Mercury Export to the Arctic Ocean from the Mackenzie River, Canada

Jun 8, 2013 - ABSTRACT: Circumpolar rivers, including the Mackenzie River in Canada, are sources of the contaminant mercury (Hg) to the Arctic Ocean, ...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/est

Mercury Export to the Arctic Ocean from the Mackenzie River, Canada Craig A. Emmerton,*,† Jennifer A. Graydon,*,† Jolie A. L. Gareis,‡ Vincent L. St. Louis,† Lance F. W. Lesack,‡ Janelle K. A. Banack,§ Faye Hicks,§ and Jennifer Nafziger§ †

Department of Biological Sciences, University of Alberta, Edmonton, Alberta T6G 2E9, Canada Department of Geography, Simon Fraser University, Burnaby, British Columbia V5A 1S6, Canada § Department of Civil & Environmental Engineering, University of Alberta, Edmonton, Alberta T6G 2R3, Canada ‡

S Supporting Information *

ABSTRACT: Circumpolar rivers, including the Mackenzie River in Canada, are sources of the contaminant mercury (Hg) to the Arctic Ocean, but few Hg export studies exist for these rivers. During the 2007−2010 freshet and open water seasons, we collected river water upstream and downstream of the Mackenzie River delta to quantify total mercury (THg) and methylmercury (MeHg) concentrations and export. Upstream of the delta, flow-weighted mean concentrations of bulk THg and MeHg were 14.6 ± 6.2 ng L−1 and 0.081 ± 0.045 ng L−1, respectively. Only 11−13% and 44−51% of bulk THg and MeHg export was in the dissolved form. Using concentration−discharge relationships, we calculated bulk THg and MeHg export into the delta of 2300−4200 kg yr−1 and 15−23 kg yr−1 over the course of the study. Discharge is not presently known in channels exiting the delta, so we assessed differences in river Hg concentrations upstream and downstream of the delta to estimate its influence on Hg export to the ocean. Bulk THg and MeHg concentrations decreased 19% and 11% through the delta, likely because of particle settling and other processes in the floodplain. These results suggest that northern deltas may be important accumulators of river Hg in their floodplains before export to the Arctic Ocean.

■

only during ice-free conditions.9,10 This paucity of studies on large circumpolar rivers is noteworthy considering Hg loadings to Arctic Ocean ecosystems have the potential to impact local indigenous communities due to bioaccumulation of MeHg in marine mammals and fish in their traditional diets.5,11 MeHg is the ultimate species of importance in ecosystems; however, inorganic Hg can fuel methylation in the water column of Arctic marine waters,12 making it important to quantify both MeHg and THg fluxes from rivers draining to the Arctic Ocean. The Mackenzie River is the largest north-flowing river in North America and the fourth largest river by discharge that empties into the Arctic Ocean.13 The Mackenzie has a hydrological regime in which basin snowmelt drives a major spring flow event. Snow in the large river basin melts south to north, delivering runoff toward still frozen downstream portions of the river. This unequal melting creates ice jamming, high water levels, and localized flooding before runoff and flow decrease toward late summer and autumn. Mackenzie River flooding is most striking throughout its delta. The river’s main stem splits into a network of distributary channels near the

INTRODUCTION The Arctic Ocean is the most estuarine of the world’s oceans, and its sensitive coastal ecosystems are strongly affected by water and nutrients delivered by circumpolar rivers.1,2 Rapid warming of the North has increased the collective freshwater flow to the Arctic Ocean from rivers by as much as 7% over the past 60 years,3 likely increasing the mass flux of nutrients and contaminants such as mercury (Hg) to Arctic regions.4 This Hg flux is potentially accentuated by long-range atmospheric transport of Hg to remote regions of the Arctic5 from industrial emissions in more populated southerly latitudes. It has recently been estimated, using seasonal atmospheric measurements and a three-dimensional ocean-atmosphere model, that circumpolar rivers are a significant source of Hg to the Arctic Ocean.6 However, the authors highlighted the lack of contemporary, seasonal sampling of Hg in large Arctic rivers. To date, only a few studies have attempted to quantify concentrations and export of total Hg (THg; all forms of Hg in a sample) and methyl Hg (MeHg; a toxic form of Hg that bioaccumulates in organisms and biomagnifies through food webs) in the largest north-flowing rivers in Eurasia and North America: THg concentrations were measured in the three largest Eurasian Arctic rivers during the late summers of 1991 and 1993;7 the lower Yukon River was sampled between 2001 and 2005;8 and the Mackenzie River was sampled between 2003 and 2005, but © 2013 American Chemical Society

