Environ. Sci. Technol. 1997, 31, 960-967
Recent Declines in Atmospheric Mercury Deposition in the Upper Midwest D A N I E L R . E N G S T R O M * ,† A N D EDWARD B. SWAIN‡ St. Croix Watershed Research Station, Science Museum of Minnesota, Marine on St. Croix, Minnesota 55047, and Minnesota Pollution Control Agency, 520 Lafayette Road, St. Paul, Minnesota 55155
Historic increases in atmospheric mercury deposition caused by anthropogenic emissions have been well documented from sediment cores from lakes and peatlands in North America and Europe. Few previous studies have addressed the question of whether mercury deposition has increased continuously to the present or whether it has declined in recent decades. We present stratigraphic data from a suite of Minnesota lakes that indicate mercury deposition peaked in the 1960s and 1970s, at least for the upper Midwest. Recent declines, which appear in both rural and urban lakes, are not evident in sediment cores from remote coastal lakes in southeastern Alaska. Because the Alaskan sites provide an integrated sample of mercury pollution in the Northern Hemisphere, these results imply that global mercury emissions have not abated and that decreased inputs to Midwestern lakes are caused by reduced emissions from regional sources. U.S. inventory data suggest that decreased emissions likely resulted from reduced industrial use of mercury, use of pollutioncontrol technologies that incidentally capture mercury, a shift from coal to natural gas for commercial and residential heating, and a decrease in uncontrolled waste incineration. Increased stack height and other factors that favor long-distance transport could be partially responsible for the trend.
FIGURE 1. Location of study lakes in eastern Minnesota (circles), western Minnesota (squares), Minneapolis, and Glacier Bay National Park in southeastern Alaska (arrow). is similar to the global average and appears regionally uniform, again implying broad-scale continental to global emission sources. However, most previous investigations focused on the magnitude of the change but paid little attention to the question of whether Hg deposition has increased continuously to the present or whether, for some geographic areas, it has declined in recent decades. Here we evaluate recent changes in atmospheric mercury deposition for three regions of the United States over which different mercury emission sources should dominate. Trends in atmospheric deposition are inferred from Hg accumulation rates in 210Pb-dated sediment cores from eight lakes in rural Minnesota, four urban lakes in Minneapolis, MN, and three remote wilderness lakes on the Gulf Coast of southeastern Alaska (Figure 1). By comparing mercury accumulation trends among these different areas, we are able to separate deposition changes occurring on local and regional scales in the upper Midwest from those driven by mercury emission trends on a global scale.
Methods Introduction Modern measurements of atmospheric mercury (Hg) concentration and deposition together with historical records from lake sediments and peat indicate that the global reservoir of atmospheric mercury has increased by a factor of 2-5 since the beginning of the industrialized period (1-8). Because mercury vapor (Hg0) has a long atmospheric residence time (9) and because mercury contamination of lacustrine food webs appears geographically pervasive (1016), mercury pollution is often viewed as a global problem that may defy state or national abatement efforts. This view is reinforced by most historic accounts of mercury deposition that assume a monotonic rise in mercury contamination that parallels the measured increase in global atmospheric concentrations (6, 17). In non-urban areas of the upper Midwest, lake-sediment records indicate that atmospheric Hg deposition has increased by a factor of 3-4 since pre-industrial times (6). The increase * Corresponding author e-mail:
[email protected]. † Science Museum of Minnesota. ‡ Minnesota Pollution Control Agency.
