Atmospheric wet and litterfall mercury deposition at urban and rural sites in China

Mercury (Hg) concentrations and deposition fluxes in precipitation and litterfall were measured at multiple sites (six rural sites and an urban site) across a broad geographic area in China. The annual deposition fluxes of Hg in precipitation at rural sites and an urban site were 2.0 to 7.2 and 12.6± 6.5 μg m−2 yr−1, respectively. Wet deposition fluxes of Hg at rural sites showed a clear regional difference with elevated deposition fluxes in the subtropical zone, followed by the temporal zone and arid/semi-arid zone. Precipitation depth is the primary influencing factor causing the variation of wet deposition. Hg fluxes through litterfall ranged from 22.8 to 62.8 μg m−2 yr−1, higher than the wet deposition fluxes by a factor of 3.9 to 8.7 and representing approximately 75 % of the total Hg deposition at the forest sites in China. This suggests that uptake of atmospheric Hg by foliage is the dominant pathway to remove atmospheric Hg in forest ecosystems in China. Wet deposition fluxes of Hg at rural sites of China were generally lower compared to those in North America and Europe, possibly due to a combination of lower precipitation depth, lower GOM concentrations in the troposphere and the generally lower cloud base heights at most sites that wash out a smaller amount of GOM and PBM during precipitation events.


Introduction
Mercury (Hg), especially its methylated form, is a potent neurotoxin to humans and wildlife.Because of its high volatility, mild reactivity, and water solubility, gaseous elemental mercury (GEM) is the dominant (> 75 %) form of total Hg in the atmosphere (Sprovieri et al., 2010;Gustin et al., 2015;Shah et al., 2016), while the majority of the remaining fractions of Hg in the atmosphere are gaseous oxidized mercury (GOM) and particulate bound mercury (PBM), which are operationally defined.GEM has an atmospheric residence time of several months to a year and therefore spreads globally before being converted to other forms and deposited to the earth's surface (Holmes et al., 2010;Driscoll et al., 2013;Amos et al., 2015).On the other hand, GOM and PBM have much higher water solubility and reactivity, and are readily deposited via wet and dry deposition pathways.
Quantifying the relationship between atmospheric Hg depositions and emissions is of prime importance in its global biogeochemical cycle.Hg in the atmosphere could be from anthropogenic and natural emission sources.At a global scale, direct and legacy anthropogenic emissions represent the predominant (∼ 2/3) sources of Hg in the atmosphere (Seigneur et al., 2004;Selin et al., 2007;Pirrone et al., 2010), and account for most of the deposition of Hg to the earth's surface (Selin et al., 2008;Amos et al., 2013).Hg is removed from the atmosphere through wet deposition and dry deposition pathways.Atmospheric Hg deposition networks have Published by Copernicus Publications on behalf of the European Geosciences Union.
X. Fu et al.: Atmospheric wet and litterfall mercury deposition been established in North America (NADP, 1994) and Europe (EMEP).These networks provided critical databases on the spatial and temporal trends in Hg wet deposition and help to constrain the regional and global atmospheric Hg deposition budget (Selin et al., 2007;L. Zhang et al., 2012).Dry deposition of atmospheric Hg, including uptake of GEM by vegetation and dry deposition of GOM/PBM, is considered to be more important than wet deposition in the total Hg deposition budget (Dastoor and Larocque, 2004;Selin et al., 2007).However, dry deposition of Hg is difficult to be quantified accurately because of the lack of proper measurement techniques.Recently, an increasing number of studies have suggested that measurements of litterfall (uptake of total gaseous mercury -TGM) and throughfall (washout of GOM and PBM deposited to foliage surfaces) could be a good indicator of Hg dry deposition to forest ecosystems (Johnson and Lindberg, 1995;St. Louis et al., 2001;Gustin, 2012).Also, empirical models have been developed to quantify dry deposition of Hg to various landscapes (Lyman et al., 2007;L. Zhang et al., 2012).Although subjected to some bias (Zhang et al., 2009;Gustin et al., 2015), these studies significantly improved our understanding of the role of dry deposition in the global Hg biogeochemical cycle.
China is the largest source region of atmospheric Hg in the world.Recent studies revealed that Hg emissions from anthropogenic and natural sources in China were in the range of 800-1200 t (Shetty et al., 2008; S. X. Wang et al., 2014;Fu et al., 2015a;Wang et al., 2016), approximately 1/6 of the global Hg emissions into the atmosphere (Pirrone et al., 2010;Song et al., 2015).The large emissions in China not only have an impact on the domestic Hg pollution (Fu et al., 2015b), but also have the potential to enhance atmospheric Hg concentration and deposition elsewhere (Seigneur et al., 2004;Strode et al., 2008;Durnford et al., 2010).To construct the mass budget of Hg in China, measurements of atmospheric Hg wet and dry deposition are essential.However, studies of wet and dry deposition in China are rather limited.Most of the previous studies in China were conducted either at a single site using bulk collectors or mostly in urban areas (Guo et al., 2008;Wang et al., 2009;Y. M. Wang et al., 2014;Fu et al., 2010b;Zhou et al., 2013;Xu et al., 2014;Zhu et al., 2014), which are not sufficient to depict the overall pattern of atmospheric Hg deposition in China.In the present study, we conducted comprehensive measurements of wet and litterfall deposition fluxes of Hg at multiple sites (six rural sites and one urban site) across a broad geographic area in mainland China.This study aims to better understand the spatial distributions of Hg deposition and mass balance of atmospheric Hg in China.We also compared these fluxes to previous observations in China and other regions worldwide and studied the potential factors influencing wet and litterfall deposition fluxes of Hg.