Received: Revised: Accepted: Published: 7644

February 14, 2013 June 3, 2013 June 7, 2013 June 8, 2013 dx.doi.org/10.1021/es400715r | Environ. Sci. Technol. 2013, 47, 7644−7654

Environmental Science & Technology

Article

Figure 1. Location of river (upstream of delta) and distributary channel (downstream of delta) sampling sites in the Mackenzie River delta region and locations of three delta lakes sampled in 2010 near Inuvik, Northwest Territories, Canada.

community of Tsiigehtchic, N.W.T., Canada and flows 200 km downstream through a large sand-silt delta containing over 45 000 floodplain lakes and wetlands14 (Figure 1). The much smaller Peel River also drains into the western delta. During high flow periods (especially in spring), river water can spill into floodplain lakes and be biogeochemically altered by autochthonous primary production, erosion, particle sedimentation, and inundation of organic material.15 Modified floodwater then exits the floodplain through distributary channels into the ocean as river levels decline. Calculations of Hg export from rivers require accurate measurements of water discharge and Hg concentrations.16−18 Numerous factors have complicated previous attempts to quantify Mackenzie River Hg concentrations and export to the Arctic Ocean. First, concentrations of Hg in river water during annual breakup of ice cover on the river have never been adequately characterized due to dangerous sampling conditions. Lack of adequate sampling during this high-flow period results in poorly constrained Hg fluxes used in mass balance assessments of Arctic marine systems because the Mackenzie

River potentially delivers more than half of all riverine sediment reaching the Arctic Ocean.13 Second, estimates of Hg export during ice-breakup are currently poor because of the large uncertainty associated with discharge measurements taken in late May and early June. During this period, winter ice breaks up and fills river channels, causing wide-scale and temporally variable ice jamming. This situation compromises the discharge−water level relationships developed at Water Survey of Canada (WSC) river gauging stations upstream of the delta. Although WSC uses several approaches to correct for ice-effects on water discharge, the error associated with the water discharge estimate is still large during this time,19,20 and this issue is yet to be resolved adequately. Finally, whereas continuous discharge of both the Mackenzie and Peel rivers has historically been monitored upstream of the delta by WSC, water exiting the delta is partitioned among more than five distributary channels, and the discharges among these channels are not presently known because of complex ocean backwater effects. For this reason, previous estimates of Mackenzie River Hg export to the Arctic Ocean employed flow records and 7645

