Hydraulic and Biochemical Gradients Limit Wetland Mercury Supply to an Adirondack Stream

Sixmile Brook (within Fishing Brook watershed; Adirondack region, USA) and a companion study area, McTier Creek (Atlantic Coastal Plain region, USA) [20] (Figure 1)exhibited elevated MeHg bioaccumulation [21] and downstream increases in fluvial MeHg concentrations and yields [7] (Figure 2). In contrast to the limited temporal variability in MeHg concentrations (or yields)(Figure 2B; [7] ) and persistent wetland-to-stream MeHg gradients observed in McTier Creek[discussed in detail in 13],Sixmile Brook demonstrated substantial variability in fluvial MeHg concentrations (and fluxes)and in wetland-stream MeHg gradients (Figure 2A; [7] ). To better understand the biogeochemical and hydrological processes that contribute to fluvial MeHg temporal variability in Sixmile Brook and in other Adirondack streams, two approaches were employed: 1) multivariate regression modeling of the encompassing Fishing Brook watershed[presented in 19] and 2) instantaneous, instream mass flux assessments in a 500 m study reach. Regression analysis of 43 fixed station samples collected in Fishing Brook downstream of Sixmile Brook during 2007-2009 suggested that seasonal temperature-driven effects on wetland Hg methylation and stream flow variations during the cold dormant season are principal controls on fluvial MeHg concentration patterns [19]. Because the lowest fluvial MeHg concentrations were observed under snowmelt, high-flow conditions, instantaneous wetland-stream hydraulic and biogeochemical gradients were hypothesized to be fundamental controls on MeHg supply in Sixmile Brook. Herein, within-reach changes in the fluxes of water, MeHg, THg, DOC, and chloride as well as lateral (groundwater to surface-water) and vertical (hyporheic-porewater)hydraulic Abstract


Introduction
Identifying methyl mercury (MeHg) source areas and the controls on MeHg supply to streams is an environmental priority due to widely-reported correlations between elevated fluvial-MeHg concentrations and fish-Hg contamination [1][2][3] , the leading cause of fish consumption advisories in streams throughout the United States (US) [4] and Canada [5].Strong correlations between wetland area and fluvial MeHg concentrations [6][7][8] , higher dissolved MeHg concentrations [7,[9][10][11] and Hg methylation rates [7,12] in wetlands compared with adjacent stream habitats, and strong wetland-stream hydrologic and concentration gradients were assessed in Sixmile Brook during snowmelt and later during the growing season, to test the hypothesis that unfavorable hydraulic and chemical gradients (i.e., insignificant or from surface-water toward groundwater) limit fluvial MeHg supply in this and in similar systems in the Adirondacks region.

Hydraulic and Biochemical Gradients Limit Wetland Mercury Supply to an Adirondack Stream
Copyright: © 2016 Bradley et al.

Piezometer Network
Out-of-channel (groundwater) piezometers (2.5 cm diameter PVC) were installed on both banks with screened intervals 30-60 cm below land surface (Figure 1; Figures S1 and S2 in File S1).Only data from 20 piezometers closest (3-10 m) to the stream channel are discussed in this paper.In-streamhyporheic piezometers (4 each at top and bottom of reach; 1 cm diameters; 30 cm screened intervals) were emplaced at the edge of water on each bank and equidistant across the channel with the top of screen set at 20 cm below the bed sediment/surface-water interface.Fine scale (1.5 cm minimum resolution) vertical solute gradients were assessed at edge-of-water (margin) and center-of-channel (channel)at the downstream reach boundary using two mini -Piezometer (USGS MINIPOINT) devices (seven 0.325 cm stainless steel tubes, 0.8 cm screened interval) [28,29].Each MINIPOINT device was installed with sampling points at approximately 1.5, 4.5, 7.5, 12.5, 17.5, 30, and 60 cm below the bed-sediment/surface water interface.