Site description
Precipitation samples were collected at six rural sites and one urban site across a broad geographic area in mainland China: Mt.Changbai (MCB,128.112N, 1041 m a.s.l.) (Fig. 1).Four sites (i.e., MCB, MDM, MLG, and MAL) are rural upland forest sites, two sites (i.e., MWLG and BYBLK) are alpine grassland sites and the remainder (i.e., GY) is an urban site (Table 1).
The MCB site is located in a temperate broadleaf and Korean pine mixed forest on the northern slope of Mt.Changbai, northeastern China, which is one of the largest national forest reserves in China and covers an area of ∼ 1965 km 2 .The forest is dominated by tree species of Pinus koraiensis, Fraxinus mandshurica, Tilia amurensis, Acer mono and Quercus mongolica (Dai et al., 2011).There are no nearby sources at the MCB site and the predominant wind directions are from southwest to northwest, which could carry air pollutants from regional industrial and urbanized centers and northern China by long-range atmospheric transport (Fu et al., 2012b).
The MDM site is located at the summit of Mt.Damei in the eastern Yangtze River Delta, eastern China, and about 15 km away from the East China Sea coast.This site is surrounded by a shrub (Styrax faberi, Lithocarpus glaber, and Indocalamus tessellates), broadleaf (Quercus fabri, Liquidambar formosana, and Platycarya Strobilacea) and conifer (Pinus massoniana) mixed forest.The Yangtze River Delta is one of the largest atmospheric Hg source regions in China (L.Zhang et al., 2015).There are no large point sources within 5 km of the sampling site.The predominant wind direction during the wet season is from the east and the site mainly receives clean marine air masses.During the dry season, the predominant wind direction changes to from the northwest, which enhances the transport of air pollutants from northern and eastern China to the sampling site (Yu et al., 2015).
The MLG site is located on top of an isolated peak (elevated approximately 1000 m relative to the surrounding terrain) of the Miaoling mountain range, southwestern China.The sampling site is surrounded by an upland shrub (Rhododendron) and deciduous broadleaf mixed forest (Cinnamomum, Acer Linn., Chaenomeles, Sinarundinaria Nakai).It is isolated from industrial sources and populated regions and frequently receives free tropospheric air from central China plain regions and the South China Sea (SCS) (Fu et al., 2010a).The MLG site is also influenced by boundary layer  air transported by plain-to-mountain winds from southwestern China during daytime.The MAL site is located at a summit of the northern edge of the Ailao Mountain National Nature Reserve in central Yunnan province, southwestern China.The reserve has an area of 677 km 2 and is predominantly (> 80 %) covered by evergreen broadleaf primary forests.Canopy species mainly include Castanopsis wattii, Lithocarpus xylocarpus, Schima noronhae, L. jingdongensis, and Hartia sinensi (Yuan et al., 2009).The MAL site is located 180 km southwest of Kuming, the capital of Yunnan province, and 200 km north of the Indochinese Peninsula.This site is frequently influenced by long-range transport of Hg released from anthropogenic sources and biomass burning in southwestern China, the Indochinese Peninsula, and South Asia (Zhang et al., 2016).
The MWLG site is situated at the summit of Mt.Waliguan at the edge of northeastern part of the Qinghai-Xizang (Tibetan) Plateau.The surrounding area is mainly dominated by naturally preserved upland grasslands and arid/semi-arid lands and there are no nearby sources.The potential source regions are industrial and urbanized centers in northwestern China and northern India (Fu et al., 2012a).
The BYNLK site is located in the Bayinbuluk upland grassland in the southern Tianshan mountains, Xinjiang Uygur Autonomous Region, central Asia, and with a total area of ∼ 23 000 km 2 .The surrounding area is rural and the population density is generally lower than 1 person km −2 .
The GY site is located in the downtown of Guiyang, the capital city of Guizhou province, southwestern China, with a population of ∼ 4.5 million.Guiyang is heavily polluted with respect to atmospheric Hg due to significant industrial emissions (Fu et al., 2011).Surrounding areas of the sampling site are mixed residential and commercial.

