Journal of Geophysical Research: Biogeosciences

Seasonal methylmercury dynamics in water draining three beaver impoundments of varying age

Authors


Abstract

[1] We monitored water chemistry including unfiltered and filtered total mercury (THg), total methylmercury (TMeHg) and redox-sensitive species in three beaver pond inlets and outlets in southwestern Quebec (Canada) on a monthly basis from March to September 2007. In-pond methylation efficiencies (percent of THg as TMeHg) peaked when ponds were ice-covered (%TMeHg/THg range: 53–80%) and in summer (%TMeHg/THg range: 34–67%). Low oxygen concentrations during these periods likely promoted reducing conditions leading to inorganic mercury methylation. Total dissolved MeHg was the predominant fraction of TMeHg in outlets (80% on average), and TMeHg values were up to 27–fold higher in outlets compared to inlets in summer. During the summer, TMeHg concentrations (up to 2.80 ng L−1) were higher in the recent pond than in the older ones. Seasons and pond age influence MeHg production dynamic in beaver ponds; beaver pond formation may increase MeHg transfer through food webs within the pond area and in downstream aquatic systems.

1. Introduction

[2] Health risks from methylmercury (MeHg) in fish have been the subject of several large epidemiological investigations leading to governmental advisories on fish consumption [Clarkson and Magos, 2006]. High MeHg levels in freshwater fish of northeastern North America have been linked to high deposition of atmospheric mercury from industrial emissions [Hammerschmidt and Fitzgerald, 2006; Harris et al., 2007]. In addition to atmospheric deposition, many factors influence amount and transport of mercury in aquatic systems, including physical characteristics of watersheds (e.g., land cover, soil type and erosion), biogeochemical controls (associations with dissolved organic carbon, ionic strength) and seasonal dynamics (storm events, temperature extremes, snowmelt) [Babiarz et al., 1998]. However, watershed disturbances made by beavers (Castor canadensis) are seldom mentioned to affect mercury chemistry in streams.

[3] By flooding forested environments, beavers create wetlands that might favor microbial activity and methylation of inorganic mercury. Driscoll et al. [1995] were the first to demonstrate a rise of MeHg concentrations in water draining a long-established beaver pond located in the Adirondacks region (NY, USA). Moreover, experimentally flooded wetland (ELARP) and upland forest (FLUDEX) in Ontario (Canada) exhibited net MeHg production in flooded peat and soils and higher MeHg concentrations in surface water than those for unflooded wetlands or lakes [St. Louis et al., 2004; Hall et al., 2005].

[4] Over the past 40 years, beaver populations in northeastern North America have been on the rise [Driscoll et al., 1998; Fortin et al., 2001]. Beavers were previously regulated by both humans and wolves, their principal natural predators. Currently, beaver densities have increased markedly in certain areas of the Canadian Shield in response to decreased predation [Fortin et al., 2001]. The resulting increase in beaver dams and flooded forested areas may favor the production and transfer of MeHg to streams and lakes.

[5] Most studies on beaver impoundments have shown that reducing conditions prevail during the summer, when ponds are sinks of sulfate and nitrate, and are sources of dissolved organic carbon, ammonium, iron and manganese [Smith et al., 1991; Cirmo and Driscoll, 1993; Driscoll et al., 1998; Margolis et al., 2001]. The occurrence of a reducing environment during summer may promote methylation of inorganic mercury in beaver ponds; Driscoll et al. [1998] found that methylation efficiency increased from 6 to 10% during the fall/winter/spring to 16% during summer in a beaver pond outflow of Adirondacks. Because THg concentrations were remarkably constant in outlet (2.5 ± 0.22 ng L−1) throughout the study period, the methylation efficiency rise was liable to the MeHg concentration increase (from 0.20 ng L−1 in spring/fall/winter to 0.40 ng L−1 in summer).

[6] In addition to season, time since flooding may also control MeHg supply to downstream systems by regulating net methylation rates. For instance, ELARP and FLUDEX projects demonstrated a rapid pulse of MeHg production following inundation (TMeHg concentrations up to 3 ng L−1 in summer) and a subsequent decline after two years of flooding (on average between 0.46 to 0.65 ng L−1) due to microbial demethylation [St. Louis et al., 2004; Hall et al., 2005].

