO4

2− concentrations were determined by ion chromatography(Dionex DX500 ion chromatograph with ASRS-ULTRA-II ion/conductivity suppressor); Cl− and SO4

2− were separated fromother anions on the basis of retention time. Water Ca2+, Mg2+,Na+, and K+ concentrations were determined by atomicabsorption spectroscopy in an air-acetylene flame at 11 p.s.i.(Varian SpectAA22 with SPS-5 autosampler). Water NO3

−2

concentrations were determined photometrically (Bran+Luebbe Continuous-Flow Technicon Autoanalyzer III set at520 nm) following sample acidification to pH 5.5 andsequential reactions with sulphanilamide and then N-1(-naphthyl)-ethylenediamine dichloride; these reactions form areddish-purple azo dye in proportion to the concentration ofNO3

− + NO2 in sample (sample NO3− converted to NO2 in

acidification step) . Water NH3 concentrations weredetermined photometrically (Bran+Luebbe Continuous-FlowTechnicon Autoanalyzer III set at 630 nm) following sampletreatment with the Berthelot Reaction which forms indophenolblue in proportion to the concentration of NH3 in sample.2.3. Mercury Bioaccumulation Trends. HgT and MeHgT

BAFs from water to amphipods were determined for each lakeby dividing concentrations in amphipods (ng kg−1 dw) byconcentrations in water (ng L−1). A three-factor Lorentzianmodel was used to assess Hg bioaccumulation across the waterHgT and DOC concentration gradients:

=+ −( )

Ya

1 X Xb

2o

where Y is the Hg concentration or BAF in amphipodcomposite, a is the height of the Lorentzian curve at theinflection point, b is a measure of the spread (standarddeviation) of the curve, and Xo (or TC, threshold concen-tration) is the inflection point along the gradient axis (x-axis).2.4. Modeling of Hg Speciation. Equilibrium speciation

calculations were undertaken with Visual MINTEQ Version3.09 to model Hg interactions with inorganic ligands and DOC.Lake chemistry data (above) were entered into MINTEQ; pHwas fixed at the value measured at sampling. The extent of Hgbinding to DOC was predicted by a NICA-Donnan model.10

This model enables simulation of cation complexation toconstituents that are highly heterogeneous with respect tobinding affinity, such as humic (HA) and fulvic (FA) acids. Theratio of dissolved organic matter (DOM) to DOC was set bydefault at 1.65, which is an average based on results for lakesand streams from the Swedish Environmental MonitoringNetwork.11 We ran the model assuming that the active DOMwas comprised of 75% FA and 25% HA, which approximatesthe HA:FA ratio observed in lakes with watershed vegetationsimilar to that in the Inuvik area.12 As modeled, each FA andHA fraction had two types of binding sites: one with a weakaffinity (FA1 and HA1) for Hg and one with a strong affinity(FA2 and HA2) for Hg.13 The weak binding sites arerepresentative of interaction with carboxyl groups while thestrong binding sites are representative of binding with organicthiol functional groups.13−15 The generic NICA-Donnan modelparameters for Hg were based those recommended by Milne et

al.10 The estimated binding affinities for Hg are described bythe equations below:

+ ⇒ +

=

+ +

K

Hg 2H O Hg(OH) 2H

log 6.2

22 2

‐ + + ⇒ ‐ +

=

+ +

K

H FA1 Hg(OH) H [FA1 Hg] 2H O

log 142 2

‐ + + ⇒ ‐ +

=

+ +

K

H FA2 Hg(OH) H [FA2 Hg] 2H O

log 302 2

‐ + + ⇒ ‐ +

=

+ +

K

H HA1 Hg(OH) H [HA1 Hg] 2H O

log 142 2

‐ + + ⇒ ‐ +

=

+ +

K

H HA2 Hg(OH) H [HA2 Hg] 2H O

log 302 2

We also ran simulations for which the binding constants forHg to the strong binding sites onto fulvic and humic acids (FA2and HA2) varied from log K = 25 to log K = 32 (Figure S1 ofthe Supporting Information, SI) to reflect variability observedin conditional constant estimates.15,16

Two-segment piecewise regression was used to analyze theMINTEQ output; the model was as follows:

