S9 (II) Understanding mercury - organic matter interactions

Hypolimnetic cycling of ambient Hg (Hg_Amb) and amended isotopic Hg (²⁰²Hg, purity 90.8%; ²⁰²HgLake) in Lake 658 in the Experimental Lakes Area of northwest Ontario was investigated through detailed water column aqueous and solid phase sampling and phase partitioning studies. Sedimenting organic matter was collected by traps situated both below the thermocline and directly above the sediment-water interface. Detailed aqueous Hg profiles in the region directly above the sediment-water interface were obtained using an in-situ close-interval sampler. Settling plankton was an important vector for the transport of Hg_Amb and ²⁰²HgT_Lake and ²⁰²MeHg_Lake to the sediment-water interface. Significant concentrations of isotopic HgT, and to a lesser extent MeHg, were found in sediment trap particles 6 days after addition of the first spike, indicating rapid particulate organic matter partitioning and transport of both forms to the sediment-water interface. Isotopic HgT and MeHg were present in the filtered phase near the sediment-water interface in the hypolimnion two weeks after the first spike, indicating diagenesis of recently-fallen organic particles. Use of Diethylaminoethyl (DEAE) anionic resin revealed that water column Hg and MeHg were dominated by DOM complexation even in the presence of sulfide. Spectrophotometric DOM characterization indicated an increase in aromaticity and molecular weight with depth in the water column, suggesting increased allochthonous carbon near the sediment-water interface.

Solid-phase speciation revealed the role of active iron transport as a delivery mechanism of HgT and MeHg and the importance of iron oxidation and dissolved organic matter occlusion in forming aggregates that remove Hg and MeHg from solution, providing a mechanism for the redelivery of Hg and MeHg to the sediment-water interface in the hypolimnion. Suspended particulate matter isolated from the oxic epilimnion (zone of active primary production), the oxic hypolimnion (zone of active Fe oxidation), and the anoxic hypolimnion (zone of active FeS formation) was sequentially extracted (1M KOH, 6M HCl, 12M HNO3) to better understand the association of Hg and MeHg to solid organic and inorganic phases across these geochemical gradients. Solid-phase speciation of Hg was dominated by base-extractable (organic) phases; however, acid-extractable (inorganic) phases exhibited enhanced importance in the anoxic hypolimnion. Recalcitrant strong acid-extractable phases predominated in the zone of active sulfide production.

In a series of laboratory experiments using radioisotopes to track methylmercury (Me²⁰³Hg) accumulation by phytoplankton, we found the concentration of dissolved organic carbon (DOC) was inversely related to the amount of Me²⁰³Hg taken up by phytoplankton. The amount of Me²⁰³Hg accumulated by phytoplankton (Cyclotella meneghiniana) was expressed as volume concentration factor (VCF), or amount of Me²⁰³Hg in cells divided by the amount of Me²⁰³Hg in an equivalent volume of water. The relationship between VCFs and DOC was non-linear, with the highest VCFs (250,000) occurring when no organic matter was added and a rapid drop in VCFs (to ~50,000) occurring as dissolved organic carbon (DOC) concentrations approached 200 µM. This relationship held true for a variety of types of organic matter. Most organic matter was isolated by XAD resins from the San Francisco Bay Delta and then reconstituted at various concentrations in the laboratory. Surprisingly, organic matter created from freshly lysed algal cells resulted in similar VCFs as the San Francisco Bay Delta isolates. Recent data from the Hudson River Estuary also shows remarkable overlap with the San Francisco Bay Delta data. These results suggest that DOC can be a useful indicator of Me²⁰³Hg bioavailability, even when reduced sulfur concentrations are not known. The fraction of the organic matter also affected VCFs. Higher VCFs in the transphilic fraction than in the hydrophobic fraction indicated that binding to aromatics could limit Me²⁰³Hg bioavailability. VCFs were also affected by Cl- concentrations. Under low DOM concentrations, algal cells grown in high (28,000 µM) Cl- concentrations had VCFs around 300,000 whereas cells grown in low (230 µM) Cl- concentrations had VCFs around 200,000. Low concentrations of Cl- combined with high organic matter resulted in VCFs < 100,000. These differences in algal VCFs were important because the amount of Me²⁰³Hg in amphipod grazers (Hyalella azteca) was proportional to the initial concentrations in phytoplankton. DOC is likely an important predictor MeHg accumulation in the food web. However, DOC concentrations did not affect the loss of Me²⁰³Hg from amphipods as seen by assimilation efficiencies. These results are an important step in understanding the factors that govern the bioavailability and subsequent trophic transfer of MeHg.

