Mercury emissions

Globally, approximately 6600 metric tons of Hg are emitted to the atmosphere annually, with 33% to 36% attributed to direct anthropogenic emissions. The remainder originates from natural sources or from past anthropogenic emissions that are rereleased (Mason and Sheu 2002). These values suggest that about two-thirds of atmospheric Hg emissions are derived from either direct or reemitted anthropogenic sources. Coal-fired power plants are the largest single category of Hg emissions, with 1450 metric tons per year, comprising about 50% of anthropogenic sources (Pacyna et al. 2003).

Total anthropogenic Hg emissions from all sources in the United States are calculated to be 103 metric tons per year, with the Northeast contributing about 4.7 metric tons per year (USEPA 1999). Mercury emissions in the United States have declined markedly over the past decade (table 1) as a result of federal regulations that mandated large reductions in Hg emissions in medical waste incinerators and in municipal incinerators (USEPA 2005). Unlike incinerator emissions, emissions from electric utilities have remained largely unchanged, and their relative contribution to total US emissions has increased from 25% to 40%. Municipal waste incinerators (23%) and electric utilities (16%) are the largest point-source categories in the Northeast.

Table 1.

Mercury (Hg) emissions (in metric tons per year), by source category, in the United States from 1990 through 2002 and in the Northeast region in 2002.

Mercury is emitted to the atmosphere from point sources in three forms: elemental Hg (Hg⁰), gaseous ionic Hg (reactive gaseous mercury, or RGM), and particulate Hg (PHg). This speciation exerts significant control over the fate of atmospheric Hg emissions and varies widely among sources (table 2). Therefore, Hg can be a local, regional, or global pollutant, depending on the speciation of the emissions and the associated residence times in the atmosphere (Dastoor and Larocque 2004).

Table 2.

Percentage of mercury species emitted, by source category.

In 1999, 57% of calculated point-source Hg emissions in the Northeast occurred as Hg⁰, 33% as RGM, and 10% as PHg (USEPA 1999). Studies indicate that emissions from coal combustion in the United States are roughly 50% Hg⁰, 40% RGM, and 10% PHg (Pacyna et al. 2003). However, emissions from coal combustion in the northeastern states have a higher percentage of RGM (68%) and a lower percentage of Hg⁰ (30%) and PHg (2%; NESCAUM 2005). The actual Hg emission speciation profile for a specific power plant depends on the type of coal used and the air pollution control technology employed (NESCAUM 2003).

Elemental Hg, which is relatively unreactive and generally slowly oxidized, constitutes by far the largest pool of Hg in the atmosphere because of its relatively long residence time (0.5 to 2 years) and long-range transport potential (tens of thousands of kilometers). However, under some conditions Hg⁰ can be rapidly oxidized and deposited locally or regionally, as observations have shown in the Arctic and Antarctic (Lindberg et al. 2002), at the marine and continental boundary layer, and in areas downwind of urban areas (Weiss-Penzias et al. 2003). Elemental Hg can also be directly deposited to forested ecosystems through stomatal gas exchange (Grigal 2002). As a result, the atmospheric lifetime of Hg⁰ is probably closer to 0.5 year than to 2 years.

Reactive gaseous Hg consists predominantly of gaseous chloride and oxide forms of ionic Hg. This species is highly soluble in water and readily deposits to surfaces within tens to a few hundreds of kilometers from emission sources. Because of RGM's short atmospheric residence time (0.5 to 2 days), elevated Hg deposition can occur near RGM emission sources.

The atmospheric residence time of PHg is also relatively short (0.5 to 3 days). Although the fraction of PHg in ambient air in remote areas is generally less than 5% of total atmospheric Hg (Horvat 1996), concentrations may be higher near Hg emission sources and under certain atmospheric conditions (Lu et al. 2001).

Atmospheric deposition

Atmospheric deposition of Hg occurs in two forms: wet deposition (the deposition of Hg associated with rain and snow) and dry deposition (the deposition of PHg and RGM, cloud and fog deposition, and stomatal uptake of Hg⁰). Although some areas have been contaminated by land disposal of Hg or discharge of Hg in wastewater effluent, the predominant input of Hg to most watersheds is atmospheric deposition. Fitzgerald and colleagues (1998) systematically rule out alternate hypotheses, such as natural weathering, as a significant cause of the observed widespread Hg contamination.

