Degassing and Cycling of Mercury at Nisyros Volcano (Greece)

Nisyros Island (Greece) is an active volcano hosting a high-enthalpy geothermal system. During June 2013, an extensive survey on Hg concentrations in different matrices (fumarolic fluids, atmosphere, soils, and plants) was carried out at the Lakki Plain, an intracaldera area affected by widespread soil and fumarolic degassing. Concentrations of gaseous elemental mercury (GEM), together with H2S and CO2, were simultaneously measured in both the fumarolic emissions and the atmosphere around them. At the same time, 130 samples of top soils and 31 samples of plants (Cistus creticus and salvifolius and Erica arborea and manipuliflora) were collected for Hg analysis. Mercury concentrations in fumarolic gases ranged from 10,500 to 46,300 ng/m3, while Hg concentrations in the air ranged from high background values in the Lakki Plain caldera (10-36 ng/m3) up to 7100 ng/m3 in the fumarolic areas. Outside the caldera, the concentrations were relatively low (2-5 ng/m3). The positive correlation with both CO2 and H2S in air highlighted the importance of hydrothermal gases as carrier for GEM. On the other hand, soil Hg concentrations (0.023-13.7 μg/g) showed no significant correlations with CO2 and H2S in the soil gases, whereas it showed a positive correlation with total S content and an inverse one with the soil pH, evidencing the complexity of the processes involving Hg carried by hydrothermal gases while passing through the soil. Total Hg concentrations in plant leaves (0.010-0.112 μg/g) had no direct correlation with soil Hg, with Cistus leaves containing higher values of Hg with respect to Erica. Even though GEM concentrations in the air within the caldera are sometimes orders of magnitude above the global background, they should not be considered dangerous to human health. Values exceeding the WHO guideline value of 1000 ng/m3 are very rare (<0.1%) and only found very close to the main fumarolic vents, where the access to tourists is prohibited.


Introduction
Volcanoes and geothermal areas are natural sources of environment pollutants potentially dangerous for human health. Paroxysmal eruptions and passive degassing emit huge amounts of gases such as CO 2 , H 2 S, SO 2 , and HF, including gaseous elemental mercury (GEM) [1][2][3][4]. Trace metals, being associated with uprising gases, are usually found at consider-able concentrations in hydrothermal fluids [5]. Even at very low concentrations, they can have a strong impact on the atmosphere and hydrosphere and consequently on the biosphere [6].
Among the volcanic trace volatile elements, mercury (Hg) is one of the most environmentally significant [7] because of its extreme mobility and toxicity [8]. The biogeochemistry of Hg is extremely complex due to the exchanges between atmospheric, terrestrial, and marine pools [9]. These processes are mainly driven by microbial activity, dark abiotic and photochemical reactions affecting Hg speciation and bioaccumulation [10]. It is emitted in several forms: elemental (metallic) Hg and inorganic and organic Hg compounds. Metallic Hg (Hg 0 ) is highly volatile due to its high vapour pressure and may experience long-range transport in the air due to its relatively long half-life in the atmosphere (1-2 years [11]). Monovalent and divalent Hg are both soluble in water; divalent Hg (Hg II ) is more stable and common in the environment than monovalent (Hg I ). This form also may undergo complexation, precipitation with inorganic ligands, and sorption onto the soil matrix. The toxicological properties of Hg for the environment and human health depend on the physical and chemical form in which it occurs. Hg vapours, for example, are very dangerous if inhaled, due to their ability to reach the lungs causing pulmonary oedema, pain, and peeling of the respiratory epithelium of the bronchi [12,13].
Mercury, as a constituent of volcanic and geothermal fluids [14,15], is discharged in water and released into the atmosphere as Hg 0 being associated with reducing noncondensable gases [16,17].
In the last decades, many authors underlined the correlation between Hg and H 2 S in discharged hydrothermal fluids (e.g., [18]), as testified by the formation of solid cinnabar (HgS) at the fumarolic vents. Hydrogen sulphide is a toxic pollutant; it is corrosive and poses severe concerns for human health [19,20].