=

− + −−

≤ ≤

− + −−

≤ ≤

⎧

⎨

⎪⎪⎪⎪

⎩

⎪⎪⎪⎪

f

m T c m c cT c

c c T

m c c m c Tc T

T c c

1( ) 2( 1)1

,

for 1

2( 2 ) 2( )2

,

for 2

C

C

C

C

C

C

where f is the percentage of water Hg bound to the FA and HAfractions, c is the water DOC concentration, c1 is the lowestDOC concentration in the series (start point), c2 is the highestDOC concentration in the series (end point), TC (iterativequasi-Newton method) is the DOC threshold concentration,m1 is the regression slope across DOC concentrations <TC(regression segment 1), and m2 is the slope across DOCconcentrations >TC (segment 2).Outlier data for all trend and modeling analyses were

identified with Grubbs test;17 statistical analyses wereperformed with STATISTICA 9 (Stat Soft Inc.). All outliersare identified on figures.

3. RESULTS AND DISCUSSIONIn contrast to some field studies with limited ranges in DOCconcentrations that have shown Hg concentrations in biota atlower trophic levels increasing in parallel with water Hgconcentration,18,19 we observed in our tundra lakes HgT andMeHgT concentrations in amphipods with distinct bell-shapedpatterns when analyzed against the water HgT gradient (0.7−4.7 ng L−1). The threshold water HgT concentration (TC,Lorentzian model inflection point) was determined to be 1.6−2.3 ng L−1 (Figure 2a). Mercury concentrations in amphipodsincreased with increasing water HgT concentration across lakeshaving water HgT concentrations < TC, but Hg concentrationsin amphipods decreased with increasing water HgT concen-

Environmental Science & Technology Article

dx.doi.org/10.1021/es403849d | Environ. Sci. Technol. 2014, 48, 3162−31683164

tration when the HgT TC was exceeded (Figure 2a). In general,HgT BAFs decreased with increasing water HgT concentration.For large amphipods, there was some evidence that Hgbioavailability also followed a bell-shaped pattern (TC = 1.5 ngL−1; Figure 2b). Mercury concentrations in the tundra lakeswere less than toxic levels; thus, the trajectory shifts in Hgbioaccumulation at the HgT TC (Figure 2) cannot be explainedby toxicological impairment. We suggest that the trajectoryshifts were caused by changes in critical limnetic conditions thatoccurred in conjunction with the HgT TC.Further examination of the bell-shaped relationship between

Hg concentrations in amphipods and water HgT concentration(Figure 2a) revealed that lake water pH shifted from alkaline toslightly acidic at the HgT TC. While Hg concentrations inamphipods were not directly correlated with pH, the alkalinelakes (mostly Thaw Lakes) comprised the upslope (HgTconcentration <TC) of the response curve, whereas the near-neutral to acidic lakes (mostly Reference Lakes) comprised thedownslope (HgT concentration >TC) of the response curve(Figure 2a). Mercury BAFs in amphipods also showed adistinct bell-shaped pattern when analyzed against the waterDOC gradient (2.2−23.1 mg C L−1), with the DOC TC being8.5−8.6 mg C L−1 (Figure 3a). As observed across the waterHgT gradient (Figure 2a), the alkaline lakes (mostly ThawLakes) comprised the upslope (DOC concentration <TC) ofthe response curve with the near-neutral to acidic lakes (mostlyReference Lakes) comprising the downslope (DOC concen-tration >TC; Figure 3a). Hg bioavailability in bacteria 20 andzooplankton 21 is typically related inversely to pH when DOC

is not a covariate of pH. Our results show, however, thatbioaccumulation cannot be predicted solely by pH in this dataset, suggesting other mediating factors, which we elaboratebelow. MeHgT BAFs also plummeted when the DOC TC wasexceeded (Figure 3a). The Hg bioaccumulation patternsobserved across the water HgT and DOC gradients weresimilar among large and small amphipods (Figures 2a and 3a);reasons for this pattern may include: (i) Hg and DOC arecotransported to the tundra lakes;18,22 and (ii) DOC complexesHg15,23 and therefore regulates Hg bioavailability.6