Judging from global models (Hudson et al. 1995), reconstructions of mass balances (Mason et al. 1994), and paleo-limnological techniques (Engstrom and Swain 1997), it appears that deposition of Hg has increased two- to threefold over the past two centuries, following increases in Hg emissions associated with industrialization and Hg use. Paleo-limnological studies in the Northeast typically show Hg deposition starting to increase in the late 1800s or early 1900s and increasing 2.5- to 15-fold by the late 20th century (1970s to 1990s) (figure 1; Kamman and Engstrom 2002). Decreases in sediment Hg deposition in the Northeast (approximately 25%) have been evident in recent years, coincident with reductions in US emissions and with static global emissions. Because inventories of Hg emissions have been limited, it is not clear what is responsible for the declines in Hg deposition over the past few decades. However, it seems likely that controls on particulate matter and sulfur dioxide from electric utilities, and reductions in consumer and industrial Hg use, are important factors (Engstrom and Swain 1997).

Figure 1.

Changes in historical deposition of mercury (Hg) to sediments in (a) Spring Lake and (b) Wallingford Pond, Vermont, from 1820 to the present (after Kamman and Engstrom 2002). The sediment patterns reflect changes in Hg emissions and deposition over time.

In the eastern United States, Hg deposition is high (USEPA 1997), but it is difficult to identify its specific sources. Of the estimated 52 metric tons of Hg deposited per year in the United States from US sources, 24 metric tons (46%) are likely to originate from domestic utility coal boilers (half of the 48 metric tons of Hg that the coal-fired utilities emit each year is likely to be deposited within the United States; USEPA 1997). Likewise, for regions of New York it is estimated that 11% to 21% of the Hg deposited is derived from emissions within New York, 25% to 49% originates from other US sources, and 13% to 19% originates from Asia (Seigneur et al. 2003). Given that most coal-fired utilities emit 50% to 70% of Hg as RGM and PHg (table 2), local sources are most likely an important component of the deposition in areas within 50 km of these sources. An analysis of emissions and deposition in southern New Hampshire shows a local region of high deposition associated with local electric utility emissions (Evers et al. 2007).

In the United States and Canada, measurements of wet Hg deposition, which are largely made through the Mercury Deposition Network (MDN), show that wet Hg deposition is highest in the Southeast (e.g., Florida, Mississippi) and lowest in the West. There are currently seven MDN sites in the Northeast, with average annual wet deposition ranging from 3.8 to 12.6 μg per m² per year ( http://nadp.sws.uiuc.edu/mdn/). There do not appear to be broad spatial patterns in wet Hg deposition across the region, but the network is sparse. Because of the placement of collectors in rural areas, the deposition values for the region do not include elevated deposition that would be expected near Hg sources and in urban areas.

Estimates of dry Hg deposition are highly uncertain because of the complex interrelationships of atmospheric conditions, collection surface characteristics and terrain, and chemical properties of the contaminants. Several modeling efforts have been used to estimate dry deposition of Hg, however. In regions of New York, estimated dry Hg deposition was 4 to 10 μg per m² per year (Seigneur et al. 2003). Another model estimate specifically for the Northeast suggests that dry deposition of RGM plus Hg⁰ was 37 μg per m² per year (Xu et al. 2000). Both studies indicate that dry deposition provides a significant pathway of Hg inputs (50% to 75% of total deposition) and agree with USEPA predictions that Hg dry deposition in the Northeast is the highest in the country, in part as a result of the abundant forests whose canopies effectively collect Hg from the atmosphere.

Because of the large surface area associated with canopy foliage, atmospheric deposition of contaminants is elevated in forests compared with other types of ecosystems. Forest studies have indicated that total atmospheric Hg deposition may be estimated using fluxes of throughfall (precipitation that passes through the canopy) plus litterfall (plant material that falls to the forest floor; Rea et al. 2001). Grigal (2002) suggests that the ratio of Hg fluxes resulting from wet deposition, throughfall, and litterfall, respectively, is 1.0 to 1.8 to 2.2. So for the 5 μg per m² per year of wet deposition that might be typical of the Northeast, anticipated through-fall would be 9 μg per m² per year, and litterfall would be 11 μg per m² per year, resulting in total Hg deposition of 20 μg per m² per year and dry deposition of 15 μg per m² per year (75% of total).

Some portion of the Hg deposited to Earth's surface is reemitted to the atmosphere. However, rates of volatilization vary widely in association with differences in vegetation, soil moisture, temperature, solar radiation, and landscape characteristics. In general, volatilization rates from soil are high immediately after inputs of ionic Hg to the soil (Schluter et al. 1995). On the basis of a review of the literature, Grigal (2002) estimated a mean rate of Hg⁰ volatilization from soil of approximately 11 μg per m² per hour. This rate is more than adequate to reemit most of the atmospheric Hg deposition. The magnitude and uncertainty of this process demonstrate the acute need for additional research on Hg reemissions.

