Mercury Exchange Flux from Two Different Soil Types and Affecting Parameters

Mercury exchange fluxes between atmosphere and soil surface were measured in two different types of soils; lawn soil (LS) and forest soil (FS). Average Hg emission from LS was higher than from FS although the soil Hg content was more than 2 times higher in forest soil. In LS, Hg emissions were much greater in warm season than in cold season; however, deposition was dominant in FS during warm season because of leafy trees blocking the solar radiation reaching on the soil surface. In both LS and FS, Hg fluxes showed significantly positive correlations with UV radiation and soil surface temperature during cold season. In addition, it was observed that emission showed positive correlation with UV radiation and soil temperature while there was negative relationship between deposition and UV radiation.


INTRODUCTION
Mercury (Hg) has been considered as a toxic pollutant.It differs from other heavy metals in that it continuously goes through the deposition and re-emission cycle in the environment because of its high vapor pressure (Poissant et al., 2000).Therefore, it is difficult to distinguish between re-emitted Hg that was previously deposited from atmosphere and Hg emitted from natural sources.According to the recent report by UNEP (2013), re-emitted Hg from soil and vegetation accounts for about 19-51% of the total Hg currently being emitted and re-emitted to the atmosphere from all sources.When Hg is deposited on soil surface through wet and dry depositions it remains in the soil as inorganic Hg(II) compounds (Schuester, 1991), evades to lakes or oceans through runoff (Ahn et al., 2010), volatilizes to atmosphere (Barkay et al., 2003;Schlüter, 2000;Poissant and Casimir, 1998), or undergoes transformation to methylated Hg, the most toxic form (King et al., 2002).Hence, Hg exchange between atmosphere and soil is a critical step to track the fate of Hg in biogeochemical cycle.
Since the soil conditions are typically favorable for the formation of Hg(II) compounds Hg should be reduced to Hg(0), the most volatile form, before volatilization.Previous studies showed that Hg emissions from soil surfaces are affected by various environmental parameters including solar radiation, temperature and soil wetness especially in controlled environments (Corbett-Hains et al., 2012;Fu et al., 2012;Kocman and Horvat, 2010;Choi and Holsen, 2009a;Yang et al., 2007;Xin et al., 2007); however, there are also many studies which did not observe any clear relationship between Hg exchange flux and those parameters or did find even opposite trends.Generally, it has been shown that Hg emission from soil surfaces are positively correlated with solar radiation and temperature, but in a few studies Hg emissions were significantly enhanced during the nighttime or on cloudy days (Zhang et al., 2008;Engle et al., 2005).Some studies found a positive correlation between Hg emissions and soil humidity (Gustin and Stamenkovic, 2005;Bahlmann et al., 2004;Scholtz et al., 2003) while others observed the opposite results (Choi and Holsen, 2009a;Gabriel et al., 2006;Wallschlager et al., 2000;Leonard et al., 1998).Other influencing parameters on Hg exchange flux include Hg contents and speciation in soil (Corbett-Hains et al., 2012;Fu et al., 2012;Kocman and Horvat, 2010;Choi and Holsen, 2009a;Yang et al., 2007;Xin et al., 2007), atmospheric Hg concentration (Engle et al., 2001), atmospheric oxidants (Engle et al. 2005), and soil properties (organic matter, pH, cation exchange capacity).There are also a number of studies which identified the biological effect on Hg reduction in soil, but it is not conclusive (Barkay et al., 2003;Schlüter et al., 2000).
To date there have been only a limited number of studies carried out to determine the combined effects of the parameters that influence Hg emissions from soil.In Korea, there was only one study to estimate Hg emission flux from soil surfaces using micrometeorological gradient method in Seoul (Kim and Kim, 1999).This study is the first study that directly measured the Hg exchange flux between soil and atmosphere using dynamic flux chamber (DFC) in Korea.In this study, Hg fluxes were measured from two different types of lands including forest and lawn in cold and warm seasons, and compared with various environmental parameters.