Precipitation collection and analysis
Precipitation samples were collected in 2011-2014at MCB and MAL, 2012-2014at MDM and MWLG, 2013-2014at BYBLK, 2012-2013at GY, and 2008-2009 at MLG. Wet-only precipitation samples were collected at all the studied sites using wet-only automatic precipitation collectors (HHG9-ZJC-III, Beijing Midwest Great Technology Co. Ltd., Beijing, China) with the exception of the MLG site.Wet-only precipitation samples at the MLG site were collected using a manual method described in Fu et al. (2010a).The HHG9-ZJC-III sampler consists of a 25 cm diameter, borosilicate glass bucket (volume: 15 L) for precipitation collection and a moisture sensor that could move a cover from and back to the bucket at the onset and the end of a precipitation event using a mechanical arm, respectively (Price et al., 2008).The cover consists of a foam pad encased in plastic wrap and forces a tight seal over the borosilicate glass bucket to prevent evaporation of the sample.The borosilicate glass bucket was baked for 1 h in a muffle furnace at a temperature of 500 • C before field sampling and rigorously rinsed by Milli-Q water before each of the new sampling cycles during the field sampling, and this aimed to avoid possible contamination during the collection of precipitation Hg (Burke et al., 1995).Immediately after a precipitation event, the precipitation sample was transferred carefully to a rigorously acidcleaned Teflon sample bottle (volume: 250 mL).Extra samples during heavy precipitation events (e.g., accumulative precipitation depth of > 6 mm) were used for anion and trace element analysis or abandoned in the field.We caution that, X. Fu et al.: Atmospheric wet and litterfall mercury deposition due to the adsorptive behavior of Hg to the wall of borosilicate glass bucket, Hg in precipitation might be lost to the bucket during the transfer processes, which consequently resulted in an underestimation of Hg concentrations in precipitation.However, the artifact during transferring of precipitation is expected to be low because pervious studies observed minor adsorption of Hg to the clean Teflon and borosilicate glass surface (Vermette et al., 1995;Landis and Keeler, 1997).In the winter at several sampling sites, i.e., MCB, MWLG, and BYBLK, surface fresh snow was collected from the upper 1 cm of the snowpack at open-air sites using a precleaned Teflon scoop.Snow samples were collected immediately after snow events.The snow samples were transferred carefully to a thoroughly acid-cleaned Teflon sample bottle (volume: 400 mL) and remained sealed at a room temperature of 10-20 • C in the dark to melt snow.After the completion of field sampling, trace metal grade HCl (to 5 ‰ of total sample volume) was added into the Teflon bottles with rain and snow samples and each sample was individually sealed into three successive polyethylene bags and kept in a refrigerator at 4 • C until Hg analysis.To ensure clean operation, polyethylene gloves were worn throughout the sample collection period.
Hg concentrations in precipitation were analyzed using US EPA Method 1631 (USEPA, 2002).Briefly, Hg in precipitation was purged from solution in a Hg-free nitrogen stream and concentrated onto a gold-coated sand trap after oxidation by BrCl followed by addition of NH 2 OH q HCl to discharge the excess BrCl and reduction of divalent Hg by SnCl 2 to Hg 0 .The trapped Hg is then thermally desorbed from the gold trap into an inert gas stream and quantified using a dual amalgamation technique followed by cold-vapor atomic fluorescence spectrometry (CVAFS) (USEPA, 2002).The data quality was controlled using laboratory blank tests (< 0.3 ng L −1 ), field blank tests (Hg conc.= 0.46 ± 0.51 ng L −1 , 1SD, n = 21), and the Interlaboratory Comparison Program initiated by the U.S. Geological Survey (USGS, https://bqs.usgs.gov/PCQA/Interlaboratory_Comparison/).The measured Hg values of the standard reference samples prepared by the USGS during 2012-2014 were overall within the quality control acceptance criteria (62 % and 90 %) of measured values were within the 10 and 20 % of the most probable values (MPV, the mean results submitted by all the laboratory participants, respectively) and showed a mean difference of 0.47 ± 1.55 ng L −1 (1 SD, n = 50) in Hg concentrations between the measured values and MPV.

Collection and analysis of litterfall samples
Litterfall samples were collected using litterfall collectors (0.25 m 2 at MLG and 1.0 m 2 at MCB, MDM, and MAL) at the sampling sites (St. Louis et al., 2001;Fisher and Wolfe, 2012;Risch et al., 2012).The collectors were constructed using an untreated wood frame and a Nylon net bottom with a pore size of 1.0 mm × 1.0 mm.At the MCB, MDM, MLG, and MAL sites, four, four, three, and eight collectors were deployed, respectively, which were located under the predominant tree species at the study sites.The litterfall collectors were elevated by approximately 20-40 cm above the ground surfaces.Collections of litterfall at the MDM, MLG, and MAL sites were conducted throughout a fullyear sampling campaign (Table 2).Collections of litterfall at the MCB site were conducted from September to October 2013, during which the accumulative litterfall biomass fluxes in the four collectors ranged from 214 to 355 g m −2 (mean: 281 ± 59 g m −2 , 1 SD, n = 4), corresponding to approximately 58 % of the annual litterfall mass flux at the sampling site (Zhou et al., 2014).We therefore assume that the measured Hg concentrations in litterfall at the MCB site were representative of the annual means.Litterfall samples were collected monthly, packed into paper bags, and air-dried in a clean environment (i.e., low atmospheric TGM concentration (< 3 ng m −3 ), relatively dry air condition and isolated from precipitation) near the sampling sites.Litterfall samples were milled using a pre-cleaned food blender and stored in the laboratory until Hg analysis.Polyethylene gloves were worn throughout the sample collections and grinding period and the blenders were cleaned with Mili-Q water and ethanol between samples to prevent potential contaminations.
Hg concentrations in litterfall samples were determined using a Lumex RA-915 b multifunctional mercury analyzer (Lumex Ltd., Russia) equipped with a pyrolysis attachment.The Lumex RA-915 b analyzer was routinely calibrated and the QA/QC of the measurements of litterfall samples were controlled by procedural blanks, certificated reference materials (GBW10020, 150 ng g −1 ), and duplicated analyses.The recoveries of certified standards ranged from 88 to 112 % (mean: 98 ± 5 %, 1 SD, n = 60) for plant samples.All samples were measured in duplicates and the mean concentrations were adopted in this study.

