Mercury fluxes, budgets and pools in forest ecosystems of China: A critical review

Abstract. Mercury (Hg) accumulation and retention in forest ecosystems play a key role in global biogeochemical cycling of Hg. Especially in China, forests are suffering highly elevated Hg loads. Numerous studies have been conducted to characterize the fluxes and pools of Hg in the terrestrial forests in China during the past decade, which provide insights into spatial distributions and estimate the Hg mass balance in forests through observations at widely diverse subtropical and temperate locations. In this paper, we present a comprehensive review of the research status of forest Hg in China to characterize the Hg budgets and pools. Averaged total Hg (THg) inputs at remote forests and rural & suburban forests in China are about 2 to 4-fold and 2.5 to 5-fold higher than the observed values in Europe and North America, respectively. The highly elevated THg inputs are mainly derived from the elevated atmospheric Hg concentrations. Additionally, production of litterfall biomass is showed to be an important influential factor raising the high Hg inputs at subtropical forests. Compared to the input, THg outputs from the forest ecosystems are relative small, which results in large amount of Hg resided in the forest soils. The annual THg retentions range from 26.1 to 60.4 µg m−2 at subtropical forests and from 12.4 to 26.2 µg m−2 at temperate forests of China, which are about 3.8- to 7.9-fold and 1.2 to 2.8-fold higher compared to those in North America. Given the large areal coverage, THg retention in forest is appropriately 69 t yr−1 in China and is much high than that in global scale estimated by models. The much higher THg retention has elevated the THg pools in Chinese subtropical forests, which poses a serious threat for large Hg pulses remitted back to the atmosphere and additional ecological risks in the forest. The current study has implication for the role of China forests in the global Hg biogeochemical cycle and the optimization of atmospheric Hg transport and deposition models.


Introduction
Mercury (Hg) is considered as a highly toxic heavy metal due to its biogeochemical properties and its toxicity that can affect the health of human and ecosystems (Kojta et al., 2015;Falandysz et al., 2015a;Zhou et al., 2015b;Du et al., 2016).Unlike other heavy metals, atmospheric Hg may exist three operational Hg forms: gaseous elemental Hg (GEM); gaseous oxidized Hg (GOM, also known as reactive gaseous Hg); and particulate-bound Hg (PBM) (Fu et al., 2015).Due to its strong stability and low water solubility, GEM is the most abundant (more than 90%) in the atmosphere and has long resistance time of 0.5-2 years, which can be transported globally and deposited to the remote eco-environment (Gustin et al., 2012;Åkerblom et al., 2015).Atmospheric Hg deposition to terrestrial and aquatic ecosystems plays a significant role in the global biogeochemical cycling of Hg (Zhou et al., 2103a;Blackwell and Driscoll, 2015).Consequently, an understanding of how Hg is transported, deposited and circled the globe is significant for a full understanding and quantifying of Hg biogeochemical cycles (Fisher and Wolfe, 2011).
Atmospheric Hg is nearly the exclusive source of Hg in forest biomass due to the limitation of root uptake (Grigal, 2003).The forest canopy is a major receptor of Hg in terrestrial forest ecosystems, which can absorb Hg through stomatal uptake of GEM, and adsorb PBM and GOM onto foliage surface (Fu et al., 2015).Therefore, atmospheric deposition by litterfall and throughfall is the largest input of Hg to forested watersheds that are not affected by natural geologic or point sources (Blackwell and Driscoll, 2015;Zhou et al., 2016b).Forest ecosystems are considered as a large sink of atmospheric Hg and an active pool of Hg, which is a vital part in the global Hg cycle (Friedli et al., 2007;Zhou et al., 2016a;Ma et al., 2016).Additionally, the large amounts of Hg inputted to the forest are sequestrated in the vegetation and soils, and have produced ecological risks on the bioaccumulation of Hg in the internal forest.For example, Hg sequestrated in the forest soil are considered as potential sources of both total Hg (THg) and methylmercury (MeHg) to downstream aquatic ecosystems (Selvendiran et al., 2008;Ma et al., 2015).Moreover, Hg in the forest soil and biomass can be directly used by forest animals that may be highly vulnerable to the increasing Hg loads (Rimmer et al., 2010;Zhou et al., 2016a).
Dynamic and budget studies quantifying Hg flux and pool in the terrestrial forests are necessary for investigating status of Hg inputs to, retention within, and output from forest ecosystems.Many studies have improved our knowledge of current Hg pools and fluxes (Larssen et al., 2008;Ma et al., 2015;Grigal et al., 2000;Grigal, 2003).
However, terrestrial forest has constantly been underestimated as sink for atmospheric Hg on a global scale (Wang et al., 2016b;Obrist, 2007).Previous reviews were mainly focused on the atmospheric Hg concentrations (Fu et al., 2015), atmospheric Hg depositions (Wang et al., 2016a;Wright et al., 2016) and air-surface fluxes (Zhu et al., 2016); than those of wet inputs, with throughfall ranging from about 1.7 to 2.5 times the wet input (Fig. 1).However, these were different with the North America forests, where throughfall Hg inputs were found to be lower than wet-only depositions in deciduous forests, but to be higher than wet-only depositions in coniferous forests (Wright et al., 2016).