dx.doi.org/10.1021/es400715r | Environ. Sci. Technol. 2013, 47, 7644−7654

Environmental Science & Technology

Article

floats. All samples were immediately stored in dark coolers until return to the laboratory (within six hours). One set of THg and MeHg samples from each site was then filtered through 0.45 μm nitrocellulose membranes in acid washed Nalgene filter units (part no. 450-0045). All THg samples were preserved with trace-metal-grade HCl equal to 0.2% of the sample volume and stored in cool, dark conditions. MeHg samples, all of which were initially collected or filtered into Teflon bottles, were frozen until analysis. At each site, separate water samples were collected concurrently for general nutrient chemistry (J. Gareis, unpublished data). A subset of floodplain lakes near Inuvik, N.W.T., Canada (Figure 1) was also sampled during the 2010 open water season for THg, MeHg, and nutrient analyses, similar to river sites (Figure S3). Analytical Methodology. Water samples were analyzed for THg using EPA method 1631 with minor modifications.25,26 Briefly, all Hg in samples was oxidized to Hg(II) by the addition of bromine monochloride (BrCl; hydroxylamine hydrochloride [NH2OH·HCl] was used for neutralization of free halogens after incubation), reduced to Hg(0) using stannous chloride (SnCl2), purged onto gold traps, and thermally desorbed and analyzed by cold vapor atomic fluorescence spectrometry (CVAFS; Tekran model 2600). Analysis was completed at the University of Alberta Biogeochemical Analytical Service Laboratory (BASL; Edmonton, AB). All samples collected for MeHg were distilled at 120 °C with potassium chloride (KCl) and sulfuric acid (H2SO4) to remove matrices that may interfere with the subsequent aqueous phase ethylation process.25,27 Five picograms of an enriched stable MeHg isotope (Me201Hg, 96.2%) was added to each still prior to distillation as an internal standard. The pH of the distillate was adjusted to 4.9 using acetate buffer. The distillate was ethylated using sodium tetraethyl borate (NaBEt4) and purged with ultra high purity nitrogen gas (N2). Volatile Hg species were trapped using Tenax, thermally desorbed, separated using a gas chromatography column, and detected by inductively coupled-plasma mass spectrometry (ICP-MS, Perkin-Elmer ELAN DRC-e). ICP-MS quantifies concentrations of individual Hg isotopes, and to calculate concentrations of ambient MeHg, an isotope other than the internal standard (i.e., Me200Hg) was used as an ambient Hg surrogate. The BASL successfully participated in international interlaboratory comparisons over the course of this study. Spike recoveries for ambient THg and MeHg samples were generally >95% and >75%, respectively. MeHg concentrations were adjusted using individual internal standard spike recoveries. Limits of detection (LODs) for analyses of bulk and dissolved THg was 0.05 ng L−1 and 0.015 ng L−1 for bulk and dissolved MeHg analyses. THg was detectable in all samples, and 12% of MeHg samples were below the LOD. THg duplicate analyses were within 10%. MeHg duplicate analyses were generally within 30%, except in some cases where concentrations were extremely low and very near the LOD. Quantification of Hg Export from the Mackenzie and Peel Rivers into the Delta. We used Hg concentrations from periodic water sampling between 2007 and 2010, daily flow records, and a modeling approach to quantify annual export of Hg from the Mackenzie and Peel rivers. We used Loadrunner28 as a user interface to automate model runs of the United States Geological Survey Load Estimator (LOADEST) model software.29 LOADEST fits calibration equations to paired measures of sample concentration and river discharge. These

water sampling programs upstream of the delta and ignored floodplain processes that may affect Hg concentrations and export. Between 2007 and 2010, we had the opportunity to address some of these issues affecting Hg concentration and export estimates, including the lack of sampling during ice-breakup and the effects of the delta floodplain on Hg chemistry in the river. The goals of this study were to: (1) measure bulk (unfiltered) and dissolved (filtered) THg and MeHg concentrations throughout the freshet and open water periods in rivers entering and distributary channels exiting the delta; (2) quantify annual THg and MeHg export from the gauged Mackenzie and Peel rivers entering the delta; and (3) estimate THg and MeHg export to the Arctic Ocean using relative changes in Hg concentrations in river water entering and exiting the delta.