Deposition fluxes of Hg in precipitation and litterfall
The volume-weighted mean (VWM) Hg concentration at each sampling site was calculated using Eq.(1): where VWM is the volume-weighted mean of precipitation Hg concentrations in ng L −1 , and Hg i and PD i are the Hg concentration (ng L −1 ) and precipitation depth (mm) of a single precipitation event i, respectively.Annual wet deposition flux of Hg at each sampling site was estimated using the VWM Hg concentration calculated from all samples and the mean annual precipitation depth.Mean litterfall deposition flux of Hg at each sampling site was calculated by averaging the litterfall Hg fluxes obtained for all litterfall collectors, which were calculated by multi- plying the Hg concentration in litters by the corresponding annual litterfall biomass.
3 Results and discussion

Precipitation Hg concentrations and deposition fluxes
Concentrations of Hg in precipitation and corresponding precipitation depth are presented in Fig. 2. A large variation in precipitation Hg concentrations was observed at all sampling sites, with the maximum concentrations up to an order of magnitude higher than the minimum concentrations.The VWM Hg concentrations in precipitation at the remote sites varied from 3.7 to 7.7 ng L −1 (mean: 5.6 ± 2.0 ng L −1 , Table 1), with the highest VWM Hg concentration observed at the BYBLK site and the lowest at the MAL and MDM sites.The VWM Hg concentration in precipitation at the urban site of GY was 11.9 ± 6.1 ng L −1 , which was 1.5 to 3.2 times higher than the values at remote sites (Table 1).We acknowledge that, due to the lack of long-term simultaneous observations, the variation in VWM Hg concentrations in precipitation among the sampling sites may have uncertainties.In the present study, precipitation samples at the urban and rural sites were collected throughout 1 to 3 years, and there might exist interannual variations in VWM Hg concentrations at each sampling site.For example, the maximum annual VWM Hg concentration (5.1 ng L −1 ) at MAL was observed during June 2013-May 2014, which was approximately 1.8 times higher than that (2.9 ng L −1 ) during June 2011-May 2012.At MCB, annual VWM Hg concentrations were highest (8.1 ng L −1 ) during August 2012-July 2013 and lowest (6.0 ng L −1 ) during August 2013-July 2014.VWM Hg concentrations in precipitation at all sites showed a clear season trend, with lower concentrations in the summer wet season and higher concentrations in the winter dry season (Fig. 3).This pattern is consistent with previous observations in rural and urban areas of China (Huang et al., 2012(Huang et al., , 2013;;Ma et al., 2016).Higher Hg concentrations in precipitation during the winter dry season were potentially due to elevated wintertime atmospheric PBM concentrations in China (Fu et al., 2008;Zhang et al., 2013;Xu et al., 2014;Zhu et al., 2014), which could be incorporated into wet deposition via scavenging processes below cloud.Lower VWM Hg concentrations in precipitation during the summer wet season were mostly associated with higher precipitation amounts at the sampling sites, suggesting increasing amounts of precipitation would dilute the Hg concentra-  tions in samples that were scavenged from the boundary layer during the onset of the precipitation (Gratz et al., 2009;Yuan et al., 2015).
Annual fluxes of Hg in precipitation at the sampling sites varied from 2.0 to 12.6 µg m −2 yr −1 (mean: 5.9 ± 3.6 µg m −2 yr −1 , Table 1).Wet deposition fluxes showed a clear urban-rural difference, with the annual deposition flux at the urban site of GY elevated by a factor of 1.8 to 6.3 compared to the values at rural sites.This could be partly due to the elevated VWM Hg concentration in precipitation at the GY site.Wet deposition fluxes at rural sites also showed a clear regional difference.The annual wet deposition fluxes of Hg in the subtropical zones in southwestern and eastern China (i.e., MAL, MLG, and MDM) were relatively higher (by a factor of 1.1 to 1.3) than that at the MCB site in the temperate zone in northeastern China, and much higher (by a factor of 3 to 3.6) than that at the MWLG and BYBLK sites, which were located in the arid/semi-arid zones in northwestern China.This regional variation could not be explained by the difference of VWM Hg concentrations in precipitation because the correlation between annual wet deposition fluxes of Hg and VWM Hg concentrations in precipitation is not significant (p > 0.05).Instead, annual wet deposition fluxes of Hg were positively correlated with annual precipitation depth at the remote sites (r 2 = 0.86, p < 0.01).This suggests that precipitation depth had a greater influence on the regional variation of wet deposition fluxes of Hg at remote sites of China than VWM Hg concentrations, which is in agreement with previous studies in North America (Risch et al., 2012;L. Zhang et al., 2012).
The VWM Hg concentrations in precipitation at the remote sites of this study were overall consistent with previous observations in China (Fig. 4).For example, VWM Hg concentrations in precipitation and wet deposition fluxes of Hg at the Nam Co and SET stations of the Tibetan Plateau and in Mt.Simian, southwestern China, ranged from 4.0 to 10.9 ng L −1 and from 1.8 to 15.4 µg m −2 yr −1 , respectively (Huang et al., 2012(Huang et al., , 2015;;Ma et al., 2016).However, the VWM Hg concentration at the GY site was 1.0-4.4times lower than the levels (12.3-52.9ng L −1 ) observed in other urban areas of China, and the wet deposition fluxes of Hg at the GY site were consequently lower than those (14.0-56.5 µg m −2 yr −1 ) in the urban areas of China, with the exception of the flux observed in Lhasa of the Tibetan Plateau (flux: 8.2 µg m −2 yr −1 ) (Wang et al., 2009(Wang et al., , 2012;;Huang et al., 2013;Xu et al., 2014;Zhu et al., 2014).

