Mercury volatilization from salt marsh sediments

Authors


Abstract

[1] In situ volatilization fluxes of gaseous elemental mercury, Hg(0), were estimated for tidally exposed salt marsh sediments in the summer at the urban/industrial Secaucus High School Marsh, New Jersey Meadowlands (Secaucus, New Jersey) and in the early autumn at a regional background site in the Great Bay estuary (Tuckerton, New Jersey). Estimated daytime sediment-air mercury volatilization fluxes at the Secaucus High School Marsh ranged from −375 to +677 ng m−2 h−1 and were positive (land to air flux) in 16 out of 20 measurement events. At the Great Bay estuary, mercury fluxes measured continuously over a 48-h period ranged from −34 to +81 ng m−2 h−1 and were positive during the day and negative at night. At both sites, mercury volatilization fluxes peaked at midday, and cumulative mercury fluxes exhibited strong positive correlations with cumulative solar radiation (r2 = 0.97, p < 0.01) consistent with a light-driven mercury volatilization efficiency of about 15 ng Hg mol PAR−1 or about 0.06 ng Hg kJ−1. No significant correlations were found between mercury fluxes and wind speed, air temperature, or tide height at either site. Thus despite a tenfold difference in sediment mercury concentration, photochemistry appears to be the dominant factor controlling mercury volatilization from these salt marsh sediments. The average mercury volatilization flux estimated for the Great Bay salt marsh in this study (17 ng m−2 h−1) compares well with other micrometeorological mercury fluxes for nonpoint source contaminated salt marsh and forest soils (8–18 ng m−2 h−1) and is more than 10 times higher than the average mercury emission flux from land (∼1 ng m−2 h−1). Annual mercury emissions from salt marsh wetlands may be comparable to individual industrial emissions sources in coastal states of the eastern United States.

1. Introduction

[2] Land-air exchange of mercury is of global concern because of mercury's atmospheric mobility, bioaccumulation potential, and toxicity. Global anthropogenic mercury emissions were estimated at 2200 t a−1 for the year 2000 and nonindustrial emissions from land, including emissions of truly natural mercury and mercury from previous anthropogenic deposition, are estimated to range from 1000 to 1600 t a−1 [Lamborg et al., 2002; Mason and Sheu, 2002]. While truly natural land-air mercury emissions occur in regions with air-exposed, mercury-rich minerals, mercury volatilization from most land surfaces, including salt marshes, represents the evasion of both natural and anthropogenic mercury. Indeed, mercury reemission is a widely distributed phenomenon, but the volatilization of mercury from tidally air-exposed salt marsh sediments has not been quantified.

[3] Mercury in aquatic systems accumulates in sediments [Gilmour and Henry, 1991] where it is precipitated as a sulfide, associated with iron sulfides, or bound to organic matter [Mantoura et al., 1978; Drobner et al., 1990; Morse and Luther, 1999]. Coastal marine sediments are covered by a thin layer of oxic sediment [Baillie, 1986], but reduced mercury (Hg0) may be formed at the sediment surface by photochemical processes or transported there from deeper in the sediment by biological activities [Canario and Vale, 2004; Robbins and Edgington, 1975; Kramer et al., 1991]. Although mercury is efficiently retained by aquatic sediments, water-level fluctuations may lead to the resuspension and short-term exposure of sediments to solar radiation [Carpi and Lindberg, 1997; Canario and Vale, 2004; Poissant et al., 2004]. In tidal systems, ebb tide brings sediments in direct contact with the atmosphere where photochemical reactions and boundary layer meteorology may enhance mercury volatilization. Additional biological processes (e.g., microbial reduction, sediment bioturbation) in wetland ecosystems may also lead to the mobilization of sediment-bound mercury [Poissant et al., 2004].

[4] Mercury volatilization from land is dominated by the flux of elemental mercury [Kim and Lindberg, 1995] and environments favoring the reduction of Hg(II), such as highly productive wetlands, are likely to support elevated Hg0 emissions [Bothner et al., 1980; Gobeil and Cossa, 1993]. The large area of tidal wetlands (more than 2.3 × 104 km2 in the eastern United States [Stedman and Dahl, 2008]) combined with the large reservoir of mercury in wetland sediments suggests that mercury reemission from wetland sediments has the potential to rival other natural and industrial emissions on a regional scale. In addition, the reduction and volatilization of inorganic Hg in salt marsh sediments may decrease the pool of reactive mercury available for methylation [Rolfhus and Fitzgerald, 2001] thereby decreasing the accumulation of monomethylmercury in aquatic animals and their consumers including humans. However, because only a few studies have examined mercury emissions from wetlands, especially salt marsh wetlands, mercury emissions from these environments are poorly constrained.

[5] Only one previous study investigating mercury volatilization from wetland sediments has employed the micrometeorological method [Lee et al., 2000] in which in situ mercury concentration gradients, atmospheric friction velocities, and turbulence correction factors are measured in situ without an enclosing chamber. In that case, sediment Hg fluxes from the Farm River salt marsh in Connecticut were estimated using a micrometeorological eddy covariance system in conjunction with a Tekran 2537A mercury vapor analyzer, as in this study. An important difference between the present study and that of Lee et al. [2000] is the presence of abundant marsh grass (Spartina patens) at the Farm River salt marsh. Other studies have investigated Hg emission from wetland sediments using dynamic flux chamber methods which eliminate the effects of atmospheric turbulence and operate at lower than atmospheric pressure. This study will contribute to the knowledge base regarding Hg dynamics in salt marsh sediments under natural conditions.