[7] Once MeHg is produced in a system, its fate in downstream aquatic food webs will be partly determined by partitioning between particulate and dissolved phases [Meili, 1997]. The dissolved MeHg fraction could be carried on a longer distance than the particulate one and be incorporated more easily by primary consumers [Hill et al., 1996; Schetagne et al., 2000]. MeHg exports from wetlands are predominantly associated with the filtered phase [Babiarz et al., 1998; Grigal, 2002], but no comparable data exist for water draining beaver ponds.

[8] Beaver ponds share many physical characteristics with other flooded wetlands, yet there are no comparative data on the seasonal behavior of beaver ponds of contrasting flood ages. The main objectives of this study are to (1) evaluate the influence of seasons, from late winter to the end of summer, and flooding time on mercury concentrations and on redox-sensitive species in water draining beaver ponds and (2) determine beaver pond influence on the dissolved mercury fractions in outflows. These results may influence the design of watershed-based management approaches in areas where beaver impoundment densities are high.

2. Methods

2.1. Study Area and Sampling Sites

[9] The study took place in the Laurentian Region of the Canadian Shield, located ca. 75 km north of Montreal (QC, Canada) (Figure 1). The area is underlain by granitic or anorthositic bedrock covered by 1–5 m of glacial tills. Soils are mostly Humic Cryorthods (U.S. classification) or Orthic Ferro-Humic podzols (Canadian classification) [Carignan et al., 2000].

Figure 1.

(left) The location of Quebec province in Canada. (right) The location of the three beaver ponds in southwest Quebec (open circles, recent, intermediate, and old ponds) with major cities (solid circles) in the surrounding area.

[10] Annual precipitation in the area averages 1100 mm, with 30% falling as snow [Carignan et al., 2000]. Precipitation was typical in 2007 with 285 mm of rain and 55 cm of snow during the sampling period between 8 March and 20 September 2007 (Figure 2). The first sample collection was during ice cover, the second and third collections occurred during snowmelt and the remaining samples were collected during summer (Figure 2). Ice-out occurred at the end of April for the three beaver ponds (Figure 2).

Figure 2.

Meteorological conditions during the sampling period from 8 March to 20 September 2007. The continuous line shows specific runoff of Lake Croche (L km−2 s−1), and the short-dashed line shows average temperature (°C). Specific runoff of beaver ponds (L km−2 s−1) are represented by dots (solid circles, recent; open circles, intermediate; solid triangles, old). Arrows represent the eight sampling events that occurred from late winter to the end of summer. The vertical dotted line shows the ice-out time.

[11] The Laurentian region is in the Nordic temperate vegetation zone and beaver pond catchments are primarily second-growth mixed forests. Dominant deciduous tree species are maple (Acer spp.), American beech (Fagus grandifolia Ehrh.), birch (Betula spp.) and trembling aspen (Populus tremuloides Michx.). Major coniferous trees include balsam fir (Abies basalmea (L.) Mill.), eastern white cedar (Thuja occidentalis L.), spruce (Picea spp.) and tamarack (Larix laricina (Du Roi) K.Koch).

[12] Three beaver ponds of different ages (recent, intermediate and old) were chosen within a limited 40 km2 area (45°53′30″–45°54′47″ N, 74°11′10″–74°01′44″ W) to minimize influence of vegetation, altitude (245 to 294 m) and bedrock on water chemistry. Ratios of drainage basin area to pond area are similar among sites, ranging from 15.94 to 18.97.

[13] Impoundment age was assessed from a qualitative classification based on visual clues derived from the literature [Woo and Waddington, 1990; Bigler et al., 2001; Ray et al., 2001] and from trappers' knowledge. The density of standing dead trees, the surface of open water zones, the dam condition and the type of macrophytes assemblage were the main visual signs used to establish the classification. The recent beaver pond was less than 10 years old and characterized by a high density of standing dead trees, few open water zones occupied only by free-floating macrophytes (e.g., Lemna minor) and a dam where growing shrubs (e.g., alder, willow) were absent. The old impoundment was over 20 years old and characterized by few remaining standing trees, large open water areas occupied by rooted macrophytes (e.g., Nuphar spp.). The last impoundment was between 10 and 20 years and had intermediate characteristics. Beavers were observed in all ponds.

[14] In the text, the terms “beaver pond” and “beaver impoundment” refer to the body of water created by beavers and formed behind the dam. Average water depths were 1.2 m, 1.6 m and 1.0 m for the recent, intermediate and the old beaver ponds, respectively. Beaver inlets and outlets refer to running water entering or flowing from beaver impoundments. Inlets from the recent and the old sites originated from small lake outflows, whereas inlet of the intermediate site was a first-order stream.