Litterfall Hg concentrations and deposition fluxes
Average Hg concentrations in litterfall at the MCB, MDM, MLG, and MAL sites were 47.0 ± 19.0, 42.3 ± 5.6, 91.1 ± 29.4, and 56.9 ± 4.4 ng g −1 , respectively (mean: 59.3 ± 22.0 ng g −1 , Table 2).Concentrations of Hg in litterfall could be affected by many factors, including atmospheric Hg concentrations, tree species, a "leaf maintenance" period, and environmental factors (Lindberg and Stratton, 1998;Frescholtz et al., 2003;Ericksen and Gustin, 2004;Millhollen et al., 2006;Poissant et al., 2008).The variation in litterfall Hg concentrations observed in different collectors (corresponding to sampling of different tree species) at each sampling site was insignificant (p values for all > 0.05, Table 2).Annual mean atmospheric TGM at the MCB, MDM, MLG, and MAL sites was 1.73 ± 0.48, 3.31 ± 1.44, 2.80 ± 1.51, and 2.09 ± 0.63 ng m −3 (Fu et al., 2015b), respectively, which was not significantly correlated with the Hg concentrations in litterfall samples (p = 0.87).The lack of significant correlation might be partly attributed to the fact that mean atmospheric TGM concentration during the "leaf maintenance" period for each tree species was not entirely equal to the annual means of atmospheric TGM at the sampling sites because of its seasonal variations (Fu et al., 2015b).In addition, other factors, including tree species, a "leaf maintenance" period, and regional environmental factors, should also play a more important role in litterfall Hg concentrations.Hg concentrations in litterfall at the MDM site were  (Huang et al., 2012(Huang et al., , 2013(Huang et al., , 2015;;Xu et al., 2014;Zhu et al., 2014;Ma et al., 2016).Black and red values above the columns correspond to VWM concentration and wet deposition of Hg, respectively.
found to increase from July to December (correlation slope = 7.0 ± 0.7 ng g −1 mon −1 , r 2 = 0.78, p < 0.01; litterfall was not collected during January-June due to little production of litterfall biomass).In contrast, significant monthly variation in Hg concentrations in litterfall samples at the MCB and MAL sites was not found (p values for both > 0.05).
Annual fluxes of Hg through litterfall at the four sampling sites ranged from 22.8 to 62.8 µg m −2 yr −1 (mean of 37.0 µg m −2 yr −1 , Table 2).The litterfall fluxes of Hg showed a clear regional distribution pattern, with the fluxes decreasing with latitude.The highest flux (62.8 µg m −2 yr −1 ) was observed at the MAL site in the southern subtropical zone in southwestern China, followed by the MLG site (flux: 39.5 µg m −2 yr −1 ) in the middle subtropical zone, the MDM site (flux: 23.1 µg m −2 yr −1 ) in the northern subtropical zone, and the MCB site (flux: 23.1 µg m −2 yr −1 ) in the middle temperate zone (Zheng et al., 2010).The relatively higher litterfall flux of Hg at the MAL and MLG sites could be explained by either higher annual biomass of litterfall or higher Hg concentrations in litterfall samples (Table 2).Deposition fluxes of Hg through litterfall in this study were comparable to those (35.5-42.9µg m −2 yr −1 ) measured in Mt.Gongga and Mt.Simian, southwestern China (Fu et al., 2010b;Ma et al., 2016), but substantially lower than that (220 µg m −2 yr −1 ) measured at Tieshanping, which was close to Chongqing, southwestern China (Wang et al., 2009).
X. Fu et al.: Atmospheric wet and litterfall mercury deposition