960
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 31, NO. 4, 1997
Sediment cores about 1 m long were retrieved by piston corer from the profundal region of each lake and sectioned in the field. Six of the eight rural Minnesota lakes were previously studied for whole-basin Hg loading using multiple sediment cores (6, 7). Data for two of the Alaskan cores are from work by Brigham (18). Lead-210 was measured by R-spectrometry methods (19), and dates and sedimentation rates were determined according to the constant rate of supply (crs) model (20, 21). Twelve to 20 depth intervals in each core were analyzed, providing a mean dating resolution of 2-8 years for the 1980s, 8-20 years for most of the 20th Century, and 15-40 years for the 1800s. Dating methods are described in detail by Engstrom et al. (7). Mercury concentrations in most cores were measured by cold vapor atomic absorption spectrometry (CVAAS) following strong-acid digestion and reduction with SnCl2 (22); dual amalgamation/cold vapor atomic fluorescence spectrometry (CVAFS) methods were used for sediments from lakes Le Homme Dieu, Carlos, and LaPerouse (23). Mercury analyses were performed at the same (or greater) stratigraphic resolution as the 210Pb dating. Reagent blanks and standard reference materials (SRMs) were run with each set of Hg analyses. Mean reproducibility of duplicate extractions for
S0013-936X(96)00089-2 CCC: $14.00
1997 American Chemical Society
TABLE 1. Geographic and Chemical Characteristics of Lakes Investigated and Comparison of Modern, Peak, and Pre-industrial Rates of Hg Accumulation Lakea
Hg flux ratioc mean surface alkalinity sediment latitude (N) longitude (W) depth (m) area (ha) (µequiv L-1) accumulationb (mm year-1) modern peak
Dunnigan Kjostad Meander Thrush
47° 42' 48° 07' 48° 08' 47° 54'
91° 38' 92° 36' 92° 09' 90° 30'
2.3 6.5 4.8 6.9
Eastern Minnesota 32.9 66 167.7 180 39.6 70 6.6 60
2.8 2.3 2.5 1.5
3.40 2.99 3.34 3.38
Cedar Mountain Carlos Le Homme Dieu
47° 04' 45° 32' 45° 57' 45° 56'
95° 10' 95° 32' 95° 22' 95° 21'
3.8 2.7 13.4 6.1
Western Minnesota 39.1 2600 15.7 4200 1020.0 3500 766.0 3500
4.5 3.6 2.0 3.3
3.57 6.10 4.58 6.69
Calhoun Cedar Harriet Wirth
44° 57' 44° 56' 44° 59' 44° 55'
93° 19' 93° 19' 93° 19' 93° 18'
10.7 6.1 8.8 3.7
Minneapolis 170.5 1940 68.9 2020 143.0 1860 15.0 2000
5.2 7.0 6.0 9.2
8.79 4.32 9.03 3.92
Dagelet Brady LaPerouse
58° 31' 58° 19' 58° 31'
137° 20' 136° 41' 137° 18'
Southeastern Alaska 2.1 20 2.6 31 2.5 397
0.9 0.9 2.5
1.82 2.10 2.05
2.42 3.04 3.53
depth of peakd (cm)
3.99 3.67 4.89 5.18
5 5 5 3
15.40 8.51 12.54 5.27
11 15 15 15
a Names of Alaskan lakes are provisional. b Mean sediment accumulation rate from 1890 to 1990. c Ratio of modern and peak Hg accumulation rates to pre-industrial Hg accumulation. d Depth below sediment surface of peak Hg accumulation.
the CVAFS methods was 8.6%, and that for the CVAAS analysis of the four urban lakes was 5.4%; SRMs were consistently within certified range. Data quality for the remaining sites (six rural Minnesota lakes and two Alaskan lakes) are described elsewhere (7, 18). To compare historic trends in atmospheric Hg deposition among sites, we calculated for each core the average Hg flux at decadal intervals from the present to 1910 and at 20-year intervals from 1910 to 1850; a single average flux was determined for 1850-1800. Hg fluxes were harmonized in this manner because the dating intervals obtained by 210Pb were different for each core; the averaged intervals are consistent with the temporal resolution of Hg analysis and 210Pb dating. We also calculated the Hg flux for each interval as a percentage of the modern (post-1980) Hg flux. These flux ratios represent unitless measures of changing Hg loading that are broadly comparable among sites and geographic regions. They are independent of individual site conditions that affect Hg sedimentation (sediment focusing) or absolute rates of atmospheric Hg deposition (e.g., rainfall) (24).
Results and Discussion Rural Minnesota Lakes. The eight rural lakes in our study fall into two distinct geographic arrays (Table 1, Figure 1): an eastern group in the Superior National Forest of northern Minnesota (Thrush, Dunnigan, Meander, and Kjostad) and a western group in west-central Minnesota (Cedar, Mountain, Le Homme Dieu, and Carlos). The eastern sites are relatively small headwater lakes that lie in mixed deciduous-conifer forest. They are fed principally by groundwater seepage and surface precipitation and are of low alkalinity. The western group consists of two small shallow lakes and two large deep basins, all of high alkalinity. Catchment vegetation ranges from conifer forest (Cedar) to deciduous woodland (Le Homme Dieu and Carlos) to native tall-grass prairie (Mountain). The watersheds of Le Homme Dieu and Carlos support numerous lakeshore homes and summer cabins along with limited agricultural development. None of the other six watersheds has been appreciably disturbed by land-use changes. The sediment records suggest that atmospheric Hg deposition to rural lakes in both regions (east and west) rose above natural background at about 1850, increased sharply