■

METHODS Discharge into the Mackenzie River Delta from the Mackenzie and Peel Rivers. Approximately 90% of the water flowing into the delta is measured by the Mackenzie River at Arctic Red River gauging station, with the balance measured at the Peel River above Fort McPherson gauging station (7%) and other unmeasured sources. Most water (∼80% of the water flowing into the delta) exits the delta through four main distributary channels: Reindeer, Middle, East, and West21,22 (Figure 1). Local runoff and precipitation are minor components of the delta’s annual hydrology.23 Sample Collection for Hg Concentrations in Rivers and Distributary Channels. Between 2007 and 2010, we collected samples for THg and MeHg concentration analyses from the Mackenzie and Peel rivers upstream of the delta and from the four major distributary channels flowing from the delta into the Arctic Ocean (Figure 1). We also collected samples at a single mid-delta station (Figures 1 and S1). Samples were collected during three periods within the annual hydrograph: (1) the spring rising limb period from the initial sharp rise in flow in late April to peak flow; (2) the falling limb period from peak flow to day 190 (latest peak flow on record; Lesack et al., submitted); and (3) the open water period from day 190 to the first ice observation in October. Freshet is defined here as the sum of periods (1) and (2). We weighted our sampling to the first two periods which allowed us to collect samples over about half of the annual runoff volume each year, except in 2009 when a reduced campaign limited our sampling to about 10% of the annual runoff volume (Figure S2). We were unable to collect samples during the winter base flow period (first ice observation in October to initial rising water in late April) for logistical reasons. Duplicate THg and MeHg samples were collected at each site into either precleaned ESS amber glass bottles used for trace contaminant sampling (part no. 0125-0150-QC) or acidwashed Teflon bottles. Strict “clean-hands/dirty-hands” sampling protocols were used to ensure samples were not contaminated.24 In mid-May, under-ice samples were collected by augering a hole through the ice at midchannel and deploying an acid-cleaned Teflon Kemmerer into the river flow to midcolumn depth and then pouring off the sample into bottles. During ice-breakup and open-water conditions, surface water from the Mackenzie and Peel rivers was collected directly into bottles upstream of helicopter floats or a boat at midchannel. Surface water from distributary channels flowing out of the delta was collected directly into bottles upstream of helicopter 7646

dx.doi.org/10.1021/es400715r | Environ. Sci. Technol. 2013, 47, 7644−7654

Environmental Science & Technology

Article

Table 1. Summary of Discharge and Bulk and Dissolved THg and MeHg Export (±1 SE) Upstream and Delta Effect-Corrected Export Downstream of the Mackenzie River Delta between 2007 and 2010 to deltaa 3

−1 d

discharge (km yr )

bulk THg (kg yr−1)

dissolved THg (kg yr−1)

bulk MeHg (kg yr−1)

dissolved MeHg (kg yr−1)

to ocean delta effect (%)b

year

MR

PR

total

2007 2008 2009 2010 2007 2008 2009 2010 2007 2008 2009 2010 2007 2008 2009 2010 2007 2008 2009 2010

330 323 334 291 3463 3152 3788 1994 404 373 424 274 20 18 21 13 9 9 9 7

15 22 23 20 222 374 400 307 20 33 35 27 1 2 2 2 0 0 1 0

346 345 357 312 3686 ± 295 3526 ± 309 4189 ± 374 2301 ± 195 424 ± 40 406 ± 42 459 ± 47 301 ± 31 21 ± 2 20 ± 2 23 ± 3 15 ± 2 10 ± 1 9±1 10 ± 1 7±1

100

81 ± 12

83 ± 12

89 ± 14

75 ± 10

total

watershed Hg yield (μg m−2)c

346 345 357 312 2976 2847 3382 1858 353 338 383 251 18 18 20 13 7 7 8 6

2.1 2.0 2.4 1.3 0.24 0.23 0.26 0.17 0.012 0.011 0.013 0.008 0.005 0.005 0.006 0.004

a

MR: Mackenzie River at Arctic Red River; PR: Peel River above Fort McPherson. bFrom eqs 1 to 3 in Methods section; see Table S1. cWatershed Hg yield is based on a combined MR and PR basin area of 1 749 700 km2. dNo net storage of water in the delta assumed over annual period.