2.2. Water Sampling

[15] Water samples were collected in streams at the inlets and outlets on a monthly basis between 8 March and 20 September 2007, i.e., eight times over a period of 195 days. Care was taken to avoid contaminating the water samples with streambed sediments. Teflon™ vials of 50 mL for total mercury (THg) and 250 mL amber glass bottles for methylmercury (MeHg) were acid-washed and thoroughly rinsed with ultrapure water (R = 18.2 MΩ cm). Nalgene HDPE bottles washed with ultrapure water were used to sample nutrients and major anions. Prior to collecting the final sample, all bottles were rinsed three times with stream water. Samples for THg analyses were preserved with 0.4% BrCl whereas TMeHg samples were preserved with 0.4% ultraclean hydrochloric acid [Parker and Bloom, 2005] within 5 h of collections. Samples for Hg analyses were sampled and handled using ultraclean protocols described by St. Louis et al. [1994].

[16] Samples for total dissolved Hg (TDHg), total dissolved MeHg (TDMeHg), DOC, and anions (Cl, NO3, SO42−) were filtered through polyethersulfone (PES) membranes (porosity: 0.45 μm) within 5 h of collections, then preserved for Hg samples, and all kept in the dark and refrigerated (<4°C) until analysis. The amber glass vials used to store dissolved organic carbon (DOC) were heated to 550°C to eliminate carbon contamination. Field and filter blanks of ultrapure water were tested to assess contamination during sampling and filtration.

[17] Chlorophyll a samples were collected monthly in 1 L Nalgene dark bottles from 25 June to 20 September 2007 and filtered through Whatman GF/C filters which were wrapped in aluminum foil and frozen to ca. −20°C. Temperature, dissolved oxygen, pH and specific conductivity were measured using a multiprobe instrument (YSI-650 DMS). All measurements were taken 1 min after stabilization and on the same days as water samples. The multiprobe was lightly shaken to avoid oxygen depletion near the membrane. Oxygen and pH probes were calibrated every sampling day.

[18] Discrete flow was measured in the outlets using a SonTek/YSI Doppler velocimeter following USGS standard methods provided with the instrument. Current velocities were measured along at least three points in the transversal section of the stream and at a depth of 0.6 of the maximum depth. In early March and April when the streams were partially obstructed by ice, outflows were estimated from the average daily flow monitored at the nearby Lake Croche V notch (lake area: 0.19 km2, watershed area: 1.08 km2, altitude: 360 m). Specific runoffs of beaver ponds and of Lake Croche were similar over the sampling period (Figure 2). Over the study period, about two thirds of the Lake Croche discharge (total discharge: 160 m3 s−1) occurred during the 6-week snowmelt period (total snowmelt discharge: 102 m3 s−1) which took place between late March and early May. Late winter and summer were characterized by base flow conditions with occasional rain events (Figure 2).

[19] Sampling was generally conducted at least 48 h before rain events to avoid resuspension of particulate materials. Only the collection in late April occurred after a rain event. Note, however, this sampling was during high flow associated with snowmelt.

2.3. Laboratory Analyses

[20] Analyses of THg and TDHg were carried out using an automated Tekran 2600 following USEPA method 1631. Briefly, excess of BrCl was neutralized with 50 μL hydroxylamine and samples were reduced in line with stannous chloride (3%) before being purged with argon and trapped on a gold trap. Hg0 was then measured by cold vapor atomic fluorescence spectrometry (CVAFS). Prior to analysis, TMeHg and TDMeHg were first distilled to remove any matrix interferences and then ethylated with NaB(C2H5)4, followed by gas chromatography separation with CVAFS [Bloom, 1989; Horvat et al., 1993]. Analytical accuracy was checked by daily analysis of TORT-2 (Lobster Hepatopancreas) certified reference material of the National Research Council (Canada). The certified value for TORT-2 is 152 ± 13 ng g−1 and the mean concentration determined was 150 ± 9 ng g−1 (n = 69; CV = 6%). Analysis of field and procedural blanks consisting of ultrapure water revealed no detectable MeHg contamination during sampling, filtration, distillation and analysis. According to the results of Bloom et al. [1997] in brown, humic-rich water, we estimated our artifact production to be less than 1%. Method detection limits (MDL) based on three times the standard deviation of ten blanks were 0.06 ng L−1 and 0.03 ng L−1 for THg and TMeHg, respectively. The average coefficient of variation (standard deviation/mean) for field triplicate determinations was 10% for THg-TDHg and 16% for TMeHg-TDMeHg. Total inorganic mercury (TIHg) was determined by the difference between THg and TMeHg and total dissolved inorganic mercury (TDIHg) by the difference between TDHg and TDMeHg.