Litterfall Hg inputs have been confirmed to be the other important pathway trapping atmospheric Hg to the forest floor via senesced leaves, needles, twigs, and branches, and other plant tissues.Concentrations of Hg in litterfall could be affected by many factors, such as tree species, lifespan, and environmental factors (e.g., solar irradiation, air temperature, altitude, etc.) (Ericksen and Gustin, 2003;Poissant et al., 2008;Blackwell and Driscoll, 2015b;Zhou et al., 2017a).However, atmospheric Hg concentrations play the most important role in Hg concentrations in litterfall, and Hg concentrations in atmosphere were deemed to be a good indicators of leaf Hg contents in forest areas (Fay and Gustin, 2007;Niu et al., 2011).Based on the available atmospheric total gaseous Hg (TGM) or GEM concentrations and litterfall Hg concentrations in 11 forested areas and 14 pairs of datasets in China, annual mean atmospheric TGM/GEM concentrations were significantly correlated with the THg concentrations in litterfall samples (Fig. 2).The significant correlation might verify that foliage can effectively trap Hg from the atmosphere by accumulation Hg through stomatal uptake of GEM (Fay and Gustin, 2007;Fu et al., 2010a, b;Laacouri et al., 2013;Zhou et al., 2017b).The mean THg and MeHg concentrations in litterfall at remote sites ranged from 12.6 to 135.1 ng g −1 (mean = 54 ng g −1 , n = 12) and from 0.28 to 0.48 ng g −1 (mean = 0.38 ng g −1 , n = 2), respectively (Table 1).
THg and MeHg concentrations in litterfall at rural & suburban areas of China were higher than those in North America and Europe, but litterfall concentrations of THg and MeHg at remote areas were compared those observed in North America and Europe, except in Mt.Leigong, Guizhou Provence (Table 1, Fig. S1).Although Mt.Leigong was relatively isolated from anthropogenic activities with lower GOM, PBM, precipitation and throughfall Hg concentrations, GEM could undergo long-range transport from emission sources.The GEM concentration was 2.80 ng m −3 in Mt.Leigong that is about 170 km to the large Hg mine of Wanshan (Fu et al., 2010a).The relatively higher GEM concentration resulted in elevated litterfall Hg concentrations.
Mean THg inputs by litterfall from 20 forests in China (41.8 µg m −2 yr −1 ) were approximately 2 to 3 times higher than those in Europe over 11 sites (14.2 µg m −2 yr −1 ) and more than 3 times higher than those in North America over 37 sites (12.9 µg m −2 yr −1 ) (Fig. 4).Since litterfall THg inputs to terrestrial ecosystems are estimated by multiplying the biomass and corresponding THg content in litterfall, both of them could influence the input fluxes.Therefore, compared to North America and Europe, higher TGM or GEM concentrations in rural & suburban forests of China resulted in the elevated litterfall Hg concentrations and corresponding higher fluxes in China.However, it should be noted that the litterfall biomass productions in forests of China (565 ± 450 g m −2 yr −1 ) were more than 2-fold higher than those observed in North America and Europe (200 ± 145 g m −2 yr −1 ).The regional differences of litterfall Hg inputs to forest ecosystems was primarily resulted by the factor of litterfall biomasses rather than litterfall Hg concentrations, as evidenced by the much stronger correlation between litterfall Hg input fluxes and litter biomass productions than that with litterfall THg concentrations (Fig. 3a and b).
The total Hg input as the sum of Hg input by litterfall and throughfall (i.e., input flux by litterfall + input flux by throughfall) to forests were ranged from 47.7 to 291.3 µg m −2 yr −1 (n=11 from 9 forests) in China (Fig. 1).Here, it should be noted that the highest Hg deposition (291.3 µg m −2 yr −1 ) was observed at Tieshanping forest from March 2005 to March 2006 (Wang et al., 2009); however, due to overestimation of litterfall biomass, the measured Hg fluxes were more than 3 times the recent studies by Luo et al. (2015a) in 2010-2011and Zhou et al. (2017b) in 2014 -2015.
The much higher Hg input at Tieshanping forest is due to it located near the center of Chongqing City (20 km), the annual atmospheric emissions of which just from coal combustion was 4.97 t (Wang et al., 2006) and Hg pollution was regarded as major environmental burdens in Chongqing (Yang et al., 2009).If we use the updated Hg inputs fluxes by Luo et al. (2015b) at Tieshanping forest, the annually mean total Hg input flux was 73.9 µg m −2 yr −1 (n=10) in China.Hg input to forest floor via litterfall was substantially comparable or greater than the throughfall input and the litterfall to throughfall input ratios range from 0.33 to 6.59 (mean= 2.14), indicating that Hg input via litterfall surpassed that by throughfall and become the major pathway of Hg input to forests in China.The observed ratios in forest ecosystem of China were much greater than those observed in North America and Europe.Ratios of litterfall Hg input to throughfall Hg input to forest ecosystems were in the range of 0.27 to 1.56 (mean=0.89;n = 9) in Europe (Schwesig and Matzner, 2000;Hultberg et al., 1995;Iverfeldt et al.,1991;Larssen et al., 2008;Lee et al., 2000;Munthe et al., 1995Munthe et al., , 1998;;Schwesig and Matzner, 2001), and in the range of 0.60 to 4.13 in North America (mean = 1.37; n = 16) (Blackwell and Driscoll, 2015b;Choi et al., 2008;Demers et al., 2007;Kalicin et al., 2008;Kolka 1999;Grigal et al., 2000;Lindberg et al., 1994;Fisher and Wolfe, 2011;Rea et al., 1996Rea et al., , 2001;;Johnson, 2002;Johnson et al., 2007;Nelson et al., 2007;St. Louis et al., 2001;Graydon et al., 2008), which was about 2.4 to 1.6 times lower compared to the ratios observed in China.The reason is the much higher litterfall biomass production in forest of China as we stated above.
Additionally, more than 90% of Hg in litterfall biomass is considered to be uptake from atmosphere, and throughfall can wash off most of the PHg and RGM on the leaf surface by previous dry depositions; therefore, litterfall and throughfall Hg inputs could be a good indicator of TGM dry deposition to forest ecosystems (Gustin et al., 2012;Zhou et al., 2013a;Fu et al., 2015).Considering dry Hg input in a forest ecosystem as the difference between total Hg input and wet Hg input (dry Hg input = total Hg input -wet Hg input), more than 80% of total Hg inputs were from dry inputs in forests of China, which was higher than those in North America and Europe (70%) but lower than those in Brazil (85%) (Wang et al., 2016).