Nisyros Island is a quiescent volcano releasing hydrothermal gases from several fumarolic emissions and also diffusively through the soil. The hydrothermal fluids of Nisyros are rich in H 2 S [21][22][23], and their diffuse emission creates an extremely acidic environment in soils affected by the hydrothermal degassing [24,25].
Here, we report the results of an extensive survey on Hg concentrations in different media (fumarolic fluids, atmosphere, soils, and plants) at the Lakki Plain, an area intensively impacted by hydrothermal degassing on Nisyros Island. Even though the geogenic degassing of Nisyros Island has been greatly studied, most of the research conducted was mainly concentrated on the gaseous C compounds and the noble gas composition of the fumaroles. The significance and novelty of this work, with respect to the existing literature, is to add Hg on the puzzle. Furthermore, to the best of our knowledge, this is so far the first study that attempts to define the Hg cycling in an active volcanic/geothermal system by taking into consideration such a great variety of media. Gaseous elemental mercury (GEM) concentrations, together with H 2 S and CO 2 in soil gas, were determined in both the fumarolic emissions and ambient air. Similarly, the relationships between fumarolic activity and Hg in the soils were investigated, comparing Hg concentrations to temperature, pH, hydrothermal gas, and the elemental concentrations of C, N, and S measured in the same soils. Leaves of two plant species (Cistus and Erica) were also collected and their Hg and S contents determined. Finally, a preliminary estimation of the Hg output to the atmosphere from the hydrothermal area of Nisyros was carried out.

Study Area
Nisyros Island ( Figure 1) is a quiescent volcano located in the easternmost volcanic group of the South Aegean Active Volcanic Arc (SAAVA [26]). The volcanic edifice developed in the last 200 ka through five distinguished stages [27,28] led to the formation of a caldera of about 4 km in diameter. The most recent activity consisted of hydrothermal explosions forming several phreatic craters, the last of which occurred in 1887 [27]. The Lakki Plain (Figure 1) represents the southeastern remnants of the calderic depression after the emplacement of a series of volcanic domes filling up the northwestern part. Fumarolic fields are currently active in this area, mainly within the hydrothermal craters strongly controlled by fracturing along the main NW-and NEtrending active fault systems [29], and are fed by a >1000 m deep hydrothermal system having a temperature of 300-350°C [30,31]. The hydrothermal craters form three main groups ( Figure 1): the oldest comprises the Kaminakia craters, the second consists of the Stefanos crater, whereas the third corresponds to the youngest area where a postcalderic dome (Lofos) is placed and includes the Phlegeton, Megalos Polybotes, and Mikros Polybotes craters [31,32]. Water vapour (91-99%) is the main component of the fumarolic fluids, followed by CO 2 and H 2 S [22]. The estimated total CO 2 and H 2 S outputs are close to 1 kg/s and <0.3 kg/s, respectively [31,32].

Materials and Methods
After the collection of few samples in 2009 and 2010, a multidisciplinary field campaign was carried out on June 2013 at the Lakki Plain, where soil gases, soils, and vegetation were sampled, and Hg, H 2 S, and CO 2 concentrations in the air were measured.
A total of 106 soil gas samples was collected at the Lakki Plain mostly in the fumarolic areas of Kaminakia, Stefanos, Mikros Polybotes, and Phlegeton craters and in the areas of Ramos and Lofos ( Figure 1). Soil gases were sampled at 50 cm depth using a Teflon tube of 5 mm ID equipped with a tight plastic syringe to avoid air contamination. Soil gas sampling sites were the same as those of the top soils. H 2 S and CO 2 analysis was carried out on the overpressurised vials using a Micro GC MSHA CP-4900 having 3 independent modules. Soil temperature was measured at 20 cm depth by using thermal probes and a digital thermometer; these measurements were carried out 10-15 min after the insertion of the thermal probe in the soil in order to achieve thermal equilibrium.