[21] Concentrations of DOC were measured by infrared detection after acidification and sodium persulfate oxidation using an Aurora TOC-1030 Analyzer or by high-temperature combustion on a platinum catalyst using a Shimadzu TOC-5000 Analyzer (MDLs: 0.10 mg L−1). Anions (Cl, NO3, and SO42−) were analyzed by ion chromatography (HPLC Waters) with respective MDLs of 0.04, 0.10 and 0.004 mg L−1. Chlorophyll a concentrations were measured using a spectrophotometer within 12 to 24 h after cold ethanol extraction (MDL: 0.5 μg L−1) [Nusch, 1980; Sartory and Grobbelaar, 1984].

2.4. Statistical Analyses

[22] Differences in water chemistry between beaver pond inlets and outlets were assessed with canonical redundancy analysis (RDA) using coded pairing variables (P. Legendre, Université de Montréal, Two-way canonical analysis of variance for paired observations, unpublished work, March 2004, available at http://biol09.biol.umontreal.ca/BIO6077/Appendix_(Deer).pdf). RDA tests of significance were made by permutation. We used nonparametric analysis of variance (Kruskal-Wallis) to assess water chemistry differences among the three beaver pond inlets and outlets over the eight sampling events. Relationships between water chemistry variables in inlets and outlets of beaver ponds were examined with Spearman correlation matrices. For values under method detection limit (MDL), one-half MDL was used in statistical analyses. All the statistical analyses were performed using the R language (Version R-2.3.1). We accepted statistical significance at p < 0.05.

3. Results

3.1. Inlet-Outlet Differences in Total Methylmercury and Total Mercury

[23] Increases occurring within beaver ponds are reported as outlet-inlet concentration ratios for each sampling event. TMeHg ratios showed a season pattern with maximum values when ponds were ice-covered in late winter (up to 18-fold for the recent) and during midsummer (up to 27-fold for the intermediate) while minimum values occurred during the snowmelt period (Figure 3). The highest values in TMeHg outlet-inlet ratios coincided with periods of high oxygen depletion (on average <1 mg L−1) in bottom waters of beaver ponds (Figure 3). For all seasons, the old beaver pond had the lowest TMeHg concentration ratios. THg concentrations ratios did not follow the pattern observed for TMeHg (Figure 3). Contrary to the two other ponds, the old impoundment showed higher THg concentrations in inlet than in outlet.

Figure 3.

Time series of outlet/inlet concentration ratios of (top) total methylmercury (TMeHg) and (bottom) total mercury (THg) of three beaver impoundments (solid circles, recent; open circles, intermediate; solid triangles, old) from March to September 2007. Dashed vertical bars and inner box represent two periods (period A and period B) with dissolved oxygen concentrations near or less than 1 mg L−1 in the three beaver pond bottoms. Short-dashed lines represent the neutral ratio 1:1 between outlet and inlet concentrations. Error bars represent standard deviation of the division of outlet concentrations on inlet concentrations (error propagation).

[24] Methylation efficiency was estimated as percent of THg as TMeHg (%TMeHg/THg). Methylation efficiency values were less than 20% in the three inlets throughout the sampling period (Figure 4). In contrast, high %TMeHg/THg was observed in the three outlets during the ice-covered period in late winter and during summertime (Figure 4). The intermediate beaver pond had the highest methylation efficiency in late winter (80%) while the recent impoundment had the greatest summer value (60%) (Figure 4).

Figure 4.

Time series of concentration of total methylmercury (TMeHg), total dissolved methylmercury TDMeHg, %TDMeHg/TMeHg and %TMeHg/THg for three (recent, intermediate, and old) beaver pond inlets (open circles) and outlets (solid circles) sampled from March to September 2007. All concentrations are in ng Hg L−1. Error bars represent standard error (n = 3).

[25] Concentrations of THg and TMeHg in the three inlets did not follow a seasonal pattern and were not significantly different among the three sites (Table 1 and Figures 4 and 5) . However, TMeHg concentrations showed an obvious seasonal dynamic in all outlets starting with a decline from the end of winter to spring melt, followed by an increase during the summer and ending with a slight decrease starting in late July (Figure 4). TMeHg concentrations among the three outlets were statistically different over the sampling period (Table 1). The recent pond had a greater TMeHg peak (2.80 ng L−1) in summer than the intermediate and old ponds (0.87 and 0.50 ng L−1, respectively; Figure 4).