Elevated mercury levels have long been reported in the foodwebs of the Florida Everglades and adjacent coastal waters. However, processes affecting mercury species distribution within the Everglades and fringing mangrove marsh remain poorly understood. Mangroves are of particular interest because their high organic matter inputs, high reported rates of sulfate reduction and high reported methyl mercury concentrations suggest methyl mercury production is high, while concentrations of mercury in fish tend to be lower with proximity to the coast.

In this study, we sought to improve our understanding of processes controlling production and transport of DOC and mercury species from tidal mangrove environments into adjacent aquatic systems, Everglades marshes, or the Gulf of Mexico. We sampled the Shark River in Everglades National Park over a two day period, and continuously measured discharge and fluorescent dissolved organic matter (FDOM) over a two week period in September of 2010. We observed strong significant relationships between FDOM and DOC (r² = 0.99), between FDOM and filtered total Hg (r² = 0.82), and between FDOM and filtered MeHg (r² = 0.90). Relationships between FDOM and particulate mercury species were not significant.

The strong relationships between the dissolved materials and FDOM allowed us to calculate the tidal fluxes as a product of the continuous measurements of FDOM and discharge. Areal fluxes were estimated as a function of tidal exchange volume and change in tide height. Calculated areal fluxes of DOC (~0.8 g C m^-2 d^-1) are similar to global estimates of DOC flux from mangroves (~0.6 g m^-2 C d^-1) and to fluxes previously reported from a Shark River mangrove mesocosm (~0.8 g C m^-2 d^-1). However, areal fluxes of dissolved mercury species – ~136 ng m^-2 d^-1 for dissolved Hg and ~0.8 ng m^-2 d^-1 for dissolved MeHg - were 10 to 100 times greater than values previously published for wetlands, but similar to tidal fluxes measured by us in a San Francisco Bay tidal wetland.

These results indicate that tidal transport of mercury species from tidal mangrove marshes can represent large sources of total and methyl mercury to local aquatic systems and coastal foodwebs, with the magnitude of transport related to the physical dynamics. These results also demonstrate the utility of using in situ optical measurements in combination with physical dynamical measurements to understand DOC and Hg processes in tidal environments.

Mercury (Hg) emitted from industrial sources has resulted in Hg-contamination of remote aquatic ecosystems. Although there is strong movement to regulate these sources, uncertainty surrounding the release of historically-deposited Hg that has accumulated in upland soils presents a regulatory challenge. Both biogeochemical and hydrological processes are expected to exert a major control on the timing and magnitude of the boreal upland response to changes in Hg loading. In this paper, we assessed the relative importance of hydrological and soil biogeochemical controls on event-scale upland mercury export in the METAALICUS (Mercury Experiment to Assess Atmospheric Loading in Canada and the U.S.) experimental catchment in northwestern Ontario. In this catchment, antecedent water storage deficits in large, soil-filled bedrock depressions exert a primary and predictable control on runoff generation by regulating subwatershed hydrologic connectivity. Surface organic soil horizons dominate the soil Hg pool, and are dynamic in terms of Hg release into soil water. Thus, the boreal shield landscape topography enhances the flushing of DOC and associated Hg from this near-surface zone. To assess the relative importance of hydrological and soil biogeochemical controls on Hg fluxes in runoff, we compared total Hg (THg) and DOC fluxes for two storm events with different antecedent moisture conditions (AMCs). We observed a distinct shift in the concentration-discharge relationship for THg and DOC with varying AMCs. The event with relatively wet AMCs exhibited nearly constant THg and DOC concentrations, whereas concentrations of THg and DOC increased with discharge for the event with dry AMCs. This difference suggests that biogeochemical processes, such as the decomposition of soil organic matter, dominate over hydrological processes in the release of Hg from soils after dry periods and may play an important role in regenerating the pool of mobile Hg in the catchment. A common assumption is that as catchments become drier in a warmer climate there will be less runoff, and hence less DOC and Hg will reach the lake. However, our results suggest that the pool of potentially mobile DOC and Hg grows as catchments become drier and that re-wetting events after warm and dry periods enhance the flushing of these pools. Therefore, correct predictions of the effects of climate warming on lake chemistry must consider the balance between runoff generation and the development of the potential DOC and Hg pools on longer time scales.