[6] The objectives of this study were to estimate mercury volatilization fluxes from contaminated and nonpoint source impacted tidally exposed salt marsh sediments. Total gaseous mercury (TGM) fluxes were estimated on six different days at the mercury-impacted Secaucus High School Marsh in the New Jersey Meadowlands and continuously over 48 h at a regional background site in the Great Bay estuary in southeastern New Jersey using the micrometeorological technique [Lee et al., 2000; Lindberg and Meyers, 2001; Lindberg et al., 2002; Goodrow et al., 2005]. In contrast to flux chamber measurements, micrometeorological methods provide spatially averaged fluxes that include the effects of ambient air turbulence, which is an important factor driving the volatilization of gases from terrestrial surfaces.

2. Site Description

2.1. Secaucus High School Marsh, New Jersey Meadowlands, Secaucus, New Jersey

[7] The Secaucus High School Marsh, at 40.80°N, 74.04°W, is a 0.15 km2 tidally restricted estuarine wetland located in the New Jersey Meadowlands (Secaucus, Hudson County, NJ; Figure 1). Secaucus High School Marsh is situated within one of the most industrial regions of the Northeastern United States and its waters and sediments have accumulated a wide range of contaminants, including mercury. The vegetation was dominated by Phragmites australis (common reed) with small patches of Spartina sp. present along the banks of the Hackensack River during field sampling in 2005. Between 2005 and 2006 the marsh was cleared to bare sediments for wetland restoration and remained bare through final experimentation in June 2007. Salinity in the Secaucus High School Marsh ranges from 3 to 12‰.

Figure 1.

Salt marsh study sites in New Jersey, United States, and their wetlands. SC, Secaucus High School Marsh; GB, Great Bay estuary.

[8] Total gaseous mercury and micrometeorological equipment were assembled on a constructed berm along the eastern edge of the marsh and approximately 1 m above it. Sample location borders include: an open wetland fetch and the Hackensack River to the north, a large expanse of wetland and the Secaucus High School to the west, a residential housing development to the south and a recreational park to the east. As a result, only data collected from the longest wetland fetch (∼0.5 km; W–NW wind direction) were analyzed. For all sampling periods, except the afternoon (1220–1330) of 7 June 2007 when an abrupt change in wind direction occurred, winds blew over exposed wetland sediments in the direction of the sensors.

2.2. Great Bay Estuary, Tuckerton, New Jersey

[9] Great Bay estuary near Tuckerton, Ocean County, New Jersey (39.51°N, 74.32°W; Figure 1) includes an 87 km2 salt marsh and 56 km2 of shallow (2 m) estuarine waters (Rutgers University Marine Field Station (RUMFS) website, locations and facilities, http://marine.rutgers.edu/rumfs/RUMFShomepage.htm, 2008). Our field site was located approximately 200 m northwest of the Rutgers Marine Field Station and its access road, both of which are elevated 3 m above the marsh surface. The Great Bay lies to the south and west, Little Egg Inlet (a conduit between Little Egg Harbor and the Atlantic Ocean) to the east, and the Great Bay Wildlife Management Area to the north of our field site near the Rutgers Marine Field Station. Mercury and micrometeorological monitoring equipment were placed in the center of an open area of the tidal estuarine wetland, with vegetated sediment exposed at low tide.

[10] Given this site's proximity to the Atlantic Ocean, its waters are saline (22–30‰) (RUMFS website, locations and facilities, http://marine.rutgers.edu/rumfs/RUMFShomepage.htm, 2008). Vegetation on the mud flats of the Great Bay estuary is dominated by dwarf Spartina alterniflora, Salicornia sp. (pickleweed), and Suaeda sp. (sea blite) [Tiner, 1987].

[11] The region is predominantly surrounded by the New Jersey Pinelands National Reserve and other state and federal wildlife refuges, making it one of the most undisturbed estuaries on the east coast (RUMFS website, locations and facilities, http://marine.rutgers.edu/rumfs/RUMFShomepage.htm, 2008). Since no known point sources of mercury are present in the Great Bay estuary and mercury mainly enters this ecosystem via atmospheric deposition, we have selected this location as our “regional background” site for comparison with the Secaucus High School Marsh.

3. Methods

3.1. Field Sampling Events

[12] Field sampling was carried out over 3- to 8-h periods on 10–11 August 2005, 24–25 May 2006, and 7 and 21 June 2007 in Secaucus and continuously for 48 h on 2–4 October 2007 at the Great Bay. Photosynthetically active radiation (PAR) irradiance (LI-COR model LI-250 light meter), air temperature, and wind speed were recorded at both sites (Table 1). In Secaucus, gaseous mercury and micrometeorological equipment was set up along the edge of the wetland to sample lower boundary layer air as it traveled approximately 0.5 km across the tidally exposed marsh. As a result of the orientation of the Secaucus High School Marsh, sampling was only conducted in Secaucus under W–NW winds. At the Great Bay, equipment was set up in the center of an open area of tidal wetland which allowed sampling from all directions. Micrometeorological equipment was mounted on 5.7 cm diameter poles with minimal tripod support to avoid aerodynamic interference.

Table 1. Micrometeorological Parameters Measured in the Secaucus High School Marsh and the Great Bay Estuary, New Jerseya
SiteDateLocal TimeTide Height (m)PAR (μmol m−2 s−1)Air Temperature (°C)Wind Speed (m s−1)
Secaucus10 Aug 20050920–10201.04519321.0
1055–11551.601550320.9
1215–13051.951680332.0
1305–14002.09-341.8
11 Aug 20050920–10500.86810351.1
1050–12101.421530370.7
24 May 20061045–11500.551710193.9
1205–12500.111790204.4
1305–1430−0.081870213.1
1530–16100.631770221.6
1705–17501.761635231.4
1805–18502.13700231.1
25 May 20061220–13150.401840252.8
1320–14350.112200242.5
1440–15400.121020253.7
1545–16550.55935252.7
7 Jun 20071040–11500.841300202.4
21 Jun 20071110–12301.031670242.6
1240–13501.501760263.1
1400–15501.851688274.1
Great Bay2 Oct 20071735–19050.3815524-
2055–22250.62023-
2235–00050.98023-
3 Oct 20070015–01451.13025-
0155–03251.02030-
0335–05050.69028-
0515–06450.38026-
0655–08250.307324-
0835–10050.6222824-
1015–11451.0745024-
1155–13251.3867324-
1335–15051.44117324-
1515–16451.16110024-
1655–18250.75380242.1
1835–20050.4153233.1
2015–21450.330231.6
2155–23250.580231.5
2335–01050.930251.9
4 Oct 20070115–02451.090302.9
0255–04251.060292.6
0435–06050.730281.9
0615–07450.4220252.2
0755–09250.35165241.6
0935–11050.60405241.2
1115–12451.00443241.5
1255–14251.341110242.4
1435–16051.371310243.1
1615–17351.15640242.6