relationships are then used with the complete flow data set30 to quantify daily and annual mass fluxes of Hg. We used the LOADEST log−linear model 1 (a0 + a1 ln[flow]) to fit paired flow and Hg concentrations from the Mackenzie and Peel rivers because, of all the equations available through the LOADEST program, it provided the strongest correlations. Discharge− concentration relationships were statistically significant for both bulk Hg (r2 = 0.38−0.68; p < 0.01) and dissolved Hg (r2 = 0.10−0.60; p < 0.01). We also used these relationships to model Hg concentrations for the winter period for which Hg concentration data were not available and the baseline discharge was relatively low. Modification of Mackenzie River Hg Concentrations by the Delta. Water discharge from distributary channels exiting the delta was not monitored due to complex hydrological regimes that are currently not well understood because of short levees, shallow depths, ice-effects, and marine storm surge influences. Therefore we could not directly compare export of Hg downstream of the delta with that entering the delta. However, using our broad sampling program, we could compare our sample Hg concentrations upstream and downstream of the delta as a semiquantitative evaluation of the delta’s influence on Hg river chemistry (i.e., “delta effect”), assuming that the volume of water entering and exiting the delta was for all intents and purposes the same. For water flowing into the delta at the Mackenzie and Peel gauging stations, we determined a flow-weighted mean concentration (FWMC) for each Hg species using all collected samples between 2007 and 2010:

and WRIVER is the fraction of total flow into the delta contributed by each river.31 For the downstream distributary channels, we were forced to use a different approach to calculating FWMCs because of the lack of continuous discharge measurements. Representative downstream concentrations of all Hg samples were calculated using an arithmetic mean approach (AMC): AMCDS = Wreindeer[∑ (Ci /n)] + Wmiddle[∑ Ci /n] + Weast[∑ Ci /n] + Wwest[∑ Ci /n]

where Ci represents sample concentrations and Wriver is the fraction of total flow from the delta contributed by each distributary channel. This approach assumes all downstream water is represented by these four channels. Downstream Wriver values were based on manual discharge measurements in 2008 by researchers from the International Polar Year project “Study of Canadian Arctic River-delta Fluxes” (IPY-SCARF) and WSC.22 To estimate a delta effect correction, we took the quotient of FWMCUS and AMCDS for each Hg species: delta effect(Hg) = AMC DS /FWMCUS

(3)

This fixed correction, calculated for each Hg species, was multiplied by upstream mass fluxes to estimate export of Hg to the Arctic Ocean during the study period.

■

RESULTS AND DISCUSSION Discharge from the Mackenzie and Peel Rivers into the Mackenzie River Delta. Annual water discharge between 2007 and 2010 from the Mackenzie River was 291−334 km3 yr−1 (Table 1), with the first three years well above the longterm mean of 291 km3 yr−1 (1973−2010).31 Peel River discharge over the same period ranged from 15 to 23 km3 yr−1. Summed total discharge into the delta was 312−357 km3 yr−1.

FWMCUS = WMackenzie[∑ (CiQ i)/∑ (Q i)] + WPeel[∑ (CiQ i)/∑ (Q i)]

(2)

(1)

where Ci represents sample concentrations, Qi represents corresponding measured daily discharge on the sampling day, 7647

dx.doi.org/10.1021/es400715r | Environ. Sci. Technol. 2013, 47, 7644−7654

Environmental Science & Technology

Article

Figure 2. THg and MeHg concentrations in river water entering (rivers) and exiting (channels) the Mackenzie River delta. Concentrations plotted on gauged discharge31 on inflowing rivers (dark gray) and on the sum of inflowing discharge for distributary channels (light gray; for illustrative purposes only; flow was not measured at downstream sampling stations). 7648

dx.doi.org/10.1021/es400715r | Environ. Sci. Technol. 2013, 47, 7644−7654

Environmental Science & Technology

Article

Figure 3. Sample mean flow-weighted concentrations of bulk and dissolved THg and MeHg from the Mackenzie and Peel rivers during the rising, falling, and open water periods of the annual hydrograph (see Methods for definitions of these periods). Winter flow-weighted mean concentrations are estimated from fit relationships between Hg concentrations and river discharge (i.e., not directly based on any samples collected during winter). Limits of detection for THg and MeHg analyses were 0.05 ng L−1 and 0.015 ng L−1, respectively. * indicates statistically significant differences in Hg concentrations relative to other periods (Kruskal−Wallis nonparametric test; α = 0.05).