Top soils were collected from the first 3 cm depth at 130 spots at the Lakki Plain. Soil samples were dried, homogenized, and powdered. An aliquot of the homogenized samples was used for the analysis of total Hg, which was performed using a DMA-80 analyser (an atomic absorption spectrophotometer, Milestone, Wesleyan University, Middletown, CT, USA). About 10 mg of dry soil was loaded into specific nickel boats and analysed according to the US-EPA 7473 method [33]. Accuracy was checked by running replicates of the reference materials NCSDC7701 2 Geofluids (0 015 ± 0 006 mg/kg) and MESS3 (0 091 ± 0 009 mg/kg). Bench quality control material was measured at the start of each analytical run (set of 15 samples) for quality assurance and control. The measured values were, on average, within ±8% of the recommended values. Total C, N, and S were analysed on powdered samples by elemental analysis (Elementar Vario EL Cube, Hanau). The technique is based on "purge and trap" separation (C, N, and S), following high-temperature incineration (induction furnace) in a pure oxygen atmosphere and at a constant temperature exceeding 1150°C for the sample, with WO 3 as catalyst. Helium was used as the carrier gas. Detection limits were 0.04 wt% for C and 0.003 wt% for N and S.
Soil pH values were measured using a specific combination electrode on soil suspensions that were made with deionized water with a soil/solution weight ratio of 1/2.5 [34].
A passive biomonitoring survey was carried out in order to determine the total Hg content in two local spontaneous plants and evaluate the possible contamination by Hg gas emissions. The collected vegetation consisted of 17 leaf samples of the genus Erica (manipuliflora and arborea spp.) and 14 leaf samples of the genus Cistus (creticus and salvifolius spp.). Both the plant species have evergreen leaves and grow widespread as small shrubs (10-50 cm height) in the Lakki Plain where soil degassing is at a lower level. To enhance the interpretation, soil samples were collected along a buffer area of few tens of centimetres close to the sampled plants.
One sample of vegetation and soil was collected outside the caldera as local background blank. The sampling sites and the sampled plants were chosen randomly; for each site, three separate plants in a buffer area of 1.5 m were sampled and merged to obtain the final sample. Vegetation samples were dried in the oven at temperature below 40°C and powdered by agate planetary ball mill to avoid contamination. Analysis of total Hg was made with the use of PerkinElmer Inc. SMS 100 Solid Mercury Analysis. Each sample was heated in an oxygen-rich furnace to release all the decomposition products including Hg. These products were then carried in a stream of O 2 to a catalytic section of the furnace; halogens and/or oxides of N and S were trapped on the catalyst. The remaining vapour was then carried to an amalgamation cell that selectively trapped Hg. After the system was flushed with O 2 to remove any remaining gas or decomposition products, the amalgamation cell was rapidly heated, releasing Hg vapour. Flowing O 2 carried the Hg vapour through an absorbance cell positioned in the light path of a single wavelength atomic absorption spectrophotometer. Absorbance was measured at the 253.7 nm wavelength as a function of the Hg concentration in the sample.
Collection of total gaseous mercury (TGM) was performed with gold-coated bead traps (Au traps), through which atmospheric air was pumped at flow rates between 0.5 and 0.6 L/min [35] over collection periods ranging from 2 to 60 min. At three fumaroles, the gas was collected downstream of a vapour condenser sucking with a  3 Geofluids graduated 100 mL syringe and sent to the same Au traps through a three-way valve [36]. The collected Hg was then measured by cold vapour atomic fluorescence spectrometry (CVAFS), based on the conventional thermaldesorption amalgamation technique (relative standard deviation < 15%; EPA Method IO-5; [33,35,37]). The results obtained were multiplied by the sampling volume and expressed in ng/m 3 .