between 1920 and 1960, and then declined, but only in the East (Figure 2). In the eastern lakes (Dunnigan, Kjostad, Meander, and Thrush), peak deposition rates of 1.2-1.5 × modern are reached during the 1960s and 1970s with clear declines to the present (Figure 3). Hg profiles from the western group suggest that Hg deposition either leveled off after 1960 (Cedar and Carlos) or increased continuously to the present (Mountain and Le Homme Dieu). None of the western lakes shows a decline; all of the eastern lakes do. Hg accumulation rates during the last decade are elevated above pre-industrial rates by a factor of 3.0-3.4 in the east and 3.6-6.7 in the west (Table 1). Peak Hg accumulation rates (1960-1980) in the eastern lakes are 4.0-5.2 times higher than pre-industrial rates. Although sedimentation rates in the eastern lakes (1.52.8 mm/yr) are somewhat lower than those of the western group (2.0-4.5 mm/yr), the Hg declines are clearly defined by 2-3 stratigraphic levels over a sediment thickness of 3-5 cm. We are somewhat puzzled by the monotonic increase in Hg accumulation in the western lakes to a modern level that is substantially higher than that in the eastern group. The most attractive explanation is a hypothesis put forward by Sorensen et al. (25), who found higher average concentrations of Hg in precipitation at two monitoring sites near the western border of Minnesota as compared to five sites to the east. Partly because all sites showed high correlations between Hg and Ca and Mg concentrations (r ) 0.62-0.87), Sorensen et al. hypothesized that the higher concentrations at the western sites may be the result of washout of soil-derived particulates associated with the agricultural activity that dominates the region’s land use. Wind-blown soil may thus account for the generally higher modern accumulation rates in our western lakes, a loading that would not immediately reflect a decrease in Hg emissions and may have obscured such a decrease. The monotonic increase in Hg accumulation in the western lakes may also reflect a geographic location upwind of Hg emission sources to the east; that is, the Hg sources that contributed to the deposition decline in eastern Minnesota have never been a significant component of the Hg flux to the western part of the state. Western Minnesota is at the low end of a Hg contamination gradient that has been documented across the Great Lakes states (26). In addition, northeastern Minnesota probably received deposition from
VOL. 31, NO. 4, 1997 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
961
FIGURE 2. Hg accumulation rates for individual strata (points) averaged at decadal intervals (bars) in 210Pb-dated sediment cores from lakes in rural and urban Minnesota and southeastern Alaska. Differences in magnitude of Hg flux within lake groups reflect core-specific differences in sediment focusing and differences in Hg contributions from the watershed (see text). Although temporal trends in Hg accumulation are broadly a variety of regional sources that had minimal impact on similar within lake groups, absolute rates vary among sites western Minnesota, including non-ferrous smelting in Onby a factor of 2 or more for any given time interval (Figure tario, taconite processing on Minnesota’s iron range, a steel 2). However, these site-to-site differences in Hg flux do not mill in Duluth, and volatilization of mercuric fungicides from reflect actual differences in atmospheric Hg deposition, but paper mills.
962
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 31, NO. 4, 1997
1960s and 1970s followed by a sharp decline during the 1980s (Figure 3). Peak Hg accumulation rates in each core are 5-15 times higher than background (pre-1850) accumulation rates and as much as 1 order of magnitude greater than those from rural Minnesota (Figure 2); modern (1980-1993) rates are 3.9-9.0 times greater than background (Table 1). The nearsurface declines are clearly defined by 3-4 stratigraphic levels with decreasing Hg concentrations over a sediment thickness of 11-15 cm. The rise in Hg accumulation above background rates is similar in timing to that found in rural Minnesota lakes, but the magnitude of the increase from background to peak rates is greater in the urban sites. The larger urban increase may reflect higher atmospheric Hg loading in Minneapolis. Alternatively, a greater portion of the atmospheric Hg deposited on urban watersheds may be exported to lakes because of hydrological alterations, including wetland destruction, pavement of streets, and construction of storm sewers. Although the Minneapolis-St. Paul metropolitan area has never supported large industrial Hg emitters (e.g., smelters, chlor-alkali plants), it is possible that the combined emissions from light industry, power plants, waste incineration, and residential heating were an additional load in the urban Hg flux. Although the century-long rise in Hg accumulation in the Minneapolis lakes may have been accentuated by watershed land-use changes, the declining Hg flux of the last decade appears to be largely an atmospheric signal that cannot be attributed to a sudden reversal of watershed disturbance. Few land-use changes have taken place in recent decades that could substantially reduce Hg export to the city lakes, and related efforts to reduce nutrient and particulate loading in runoff have not substantially changed sedimentation rates or trophic conditions (27). The fact that all four lakes show a mercury decline of similar timing and magnitude argues strongly for an urban-wide change in atmospheric Hg loading that was independent of local watershed control. FIGURE 3. Average Hg accumulation rates at decadal intervals as a percent of modern (1980-1993) Hg accumulation rates in sediment cores from lakes in rural and urban Minnesota and southeastern Alaska. rather (i) lake-to-lake differences in watershed inputs of Hg and (ii) core-to-core differences in Hg sedimentation (6, 7). Because lakes receive a portion (roughly 20-25%) of the atmospheric Hg deposited to their watersheds, lakes with larger catchments (relative to lake area) have higher total Hg loading. Mercury concentrations and sedimentation rates are also spatially variable within a lake basin, and mercury accumulation rates at a single core site cannot be automatically extrapolated to the entire