[13] Total gaseous mercury measurements were used to determine the difference in concentration over two heights (“concentration gradient”) in the lower boundary layer [Edwards et al., 2005]. In August 2005, TGM was monitored at 0.2 m and 1.5 m above the top of the Phragmites in Secaucus. For the May 2006 and June 2007 Secaucus sampling events, TGM was monitored between 0.05 m and 1 m above unvegetated sediment for the lower height and between 4.1 m and 4.4 m for the upper height. At the Great Bay, TGM was monitored at 0.2 m and 3.2 m above the marsh surface. TGM sample lines were the same length for both heights to minimize sampling bias. Micrometeorological data were collected at 0.8 m above the top of the Phragmites in August 2005 and at 3.1 m and 4.4 m above the sediment surface in May 2006 and June 2007, respectively, in Secaucus. Micrometeorological data were collected at 3.0 m above the sediment surface at the Great Bay. Sampling heights differed between collection dates in an effort to optimize mercury gradient measurements and adapt to site conditions.

3.2. Flux Calculations

[14] Vertical fluxes of gaseous mercury were estimated from measured vertical concentration gradients of TGM and calculated friction velocities and atmospheric stability correction factors in a modified form of the Thornthwaite-Holzmann equation [Majewski et al., 1991; Goodrow et al., 2005]:

display math

where FHg is the land-air mercury flux (ng m−2 h−1), u* is the friction velocity, κ is the von Karman constant, Hg1 and Hg2 are the TGM concentrations (ng m−3) at heights z1 (lower) and z2 (upper) above the ground, and ϕW is the atmospheric stability correction factor for water vapor [Thornthwaite and Holzman, 1939; Dyer and Hicks, 1970; Majewski et al., 1991]. The atmospheric stability correction factor for water vapor was used as a surrogate for that of mercury. Measurement uncertainties in u* and ϕw and the TGM concentration difference (Hg1 and Hg2) were propagated through the flux equation. Atmospheric mercury exists primarily (up to 99%) in the gaseous elemental (Hg0) form, [Zillioux et al., 1993; Lin and Pehkonen, 1999; Aucott et al., 2009], but reactive gaseous mercury may be present at greater than 1% under certain conditions [Sheu and Mason, 2001]. Although we did not measure gaseous mercury speciation in this study, elemental mercury represented more than 99.4% of the total gaseous mercury measured continuously over a two and a half year period (2004–2006) at an urban/industrial site in Elizabeth, New Jersey [Aucott et al., 2009] 23 km to the south–southwest of the Secaucus site. Total gaseous mercury concentrations are therefore assumed to be dominated by elemental mercury and the vertical fluxes estimated using equation (1) are assumed to be representative of gaseous elemental mercury.

[15] Atmospheric stability correction factors were calculated using the Monin-Obukhov length scale for turbulent mixing (L) [Obukhov, 1946]:

display math

where ρ is the air density, Cp is the specific heat of air at constant pressure, θ is the potential temperature, g is the acceleration of gravity, H is the directly measured sensible heat flux, and LE is the latent heat flux. Turbulent mixing length scale was further used to calculate nonadiabatic atmospheric stability correction factors according to Dyer and Hicks [1970]:

display math

where z is the height of eddy correlation measurements (3.08 m). Sign designation for TGM fluxes followed the convention of positive values corresponding to upward fluxes (mercury emission to the atmosphere) and negative values corresponding to downward fluxes (mercury deposition).

3.3. Micrometeorology

[16] Micrometeorological parameters were measured using an eddy correlation system (Campbell Scientific, Logan, Utah, USA). The eddy correlation (EC) system was used to resolve turbulent fluctuations in vertical and horizontal velocity, temperature, and specific humidity in the near surface atmosphere. The EC system consisted of a CSAT3 three-dimensional sonic anemometer, a KH2O ultraviolet krypton hygrometer, and a FW05 fine wire thermocouple. The CSAT3 sonic anemometer was used to determine momentum fluxes and sensible and latent heat fluxes at a rate of 10 Hz, with noise in horizontal and vertical velocities of 1 mm s−1 and 0.5 mm s−1, respectively. High-frequency (10 Hz) water vapor measurements were determined using the KH2O hygrometer and high-precision (±0.002°C) temperature measurements were made using a 0.0013 cm FW05 fine wire thermocouple. Micrometeorological measurements were processed to give 10 min averaged friction velocities and sensible and latent heat fluxes. Horizontal wind speeds and direction were also measured with Y.M. Young Wind Sentry (Campbell Scientific, Logan, Utah, USA) cup anemometers and vanes at both Hg sample heights with a range of 0 to 50 m s−1 and a threshold value of 0.5 m s−1. All micrometeorological instruments were calibrated and their analytical uncertainties determined by Campbell Scientific prior to deployment in the field.

