Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-atmospheric mercury sources

STOTEN-19338; No of Pages 13 Science of the Total Environment xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-atmospheric mercury sources Joseph Domagalski a,⁎, Michael S. Majewski a, Charles N. Alpers a, Chris S. Eckley b, Collin A. Eagles-Smith c, Liam Schenk d, Susan Wherry e a

U.S. Geological Survey, California Water Science Center, 6000 J Street, Placer Hall, Sacramento, CA 95819, United States U.S. Environmental Protection Agency, Office of Environmental Assessment, EPA-Region 10, 1200 6th Ave., Suite 900, Seattle, WA 98101, United States c U.S. Geological Survey, Forest and Rangeland Ecosystem Science Center, 3200 SW Jefferson Way, Corvallis, OR 97331, United States d U.S. Geological Survey, Oregon Water Science Center, 2795 Anderson Ave., Suite 106, Klamath Falls, OR 97603, United States e U.S. Geological Survey, Oregon Water Science Center, 2130 SW 5th Ave., Portland, OR 97201, United States b

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Watersheds have geologic, anthropogenic, and global atmospheric sources of Hg. • This study investigated sources of stream Hg loads in the Western U.S. and Canadian-Alaskan Arctic. • Abandoned mines increased annual stream load relative to deposition. • Watersheds in urban areas had N Hg loads/unit area than vegetated areas. • River loads are attenuated in watersheds with forests and other natural vegetation.

a r t i c l e

i n f o

Article history: Received 25 October 2015 Received in revised form 15 February 2016 Accepted 16 February 2016 Available online xxxx Keywords: Western North American Mercury Synthesis WNAMS Urban contaminants Stream loads historical mining

a b s t r a c t Annual stream loads of mercury (Hg) and inputs of wet and dry atmospheric Hg deposition to the landscape were investigated in watersheds of the Western United States and the Canadian-Alaskan Arctic. Mercury concentration and discharge data from flow gauging stations were used to compute annual mass loads with regression models. Measured wet and modeled dry deposition were compared to annual stream loads to compute ratios of Hg stream load to total Hg atmospheric deposition. Watershed land uses or cover included mining, undeveloped, urbanized, and mixed. Of 27 watersheds that were investigated, 15 had some degree of mining, either of Hg or precious metals (gold or silver), where Hg was used in the amalgamation process. Stream loads in excess of annual Hg atmospheric deposition (ratio N 1) were observed in watersheds containing Hg mines and in relatively small and medium-sized watersheds with gold or silver mines, however, larger watersheds containing gold or silver mines, some of which also contain large dams that trap sediment, were sometimes associated with lower load ratios (b0.2). In the non-Arctic regions, watersheds with natural vegetation tended to have low ratios of stream load to Hg deposition (b0.1), whereas urbanized areas had higher ratios (0.34–1.0) because of impervious surfaces. This indicated that, in ecosystems with natural vegetation, Hg is retained in the soil and may be transported subsequently to streams as a result of erosion or in association with dissolved organic carbon. Arctic watersheds (Mackenzie and Yukon Rivers) had a relatively elevated ratio of stream load to atmospheric deposition (0.27 and

⁎ Corresponding author. E-mail address: [email protected] (J. Domagalski).

http://dx.doi.org/10.1016/j.scitotenv.2016.02.112 0048-9697/Published by Elsevier B.V.

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112

2

J. Domagalski et al. / Science of the Total Environment xxx (2016) xxx–xxx

0.74), possibly because of melting glaciers or permafrost releasing previously stored Hg to the streams. Overall, our research highlights the important role of watershed characteristics in determining whether a landscape is a net source of Hg or a net sink of atmospheric Hg. Published by Elsevier B.V.

1. Introduction Mercury (Hg) is a globally distributed contaminant in aquatic systems that can impact various species through bioaccumulation (Selin, 2009) with potential human health effects, especially from fish consumption (Ullrich et al., 2001; Selin, 2009). Within the United States, Hg is currently the seventh most common contaminant listed under the federal Clean Water Act as impairing rivers and streams, and is the most prevalent chemical for which a related management plan, known as a Total Maximum Daily Load (TMDL), is either underway or planned (http://iaspub.epa.gov/waters10/attains_nation_cy.control?p_ report_type=T, accessed June 11, 2015). TMDLs for Hg are designed to reduce levels in top trophic level fish. Understanding the sources or pathways, and forms of Hg in water bodies is critical for a successful TMDL. Major sources or pathways of Hg to the aquatic environment include atmospheric deposition, erosion from geologic material, urban discharges, agricultural and industrial discharges, mining locations, and combustion (Wang et al., 2004; Pirrone et al., 2009). In Eastern North America, it is known that atmospheric deposition of Hg from power plant emissions is a major pathway to the landscape and streams (Wentz et al., 2014). The predominant form of atmospheric Hg (N 95%) is gaseous elemental (Hg(0)) (Schroeder and Munthe, 1998; Fitzgerald, 1986). Gaseous oxidized Hg, and particulate bound Hg are also present, but at lower concentrations. Hg(0) has an atmospheric half-life of 6 months to one year (Schroeder and Munthe, 1998) whereas gaseous oxidized Hg and particle bound Hg have much shorter half-lives, on the order of days, and are rapidly deposited. The primary removal mechanism for Hg(0) is by oxidation to the gaseous oxidized form followed by deposition, either wet or dry. Gaseous oxidized Hg can also be reduced back to elemental Hg allowing re-entry to the atmosphere. These processes also occur at Hg mine and amalgamation sites. Evasion, wind erosion, and runoff are potential release mechanisms for Hg from these areas to streams. Anthropogenic uses of Hg include industrial processes for chemical manufacturing, fluorescent light bulbs, and dental fillings, but these are in the process of being phased out. Within Western North America, one of the largest sources of Hg to the environment was the historical processing or recovery of gold and silver by amalgamation (Nriagu, 1994; Hylander and Meili, 2003). Although largely eliminated in North America, amalgamation is still used by artisanal miners in many developing countries (Telmer and Veiga, 2009). Amalgamation was used historically in California from the mid to late 19th century to the mid 20th century (Alpers et al., 2005). It has been estimated that about 4,535,000 kg of Hg were lost to the environment from placer mines and an additional 1,360,000 kg from hard rock mining, within California, as a result of its use in amalgamation (Churchill, 2000). Previous studies, such as Brigham et al. (2009), have compared atmospheric Hg deposition to Hg transport in streams to provide insight on watershed processes that affect Hg transport. Their study included locations in the Western, Mid-Western, and Eastern U.S., and estimated that 3–44% of the atmospherically deposited Hg was exported from the watersheds on an annual basis. In this study, we compare annual riverine loads of total Hg (THg) at selected river locations in the Western U.S. and the Canadian-Alaskan Arctic to annual estimates of wet and dry deposition over the upstream watersheds. Land use in these watersheds ranges from largely pristine to having variable levels of anthropogenic disturbance including urbanization, agriculture, and mining. Climatic conditions for the selected watersheds also varied, especially with respect to rainfall. Comparisons of this sort provide a first step at

quantifying the relative importance of atmospheric and nonatmospheric Hg sources or pathways to stream loads, as well as a way to estimate the amount of Hg stored in watersheds. Stream flow typically transports considerably less THg from the watershed than is deposited by wet and dry depositional processes (Shanley et al., 2008; Brigham et al., 2009; Journey et al., 2012). Runoff contributions of the atmospherically deposited Hg to river loads, however, are variable and difficult to quantify. The amount of Hg transported to streams and rivers by runoff attributed to atmospheric deposition is dependent upon the annual rainfall amount, the amount of Hg retained by the soil and vegetation, and the density of impervious surfaces within the watershed (Shanley et al., 2008; Brigham et al., 2009; Journey et al., 2012; Tsai and Hoenicke, 2001). Watersheds of Western North America have a variety of Hg pathways because of variable geographic location, geology, and land uses. Atmospheric deposition originates from a combination of local sources and trans-ocean atmospheric transport from Asia (Strode et al., 2008). Western North America also has a number of mining districts for precious metals (gold and silver), base metals, and Hg (Brobst and Pratt, 1973). Mercury deposits in the Western U.S. are found primarily in California, Nevada, and Oregon associated with either altered mafic volcanic rocks or with epithermal hydrothermal systems (Gustin et al., 2000; Rytuba, 2003). Gold and silver deposits occur throughout the Western U.S. (Long et al., 1998), and are large potential sources of Hg to water, because of discharges from historical mining that used the Hg amalgamation process for metal recovery. A study of this nature has a considerable uncertainty, some of which is difficult to quantify. Measurements of Hg in wet deposition are available, but limited. The Mercury Deposition Network (MDN) (http:// nadp.sws.uiuc.edu/mdn/, accessed June 17, 2015), maintains more than 100 sites throughout the continental U.S., Alaska, and small portions of southern Canada for the purpose of collecting and measuring precipitation samples for Hg concentrations and fluxes. The majority of the sites, however, are east of the Rocky Mountain States. Current monitoring methods are not able to provide spatially comprehensive information on the wet and dry deposition of Hg, so modeling has become an important tool to estimate atmospheric deposition of Hg over wide areas (Gbor et al., 2007; Ryaboshapko et al., 2007; Lin et al., 2006). Model estimates of Hg dry deposition range from b 1 to 3 times the rate of wet deposition (St. Louis et al., 2001; Evers and Clair, 2005; Seigneur et al., 2004). These results suggest that atmospheric dry deposition of Hg can be more significant than previously thought (Risch et al., 2012; Zhang et al., 2012), but there is no national sampling network and fewer direct measurements have been made. This study relied on only one year of modeled dry deposition. We recognize that year-to-year variability of dry deposition almost certainly occurs, therefore, the dry deposition estimate is a limitation that needs to be pointed out. In addition, local sources of potentially high Hg emissions can contribute to stream loads and may not be captured by the Hg deposition network or the dry deposition model. Uncertainty of stream loads can be calculated using statistical techniques, and the uncertainty usually decreases with longer term monitoring, especially monitoring that takes into account variation in stream flow throughout the year. Climatic effects must also be considered. Some locations in this study had only a few years of water sample collection for Hg analysis, whereas others had up to 20 years. If sample collection took place during drought or during years of above average rainfall, stream load estimates, although accurate, would not fully explain the longer-term transport processes. Effects of climate can be better understood for sites with longer periods of record, but are difficult to