lake bottom. Atmospheric deposition rates for mercury can be determined from sediment studies, but such calculations usually require that many cores be analyzed from a single lake (24). Urban Minnesota Lakes. The four Minneapolis lakes (Harriet, Calhoun, Cedar, and Wirth) are located in established residential neighborhoods in the western part of the city. Their watersheds support a narrow buffer of urban parkland, with the remaining landscape covered with single-family homes, apartments, and light commercial development. Urbanization, which began in the late 1800s, lowered water quality and increased sedimentation in the lakes, although conditions have been relatively stable since about 1970 (27). Three of the four lakes are of moderate size and depth (Wirth is small and shallow), and all lakes are moderately alkaline (Table 1). Profiles of Hg accumulation for the Minneapolis lakes show a continuous rise in Hg flux from background levels prior to about 1850 to peak values (1.3-2.0 × modern) during the
Alaskan Lakes. Three lakes from southeastern Alaska were included in this study to provide a global reference against which Midwestern trends in atmospheric Hg deposition might be compared. The study sites are located about 2 km from the Gulf of Alaska in Glacier Bay National Park (58-59° N latitude). They are relatively small headwater lakes of low alkalinity in watersheds of undisturbed temperate rainforest (Table 1). There are no significant anthropogenic Hg sources anywhere in southeastern Alaska and very little human development within several hundred kilometers of these lakes. The nearest potential volcanic sources for Hg (Lake Clark and Katmai National Parks) are >800 km to the northwest. Because these sites sample air directly off the Gulf of Alaska, Hg accumulation in the sediments can be viewed as an integrated sample of global Hg pollution in the Northern Hemisphere and, thus, a component of the Hg deposition experienced at our Midwestern sites. Asian sources of anthropogenic Hg (China, Siberia) are so distant from southeastern Alaska (8000 and 3000 km, respectively) that they probably contribute only to the background hemispheric Hg pollution reaching western North America. Sediment records show that Hg accumulation in the Alaskan lakes has increased by a factor of 1.8-2.1 since preindustrial times and that the increase has been continuous to the present (Figures 2 and 3). The data indicate that atmospheric Hg deposition rose above background in the mid-1800s, increased more gradually than in the Midwest, but did not decline in recent decades. Hg profiles from lakes in northern Sweden, Finland, and Canada also show a modern Hg flux 2-3 times background with no evidence of recent declines (2, 3, 28, 29). The similarity of these trends from remote sites in the Northern Hemisphere implies that global
VOL. 31, NO. 4, 1997 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
963
Hg emissions have not abated and that the recent declines observed in Midwestern lakes are of regional origin. Regional Comparisons. The regional component of Midwestern Hg deposition can be estimated by subtracting the global deposition increase as documented in southeastern Alaska from the increase in Hg deposition found in rural Minnesota. According to calculations by Swain et al. (6), atmospheric Hg deposition in the rural Midwest has increased from 3.7 µg m-2 year-1 in 1850 to 12.5 µg m-2 year-1 in 1990, a factor of 3.4. Assuming that average Hg deposition in the Northern Hemisphere doubled during the same time period (based on the Alaskan data), global Hg emissions from natural and anthropogenic sources currently account for 7.4 µg m-2 of annual Hg deposition in rural Minnesota. The remaining deposition, 5.1 µg m-2 year-1sabout 40% of the total (natural plus anthropogenic)smust derive principally from anthropogenic Hg sources in the Midwest and eastern United States. These calculations assume that mineral sources of Hg in watershed soils are negligible for all regions and that Hg sediment profiles provide an accurate assessment of changing atmospheric deposition. We estimated previously that erosion of mineral soils contributed no more than 12% of the pre-industrial Hg flux to our Minnesota lakes (7). Similar calculations based on the Hg content of soil parent material from our Alaskan watersheds indicate that local geologic sources account for less than 8% of the pre-industrial Hg flux to these lakes. The Hg concentrations in the inorganic silt + clay at the base of the Alaskan cores is 2.7-4.9 ppb. These glaciofluvial deposits represent an integrated sample of the unweathered parent soil in the local watersheds. Assuming a similar Hg content for the mineral matter in the overlying lake sediments, erosion of mineral soils contributed at most 3-8% of the pre-industrial sedimentary Hg flux and 1.5-4% of the modern Hg flux to these sites. The recent declines in Hg accumulation in the Midwestern lakes cannot be explained by local land-use changes, nor is it likely that bioturbation or Hg diffusion has played a role in generating these trends. Surface enrichment of Hg in some sediment cores has been attributed to upward diffusion of Hg along Fe-rich redox gradients (30, 31). However, such mechanisms cannot explain our subsurface Hg peaks, as most occur well below the depth of active redox cycling. Calculations by Hurley et al. (32) and Gobeil and Cossa (33) indicate that the diffusive flux of Hg is too slow to significantly modify historic trends in Hg accumulation where sedimentation rates are comparable to those observed here. Moreover, laboratory studies with incubated cores from one of our sites show that Hg peaks once emplaced do not diffuse, nor are they generated in homogenized