Sampling Event Hydrologic Setting
Lagrangian (following same average parcel of water downstream through the study reach) surface-water flux assessments (described in detail below) were initiatedin2009 during snowmelt in May and during the growing season in August.Lateral solute concentration gradients were assessed by sampling surface water (0-3h from start of sampling), followed immediately by in-channel pore water (3-8 h) and then out-of-channel groundwater (6-34 h).MINIPOINT sampling was conducted during the May assessment.

Stream flow Data Collection
Instantaneous stream discharge was measured during surface water sampling at a location immediately downstream of the study reach using acoustic Doppler velocimetry (ADV) [22].Surface-water samples were collected at the mid-time of three consecutive discharge measurements (1 h each).Subsequently, paired Lagrangian discharge measurements were made in triplicate (one h intervals), with upstream and downstream measurements offset by the average downstream transport time (25 and 40 min) through the reach.Surface-water transport time was calculated using the average stream flow velocity measured immediately downstream of the reach before sample collection.

Water-Quality Data
For each assessment, triplicate surface-water samples were collected at upstream and downstream boundaries using a Lagrangian sampling approach [13].Surface-water was collected at one hour intervals at the upstream margin without entering the stream, using a pole to position sample bottles at mid depth in the center of flow until full (Figure S2 in File S1).Corresponding downstream samples were collected following a delay (25 and 40 min) equivalent to the average surface-water transport time through the reach.Ultra-trace-level clean-sampling, processing, and Hg analysis procedures were as in [6,9,30,31].Filtered (0.7 µm quartz-fiber filter) samples were analyzed for FMeHg by gas chromatographic separation with cold vapor atomic fluorescence spectrometry [32] and for FTHg by oxidation, purge and trap, and cold vapor atomic fluorescence spectrometry [33] at the USGS Mercury Research Laboratory (Middleton, Wisconsin).The reporting limit for FMeHg and FTHg was 0.04 ng/L.Concentrations of DOC [34]and chloride [35]were determined as cited.
Vertical hydraulic gradients in the stream channel at the time of sample collection were estimated as the potentiometric difference between hyporheic pore water (water level inside in-channel piezometers) and surface water (water level outside the piezometers) using an electric water level tape.Porewater samples were collected from in-channel piezometers using a peristaltic pump (flow less than 50 mL min -1 ) and low volume (200 mL purge; 1 L total pumped volume) sampling techniques [13].An assessment of water quality parameters (pH, conductivity, and temperature) over time was conducted to verify stream-water entrainment did not occur under these pumping conditions [13].Out-of-channel groundwater wells were purged (3 well volumes) on day 1, immediately after water-elevation measurements.Groundwater was sampled and processed as described for in-channel porewater samples.Ultra-low-flow (approximately 2 mL min -1 ) MINIPOINT sampling was conducted [as described in detail elsewhere 28] to collect porewater from the shallow hyporheic-flow and underlying groundwaterhyporheic porewater mixing zones using a peristaltic pump.All samples were filtered and preserved with 6.5 N trace-metal grade HCl in the field in a temporary glove box [13](Figure S3 in File S1), stored on ice, and shipped overnight to the lab.Fresh gloves, glove-box bags, and pre-cleaned filter assemblies were used for each sample.Peristaltic tubing was acid cleaned (ultrapure 5% HCl) and rinsed (distilled water) between samples [6,13,31].Additional sampling details provided in File S1.

Statistical Analyses and Interpretive Framework
Statistical significance metrics and percentage reach flux increases provided direct, quantitative measures of the importance of reach wetlands as sources of water and solutes to the aquatic habitat during each 3 h surface-water flux assessment.Reach discharge results were then combined with hyporheic zone hydraulic gradient data to assess the relative importance of advective and diffusive wetland-stream solute exchange.Solute concentration gradients (e.g., groundwater to surface water, in-channel porewater to surface water) clarified the effect of groundwater discharge (dilution or concentration of solutes) on stream solute concentrations and provided insight into the direction and magnitude of diffusive exchange in the event that groundwater discharge was insignificant.
The One-tailed paired t-test was employed to assess flux differences between the top and bottom of the study reach.Kruskal-Wallis ANOVA on Ranks was combined with Dunn's Multiple Comparison Method to identify significant differences between four lateral gradient groups: groundwater, porewater at the channel margin, porewater at the center of channel, and stream surface water.A significance level (α) of 0.05 was employed for all comparisons.