Figure 5.

Time series of concentration of total mercury (THg), total dissolved Hg (TDHg) and %TDHg/THg for three (recent, intermediate, and old) beaver pond inlets (opened circles) and outlets (solid circles) sampled from March to September 2007. All concentrations are in ng Hg L−1. Error bars represent standard error (n = 3).

Table 1. Differences in Water Chemistry Between the Inlet and the Outlet for the Three Beaver Ponds and Differences Among the Three Inlets and the Three Outletsa
VariableRecent PondIntermediate PondOld PondAmong InletsAmong Outlets
  • a

    For total mercury (THg), total dissolved mercury (TDHg), %TDHg/THg, total inorganic mercury (TIHg), total dissolved inorganic mercury (TDIHg), total methylmercury (TMeHg), total dissolved methylmercury (TDMeHg), %TDMeHg/TMeHg, %TMeHg/THg, dissolved organic carbon (DOC), sulfate (SO42−), nitrate (NO3), chloride (Cl), dissolved oxygen (O2), pH, and specific conductivity. I, inlet; O, outlet; NS, nonsignificant; *p < 0.05; **p < 0.01; n = 8.

THgI < O**I < O**I > O**NSp = 0.0012
TDHgI < O**I < O**NSp = 0.0149p = 0.0011
%TDHg/THgNSI < O**I < O*p = 0.0060NS
TIHgI < O**NSI > O**NSp = 0.0042
TDIHgI < O*I < O**I > O*p = 0.0116p = 0.0040
TMeHgI < O**I < O**I < O**NSp = 0.0147
TDMeHgI < O**I < O**I < O**NSp = 0.0143
%TDMeHg/TMeHgI < O*NSNSNSNS
%TMeHg/THgI < O**I < O**I < O**p = 0.0385NS
DOCI < O**I < O**I < O**p = 0.0004NS
SO42−I > O**I > O**I > O**p = 0.0007p = 0.0005
NO3NSI > O**I > O**NSNS
ClNSNSI > O**p < 0.0001p = 0.0001
O2I > O**I > O**I > O**NSNS
pHI > O**I > O*NSp = 0.0289NS
ConductivityNSI > O**NSp = 0.0113p = 0.0002

[26] THg levels peaked in the recent (4.96 ng L−1) and in the intermediate (2.56 ng L−1) pond outlets in late June (Figure 5). Exceptionally high concentrations of both TMeHg and THg occurred in early April at the intermediate inlet (TMeHg: 0.41 ng L−1; THg: 3.31 ng L−1) and outlet (TMeHg: 2.08 ng L−1; THg: 3.96 ng L−1). THg concentrations in the old beaver pond outlet showed relatively small variation over the seasons (THg range: 0.50–1.22 ng L−1) (Figure 5).

3.2. Inlet-Outlet Differences in Total Dissolved Methylmercury and Total Dissolved Mercury

[27] Total dissolved MeHg (TDMeHg) and total dissolved mercury (TDHg) in inlets and in outlets followed nearly the same seasonal patterns as TMeHg and THg (Figures 4 and 5). In outlets, TDMeHg accounted on average for more than 80% of TMeHg over the sampling period (recent: 74 ± 13%; inter: 84 ± 14%; old: 84 ± 16%; mean ± standard deviation). Because TDMeHg was a large part of TMeHg also in inlets (recent: 52 ± 29%; inter: 58 ± 34%; old: 78 ± 21%; mean ± standard deviation), the proportion of TMeHg as TDMeHg (%TDMeHg/TMeHg) only increased significantly in the recent pond outlet (Table 1).

[28] The proportion of THg as TDHg (%TDHg/THg) increased significantly from the inlet to outlet for the intermediate and the old impoundments, but not for the recent pond (Table 1 and Figure 5). Average %TDHg/THg per beaver pond ranged from 30 to 60% in inlets and from 60 to 70% in outlets over the sampling period.