Mercury transport and fate in upland forest ecosystems

Although there have been few direct studies of soil sequestration of Hg, immobilization of Hg in forest soil is known to correspond with the retention of organic carbon (Schwesig et al. 1999). Pools of Hg in upland soil in northern temperate regions are about 7 mg per m², although higher levels have been reported in central Europe (Grigal 2003).

The export of Hg by waters draining upland soils to surface waters is generally low. Concentrations and fluxes of Hg in soil waters, as in soil, are closely related to dissolved organic carbon (DOC; Schwesig et al. 1999). In northern forests, concentrations of total Hg are highest in waters draining the upper soil, coinciding with high concentrations of DOC. Concentrations and fluxes of total Hg decrease as DOC is immobilized with depth in mineral soil (Grigal 2002).

Limited studies suggest that MeHg concentrations in upland soils and groundwaters are generally low, although higher concentrations occur in upper soil waters and decrease with soil depth (Grigal 2002). Low concentrations and fluxes of MeHg in drainage waters suggest that rates of methylation are low, and freely draining upland soils are generally not important in the supply of MeHg to downstream surface waters, with the possible exception of recently harvested forests (Porvari et al. 2003).

Transport and transformation of mercury in wetlands

Wetlands are important features of the landscape that influence the supply of different Hg species to adjacent surface waters. Wetlands are typically net sinks of total Hg and sources of MeHg (Grigal 2002, 2003). Rates of total Hg accumulation are greater in wetlands than in upland soils because of the strong association of Hg with organic matter (Grigal 2003). Annual rates of MeHg production in wetlands are approximately 0.1 to 1 μg per m² per year (Galloway and Branfireun 2004). The factors controlling methylation of Hg in wetlands are not completely understood, but they most likely involve the amounts and types of organic matter, hydrologic flow paths, and rates of microbial activity (Galloway and Branfireun 2004). Wetlands are also a major source of DOC. Organic matter produced in wetlands forms complexes with both ionic Hg and MeHg, enhancing the transport of these Hg species to surface waters but decreasing their bioavailability (Hudson et al. 1994). An elevated supply of DOC to downstream surface water could also stimulate methylation and limit photodegradation of MeHg and photoreduction of ionic Hg. Furthermore, wetlands support sulfate-reducing bacteria, which appear to be largely responsible for Hg methylation (Benoit et al. 2003). Concentrations of MeHg in wetland porewaters (waters filling the spaces between solid material in sedimentary deposits) and surface waters vary seasonally, with the highest concentrations evident during the late summer, presumably as a result of warmer temperatures, higher rates of microbial activity, and longer hydraulic residence times (Galloway and Branfireun 2004).

Mercury concentrations and transformations in surface waters

Freshwater ecosystems are among the most sensitive to Hg pollution. Total Hg concentrations in surface waters in the Northeast vary by more than an order of magnitude, from less than 0.5 to 12.7 nanograms per liter (5th to 95th percentile; figure 2; Dennis et al. 2005). Most of the Hg in surface water occurs as ionic Hg, with MeHg ranging from 1% to 35% of total Hg (figure 3). Under conditions of high total Hg loading, MeHg production can vary widely, depending on the methylation efficiency of a particular ecosystem (Krabbenhoft et al. 1999).

Figure 2.

Average water mercury (Hg) concentrations within 18-minute grid cells for lakes and streams across northeastern North America. Inset shows the distribution of Hg concentrations comprising the mean for each quintile.

Figure 3.

Box and whisker plots of mercury (Hg) concentrations in water and aquatic biota in eastern North America. Also shown are the ranges for the percentage of total Hg occurring as methylmercury (MeHg). All values were obtained from NERC (Northeastern Ecosystem Research Cooperative) data and represent wet weight, except those for phytoplankton, which were obtained from Watras and colleagues (1998).

Mercury enters remote surface waters through direct atmospheric deposition and through soil water, wetland, or groundwater drainage. Streams and rivers can exhibit marked temporal variation in Hg concentrations, which is associated with variations in concentrations of DOC or suspended matter. Large increases in Hg concentrations can occur during high flow events (Shanley et al. 2005).