Results and Discussion
The demonstrated importance of wetland areas as sources of MeHg and THg to the stream and the substantial temporal variability in MeHg supply to the stream during 2007-2008 (Figure 2; [7] )indicated that Sixmile Brook was a suitable location to evaluate wetland-stream hydraulic and chemical gradients as controls on short-term wetland-stream solute transport under different hydrologic and geochemical conditions.These short (25 and 40min per Lagrangian sample pair, approximately 3 h total) reach synoptics were intended to rigorously assess "instantaneous" gradients and wetland-stream fluxes and were not intended to represent longer-term wetland-stream dynamics[assessed previously in 7].

Change in Sixmile Brook Reach Fluxes
During May 2009 snowmelt event, a significant (p<0.005)increase in stream discharge of 7.1±1.3% was observed within the Sixmile Brook study reach (Table 1).Because no tributaries existed within the reach, increases in discharge between upstream and downstream measurements were attributed to groundwater fluxes.Comparable (8.0±1.3% and 9.0±1.3%)significant (p<0.005)increases in the fluxes of DOC and dissolved chloride (respectively) were also observed within the reach, demonstrating efficient hydrologic transport of dissolved constituents from wetland sediments to the stream habitat.However, no statistically significant (p>0.16) changes in the fluxes of FMeHg or FTHg were observed, demonstrating that discharging groundwater was not a significant source of Hg within the study reach during this short surface-water flux assessment.Because significant increases in fluxes of water, DOC, and chloride within the 25 min downstream travel time indicated good wetland-stream hydrologic connectivity and a favorable hydraulic gradient, the lack of significant net fluxes of FMeHg and FTHg was hypothesized to be due to dilute Hg concentrations in groundwater discharge within the study reach during the May assessment.
During the August 2009 growing-season assessment, no statistically significant changes in discharge or in the fluxes of FMeHg, FTHg, DOC or chloride were observed in the Sixmile Brook study reach, indicating that the adjacent riparian wetlands were not a significant source to the reach aquatic habitat during the 40 min downstream travel time(Table 1