3.3. Inlet-Outlet Differences in Ancillary Variables

[29] Dissolved organic carbon (DOC) concentrations did not show a similar seasonal pattern to that of TMeHg or THg; DOC concentrations increased considerably through the sampling period in outlets (Figure 6) and were always greater in outlets than in inlets (Table 1). DOC concentrations were between threefold and eightfold higher in the outlet than in the inlet of the intermediate beaver pond (outlet/inlet concentration ratios). The recent and the old beaver impoundments had lower increase factors ranging from 1.1 to 2.5-fold. Considering all inlets and outlets, DOC was only correlated to THg in the recent beaver pond outlet (rs = 0.76; p < 0.05; n = 8). Dissolved oxygen concentrations were strongly and negatively correlated to DOC concentrations in the three outlets (same correlation coefficients for the three outlets; rs = −0.88; p < 0.01; n = 8).

Figure 6.

Time series of concentration of dissolved organic carbon (DOC) in mg C L−1, dissolved oxygen in mg L−1 and pH for three (recent, intermediate, and old) beaver pond inlets (open circles) and outlets (solid circles) sampled from March to September 2007. Error bars represent standard error (n = 3).

[30] Compared to the three inlets, dissolved oxygen concentrations were significantly lower in outlets (Table 1 and Figure 6). The recent pond outlet showed negative correlations between dissolved oxygen concentrations and THg (rs = −0.90; p < 0.001; n = 8), TDHg (rs = −0.98; p < 0.0001; n = 8), TMeHg (rs = −0.69; p < 0.05; n = 8), TDMeHg (rs = −0.69; p < 0.05; n = 8), and TDIHg (rs = −0.95; p < 0.001; n = 8). pH levels were lower in outlets of the recent and the intermediate pond although there was no difference for the old pond (Table 1 and Figure 6).

[31] Sulfate (SO42−) concentrations in outlets were at all times lower than in inlets (Table 1 and Figure 7). SO42− concentrations remained constant in all inlets over the seasons but decreased considerably from March to September in outlets (Figure 7). Dissolved oxygen concentrations were positively correlated to SO42− concentrations in outlets of the intermediate and the old impoundments (intermediate: rs = 0.76; old: rs = 0.69; p < 0.05; n = 8).

Figure 7.

Time series of concentration of sulfate (SO42−) in mg SO42− L−1, nitrate (NO3) in mg NO3 L−1, chloride (Cl) in mg Cl L−1 and conductivity in μS cm−1 for three (recent, intermediate, and old) beaver pond inlets (open circles) and outlets (solid circles) sampled from March to September 2007. Values are missing for conductivity in early March. Error bars represent standard error (n = 3).

[32] Nitrate (NO3) concentrations were significantly lower in outlets than in inlets for the intermediate and the old impoundments (Table 1 and Figure 7). NO3 concentrations were below the method detection limit for the old beaver pond outlet from late May to late September.

[33] Concentrations of chloride (Cl) were lower in streams of the recent and the intermediate ponds than for the old one (Figure 7). Concentrations were not significantly different between inlets and outlets for the recent and intermediate beaver ponds; however the old impoundment showed higher Cl concentrations in the inlet than in the outlet (Table 1).

[34] Conductivity showed no clear dynamics over the seasons. The recent pond had higher conductivities values (∼100 μS cm−1) than in the two others (∼40–60 μS cm−1) (Figure 7).

[35] Average chlorophyll a concentrations in ponds from late June to late September were 37.1 ± 12.7, 14.8 ± 7.9, 2.1 ± 0.6 μg L−1 (mean ± standard deviation; n = 4) for the recent, the intermediate and the old beaver pond respectively. Chlorophyll a concentrations were stable throughout the summer in the old beaver pond, but peaked in late August at 53.0 ± 3.9 μg L−1 (mean ± standard error; n = 3) in the recent and in late September at 24.4 ± 1.2 μg L−1 (mean ± standard error; n = 3) in the intermediate pond.

4. Discussion

4.1. High Methylmercury Concentrations in Water Draining Beaver Impoundments

[36] The high TMeHg concentrations (up to 2.8 ng L−1) measured in the recent pond during midsummer base flow conditions were comparable to the highest values reported for 474 streams in northeastern North America [Shanley et al., 2005] and to those reported in experimental reservoirs in Ontario (Canada) [Hall et al., 2005] with TMeHg concentrations up to 3.2 ng L−1. At the opposite, TMeHg concentrations in the old impoundment were comparable to a natural basin wetland outflow in Ontario (means TMeHg concentrations range over 3 years: 0.456 to 0.727 ng L−1) [St. Louis et al., 1996].