3.4. Tekran 2537A Mercury Vapor Analyzer

[17] Ambient air was continuously sampled at two heights above the tidally exposed wetland sediments using an automated mercury sampler (Tekran 2537A, Toronto, Canada). TGM was collected by gold amalgamation and analyzed by cold vapor atomic fluorescence spectrometry (CVAFS) after thermal desorption [Schroeder et al., 1995]. The Tekran sampled ambient air at a flow rate of 1.5 L min−1 for a period of 5 min (7.5 L sample volume) in an effort to obtain optimum instrument detection limits (0.1 ng m−3). Two 5 min samples were taken at each height, one on each of the gold traps (A and B), during monitoring using a Synchronized Two-Port Sampler (Tekran, Model 1110). TGM concentrations from A and B cartridges were averaged for each of the upper and lower heights in an effort to remove minor systematic cartridge biases. Since TGM concentrations were not measured simultaneously at the two heights, vertical TGM concentration gradients were calculated over 40 to 80 min periods to obtain representative averages. The Tekran unit was housed in a temperature controlled tent to prevent overheating and was calibrated regularly using gas injection calibrations in the laboratory prior to field deployment and daily internal calibrations in the field to assure analyzer performance. For field sampling periods, sample line blanks with zero air gave no detectable peaks and instrument response factors ranged from 1.7 × 106 to 3.7 × 106 peak area ng Hg−1 and varied by less than 15% within each sampling day.

3.5. Sediment Mercury Concentration

[18] Composite sediment samples (10–20 samples per composite) were collected by shovel, placed in sealed 5 gallon HDPE containers and stored, untreated, in a cold chamber at 10°C until analysis. Care was taken to minimally disrupt the ecosystem during sample collection.

[19] Total mercury was determined for each homogenized, composite sediment sample by EPA Method 1631 after aqua regia digestion. Digested sediment samples were diluted and reduced with 0.5 M SnCl2. Elemental mercury was purged with Hg-free argon, collected onto gold-coated sand traps, passed through a dual amalgamation system, and subsequently thermally desorbed to a Tekran model 2500 (Toronto, Canada) CVAFS detector [Fitzgerald and Gill, 1979; Bloom and Crecelius, 1983].

[20] Bubbler and reagent blanks were run for quality control purposes. A gas injection calibration curve was performed and compared with liquid standard spikes to determine recovery. Average recoveries were 100% ±10%.

3.6. Statistical Analyses

[21] Percent gradients were calculated as:

display math

where Hg1–Hg2 is the vertical TGM concentration gradient (ng m−3) and Hg1 is the TGM concentration (ng m−3) at the lower sample height [Kim and Kim, 1999; Goodrow et al., 2005]. Samples with percent TGM gradients less than 2.6% (twice the relative analytical uncertainty) in the Secaucus High School Marsh and less than 0.72% in the Great Bay estuary were assigned vertical TGM fluxes of zero. Analytical uncertainty was estimated as the minimum relative standard deviation of field measurements collected over a 1-h period at the lower sample height. Given these uncertainties and ambient TGM concentrations during sampling, minimum detectible TGM gradients (absolute value) were ≤0.09 ng m−3 in Secaucus and ≤0.02 ng m−3 at the Great Bay. Depending on friction velocities and atmospheric stability factors, minimum detectible fluxes were 16 to 48 ng m−2 h−1 in Secaucus and 3 to 10 ng m−2 h−1 at the Great Bay.

[22] TGM concentrations at the two sites were compared using a two-tailed t test for unequal variances. Regression analyses were used to analyze relationships between mercury fluxes and wind speed or air temperature and between cumulative mercury flux and cumulative irradiance.

4. Results

[23] The concentrations of total mercury in sediments at the two field sites were 7.1 mg kg−1 and 0.45 mg kg−1 (both dry wt.) at the Secaucus High School Marsh and Great Bay estuary, respectively. Ambient TGM concentrations measured 4.3 m above the Secaucus High School Marsh ranged from 1.4 to 5.1 ng m−3 (Table 2). The average summertime concentration of TGM at the Secaucus site (2.5 ng m−3) was similar to that observed in Bayonne, New Jersey (2.2 ng m−3) 18 km to the south [Goodrow et al., 2005] and Elizabeth, New Jersey (2.3 ng m−3) 23 km to the south–southwest [Aucott et al., 2009]. Ambient TGM concentrations measured 3.2 m above the marsh surface at the Great Bay estuary site ranged from 2.3 to 3.4 ng m−3 (Table 3). The average concentration of TGM at the Great Bay site (3.0 ng m−3) was higher than that in Secaucus (p = 0.06, two-tailed t test for unequal variances), but Secaucus TGM concentrations were more variable than those at the Great Bay. Indeed, the highest ambient TGM concentrations (4.9 to 5.1 ng m−3) were recorded in Secaucus on 11 August 2005, a day when daytime air temperatures ranged from 35 to 37°C. Average ambient concentrations measured above both sites were elevated compared with global background concentrations although measurements were only recorded during summer/early fall and may not be representative of actual average annual concentrations.

Table 2. Measurements at Secaucus High School Marsh, New Jerseya
DateLocal Timeu*equation imagewTGM Gradient (ng m−3)Ambient TGM (ng m−3)TGM Flux (ng m−2 h−1)
  • a

    Friction velocities (u*), atmospheric stability correction factors (ϕw), average vertical concentration gradients of total gaseous mercury (TGM), ambient TGM concentrations, and sediment-air mercury fluxes. Values are means ± SD for ambient TGM, u*, and ϕw and means ± propagated error for TGM gradients and fluxes. Atmospheric correction factors for water vapor were calculated using the Monin-Obukhov length scale [Obukhov, 1946] according to Dyer and Hicks [1970]. Samples with TGM percent gradients <2.6% (twice the relative analytical uncertainty) were assigned vertical fluxes of zero. Hg concentrations measured at the upper sample height (4.3 m) were taken as ambient.