Relative contribution of wet and litterfall deposition to total Hg deposition in forests
Ratios of annual mean litterfall deposition flux to annual wet deposition flux of Hg at the four sampling sites ranged from 3.9 to 8.7 (mean: 5.8 ± 2.3).The ratios were overall consistent with the previous observations (2.8 to 7.6) in China (Wang et al., 2009;Fu et al., 2010b;Ma et al., 2016).On the other hand, the observed ratios in China were much greater than those observed in North America and Europe.Rich et al. ( 2012) collected litterfall at 23 remote sites in the eastern USA and found that the mean ratio of litterfall Hg deposition to Hg wet deposition was 1.3 (ranged from 0.4 to 2.6), which was 3.0 to 6.7 times lower compared to the ratios observed in China.In Europe, ratios of litterfall Hg deposition to wet Hg deposition were in the range of 0.4 to 2.6 (mean: 1.2 ± 0.8; n = 5) (Iverfeldt, 1991;Munthe et al., 1995;Lee et al., 2000;Schwesig and Matzner, 2000).Hg in litterfall biomass has been suggested to be mostly from uptake of atmospheric TGM, and therefore litterfall deposition could be a good indicator of TGM dry deposition to forest ecosystems (Frescholtz et al., 2003;Gustin, 2012;L. Zhang et al., 2012).In addition to litterfall and wet deposition, dry deposition of PBM and GOM to the forest floor and other surfaces could also contribute to the total Hg deposition to a forest.Given the measured atmospheric PBM and GOM concentrations (Fu et al., 2015b;Yu et al., 2015), the dry deposition fluxes of PBM and GOM at the MCB, MDM, and MAL sites were estimated to be 3.0, 9.6, and 2.3 µg m −2 yr −1 , respectively, using the average dry deposition velocities of PBM and GOM over forests modeled by L. Zhang et al. (2012).At the MLG site, annual dry deposition flux of PBM and GOM was estimated to be 4.4 µg m −2 yr −1 using the comparison of precipitation and throughfall data collected side by side (St. Louis et al., 2001;Fu et al., 2010a;Gustin, 2012).The importance of litterfall in the total deposition of Hg has been highlighted by many previous studies (St. Louis et al., 2001;Lindberg et al., 2007;Risch et al., 2012;L. Zhang et al., 2012).In this study, we estimate that litterfall deposition represented 60-87 % (mean: 74.5 ± 11.4 %) of total Hg deposition to the four studied forests, which were much higher compared to those (mean: 46.2 ± 12.5 %) over rural forests in North America and Europe (Munthe et al., 1995;Rea et al., 1996;Grigal et al., 2000;Lee et al., 2000;St. Louis et al., 2001;L. Zhang et al., 2012), whereas the wet deposition played a minor role (mean: 13.9 ± 3.5 %) in the total Hg deposition budget.Therefore, Hg deposition through litterfall played a predominant role in the total Hg deposition budget in forest ecosystems in China.(e) atmospheric gaseous oxidized mercury (GOM) concentrations between China and North America and Europe.Note that atmospheric PBM in Europe is referred to as total particulate bound mercury and in the remaining regions is referred to as particulate bound mercury on particles with an aerodynamic diameter of < 2.5 µm.Data are from this study, the literature, and references therein (EMEP; Munthe et al., 1995Munthe et al., , 2003;;Lee et al., 2000;Schwesig and Matzner, 2000;St. Louis et al., 2001;Pirrone et al., 2003;Swartzendruber et al., 2006;Wang, 2006;Yatavelli et al., 2006;Demers et al., 2007;Lindberg et al., 2007;Valente et al., 2007;Bushey et al., 2008;Choi et al., 2008;Larssen et al., 2008;Li et al., 2008;Fain et al., 2009;Peterson et al., 2009;Prestbo and Gay, 2009;Song et al., 2009;Wang et al., 2009;Engle et al., 2010;Sprovieri et al., 2010;Liu et al., 2011;Fisher and Wolfe, 2012;Fu et al., 2012b;Juillerat et al., 2012;Risch et al., 2012; X. T. Zhang et al., 2012;Chen et al., 2013;Zhu et al., 2014;Fu et al., 2015bFu et al., , 2016;;Ma et al., 2016).(Wang, 2006;Fu et al., 2011Fu et al., , 2015b;;Huang et al., 2012;Wang et al., 2012;Zhu et al., 2012Zhu et al., , 2014;;Huang et al., 2013;Xu et al., 2014Xu et al., , 2015)).