Some inputs of Hg to lakes are removed from the water column by the volatilization of Hg⁰ and by sediment deposition. In freshwater lakes, photochemical processes are largely responsible for the reduction of ionic Hg to Hg⁰ (Amyot et al. 1997). Microbial reduction has been observed in laboratory studies, but only at higher than ambient concentrations of Hg (Morel et al. 1998). Biogeochemical processes in lakes also result in net production of MeHg due to methylation in anoxic sediments and in the water column.

The geographic distribution of average surface water Hg concentrations in the Northeast (figure 2) shows landscape-level heterogeneity in lake and river Hg concentrations, and areas where concentrations are elevated across several contiguous 18-minute grid cells. Areas of elevated Hg concentrations in surface waters can be explained by high concentrations of DOC, as in the Adirondacks; by high inputs of suspended solids, from rivers along Lake Champlain, related to high flow events; and by elevated atmospheric Hg deposition, as in lakes in southeastern New Hampshire and eastern Massachusetts. A large portion of the variation in total Hg and MeHg across the region can be explained by variation in DOC (Dennis et al. 2005). Areas with the highest mean surface water Hg concentrations also have the greatest range in Hg concentrations (figure 2). This variation may be attributed to heterogeneity in watershed characteristics or to high flow events (Shanley et al. 2005).

Other factors controlling mercury dynamics in surface waters

Other factors, such as water chemistry, land cover and land use, and watershed disturbances, alter the transport, transformation, and bioavailability of Hg in surface waters.

The Northeast receives elevated loading of acidic deposition as well as Hg deposition, and contains a relatively large number of acidified surface waters. Acidic deposition and the associated sulfur alter the acid–base status of surface waters, thereby influencing Hg transformation and accumulation in fish. Sulfur transformations are closely coupled with Hg dynamics. The solubility of Hg increases with increasing sulfide concentrations in anoxic waters through complexation reactions, potentially increasing the pool of Hg available for methylation (Benoit et al. 2003). Experimental observations show that when sulfate is added to wetlands or lakes, sulfate reduction is enhanced, leading to increased methylation and MeHg export (Branfireun et al. 1999, Watras et al. 2006).

Widespread observations show an inverse relationship between fish Hg concentrations and surface water pH (e.g., Kamman et al. 2004). Hrabik and Watras (2002) used reference data and observations from a lake experimentally acidified with sulfuric acid to examine the relative contribution of atmospheric Hg deposition and acidic deposition to Hg concentrations in fish. They found that half of the decrease in fish Hg over a six-year period during which the lake was recovering from acidification could be attributed to decreases in sulfuric acid loading.

In a study of 21 river basins nationwide, watersheds with mixed agriculture and forest land cover had the highest methylation efficiency, even where these watersheds had low total Hg in sediments (Krabbenhoft et al. 1999). Some waters draining largely agricultural lands have relatively high concentrations of total Hg and MeHg, but lower concentrations in fish, presumably due to algal “bloom dilution” associated with high phosphorus loading (Kamman et al. 2004; see below) or elevated DOC concentrations (which could stimulate methylation but limit bioaccumulation), or both.

Land disturbance influences Hg export and availability for methylation. Forest harvesting has been shown to increase export of total Hg and MeHg (Porvari et al. 2003). Fire results in a complex pattern of Hg loss from watersheds. During and shortly after fire, elevated Hg losses are associated with volatilization and drainage losses (Grigal 2002). Over the longer term, Hg transport to surface waters is reduced in burned areas as a result of decreases in soil carbon and DOC concentrations.

In reservoirs, rates of Hg methylation can be altered by water level fluctuation associated with hydropower production or flood control. Many large bodies of water in the Northeast are impounded to increase their storage or daily peaking capacity, and these water bodies may fluctuate tens of centimeters on a daily basis or several meters over the course of a summer. As the littoral zone experiences periodic wetting and drying, varying cycles of reduction and oxidation may enhance the production of MeHg, depending on a variety of factors (Sorensen et al. 2005, Evers et al. 2007).

Trophic transfer of mercury in surface waters of the Northeast

Concentrations of total Hg or MeHg in surface waters often do not correlate well with the Hg content of freshwater biota, such as fish. There are many physical, chemical, ecological, and land-use factors controlling the trophic transfer of MeHg, which are key to predicting MeHg concentrations in fish and other freshwater organisms.