concentration (0.081 ± 0.045 ng L−1) is the first reported in the Mackenzie River for the freshet-open water period. Mackenzie dissolved THg (1.6 ± 0.6 ng L−1) and MeHg (0.033 ± 0.016 ng L−1) were lower than observed previously in the river (2.8 ng L−1 and 0.076 ng L−1, respectively, for THg and MeHg).9 Bulk THg and MeHg concentrations in the Peel River (22.8 ± 11.8 ng L−1 and 0.125 ± 0.090 ng L−1) were higher than those in the Mackenzie, whereas dissolved THg and MeHg concentrations in the Peel (2.0 ± 0.8 ng L−1 and 0.028 ± 0.017 ng L−1) were similar to those in the Mackenzie. Most bulk THg in the Mackenzie and Peel rivers was particlebound (87.7 ± 5.9% and 89.9 ± 5.1%, respectively), similar to findings from the Yukon River.8 MeHg concentrations in the Mackenzie and Peel rivers were less particle-bound (51.9 ± 26.1% and 69.0 ± 22.1%, respectively) than bulk THg. The percent bulk THg present as bulk MeHg (%MeHg; a proxy for rates of MeHg production in watersheds) was low in the Mackenzie (0.2−1.5%) and Peel (0.2−1.3%) rivers during the freshet-open water seasons. Shuster et al.8 showed similar to higher %MeHg values in samples from the Yukon River (0.1− 9%). Concentrations of all Hg species in the Mackenzie and Peel rivers corresponded with river discharge (Figures 2−4). Flowweighted mean concentrations of bulk THg and MeHg in each river were highest during the rising limb of the hydrograph each spring during melt, and decreased during the subsequent falling limb of the hydrograph and during the open water summer period (Figure 3). However, there were no clear patterns of hysteresis (i.e., systematic differences between rising and falling limbs of the hydrograph) in the Hg concentration data (Figure S4). The percent particle-bound THg in the Mackenzie changed very little between the rising limb and open water periods (3% increase), whereas the percent particle-bound MeHg decreased 16% over the same period. %MeHg decreased

Generally, seasonal patterns of flow in the Peel River were similar to the Mackenzie River; however the Peel showed sharper seasonal transitions due to its mountainous watershed and flashier hydrology (Figure S2). At the start of May of each year, flow increased rapidly in the Mackenzie and Peel rivers due to snowmelt throughout the basin (Figure S2). Total flow from both rivers during the rising period (which lasted between 16 and 34 days) was between 36 and 41 km3, representing only 11−12% of annual flow. In the Mackenzie River, the peak mean daily flow was between 24 200 and 30 000 m3 s−1 and occurred between May 19 and May 30 over the four year study period. Ice breakup occurred during this time, causing variable ice jamming and water level fluctuation. Ice cleared from the river approximately 1−3 weeks after peak flow. The duration of the falling period in the Mackenzie and Peel rivers was 39−51 days, during which time 75−96 km3 of water was delivered to the delta (23−27% of annual flow). The open water period began after day 190 and lasted between 87 and 108 days. Between 125 and 138 km3 of water from Mackenzie and Peel rivers was delivered to the delta, representing 37−40% of total annual flow. Precipitation events in each river basin during open water periodically increased river flow. The winter period was the longest of the hydrological seasons, lasting between 181 and 207 days. Mean daily flow in the Mackenzie River during the winter period averaged only 5000 m3 s−1, and total water discharge to the delta was 76−98 km3 (23−28% of annual flow). Hg Concentrations in the Mackenzie and Peel Rivers Upstream of the Delta and Distributary Channels Exiting the Delta into the Arctic Ocean. Mackenzie River flow-weighted (mean ±1 SD) bulk THg concentration (14.6 ± 6.2 ng L−1) from our study was considerably larger than that reported for the river in a recent summary of Hg concentrations in large circumpolar rivers (7.2 ng L−1).8 The mean bulk MeHg 7649

dx.doi.org/10.1021/es400715r | Environ. Sci. Technol. 2013, 47, 7644−7654

Environmental Science & Technology

Article

Figure 4. Discharge−Hg concentration relationships for samples collected at the Mackenzie River at Arctic Red River and Peel River above Fort McPherson. Best-fit relationships were generated using LOADEST model 1 (α = 0.05).