The simultaneous real-time measurements of gaseous elemental mercury (GEM), CO 2 , H 2 S, and meteorological parameters (air temperature, pressure, and relative humidity) were carried out by coupling portable instruments (similarly to what was proposed by [20]). GEM was measured with a Lumex® RA-915M, which is an atomic absorption spectrometer with a Zeeman effect with high-frequency modulation of light polarization (ZAAS-HFMLP). The separation of the spectral lines (at λ = 254 nm) is operated by a permanent external magnetic field, into which a source of radiation (Hg lamp) is placed [38,39]. The Zeeman background correction and the multipath analytical cell provide high selectivity and sensitivity [39]. The instrument operates at a flow rate of 10 L/min, whereas its rechargeable battery allows up to 8 h of continuous measurements. The detection limit is 2 ng/m 3 , while the accuracy of the method is 20% from 2 to 50,000 ng/m 3 [38,39]. The remaining parameters were measured with a Multi-GAS analyser manufactured by INGV-Palermo. Atmospheric gas was drawn into the sampler with an air pump at 1.2 L/min through a 1 μm Teflon membrane particle filter and was pumped through a CO 2 /H 2 O gas detector (Licor LI-840 NDIR closed-path spectrometer) and a series of electrochemical sensors for SO 2 (0-200 ppm; 3ST/F electrochemical sensor by City Technology Ltd.) and H 2 S (0-50 pm; EZ3H electrochemical sensor by City Technology Ltd.) detection. The sensors were housed in a weather-proof box mounted on a backpack frame and were calibrated, before and after fieldwork, with standard calibration gases (200 ppm SO 2 , 50 ppm H 2 S, and 3014 ppm CO 2 ) mixed with ultrapure nitrogen to provide a range of desired concentrations [4,40].
The spatial coordinates for each concentration value were simultaneously acquired through a GPS signal. All instruments were synchronized and set to high-frequency acquisition (every two seconds: 0.5 Hz). Measurements were carried out along four (Polybotes, Kaminakia, Stefanos, and Lofos) transect walks (about 15 km path, with a mean speed of 1.5 km/h) across the Lakki Plain caldera. The raw data have been processed by a dedicated software (RatioCalc [41]) that allows a derivation of mass ratios of various compounds (e.g., CO 2 /H 2 S, GEM/CO 2 , and GEM/H 2 S).
Dataset from the gas soil and air surveys were used to define the threshold values of Hg, H 2 S, and CO 2 . Data were processed following Sinclair's portioning method extracting the main populations [42]. This method consists in the definition of single populations through the inflection points (main populations) or changes in direction (secondary populations) of the curvature on the probability plot by visual analysis. Finally, data were plotted by using the GIS platform; distribution maps were drawn and ranked according to the identified populations.

Fumarolic Gases and
Atmosphere. In 2009, TGM was measured with Au traps in the atmosphere at 9 sampling sites at different distances from the main fumarolic vents. These sampling sites were previously investigated by D'Alessandro et al. [43] measuring H 2 S concentrations in the atmosphere with passive samplers. The traps were placed in a range of distance always longer than 10 m and up to about 2 km from the fumarolic vent. Mercury values ranged from 9.4 to 420 ng/m 3 (Table 1). Although not directly comparable due to the different and sometimes not overlapping measuring time intervals (2-60 min for TGM and 4 hours to 5 days for H 2 S), the two datasets show a positive correlation ( Figure 2), with the highest values close to the fumaroles and the lowest on the caldera rim.
In 2010, further 5 measurements with Au traps were performed. Two measurements were made in the atmosphere at less than 1 m far from the two main fumarolic vents and gave concentrations of 2360 and 4530 ng/m 3 (Table 1). Three fumarolic vents were investigated by performing measurements on the undiluted hydrothermal fluid collected at the outlet of a condenser; results provided values from 10,500 to 46,300 ng/m 3 ( Table 1). The composition of the contemporaneously collected gases was published by [44].