Higher dry and wet depositions resulted in higher total Hg inputs to Chinese forests, which averaged 78.4 µg m −2 yr −1 at remote forests and 106.5 µg m −2 yr −1 at rural & suburban forests, and ranged from 47.7 to 119.5 µg m −2 yr −1 (n= 5) and from 56.0 to 291.3 µg m −2 yr −1 (n= 6), respectively.We have also reviewed the THg inputs by throughfall and litterfall in the Europe and North America (Fig. 4), and the results showed that THg inputs were significantly lower than those observed in China (p<0.05 for Europe and p<0.01 for North America).Mean THg input was about 39.2 µg m −2 yr −1 (n= 9) in the Europe, which was about 2.0-and 2.5-fold lower than that observed at remote forests and rural & suburban forests in China.Even lower THg input was found in the North America (20.2 µg m −2 yr −1 , n= 17) and was about 4-and 5-fold lower than that at remote forests and rural & suburban forests in China.

Exports from surface runoff and underground runoff
The dominate pathways of Hg output from forest catchments were runoffs and soil-atmosphere exchange fluxes.
The output fluxes of THg and MeHg via surface runoff measured in China are showed in Table 2.The mean THg and MeHg concentrations in surface runoff ranged from 2.3 to 17.2 ng L −1 (mean = 6.0 ± 4.1 ng L −1 , n = 11) and from 0.2 to 0.25 ng L −1 (mean = 0.23 ng L −1 , n = 2), respectively.Comparing to the THg (40.5± 19.6 ng L −1 ) and MeHg (0.32 ng L −1 ) in throughfall, the corresponding Hg concentrations in surface runoffs were seemed much lower, which was consistent with the general concept that forests had the filtering function between atmosphere and hydrosphere (Ericksen et al., 2003;Larssen et al., 2008).The export fluxes of THg via surface runoffs and/or stream waters ranged from 3.0 to 8.6 µg m −2 yr −1 (mean = 4.8 ± 2.6 µg m −2 yr −1 , n = 6).Luo et al. (2014) collected 117 stream water samples in China, including 42 streams from 9 sites in the northeastern forests and 75 streams from 16 sites in the southern forests, and the result showed that THg concentration was higher in northeastern forests (17.2 ± 11.0 ng L −1 ) than that in the southern forests (6.2 ± 6.4 ng L −1 ).The THg concentrations in stream water were positively correlated to DOC concentrations, suggesting that the DOC may facilitate the Hg mobility.Due to cool and dry climate in northern forests, litter decomposed more slowly and resulted in deeper litter and organic layers than those in southern forests (Zhou et al., 2015a(Zhou et al., , 2017a)).Therefore, soil erosion in northern forests with higher DOC in stream waters resulted in higher THg concentrations.
No statistically significant correlations were showed between THg concentrations in stream water and throughfall (r 2 = 0.00, p>0.05, n = 9), and between throughfall Hg inputs and stream water exports (r 2 = 0.03, p>0.05, n = 6), implying that THg output from stream water was regulated directly by processes other than current deposition input in these forested catchments.However, THg export fluxes via runoff and/or stream waters were significantly correlated with THg concentrations in surface soils (organic layer or top 10 cm) (r 2 = 0.52, p < 0.05, Fig. S2).Higher THg depositions have resulted in much higher soil THg concentrations at forest sites of China.Although soils in forests have been suggested as filters between throughfall and stream waters, but THg in stream waters also can desorb from soils (Xue et al., 2013).Yin et al. (1997) suggested that higher Hg concentrations in the water of prefiltration and soils both could be resulted in higher Hg concentrations in the leachate.Therefore, higher soil Hg contents caused by higher deposition at forests of China caused high Hg concentrations in the stream water.Since the adsorption and desorption of THg in soils cloud also depend on other factors, including the soil physical and chemical properties (pH, organic matter, consistency) and leachate properties (pH, dissolved organic matter, salinity) (Yin et al., 1997;Xue et al., 2013;Liao et al., 2009), the deduction may have large uncertainties.
The direct measurements of THg in underground runoffs were not conducted in any forests of China, but they played important roles in the THg export from forests due to both of the amounts and THg concentrations usually higher than those of surface runoffs in subtropical forests (Liu, 2005;Luo et al., 2015b).Several studies have measured THg concentrations in solutions of soil profiles in subtropical forest of Tieshanping, which was averaged 21.8 ng L −1 and ranged from 1.98 to 60 ng L −1 (Wang et al., 2009;Zhou et al., 2015;Luo et al., 2015b).The observed THg concentrations of soil solution was higher than those in five Swiss forest soils, and the reason may be due to higher THg loads and soil THg content in this Chinses forest.Although no studies directly measured the export flux of THg via underground runoff, we roughly estimated the flux based on the THg in soil solutions and runoff amount in Tieshanping forest, which is 6.0 µg m −2 yr −1 ; therefore, the total Hg output by runoffs as the sum of Hg output by surface runoff (3.5 µg m −2 yr −1 ) and underground runoff (6.0 µg m −2 yr −1 ) was 9.5 µg m −2 yr −1 .Table 3 shows the statistical summary of soil-atmosphere Hg exchange fluxes and associated site information in the 30 forest sites.Mean soil-atmosphere Hg exchange fluxes at remote forests were in the range of 1.6-4.77ng m −2 hr −1 (mean = 3.3 ± 3.4 ng m −2 hr −1 , n = 12), and those at rural & suburban forests were significantly higher (T test, p < 0.05) and ranged from -0.8 to 17.8 ng m −2 hr −1 (mean = 8.3 ± 7.1 ng m −2 hr −1 , n = 18).Generally, soilatmosphere Hg exchange fluxes are bi-directional.Nevertheless, only one site showed overall net deposition of −0.8 ng m −2 hr −1 in the wetland of Tieshanping forest and the other forest soils showed overall net emissions in China.