sediments by redox cycling at the sedimentwater interface (34). Well-preserved Hg records with subsurface peaks have also been found in sediment cores from lacustrine and riverine systems where point-source Hg inputs have been curtailed (35-37). It also seems highly improbable that temporally concordant patterns of Hg accumulation could be generated in numerous sediment profiles of varying thickness (Table 1) by post-depositional sedimentary processes. Additional evidence for recent declines in atmospheric Hg deposition in eastern Minnesota comes from an ombrotrophic peatland near Duluth (8). Here average Hg accumulation rates in five peat cores declined from 37.8 to 24.5 µg m-2 year-1 between 1950-1980 and 1980-1991. It is unlikely, however, that such reductions are geographically limited to Minnesota or even the upper Midwest. Multiple cores from Little Rock Lake, a remote seepage lake in northern Wisconsin, all show declines in Hg concentration in sediments deposited after about 1970 (32). A recent survey of 30 seepage lakes in northern Wisconsin and the Upper Peninsula of Michigan revealed substantial near-surface declines in sediment-Hg concentrations in more than 80% of the lakes (38). A 52% decrease in the Hg content of Canadian domestic
964
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 31, NO. 4, 1997
cigarette tobacco between 1968 and 1988 provides independent evidence for decreasing Hg contamination in the eastern Great Lakes region (39). Canadian tobacco is grown in a small geographic area of southern Ontario, and its Hg content is thought to reflect ambient Hg levels in air from this region. U.S. Industrial Hg Use and Emission Trends. The declining Hg flux recorded in our Midwestern sites may represent a general decrease in Hg emissions in the industrialized areas of eastern North America. On a global scale, manufacturing, coal combustion, waste incineration, and non-ferrous metal smelting are the dominant anthropogenic sources of atmospheric Hg (40). In the United States these emission sources have changed markedly over the last few decades, with substantial reduction in industrial Hg consumption, increased coal combustion, and centralized waste incineration. The most likely explanation for a mid-1970s decline in Hg deposition is a reduction in the industrial use of Hg and a coincident shift in coal combustion to electric utilities. Total U.S. consumption of Hg for industrial purposes rose dramatically following World War II, peaked during the 1960s, and declined thereafter to record lows in the 1990s (Figure 4A). Worldwide Hg production shows virtually the same pattern, with U.S. Hg consumption averaging 25% of global production for most of this century (Figure 4B). The overall decline in U.S. Hg consumption is consistent with the recent decrease in atmospheric Hg deposition inferred from our lake-sediment cores. However, it is difficult to ascribe trends in atmospheric deposition to any single industry because of simultaneous changes among multiple industrial uses and the uncertain fate of the Hg in many applications. U.S. industrial Hg consumption peaked between 1963 and 1969 at an average of 2,600 metric tons annually (t/year) (Figure 4A). Much of this Hg was used in electrical devices and control instruments (727 t/year) and may not have been introduced directly into the environment. The most dramatic decline in industrial consumptionsHg use in batteriesshas occurred only since 1986 (41). Other industrial applications, especially Hg fungicides, went directly into the environment. Hg fungicide use in agriculture and the paper industry peaked in 1956 at 342 t/year and declined below 100 t/year in 1970 (Figure 4A). Hg fungicides were added to paint in significant quantities between 1960 and 1990. The sum of all Hg fungicide use was above 350 t/year between 1961 and 1970, peaking at 484 t/year in 1968. Presumably much of the Hg ultimately volatilized to the atmosphere as Hg0, although little is known about this process. Hg consumption by the chlor-alkali industry peaked at 525 t/year between 1967 and 1976 and declined to 295 t/year in the 1980s and to 190 t/year between 1990 and 1994 (Figure 4A). Part of this decline can be attributed to decommissioning of Hg-based chlor-alkali plants and the balance to increased efficiency of chlorine production per quantity of Hg consumed (42). Decreased Hg consumption has probably led to decreased emissions, although releases to air and water reported by the industry are a small fraction of that consumed. Reported consumption of Hg for chlorine production was over 200 t/year as recently as 1992, whereas the industry reported releases to air and water of about 6 t/year (43) (this discrepancy was pointed out to the authors by R. U. Ayres, unpublished material). The difference is unaccounted for, although it seems likely that Hg-bearing sludges were being accumulated on-site. Waste Hg stored or land-filled as solidified sludges can be a significant secondary source of atmospheric Hg emissions from chlor-alkali facilities (44). Potential Hg emissions from coal combustion peaked in the United States during three time periods: 1916-1929, 1940-1947, and from 1980 to the present (Figure 4C). Coal now consumed in the United States averages about 0.1 mg of Hg/kg of coal (45), although the Hg content of coal consumed in the past may have averaged as high as 0.2 mg/