Our results indicate that DOC may both promote and inhibitHg bioavailability in these Arctic tundra lakes, such that: (i)DOC at low concentrations (<TC) promoted Hg bioavail-ability; and (ii) DOC at high concentration (>TC) inhibited Hgbioavailability (Figure 3a). As Hg is transported to the tundralakes with DOC,15,18,22,23 DOC may promote Hg bioaccumu-lation by increasing the supply of Hg to the lakes, and bystimulating microbial heterotrophic activity responsible forDOC degradation, thereby releasing bound Hg.24 Wehypothesize that the bioavailability of Hg in the high-DOC(>DOC TC) lakes (mostly Reference Lakes) was likely reduceddue to the formation of Hg species that are not readily availablefor direct biological uptake from water, which is the primarymode of Hg uptake at low and intermediate trophic levels.6

In support of our proposed mechanism, we used athermodynamic equilibrium model (MINTEQ) that predicteda pronounced increase in Hg adsorption to strong binding siteson humic acids (HA2, see Methods section) and reducedadsorption to strong binding sites on fulvic acids (FA2) whenthe critical DOC concentration (11.7 mg C L−1) was exceeded(Figure 3b). Interestingly, sensitivity analyses showed that theshift from fulvic to humic acids strong binding sites was mostapparent when log KHg‑FA was ≥ log KHg‑HA (Figure S1 of theSI). The resulting shift in Hg binding from fulvic acids to humicacids may result in reduced Hg bioaccumulation in aquaticbiota due to greater Hg bioavailability in the presence of lowmolecular weight (fulvic) organic acids,25 such as thoseassociated with the pool of fulvic acids.26 Alternatively, shiftsin Hg bioaccumulation may be linked to the complexmechanisms that drive Hg-DOC photoredox chemistry (andtherefore speciation), for which threshold-type relationshipshave been observed over DOC gradients.27

Analysis of the chemical modeling data identified abreakpoint at ca. 11.1 mg C L−1 (Figure 3b). This criticalDOC concentration corresponds well with the DOC TC (8.6−8.8 mg C L−1) estimated for the Hg BAF versus DOCrelationship (Figure 3a). In further support of our proposedmechanism, MeHgT was not detected in the water column ofhigh-DOC lakes; MINTEQ predicted that virtually all of thewater column Hg in these lakes would be bound to DOC, andtherefore much of the Hg in lakes with high DOC levels maynot be readily available to bacteria for methylation.The results of earlier studies have led some researchers to

speculate that there might be a threshold-type relationshipbetween Hg bioaccumulation/bioavailability and DOC infreshwater ecosystems.18,28 Our study quantitatively demon-strates these thresholds and shows that DOC both promotes(when concentration is <DOC TC) and inhibits (whenconcentration is >DOC TC) Hg bioavailability in naturalwaters. One of the most widely cited papers speculating aboutthe possibility of a DOC threshold is that by Driscoll et al.,28

who studied Hg bioaccumulation in perch (Perca f lavescens)from Adirondack Park (U.S.) lakes. They showed a positive

Figure 2. Mercury bioaccumulation in large and small amphipods inrelation to lake water HgT concentration. (a) Hg concentrations inamphipods versus water HgT concentration; MeHgT trends in largeamphipods were not determined (insufficient sample size). (b) HgTBAFs to amphipods versus water HgT concentration. Model parameterstatistics (three-parameter Lorentzian functions): pa all <0.001 and pb= <0.001 − 0.009 (pXo shown as TC above panels). Model parameterstatistics (linear function): pintercept <0.001, pslope <0.001. Modelparameters remain significant (p < 0.05) when outliers are maintainedin analysis. Model parameters are defined in the Methods section.

Environmental Science & Technology Article

dx.doi.org/10.1021/es403849d | Environ. Sci. Technol. 2014, 48, 3162−31683165

linear relationship between Hg concentration in perch andDOC concentration across lakes having DOC concentrations<∼10 mg C L−1, but that Hg concentrations in perch fromRock Pond Lake (∼27 mg C L−1) deviated far below thatexpected from the linear trajectory. Driscoll et al.,28 whilesuggesting the possibility of a DOC threshold, excluded theRock Pond Lake datum as an outlier because they hadinsufficient data to statistically confirm the downslope of athreshold-type response trajectory. Data presented by Belgerand Forsberg 29 also showed that Hg bioaccumulation inAmazonian wolf fish (Hoplias malabaricus) increased linearlywith increasing DOC concentration but, like Driscoll et al.,28

they excluded a high-DOC system from their statistical analysis,which suggested a threshold-type relationship was also presentin their Amazonian study system. In contrast to previousstudies, the environmental data offered by the western Arcticlake system provided us with an exceptional DOC gradient toassess the relationship between Hg bioaccumulation and DOC(Figures 2a and 3a).In a companion paper,30 we assessed the role of DOC on