[37] The percentage of THg that is MeHg (%MeHg/THg) is a good relative indicator of MeHg production rates in ecosystems [St. Louis et al., 2004]. Methylation efficiencies in our three outflows were high during both late winter (%TMeHg/THg range: 53–80%) and summer (%TMeHg/THg range: 34–67%). Levels of %TMeHg/THg are usually in the 5–15% range under low-flow conditions in streams [Balogh et al., 2004] and in water draining wetlands [St. Louis et al., 1994]. More similar to our sites, the amount of THg present as TMeHg peaked at 45% in the FLUDEX upland reservoir outflow midway through the second flooding season, and at 95% in water exiting the ELARP wetland reservoir during the first year of flooding (ON, Canada) [Hall et al., 2005]. Elevated %TMeHg/THg from ELARP wetland reservoir during the summer period persisted 9 years after flooding, although the amplitude of the peak became somewhat muted as time passed [St. Louis et al., 2004]. However, 10 or 20 years after flooding, our aged beaver ponds maintained high methylation efficiencies in outflows, particularly before snowmelt and during summer. This long-term generation of TMeHg in beaver dams compared to other flooded wetlands may be partly the result of the large woody biomass transferred by beavers in ponds in the course of their food gathering and building activities (i.e., dam and hut). The slow bacterial decomposition of this biomass may fuel mercury methylation over many years.

4.2. Effect of Season and Pond Age on Mercury Concentrations

[38] We observed higher TMeHg outlet/inlet ratio (up to 27–fold) and greater methylation efficiencies in outflows of ice-covered ponds and during midsummer than at any other sampling collection. Low dissolved oxygen concentrations measured in bottom waters of ponds at those dates may indicate that TMeHg was produced and discharged from areas where anoxic/anaerobic conditions prevailed. Mercury methylation has been shown to be predominant in bottom water and in surface sediments at the oxic/anoxic interface [Ullrich et al., 2001]. Most annual studies on beaver impoundments have reported water chemistry variations mainly during summer [Smith et al., 1991; Cirmo and Driscoll, 1993; Driscoll et al., 1998; Margolis et al., 2001] and few observed noticeable effects during the ice-covered period on redox-sensitive species [Devito and Dillon, 1993]. Because our study include only one sampling collection during late winter, further research should be done to evaluate the whole wintertime effect on oxygen and TMeHg concentrations in beaver ponds.

[39] TMeHg outlet/inlet ratios were lowest (near 1, the neutral ratio) during high-flow conditions when concentrations were diluted, and highest during base flow conditions. Consequently, large amount of mercury may have been flushed from beaver ponds during snowmelt, when approximately two thirds of the discharge over the sampling period occurred. Further, the observation of high TMeHg outlet/inlet ratios was favored under summer base flow conditions, as a result of prolonged water residence time in the impoundments. Many studies have shown that discharge conditions may drive mercury flux patterns in streams with no beaver dams [Balogh et al., 2004; Shanley et al., 2005; Balogh et al., 2006]; here our discrete sampling collections are incompatible with a complete flux measurement analysis. However, this study shows that low oxygen concentrations and warm summer temperature seem to be key factors enhancing mercury concentrations in beaver pond outlets even though base flow conditions may also play a role.

[40] The lower pH, dissolved oxygen, NO3, SO42−, and higher DOC levels measured in outlets compared to inlets throughout the sampling period indicate that biological activity was high in all ponds. Because SO42− concentrations were always lower in outlets, we hypothesize that sulfate-reducing bacteria (SRB) could be responsible of the increased TMeHg levels in the three beaver impoundment outlets. SRB may have been more active and abundant in the recent pond where the primary productivity and TMeHg concentrations were highest [Jin et al., 1996; Bank et al., 2007]. Enhanced decomposition rates promoted by the flooding of fresh and labile organic matter in the recent pond presumably contributed to higher TMeHg production in this system than in the two others. The shadowing effect of the high density of standing dead trees and the water stained by dissolved organic matter and by the elevated primary productivity may have also reduced photodegradation of TMeHg in the recent beaver impoundment [Sellers et al., 1996]. However, future research concerning the complex and paired interactions between flooding time and TMeHg degradation is required to further evaluate these hypotheses.