10 Aug 20050920–10200.19 ± 0.060.59 ± 0.18−1.44 ± 0.053.25 ± 0.08−375 ± 165
1055–11550.26 ± 0.080.65 ± 0.17−0.83 ± 0.053.30 ± 0.55−269 ± 109
1215–13050.31 ± 0.090.69 ± 0.151.04 ± 0.041.46 ± 0.15203 ± 74
1305–14000.36 ± 0.110.37 ± 0.200.15 ± 0.052.62 ± 0.1863 ± 45
11 Aug 20050920–10500.23 ± 0.050.75 ± 0.1305.05 ± 0.42-
1050–12100.20 ± 0.040.54 ± 0.1004.87 ± 0.06-
24 May 20061045–11500.30 ± 0.020.46 ± 0.070.44 ± 0.042.09 ± 0.0697 ± 19
1205–12500.29 ± 0.040.41 ± 0.070.22 ± 0.052.31 ± 0.0351 ± 15
1305–14300.30 ± 0.050.40 ± 0.070.55 ± 0.041.98 ± 0.04136 ± 34
1530–16100.27 ± 0.020.39 ± 0.010.71 ± 0.041.82 ± 0.04166 ± 15
1705–17500.27 ± 0.020.55 ± 0.080.25 ± 0.031.49 ± 0.0576 ± 16
1805–18500.32 ± 0.020.94 ± 0.110.17 ± 0.031.38 ± 0.0036 ± 7
25 May 20061220–13150.30 ± 0.020.54 ± 0.070.22 ± 0.052.59 ± 0.0475 ± 20
1320–14350.31 ± 0.020.69 ± 0.070.26 ± 0.062.92 ± 0.2071 ± 18
1440–15400.30 ± 0.020.70 ± 0.080.24 ± 0.052.53 ± 0.0863 ± 15
1545–16550.29 ± 0.010.55 ± 0.020.44 ± 0.052.39 ± 0.05142 ± 18
7 Jun 20071040–11500.24 ± 0.150.31 ± 0.220.32 ± 0.052.44 ± 0.13227 ± 215
21 Jun 20071110–12300.30 ± 0.090.42 ± 0.150.93 ± 0.041.90 ± 0.10677 ± 310
1240–13500.33 ± 0.070.45 ± 0.100.49 ± 0.042.14 ± 0.13364 ± 117
1400–15500.36 ± 0.090.54 ± 0.140.55 ± 0.041.67 ± 0.12374 ± 139
Table 3. Measurements at Great Bay Estuary, New Jerseya
DateLocal Timeu*equation imagewTGM Gradient (ng m−3)Ambient TGM (ng m−3)TGM Flux (ng m−2 h−1)
  • a

    Friction velocities (u*), atmospheric stability correction factors (ϕw), average vertical concentration gradients of total gaseous mercury (TGM), ambient TGM concentrations, and sediment-air mercury fluxes. Values are means ± SD for ambient TGM, u*, and ϕw and means ± propagated error for TGM gradients and fluxes. Atmospheric correction factors for water vapor were calculated using the Monin-Obukhov length scale [Obukhov, 1946] according to Dyer and Hicks [1970]. Samples with TGM percent gradients <0.72% (twice the relative analytical uncertainty) were assigned vertical fluxes of zero. Hg concentrations measured at the upper sample height (3.2 m) were taken as ambient.

2 Oct 20071735–19050.33 ± 0.030.79 ± 0.000.05 ± 0.043.43 ± 0.0510 ± 4
2055–22250.21 ± 0.020.76 ± 0.050.08 ± 0.043.39 ± 0.1112 ± 3
2235–00050.17 ± 0.020.61 ± 0.050.06 ± 0.033.27 ± 0.138 ± 3
3 Oct 20070015–01450.19 ± 0.020.68 ± 0.060.08 ± 0.033.19 ± 0.2711 ± 3
0155–03250.22 ± 0.020.77 ± 0.05−0.23 ± 0.033.26 ± 0.05−34 ± 4
0335–05050.21 ± 0.010.75 ± 0.02−0.10 ± 0.033.37 ± 0.07−15 ± 3
0515–06450.21 ± 0.020.74 ± 0.06−0.11 ± 0.033.36 ± 0.24−16 ± 3
0655–08250.21 ± 0.020.75 ± 0.050.20 ± 0.033.23 ± 0.0630 ± 5
0835–10050.15 ± 0.040.57 ± 0.140.41 ± 0.043.20 ± 0.0558 ± 21
1015–11450.17 ± 0.040.64 ± 0.120.29 ± 0.033.23 ± 0.0742 ± 12
1155–13250.26 ± 0.040.80 ± 0.030.39 ± 0.033.09 ± 0.0768 ± 10
1335–15050.29 ± 0.020.79 ± 0.020.29 ± 0.033.14 ± 0.1156 ± 5
1515–16450.23 ± 0.040.76 ± 0.040.20 ± 0.033.14 ± 0.0832 ± 6
1655–18250.22 ± 0.030.74 ± 0.110.09 ± 0.033.10 ± 0.1315 ± 4
1835–20050.34 ± 0.050.89 ± 0.0302.80 ± 0.13-
2015–21450.17 ± 0.030.59 ± 0.16−0.02 ± 0.032.70 ± 0.02−3 ± 2
2155–23250.16 ± 0.040.59 ± 0.1202.51 ± 0.06-
2335–01050.23 ± 0.110.75 ± 0.13−0.04 ± 0.022.37 ± 0.05−6 ± 4
4 Oct 20070115–02450.31 ± 0.080.76 ± 0.2802.52 ± 0.07-
0255–04250.34 ± 0.040.79 ± 0.0402.60 ± 0.12-
0435–06050.29 ± 0.050.85 ± 0.1302.35 ± 0.02-
0615–07450.16 ± 0.020.62 ± 0.1102.29 ± 0.06-
0755–09250.14 ± 0.020.45 ± 0.130.04 ± 0.032.44 ± 0.286 ± 3
0935–11050.13 ± 0.050.32 ± 0.150.17 ± 0.032.50 ± 0.0637 ± 22
1115–12450.17 ± 0.040.40 ± 0.110.21 ± 0.032.50 ± 0.0648 ± 17
1255–14250.27 ± 0.040.43 ± 0.050.25 ± 0.032.52 ± 0.0581 ± 16
1435–16050.35 ± 0.030.56 ± 0.040.15 ± 0.032.55 ± 0.0451 ± 7
1615–17350.33 ± 0.080.54 ± 0.060.07 ± 0.032.49 ± 0.0522 ± 7