Both soil solution pH and soil particle size distribution have been shown to influence the adsorption of Hg_D to the soil matrix and therefore the amount of Hg_D in soil solution available to be transported to the stream. While prominent in laboratory literature, no watershed-scale studies have sought to determine the effects of basic soil properties and pH on Hg transport. We monitored three watersheds of similar size with comparable climate, seasonal temperature, land use/land cover (no wetlands in any of the sites), soil organic carbon content, watershed slope, mean discharge, and Hg deposition, but with different underlying bedrock geology. Differences in weathering capacity of bedrock result in a gradient in soil composition (clays to sands) as well as acid neutralizing capacity (ANC), resulting in a range of soil pH at the sites. From March 2008 through October 2010, 74-131 samples were collected at each of the three streams including baseflow (every 2 weeks) and high flow (storm events sampled every two hours). Samples were analyzed for Hg_D, Hg_P, DOC, UV absorbance at 254nm and pH. Across all sites there was a significant positive correlation between Hg_D and DOC and the linear best fit lines were statistically similar. Relationships improved between Hg_D and UV absorbance indicating that within a site the optical qualities associated with DOC (typically assumed to reflect aromaticity) also impact the amount of Hg transported. Overall, SUVA₂₅₄ was significantly lower in the more acidic system, indicating less aromatic DOC as expected, yet the amount of Hg_D transported per unit DOC was the same at all sites. As a result, the more acidic stream transported more Hg_D per unit UV absorbance, indicating that watershed characteristics other than DOC quantity and quality are influencing export. Based on direct evidence from laboratory studies and indirect field based evidence, we believe the sandy soils are less competitive than high surface-area clay soils for Hg binding and therefore a greater fraction of the Hg becomes associated with organics which are subsequently transported downstream at the more acidic, sand dominated site. This research illustrates that in addition to DOC quantity and quality, physical soil characteristics may also play an important role in Hg transport in catchment stream systems.

Mercury (Hg) ecosystem cycling and the associated biogeochemical relationships are affected by land use. However, studies concerning developed ecosystems are limited with previous research focused on non-urban, remote ecosystems. Developed areas experience increased atmospheric loading but also experience additional loading from impervious surface runoff, sewer overflows and wastewater effluent. Additionally, alterations in biogeochemical carbon cycling and increases in road salt application can impact Hg mobilization and flux. We investigated the effects of land use and combined sewer overflows on watershed Hg loading and association with organic carbon in the Park River watershed (Hartford, CT). Samples were collected from a series of sites along the river to target and separate influences from impervious cover runoff and CSOs as well as from three sites within the combined sewer system (CSS). Event samples were collected across the hydrograph targeting base, rise, peak and fall. Under elevated flow conditions, total Hg (THg) and methyl Hg (MeHg) concentrations in the river samples varied from 0.5-14.9 ng/L and 0.04-0.29 ng/L, respectively. While largely associated with particulates, the dissolved THg increased to 1.4 ng/L during the hydrograph peak at the suburban site with peak dissolved THg concentration decreasing downstream. In the CSS samples, THg concentrations reached 897 ng/L with a maximum dissolved THg of 23 ng/L. Dissolved THg in the CSS samples increased with DOC. However, SUVA decreased as DOC increased, suggesting a shift towards less aromatic organic carbon. A similar relationship was observed for baseflow samples, although these were much lower in Hg species and DOC. We are currently characterizing the DOC characteristics and potential for Hg species binding for the wastewater, impervious runoff and forested inputs to relate mobilization with source. Additionally, sorption experiments are being conducted with purified DOC to assess potential differences in Hg binding.