Export of soil-atmosphere exchange fluxes
Many studies have identified factors that correlate with the magnitude and direction of soil-atmosphere Hg exchange fluxes, including atmospheric and soil physicochemical properties.The well-known factors studied in the previous researches influencing soil-atmosphere Hg exchange fluxes included substrate Hg concentration, air and soil temperature, measurement methodology, as well as environmental variables (e.g.forest type, terrain type and soil cover).The most commonly promoting Hg 0 production is solar radiation that is reported with positive correlations in all the studied forests in China (n = 30).The relationship is mainly attributed to photochemical reduction of soilbound Hg, which converts soil Hg 2+ to volatile Hg 0 (Amyot et al., 1994(Amyot et al., , 1997;;Carpi and Lindberg, 1997;Moore and Carpi, 2005;Xin et al., 2007;Zhou et al., 2017b).Photo-reduction is a major driver of Hg 0 generation and evasion from soils (Choi and Holsen, 2009;Engle et al., 2001;Zhou et al., 2015aZhou et al., , 2017b)), although other abiotic and biotic processes also resulted in translation of Hg 2+ to Hg 0 production, including reduction by humic acids (Alberts et al., 1974;Allard and Arsenie, 1991) and iron oxides under anoxic conditions (Lin and Pehkonen, 1997) as well as reduction by microorganisms (Siciliano et al., 2002;Agnan et al., 2016) and/or microbial exudates (Poulain et al., 2007(Poulain et al., , 2004;;Fritsche et al., 2008).Additionally, other important correlation was identified with soil or air temperature, which is also significantly correlated to the Hg 0 production and observed with soil-atmosphere Hg flux in all the forests in China (n=30).Soil temperature was generally stimulated directly to activation energy of Hg 0 (Gustin et al., 1997;Edwards and Howard, 2013) or stimulation Hg 0 evasion by action of soil microorganism activity (Pannu et al., 2014).Agnan et al. (2016) showed that substrate Hg concentration was significantly correlated with soil-atmosphere Hg fluxes across Hg-enriched sites by large global data set (n = 538), but an apparent lack of correlation between substrate Hg concentrations and soil-atmosphere Hg fluxes across all background soils (n = 307) that defined as substrate Hg concentrations ≤ 300 ng g −1 and atmospheric Hg 0 concentrations ≤ 3 ng m −3 .Across all vegetationcovered soils (forest and wetland) of China, the correlation between soil Hg concentrations and soil-atmosphere exchange fluxes also did not show significantly across the entire database (r 2 = 0.02, p > 0.05, n = 25), which was Alternatively, forest areas showed a much narrower range of soil Hg content compared to Hg-enriched substrates, which influenced the fluxes inconspicuously.However, Zhou et al. (2016c) reported strongly positive correlations between soil Hg contents and fluxes at individual forest of Tieshanping subtropical forest (r 2 =0.97, p<0.001) due to the sampling locations that were nearby and have similar other environmental factors.
According to the two-resistance exchange interface model, the exchange fluxes are caused by the gradient of Hg 0 concentrations on both interfaces (Zhang et al., 2002); therefore, high Hg 0 concentrations in the atmosphere will reduce the potential of Hg 0 produced in the soil and diffusion to atmosphere.Laboratory and filed simulation studies showed that elevated atmospheric Hg concentrations significantly inhibited soil Hg volatilizations (Zhou et al., 2017b;Ericksen and Gustin, 2004;Hanson et al., 1995;Poulain et al., 2004).Atmospheric compensation point for Hg 0 flux was firstly presented by Hanson et al. (1995), which is the atmospheric Hg concentration at which no net flux occurs between soil and air (flux to be 0).A previous study using the global database set in background areas showed significant correlation between atmospheric Hg and soil-atmosphere exchange fluxes (p < 0.001, n = 263) (Agnan et al., 2016).In contrast, based on the database combining all forest-covered soils in China, correlation between atmospheric Hg concentrations and soil-atmosphere exchange fluxes was not significant (r 2 = 0.05, p > 0.05, n = 28), which was inconsistent to the concept of the compensation point.The no correlation was contributed to the variations of environmental factors and Hg emissions at forest sites that resulted in a different buildup of GEM/TGM near the surface in the boundary layer.Thus, high soil emissions caused high GEM/TGM concentrations and not vice versa via a control of air GEM/TGM concentrations on soil-atmosphere exchange fluxes.However, in individual forests, studies showed that compensation points at subtropical forests were in the range of 3.89-6.90ng m −3 in Tieshanping forest stands (Du et al., 2014;Zhou et al., 2016c) and 7.75 ng m −3 in Qianyanzhou forest (Luo et al., 2015a), which were much higher than that calculated according to the global database in background sites (2.75 ng m −3 , Agnan et al., 2016).Higher compensation points observed in China also imply that natural surface contribute larger atmospheric Hg pools in China.
Additionally, studies have observed predictable influences of environmental variables on Hg 0 exchange across multiple forests when using consistent measurement methodology, such as significant correlations with air relative humidity (Ma et al., 2013(Ma et al., , 2015;;Du et al., 2014;Luo et al., 2015a).However, it should be noted that the correlation between air humidity and air temperature were also observed, indicating that air temperature may control the air and soil humility.Furthermore, soil moisture stimulated soil Hg emissions at Qianyanzhou and Zhuzhou forests (Luo et al., 2015a;Du et al., 2014) but reduced emissions at Tieshanping forest stands (Du et al., 2014;Zhou et al., 2016c).
Previous studies suggested that soil moisture contributed to TGM flux had optimum interval and should be under intermediate conditions, neither under fairly dry nor very wet (Gustin and Stamenkovic, 2005;Lin et al., 2010;Pannu et al., 2014;Obrist et al., 2014;Zhou et al., 2017b), which can elucidate the different correlations at different forest ecosystems.