In 2013, GEM measured with Lumex® gave atmospheric concentrations from 2 to 7132 ng/m 3 . Values measured with the Multi-GAS ranged from 393 μmol/mol, which is the background atmospheric value, up to saturation of the sensor (~4000 μmol/mol) for CO 2 , from 0.2 μmol/mol up to saturation of the sensor (~60 μmol/mol) for H 2 S, and from 0.39 to 1.34 μmol/mol for SO 2 . Saturated values, about 0.1% for CO 2 and 1.6% for H 2 S, were not considered for Hg/CO 2 and Hg/H 2 S calculations.

4.2.
Soils. All parameters measured in the soil samples of the fumarolic areas of the Lakki Plain are shown in Table 2. Total Hg ranged from 0.023 to 13.7 μg/g of dry soil. Soil temperatures measured at 20 cm depth varied between 25.5 and 100°C, while concentrations of H 2 S and CO 2 at 50 cm depth ranged from <0.001 to 17.8% vol and from 0.28 to 75.3% vol, respectively. Elemental C, N, and S contents in the soil were in the range 0.06-2.63, 0.004-0.23, and 0.014-56.3 weight %, respectively. Soil pH varied from 0.71 to 7.30.

Plants.
Total Hg concentrations in the leaves collected from the Lakki Plain are summarized in Table 3. Data analysis showed the Hg range from 0.014 to 0.066 μg/g for Erica leaves and from 0.010 to 0.112 μg/g for Cistus leaves. The Hg concentrations in the soils sampled near the plants vary from 0.045 to 0.619 μg/g with a pH ranging from 4.21 to 5.55 ( Table 2). The highest pH value is 7.30 and was measured outside the caldera; the value is representative of the local background for the soils collected close to the plants ( Table 2).

Mercury in the Fumarolic Fluids of Nisyros.
Analyses of Hg directly on fumarolic fluids have rarely been performed. 4 Geofluids The values reported in literature related to several fumaroles worldwide [3,35,[45][46][47][48][49] cover a broad range from 1400 to 1,828,000 ng/m 3 ( Table 4). The values resulting from the three fumaroles sampled at Nisyros fall within this range.
Correlating the obtained Hg values with the major composition of the fumarolic gases [44], Hg/H 2 S and Hg/CO 2 ratios ranging from 0 36 × 10 −7 to 3 3 × 10 −7 and from 0 76 × 10 −8 to 3 5 × 10 −8 , respectively, were observed. The results obtained are coherent with values provided by other fumarolic fields at world scale ( Table 4). The measured Hg/H 2 S and Hg/CO 2 ratios should be regarded as lower limit values since water soluble Hg II species plausibly are lost within the fumarolic condensate collected by the sampling device. As evidenced in previous studies, the lost Hg fraction could represent a significant part of the total emitted Hg. Bagnato et al. [49] suggested that up to 70% of the total Hg emitted from the Bocca Grande (BG) fumarole (Phlegrean Fields, southern Italy) remained in the collected condensate. However, of the 70 fumarolic samples in which Nakagawa [36,45,46] measured Hg both in the gas and in the condensed vapour, only in 8 of the samples was the Hg found in the condensed fraction that represented more than 20% of the total. Nevertheless, further studies will be necessary to ascertain the quantity of Hg lost in the condensed steam in the fumaroles of Nisyros.

Mercury in the
Atmosphere of the Lakki Plain Area. Background values of atmospheric GEM in pristine unpolluted areas in the northern hemisphere are below 2 ng/m 3 , though a decreasing trend in the last decades was recognized [50]. In volcanic/geothermal areas, measured values are often significantly higher than the natural background (tens to hundreds of ng/m 3 , Table 5). This holds true also for the fumarolic area of the Lakki Plain at Nisyros both for the point measurements with Au traps and for measurements performed along transects with Lumex®.
The data acquired in the air with the Lumex® and Multi-GAS were plotted as described by Sinclair [42] and are shown in Figure 3. The CO 2 probability plot (Figure 3(a)) identifies three main populations (A, B, and C). The A population comprises 64.6% of the data with values ranging from 392 to 431 ppm and refers to the local atmospheric CO 2 background, which is close to the global atmospheric value of unpolluted air in 2013 (395 ppm [51]). Population B (33.3% of the data; CO 2 431-561 ppm) is the population with CO 2 level slightly higher than average atmospheric air, probably due to diffuse soil degassing. Population C (2.1% of the data) includes the highest values (up to >4000 ppm), indicating a significant fumarolic CO 2 contribution to the atmosphere.