In the Brazilian Amazon, studies have shown that recent deforestation contributes to soil mercury (Hg) release trough terrigenous fluxes of humus and soil materials to the aquatic environment. As mercury contamination has become a problem to human health trough fish consumption, it is of prior importance to elucidate the dynamics of this diffuse pollution. Terrestrial organic matter (TOM) is known as an important transportation vector of soil Hg. The aim of this study was to investigate the movements of organic matter and its associated Hg from the watershed to aquatic systems. The natural dynamics of drainage basins are being modified by local small-scale farmers using slash-and-burn practices to convert forest into agricultural lands, pastures and fallows. Lignin biomarkers were used to trace back the respective inputs of TOM from each land use to the water system of the Tapajos River (state of Pará). Samples of vegetation (leaves and bark) were collected in order to establish the specific lignin signature of the different land covers (agrarian soils, pasture, fallows and forest). Six sediment cores from the floodplain lakes and 33 soil cores from the region were retrieved and analyzed for Hg and lignin biomarkers. In complement, suspended particulate matter was collected with tangential flow filtration in 4 lakes of the watershed and analyzed in a similar way. Our results show a link between TOM fluxes and sedimentary Hg concentrations. A general increase of THg concentrations in most recent sediment was noted, with maximum values ranging up to 310 ng/g in the most deforested areas. Establishing the dynamic portrait of Hg movements in the newly colonized watersheds of the Amazon could lead to the development of conservation measures adapted to this environment.

The spatial variability of methyl mercury production in freshwater-influenced bays of the Great Salt Lake was examined via collection of sediment and water samples from multiple transects at the north and south ends of Farmington Bay in summer and fall 2009, and the north end of Ogden Bay in summer 2010. Subsamples were spiked with ²⁰⁴Hg²⁺ to examine net production of methyl ²⁰⁴Hg over time. Mercury concentration and isotopic signals for the incubated sub-samples were detected using cold vapor atomic fluorescence spectrometry in line with inductively couple plasma mass spectrometry. Concentrations of methyl ²⁰⁴Hg and ²⁰²Hg were used to develop first order methylation and demethylation rate constants (k_meth, k_demeth) for each site. Methylation was not significant in water samples, but was significant in sediment samples, and showed spatial variation that did not correspond to methyl or total Hg concentrations, or water column salinity or dissolved oxygen. However, a positive correlation with sediment organic matter content was observed for all three transects that showed significant methylation, suggesting that methylation in the wetlands of Great Salt Lake is controlled by sediment organic carbon content.

Dissolved organic carbon (DOC) concentrations in south-western Nova Scotia streams, sampled at weekly to biweekly intervals, were weel synchronized but varied spatially and temporally from <5 to 40 mgL^-1 (average=2.2mgL^-1), being highest midsummer to fall and lowest during winter to spring. The DOC concentrations in upland soil leachates near the streams ranged from 0.2 to 680 mgL^-1 (average=2.2mgL^-1), with highest values after spring thaw. In comparison, DOC concentrations in shallow-well water flowing towards the streams were generally low (average=2.2mgL^-1). A 3-parameter model (DOC-3) was proposed to project weekly to biweekly stream DOC concentrations from daily soil temperature and moisture estimations, year-round. In this model, basin-specific stream-water DOC is in part inferred from the wet-area percentage per basin, while the water that that would seep through the upland soils and underlying bedrock towards the streams would be clear. The parameters of this model refer to (i) a basin-specific DOC release parameter “k_DOC”, related to the wet-area percentage per basin, (ii) the lag time “τ” between DOC production and subsequent stream DOC emergence, related to total stream length or catchment area above the stream sampling location; and (iii) the activation energy “Ea”, to deal with the temperature effect on DOC production. This model was calibrated with the weekly to biweekly DOC data from 3 streams (Pine Marten, Moosepit Brook, and the Mersey River sampled at or near Kejimkujik National Park) and was extended to project the DOC concentrations and fluxes within the 8 streams and lake of the Pockwock-Bowater Watershed Project near Halifax, Nova Scotia. The best-fitted model calculations captured the stream-water DOC concentrations at R²=0.6, 0.7, and 0.8 at the weekly, monthly, and annual scales, respectively. Within Pockwock Lake, DOC concentrations remained synchronized with the stream-water DOC along the shore, but dropped to a fairly constant value away from the shore regardless of lake depth and season.