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112

J. Domagalski et al. / Science of the Total Environment xxx (2016) xxx–xxx

understand for shorter time frames. In addition, many sampling programs target sites downstream of potential sources of contamination. In this study, a total of 27 watersheds had sufficient data to calculate annual loads of THg at stream flow gauging sites. Of those 27, 15 were mining influenced to some degree, two were mostly urban, and 10 were either in pristine regions or areas with low levels of development. Watershed sizes, within the continental U.S. ranged from small (b 1 km2) to large (69,455 km2). The two Arctic watersheds were large (852,257 and 1,289,036 km2). The sites also span a range of climatic conditions from relatively wet to dry to Arctic. In addition, some of the mining locations were nested within larger watersheds, where it was possible to calculate THg loads, thereby allowing for an examination of the effect of watershed area on transport. 2. Study area, materials & methods 2.1. Atmospheric deposition and study area description Spatial maps of Hg wet deposition for the non-Arctic locations were obtained from the NADP (http://nadp.sws.uiuc.edu/mdn/, accessed January 14, 2015) and uploaded to a geographic information system (ArcGIS). The Hg deposition network (MDN) utilizes measurements of Hg in precipitation at a number of locations and then uses precipitation from the PRISM (http://www.prism.oregonstate.edu) database to calculate estimated deposition to the land surface. Annual data of cumulative THg wet deposition were available from 2000 to 2013. The pixel size for the map is approximately 800 × 800 m2. The spatial analyst tool of ArcGIS was used to calculate the annual amount of THg wet deposition over each delineated watershed for each year. The Community Multiscale Air Quality (CMAQ) model, a complex computer model designed to simulate a wide range of physical and chemical processes that occur at different scales in the lower atmosphere (https://www. cmascenter.org/cmaq/; Bash, 2010; Lin et al., 2006), was used to estimate dry deposition of Hg for this analysis. This model computes bidirectional flux between a surface and the atmosphere, in order to estimate net flux during a given time period using the U.S. Environmental Protection Agency's national emission inventory (http://www3.epa. gov/ttn/chief/net/2008inventory.html) coupled with the Weather Research and Forecasting meteorology model (http://www.wrf-model. org/index.php). The only available Hg dry deposition estimates are for 2009. The pixel size of the dry deposition (40 km × 40 km) was considerably coarser relative to the wet. Because of the lack of regional dry deposition data, we assumed that dry deposition amounts did not vary appreciably over the time frame of the study from 2000 to 2013, and used the 2009 model as an estimate for all years. Wet Hg deposition maps for the individual years are shown with the Supporting Information (SI) (Figs. SI1–14). Maps of dry Hg deposition produced from the CMAQ model are also shown for each month of 2009 (Figs. SI15–26). An estimate of wet and dry atmospheric Hg deposition for the Arctic watersheds was obtained from Dastoor et al., (2015), who utilized the Global/Regional Atmospheric Heavy Metals Model (GRAHM) (Dastoor and Davignon, 2009). Some limited sampling over the watersheds suggest agreement within about 25% between modeled and actual (Dastoor and Davignon, 2009). Watersheds used for the analysis range from small (b1 km2) to large (up to 1,789,036 km2) (Table 1). Selected watersheds with sufficient data to calculate THg loads are located in California, Oregon, Nevada, and Colorado, plus there is one watershed entirely in Canada (Mackenzie River) and one that originates in Northwest Territories, Canada and then flows through Alaska (Yukon River). Land uses or land cover in these watersheds ranged from mixed with mining, agriculture, forests, and urban uses, to mostly natural with one located entirely within a National Park. Detailed land use data from the 2011 National Land Cover Database (http://www.mrlc.gov) and mine density (including gold, silver, and mercury) are summarized in Table 2 and shown in Figs. SI27–30.

3

Historical Hg mines are less numerous than gold and silver mines (Alpers et al., 2005), but can be important ongoing sources of Hg contamination (Rytuba, 2003). Fifteen of the 27 watersheds have mining as one of the land uses with an additional two that are adjacent to Hg mines. Although the land area associated with mining is typically small compared to the other land uses or land covers, there may be a relatively large contribution to the THg stream load if easily erodible material contaminated with mining waste is present, especially following large precipitation events. A map with watershed locations and one year of wet deposition (2006) is shown in Fig. 1A and locations of the Arctic watersheds are shown in Fig. 1B. Some of the watersheds, such as Sagehen and Furnace Creeks, as well as the Guadalupe River, the Z4LA Drainage, the Trask sub-watersheds, and a few others, are too small to show at this scale. The Sagehen Creek watershed, located just slightly east of the Bear, Yuba, and Greenhorn watersheds, is shown in Fig. SI31. Furnace Creek is a tributary to the Coast Fork of the Willamette River, situated in the southern portion of the Willamette River watershed. Both Guadalupe River and the Z4LA watersheds are located to the southwest of the Sacramento watershed, near San Francisco Bay. The Guadalupe River watershed is shown in Fig. SI32. Four other watersheds that are too small to show on Fig. 1A are within the larger Trask River watershed in Oregon. These watersheds, Gus, Pothole, Upper Main, and Rock Creeks, are each b 1 km2 in area. Locations of those watersheds are shown in Fig. SI33. These 4 watersheds are located slightly to the west of the Willamette River watershed (Fig. 1A). In some cases, reservoirs are present downstream of the mine sites and may contribute to a reduction in THg loads because of trapping the transported suspended sediment containing THg behind the dams. For the lower Sacramento River, one of the largest watersheds under consideration, several reservoirs are present between the river sampling location and most of the upstream sites of mining-related Hg. The same is true for the American River at Sacramento site. Some of the sampled locations are downstream from extensive gold or Hg mines. The Yuba River and Bear River sites are within the Sierra Nevada Foothills Gold Province (Ashley, 2002; Alpers, 2015), which includes the famous Mother Lode of California. The Cache Creek and Guadalupe River sites in California are downstream of Hg mining locations, not all of which have been successfully or completely remediated. The Carson River site is downstream of extensive silver mines in the Comstock Lode area near Virginia City, Nevada, where Hg amalgamation was used. The Furnace Creek and Coast Fork Willamette sites in Oregon are located in a very small watershed affected by the Black Butte Mine (Hg). A few sites are located in watersheds with minimal impact from anthropogenic activities. Two of these are Big Thompson River at Moraine Park, located within Rocky Mountain National Park in Colorado, and Sagehen Creek, a small watershed located near the crest of the Sierra Nevada in California. A second site further downstream on the Big Thompson River is located in the city of Loveland, so its watershed has some urban influence. Two sites are predominantly urban: the Beaverton Creek site in Oregon and the Z4LA watershed in California, which drains into San Francisco Bay. The two Arctic watersheds are very large, with the Yukon occupying 852,257 km2 and the Mackenzie 1,789,036 km2. Both are free-flowing rivers with only minimal development. Land cover is dominated by forests, shrublands, barren land, wetlands, tundra, and permafrost (Table 2). Historical mining did occur including the Klondike gold rush (https://content.lib.washington.edu/extras/goldrush.html). Some mining still occurs, but with strict environmental standards (Brabets et al., 2000). 2.2. Methods for calculation of THg load in streams and rivers Mercury loads in streams and rivers were obtained from published literature or were calculated using concentration and stream flow data

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112

4

J. Domagalski et al. / Science of the Total Environment xxx (2016) xxx–xxx

Table 1 Locations of stream water quality and gauging sites, upstream watershed areas, period of record of water quality data, and predominant land uses within the watersheds. Site

California Sacramento River at Freeport

Years of record: water quality and discharge

Predominant land uses

69,455

1994–2013

−121.503628 −121.730477 −121.104675 −121.274608

5088 3016 806 2874

1994–2013 2010–2012 2001–2003 2001–2003

Forested, urban, agricultural, mining Forested, urban, mining Forested, agricultural, mining Forested, mining Forested, mining

Latitude, decimal degrees (NAD83)

Longitude, decimal degrees (NAD83)

38.45417

−121.4983

38.600932 38.726696 39.292115 39.235781

Upstream drainage area, km2

American River at Discovery Park Sacramento Lower Cache Creek at Road 102 near Yolo South Yuba River at Jones Bar near Grass Valley Yuba River below Englebright Dam near Smartville Guadalupe River Bear River below Wolf Creek near Lucas Hill Greenhorn Creek at You Bet Road near Nevada City Bear River near Wheatland Sagehen Creek Z4LA Drainage

37.37383 39.018507 39.1874

−121.933013 −121.173561 −120.9413

471 606 98

2003–2006, 2010 2001–2002 2000–2003

Urban, mining, forested Forested, mining Forested, mining

38.999895 39.432462 37.629630

−121.406624 −120.238001 −122.106181

756 49 4.17

2000–2002 2009 2007–2010

Forested, mining Forested Urban

Nevada Carson River near Fort Churchill

39.292996

−119.317675

1997–2013

Forested, mining

Oregon Beaverton Creek Lookout Creek Coast Fork Willamette below Cottage Grove Dam Coast Fork Willamette at London Furnace Creek near Black Butte Mine Dennis Creek near Black Butte Mine Upstream Garoutte Creek near Black Butte Mine Gus Creek Pothole Creek Upper Main Creek Rock Creek