Hydraulic and Biochemical Gradients Limit Wetland Mercury Supply to an Adirondack Stream
Copyright: © 2016 Bradley et al.
sampling event, however, the apparent lack of net exchange of Hg and other solutes was attributed to insignificant advective transport between shallow groundwater and the stream.Because the results of the May flux assessment demonstrated good wetland-stream hydrologic connectivity within the reach, the lack of net groundwater discharge during the August event was hypothesized to be due to insignificant wetland-stream hydraulic gradient.In the absence of substantial net hydrologic transport to the stream, solute exchange would be limited to hyporheic zone mixing and diffusive processes at the sediment/surface-water interface, with minimal contribution to reach fluvial chemistry during the approximate 40 minute surface-water travel time between upstream and downstream reach boundaries.The lack of significant change in DOC and dissolved chloride in the absence of an increase in discharge corroborates the previous conclusion that DOC and chloride flux increases observed in May 2009 were attributable to groundwater discharge within the Sixmile reach.
The Sixmile study reach represented less than 4% of the total upstream basin area [20].The influx of upstream surface-water was the primary (generally greater than 90%) source of water and solutes (FMeHg, FTHg, DOC, and chloride) to the study reach for both flux assessment events.The mean percentage increases (7-9%) in fluxes of surface water, DOC, and chloride observed in the study reach in May 2009, however, were approximately two times greater than the corresponding increase in basin area, indicating that the riparian wetlands within the reach were important sources of water and solutes to the stream habitat.In August 2009, the flux of water and solutes at the downstream margin of the study reach was entirely attributable to surfacewater influx at the upstream margin, demonstrating that contributions from discrete wetland areas vary substantially over time.There was no evidence of significant supply of FMeHg or FTHg from the adjacent wetlands to the aquatic habitat of the Sixmile Brook study reach (Table 1) during either flux assessment, despite the spatial extent of the riparian wetlands bordering the reach (Figure 1) and the potential for solute exchange (DOC and chloride) demonstrated during May (Table 1, Figure 3A-3D).The elevated concentrations of MeHg observed in several wetland piezometers (compared with hyporheic piezometers at channel margins, see Figures 3A and [7] ) and the observed fluvial transport of MeHg from wetland-dominated reaches upstream in Sixmile Brook are consistent with wetland areas as primary sources of MeHg to the stream throughout the growing season (Figures 2A and 3A, [7] ).The current results verify the substantial temporal variability in wetland-stream MeHg supply reported previously (Figure 2) [7] and indicate the importance of wetland-stream hydraulic and chemical gradients as important contributors to temporal variability.
The Sixmile Brook flux results are in marked contrast to the consistent, significant (p < 0.0001) increases (generally 10 % or more) in both water and dissolved constituents (FMeHg, FTHg, DOC, chloride) previously reported in sister study area at McTier Creek during two 2009 sampling events(Table1; [13]).To better understand the comparative lack of net FMeHg and FTHg exchange between the adjacent riparian wetlands and the aquatic habitat within the Sixmile Brook study reach during the 2009, surface-water flux assessments, geochemical gradients between the wetland and the stream channel, as well as stream channel vertical hydraulic gradients were assessed.

Groundwater to Surface Water Gradients
Differences in solute concentrations between surface water samples and samples of groundwater collected from wetland piezometers (located 3-10 m from the stream bank) were evaluated to assess the potential importance of shallow wetland groundwater as a longer-term source of solutes to the stream.Results from piezometers installed across the stream channel at the upstream and downstream reach boundaries were used to assess the potential importance of groundwater discharge as a proximal source of solutes to the stream at the time of sample collection.Piezometers emplaced at the channel margins provided insight into the importance of shallow wetland groundwater discharge to the stream, while piezometers placed near the center of the channel reflected the contribution of comparatively deeper groundwater [13].
During May 2009, net positive (toward surface water compartment) vertical hydraulic gradients ranging from 0.025-0.05(0.5-1.0 cm water level change over 20 cm distance; data not shown) were observed in piezometers at the channel margins of Sixmile Brook, consistent with ongoing discharge of shallow wetland groundwater to the stream and the observed significant increase in surface-water flux within the reach.No vertical hydraulic gradients were detectable (potential difference less than 2 mm) in piezometers located in the center of the channel.Concentrations of DOC and chloride in channel margin piezometers were comparable to those observed in surface water, consistent with the significant increases in DOC and chloride fluxes within the study reach (Figure 3C &3D).In contrast, FMeHg concentrations (Figure 3A) in channel margin piezometers were significantly (t-test; p < 0.0001) less than in surface water (mean concentration approximately 15% of mean surface water concentration), consistent with the lack of detectable change in FMeHg flux within the reach (Table 1).Similarly, FTHg concentrations (Figure 3B) in channel margin piezometers were significantly (t-test; p = 0.003) less than in surface water (mean channel margin concentration approximately half of mean surface water concentration), consistent with the lack of significant change in FTHg flux within the reach during the May event (Table 1).It is noteworthy that surface-water FMeHg and FTHg concentrations were within the range (not statistically different; Kruskal-Wallis One-Way ANOVA on ranks; p > 0.05) observed in groundwater piezometers located 3-10 m from the stream channel (Figure 3A &3B), indicating that shallow groundwater was an important source of FMeHg and FTHg to the stream over longer time periods.However, the low concentrations of FMeHg and FTHg observed in channel margin piezometers indicated that dilute-Hg groundwater was discharging to the stream reach during the May stream-water flux assessment.
No vertical hydraulic gradients were detected in stream piezometers (margin or center of channel) during August 2009 (data not shown), suggesting minimal groundwater discharge and consistent with the lack of significant change in the fluxes of water, FMeHg, FTHg, DOC, or chloride within the Sixmile Brook reach.No significant differences in DOC concentrations were observed between groundwater, hyporheic porewater, and surface water (Figure 3G), indicating the potential for groundwater as a substantial source of DOC to the stream under favorable hydrologic conditions.Combined with the lack of significant change in discharge within the reach, this result demonstrates that the lack of increased DOC flux within the study reach was due to little or no hydraulic gradient and a corresponding lack of advective hydrologic transport from wetlands to the stream.In contrast, concentrations of FMeHg, FTHg, and chloride were statistically significantly higher in surface water than in groundwater or porewater.The lack of significant changes in fluxes within the reach, the lack of detectable vertical hydraulic gradients, and lower concentrations in groundwater than surface water indicated that the extensive riparian wetlands within the reach were not a significant source of Hg to the stream reach during the short August 2009 streamwater flux assessment period.
In contrast to Sixmile Brook, concentrations of FMeHg, FTHg, DOC, and chloride in groundwater and in-channel pore water piezometer samples were comparable to or greater than observed in surface-water samples at McTier Creek (Figure 4A-4D), consistent with groundwater as a substantial source of these solutes to the stream habitat under the favorable hydrologic conditions observed in this study.Mass balance calculations and water-stable-isotope mixing results indicated that 80% or more of the observed increases in discharge and solute fluxes within the McTier Creek study reach were attributable to groundwater discharge [13].