Hg(II) bioavailability to a whole cell biosensor using definedmedia and field samples from Canada’s western Arctic spanninga wide range of DOC levels. The results obtained with thebiosensor are consistent with what was observed here foraquatic invertebrates. The biosensor data also further supportthe kinetically important and complex nature of Hg (II)

interaction with DOM that cannot solely be predicted byinteraction at equilibrium. Together these two studiesdemonstrate how the biological uptake of Hg in lakes isdetermined by both complexation kinetics and bindingthresholds on DOC, a water quality variable predicted tochange markedly with future environmental change.These results have implications for Hg bioaccumulation in

many lakes characterized by low DOC (<8 mg L−1,31−33 thathave rising DOC expected due to humification.34 Our resultssuggest a 2−3 fold increase in Hg bioaccumulation as theselakes approach the threshold DOC TC of 8 mg L−1, adding tothe existing high Hg burdens in some northern environments.The importance of threshold-type responses linking mercury

toxicity to DOC concentrations have hitherto received littleattention. Given that DOC levels mediate numerous ecosystemfunctions in lakes and relate closely to rapidly changing climatefactors such as temperature and precipitation,13 our resultsemphasize the critical role of DOC on determining the fate andbioavailability of mercury in a rapidly changing climate.

■ ASSOCIATED CONTENT

*S Supporting InformationSensitivity analysis testing the interaction of Hg with dissolvedorganic matter as the binding constant for Hg to the strongbinding sites of humic acids (HA2) varied from log K = 25 to

Figure 3. Threshold analyses of Hg BAFs in large and small amphipods in relation to lake water DOC concentration and percentage of water HgTbound to fulvic (FA) and humic (HA) fractions in relation to DOC concentration. (a) Top two panels show Lorentzian fit of HgT BAFs in large andsmall amphipods in relation to DOC concentration. Model parameter statistics (three-parameter Lorentzian functions): pa all <0.001 and pb both =0.003 (pXo shown as TC above panels). Blue vertical line on bottom panel illustrates that MeHgT BAFs plummeted when DOC concentrationsexceeded about 11 mg C L−1. Trends for MeHgT BAFs in large amphipods were not quantified (insufficient sample size). (b) Piecewise fit of %HgTbound to fulvic (FA1 and FA2, weak and strong binding sites, respectively) and humic (HA1 and HA2) fractions versus water DOC concentrationMINTEQ (see Methods section). Model parameters are defined in the Methods section.

Environmental Science & Technology Article

dx.doi.org/10.1021/es403849d | Environ. Sci. Technol. 2014, 48, 3162−31683166

log K = 32 (Figure S1). This material is available free of chargevia the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*E-mail: Jules.Blais@uottawa.ca.

Present Address⊥School of Earth and Ocean Sciences, University of Victoria,Victoria, BC, V8W 3 V6, Canada.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSFinancial and logistical support were provided by NSERCDiscovery and Strategic grants to J.M.B. and J.P.S., NorthernContaminants Program, Polar Continental Shelf Program,Northern Scientific Training Program, Cumulative ImpactMonitoring Program, Indian and Northern Affairs Canada,Environment Canada, and Aurora College. Michael Pisaric,Joshua Thienpont, William Hurst, Kathleen Ruhland, and PeterdeMontigny provided critical field, project-design, and technicalsupport. Emmanuel Yumvihoze provided technical assistancewith Hg analyses. This article is NWT Geoscience OfficeContribution #78.