45.520404 44.2133 43.720934 43.640578 43.5784 43.58223 43.57832 45.37469 45.37312 45.35822 45.34950

−122.898587 −122.2533 −123.050258 −123.087385 −123.072 −123.071 −123.073 −123.48271 −123.51945 −123.46779 −123.47807

2004–2005 2003–2005 2011–2014 2012–2013 2013 2014 2013 2012–2013 2012–2013 2012–2013 2012–2013

Urban Forested Forested, mining Forested, mining Mining Forest, shrubs, upstream Hg mine Forest, shrubs, upstream Hg mine Forest Forest, developed Forest, grasses Forest

Colorado Big Thompson River at Moraine Park Big Thompson River at Loveland

40.353873 40.378595

−105.584168 −105.061091

103 1377

2008–2011 2006–2011

Forested Forested, urban

Arctic locations Lower Mackenzie River Lower Yukon River

68.472711 61.933611

−135.495633 −162.883056

1,789,036 852,257

2007–2010 2001–2005

Forested, ice and snow Forests, shrubs

from publicly available databases (mostly the National Water Information System of the U.S. Geological Survey, http://nwis.waterdata.usgs. gov/nwis). Annual loads of THg were calculated using either the LOADEST model (Cohn et al., 1989; Crawford, 1996; Runkel et al., 2004) or the newly developed EGRET model (Hirsch and De Cicco, 2014). The model used for load calculation or the literature sources of the previously calculated loads are shown in the Supporting Information (Table SI1). Both LOADEST and EGRET estimate daily or annual loads by developing a model that regresses concentration against discharge, time, and seasonality (using trigonometric functions of time). The EGRET Model uses the following regression: ln ðcÞ ¼ β0 þ β1 t þ β2 ln ðQ Þ þ β3 sinð2πt Þ þ β4 cosð2πt Þ þ ε

ð1Þ

3377

98 64 277 185 0.12 2.8 38.2 0.27 0.49 0.45 0.45

results are produced when multiple years are brought into the calibration of the model. The EGRET model can also be used to assess trends in concentration or load, and includes an option to calculate flownormalized loads. Flow normalization takes into account all of the measured discharges on a given day for the 20 or more years of record and produces an average flow, which can then be used to calculate either a flow-normalized concentration or load (Hirsch and De Cicco, 2014). The flow-normalized concentration or load takes away the effect of varying discharge on trends, and is useful to determine the results of management actions, such as mine remediation. The EGRET model includes estimates of bias in the load calculation (B, flux bias statistic) using the following equations: B ¼ ðP  OÞ=P where;

In Eq. (1), c is Hg concentration, Q is river discharge, and t is decimal time. The last term in the equation, ε, is an error term (unexplained variation). A similar equation can be used in LOADEST, and in addition, several other equations can be used to obtain the best statistical fit to the model. The regression coefficients (β0 − β4) are fixed for LOADEST calculations but vary with time for EGRET calculations. The EGRET model provides reliable estimates, as indicated by low values of the model bias, discussed below, and requires at least 20 years of data, whereas LOADEST can be used on smaller time frames. LOADEST requires at least one year of data with a minimum of about 20 samples collected over the year in order to capture the range of flow conditions. Better

n

n

O ¼ ∑i¼1 Li ¼ ∑i¼1 ci Q i P¼

Xn

̂¼

L i¼1 i

Xn i¼1

kcî Q i :

ð2Þ ð3Þ ð4Þ

For Eqs. (2) through (4), O is the observed load, P is the predicted load, Li is the observed load on the ith-sampled day in kg/day, ^Li is the estimated load on the ith-sampled day in kg/day, ĉ is the estimated concentration on the ith-sampled day in ng/L, k is a unit conversion factor, ci is the measured concentration on the ith-sampled day in ng/L, Qi is the

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112

J. Domagalski et al. / Science of the Total Environment xxx (2016) xxx–xxx

5

Table 2 Land cover and mine density. Land cover for the non-Arctic locations from the National Land Cover Data Base, 2011. Land cover for the Mackenzie River watershed from Pietroniro and Soulis, 2001; Land Cover for the Yukon River watershed from Brabets et al., 2000."NA, not available" Site

Open water

Perennial ice/Snow

Developed land

Barren land

Forest

Grasses

Crops

Wetlands Mines/km2

California Sacramento River American River Cache Creek South Yuba River Yuba River below Englebright Dam Guadalupe River Bear River below Wolf Creek Greenhorn Creek Bear River near Wheatland Sagehen Creek Z4LA Drainage

1.3% 1.8% 5.8% 1.9% 1.4% 0.7% 0.8% 0.2% 1.5% 0.0% 0.0%

0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

4.4% 6.1% 5.2% 1.1% 1.2% 50.9% 9.9% 4.8% 8.3% 0.0% 100.0%

1.0% 1.2% 0.7% 0.9% 0.4% 0.0% 0.0% 0.0% 0.2% 0.0% 0.0%

44.2% 63.7% 17.8% 66.8% 75.1% 36.9% 74.4% 84.3% 66.5% 88.1% 0.0%

38.4% 26.7% 64.0% 29.0% 21.9% 11.0% 14.8% 10.6% 22.3% 11.8% 0.0%

9.6% 0.1% 5.8% 0.0% 0.0% 0.2% 0.0% 0.0% 1.1% 0.0% 0.0%

1.2% 0.4% 0.6% 0.2% 0.1% 0.3% 0.1% 0.0% 0.1% 0.1% 0.0%

0.09 0.31 0.01 0.42 0.47 0.11 0.87 0.45 0.71 0.00 0.00

Nevada Carson River

0.2%

0.0%

4.8%

1.2%

17.8%

74.0%

0.4%

1.6%

0.10

Oregon Beaverton Creek Lookout Creek Coast Fork Willamette, Cottage Grove Coast Fork Willamette, London Furnace Creek near Black Butte Mine Dennis Creek near Black Butte Mine Upstream Garoutte Creek near Black Butte Mine Gus Creek Pothole Creek Upper Main Creek Rock Creek

0.0% 0.0% 1.2% 1.2% 0.0% 0.0% 0.0%

0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

87.9% 0.0% 1.4% 1.4% 0.0% 0.0% 0.0%

0.0% 0.0% 0.4% 0.4% 0.0% 0.0% 0.0%

9.1% 93.8% 61.6% 61.6% 20.0% 20.0% 20.0%

2.0% 6.2% 34.4% 34.4% 80.0% 80.0% 80.0%

0.0% 0.0% 0.5% 0.5% 0.0% 0.0% 0.0%

0.9% 0.0% 0.5% 0.5% 0.0% 0.0% 0.0%

0.00 0.00 0.02 0.03 0.50 0.25 0.00

0.0% 0.0% 0.0% 0.0%

0.0% 0.0% 0.0% 0.0%

0.0% 30.0% 0.0% 0.0%

0.0% 0.0% 0.0% 0.0%

100.0% 70.0% 60.0% 100.0%

0.0% 0.0% 40.0% 0.0%

0.0% 0.0% 0.0% 0.0%

0.0% 0.0% 0.0% 0.0%

0.00 0.00 0.00 0.00

Colorado Big Thompson River., Moraine Big Thompson River. Loveland

0.4% 15.3% 0.6% 2.4%

0.8% 3.7%

17.5% 4.1%

44.8% 56.3%

19.3% 29.2%

0.0% 1.7%

1.9% 1.9%

0.00 0.01

Barren land/lichens

Wetlands Other

Mine density

0.4% 12.5%

NA 9.2%

0.0 0.0

Arctic locations Site name

Water

Ice/snow or Tundra

Lower Mackenzie River Lower Yukon River

7.3% 15.7% (Tundra) b1% 2.8% (Ice and Snow)

Mixed forest 59.6% 62.1%

Shrublands or grasslands 14.7% 12.7%

discharge on the ith-sampled day (in m3/s), and n is the number of sampled days. Output from the LOADEST program include a bias statistic and the Nash-Sutcliffe Index. The Nash-Sutcliffe Efficiency Index is described by Krause et al. (2005). The formula for the Nash Sutcliffe Efficiency (NSE) Index is: n X

ðOi −P i Þ2

NSE ¼ 1− i¼1 n  X

2 Oi −Ō

ð5Þ

i¼1

where Oi is an observed value, Pi is its corresponding predicted value, and Ō is the average of the observed values. Models for stream loads were chosen based on the best results for either NSE index or bias statistics. In some cases, calculated loads were used from the published literature, most of which were completed with LOADEST or a similar method that regresses concentration and stream flow from which a daily, weekly, monthly, or annual load can be derived. Those sites included Sagehen Creek (Faïn et al., 2011), the Z4LA Drainage (McKee and Gilbreath, 2015), Guadalupe River (McKee et al., 2005; McKee et al., 2010), Beaverton Creek (Brigham et al., 2009), Lookout Creek

Agricultural lands 2.2% b1.0%

0.1% b1%

(Brigham et al., 2009), the lower Yukon River (Schuster et al., 2011), and the lower Mackenzie River (Emmerton et al., 2013). 3. Results and discussion 3.1. Wet and dry deposition, river loads, ratios of stream load to deposition Wet deposition of THg within the Western U.S. for one year (2006) is shown in Fig. 1A. Although no year had close to average precipitation for all locations, 2006 is shown because it is near the middle of the period of record and most watersheds had close to average precipitation during that year. The deposition pattern reflects primarily how precipitation was distributed over the Western U.S. during that year. All years of wet THg deposition from the MDN considered in this study (2000 −2013) are shown with the Supporting Information (Figs. SI1– SI14). Throughout the Western U.S., there was a statistically significant upward trend of wet Hg deposition over time (p = 0.002, slope = 0.13). For the states that had streams with THg data (California, Oregon, Nevada, and Colorado), two had statistically significant upward trends in wet THg deposition, one had no trend, and one had a significant downward trend. The trend in wet deposition in Oregon was upward (p = 0.008, slope = 0.23) as well as in Colorado (p = 0.03, slope = 0.12). The trend in wet deposition in Nevada was not significant (p = 0.16). The trend in wet deposition in California was downward for this period (p = 0.05, slope = -0.23).