FIGURE 4. (A) Trends in industrial consumption of Hg in the United States, 1910-1993 (41, 48); detailed information is not available prior to 1941; graph stacked in sequence shown in legend. Although the data depict trends in industrial use, they are not literally consumption, but rather reflect purchases of mercury for the various categories of use as reported voluntarily to the U.S. Government. (B) Trends in world production of Hg (t/year) and U.S. industrial consumption as percent of world production (41, 48). (C) Trends in U.S. coal consumption (million t/year) and potential Hg emissions (t/year) calculated assuming 0.1 mg of Hg/kg of coal (41, 49, 50). (D) Emissions trends in SO2 and particulate matter (million t/year) from U.S. electric utilities (51). (E) Emissions trends in SO2 and particulate matter (million t/year) from U.S. non-ferrous smelting (51). kg, due to less coal cleaning and a higher proportion of eastern coals, which tend to be higher in Hg. Coal consumption has also shifted to electric utilities, which consumed only 10% of U.S. coal in 1940 but 81% in 1980. Because pollution-control devices on power plants have become increasingly effective, particulate emissions have declined by 60% since 1960, and SO2 emissions have remained constant since 1970, despite a 2.5-fold increase in coal consumption by electric utilities. Some of the control in SO2 emissions was accomplished by increasing use of western low-sulfur coal (Figure 4D). The control devices for SO2 and particulates incidentally capture Hg, with efficiencies ranging from near zero to over 90% (45). In 1990, coal-burning electric utilities emitted about 40 t of Hg (45) out of potential emissions of 70 t (Figure 4C), yielding an average control efficiency of 40%. It is thought that the Hg that is captured is mainly HgII and that emitted
is mainly Hg0. The latter is more slowly removed from the atmosphere and therefore transported farther from sources before deposition (40). The net result of the shift in coal combustion from industry, residential heating, and railroads to electric utilities is to capture the Hg forms more likely to be deposited near the combustion source and to emit the balance of the Hg from tall stacks that encourage long-distance transport. Peak emission of Hg from coal combustion (particularly Hg likely to deposit near emission sources) was probably not in the high coal-consumption years since 1980, but rather during earlier periods of high coal use when the Hg content of coal was higher and there was little incidental Hg removal by particulate and SO2 controls. Air-pollution regulations in early 1970s gradually eliminated uncontrolled Hg emission sources that were common in urban areas. Emission of particulate matter in Minneapolis
VOL. 31, NO. 4, 1997 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
965
was progressively regulated and reduced, beginning with City Ordinances in the 1960s, state-wide regulations in Minnesota in 1969 and 1976, and the Federal Clean Air Act of 1970 and its Amendments of 1977. At the same time, the heating of homes and commercial buildings changed from coal to mainly natural gas. Uncontrolled burning was restricted statewide after 1972, and solid waste was largely diverted to landfills from open dumps (that sometimes burned) and incinerators lacking secondary combustion (residential, commercial, and municipal). Similar changes took place nationwide. By 1992, 190 municipal waste-combustion plants were operating in the United States, burning about 17% of the solid waste (46). Twelve plants are currently operating in Minnesota, all built between 1982 and 1989. Whether these changes increased or decreased Hg deposition probably depends on locality. The elimination of numerous small incinerators and coal burners (with low stack height and no particulate controls) should have reduced local Hg deposition in urban areas like Minneapolis, while recent construction of large municipal waste-combustion plants would tend to oppose that trend, especially at sites more distant from urban centers. Smelting of lead, copper, and zinc ores incidentally emits Hg to the atmosphere in annual quantities estimated at 100 t globally (4% of total) and about 9 t in the United States (3% of total) (47). Smelting of non-ferrous metals peaked in the 1970s (41). A modest (15-25%) decline in smelting combined with progressive application of pollution control equipment has reduced SO2 emissions by 80% and particulate emissions by 85% as compared to 1970 rates (Figure 4E). This pollution control effort captures an unknown proportion of the Hg that had been emitted to the atmosphere. There are several non-ferrous smelting facilities located in the Lake Superior region of Ontario (north and east of the eastern study lakes) but none within Minnesota itself. We conclude that the regional component of atmospheric Hg deposition in the upper Midwest has declined substantially in the last two decades. The decrease is most closely correlated with (i) a reduction in the industrial use of Hg; (ii) the increasing use at coal-burning utilities and metal smelters of technologies that incidentally reduce Hg emissions, including cleaner fuels and particulate and SO2 controls; (iii) a shift from coal to natural gas and oil for industrial and residential heating; and (iv) a decrease in uncontrolled waste incineration. Alternatively, factors favoring long-distance transportsa smaller proportion of HgII in emissions, increased stack height, a decrease in the size of emitted particulates, or a decrease in atmospheric oxidants such as ozonescould be at least partially responsible for the trend. At the same time, the global reservoir of atmospheric Hg has continued to increases0.6% annually according to recent estimates (1)sand Hg deposition in more remote parts of North America has not abated. Global anthropogenic emissions, which by our calculations currently account for about 30% of the Hg entering Midwestern lakes, may soon surpass regional anthropogenic contributions (presently 40% of total loading) if current trends continue.