Based on the probability plot (Figure 3(b)), the H 2 S dataset can be divided into four populations: A includes very low H 2 S concentrations (<0.47 ppm, 5% of the values); B population with values from 0.47 to 2.11 ppm (50% of the data) indicating a slight fumarolic contribution; C Sampling sites of 2009 and IDs in brackets are the same as in [43]; fumaroles sampled in 2010 are the same as in [44]. n.a. = not applicable.    43.5% of the data) indicating a significant H 2 S input into the atmosphere; D population (1.5% of the data) includes values > 57 ppm of H 2 S up to saturation of the sensor in the vicinity of the active fumaroles, suggesting a significant contribution of the hydrothermal fluids released from the subsurface.
Three distinct populations can be recognized (A < 8 5, B from 8.5 to 44.9, and C up to 7132 ng/m 3 ; 12.2%, 82.8%, and 5.0%, respectively) for GEM concentrations in the air (Figure 3(c)) that were measured with the Lumex® instrumentation.
In Figure 4, an example of H 2 S, CO 2 , and GEM concentrations measured in the atmosphere through transect walk within the Lakki Plain is shown. GEM, H 2 S, and CO 2 concentration peaks, in correspondence with the fumaroles of Phlegeton, show a good match, confirming the interdependence of these gaseous compounds and their common origin from the fumaroles. The main fumarolic emissions were clearly highlighted by anomalously high Hg concentrations (up to~600 ng/m 3 ) with respect to the surrounding air masses (~30 ng/m 3 ). GEM concentrations above background values measured away from the main fumarolic vents were probably due to soil degassing. The Lakki Plain and especially the main hydrothermal craters are sites of strong hydrothermal degassing with CO 2 fluxes up to 6175 g/m 2 /day [52]. Mercury fluxes from the soil have not   Table 2 were collected close to the plant samples with the above corresponding numbers.  9 Geofluids been measured at Nisyros, but as reported from the literature [3,49,53,54], soils at geothermal and hydrothermal systems also emit gaseous Hg. Nevertheless, the previous authors found Hg fluxes (1-2000 μg/m 2 /day) that are many orders of magnitude lower than those of CO 2 , with the latter gas often acting as carrier for Hg.

Mercury in the Soil.
The soils of the Lakki Plain are strongly weathered by past and present fumarolic activity. This can be recognized by the widespread presence of secondary alteration minerals, mainly sulfates [24,25]. The main drivers of the alteration process are fumarolic H 2 O and H 2 S. The former is the main carrier of thermal energy, which is reflected in soil temperatures reaching up to the boiling temperature of water. Soil temperatures provide indications regarding the hydrothermal uprising gases, allowing the identification of the actively degassing areas. High temperatures are to be considered related to both high fluxes of hydrothermal fluids and the enrichment of the hydrothermal component in the soil gases. The temperature distribution map at 20 cm depth indicates temperatures above 30°C in all the investigated sites, except for some points along the western flank of the Kaminakia crater ( Figure 5(a)). Higher temperatures, from 50 to 100°C, were recorded at the southern part of the Stefanos crater, at Phlegeton and Mikros Polybotes (Figure 5(a)).
Hydrogen sulphide in the fumarolic gases of Nisyros is the third most abundant species after H 2 O and CO 2 [22].

Geofluids
In the soil gases, on a dry basis, it represents up to nearly 18% vol. Within the soil, H 2 S is oxidised by atmospheric O 2 forming sulfuric acid [55,56], by the net reaction Such reaction is responsible for the high S contents of the analysed soils (median~2%, max~56% (Table 2)) and the very low values of soil pH (from 0.71 to 5.55 (Table 2)). Both parameters show an inverse relationship (Figure 6(a)) indicating that, at the most actively exhaling zones, more sulfuric acid is produced and more sulfur is deposited.