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112

6

J. Domagalski et al. / Science of the Total Environment xxx (2016) xxx–xxx

Fig. 1. A) Map of study area showing Western State boundaries, wet mercury deposition for 2006, and locations of watersheds studied. Wet deposition from the National Atmospheric Deposition Program (http://nadp.sws.uiuc.edu/mdn/); B.) Locations of Arctic watersheds: Yukon and Mackenzie Rivers.

Dry deposition in individual watersheds ranged from 20% to twice the amount of wet deposition. Dry deposition exceeded wet deposition at the Big Thompson watershed (Colorado) by a factor of 2. There was no clear pattern by location in the ratio of dry to wet deposition, and the overall average indicated that dry deposition is about 80% of the wet for the watersheds studied. In a study completed in Nevada in 2005 and 2006, it was shown that the ratio of dry to wet deposition was between 10 and 90% depending on location (Lyman et al., 2007). A summary of THg deposition, average river loads of THg, and the ratios of stream load to atmospheric deposition are shown in Table 3. Variation in the ratios of stream load to atmospheric deposition by watershed size and mine density or land use are shown in Fig. 2. Summary statistics for load calculations completed as part of this analysis are shown in the Supporting Information (Table SI1). Similar to previous comparisons of Hg atmospheric deposition and stream export (Brigham et al., 2009), the ratio of annual steam load to THg deposition varies considerably across these locations for 2000– 2013. In the study by Brigham et al. (2009), watersheds with mines were not studied, and as a result, the highest ratio of stream load to atmospheric deposition was 0.44. Any site with a ratio approaching or over 1 suggests that THg sources other than atmospheric deposition are present (Fig. 2). As the ratio increases, it is more likely that sources or pathways in addition to atmospheric deposition are present or occurring. The lowest ratio for a given year was 0.002 for Big Thompson River at Loveland (Colorado) in 2007 and the highest was 385 for Furnace Creek near Black Butte Mine (Oregon) in 2013. Furnace Creek is situated on a small watershed draining the Black Butte Mine, a Hg mine that was active from the 1890's to the 1960's (U.S. EPA, 2012), and was declared a United States Environmental Protection Agency (U.S. EPA) Superfund site in 2010 due to high levels of Hg found in soils surrounding the mine site. Furnace, Dennis, and Garoutte Creeks are tributaries to the Coast Fork Willamette River. Both Furnace and Dennis Creek drain directly through the contaminated mine site. As a result, very high THg

concentrations have been measured. The maximum concentration measured at Furnace Creek was 93,000 ng/L. The sampling site on Garoutte Creek was upstream of the mine, consistent with the ratio being less than one. Although there are no Hg mines upstream of the Garoutte Creek sampling site, the area may have been affected by historical atmospheric deposition of Hg related to Hg retorting associated with the Black Butte Mine and Hg amalgamation at other mines in the area. For example, amalgamation is known to have occurred in a nearby watershed (Ambers and Hygelund, 2001). The Coast Fork Willamette River flows into a small reservoir and the site of the Coast Fork Willamette at Cottage Grove is downstream of the reservoir. The reservoir likely traps Hg-contaminated sediment as the average stream load to atmospheric Hg deposition ratio drops to 0.4 below the Cottage Grove dam. Other sites with ratios N1 include Carson River near Fort Churchill (Nevada), the Z4LA Drainage, Greenhorn Creek at You Bet Road near Nevada City, Guadalupe River (California), and lower Cache Creek (California). The Carson River site is downstream of the Comstock Lode where silver was discovered in the mid-19th century. Gold was initially mined starting in 1849, and then silver was discovered in 1859 (Lincoln, 1923). The Carson River site is one of the most heavily contaminated with THg of all the locations considered in this study (Thodal et al., 2015). It has been estimated that between 1860 and 1900, 6,350 metric tons of Hg were released into the Carson River (Lechler et al., 1997; Van Denburgh, 1973; Smith, 1943) as a result of the amalgamation process. The Carson River was declared a Superfund site in 1990 by the U.S. EPA (Lechler et al., 1997) due to these high levels of Hg contamination. Because of the extensive Hg contamination at the Carson River and Black Butte Mine (Oregon), it is not surprising to find THg loads in excess of atmospheric deposition at those two locations. Some of the highest Hg concentrations were measured at the Carson River site. The minimum THg concentration for the Carson River site was 34 ng/L, the maximum

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112

J. Domagalski et al. / Science of the Total Environment xxx (2016) xxx–xxx

7

Table 3 Average wet and dry deposition, average deposition by area, average river load, average river yield, average ratio of load to deposition, and range of the ratios. Only total Hg atmospheric deposition is available for Arctic locations. "NA, not available". Site

California Sacramento River at Freeport American River at Discovery Park Sacramento Lower Cache Creek South Yuba River at Jones Bar near Grass Valley Yuba River below Englebright Dam near Smartville Guadalupe River

Average river load, kg

Period of record

Average wet/average dry deposition, kg

Average deposition, μg/m2

1994–2013 1994–2013 2010–2012 2001–2003 2001–2003

677/542 71/53 22/33 15/11 52 /41

17 24 18.2 32.3 32.4

117 6.8 48 2.7 2.6

4/4

17.0

8/7 1.5/1.4

Average river yield, kg/km2

Average ratio of load to deposition

Ratio range

0.0017 0.0013 0.016 0.003 0.0009

0.1 0.05 0.9 0.1 0.03

0.04–0.24 0.02–0.14 0.26–2 0.06–0.13 0.02–0.04

35

0.07

4.5

0.8–15

25.0 30.6

2.3 200.8

0.004 2.05

0.16 70

0.08–0.23 7–201

4.8/8.4

17.2

2.3

0.003

0.18

0.008–0.39

0.16/0.01 0.013/0.01

12.2 4.8

0.01 0.021

0.0002 0.005

0.03 1.0

NA 0.95–1.25

1997–2013

19.5/25

13.3

2004–2005 2003–2005 2011–2014

0.46/0.52 0.51/0.26 3.32/0.56

10 12 14

0.34 0.11 1.7

0.003 0.0002 0.006

2012–2013 2013 2014 2013

2.5/0.4 0.0014/0.0002 0.028/0.004 0.44/0.063

15.7 13 11.5 12.5

6.8 0.6 0.39 0.16

0.04 5 0.14 0.004

2012–2013 2012–2013 2012–2013 2012–2013

0.0075/0.0025 0.013/0.004 0.012/0.004 0.011/0.004

36.9 33.9 32.5 30.7

0.00041 0.00032 0.00021 0.0005

0.0015 0.0007 0.0005 0.001

0.04 0.019 0.013 0.03

0.02–0.06 0.01–0.03 0.01–0.018 0.03–0.035

Colorado Big Thompson River at Moraine Park Big Thompson River at Loveland

2008–2011 2006–2011

1.2/0.028 9.0/21.0

11.7 21.8

0.23 0.12

0.002 0.00009

0.19 0.004

0.16–0.22 0.002–0.01

Arctic locations Lower Mackenzie River Lower Yukon River

2007–2010 2001–2005

12,500 5970

0.002 0.005

0.27 0.74

0.18–0.33 NA

2003–2006, 2010 Bear River Below Wolf Creek Near Lucas Hill CA 2001–2002 Greenhorn Creek at You Bet Road near Nevada 2000–2003 City Bear River at Wheatland Bear River at 2000–2002 Wheatland Sagehen Creek 2009 Z4LA Drainage 2007–2010 Nevada Carson River near Fort Churchill Oregon Beaverton Creek Lookout Creek Coast Fork Willamette below Cottage Grove Dam Coast Fork Willamette at London Furnace Creek near Black Butte Mine Dennis Creek, near Black Butte Mine Upstream Garoutte Creek near Black Butte Mine Gus Creek Pothole Creek Upper Main Creek Rock Creek

was 35,900 ng/L and the median was 1080 ng/L (Data from National Water Information System http://nwis.waterdata.usgs.gov/nwis). Cache Creek (California) also has abandoned Hg mines, and THg stream loads and concentrations are elevated, particularly in response to storm water runoff compared to watersheds without Hg mineralization or mine sites. Observed concentrations for this study ranged from 1.4 ng/L at low river flow to just over 700 ng/L during runoff. (Data from National Water Information System http://nwis.waterdata.usgs. gov/nwis). The median observed concentration was 24 ng/L. Mercury is present in the cinnabar form in the geological deposits associated with these mines and was then processed to produce elemental Hg, principally for amalgamation at mines in the Western U.S., mostly in California and Nevada. Mine wastes (calcines) resulting from the cinnabar roasting process erode following precipitation events and transport the mine solids to downstream locations (Rytuba, 2003). In watersheds with Hg mineralization, such as Cache Creek, natural sources to streams (soils or geothermal discharges) are also present (Domagalski et al., 2004), and it is likely that stream concentrations are always elevated relative to non-mineralized watersheds at comparable flow conditions. The Guadalupe River (California), downstream of the New Almaden Hg mining district, had an estimated average ratio of THg river load to atmospheric deposition of 4.5. THg loads on the Guadalupe River varied between 8 and 116 kg/yr (McKee et al., 2010). A TMDL target for this watershed has been established at 106.5 kg/yr (Austin et al., 2008).