Acknowledgments We thank M. Brigham for Hg data from two of the Alaskan lakes and N. Bloom for Hg analysis of three other cores used in this study. M. Keating, E. Nater, and D. Porcella provided helpful discussions and reviews of the manuscript. Support for the various stratigraphic studies compiled in this paper was provided by the Legislative Commission on Minnesota Resources, the Minnesota Pollution Control Agency, the University of Minnesota Water Resources Research Center, the Douglas County (MN) Office of Land and Resource Management, and the National Science Foundation (BSR8705371).
966
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 31, NO. 4, 1997
Literature Cited (1) Mason, R. P.; Fitzgerald, W. F.; Morel, F. M. M. Geochim. Cosmochim. Acta 1994, 58, 3191. (2) Johansson, K. Verh. Int. Ver. Theor. Angew. Limnol. 1985, 22, 2359. (3) Verta, M.; Mannio, J.; Iivonen, P.; Hirvi, J.-P.; Ja¨rvinen, O.; Piepponen, S. In Acidification in Finland; Kauppi, P., Kentta¨mies, K., Anttila, P., Ed.s; Springer-Verlag: Heidelberg, 1990; p 883. (4) Rada, R. G.; Wiener, J. G.; Winfrey, M. R.; Powell, D. E. Arch. Environ. Contam. Toxicol. 1989, 18, 175. (5) Steinnes, E.; Andersson, E. M. Water Air Soil Pollut. 1991, 56, 391. (6) Swain, E. B.; Engstrom, D. R.; Brigham, M. E.; Henning, T. A.; Brezonik, P. L. Science 1992, 257, 784. (7) Engstrom, D. R.; Swain, E. B.; Henning, T. A.; Brigham, M. E.; Brezonik, P. L. In Environmental Chemistry of Lakes and Reservoirs; Baker, L. A., Ed.; American Chemical Society: Washington, DC, 1994; p 33. (8) Benoit, J. M.; Fitzgerald, W. F.; Damman, A. W. H. In Mercury Pollution: Integration and Synthesis; Watras, C. J., Huckabee, J. W., Eds.; Lewis Publishers: Boca Raton, FL, 1994; p 187. (9) Fitzgerald, W. F. In The Role of Air-Sea Exchange in Geochemical Cycling; Baut-Menard, P., Ed.; D. Reidel: Boston, 1986; p 363. (10) Swain, E. B.; Helwig, D. D. J. Minn. Acad. Sci. 1989, 55, 103. (11) Lathrop, R. C.; Rasmussen, P. W.; Knauer, D. R. Water Air Soil Pollut. 1991, 56, 295. (12) Wren, C. D.; MacCrimmon, H. R.; Loescher, B. R. Water Air Soil Pollut. 1983, 19, 277. (13) Verta, M. Mercury in Finnish Forest Lakes and Reservoirs: Anthropogenic Contribution to the Load and Accumulation in Fish; National Board of Waters and the Environment (Finland): Helsinki, 1990. (14) Lambou, V. W.; Barkay, T.; Braman, R. S.; Delfino, J. J.; Jansen, J. J.; Nimmo, D.; Parks, J. W.; Porcella, D. B.; Rudd, J.; Schultz, D.; Stober, J.; Watras, C.; Wiener, J. G.; Gill, G.; Huckabee, J.; Rood, B. Mercury Technical Committee Interim Report; Center for Biomedical and Toxicological Research and Waste Management, Florida State University: Tallahassee, 1991. (15) Lindqvist, O.; Johansson, K.; Aastrup, M.; Andersson, A.; Bringmark, L.; Hovsenius, G.; Håkanson, L.; Iverfeldt, Å.; Meili, M.; Timm, B. Water Air Soil Pollut. 1991, 55, 1. (16) Cunningham, P. A.; Smith, S. L.; Tippett, J. P.; Greene, A. Fisheries 1994, 19, 14. (17) Slemr, F.; Langer, E. Nature 1992, 355, 434. (18) Brigham, M. E. M.S. Thesis, University of Minnesota, Minneapolis, 1992. (19) Eakins, J. D.; Morrison, R. T. Int. J. Appl. Radiat. Isot. 1978, 29, 531. (20) Appleby, P. G.; Oldfield, F. Catena 1978, 5, 1. (21) Binford, M. W. J. Paleolimnol. 