Fig. 5 shows the seasonal variations of soil-atmosphere Hg exchange fluxes at forest areas in China.The variations can be classified into two distinct types: evergreen forest and deciduous forest.At evergreen forests, the mean exchange fluxes in warm seasons (summer and spring) were relative higher than those in cold seasons (winter and autumn, t test: p < 0.05 for all).Solar radiation over the forest canopy was much higher in the warm seasons, but the branches and leaves were also luxuriant, so soils received similar sunlight with other seasons at the subtropical evergreen forests (Ma et al., 2013).Therefore, elevated soil-atmosphere Hg exchange fluxes in warm seasons under the evergreen canopy were mainly caused by the increasing soil/air temperature.In contrast, in the deciduous forests, such as larch, mixed broadleaf forest and wetland in Mt.Dongling, the means of soil-atmosphere Hg exchange fluxes were significantly higher in cold seasons (leaf-off period) than that in the other seasons (t test: p < 0.01).Solar radiation was the maximum amount reaching the forest floor during leaf-off periods in winter, which was approximately 300 W m -2 and promoted Hg 0 production.Whereas during leaf-on periods in summer, the maximum solar radiation at the forest floor was only about 116 W m -2 .In summary, our results suggested that soil-atmosphere Hg exchange fluxes are highly dependent on temperature at the evergreen forests, which increased the rate of reduction of Hg 2+ by thermal processes, biological activities and stimulating Hg 0 evasion (Choi and Holsen, 2009;Engle et al., 2001;Poissant et al., 1998;Zhang et al., 2001).In the deciduous forests, the fluxes were similar to evergreen forests during leaf-on periods, whereas the exchange fluxes are dependent on solar radiation during leaf-off periods because that can directly reach to the forest floor.Although soil received direct solar radiation at forests in north China during leaf-off periods that can be lasted for about half a year (November to April), the exchange fluxes displayed a spatial pattern with significantly lower fluxes in the temperate zones in north China than those at subtropical zones in south China (t test, p<0.01) due to lower temperature at temperate zones.Additionally, the remote forests in the temperate zones in north China had similar exchange fluxes to Europe and North America, due to similar forest type, soil properties, TGM concentrations and environmental factors at those forests.However, the fluxes at subtropical zones of remote, rural & suburban forests were generally higher compared to those observed in North America, Europe and South America.The reason may be that forest soils at these areas have higher THg concentrations and receive more solar radiation and causing higher temperature than those at boreal and temperate forests in Europe and North America.

Hg budgets
The ultimate fate of Hg deposited to the forest ecosystem may depend on its delivery and incorporation into the forest floor.Input of THg to the forest fields included net throughfall and litterfall depositions and output pathway from the forest ecosystem included runoff outflow and soil Hg emission back to atmosphere (St. Louis et al., 2001;Fu et al., 2010a).A synthesis of Hg input into and output from forests, we conclude the Hg retentions in forest soils in four subtropical forests in south China, including Tieshanping forest, Mt.Gongga, Mt.Simian and Qianyanzhou forest (Fig. 6a).To identify how the Hg retention in the temperate forests in north China, we have also estimated the budgets in three forest stands at Mt. Dongling in north China (Fig. 6b).
Due to no studies estimated the THg export by underground runoff in China, the underground runoff fluxes in the four subtropical forests in south China was estimated according to the runoff amounts and THg concentrations.
The runoff amount was estimated to 25% rainfall amount (Liu et al., 2005) and THg concentration in runoff was estimated to same as that in Tieshanping due to similar soil THg concentrations in these areas.The estimated export fluxes by underground runoffs were ranged 6.0 to 9.9 µg m −2 yr −1 in the four forests.Base on the budget calculation, the THg retention (= throughfall + litterfall -runoff outflow (surface and underground) -soil-atmosphere exchange fluxes) at the subtropical forests ranged from 26.1 to 60.4 µg m −2 yr −1 , accounted for ranging from 46.6% to 62.8% of THg inputs (Fig. 6a).Evasion of Hg from forest soil was the dominated pathway of THg outputs from the forest compared to runoff outflow.By comparison, the annual loading of THg to subtropical forests in China were much higher compared to some forest catchments in Europe and North America (Larssen et al., 2008;Grigal et al., 2000).
Since atmospheric Hg distributions at subtropical areas indicated rural to suburban areas suffered heavy regional Hg emissions from industrial and urban areas (Fu et al., 2015), we infer anthropogenic emissions caused the elevated loading of Hg to subtropical forests in China.
In a study on Hg input at a remote temperate forest ecosystem in Mt.Changbai, northeastern China, THg concentrations in throughfall was approximately 17 ng L −1 (Wan et al., 2009a).The forest types at Mt. Changbai were similar to Mt. Dongling in Beijing: mixed forest (600-1100 m a.s.l.), coniferous forest (1100-1700 m a.s.l.), and mountain birch zone (1700-2000 m a.s.l.).Additionally, the TGM concentrations were between 1.60 ± 0.51 ng m −3 and 3.58 ± 1.78 ng m −3 (Wan et al., 2009b;Fu et al., 2012), which were comparable with the concentration of 2.5 ± 0.5 ng m −3 at Mt. Dongling (Zhou et al., 2017a).If we hypothesized the THg concentration in throughfall at Mt. Dongling was also similar to that in Mt.Changbai and throughfall amount were estimated through the mean interception of water-holding capacity of canopy measured by Fei et al. (2011).The estimated inputs of THg deposition were ranged from 21.40 to 28.73 μg m −2 yr −1 at Mt. Dongling.As forest types in Mt.Dongling and Changbai are similar, the forest soil types are also similar, which are both mountain brown forest soil (Wang et al., 2013;Zhou et al., 2017a).Therefore, we also referred the Hg concentrations in runoff (5.75 ng L −1 ) at Mt. Changbai (Wang et al., 2013) and runoff volume were used a previous study in the three stands at Mt. Dongling (Fei et al., 2011).Based on our measured THg concentrations in soil solution (9.2 ng L −1 , our unpublished data) and the amounts of underground runoffs in the three stands (Wang et al., 2012), the export fluxes by underground runoffs were estimated.Studies in the Chinese pine plantation, larch plantation and mixed broad-leaved forest found that the annual emission by soil volatilization measured by dynamics chamber and were from 0.87 to 4.03 µg m −2 yr −1 (Zhou et al., 2016c), and the total Hg outputs of which were 3.1, 2.5 and 9.0 µg m −2 yr −1 , respectively.Therefore, the annual net retention Hg from the atmosphere was 21.7 µg m −2 yr −1 for Chinese pine plantation, 26.2 µg m −2 yr −1 for larch plantation and 12.4 µg m −2 yr −1 for mixed broad-leaved forest in north China.The ratios of THg retentions to the THg inputs were much higher than these at subtropical forests (t test, p<0.05), which accounted for 57.9% to 91.3% of THg deposition.However, it should be noted that the Hg input by throughfall and output by runoff have relative greater uncertainties, so the Hg budget in the temperate forest is roughly estimated in the current study.