Comparison with observations in other regions worldwide
Figure 5 shows the comparison of wet deposition and litterfall fluxes of Hg as well as TGM/GEM, PBM, and GOM concentrations in China, North America, and Europe.The mean wet deposition flux of Hg at remote sites in China was 5.6 ± 4.2 µg m −2 yr −1 (Fig. 5a, data from this study and the literature, Huang et al., 2012Huang et al., , 2015;;Ma et al., 2016), which is 4.4 times lower than the mean (24.8 ± 17.8 µg m −2 yr −1 ) at urban sites of China (Fig. 5a, data are from this study and the literature, Wang et al., 2009Wang et al., , 2012;;Huang et al., 2013;Xu et al., 2014;Zhu et al., 2014).The mean wet deposition fluxes of Hg in North America and Europe were 9.5 ± 4.2 and 6.8 ± 3.2 µg m −2 yr −1 (Fig. 5a), respectively (EMEP; Prestbo and Gay, 2009).In contrast to the observations in China, the urban-rural variation in the wet deposition fluxes of Hg was insignificant in North America (L.Zhang et al., 2012).
The observations from this study and the literature suggested that wet deposition fluxes of Hg in urban areas of China were highly elevated (by a factor of 2.6 to 3.6) compared to North America and Europe.In China, wet deposition fluxes of Hg were significantly correlated with VWM Hg concentrations in precipitation (r 2 = 0.87, p < 0.01, Fig. 6a), whereas no significant correlation existed between wet deposition fluxes of Hg and annual precipitation depth (r 2 = 0.02, p = 0.65).Elevated wet deposition fluxes of Hg at urban sites of China were associated with the elevated VWM Hg concentrations in precipitation (Fig. 6a).Wet deposition fluxes of Hg in China were also positively correlated with ground-level TGM/GEM, PBM, and GOM concentrations (Fig. 6b, c, and d).Wet deposition of Hg has been suggested to result from the scavenging of PBM and GOM in cloud (i.e., rainout) and below cloud (i.e., washout) (Seigneur et al., 2004;Lin et al., 2006).In North America, a modeling study suggests that scavenging of GOM in and below cloud contributed mostly (∼ 89 %) to wet deposition of X. Fu et al.: Atmospheric wet and litterfall mercury deposition Hg, with ∼ 41 % contributed by washout (Selin and Jacob, 2008).In China, ground-based measurements of GOM in urban areas found that the mean GOM concentrations (means: 47.9 pg m −3 , Fig. 5e, Fu et al., 2011;Xu et al., 2015) were 5.4 times greater than the mean (8.9 pg m −3 ) in North America (Swartzendruber et al., 2006;Yatavelli et al., 2006;Valente et al., 2007;Fain et al., 2009;Peterson et al., 2009;Song et al., 2009;Engle et al., 2010;L. Zhang et al., 2012).Scavenging of GOM in the continental boundary layer (i.e., washout) would therefore contribute to the elevated wet deposition fluxes of Hg at urban sites of China.It should be noted that PBM concentrations were also highly elevated (mean: 239±102 pg m −3 ) in the urban areas of China, which were 10-20 times greater than the levels observed in North America and Europe, and approximately 5 times greater than the mean GOM concentrations at the same locations (Fig. 5d  and e). Lee et al. (2001) estimated that washout of PBM contributed approximately 1.0 µg m −2 yr −1 to the total Hg deposition in the United Kingdom at a background PBM concentration of 10 pg m −3 .Given the mean PBM concentration in urban areas of China, the mean flux of washout of PBM below cloud is roughly estimated to be 24 µg m −2 yr −1 , which explains approximately 90 % of the mean wet Hg deposition flux in the urban areas of China.Since scavenging of PBM below cloud also depends on other factors, including the vertical distribution of PBM, intensity of precipitation, and cloud base height (Tanner et al., 1997;Hicks, 2005;Brooks et al., 2014), the estimate may have large uncertainties.Nevertheless, the estimate is in agreement with the measured fraction of particulate mercury (Hg p ) in wet deposition flux of Hg at an urban site in China.Huang et al. (2013) found that ∼ 86 % of the annual wet deposition of Hg in Lhasa of the Tibetan Plateau was associated with Hg p , much higher than that at a rural site in the Tibetan Plateau (55 %, Huang et al., 2015) as well as at rural and urban sites in North America (26-63 %, Burke et al., 1995;Lamborg et al., 1995;Poissant and Pilote, 1998).These suggest the scavenging of PBM below cloud was an important contributor to the elevated wet deposition fluxes of Hg at urban sites of China.
On the other hand, mean wet deposition flux of Hg (5.6 ± 4.2 µg m −2 yr −1 ) at the rural sites of China was relatively lower (by a factor of 1.2 to 1.7) compared to those measured in North America (9.5 ± 4.2 µg m −2 yr −1 ) and Europe (6.8 ± 3.2 µg m −2 yr −1 ) (EMEP; Prestbo and Gay, 2009).This regional pattern is different from model results that predicted higher wet deposition in China because of large anthropogenic Hg emissions (Bergan et al., 1999;Dastoor and Larocque, 2004).There are several possible explanations for the lower wet deposition fluxes of Hg observed in the rural areas of China.Wet deposition fluxes of Hg at the rural sites of China were mostly observed in arid, semi-arid, and subhumid climate zones in northwestern and northeastern China (i.e., MWLG, MCB, and BYBLK in this study, Nam Co, and SET stations), where the precipitation depth is generally low (260-975 mm, data from this study and the literature, Huang et al., 2012Huang et al., , 2015) ) and anthropogenic Hg sources are scarce (X.T. Zhang et al., 2015).The remaining four rural sites (i.e., MDM, MLG, and MAL in this study, and Mt. Simian, Ma et al., 2016) were all located in mountaintop forests.Although the observations at some of these sites showed elevated PBM concentrations (mean: 31-154 pg m −3 ) (Fu et al., 2015b;Yu et al., 2015), washout of PBM below cloud was not expected to contribute significantly to Hg in precipitation because of low cloud base heights (Ray et al., 2006).In addition, observations of GOM at high-altitude sites in China (i.e., MWLG, MAL, and Shangri-La) showed mean concentrations of 2-8 pg m −3 (Fu et al., 2012a, b;H. Zhang et al., 2015), significantly lower than those (20-87 pg m −3 ) measured at highaltitude sites in North America and Europe (Swartzendruber et al., 2006;Fain et al., 2009;Weiss-Penzias et al., 2009;Fu et al., 2016).Relatively lower GOM concentrations at the high-altitude sites in China were possibly due to the elevated atmospheric particulate matters in China that facilitate the partitioning of GOM to the particulate phase (Slemr et al., 2009;Swartzendruber et al., 2009;van Donkelaar et al., 2010;Amos et al., 2012;Zhang et al., 2013).Since the scavenging of GOM in the free troposphere and continental boundary layer is an important source of wet deposition of Hg (Selin and Jacob, 2008), the lower GOM concentrations in the rural areas of China could be responsible for the lower wet deposition fluxes of Hg observed in the rural areas of China.
2008; Fisher and Wolfe, 2012;Juillerat et al., 2012;Risch et al., 2012), which is the dominant factor causing the difference in litterfall Hg fluxes between China and North America/Europe.It is worth noting that most (five out of the six) observations at the rural sites of China were made in subtropical moist forests, where the litterfall biomass productions are larger than those in the temperate and boreal forests in North America and Europe (Xiong and Nilsson, 1997;Running et al., 2004;Wang et al., 2008).Additionally, mean Hg concentrations in litterfall at the rural sites of China (63.3 ± 29.0 ng g −1 ) were elevated by a factor of 1.4 compared to that (44.0 ± 10.4 ng g −1 ) in North America and Europe (Munthe et al., 1995;Rea et al., 1996;Lee et al., 2000;Schwesig and Matzner, 2000;St. Louis et al., 2001;Lindberg et al., 2007;Larssen et al., 2008;Fisher and Wolfe, 2012;Juillerat et al., 2012;Risch et al., 2012).This could be partly attributed to the elevated TGM concentrations (Fig. 5c) and longer "leaf maintenance" period at most rural sites in China (Frescholtz et al., 2003;Poissant et al., 2008;Gustin, 2012;Fu et al., 2015b).