[24] Vertical gaseous mercury concentration gradients at the Secaucus site varied from −1.4 to 1.0 ng m−3 and were positive (higher near the sediment surface) in 16 out of 20 measurement events (Table 2). During two time periods, vertical TGM gradients were negative indicating net absorption of atmospheric TGM by the marsh, and in two others, TGM gradients were below detection (Table 2). The below detection vertical concentration gradients were recorded on 11 August 2005 when ambient TGM concentrations were elevated thus minimizing the concentration difference near the land surface. Estimated daytime land-air mercury fluxes ranged from −375 to +677 ng m−2 h−1 at the Secaucus High School Marsh (Table 2). Note that fluxes determined for both sites are representative of the summer/early fall only and should not be used to estimate winter or annual fluxes.

[25] Positive vertical gaseous mercury concentration gradients were observed in 17 out of 28 measurement periods at the Great Bay site and negative gradients were observed during five of these periods (Table 3). Six gradients at the Great Bay site were below detection. Vertical TGM concentration gradients at the Great Bay site exhibited a diurnal pattern with the highest positive gradients occurring during midday and the lowest gradients measured at night (Figure 2). Early fall land-air TGM fluxes at the Great Bay site ranged from −34 to +81 ng m−2 h−1 (Table 3).

Figure 2.

Diel trends in the vertical total gaseous mercury (TGM) concentration gradient from 0.2 m to 3.2 m above the Great Bay salt marsh mud flats (diamonds) and ambient TGM measured at 3.2 m (triangles) in Tuckerton, New Jersey, 2–4 October 2007.

[26] For the Secaucus High School Marsh, relative propagated uncertainties in mercury fluxes accounting for uncertainties in micrometeorological and TGM measurements ranged from 9% to 95%, but were less than 45% for 15 out of 18 measurement periods. Mercury fluxes estimated for the Great Bay salt marsh had relative propagated uncertainties that ranged from 9% to 67% and were less than 45% for 18 out of 22 measurement periods.

[27] Small fluxes of <10 ng m−2 h−1 accounted for about half of all values in the Great Bay estuary. By comparison, only 10% of estimated mercury fluxes in Secaucus were <10 ng m−2 h−1. Negative fluxes, indicating net movement of mercury from the air to the marsh surface, were only observed in the morning in the presence of Phragmites in Secaucus and at night at the Great Bay site. In Secaucus, the observed negative fluxes also coincided with elevated ambient air concentrations of >3 ng m−3, potentially suppressing Hg volatilization and enhancing deposition.

[28] The comparison of cumulative mercury fluxes and cumulative solar radiation is a new technique for mercury volatilization data analysis that minimizes time lag effects between photochemical reduction of Hg(II) and physical volatilization and permit the estimation of in situ efficiencies of light-driven mercury volatilization [Wollenberg and Peters, 2009]. Cumulative mercury fluxes were positively correlated with cumulative solar radiation from 24 May 2006 1045 through 25 May 2006 1655 at the Secaucus site (r2 = 0.97, p < 0.01; Figure 3) and from 3 October 2007 0835 through 4 October 2007 1735 at the Great Bay site (r2 = 0.97, p < 0.01). No significant correlations were found between mercury fluxes and wind speed or air temperature at either site.

Figure 3.

Correlations between cumulative mercury flux and cumulative photosynthetically active solar radiation (PAR) from 24 May 2006 1045 through 25 May 2006 1655 in the Secaucus High School Marsh (n = 20; p < 0.01) and from 3 October 2007 0835 through 4 October 2007 1735 in the Great Bay estuary (n = 29; p < 0.01).

5. Discussion

5.1. Abiological Factors Driving Mercury Volatilization From Wetlands Sediments

[29] Mercury concentration in sediment and soil is a first-order factor controlling land-air gaseous mercury fluxes [Gustin et al., 2000]. The sediment mercury concentration at the Great Bay site (0.5 mg kg−1) is at the high end of the range observed in nonpoint source impacted (inputs from atmospheric deposition only) sediments (0.007–1 mg kg−1) [Gilmour et al., 1992; Cai et al., 1997; Leonard et al., 1998a, 1998b; Hung and Chmura, 2006]. In contrast, the concentration of total mercury in Secaucus High School Marsh sediments (7 mg kg−1) is greater than that in nontidal sediments of the New York/New Jersey Harbor and Newark Bay (0.3 to 2.0 mg kg−1 [Hammerschmidt et al., 2008]), but similar to that of contaminated areas of the Savannah River in South Carolina where sediment total mercury concentrations are as high as 10 mg kg−1 [Kaplan et al., 2002]. Although mercury concentrations were tenfold higher in Secaucus sediments than at the Great Bay site, another environmental factor, sunlight, appears to be a more important control of mercury volatilization from these sediments.

[30] Mercury volatilization fluxes for the Secaucus High School Marsh and the Great Bay estuary showed clear diurnal trends with peak emissions in the mid to late afternoon and small bidirectional fluxes overnight and into the morning (Tables 2 and 3; Figure 2). The observed diurnal patterns are similar to those of previous studies and is likely due to photochemical reduction of Hg(II) [Kim et al., 1995; Canario and Vale, 2004; Feng et al., 2005] which is present in sediments in photochemically active forms including organic matter complexes [Zhang and Lindberg, 1999] and solid phase HgS [Nriagu, 1994]. The very strong correlations (r2 = 0.97) of cumulative mercury volatilization fluxes and cumulative PAR irradiances for both sites (Figure 3) provide additional evidence that photochemistry drives mercury volatilization from these salt marsh sediments. Indeed, the slopes of the cumulative mercury flux-cumulative PAR relationships for Secaucus and Great Bay are nearly identical and provide an estimate of the in situ efficiency of light-driven mercury volatilization of about 15 ng Hg mol PAR photons−1 or about 0.06 ng Hg kJ−1. The fact that these two sediments from strikingly different environments and with mercury concentrations that differ by a factor of 10 have nearly identical cumulative light-driven mercury volatilization efficiencies indicates that they may support similar mechanisms of mercury photoreduction.