0.23% during the freshet-open water period. Modeled flowweighted mean bulk THg and MeHg concentrations in the Mackenzie River during winter (Figures 3 and 4) were very low (1.4 ± 0.3 ng L−1 and 0.015 ± 0.005 ng L−1), with 75 ± 28% and 11 ± 5% of the THg and MeHg bound to particles, respectively. Concentrations of Hg species in distributary channels primarily discharging Mackenzie River water (Reindeer, Middle, East channels) from the delta had lower Hg concentrations compared to concentrations measured upstream of the delta (Figure 2, Table S1). The West Channel, which primarily discharges Peel River water from the delta, also had lower Hg concentrations compared to those upstream of the delta. General seasonal patterns in Hg concentration were similar in distributary channels and upstream rivers (Figure S5). THg and MeHg Export from the Mackenzie and Peel Rivers into the Delta. Using statistically strong flow-Hg concentration relationships (Figure 4), we quantified export of Hg from the Mackenzie and Peel rivers into the delta (Tables 1, S2). Export of bulk THg from the Mackenzie and Peel rivers into the delta was between 2301 and 4189 kg yr−1 from 2007 to 2010, with 54−64% percent of the annual flux delivered during freshet and 32−41% during open water conditions. Dissolved THg comprised only 11−13% of annual THg export. The Mackenzie and Peel rivers delivered 15−23 kg yr−1 of bulk MeHg to the delta from 2007 to 2010 (Table 1), with 50−58% of the flux occurring during freshet and 36−42% during open water conditions. Dissolved MeHg comprised a larger proportion of the annual bulk MeHg flux (44−51%) than did dissolved THg to the bulk THg flux (above). Rivers with strong seasonal changes in flow, including northern rivers, often show strong positive correlations between river Hg concentrations,

particles, dissolved organic carbon (DOC), and water flow.8,32−36 All bulk and dissolved Hg species in the current study were significantly correlated with river flow (Figure 4), DOC (r2 0.14−0.54, p < 0.03), and all but dissolved MeHg were correlated with total suspended solids (TSS; r2 0.17−0.94, p < 0.02). Annual snowmelt in the Mackenzie River basin delivers a flush of sediments from surface organic soil layers, erodible terraces in mountainous regions, and its own river banks during ice breakup.37,38 These processes also transport Hg bound to organic and inorganic material.8,39 Carrie et al.40 found that the majority of the annual Hg flux in the Mackenzie River originated from mountainous tributaries delivering mineral-bound Hg during freshet. Our results are consistent with this finding, as we observed higher bulk THg yield from the almost exclusively mountain-draining Peel River (3−6 μg m−2 yr−1; basin area 70 600 km2) compared to the Mackenzie River which has a mountainous, but more diverse watershed (1−2 μg m−2 yr−1; basin area 1 679 100 km2). Our observations of relatively low MeHg export and low %MeHg in the Mackenzie River also suggest a dominant mountain-based source of Hg from that river. Several studies have positively associated MeHg export from rivers with percentage wetland coverage (organic matter stores) in river basins,41−43 because anoxic wetland sediments are primary sites of Hg methylation. In the Mackenzie River basin, the largest wetland complex is located at the upstream Peace-Athabasca delta.44 Any methylation that may occur there would unlikely be reflected in the Mackenzie River as this MeHg would have to first pass through the Slave River and Great Slave Lake before discharging to the Mackenzie River. Alternatively, mountainous watersheds would likely not sustain significant MeHg production because of their relief, low wetland coverage, 7650