7 7

1840

3425 4400

0.5

24.5

0.34 0.14 0.4 2.1 385 12 0.32

2–92

0.24–0.45 0.13–0.14 0.2–0.6 0.4–3.8 NA NA NA

Several watersheds in California have historic gold mining operations where elemental Hg was used to recover the gold (Alpers et al., 2005). This is true of the Yuba and Bear River watersheds as well as Greenhorn Creek, a tributary to the Bear River, and the upper American River, all located within the greater Sacramento River watershed. Two urban watersheds, one in Oregon (Beaverton Creek) and one in California (Z4LA Drainage) have elevated ratios of stream load to atmospheric deposition. As noted by Brigham et al. (2009), atmospheric deposition is likely the major pathway of THg to Beaverton Creek. That is likely the case for the Z4LA Drainage as well, although secondary sources might also be possible. Other sources of THg in urban environments might also contribute to the stream loads including consumer products, sewage, and industrial discharges. In contrast to locations with known Hg contamination because of mining and amalgamation processes, several watersheds are in locations away from mining and are located in largely undeveloped regions: Sagehen Creek in California, Big Thompson River at Moraine Park in Rocky Mountain National Park, Colorado, Lookout Creek (Oregon), and four small basins within the larger Trask River watershed (Oregon). Pathways of Hg at those locations should also be predominantly atmospheric and that from weathering of rock. Rock types in these areas are igneous (extrusive and intrusive). Median observed concentration of THg at the Big Thompson River at Moraine Park site was 1.7 ng/L

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112

8

J. Domagalski et al. / Science of the Total Environment xxx (2016) xxx–xxx

within the Trask River, Sagehen Creek and the lower Big Thompson River suggest significant retention of the annual deposition of THg most likely onto soil. Mercury has been shown to be strongly retained by soils, and the sorption increases with increasing organic matter content (Liao et al., 2009). The Arctic watersheds have much larger loads of Hg because of their larger size and discharge. Stream flow for the Mackenzie River during the time frame of the study was about 300 km3/yr, while that of the Yukon was about 200 km3/yr. Although there are some mines located in these two watersheds, most of the THg in the rivers is from atmospheric deposition or geological formations. Both rivers have elevated ratios of THg load relative to atmospheric deposition compared to other mostly undeveloped watersheds of this study. The average ratio for the Mackenzie was 0.27 whereas that for the Yukon was 0.74. 3.2. Climatic effects on THg deposition and stream load

Fig. 2. Relation between mine density and ratio of stream load to atmospheric mercury deposition (A: all watersheds; B: small watersheds; C: large watersheds).

with a range of 0.7–11.6 ng/L (Data from, National Water Information System http://nwis.waterdata.usgs.gov/nwis). At the Sagehen Creek most concentrations were low, ranging from 0.5 to 2.0 ng/L during peak runoff (Faïn et al., 2011). Ratios of stream THg load to atmospheric Hg deposition, however, are very different at these two locations. At Sagehen Creek, the ratio was 0.03, whereas for the Big Thompson River at Moraine Park the average ratio was 0.19 and the highest observed was 0.22. THg concentrations at Lookout Creek ranged from 0.06 to 1.40 ng/L with a median concentration of 0.08 ng/L. The low load-to-deposition ratio at some of these watersheds, especially those

Precipitation affects how much THg will be deposited onto watersheds. Mean annual precipitation for 2000 through 2014 by climate division is shown in Fig. 3. Climate Divisions (CDs, http://www.esrl.noaa. gov/psd/data/usclimdivs/data/map.html) were developed for statewide, regional, national, and population-weighted monitoring of drought, temperature, precipitation, and heating/cooling degree-day values. Each of the 48 contiguous US states is subdivided into as many as 10 CDs, depending upon the size of the state. The CDs generally coincide with county boundaries and cover the total area of the state. In mountainous states, topographic features determine the boundaries. CDs allow for easy calculation of regional averages, and comparison of recent climate anomalies against a century-long record (Guttman and Quayle, 1996). Locations of CDs for the states of this study are shown in the Supporting Information (Figs. SI34–37). The Oregon sites (151 ± 24.7 mm) receive the greatest amount of rainfall followed by California (54.5 ± 13.8 mm) and Colorado (36.5 ± 5.7 mm). Nevada precipitation (18.1 ± 4.0 mm) is substantially less (Fig. 3). Precipitation in the Western U.S. has high year-to-year variability with frequent extended dry periods. Another method that can be used to evaluate the “wetness” of a CD is the Palmer Drought Severity Index (PDSI). The PDSI uses readily available temperature and precipitation data to estimate relative dryness and potential evapotranspiration. It is a standardized index that spans −10 (dry) to +10 (wet) years, and is reasonably successful at quantifying long-term drought (http://www. cpc.ncep.noaa.gov/products/monitoring_and_data/drought.shtml). Some of its limitations include the lack of multi-timescale features that make correlation with specific water-resource variables such as runoff, snowpack, and reservoir storage difficult. It also does not account for snow or ice (delayed runoff) and assumes precipitation is immediately available. (https://climatedataguide.ucar.edu/climate-data/palmerdrought-severity-index-pdsi#sthash.uhpO4Dkn.dpuf; Dai, 2011). Its usefulness becomes apparent when trying to determine the dominant Hg atmospheric input route (wet or dry deposition), especially given the paucity of direct and modeled dry deposition data (Fig. 4, SI 38–41). Oregon watersheds (Trask River sub-watersheds, Beaverton Creek, Lookout Creek, Furnace Creek and the two watersheds within the Coast Fork of the Willamette River) are distributed primarily throughout or very near the Willamette River Valley and encompass three different CDs (1, 2, and 4). Fig. SI39 shows the PDSI for the three CDs of interest in Oregon for 2000 through 2014. With few exceptions, the CDs track one another reasonably well. From mid-2001 to 2002 Division 1 recovered from an extreme drought quicker than the other two and from mid-2007 to 2010 Division 4 was wetter than the other two CDs. Overall, the annual PDSI for all three CDs oscillated mostly in the near normal range over this period of record. The 15-year tendency for all three CDs, however, was towards more wet conditions, as can be seen by the positive slopes of the regression lines, with Division 4 having the largest (slope = 0.021) The effect of this trend on Hg atmospheric deposition in these watersheds, if it continues, will be to increase the

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112

J. Domagalski et al. / Science of the Total Environment xxx (2016) xxx–xxx

9

Fig. 3. Mean annual precipitation for selected climate divisions. Climate division data are from: http://www7.ncdc.noaa.gov/CDO/CDODivisionalSelect.jsp# (accessed July 23, 2015). Climate division maps are available at http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/regional_monitoring/CLIM_DIVS/states_counties_climate-divisions.shtml.

amount and proportion of wet deposition and runoff on THg loads in the various rivers and streams. The wet THg deposition trend over the period of this study in Oregon is positive, as previously mentioned, and is consistent with increasing wetness in the PDSI. The Colorado watershed (Big Thompson River) includes a portion of the Rocky Mountain National Park and extends east into the high plains area to include the city of Loveland. Fig. SI41 shows the PDSI for the CD for 2000 through 2014. This CD experienced severe drought conditions from mid-2002 to mid-2003 and from mid-2012 to mid-2013, and for a brief period in mid-2006. An extremely moist period began in early 2009 and lasted ~1.5 y followed by an extreme drought in 2012–2013. This drought period was followed by another extremely wet period. These two extremely moist periods are driving the 15-y tendency to wetter conditions. It is also evident that during this 15-y period that there was a pattern of major oscillations between the extremes. These extreme events dictate which form of atmospheric deposition, wet or dry, will be the predominant input route for airborne Hg. Overall, there was a slight increase in THg wet deposition in Colorado during this period of study. The majority of the watersheds in California are located within CD 2 with the small Sagehen Creek watershed within CD 3. Fig. SI38 shows the PDSI for the two CDs for 2000 through 2014. The PDSIs for these two CDs track each other closely (p b 0.001). Division 3, however, had several instances of severe to extreme drought from 2001 to 2003, 2007 to 2009, and again from 2012 through 2014. The general pattern observed for this 15-year period is several years of severe to extreme drought followed by several years of near normal conditions followed

Fig. 4. Palmer Drought Severity Index (PDSI) for selected climatic divisions in the study area. Zero indicates a normal year, negative 2 indicates a moderate drought, negative 3 indicates a severe drought, and negative 4 and below indicates an extreme drought.

by several years of very to extremely moist. The overall tendency for Division 2 is towards decreasing wetness, as indicated by the slope of the trend line (slope = −0.003), but it is within the lower bounds of near normal. The tendency for Division 3 is a slight increase towards more wet (slope = 0.002), but it is at the very lower limit of the near normal range. Overall, there was a negative trend in THg wet deposition in California during this period of study. The Nevada watersheds are clustered together and encompass two CDs (1 & 3) in Nevada, and one in California (3). Fig. SI40 shows the PDSI for the three CDs for 2000 through 2014. This 15-year period includes approximate repeating 5-year cycles from extreme drought to extreme moist with very little within the near normal range. Although the PDSI for these three CDs track reasonably close together, they are not correlated (CA_CD3/NV_CD1 p = 0.09; NV_CD1/NV_CD3 p = 0.95), with the two Nevada CDs showing a slight downward tendency to more dry conditions (slopes = − 0.0072 and − 0.003 for CD1 and CD3, respectively) and the California CD showing a slight upward tendency (slope = 0.002) to more moist conditions. They all are, however, at the limit of the near normal range. Based on the analysis of these limited data, dry is the dominant form of atmospheric Hg inputs to these watersheds under severe to extreme drought conditions. There was no trend in THg wet deposition in Nevada during this period of study. Because of how water is managed, especially in the Western United States, stream flow records from sites with long-term data sets, such as the Carson River, the Sacramento River, and the American River, indicate that either climate or management has affected stream flow (Figs. SI42, SI43, SI44). In all three cases, mean daily stream flow has declined over the period of sampling. Climatic effects on THg stream load can be shown by comparing calculated annual loads to flow-normalized loads (Figs. SI45–47). Flow normalization averages all loads for a period of record. A sufficiently long period of record, at least 20 years, is necessary for this type of calculation. Both the Sacramento River and American River show a statistically significant (p b 0.05) downward trend in THg load, which can be partly explained by the long-term trends in decreasing stream flow. A plot of annual Hg wet deposition for selected watersheds is shown in Fig. 5. Wet deposition of THg does not show as much variability from year to year, as opposed to the variation of the PDSI. Wet deposition of THg does follow the drought index in most cases with slightly elevated deposition in wetter years and less in drier years, as it should because the drought index is directly related to rainfall quantity. There are increases in wet deposition of THg in 2005 and 2006 and again in 2011 when the drought index indicated normal to wet conditions. There is a noticeable reduction in 2013 at most sites except Big Thompson (Colorado) for 2013, as the west coast in North America was in drought conditions. The Western states wet Hg deposition map for 2013, shown in the Supporting Information (Fig. SI14), also shows a definite drop in wet deposition relative to the previous years. It is clear from the PDSI graphs (Fig. 4) that there are times during the study period when dry deposition was likely to have been a more important input pathway than wet.