The THg retention at subtropical forests in south China were about 2.5 times these at temperate forests in north China.If we hypothesis the total input fluxes of Hg were 20.2 and 39.2 µg m −2 yr −1 and output were 11.3 µg m −2 yr −1 (8.6 for soil emission flux, 2.7 for runoff flux) and 8.8 µg m −2 yr −1 (soil emission flux: 6.7, outflow flux: 2.1) for North America and Europe, respectively, according to the average fluxes for each item, the calculated retention were 8.9 and 30.4 µg m −2 yr −1 , respectively.The THg retention at subtropical forests was higher compared to these in North America (3.8 to 7.9 folds) and Europe, and the retention in the temperate forest was lower compared to those in the Europe but higher compared to those in North America (1.2 to 2.8 folds).Highly elevated THg contents in forest top soils were mostly likely originated from atmospheric depositions via litterfall and throughfall, whereas very limited source was originated from geological sources (Obrist et al., 2011).

Hg storage and risk assessment
Table S1 summarizes all studies of soil Hg concentrations and pools at forests of China from the literature.However, it is should be note that the attempts to compare soil Hg concentrations and pools with the data from each other and some other studies are facing difficulties, because these studies either reported the amounts of THg accumulated in different horizons or calculated THg pools stored in soil profiles of different depths, which were inconsistent with each other.
Declining Hg concentrations with soil depth are generally observed in organic to mineral layers and did not vary in the lower mineral soils from all the soil profiles in Chinese forests.Highest THg concentrations observed in litter and upper soils are indicative of Hg sorption from atmospheric deposition to upper soil horizons.As organic soils are net traps of deposited atmospheric Hg and topsoil concentrations reflect recent Hg depositions from the atmosphere, we concluded THg concentrations from topsoil (most in the organic horizons) in the Fig. S3.The soil THg concentration at remote forests averaged 150 ng g -1 and the median concentration was 104 ng g -1 , ranging from 59 to 353 ng g -1 (n = 18).The concentrations were slight higher than those observed in remote areas of North America, which were generally less than 150 ng g -1 for surface soils (Larssen et al., 2008;Obrist et al., 2011;Tabatchnick, 2012).The THg concentrations at rural & suburban forests were much higher than these observed at remote forests, which ranged from 76 to 332 ng g -1 (mean: 198 ng g -1 ; median: 196 ng g -1 ).This is in a good agreement with the elevated atmospheric Hg concentrations and higher loading of Hg in at rural & suburban forests of China, which can be proved by the significant correlation between Hg retentions and soil THg concentrations (r 2 =0.62, p< 0.05, n=7).
Predictably, higher THg depositions and soil THg concentrations has resulted higher THg pools in forest soils.For example, in the remote forests of Mt.Gongga and Mt.Ailao, the THg storage were up to 152.3 and 191.3 mg m -2 in the soil profiles of 90 and 80-cm depth, which were much higher than these in the upland forest of central Adirondack

Mountain of USA and (64 mg m
-2 in 0-90 cm depth) (Selvendiran et al., 2008) and upland forest of Steinkreuz, Germany (19 mg m -2 in 0-60 depth) (Schwesig and Matzner, 2000).However, THg storage in forest soils of temperate forests and Tibet Plateau with relative lower atmospheric Hg deposition (Zhou et al., 2017a;Gong et al., 2014), were comparable to that in North America and Europe.

Hg storage in biomass
Vegetation is known to exert significant influence the dynamics of Hg in the forest ecosystem including atmospheric Hg input and output in the terrestrial ecosystem (Ma et al., 2016;Zhou et al., 2016a).Two studies investigated the Hg distribution in the tissues of vegetation at the subtropical forest (Tieshanping forest, Zhou et al., 2016) and temperate forest (Mt.Dongling, Zhou et al., 2017a) and showed that the THg concentration followed the order of Oa > Oe > Oi > litterfall > leaf/needle > root > bark > branch > bole wood for each species.Highest THg concentrations are observed in the O horizons compared to THg in the other biomass, because organic matter was enhanced during natural processes of litterfall decomposition and transformation, in which organic matter binding Hg compounds are usually more stabilized via complexing, humification and adsorption to clay minerals (Demers et al., 2007;Pokharel and Obrist, 2011;Zhou et al., 2017a).Sequentially, relative higher THg was observed in the litterfall and leaf due to canopy leaf can effectively capture Hg in atmosphere, which can uptake Hg by stomata (Fu et al., 2015).
Root is contacted with mineral soil directly, likely to higher concentration than that of aboveground wood (Grigal, 2003).THg concentrations in roots of Norway spruce in southern Sweden were 40 ng g -1 (Munthe et al., 1998), which was much lower than that in the root of Masson pine in southwestern China (71 ng g -1 , Zhou et al., 2016a) due to large THg loading in this area.Mass of tree roots is about one-fifth that of aboveground material (Wharton and Griffith, 1993;Whittaker and Marks, 1975) and combined with high THg concentration, roots may store much higher THg biomass compared to other plant components.However, data are rare for these pools of THg at forests.Only Zhou et al. (2016) estimated the THg pools in roots that accounted for about 34% of the overstory THg pools.Bole wood had the largest biomass of vegetation in the forest, but lowest THg concentrations were observed.A previous study suggested that the source of the THg in wood was translocated from foliage (Barghigiani et al., 1991).