Trophic transfer of Hg in freshwater food webs begins with the bioaccumulation of ionic Hg and MeHg by primary producers. Bioaccumulation factors in the transfer of Hg from water to algae are by far higher (approximately 10⁵ to 10⁶) than at subsequent trophic levels (figure 3). Although both ionic Hg and MeHg are taken up by aquatic organisms, MeHg is assimilated four times more efficiently than ionic Hg (Mason et al. 1994). However, the absolute and relative assimilation efficiencies of ionic Hg and MeHg vary with trophic level, uptake pathway, and water chemistry conditions. Freshwater grazers and predators acquire MeHg mainly from their food rather than from water (Harris and Bodaly 1998). Methylmercury is efficiently transferred to the higher levels of the food web and largely incorporated within proteins, as in muscle tissue.

The NERC data show that MeHg increases in concentration and comprises a greater percentage of the total Hg in freshwater consumers and predators as it progresses up the food chain (figure 3). Thus organisms consuming prey at higher trophic levels are exposed to higher concentrations of total Hg and MeHg (Vander Zanden and Rasmussen 1996). Fish Hg occurs almost entirely as MeHg.

A variety of physical, chemical, and biological factors influence the biomagnification of MeHg. Fish Hg concentrations tend to vary positively with lake or watershed area and negatively with pH, acid neutralizing capacity (ANC), nutrient concentrations, zooplankton density, and human land use (Chen et al. 2005). Furthermore, the Hg added to the lake surface each year appears to be more available for conversion to MeHg than Hg that has been in the ecosystem for longer periods (Gilmour et al. 2003).

Both experimental and field studies show that nutrient enrichment diminishes Hg bioaccumulation in phytoplankton through the biodilution of Hg under algal bloom conditions (Pickhardt et al. 2002). Mercury concentrations in zooplankton also decrease with increasing zooplankton densities that in turn are correlated with lower Hg concentrations in fish (Chen and Folt 2005). Growth dilution in fish, also under conditions of high productivity and food availability, may be related to lower Hg concentrations in fish (Essington and Houser 2003).

Within given fish populations, Hg burdens increase with the age and size of individuals in part because of the slower rates of elimination and longer exposure in larger individuals, and in part because of the consumption of higher-trophic-level foods by older and larger individuals (Wiener and Spry 1996). Mercury concentrations in top predator fish are higher in food webs with longer chain lengths and less omnivory (Stemberger and Chen 1998).

Indicators of mercury sensitivity

Four simple and common measures of water quality—DOC, ANC, pH, and total phosphorus—have been shown by Chen and colleagues (2005) and many others to be related to fish Hg concentrations. To develop indicators of Hg sensitivity, we combined data from two stratified, random-probability surveys of northeastern lakes (USEPA EMAP [Environmental Monitoring and Assessment Program], Northeast Lakes Program, 1991–1994, and Vermont–New Hampshire REMAP [Regional EMAP], 1998–2000) with the survey data sets of Chen and colleagues (2005) to examine these four water-chemistry characteristics in lakes with standard-age yellow perch (Perca flavescens) whose tissue contained mean concentrations of Hg above and below the USEPA criterion (0.3 μg per g; figure 4). The standard age for yellow perch examined in this analysis was 4.6 years (Kamman et al. 2004). This analysis showed that lakes with Hg levels above 0.3 μg per g in yellow perch had significantly higher DOC (t = −3.099, p = 0.003) and lower pH (t = −6.282, p < 0.001), ANC ( t = 2.835, p = 0.007), and total phosphorus (t = 3.840, p < 0.001) than lakes with fish Hg concentrations below 0.3 μg per g. As yellow perch have low to moderate Hg concentrations, these thresholds are conservative and help identify the most sensitive lakes.

Figure 4.

Relationship between methylmercury (MeHg) concentrations in standard-length yellow perch and total phosphorus concentration in lakes (a), and box and whisker plots of concentrations of dissolved organic carbon (b), pH (c), and acid neutralizing capacity (d) for lakes in the northeastern United States containing average concentrations of standard-age yellow perch with MeHg concentrations less than and greater than 0.3 μg per g.

Twenty percent of lakes in the region had total phosphorus concentrations above 30 μg per L. In those lakes, Hg concentrations in yellow perch were below 0.3 μg per g. In the remaining 80%, we found that most lakes (75%) had yellow perch Hg concentrations exceeding 0.3 μg per g when surface waters had a DOC level of more than 4.0 mg carbon per L, a pH of less than 6.0, or an ANC of less than 100 micro-equivalents (μeq) per L. These commonly monitored indicators provide natural resource managers with a useful tool for evaluating the likelihood of high fish Hg concentrations in individual lakes.