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112

10

J. Domagalski et al. / Science of the Total Environment xxx (2016) xxx–xxx

Fig. 5. Annual wet deposition of THg at selected watersheds.

3.3. Land-use effects on stream THg loads Several locations in California, principally the Yuba and Bear River sites within the Sierra Nevada, have varying loads and ratios of load to atmospheric deposition. We investigated whether an elevated ratio of stream load to atmospheric deposition in these areas is correlated to watershed mine density. The ratios and mine densities on Fig. 2 are separated out to show all watersheds (Fig. 2A), small watersheds (Fig. 2B), and large watersheds (Fig. 2C). The best correlation of mine density with the ratio of load to deposition is for small watersheds (r2 = 0.93, p = 0.003. As shown in Fig. 2, ratios vary across these sites, especially with the larger watersheds. Some areas with relatively high mine density, such as the Yuba River, have relatively low ratios of runoff to deposition, and some with low mine density, such as Cache Creek and the Carson River, have elevated ratios. Watershed size may be of some importance here. The Greenhorn Creek and Furnace Creek watersheds both have elevated ratios of runoff relative to deposition. Greenhorn Creek (98 km2) is the smallest watershed considered within the Bear and Yuba system, and has an elevated ratio. The other 5 watersheds within the Bear and Yuba system have areas ranging from 509 to 2874 km2. Watershed processes may attenuate the concentrations of THg as area increases. How Hg was used at individual mines probably also plays a role on subsequent runoff, as well as the hydrological connectivity of contaminated material to the stream. For example, mine tailings are directly in the streambed of Furnace Creek. At the Carson River system in Nevada, it was estimated that 6350 metric tons of Hg were lost to the watershed during the time frame of the mining operation (Lechler et al., 1997; Smith, 1943). Some locations, such as Cache Creek, have had some successful remediation in the years since mining operations ended. Watershed properties, especially land use, also affect the ratio of stream load to atmospheric deposition (Domagalski et al., 2004; Alpers et al., 2005; Alpers, 2015, Brigham et al., 2009). Two sites with predominantly urban land use, and therefore, a significant amount of

the watershed with impervious surfaces, are Beaverton Creek (Oregon) and Z4LA Drainage (California). The average ratio at Beaverton Creek (88% urban) was 0.34, and that of the Z4LA Drainage (100% urban) was 1. Atmospheric deposition is likely the main pathway of Hg to the landscape in these two watersheds. Urban watersheds are likely more efficient transporters of atmospherically deposited Hg than forested watersheds because of the preponderance of impervious surfaces in urban areas. These ratios are higher than those found in watersheds with a preponderance of natural vegetation land cover as those tend to be near or below 0.1. Runoff of THg in urban environments is likely related to the density of impervious surfaces. The Z4LA Drainage is small and mostly covered with impervious surfaces, whereas the Beaverton Creek watershed is in a suburban area with some open space including residential areas and parks. One watershed, Big Thompson River at Moraine Park was unique in having a large amount of land cover (about one-third of the watershed area) as snow and ice or barren land. The only likely Hg pathway was atmospheric deposition, and the average ratio was 0.19, higher than most forested sites without mines. There were two sites on Big Thompson River where it was possible to calculate THg stream loads. The THg load was attenuated at the downstream site as the average load at the downstream site was about one half that of the upstream, indicating that THg possibly settled to the streambed. The area of the upstream site is only 7% of the area of the combined upstream and downstream portions of the Big Thompson River watershed at the Loveland stream flow gauging site. The average ratio of stream load-to-atmospheric deposition is greatly attenuated with an upstream ratio of 0.19 and a downstream ratio of 0.004. About 75% of the greater Big Thompson River watershed is either forested or has natural vegetation. Attenuation or retention of atmospherically deposited THg in high-altitude forested watersheds was also observed at Sagehen Creek (Faïn et al., 2011). At Sagehen Creek, forests were about 88% of the land cover and the remainder was mostly grasses. Unlike the upper Big Thompson River site, there was no perennial ice and snow or barren land. Collectively, natural vegetation (forest and grasses) is the main land cover in this watershed, so it is possible that much of the atmospherically deposited Hg is retained in the soils or vegetation, similar to the situation observed at the Sagehen Creek watershed (Faïn et al., 2011). Residual Hg from the Black Butte Mine in the southern portion of the Willamette River watershed has affected stream loads for a portion of the Coast Fork Willamette River. Mercury-laden waste from the abandoned mine enters Furnace Creek, which then discharges into another tributary before it enters the Coast Fork Willamette River. Furnace Creek is a small stream draining the mine site. One year of monitoring data on Furnace Creek (2013) indicated that about 0.6 kg of Hg was transported out of Furnace Creek. The first downstream sampling site on the Coast Fork Willamette River is located about 10 km from the mine site, and the ratio of the stream load to atmospheric deposition at that location exceeded 2, indicating that the significant presence of residual Hg is still present. The Coast Fork Willamette River travels another 7.5 km before discharging into a small reservoir. The reservoir traps some of the Hg, but the reservoir outflow still has a ratio of stream load to atmospheric deposition of 0.4. 3.4. Watershed yields of THg A plot of average annual atmospheric total Hg deposition versus THg stream loads (Fig. 6A) shows the effect of watershed type and size on stream load. Area-normalized atmospheric Hg deposition and watershed THg yield are shown in Fig. 6B. Watershed yield is the load of THg divided by watershed area. Diagonal lines in Fig. 6A and B indicate constant values of the ratio of stream load or yield to atmospheric deposition. Sites with either Hg mines or Au-Ag mines where Hg amalgamation methods were used had the highest watershed yields, although some sites, particularly the amalgamation sites, were only slightly elevated or overlapped with reference or low-impact sites. Only mining

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112

J. Domagalski et al. / Science of the Total Environment xxx (2016) xxx–xxx

Fig. 6. Plots of annual average total atmospheric mercury deposition versus annual average watershed load or yield of total mercury. A: total mercury load, B: mercury mass normalized to watershed area (Yield). Symbols denoting watershed type and size as in Fig. 2.

impacted sites had watershed yields above 0.01 kg/km2. Large watersheds such as the lower Sacramento and American Rivers had areanormalized THg yields indistinguishable from those of the reference sites. Despite having mines and urban centers, elevated yields were not apparent. Small urban sites were elevated above the mixed land use sites and reference sites. One of the Arctic sites, the Mackenzie River, had a yield similar to one of the higher reference sites. The Yukon River had a higher yield than the Mackenzie River and all of the reference sites. Both of the Arctic sites had similar yields and stream/atmospheric ratios relative to the urban sites (Fig. 6B). As explained by Schuster et al., 2011, these high yields may be attributable to climate change. Melting of permafrost, a sink of THg, will contribute to Hg yields or export to downstream water bodies with warming temperatures. 4. Summary Similar to previous studies (e.g. Brigham et al., 2009), stream export of THg from watersheds relative to atmospheric deposition is highly