1 Measurements
Air-soil Hg fluxes were measured from the forest soils (FS) and lawn soil (LS) in the campus of Kangwon National University in South Korea using dynamic flux chamber (DFC).The lawn site was completely covered by about 3 cm tall grasses, and the mixed forest site (deciduous and coniferous forest canopy) was covered by a 1~2 cm of fallen leaves.The sampling periods were Feb. 22~25 and May 28~30, 2011 for the lawn site, and Feb. 25~27 and June 13~15, 2011 for the forest site, including cold and warm seasons.
A cylindrical polycarbonate DFC (3.78 L) was used in this study similar to that used by Choi andHolsen (2009a, 2009b).The diameter of the DFC opening over the soil surface was 18 cm, and four inlet holes (1 cm diameter) were evenly distributed around the chamber wall to insure the chamber was well mixed.A detailed description and diagram can be found in Choi and Holsen (2009a).The edges of chambers were placed 3 cm under the soil surface, and deployed one week before measurements started for each sampling period.Hg concentrations in inlet (C in ) and outlet (C out ) were alternatively measured using Tekran 1110 synchronized two-port sampling unit and Tekran 2537B analyzer (cold vapor atomic fluorescence spectrophotometry), and 10 min average of C in and C out were measured to calculate the Hg emission flux every 20 min using equation (1).
where F is Hg emission flux in ng m -2 h -1 , C out and C in are concentrations of Hg at the outlet and inlet in ng m -3 , respectively, Q is the flushing flow rate through the chamber in m 3 h -1 , and A is the surface area of the soil exposed in the chamber in m 2 .The optimum flushing rate, Q of 5 Lpm was chosen since the best consistent recovery of standard Hg vapor was yielded as in Choi andHolsen (2009a, 2009b).A positive value of F indicates the net Hg emission from the soil surface while a negative value indicates the net Hg deposition to the soil surface.It should be noted that the transmission of UVA, UVB and UVC through a polycarbonate chamber are 88, 28, and 0%, respectively (Skinner, 1998); therefore, there is a possibility that Hg emission flux was underestimated in field measurements (Carpi et al., 2007).LS and FS were collected 5~7 cm below the surface to measure soil properties and Hg contents in soils.After collected soil was freeze-dried and homogenized, the total Hg content in soil was measured using a DMA 80 automatic Hg analyzer (Milestone, Inc.).Soil organic matter and organic carbon were determined by loss on ignition (LOI) (Nelson and Sommers, 1996).Soil pH was measured by soaking in distilled water.Soil particle size distribution was measured using the hydrometer method (Bohn and Gebhardt, 1989;Bouyoucos, 1962).

2 Environmental Parameters
Environmental parameters including soil surface temperature and soil relative humidity were measured every 5 min on site using a HOBO data logger (U23-001, onset, INC., USA) which was placed 3 cm under the soil surface.UVA was measured with a UV-A light meter (YK-37UV3D, Lutron, Inc., Taiwan).Meteorological data including solar radiation, air temperature, and atmospheric relative humidity were obtained from Chuncheon meteorological observatory in Korea where is located approximately 3.46 km north of the sampling site.Ozone (O 3 ) concentration was obtained from national monitoring site located approximately 1.5 km south of the sampling site.

3 QA/QC
The Tekran 2537B was calibrated with its internal permeation source before the sampling.It was also manually calibrated using injection of Hg 0 saturated vapor.R 2 was ¤0.9995 using five point manual injection, and the recoveries were 108.1±1.6% and 102.5 ±2.7% for cartridge A and B, respectively.RSD (relative standard deviation) was 3.3±1.3%from 7 injections of Hg 0 saturated vapor.The concentration difference between inlet and outlet was less than 5% when DFC was placed on the baked aluminum foil with no light.
For analysis of Hg content in soils, a standard curve was obtained (0.5 ppb to 2 ppm) and R 2 was ¤0.999.SRM (standard reference materials: MESS 3, marine sediment, NIS) was used to calculate the accuracy and precision, and recovery ranged from 102 to 108%.The method detection limit (MDL) was 0.029 μg kg -1 , cal-culated as three times the standard deviation of seven sequential analyses of SRMs.

1 Basic Properties of Soils
The Hg content in the FS (77.3 μg kg -1 in average) was more than 2 times higher than that in the LS (34.8 μg kg -1 ), and there was no difference between the soils collected during different times of the year (Table 1).Organic matter (OM) and organic carbon (OC) content were also higher for FS than for LS.Soil OM has been suggested as effective binding site for Hg(II), resulting in a positive correlation between Hg content and OM content in soil (Yang et al., 2007;Meili, 1991).The contribution of sand was higher for FS than for LS, indicating that the porosity of FS was probably greater than for the LS.The average pH was 6.6 and 4.6 for the LS and FS, respectively.