Concentrations of Hg were positively correlated in 11 pairs of leaf and adjacent bole wood samples of different tree species at forests of China (Fig. S4).It is reasonable for their correlation because leaf and bole wood are both exposed, one directly and the other indirectly to the same atmospheric pool of Hg.Grigal (2003) suggested that THg in bark is probably from long-term dry deposition, and they summarized 15 pairs of bark and adjacent wood-only samples and found significant correlations.However, no significant correlation was observed between THg concentrations in bark and bole wood or leaf, probably due to that the THg accumulation rates were differed in the barks of different tree species.
THg concentrations of each component at the suburban forest of Tieshanping at subtropical zone was much higher than those at the remote forest of Mt.Dongling at temperate zone.Accordingly, much higher THg pool of 103.5 mg m -2 showed in suburban forest of Tieshanping than that of 7.3-10.8mg m -2 in remote forest of Mt.Dongling (Fig. S5).The THg pools in North America were much lower than those at subtropical forest of China and comparable to those at temperate forest of China (Friedli et al., 2007;Obrist et al., 2009;Richardson et al., 2013).Nonetheless, soil THg pools accounted for over 90% of the total ecosystem Hg pools forests around the world.For example, over 97% and 99% of the THg resides in soil layers (0-40 cm) at Mt. Dongling and Tieshanping forest in China; more than 99% of the THg pool were stored in the soil depth of top 60 cm at the coniferous and deciduous upland forest in Vermont, USA (Richardson and Friedland, 2015); THg pools at upland forest in Sierra Nevada, showed soil of top 40 cm constituted over 94% of the total ecosystem Hg storage (Obrist et al., 2009); THg pools in the soils exceed more than 90% of the total ecosystem Hg pools at Sierra Nevada forest (Engle et al., 2006;Obrist et al., 2009); and THg resided in organic soils accounted from 93 to 97% of ecosystem THg at two subtropical forest stands in Canada (Friedli et al., 2007).

Risk assessment
The studies summarized in this review showed significant inputs and retention of Hg in forest ecosystems in China.The apparent accumulation and storage of THg may present an important ecological risk.Firstly, the Hg in forest soil could be re-emitted back to the atmosphere.Organic matter has a high binding ability of Hg in forest surface soils, but the Hg bonded organic carbon would probably be released to the environment as the decomposition of organic matter occurs.Studies on climate change showed that the accelerated global warming would accelerate the decomposition of organic carbon (Schimel et al., 1994), which could probably accelerate Hg emission from soil (Obrist, 2007;Fu et al., 2010a).Additionally, the increasing of global temperature would aggravate the occurrence of potential fires and causing large pulses of Hg to the global atmospheric pool (Zhou et al., 2016a(Zhou et al., , 2017a)).The average THg emission from forest wildfires was 0.78 t yr -1 during the first decade of this century in China (Chen et al., 2013), which was accounting for about 12.8% of total Hg emissions from biomass burning.Zhou et al. (2016a) estimated the THg emission from the subtropical forest of Tieshanping was about 0.82 mg m -2 , which was lower than the mean value of 1.22 mg m -2 (range: 0.68-1.70mg m -2 ) in the temperate forest of Mt.Dongling (Zhou et al., 2017a).
In contrast, the THg pools in the fuel biomass were much higher at the subtropical forest compared the temperate forest as we showed in the above section.Therefore, it should be noted that THg emission rate from different plant components and soil layers differed greatly due to combustion completeness that is defined as the ratio of THg combustion in China (Zhou et al., 2016a(Zhou et al., , 2017a)).
Secondly, the Hg retention in the forest soils would accumulate through food webs, threatening the balance of forest ecosystems (Rimmer, 2010).However, the relevant studies in China were rare.Many studies showed that mushroom had high accumulation ability of THg and MeHg from substrate (like soil, litter and wood) and strong translocation to the fruiting bodies (Fischer, et al., 1995;Árvay et al., 2014;Falandysz et al., 2015aFalandysz et al., , b, 2016;;Ostos et al., 2015).Studies in southwestern China showed that THg concentrations in the Fungi Boletus species and genus Leccinum species were up to 3500-4800 ng g -1 (mean 42000 ng g -1 ) and 4900-22000 ng g -1 (10900 ng g -1 ) dry matter, respectively (Falandysz et al., 2015a, b).Similarly, a study in Poland also showed efficient accumulation of THg in the Leccinum mushrooms, but the average Hg concentrations being an order of magnitude lower because of lower concentrations of THg in surface forest soil of Poland.Although some lowly cumulative species of mushroom were observed in the subtropical forests (Kojta et al., 2015;Wiejak et al., 2014), mushroom is an important food item in southwestern China, and high rates of consumption can deliver relatively high doses of Hg to local human beings (Kojta et al., 2015;Falandysz et al., 2015aFalandysz et al., , b, 2016)).If according to the value of the provisionally tolerable weekly intake (PTWI) or the reference dose (RfD), the most edible mushrooms from Yunnan provide a high dose of Hg when consumed at a rate higher than 300 g per week, which will post a higher health risks to consumers (Falandysz et al., 2016).
Additionally, the ecological stress to forest insect were investigated in an suburban forest (Tieshangping) in China, which showed that insect living in the soil has two to three orders of magnitude higher THg accumulation than that living on the plant due to large Hg pools in the forest soils (Zhou et al., 2016a).Although animals in the high position of the food chain were not studied in forest of China, Rimmer et al. (2010) showed that food web reflected the transfer of Hg from lower to higher trophic levels with a resulting increase in Hg burden.Therefore, we can infer that Hg will be seriously bioaccumulated along the food chain and pose risk to the local creatures by physiological toxicity.