[31] While photochemistry appears to be the primary driver of mercury volatilization from these salt marsh sediments, other physical factors may also be important under different environmental conditions. Low nighttime fluxes of mercury, as observed in this study, may result from a stable, stratified atmospheric boundary layer which causes transport to become dominated by molecular diffusion rather than turbulent mixing [Kim et al., 1995]. However, meteorological parameters did not exhibit strong diurnal trends in this study, indicating that they did not play a major role in controlling mercury volatilization fluxes. For example, for the period of 1155 on 3 October 2007 to 0425 on 4 October 2007, friction velocities and atmospheric stability correction factors were relatively constant (Table 3). Over this same period, however, the vertical gaseous mercury concentration gradient decreased from 0.4 to zero indicating that the source of volatile elemental mercury from reduction of Hg(II) in the sediments was effectively turned off over night (Table 3 and Figure 2).

[32] Temperature has been considered an important environmental factor driving the reemission of mercury from terrestrial surfaces [Feng et al., 2005] and positive correlations between mercury evasion fluxes and air temperature have been observed in field studies [Kim et al., 1995; Canario and Vale, 2004]. However, correlations between mercury flux and temperature may ultimately reflect the correlation between mercury flux and solar radiation [Poissant and Casimir, 1998; Feng et al., 2005; O'Driscoll et al., 2007] and mercury volatilization fluxes are often independent of air temperature [Gustin et al., 1997, 2002; Bahlmann et al., 2004; Feng et al., 2005]. In this study, mercury fluxes were not correlated with air temperature. For example, mercury fluxes at both sites decreased in the afternoon as sunlight decreased, but ambient air temperatures remained relatively constant (Tables 13). Air temperature was uncorrelated with TGM flux in Secaucus and was weakly negatively correlated in the Great Bay (r2 = 0.24, p < 0.01). Sediment temperature was not measured in our study, but for an uncontaminated forest site, mercury volatilization was only weakly correlated with soil temperature [Kim et al., 1995].

[33] Higher salinity at the Great Bay than Secaucus might have affected photochemical (or biological) reduction of Hg(II), and therefore mercury volatilization, through greater complexation with sulfide at the Great Bay, but the nearly identical light-driven mercury volatilization efficiencies for the two sites suggests that this was a minor effect. The degree of tidal flooding did not appear to affect mercury volatilization as tide heights were poorly correlated with mercury fluxes (r2 = 0.03 for Secaucus; r2 = 0.25 for Great Bay). Tidal inundation was also found to have no effect on mercury flux from the Farm River salt marsh [Lee et al., 2000] (see Table 4 and section 5.3).

Table 4. Average Land-Air Gaseous Mercury Volatilization Fluxes for Wetland Sediments, Two Soils, and Mine Waste Determined Using the Micrometeorological Methoda
SiteDescriptionSediment Hg (mg kg−1)TGM Flux (ng m−2 h−1)
Secaucus H.S. Marsh, NJbNonvegetated salt marsh sedimentsc7.185
Great Bay estuary, NJbdwarf Spartina grass salt marsh0.4517
Farm River salt marsh, CTdtall Spartina salt marsh, before leaf-out0.2–0.513
Farm River salt marsh, CTdtall Spartina salt marsh, full foliage0.2–0.5−3.3
Walker Branch, Oak Ridge, TNeforest soil0.5–0.78–18
East Fork Poplar Creek, TNfcontaminated forest soil20–6010–200
Carson River, NVgcontaminated mine waste4–255115–4152

5.2. Biological Factors Affecting Mercury Volatilization

[34] Plants have been implicated in land-air mercury exchange [Lindberg, 1996; Leonard et al., 1998a; Ericksen et al., 2006] and foliar surfaces may act as sources or sinks of gaseous mercury depending on atmospheric concentration [Hanson et al., 1995]. Photosynthesis is highest at midday, coincident with peak Hg fluxes, so it may be difficult to separate the effects of direct sediment photoreduction from plant-induced processes in heavily vegetated salt marshes. The results of 10 August 2005 indicate net absorption of atmospheric mercury by the Secaucus salt marsh in the presence of the tall grass Phragmites (Table 2). However, net absorption was only observed in the morning hours before solar irradiance and the sun's elevation reached their maxima (1330 local time). Definitive conclusions regarding the role of Phragmites in mercury land-air exchange cannot be drawn given the small number of observations with Phragmites at the Secaucus site, but gaseous mercury uptake was observed at the Farm River salt marsh in summer with the marsh grass Spartina patens at full foliage [Lee et al., 2000] (see section 5.3). Phragmites could not have played a role in mercury absorption or volatilization at the Secaucus site in 2006 and 2007 since the marsh had been cleared to bare sediment in preparation for wetland restoration. At the Great Bay, dwarf Spartina alterniflora and the waxy Salicornia and Suaeda are not expected to play major roles in mercury gas exchange because of their small surface areas relative to sediment and the low gas exchange rates of these salt-tolerant C-4 plants [Maricle et al., 2007; Voznesenskaya et al., 2007]. Nonetheless, the specific roles of vascular plants in mercury cycling at the land-air interface, especially for tall grass wetlands, are incompletely understood and require additional study.