2 Hg exchange in Lawn Soil
Hg emission-deposition fluxes varied significantly with sampling location and time.Twenty-min fluxes were smoothed using the Savitzky-golay method in this study (Savitzky and Golay, 1964).For the lawn soil (LS) the Hg emission was statistically larger in May, 2011 (7.8±12.8ng m -2 hr -1 ) than in February 2011 (0.2±5.4 ng m -2 hr -1 ) (Fig. 1).In February the soil temperature dropped to nearly 0� C at night in the lawn, resulting in low emission fluxes and the occurrence of frequent deposition as suggested in previous studies (Corbett-Hains et al., 2012;Choi and Holsen, 2009a).On the other hand, most of Hg flux occurred as emission in May when soil temperate ranged 15 to 33� C and UV-A radiation increased to 2.8 mW cm -2 .In February, Hg emission flux generally increased with increases in UVA radiation and soil temperature during daytime; however, there was no distinct diurnal variation observed in May (Table 2, Fig. 1).Previous researches suggested that Hg emission flux from soils was positively correlated with soil surface temperature and solar radiation especially in laboratory investigation (Kocman and Horvat, 2010;Lin et al., 2010;Choi and Holsen, 2009a); however, it was also often observed that there was no significant relationship between Hg emission flux and environmental parameters in field experiments (Stamerkovic et al., 2008;Ericksen et al., 2006).For example, a peak of Hg emission flux (58.1 ng m -2 hr -1 ) occurred from LS during early morning (around 2 am) on May 29 th when soil temperature and UV-A radiation were low; however, early morning peak was not shown on May 30th (Fig. 1).Several studies also found a significant increase of Hg emission at nighttime (Zhang et al., 2008;Engle et al., 2005;Zhang and Lindberg, 1999) and suggested the role of atmospheric oxidants as a possible reason.In this study, there was no positive relationship between O 3 concentration and Hg emission flux throughout the whole sampling period; however, generally higher ozone concentrations were observed during 1-11 am on May 29 th (37.1 ppb) when Hg emission flux increased (Fig. 1) than those during the same time on May 30th (19.2 ppb).According to Engle et al. (2005), O 3 showed similar influence on increasing Hg 0 emissions from substrates to light especially when the substrate was relatively dry.Soil RH during early morning on May 30 was significantly lower than on May 29, suggesting that both low soil RH and high atmospheric O 3 concentration could derive the increased Hg emission flux from soil surface.The chemical mechanisms that produced the trends between O 3 and Hg emission are not known but potentially involve heterogeneous reactions between O 3 , the substrate, and Hg (Engle et al., 2005).
In addition, Poissont and Casimir (1998) found that there was a positive correlation between water vapor flux and Hg emission flux, suggesting that water vapor is a good tracer gas to derive the turbulent mixing coefficient for Hg.The atmospheric RH even exceeded the RH in soil during early morning on May 30, possibly limiting the volatilization of water vapor and associated Hg from the soil surface in this study.On the other hand, Erickson et al. (2006) indicated that increase of Hg emission was the result of the soil Hg pool replenishing in the absence of light.These results indicate that the emission-deposition process is affected by many multiple and interacting factors; therefore, it did not often show strong correlation with just one individual parameter in this present study.