11

variable. In this study, stream loads of THg, especially in smaller watersheds with Hg mining or precious metals mining that used Hg amalgamation, are higher relative to what is deposited over the watershed by precipitation. In watersheds away from mines, lower ratios of stream export to deposition indicated that Hg is retained in the watersheds. Retention is attributable to soil absorption of the deposited Hg, less precipitation mobilizing Hg, or trapping of sediments by reservoir dams. It has been previously shown (Liao et al., 2009) that Hg is strongly adsorbed onto soil, with the adsorption increasing with increasing amounts of organic matter. In watersheds where abandoned Hg and precious metals mines that used Hg amalgamation are located, especially smaller watersheds with mines, stream loads may exceed that of atmospheric deposition, on an annual basis. Watershed area is important, as smaller watersheds with mining have more elevated ratios of stream load relative to atmospheric deposition as well as higher watershed yield. This was true for locations with mines that utilized Hg amalgamation for gold or silver recovery, and for locations with Hg mines. Locations with no or minimal mine-site remediation will have elevated loads of THg, especially during rainfall events that mobilize easily erodible sediment, such as previously shown by Domagalski et al. (2004) and Thomas et al. (2002). Larger watersheds with Hg mines and other mixed land uses, such as the Sacramento River, have lower ratios of stream load relative to atmospheric deposition, during years of low rainfall, that are similar to non-mining locations. The ratios tend to be elevated during wet years, indicating that either the previously deposited atmospheric Hg or erodible sediments contaminated with mining debris, as well as natural sources of Hg, contribute to the annual THg load under wetter conditions. At least one large watershed, the Carson River in Nevada, has significant contamination because of the amount of Hg mobilized to the watershed as a result of silver mining with amalgamation. THg loads in that watershed are likely to be elevated for a long time. Watersheds with mostly natural vegetation (forests and grasses), away from mining influences, retained Hg, presumably in surficial soils, as the ratios of stream load to atmospheric deposition tended to be close to or below 0.1 in most cases. The Arctic locations, especially the Yukon River, had elevated ratios of stream load relative to atmospheric deposition, especially when compared to other low-impacted watersheds in California and Oregon. Changing climate in this region is likely causing increased loads and mobilization of Hg due to melting permafrost. Annual Hg atmospheric deposition amounts for each watershed are correlated with the total rainfall amounts and PDSI values for each of the associated CDs, but are attenuated and do not show any large interannual variations. Wet deposition of THg throughout the Western U.S. showed a statistically significant positive trend over the period of the study, but not in all states. The trend was significant and downward in California. That is consistent with a decreasing trend of precipitation in California for that period. Both Colorado and Oregon showed a trend towards wetter conditions as indicated by the Palmer Drought Severity Index, as well as a trend of higher THg wet deposition. In this study, only one year of modeled Hg dry deposition was available. The limited data nonetheless suggested that, on average, dry deposition of Hg is about 80% of the wet deposition across the Western U.S. That data gap is an area of research that needs to be expanded upon. Drought conditions are common for portions of Western North America suggesting that dry deposition will be the dominant atmospheric input of Hg during droughts. Although there are locations in Western North America where wet deposition of Hg is measured, the spatial density of sites is low and represents another area with a significant data gap. Acknowledgements This work was conducted as a part of the Western North American Mercury Synthesis Working Group supported by the John Wesley Powell Center for Analysis and Synthesis, funded by the U.S. Geological Survey. We also acknowledge a grant from the Region-10 U.S.

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112

12

J. Domagalski et al. / Science of the Total Environment xxx (2016) xxx–xxx

Environmental Protection Agency (RARE). We also gratefully acknowledge the contribution from Dr. Jerry Lin (Lamar University) for providing output from the Community Multi-scale Air-quality (CMAQ) model of 2009 Hg dry deposition. We are also grateful to the Sacramento Coordinated Monitoring Program (California) for sharing their Hg data for the Sacramento and American Rivers and the Regional Monitoring Program for Water Quality in the San Francisco Bay. We thank Carl Thodal and Eric Morway, of the Nevada Water Science Center (USGS), for sharing data on the Carson River watershed. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scitotenv.2016.02.112. References Alpers, C.N., 2015. Arsenic and mercury contamination from historical gold mining in the Sierra Nevada, California. 27th International Applied Geochemistry Symposium (IAGS) Hosted by the Association of Applied Geochemists (AAG) in Tucson, AZ, April 20–24, 2015. Alpers, C.N., Hunerlach, M.P., May, J.T., Hothem, R.L., 2005. Mercury Contamination from Historical Gold Mining in California. U.S. Geological Survey Fact Sheet 2005–3014. http://pubs.usgs.gov/fs/2005/3014/. Ambers, R.K.R., Hygelund, B.N., 2001. Contamination of two Oregon reservoirs by cinnabar mining and mercury amalgamation. Environ. Geol. 40 (6), 699–707. Ashley, R.P., 2002. Geoenvironmental model for low-sulfide gold-quartz vein deposits. U.S. Geological Survey Open-file Report 02-195K , pp. 176–195. http://pubs.usgs. gov/of/2002/of02-195/OF02-195K.pdf. Austin, C.M., Potter, S.M., Ponton, J.D., Whtye, D.C., Mumley, T.E., 2008. Guadalupe River Watershed Mercury Total Maximum Daily Load (TMDL) Project, Staff Report For Proposed Basin Plan Amendment. San Francisco Bay Region, California Regional Water Quality Control Board. http://www.waterboards.ca.gov/sanfranciscobay/ water_issues/programs/TMDLs/guadalupe_river_mercury/C1_Guad_SR_Sep08.pdf. Bash, J.O., 2010. Description and initial simulation of a dynamic bidirectional air-surface exchange model for mercury in Community Multiscale Air Quality (CMAQ) model. J. Geophys. Res. 115, D06305. http://dx.doi.org/10.1029/2009JD012834. Brabets, T.P., Wang, B., Meade, R.H., 2000. Environmental and hydrologic overview of the Yukon River Basin, Alaska and Canada. U.S. Geological Survey Water-Resources Investigations Report 99-4204. Anchorage, Alaska, p. 106. http://pubs.usgs.gov/wri/ wri994204/. Brigham, M.E., Wentz, D.A., Aiken, G.R., Krabbenhoft, D.P., 2009. Mercury cycling in stream ecosystems. 1. Water column chemistry and transport. Environ. Sci. Technol. 43, 2720–2725. Brobst, D.A., Pratt, W.P. (eds)., 1973 United States mineral resources: mercury. U.S. Geological Survey Prof. Paper 820. 736 p. Churchill, R.K., 2000. Contributions of mercury to California's environment from mercury and gold mining activities—insights from the historical record. Proceedings and Summary Report, Workshop on Assessing and Managing Mercury from Historic and Current Mining Activities, November 28–30, 2000. U.S. Environmental Protection Agency, San Francisco, California EPA/625/R-04/102. Cohn, T.A., Delong, L.L., Gilroy, E.J., Hirsch, R.M., Wells, D.K., 1989. Estimating constituent loads. Water Resour. Res. 25, 937–942. Crawford, C.G., 1996. Estimating mean constituent loads in rivers by the rating-curve and flow-duration, rating-curve methods. Bloomington, Indiana, Indiana University Ph.D. dissertation, p. 245. Dastoor, A.P., Davignon, D., 2009. Global mercury modelling at environment Canada. In: Pirrone, N., Mason, R. (Eds.), Mercury Fate and Transport in the Global Atmosphere: Emissions Measurements and Models. Springer-Verlag, New York, pp. 519–532. Dastoor, A., Ryzhkov, A., Durnford, D., Lehnherr, I., Steffen, A., Morrison, H., 2015. Atmospheric mercury in the Canadian Arctic. Part II: insight from modeling. Sci. Total Environ. 509-510, 16–27. Dai, A., 2011. Drought under global warming: a review. Wiley Interdisciplinary Reviews: Climate Change http://dx.doi.org/10.1002/wcc.81. Domagalski, J.L., Alpers, C.N., Slotton, D.G., Suchanek, T.H., Ayers, S.M., 2004. Mercury and methylmercury concentrations and loads in the Cache Creek watershed, California. Sci. Total Environ. 327, 215–237. Emmerton, C.A., Graydon, J.A., Gareis, J.A.L., St. Louis, V.L., Lesack, L.F.W., Banack, J.K.A., Hicks, F., Nafziger, J., 2013. Mercury export to the Arctic Ocean from the Mackenzie River, Canada. Environ. Sci. Technol. 47, 7644–7654. Evers, D.C., Clair, T.A., 2005. Mercury in northeastern North America: a synthesis of existing databases. Ecotoxicology 14 (1–2), 7–14. Faïn, X., Obrist, D., Pierce, A., Barth, C., Gustin, M.S., Boyle, D.P., 2011. Whole-watershed mercury balance at Sagehen Creek, Sierra Nevada, CA. Geochim. Cosmochim. Acta 75, 2379–2392. Fitzgerald, W.F., 1986. Cycling of mercury between the atmosphere and oceans. In: BrentMenard, P. (Ed.), The Role of Air-Sea Exchange of Geochemical Cycling. D. Reidel, Dordrecht, pp. 363–408. Gbor, P.K., Wen, D., Meng, F., Yang, F., Sloan, J.J., 2007. Modeling of mercury emission, transport and deposition in North America. Atmos. Environ. 41, 1135–1149.