Environmental implication and research needs
The estimates the mean dry deposition of 12.3 mg m -2 yr -1 , which converted to the total Hg deposition in China is <121.0t yr -1 (Wang et al., 2014).Given that more than 80% of the THg deposition was from dry deposition, the THg dry deposition was 153.8 t yr -1 in forest ecosystems of China, which is even higher than the total Hg deposition in the whole mainland China.Therefore, a large underestimation compared to the observation-based estimate just from forest areas of China in this study.Therefore, future model studies should consider the THg dry deposition in forested areas individually.
Hg sequestrated in forest litters and surface soil by legacy Hg retention can be quickly volatilized to the atmosphere by soil-atmosphere exchanges.Recent global Hg models suggested that soils not only act as net sinks but also as net sources for atmospheric Hg in global Hg cycling (Amos et al., 2013), and the role of forest ecosystems as atmospheric Hg sink or a source are existing confliction (Lindberg et al., 1991(Lindberg et al., , 1998;;Pirrone et al., 2010;Gustin et al., 2008).Using the global database of terrestrial surface−atmosphere fluxes, forest ecosystems appear a net deposition of 59 t yr -1 , but the estimation existed large uncertainties and ranged (37.5th−62.5thpercentiles) from a deposition of 727 t yr -1 to an emission of 703 t yr -1 (Agnan et al., 2016).Base on the field observations of THg retention in Chinese forests, the THg retention in forest soils was 69 t yr -1 just in China, which was much higher than the global data of 59 t yr -1 (Agnan et al., 2016).Such difference is mainly resulted from the variation of reported atmospheric Hg uptake by foliage and the limited geospatial representation of available data (Wang et al, 2016;Zhu et al., 2016;Agnan et al., 2016).Thus, more studies should be conducted to character the whole-ecosystem fluxes and to question to what degree the ecosystems are net sinks or sources of atmospheric Hg.
To better assess the role of forest ecosystems in the global Hg cycling, it is also essential to understand the THg pools in the branches, stems and roots that can be translated from the atmosphere by the foliage uptake.A previous study estimated that approximately 139 t yr -1 Hg was stored in bole woods (Obrist et al., 2007).However, there is no study exactly quantifying the amount of Hg translocation after Hg uptake by leaves, and the THg storage in biomass are scarce and need more data.Further studies concerning the transformation and migration processes after vegetation uptake will benefit to constrain atmospheric Hg sink in forest ecosystems.
In addition, the large '' active'' soil pool at forests is a potential short-term and long-term source of THg and MeHg to downstream aquatic ecosystems (Selvendiran et al., 2008;Ma et al., 2015).However, there is no study reporting the accumulation of THg and MeHg in aquatic ecosystem after output from the forest ecosystem.The data become available.

Conclusions
As the largest energy consumer and anthropogenic Hg emission in the world, much attention has been paid to characterize the behavior of Hg in China.Forests are regarded as large pools of Hg in the global Hg cycle.In this paper, an integrated review of the knowledge reported in peer-reviewed literature is provided.Hg deposition and pools have been found to be substantially elevated in both remote, rural & suburban forests of China compared to those observed in North America and Europe.A strong spatial variation in Hg pools was observed, with high storage related to regional atmospheric Hg concentrations in southern China.The large Hg storage in the forests pose a serious threat for large pluses to the atmospheric Hg during accelerated organic matter decomposition and potential wildfires, and additional ecological stress to forest animals.However, very few studies are attempted to study the ecological risk of Hg in the forest ecosystem in China, which are suffering highly Hg depositions.
The forests play important roles in the geochemical cycles of Hg in China.According to the budget calculation, the THg retention ranged from 26.1 to 60.4 µg m −2 yr −1 at the subtropical forests in southern China, accounted for ranging from 46.6% to 62.8% of THg inputs, and ranged from 12.4 to 26.2 µg m −2 yr −1 at the temperate forests in northern China.The Hg retention and storage pools in at the subtropical forests were much higher than those in North America, but those in the temperate forests were comparable to Europe and North America.The result of the current review may answer the question to what degree the ecosystems are net sinks or sources of atmospheric Hg in China.
Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2017-794Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 23 April 2018 c Author(s) 2018.CC BY 4.0 License.consistent with the global database set in background soils (Agnan et al., 2016).The lack of correlation between substrate Hg concentrations and soil-atmosphere Hg fluxes may indicate either little control of soil Hg content on the exchange fluxes across forested areas, or that other parameters prevailed over the effects of soil Hg content.
concentration loss by wildfire to THg concentration before burn(Melendez-Perez et al., 2014).Due to the large amount of THg retention in Chinese forests, a hectare of forest combustion equals about from 104.4 to 261.5 t coal Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2017-794Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 23 April 2018 c Author(s) 2018.CC BY 4.0 License.
large THg retention of in the forest ecosystem suggested strong adsorption and absorption of Hg by vegetation that was underestimated by global modeling of previous studies.If we roughly estimated the THg deposition at forests of China using the average THg depositions (92.45 µg m −2 yr −1 ) by present studies and the forest area (2.08×10 12 m 2 ) in 2015, the THg deposition would be 192.3t yr -1 in forest areas of China.GEOS-Chem model Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2017-794Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 23 April 2018 c Author(s) 2018.CC BY 4.0 License.
processes of Hg methylation, transformation and translocation may be different from those in North America and Europe because of the larger Hg deposition and storage in China, which requires further investigation when more Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2017-794Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 23 April 2018 c Author(s) 2018.CC BY 4.0 License.

Fig. 2 .
Fig. 2. Relationship analysis between the GEM or TGM concentrations verses the litterfall Hg concentrations for field trap measurements.

Fig. 4 .
Fig. 4. Box chart for Hg inputs to forest ecosystems in China, Europe and North America.

Fig. 5 .
Fig. 5. Box chart for soil-atmosphere Hg exchange fluxes in deciduous and evergreen forest ecosystems in China (CHI, including four seasons), North America (NA), Europe (Eur) and Brazil (Bra).