Hyporheic Vertical Profiles
MINIPOINT vertical concentration profiles obtained from the margin and center of the Sixmile Brook stream channel

Hydraulic and Biochemical Gradients Limit Wetland Mercury Supply to an Adirondack Stream
Copyright: © 2016 Bradley et al.
groundwater with dilute Hg and the lack of significant change in FMeHg and FTHg fluxes within the study reach during the May event.The fact that concentration profiles from the margin and center of the channel were comparable for the conservative chloride ion (Figure 5D) as well as for FTHg and DOC (Figure 4A-C) is also consistent with surface water as the primary control on these shallow, hyporheic-porewater concentrations during the May flux assessment.However, concentrations of FMeHg in the hyporheic (0-20 cm) zone were generally higher at the channel margin than at the channel center (Figure 5A), consistent with previous conclusions that wetlands are longer-term sources of FMeHg to Sixmile Brook [7].during May provided additional insight into the supply of solutes to the stream reach (Figure 5A-D).Because MINIPOINT assessments were conducted once and thus lacked estimates of sample variability, these concentration profiles were used only to 1)assess centimeter-scale trends in vertical hyporheic concentration gradients, 2) identify the depth of the surface-water mixing (hyporheic flow) zone, and 3) verify relative differences in chemical signatures in shallow groundwater samples collected at the channel margin and comparatively deeper groundwater samples collected from the channel center.
Chloride profiles indicated a well-mixed hyporheic-flow porewater layer (indistinguishable from surface water to a depth of approximately 20 cm below the sediment/surface water interface) and decreasing concentrations with increasing depth from 20-60 cm (Figure 5D).FMeHg, FTHg, and DOC concentrations also generally decreased with increasing depth, consistent with the previous conclusion that the primary source of these solutes to the stream reach was upstream surface water.The fact that channel margin, porewater FMeHg and FTHg concentrations below the surface mixed (hyporheic flow) layer were approximately an order of magnitude less than in the overlying surface water is consistent with the discharge of deeper