Gustin, M.S., Lindberg, S.E., Austin, K., Coolbaugh, M., Vette, A., Zhang, H., 2000. Assessing the contribution of natural sources to regional atmospheric mercury budgets. Sci. Total Environ. 259, 61–71. Guttman, N.B., Quayle, R.G., 1996. A historical perspective of U.S. climate divisions. Bull. Amer. Meteor. Soc. 77, 293–304. Hirsch, R.M., De Cicco, L.A., 2014. User guide to Exploration and Graphics for RivEr Trends (EGRET) and dataRetrieval: R packages for hydrologic data (version 2.0, February 2015). U.S. Geological Survey Techniques and Methods book 4 chap. A10, 93 p http://dx.doi.org/10.3133/tm4A10. Hylander, L.D., Meili, M., 2003. 500 years of mercury production: global annual inventory by region until 2000 and associated emissions. Sci. Total Environ. 304, 13–27. Journey, C.A., Burns, D.A., Riva-Murray, K., Brigham, M.E., Button, D.T., Feaster, T.D., Petkewich, M.D., Bradley, P.M., 2012. Fluvial transport of mercury, organic carbon, suspended sediment, and selected major ions in contrasting stream basins in South Carolina and New York, October 2004–September 2009. U.S. Geological Survey Scientific Investigations Report 2012–5173 , p. 125. http://pubs.er.usgs.gov/publication/ sir20125173. Krause, P., Boyle, D.P., Bäse, F., 2005. Comparison of different efficiency criteria for hydrological model assessment. Adv. Geosci. 5, 89–97. Lechler, P.J., Miller, J.R., Hsu, L.C., Desilets, M.O., 1997. Mercury mobility at the Carson River Superfund Site, west-central Nevada, USA: interpretation of mercury speciation data in mill tailings, soils, and sediments. J. Geochem. Explor. 58, 259–267. Liao, L., Selim, H.M., DeLaune, R.D., 2009. Mercury adsorption-desorption and transport in soils. J. Environ. Qual. 38, 1608–1616. Lin, C.J., Pongprueksa, P., Lindberg, S.E., Pehkonend, S.O., Byune, D., Jang, C., 2006. Scientific uncertainties in atmospheric mercury models I: model science evaluation. Atmos. Environ. 40, 2911–2928. Lincoln, F.C., 1923. Mining Districts and Mineral Resources of Nevada, reprinted 1982. Nevada Publications, Las Vegas, Nevada, pp. 222–233. Long, K.R., DeYoung Jr., J.H., Ludington, S.D., 1998. Database of significant deposits of gold, silver, copper, lead, and zinc in the United States. U.S. Geological Survey Open-File Report 98-206-A,B, p. 60, Available at https://pubs.er.usgs.gov/publication/ofr98206AB. Lyman, S.N., Gustin, M.S., Prestbo, E.M., Marsik, F.J., 2007. Estimation of dry deposition of atmospheric mercury in Nevada by direct and indirect methods. Environ. Sci. Technol. 41, 1970–1976. McKee, L., Leatherbarrow, J., Oram, J., 2005. Concentrations and loads of mercury, PCBs, and OC pesticides in the lower Guadalupe River, San Jose, California: water years 2003 and 2004. A Technical Report of the Regional Watershed Program: SFEI Contribution 409. San Francisco Estuary Institute, Oakland, CA, p. 72. McKee, L.J., Hunt, J.A., Greenfield, B.J., 2010. Concentration and loads of mercury species in the Guadalupe River, San Jose, California, Water Year 2010. A Report Prepared for the Santa Clara Valley Water District in Compliance with California Regional Water Quality Control Board San Francisco Bay Region Order Number 01-036 as Amended by Order Number R2-2009-0044, Requirement D. October 29, 2010. San Francisco Estuary Institute. http://www.sfei.org/documents/concentrations-and-loads-mercuryspecies-guadalupe-river-san-jose-california-water-year. McKee, L.J., Gilbreath, A.N., 2015. Concentrations and loads of suspended sediment and trace element pollutants in a small semi-arid urban tributary. San Francisco Bay, California. Environ. Monit. Assess. 187, p. 499. Nriagu, J.O., 1994. Mercury pollution from the past mining of gold and silver in the Americas. Sci. Total Environ. 149, 167–181. Pietroniro, A., Soulis, E.D., 2001. Comparison of global land-cover databases in the Mackenzie basin, Canada. In: Owe, M., Brubaker, K., Ritchie, J., Rango, A. (Eds.), Remote Sensing and Hydrology 2000. Proceedings of symposium held at Santa Fe, New Mexico, USA, April 2000, IAHS Publication No. 267, 2001, pp. 552–557. Pirrone, N., Cinnirella, S., Feng, X., Finkelman, R.B., Friedli, H.R., Leaner, J., Mason, R., Mukherjee, A.B., Stracher, G., Streets, D.G., Kevin Telmer, K., 2009. Global mercury emissions to the atmosphere from natural and anthropogenic sources, chapter 1. In: Pirrone, N., Mason, R. (Eds.), Mercury Fate and Transport in the Global Atmosphere: Emissions Measurements and Models. Springer-Verlag, New York, pp. p 3–50. Risch, M.R., DeWild, J.F., Krabbenhoft, D.P., Kolka, R.K., Zhang, L., 2012. Litterfall mercury dry deposition in the eastern USA. Environ. Pollut. 161, 284–290. Runkel, R.L., Crawford, C.G., Cohn, T.A., 2004. Load Estimator (LOADEST): a FORTRAN program for estimating constituent loads in streams and rivers. U.S. Geol. Surv. Techniques and Methods, Book 4, Ch. A5, p. 69. Ryaboshapko, A., Bullock Jr, O.R., Christensen, J., Cohen, M., Dastoor, A., Ilyin, I., Petersen, G., Syrakov, D., Travnikov, O., Artz, R.S., Davignon, D., Draxler, R.R., Munthe, J., Pacyna, J., 2007. Intercomparison study of atmospheric mercury models: 2. modelling results vs. long-term observations and comparison of country deposition budgets. Sci. Total Environ. 377, 319–333. Rytuba, J.J., 2003. Mercury from mineral deposits and potential environmental impact. Environ. Geol. 43, 326–338. Schroeder, W.H., Munthe, J., 1998. Atmospheric mercury—an overview. Atmos. Environ. 32, 809–822. Schuster, P.F., Striegl, R.G., Aiken, G.R., Krabbenhoft, D.P., Dewild, J.F., Butler, K., Kamark, B., Dornblaser, M., 2011. Mercury export from the Yukon River Basin and potential response to a changing climate. Environ. Sci. Technol. 45, 9262–9267. Selin, N.E., 2009. Global biogeochemical cycling of mercury: a review. Annu. Rev. Env. Resour. 34, 43–63. Seigneur, C., Vijayaraghavan, K., Lohman, K., Karamchandani, P., Scott, C., 2004. Global source attribution for mercury deposition in the United States. Environ. Sci. Technol. 38, 555–569. Shanley, J.B., Mast, M.A., Campbell, D.H., Aiken, G.R., Krabbenhoft, D.P., Hunt, R.J., Walker, J.F., Schuster, P.F., Chalmers, A., Aulenbach, B.T., Peters, N.E., Marvin-DiPasquale, M., Clow, D.W., Shafer, M.M., 2008. Comparison of total mercury and methylmercury cy-

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112

J. Domagalski et al. / Science of the Total Environment xxx (2016) xxx–xxx cling at five sites using the small watershed approach. Environ. Pollut. 154, 143–154. St. Louis, V.L., Rudd, J.W.M., Kelly, C.A., Hall, B.D., Rolfhus, K.R., Scott, K.J., Lindberg, S.E., Dong, W., 2001. Importance of the forest canopy to fluxes of methyl mercury and total mercury to boreal ecosystems. Environ. Sci. Technol. 35, 3089–3098. Smith, G.H., 1943. The history of the Comstock, 1850–1920. Univ. Nevada Bull. 37, 41–47. Strode, S.A., Jaeglé, L., Jaffe, D.A., Swartzendruber, P.C., Selin, N.E., Holmes, C., Yantosca, R.M., 2008. Trans-Pacific transport of mercury. J. Geophys. Res. 113, D15305. http:// dx.doi.org/10.1029/2007JD009428. Telmer, K.H., Veiga, M.M., 2009. World emissions of mercury from artisanal and small scale gold mining, Chapter 6. In: Pirrone, N., Mason, R (Eds.), Mercury Fate and Transport in the Global Atmosphere,: Emissions, Measurements and Models. SpringerVerlag, New York, pp. 131–172. Thodal, C.E., Morway, E., Thomas, K.A., 2015. Estimation of suspended sediment and total mercury loads and application of flow-adjusted trend analyses to assess floodplain restoration, Carson River, Nevada. Proceedings, 3rd Joint Federal Interagency Conference on Sedimentation and Hydrologic Modeling, April 19–23, 2015, Reno, Nevada, USA, pp. 717–726. Thomas, M.A., Conaway, C.H., Steding, D.J., Marvin-DiPasquale, M., Abu-Saba, K.E., Flegal, A.R., 2002. Mercury contamination from historic mining in water and sediment, Guadalupe River and San Francisco Bay, California. Geochemistry: Exploration, Environment, Analysis. 2, 217–221. http://dx.doi.org/10.1144/1467-787302-024. Tsai, P., Hoenicke, R., 2001. San Francisco Bay atmospheric deposition pilot study part 1: Mercury: San Francisco Estuary Institute. SFEI Contribution 72. http://www.sfei.org/ sites/default/files/ADHg_FinalReport.pdf.

13

Ullrich, S.M., Tanton, T.W., Abdrashitova, S.A., 2001. Mercury in the aquatic environment: a review of factors affecting methylation. Crit. Rev. Environ. Sci. Technol. 31, 241–293. U.S. EPA, 2012. Optimization Review Black Butte Mine Superfund Site Lane County, Oregon. EPA-542-R-12-003. Office of Solid Waste and Emergency Response, Office of Superfund Remediation and Technology Innovation, Washington D.C. Van Denburgh, A.S., 1973. Mercury in the Carson and Truckee River basins of Nevada. U.S. Geological Survey Open-file Report 1973–352 http://pubs.usgs.gov/of/1973/0352/ report.pdf. Wentz, D.A., Brigham, M.E., Chasar, L.C., Lutz, M.A., Krabbenhoft, D.P., 2014. The Quality of Our Nation’'s Waters Mercury in the Nation’'s Streams—Levels, Trends, and Implications. U.S. Geological Survey Circular 1395, p. 90. http://pubs.usgs.gov/circ/1395/ pdf/circ1395.pdf. Wang, Q., Kim, D., Dionysiou, D.D., Sorial, G.A., Timberlake, D., 2004. Sources and remediation for mercury contamination in aquatic systems–a literature review. Environ. Pollut. 131, 323–336. Zhang, L., Blanchard, P., Gay, D.A., Prestbo, E.M., Risch, M.R., Johnson, D., Narayan, J., Zsolway, R., Holsen, T.M., Miller, E.K., Castro, M.S., Graydon, J.A., St. Louis, V.L., Dalziel, J., 2012. Estimation of speciated and total mercury dry deposition at monitoring locations in eastern and central North America. Atmos. Chem. Phys. 12, 4327–4340.

Please cite this article as: Domagalski, J., et al., Comparison of mercury mass loading in streams to atmospheric deposition in watersheds of Western North America: Evidence for non-a..., Sci Total Environ (2016), http://dx.doi.org/10.1016/j.scitotenv.2016.02.112