1990, 3, 253. (22) Environment-Canada. Mercury: Methods of Sampling, Preservation, and Analysis; Economic and Technical Review Report EPS 3-EC-81-4; Environment-Canada: Ottawa, 1981. (23) Bloom, N. S.; Crecelius, E. A. Mar. Chem. 1983, 14, 49. (24) Protocol for Estimating Historic Atmospheric Mercury Deposition EPRI/TR-106768; Electric Power Research Institute: Palo Alto, CA, 1996. (25) Sorensen, J. A.; Glass, G. E.; Schmidt, K. W. Environ. Sci. Technol. 1994, 28, 2025. (26) Nater, E. A.; Grigal, D. F. Nature 1992, 358, 139. (27) Minneapolis Chain of Lakes Clean Water Partnership Project, Phase IsDiagnostic Report; Minneapolis Park and Recreation Board: 1993. (28) Hermanson, M. H. Water Sci. Technol. 1993, 28, 33. (29) Lucotte, M.; Mucci, A.; Hillaire-Marcel, C.; Pichet, P.; Grondin, A. Water Air Soil Pollut. 1995, 80, 467. (30) Strunk, J. L. M.S. Thesis, Michigan State University, East Lansing, 1991. (31) Rasmussen, P. E. Environ. Sci. Technol. 1994, 28, 2233. (32) Hurley, J. P.; Krabbenhoft, D. P.; Babiarz, C. L.; Andren, A. W. In Environmental Chemistry of Lakes and Reservoirs; Baker, L. A., Ed.; American Chemical Society: Washington, DC, 1994; p 425. (33) Gobeil, C.; Cossa, D. Can. J. Fish. Aquat. Sci. 1993, 50, 1794. (34) Henning, T. A. M.S. Thesis, University of Minnesota, Minneapolis, 1989. (35) Syers, J. K.; Iskandar, I. K.; Keeney, D. R. Water Air Soil Pollut. 1973, 2, 105. (36) Smith, J. N.; Loring, D. H. Environ. Sci. Technol. 1981, 15, 944.
(37) Breteler, R. J.; Bowen, V. T.; Schneider, D. L.; Henderson, R. Environ. Sci. Technol. 1984, 18, 404. (38) Schumaker, R. J.; Brezonik, P. L. Abstracts of Papers; Conference on Mercury Pollution in the Upper Great Lakes Region, Minneapolis, MN, 1995. (39) Rickert, W. S.; Kaiserman, M. J. Environ. Sci. Technol. 1994, 28, 924. (40) Mercury Atmospheric Processes: A Synthesis Report EPRI/TR104214; Electric Power Research Institute: Palo Alto, CA, 1994. (41) Minerals Yearbook; U.S. Bureau of Mines: Washington DC, 19321991. (42) North American Chlor-alkali Industry Plants and Production Data Book; Chlorine Institute: Washington, DC, 1995. (43) 1992 Toxics Release Inventory; EPA-745/R-94-001; U.S. EPA: Washington, DC, 1994. (44) Lindberg, S. E.; Turner, R. R. Nature 1977, 286, 133. (45) Electric Utility Trace Substances Synthesis Report, Volume 3: Mercury in the Environment; Electric Power Research Institute: Palo Alto, CA, 1994.
(46) Kaiser, J. V. L. Waste Age 1992, 23, 26. (47) Nriagu, J. O.; Pacyna, J. M. Nature 1988, 333, 134. (48) Mineral Industry Surveys: Mercury in 1993; U.S. Bureau of Mines: Washington, DC, 1994. (49) Annual Energy Review 1991; Energy Information Administration, U.S. DOE: Washington, DC, 1992. (50) Quarterly Coal Report; Energy Information Administration, U.S. DOE: Washington, DC, 1994. (51) National Air Pollutant Emission Trends, 1900-1992; EPA-454/ R-93-032; U.S. EPA: Washington, DC, 1993.
Received for review January 29, 1996. Revised manuscript received November 15, 1996. Accepted November 18, 1996.X ES9600892 X
Abstract published in Advance ACS Abstracts, February 1, 1997.
VOL. 31, NO. 4, 1997 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
967