3 Hg Exchange in Forest Soil
In forest soil (FS) the Hg emission fluxes were higher in winter (Feb. 25-28, 2011) than in summer (June 13-15, 2011) (Fig. 2), which was opposite of those observed in LS (Fig. 1).Also, although the Hg content in FS was more than 2 times higher than in LS, Hg emission flux was not observed to be higher in FS.Organic matter has been suggested to either facilitate or inhibit Hg reduction in previous researches, but it typically shows the extremely high affinity with Hg, resulting in low availability/mobility in soil (Mauclair et al., 2008;Skyllberg et al., 2006) while Hg reduction often increases with an increase of organic matter in natural water (Ahn et al., 2010;Zhang and Lindberg, 2001;Allard and Arsenie, 1991).In this study, higher organic matter content in FS possibly reduced the Hg emission despite of higher Hg content in soil than in LS.In addition, UV radiation was relatively low for the FS due to the cloudy condition during sampling, and fal-len leaves and vegetation left inside chamber probably limited the solar radiation reached on the soil surface.During winter, high Hg emission flux was observed when soil temperature and UV-A radiation increased; however, throughout the sampling, the variation of Hg emission flux did not clearly follow the variation of temperature and/or UV radiation for the forest site (Fig. 2).When Hg emission flux was measured in Adirondack forest, USA (Choi, 2007) its correlation with temperature or solar radiation was not observed in coniferous forest while there was a significant correlation in deciduous forest, and it was suggested that the bacteria in the coniferous soils primarily engage in the oxidation of Hg rather than reduction of Hg (Smith et al., 1998).The forest in this study was mixed with coniferous and deciduous trees, possibly resulting in unclear correlation of Hg emission flux with temperature or UV radiation.In addition, there is a possibility that bacteria can alter soil properties including soil In June Hg flux mostly appeared as deposition direction despite of high soil temperature (Fig. 2).UV radiation reached on soil surface was zero for most of times because the forest canopy was shading the forest floor, but the direction of flux changed to emission from deposition when UV radiation increased to 1.3 mW cm -2 during a very short time in the afternoon on June 14 th (Fig. 1).Soil surface temperature also increased during that short time; however, it increased again in afternoon of June 15 th when Hg emission flux and UV radiation were not observed (Fig. 2).This result indicates that radiation is indeed an important factor for Hg emission from the soil surface.Increased Hg deposition to soil surfaces during summer was also reported in previous researches (Choi and Holsen, 2009a;Choi, 2007).
In previous studies, Hg exchange fluxes between atmosphere and soil were measured in different soil types.In Korea, there was only one study to estimate the Hg exchange flux, and the average emission flux was 96.0 ng m -2 hr -1 (Kim and Kim, 1999), a much higher flux than those found in this study (Table 2).In the study of Kim and Kim (1999), Hg exchange fluxes were measured using gradient measurement method while DFC was used in this study.Also, the sampling site of Kim and Kim (1999) was located in the residential area of the metropolitan (Seoul), which probably caused greater Hg accumulation in soil due to the nearby anthropogenic Hg sources when compared with the sites in this study.Hg emission fluxes in forest soil were consistently less than those in bare soil in Guangdong and Mt.Gongga in China while the Hg concentration in soils were comparable between bare and forest soils (Table 2), indicating that Hg emission flux cannot be estimated only by soil Hg content.Higher Hg emission fluxes in LS than in FS were also observed in this study although the soil Hg contents showed the opposite order.The Hg emission fluxes in LS in this study were generally higher than those observed in USA and Canada, but clearly lower than in urban sites in China (Guangdong and Guiyang, China).

4 Relationship with Environmental
Parameters Relative frequency distribution of hourly soil Hg flux in FS and LS for each season was shown in Fig. 3.In FS, the bulk of the soil Hg flux measurements were between -0.5 and 0.5 ng m -2 hr -1 in February; however, most of Hg exchange was shown as depositional flux in June because the forest canopy was shading the forest floor, resulting in almost no solar radiation reached the soil surface.The lowest minimum hourly Hg flux was -2 and -12 ng m -2 hr -1 in FS in February and June, respectively.Frequency distribution in LS was very similar to FS in February, showing -0.5 -0.5 ng m -2 hr -1 as a dominant flux (Fig. 3).On the other hand, significant shift to the emission flux was observed in warm season (May) in LS, showing the opposite seasonal trend to FS.In May, the emission flux higher than 15 ng m -2 hr -1 was dominant in frequency distribution in LS, having the highest maximum hourly Hg flux of 33 ng m -2 hr -1 .
In order to identify the relationship of Hg flux with environmental variables the Pearson correlation coefficients were calculated.During winter (Feb.2011), Hg flux showed significantly positive correlation with UV radiation and soil surface temperature in both sites (Table 3), suggesting that UV radiation and/or soil temperature were important for Hg emission when the soil surface temperatures were low (the soil surface temperatures were 8.3� C and 3.2� C in LS and FS on average, respectively, and often dropped below 0� C). affecting on Hg emission.In May Hg emission flux was observed to be the highest in the LS; however, it showed a statistical negative correlation with soil temperature and a positive correlation with soil RH (Table 3).As discussed earlier, a peak of Hg emission appear-   2011.02. 2011.06. 2011.02. 2011.05. 2011.02. 2011.06.
A less significant result for correlations between Hg emission flux and environmental parameters was derived either by short sampling duration or by effect of multiple interacting factors.One possible factor is the oxidation effect of bacteria for FS (Choi, 2007;Smith et al., 1998;Hamlett et al., 1992;Summers and Silver, 1972).Since emission-deposition cycle greatly affects on the transport and fate of Hg in multimedia environments the averaged Hg emission and deposition fluxes for each sampling event were separately compared with environmental parameters for each sampling event in LS and FS (Fig. 4).Average volatilization flux generally increased with increases on UV radiation and soil surface temperature for LS and FS, showing higher flux in warm season while the deposition flux showed negative relationship with UV-A radiation (Fig. 4).In FS, the highest emission and deposition fluxes concurrently occurred during warm season (June, 2011), indicating that deposition may be as important as emission for the forest floors in warm season because of the limited solar radiation reaching on soil surface due to the leafy trees.On the other hand, deposition flux for LS was relatively consistent over the sampling periods.In overall, the deposition flux was not affected by soil surface temperature.
Daily averaged net flux showed higher emission flux with an increase of soil surface temperature (Spearman rank order correlation coefficient, r s = =0.673,p-value = =0.033)and soil humidity (r s = =0.709,p-value= =0.022) for LS while it did not have significant correlation with any environmental parameters for FS.This result possibly suggests that fallen leaves on forest floor acted as an inhibitor for showing strong relationship between Hg emission flux and environmental parameters by limiting the amount of heat and solar radiation reaching the forest floor.