■ REFERENCES(1) Lamborg, C. H.; Fitzgerald, W. F.; O’Donnell, J.; Torgersen, T. Anon-steady-state compartmental model of global-scale mercurybiogeochemistry with interhemispheric atmospheric gradients. Geo-chim. Cosmochim. Acta 2002, 66, 1105−1118.(2) Streets, D. G.; Devane, M. K.; Lu, Z. F.; Bond, T. C.; Sunderland,E. M.; Jacob, D. J. All-time releases of mercury to the atmosphere fromhuman activities. Environ. Sci. Technol. 2011, 45, 10485−10491.(3) Cole, A. S; Steffen, A. Trends in long-term gaseous mercuryobservations in the Arctic and effects of temperature and otheratmospheric conditions. Atmos. Chem. Phys. 2010, 10, 4661−4672.(4) Dietz, R.; Outridge, P. M.; Hobson, K. A. Anthropogeniccontributions to mercury levels in present-day Arctic animals Areview. Sci. Total Environ. 2009, 407, 6120−6131.(5) Kokelj, S. V.; Jenkins, R. E.; Milburn, D.; Burn, C. R.; Snow, N.The influence of thermokarst disturbance on the water quality of smallupland lakes, Mackenzie Delta region, Northwest Territories, Canada.Permafrost Periglac. Proc. 2005, 16, 343−353.(6) Barkay, T.; Gillman, M.; Turner, R. R. Effects of dissolved organiccarbon and salinity on bioavailability of mercury. Appl. Environ.Microbiol. 1997, 63, 4267−4271.(7) Jeremiason, J. D.; Engstrom, D. R.; Swain, E. B.; Nater, E. A.;Johnson, B. M.; Almendinger, J. E.; Monson, B. A.; Kolka, R. K. Sulfateaddition increases methylmercury production in an experimentalwetland. Environ. Sci. Technol. 2006, 40, 3800−3806.(8) Watras, C. J.; Morrison, K. A. The response of two remote,temperate lakes to changes in atmospheric mercury deposition, sulfate,and the water cycle. Can. J. Fish. Aquat. Sci. 2008, 65, 100−116.(9) Gustafsson, J. P. Visual MINTEQ, A Free Equilibrium SpeciationModel. KTH, Department of Land and Water Resources Engineering,2012.(10) Milne, C. J.; Kinniburgh, D. G.; Van Riemsdijk, W. H.; Tipping,E. Generic NICA-Donnan model parameters for metal-ion binding byhumic substances. Environ. Sci. Technol. 2003, 37, 958−971.(11) Sjostedt, C. S.; Gustafsson, J. P.; Kohler, S. J. Chemicalequilibrium modeling of organic acids, pH, aluminum, and iron inSwedish surface waters. Environ. Sci. Technol. 2010, 44, 8587−8593.(12) Reinikainen, J.; Hyva rinen, H. Humic- and fulvic-acidstratigraphy of the Holocene sediments from a small lake in FinnishLapland. The Holocene 1997, 7, 401−407.