dx.doi.org/10.1021/es400715r | Environ. Sci. Technol. 2013, 47, 7644−7654

Environmental Science & Technology

Article

lakes and wetlands could be notable sources of MeHg to distributary channels during the summer when it is warmest, even though lake drainage occurs in only about 12% of the lakes at that time.52 For example, increases in bulk MeHg concentrations and %MeHg in lake water at peak summer temperatures (mid-July) have been previously observed in six floodplain lakes.10 Finally, there is evidence of postflood storm surges entering the delta from the coast, which occur later in summer when coastal sea ice recedes.53 These surges can raise local river levels and flush low-lying delta lakes. In 2008 and 2009, storm surge related increases in water levels in the delta occurred in midsummer, at a time when in-lake methylation has previously been observed to be at its peak.10 Overall, the net effect of processes removing MeHg from river water appeared to be larger than those adding to it. Estimates of Hg export downstream of the delta are crucial for large-scale Hg modeling in the Arctic Ocean. We applied our delta effect corrections from eq 3 to our upstream Hg import fluxes to provide adjusted export estimates to the coastal Arctic Ocean (Table 1). Export of bulk THg from the delta to the Arctic Ocean was 1850−3400 kg yr−1 from 2007 to 2010, of which 250−380 kg yr−1 was in the dissolved form (Table 1). Export of bulk MeHg from the delta during the same period was 13−20 kg yr−1, of which 6−8 kg yr−1 was in the dissolved form (Table 1). Future advancements in the understanding of the hydrology of downstream distributary channels should increase the precision of our estimates and provide stronger quantification of direct Hg export to the Arctic Ocean. Hg has historically been undersampled in northern rivers, and periodic ice-effects hinder accurate annual discharge measurements. Therefore, reconstructions of Hg export from northern rivers have been problematic.8 To support future Hg studies, we attempted to bracket historical Hg export from the Mackenzie River delta using historical discharge records from the Mackenzie and Peel river gauging stations and the strong interannual Hg-discharge relationships we established with our current study (which incorporated 35% of the historical range of flow from the rivers; Figure S6; Table S3). Historically (1975−2010), the Mackenzie and Peel rivers delivered between 228 and 357 km3 of water to the Arctic Ocean31 with a mean of 310 km3 (assuming net zero floodplain storage over a full year). This results in an estimated range of bulk THg and MeHg delivery to the coastal Arctic for the last 35 years of between 450 and 3350 kg yr−1 (mean 1944 ± 267 [95% C.I.] kg yr−1) and 5.0−20.0 kg yr−1 (mean 13.2 ± 1.4 kg yr−1), respectively. The large interannual range of Hg export is supported by longterm export estimates of TSS by Carson et al.38 The authors quantified a range of TSS export of 40−160 Mt yr−1 (factor of 4) between 1974 and 1994. Our bulk THg export estimates show larger relative ranges (factor of 7); however our equations were developed on data collected during higher water years and over a shorter time period compared to the Carson et al. study. Fisher et al.6 suggested that large circumpolar rivers may provide a substantial source of Hg to the Arctic Ocean based on ocean-atmosphere modeling of Hg in the Arctic. The authors estimated that 80 000 kg yr−1 of Hg may be delivered to the Arctic Ocean by circumpolar rivers. Our results suggest that the Mackenzie River, which delivered 1850−3400 kg yr−1 of THg between 2007 and 2010, is only a small contributor of this total circumpolar river Hg flux. Further, Hg in the Mackenzie and Peel rivers is predominantly particle-bound and influenced by physical processes in the Mackenzie River delta and estuary that decrease mass export delivered past continental shelves.

shallow soils, cold temperatures, and turbulent, oxygenated waters. Increases in the export of dissolved THg and MeHg with flow from the Mackenzie River are also likely linked to flushing of soils and delivery of organic and inorganic particles downstream. Hg tends to bind to small organic and inorganic particles, some of which may have passed through our 0.45 μm filters. Whitehouse et al.45 found that a significant portion of Mackenzie River total organic carbon (∼20%) was in the colloidal (