CONCLUSIONS
In this study, Hg exchange fluxes between atmosphere and soil surface were observed in two types of soils (LS and FS) in cold (February) and warm (May, June) seasons.Hg emission was much higher during warm season in LS; however, in FS, Hg deposition was frequently observed in warm season probably because of leafy trees blocking the solar radiation.When UV radiation suddenly increased during a very short time in FS in June, the direction of flux abruptly changed from deposition to emission, showing the influence of radiation on Hg emissions.
In cold seasons, Hg fluxes showed significantly positive correlations with UV radiation and soil surface temperature in both soils; however, concurrent variations between Hg flux and environmental parameters were not clearly observed in warm season.In LS, Hg emission flux was possibly affected by enhanced atmospheric oxidants as well as other environmental parameters in warm season, resulting in ambiguous diurnal variation.When the Hg emission and deposition fluxes for each sampling event were separately compared with environmental parameters average volatilization flux generally increased with increases on UV-A radiation and soil surface temperature for both sites while the deposition flux showed negative relationship with UV-A radiation.There were also statistically significant relationships between daily averaged net Hg flux with soil surface temperature and soil humidity in LS.
Average Hg emission from LS was higher than from FS although the Hg content in soil was more than 2 times higher in FS, indicating that total amount of Hg in soil was not a primary indicator for Hg emission flux in this study.Hg emission fluxes from LS in this study were generally lower than in urban sites in Korea and China, but higher than those measured in USA and Canada.It should be noted that a polycarbonate chamber possibly underestimated the Hg emission flux because it blocks most of the UV-B light (Carpi et al., 2007) although in the study of Choi and Holsen (2009a) it did not impact the results significantly.In addition, the relatively higher flushing rate compared with small DFC volume could enhance the emission flux (Eckley et al., 2010;Lindberg et al., 2002), bringing the possibility of overestimation on flux measurement in this study.

Fig. 1 .
Fig. 1.Hg exchange flux between soil and air at the lawn soil (LS).The positive and negative values indicate emission and deposition, respectively.

MercuryFig. 2 .
Fig. 2. Hg exchange flux between soil and air at the forest soil (FS).The positive and negative values indicate emission and deposition, respectively.

Fig. 3 .
Fig. 3. Histograms of relative frequency of soil Hg flux at FS and LS in each season.
morning (around 2 am) on May 29 th was possibly explained by high O 3 concentration and low soil RH.In FS, there was a significantly positive relationship of Hg emission flux with soil temperature and/or UV radiation in February and June (Table

Table 1 .
Mercury Exchange between Soil and Air201 Selected soil properties and total Hg content in soil surface at the sampling site.

Table 2 .
Comparison of Hg emission fluxes from soil surfaces with other studies.
*Fluxes were estimated based on Hg concentration gradients at two heights.For all others, fluxes were directly measured using DFC

Table 3 .
Mercury Exchange between Soil and Air 205 Correlation coefficient (p-value) between Hg emission flux with environmental parameters.