Hydraulic and Biochemical Gradients Limit Wetland Mercury Supply to an Adirondack Stream
Copyright: © 2016 Bradley et al.
In contrast, only limited changes in concentrations with depth were observed in MINIPOINT samples from McTier Creek in July 2009, and solute concentrations in porewater and surfacewater samples were comparable (Figure 5E-H).These results are in agreement with the documented importance of groundwater discharge in the McTier Creek study reach [13] under nonflood conditions.FMeHg, FTHg, and DOC concentrations were consistently higher at the channel margin than at the center of channel, further supporting the conclusion that discharge of shallow groundwater from adjacent wetland/riparian floodplain areas is the primary source of these constituents to the stream.In contrast, chloride concentrations were consistently higher at the center channel location than at the channel margin and comparable to surface-water concentrations, indicating that the primary sources of chloride in the reach were surface water from upstream of the reach and discharge of deeper groundwater within the reach.The fact that surface-water concentrations of FMeHg, FTHg, and DOC were intermediate to MINIPOINT concentrations from the margin and center locations (Figure 5E-5G) is consistent with the single-screen in-channel Piezometer results (Figure 4A-4C) and indicates that surface water in the stream reach is a mixture of shallow and comparatively deeper groundwater discharge.

Implications for MeHg Supply
The results of this study verify that the extensive riparian wetlands that border Sixmile Brook are hydrologically well connected to the stream.Consequently, the results are compatible with the earlier conclusion that adjacent and upstream wetland areas are important long-term sources of water and solutes to the Sixmile Brook aquatic environment [7].These results also support the hypothesis that unfavorable instantaneous wetland-stream hydraulic and chemical gradients (direction and magnitude) limit the importance of these wetland areas as sources of Hg and other solutes to Adirondacks-region stream habitats in the short term.
Hg-risk management strategies are applied at regional to national scales, while foundational research on environmental Hg processes and dynamics is most often conducted at focused-reach to small-watershed scales.The development of mechanistic models that extend research-scale findings to Hgrisk-management scales is challenged by competing needs to capture those Hg processes which are most essential to system simulation while maintaining model usability (i.e., avoiding parameter inflation and excessive processing time).Numerical model improvements that allow more accurate prediction of temporal and spatial variations in fluvial MeHg supply are needed to better manage Hg-bioaccumulation risk in the Adirondacks and other regions of North America [19,36].The current results suggest that continued investigation of factors that affect temporal variations in wetland-stream gradients and improved near-stream hydrologic models may substantially improve Hgrisk-management models.

Figure 3 :
Figure 3: Concentrations of filtered methylmercury (A, E), filtered total mercury (B, F), dissolved organic carbon (C, G), and dissolved chloride (D, H) at Sixmile Brook in May (A-D) and August (E-H) 2009.Boxplots are groundwater (G) collected from wells 3-10 m from stream bank, porewater collected from the margin (Ch M ) and center (Ch C ) of the channel, and stream water (S)(number of samples in parentheses).Dashed red line is the FMeHg reporting limit of 0.04 ng L -1 .Censored data were set at 0.02 ng L -1 .Boxes, centerlines, whiskers, and circles (•) indicate interquartile ranges, medians, 10 th and 90 th percentiles, and data, respectively.Different letters indicate statistically different groups (p < 0.05).

Figure 4 :
Figure 4: Concentrations of filtered methylmercury (A), filtered total mercury (B), dissolved organic carbon (C), and dissolved chloride (D) at McTier Creek (April and July data did not differ significantly and are combined) in 2009.Other labeling as in Figure 3.

Table 1 :
Change (%) in mass fluxes (means±standard deviations; n=3) of water and solutesbetween upstream and downstream margins of the Sixmile Brook (NY) and McTier Creek (SC) study reaches during 2009.Negative values indicatea net decrease andp-values in bold are statistically significant (one-tailed paired t-test; p<0.05).