Conclusions
In this study, we analyzed the concentrations and deposition fluxes of Hg through precipitation and litterfall at multiple sites across a broad geographic area in mainland China.The VWM Hg concentrations and deposition fluxes of Hg in precipitation at rural sites were 3.7 to 7.7 ng L −1 (mean: 5.6 ± 2.0 ng L −1 ) and 2.0 to 7.2 µg m −2 yr −1 (mean: 4.8 ± 2.2 µg m −2 yr −1 ), respectively, lower than the VWM Hg concentration (11.9 ± 6.1 ng L −1 ) and deposition fluxes of Hg (12.6 ± 6.5 µg m −2 yr −1 ) at the urban GY site.Elevated wet deposition fluxes of Hg in urban areas of China corresponded to the elevated ground-level atmospheric Hg concentrations, indicating Hg scavenging below cloud was an important source of wet deposition of Hg in the urban areas of China.The wet deposition fluxes of Hg at the rural sites in this study, however, were lower than the observations in North America and Europe, and different from previous model results, possibly due to a combination of low precipitation depth and low cloud base heights and the low GOM concentrations in the troposphere in China.Hg fluxes through litterfall at the sampling sites ranged from 22.8 to 62.8 µg m −2 , significantly higher (by a factor of 3.9 to 8.7) than the wet deposition fluxes in China and the litterfall deposition fluxes in North America and Europe, indicating that litterfall deposition contributes significantly to the total Hg deposition budget in forest ecosystems in China.

Data availability
Mercury (Hg) concentrations in precipitation and litterfall at the sampling sites were measured by the authors of this paper, and the data sets of rain depth were provided by the filed stations monitored in this study.These data sets can be obtained by contacting Xuewu Fu (fuxuewu@mail.gyig.ac.cn).

Figure 1 .
Figure 1.Locations of the sampling sites in this study.

Figure 3 .
Figure 3. Monthly variations in volume-weighted mean (VWM) Hg concentrations at rural and urban sites in China.

Figure 4 .
Figure 4. Annual VWM Hg concentrations in precipitation and wet deposition fluxes of Hg in this study and the literature(Huang et  al., 2012(Huang et al., , 2013(Huang et  al.,  , 2015;;Xu et al., 2014;Zhu et al., 2014;Ma et al., 2016).Black and red values above the columns correspond to VWM concentration and wet deposition of Hg, respectively.

Table 1 .
Sampling site information, volume-weighted mean (VWM) concentrations and annual wet deposition fluxes of Hg in precipitation at remote sites and an urban site of China.

Table 2 .
Sampling site information, concentrations and annual litterfall deposition fluxes of Hg in litterfall at remote forest sites of China.