[35] Bioturbation has the potential to facilitate the transport of elemental mercury from reduced sediment layers to the surface [Canario and Vale, 2004; Robbins and Edgington, 1975; Kramer et al., 1991] and was evident in surface sediments at both sites by the activities of snails, fiddler crabs, and other sediment-dwelling invertebrates. At the Great Bay site, bioturbation did not directly drive mercury volatilization since no mercury was emitted to the atmosphere at night, but bioturbation could have brought Hg(II) to the surface overnight, followed by subsequent photoreduction and volatilization the next day.

[36] Since overnight mercury fluxes were zero at the Great Bay, mercury reduction by nonphotosynthetic bacteria does not appear to have been a significant source of volatile mercury at this site. Photosynthetic microalgae and cyanobacteria are capable of reducing mercury [Lefebvre et al., 2007], but algal or cyanobacterial mats were not observed on the sediment surfaces in either Secaucus or the Great Bay.

5.3. Comparisons With Other Micrometeorological Mercury Emissions Studies

[37] There are only a few micrometeorological studies of mercury volatilization from sediment/soil surfaces to the atmosphere and only one previous study [Lee et al., 2000] employed this technique in a salt marsh (Table 4). Micrometeorological mercury fluxes were positive (land to air mercury volatilization) and of similar magnitude for the Great Bay estuary, the Farm River salt marsh before leaf-out in the spring, and the Walker Branch watershed (Table 4). The negative mercury fluxes (uptake by land) observed at the Farm River salt marsh occurred after the marsh grass Spartina patens had achieved full foliage [Lee et al., 2000]. Prior to leaf-out, mercury volatilization at Farm River was strongly correlated with solar radiation, but after leaf-out, correlations between mercury volatilization and either soil temperature or solar radiation weakened [Lee et al., 2000]. Thus, the results of Lee et al. [2000] and the present study indicate that photochemistry plays an important role in mercury volatilization from salt marshes unless tall, leafy vegetation is present, in which case mercury fluxes will be small or negative (from air to land). Soil mercury fluxes at the Walker Branch site were correlated with air temperature and weakly with soil temperature [Kim et al., 1995]. Solar radiation was not directly evaluated in the Walker Branch study, but peak mercury emissions were consistently observed during the early afternoon indicating a possible role for light. The very high mercury volatilization fluxes estimated for contaminated forest soils in East Fork Poplar Creek, Tennessee and contaminated mine waste in Carson River, Nevada reflect the high levels of mercury at these sites, but were also shown to be driven by light [Gustin et al., 1997, 2002].

5.4. Annual Mercury Emissions From Salt Marsh Wetlands

[38] The potential contribution of salt marsh wetlands to annual mercury emissions was evaluated for New Jersey and the eastern United States. Although limited in number and seasonal coverage, our summertime–early autumn measurements provide some of the only data with which to constrain sediment-air volatilization of mercury from salt marsh ecosystems. Annual salt marsh mercury emissions were estimated using the 24-h mean mercury flux at the Great Bay regional background site (17 ± 4 ng m−2 h−1) and the areas of tidal salt marsh in New Jersey (600 km2) [Dahl, 1990] and the eastern United States (including the Gulf of Mexico; 2.3 × 104 km2) [Stedman and Dahl, 2008]. Since the results of this study and Lee et al. [2000] suggest that mercury volatilization from salt marshes tends to decrease in the presence of tall grasses at full foliage and since the Great Bay site had minimal vegetation, the annual emissions estimated here likely represent an upper limit on the contribution of salt marsh emissions in these areas. For New Jersey, tidal salt marshes may account for mercury emissions of as much as 90 ± 20 kg a−1. This value is within the range of mercury emissions from various industrial sources in New Jersey (2–450 kg a−1). Mercury emissions from tidal salt marshes in the eastern United States may be as high as 3400 ± 800 kg a−1 which represents about 3% of industrial emissions in the eastern United States [Walcek et al., 2003]. The annual volatilization flux of mercury from the Great Bay salt marsh (150 μg m−2 a−1) is approximately 10 times greater than the direct atmospheric deposition flux of mercury in central New Jersey (14 μg m−2 a−1) [Zhuang, 2003]. A greater emission than direct atmospheric deposition flux occurs in this system because estuarine sediments concentrate atmospheric mercury deposition from watersheds that are 10 to 30 times larger in area as well as the coastal ocean.

6. Conclusions and Implications

[39] Our results suggest that mercury volatilizes to the atmosphere from tidally exposed, estuarine wetlands where sediments are exposed directly to solar radiation and the atmosphere. We conclude that, in the absence of tall leafy grasses, solar radiation is the most important factor controlling in situ volatilization of mercury from tidally exposed sediments. Specific effects of other environmental parameters on mercury volatilization from salt march sediments, including salinity, tidal inundation, vegetation type, and sediment temperature appear to be of secondary importance, but require further examination. The average mercury volatilization flux estimated for the Great Bay salt marsh in this study (17 ng m−2 h−1) compares well with other micrometeorological mercury flux estimates for uncontaminated salt marsh and forest soils (8–18 ng m−2 h−1) and is more than 10 times higher than the average mercury emission flux from land (∼1 ng m−2 h−1) [Fitzgerald and Mason, 1996]. Annual mercury emissions from salt marsh wetlands may be comparable to individual industrial emissions sources in coastal states of the eastern United States. More quantitative and mechanistic studies of the volatilization of gaseous mercury from wetlands are needed to evaluate its importance to land-air mercury cycling and the physical and biological factors that modulate this flux.

Acknowledgments

[40] We would like to thank Rose Petrecca and Kenneth Able of the Rutgers University Marine Field Station (RUMFS) for facilitating our study on the Great Bay estuary, Robert Miskewitz for help with the Secaucus field work, Sandy Goodrow for help creating the site map, and Steve Peters for helpful comments. Support for this work was provided by the Hudson River Foundation, the New Jersey Department of Environmental Protection, the Department of Education through a Graduate Assistance in Areas of National Need grant, and a Hatch/McIntyre-Stennis grant through the New Jersey Agricultural Experiment Station.

Ancillary