(13) Drexel, R. T.; Haitzer, M.; Ryan, J. N.; Aiken, G. R.; Nagy, K. L.Mercury(II) sorption to two Florida Everglades peats: Evidence forstrong and weak binding and competition by dissolved organic matterreleased from the peat. Environ. Sci. Technol. 2002, 36 (19), 4058−4064.(14) Ravichandran, M. Interactions between mercury and dissolvedorganic matterA review. Chemosphere 2004, 55 (3), 319−331.(15) Haitzer, M.; Aiken, G. R.; Ryan, J. N. Binding of mercury(II) toaquatic humic substances: Influence of pH and source of humicsubstances. Environ. Sci. Technol. 2003, 37, 2436−2441.(16) Dong, W.; Bian, Y.; Liang, L.; Gu, G. Binding constants ofmercury and dissolved organic matter determined by a modified ionexchange technique. Environ. Sci. Technol. 2011, 45, 3576−3583.(17) Coucke, W.; China, B.; Delattre, I.; Lenga, Y.; Van Blerk, M.;Van Campenhout, C.; Van de Walle, P.; Vernelen, K.; Albert, A.Comparison of different approaches to evaluate external qualityassessment data. Clin. Chim. Acta 2012, 413, 582−586.(18) Moore, D. R. J. et al. Developing Ambient Water QualityCriteria for Mercury: A Probabilistic Site-Specific Approach; WaterEnvironment Research Foundation: Alexandria, VA, 2003; p 184.(19) Adams, R. M.; Twiss, M. R.; Driscoll, C. T. Patterns of mercuryaccumulation among seston in lakes of the Adirondack Mountains,New York. Environ. Sci. Technol. 2009, 43, 4836−4842.(20) Kelly, C. A.; Rudd, J. W. M.; Holoka, M. H. Effect of pH onmercury uptake by an aquatic bacterium: Implications for Hg cycling.Environ. Sci. Technol. 2003, 37 (13), 2941−2946.(21) Chetelat, J.; Amyot, M.; Garcia, E. Habitat-specific bioaccumu-lation of methylmercury in invertebrates of small mid-latitude lakes inNorth America. Environ. Pollut. 2011, 159 (1), 10−17.(22) Hill, J. R.; O’Driscoll, N. J.; Lean, D. R. S. Size distribution ofmethylmercury associated with particulate and dissolved organicmatter in freshwaters. Sci. Total Environ. 2009, 408, 408−414.(23) O’Driscoll, N. J.; Siciliano, S. D.; Peak, D.; Carginan, R.; Lean,D. R. S. The influence of forestry activity on the structure of dissolvedorganic matter in lakes: Implications for mercury photoreactions. Sci.Total Environ. 2006, 366, 880−893.(24) Porcal, P.; Koprivnjak, J. F.; Molot, L. A.; Dillon, P. J. Humicsubstances-part 7: The biogeochemistry of dissolved organic carbonand its interactions with climate change. Environ. Sci. Pollut. Res. 2009,16, 714−726.(25) Golding, G. R.; Kelly, C. A.; Sparling, R.; Loewen, P. C.; Rudd,J. W. M.; Barkay, T. Evidence for facilitated uptake of Hg(II) by Vibrioanguillarum and Escherichia coli under anaerobic and aerobicconditions. Limnol. Oceanogr. 2002, 47, 967−973.(26) Remucal, C. K.; Cory, R. M.; Sander, M.; McNeill, K. Lowmolecular weight components in an aquatic humic substance ascharacterized by membrane dialysis and orbitrap mass spectrometry.Environ. Sci. Technol. 2012, 46, 9350−9359.(27) Garcia, E.; Amyot, M.; Ariya, P. A. The relationship betweenDOC photochemistry and mercury redox transformations in temper-ate lakes and wetlands. Geochim. Cosmochim. Acta 2005, 69, 1917−1924.(28) Driscoll, C. T.; Yan, C.; Schofield, C. L.; Munson, R.; Holsappl,J. The mercury cycle and fish in the Adirondack Lakes. Environ. Sci.Technol. 1994, 28, 136−143.(29) Belger, L.; Forsberg, B. R.. Factors controlling Hg levels in twopredatory fish species in the Negro River basin, Brazilian Amazon. Sci.Total Environ. 2006, 367, 451−459.(30) Chiasson-Gould S. A.; Blais, J. M.; Poulain, A. J. Dissolvedorganic matter kinetically controls mercury bioavailability to bacteria.Environ. Sci. Technol. 2014, (Companion to this paper) doi: 10.1021/es4038484.(31) Pienitz, R.; Smol, J. P.; Lean, D. R. S. Physical and chemicallimnology of 59 lakes located between the southern Yukon and theTuktoyaktuk Peninsula, Northwest Territories (Canada). Can. J. Fish.Aquat. Sci. 1997, 54, 330−346.(32) Pienitz, R.; Smol, J. P.; Lean, D. R. S. Physical and chemicallimnology of 24 lakes located between Yellowknife and Contwoyto

Environmental Science & Technology Article

dx.doi.org/10.1021/es403849d | Environ. Sci. Technol. 2014, 48, 3162−31683167

Lake, Northwest Territories (Canada). Can. J. Fish. Aquat. Sci. 1997,54, 347−358.(33) Lim, D. S. S.; Douglas, M. S. V.; Smol, J. P.; Lean, D. R. S.Physical and chemical limnological characteristics of 38 lakes andponds on Bathurst Island, Nunavut, Canadian High Arctic. Internat.Rev. Hydrobiol. 2001, 86, 1−22.(34) Macdonald, R. W.; Harner, T.; Fyfe, J. Recent climate change inthe Arctic and its impact on contaminant pathways and interpretationof temporal trend data. Sci. Total Environ. 2005, 342, 5−86.

Environmental Science & Technology Article

dx.doi.org/10.1021/es403849d | Environ. Sci. Technol. 2014, 48, 3162−31683168