[41] The old impoundment of this study displayed the same THg and TMeHg patterns in downstream waters as those observed by Driscoll et al. [1998] in an old beaver pond. Both mature ponds showed very similar maximum TMeHg levels in summer, at 0.4 ng L−1 in the Adirondack region and at 0.5 ng L−1 in the Laurentian region. However, our old site exhibited higher methylation efficiencies both before snowmelt (60%) and in midsummer (40%) than did the Adirondack beaver pond (winter: 6–10%; summer: 16%). This gap between these two old beaver ponds of northeastern North America reveal that pond age is not the only factor controlling MeHg production. Differences in sediment organic matter content, microbial activity, hydrologic condition and atmospheric mercury deposition could contribute to MeHg production variability in these aquatic systems.

4.3. Beaver Pond Influence on Dissolved Mercury

[42] Beaver ponds contributed generally to increase the proportion of TDHg on THg and TDMeHg on TMeHg in outlets. Higher standard deviation values for average %TDMeHg/TMeHg in inlets than in outlets reflects the influence of beaver ponds in buffering the variability of filtered/unfiltered MeHg partition over seasons. Given the possibly low particulate loading in inlets under base flow conditions, it is not surprising that the dissolved fraction accounted also for the majority of the TMeHg in inlets.

[43] The high TDMeHg concentrations observed in outflows (up to 2.06 ng L−1) suggest an increase in the potential for bioavailable mercury to be transported to and contaminate downstream aquatic systems. No data on TDMeHg in beaver pond outflow exist, but Paterson et al. [1998] measured comparable TDMeHg concentrations (up to ∼1.9 ng L−1) in water of a flooded wetland reservoir (ELARP). The fate and bioavailability of the high TDMeHg concentrations occurring in beaver pond outlets should be further examined since primary producers and primary consumers are critical intermediaries in the movement and biomagnification of mercury from water to upper trophic levels [Hill et al., 1996].

4.4. Mercury and DOC Dynamics

[44] The lack of correlation between mercury (THg and TMeHg) and DOC concentrations in this study may suggest that mercury dynamic in some cases may be more related to reducing conditions than to DOC pattern. Even though DOC production increased through the summer, TMeHg production peaked during periods of oxygen depletion in ponds. This observation is inconsistent with the findings of Driscoll et al. [1998] where TMeHg concentrations coincided with increases in DOC concentrations at the same time in inlet and outlet of a beaver pond. Note that DOC concentrations in the three beaver pond outlets of this study (DOC range: 3–9 mg L−1) were similar to those reported by Driscoll et al. [1998] (DOC range: 4–9 mg L−1) and Margolis et al. [2001] (DOC range: 1–11 mg L−1) in beaver impoundments outflows. It is unclear why mercury and DOC were not related, but the quality and source of DOC and complexation with particulate organic carbon (POC) should be further investigated as explanatory variables.

5. Conclusion

[45] Peaks of MeHg concentrations in water draining three beaver impoundments of varying age occurred during midsummer, but also at the end of winter when the ponds were ice-covered, which was unexpected. Depletion of dissolved oxygen concentrations in beaver pond beds at those periods likely promoted MeHg production, although the highest TMeHg concentrations coincided with increases in temperature during summer and presumable increases in microbial activity.

[46] Age of flooding was a key determinant, with recent and intermediate ponds being particularly important MeHg sources and sites of Hg transformations. Mercury patterns in recent beaver ponds are likely similar to those observed in the first years of flooded wetland and forest, while mercury patterns in older impoundments are closer through time to natural wetlands.

[47] By promoting the release of large concentrations of MeHg in the dissolved form, these natural reservoirs may favor food web contamination and downstream transport on long distances. Mercury content of aquatic organisms living and growing in beaver ponds and in downstream waters may show variable mercury contents as a function of flooding time and season.

[48] Watershed managers of northeastern United States and southeastern Canada regions, where the occurrence and effects of beavers is pervasive and increasing, should integrate landscape modifications made by beavers in their strategies to reduce mercury supply in aquatic systems. Understanding the immediate and long-term effects of beaver dam construction on water quality, and specifically, on mercury contamination of surface waters may be critical for accurately predicting the ultimate effectiveness of reductions in mercury deposition and other land management decisions. This study lays the groundwork for future research to explore the variability of net MeHg flux among several beaver ponds of contrasting age and over all seasons (fall and winter included) to compare them to other wetland type.

Acknowledgments

[49] This research was supported by NSERC postgraduate scholarship to V.R. and NSERC Discovery grants to M.A. and R.C. We thank Valérie Girard, Ariane Cantin and Simon Legault for support in the field and Dominic Bélanger, Marie-Claude Turmel and Kwélé Bonge-Boma for support in the lab. The comments of two anonymous reviewers greatly improved the manuscript.