The total amount of Hg trapped in the soil was plotted in a probability plot. As for the Hg concentrations in the air, three populations were detected ( Figure 7); a secondary population containing the higher values was detected as indicated by the black arrow in Figure 7. High Hg concentrations were measured in the soil close to the main fumarolic vents (Figure 5(b)). As for sulfur, the inverse relationship between soil Hg and soil pH (Figure 6(b)) supports the transport and the deposition of Hg by fumarolic fluids. Daskalopoulou et al. [24] evidenced that also other volatile elements like As, Bi, Pb, Sb, Se, and Te are enriched in the soils of the Lakki Plain that are mostly affected by hydrothermal gases.
The accumulation of Hg in the soil matrix does not depend solely on the amount of Hg carried by the uprising hydrothermal gases but also on the soil retention capacity. Many studies demonstrated that Hg in soils shows generally a good correlation with soil organic matter (SOM). Ottesen et al. [57] evidenced such positive correlation at the continental scale, based on the analysis of more than 4000 soil samples from 33 countries in Europe. Martin et al. [58] found that the same correlation holds true for the volcanic soils of Mt. Etna. Elemental carbon, which can be considered a proxy for SOM, in the soils of the Lakki Plain does not show any correlation with the measured Hg values. This probably depends on the fact that most of the sampled soils are totally devoid of vegetation, limiting the presence of SOM to some vegetal debris and microbial communities. The low content of SOM may therefore not contribute much to Hg accumulation in the soils of the Lakki Plain. Sulphide, which could also react with Hg, is also very scarce in the surface soil levels due to the oxidising environment [24].
Soil temperatures are thought to play an important role in Hg retention in soils. This parameter has probably a contrasting effect because higher soil temperatures indicate stronger hydrothermal gases transporting more Hg from the depth, but at the same time, higher soil temperatures presuppose also a faster remobilization of Hg from the soils because of its high volatility [3].
Although hydrothermal gases like CO 2 and H 2 S show their maxima close to the main fumarolic areas (Figures 5(d) and 5(e)), no correlation with soil Hg could be evidenced (p value 0.77 and 0.83, respectively, and correlation index 0.04 and 0.02, respectively). However, even though both soil and gas samples were collected at the same time and in the same place, they are not totally comparable as CO 2 and H 2 S were measured on the gas phase collected at the 50 cm depth while Hg was measured on the solid phase at the soil surface. Nevertheless, a better correlation was expected.

Mercury in the Plants.
Plant leaves present lower concentrations with respect to the soils where they grow, while the local background in both plant leaves and soils shows very low concentrations in total Hg. Results show a range of values regarding the total Hg concentrations that seem to be positively correlated with the intensity of the fumarolic activity. The highest concentrations were measured in sites located close to the hydrothermal drill and the craters of Stefanos, Phlegeton, Mikros Polybotes, and Kaminakia, confirming the impact of the hydrothermal activity on the surrounding environment.
Regarding the availability of soil Hg to plants, it is considered to be low as there is a tendency for Hg to accumulate in 11 Geofluids the roots, indicating that the roots serve as a barrier to Hg uptake [59,60]. Mercury concentration in aboveground parts of plants appears to depend largely on foliar uptake of Hg 0 volatilized from the soil [61,62] and therefore on the age of the plant and the time of day and year [63]. The transfer of Hg (gaseous forms) from the atmosphere occurs by dry and wet deposition (rain and snow) and enters into the organism by the stomata of the leaves through the transpiration process [59,61,64,65]. The soil-plant correlation diagram (Figures 8(a) and 8(b)) shows that Cistus samples were enriched in Hg with respect to Erica presenting a moderately good positive correlation.  12 Geofluids Taking into consideration that Hg uptake is mostly through the leaves and the samples of both species were collected on the same day, the relatively high concentrations of total Hg in the Cistus samples can be justified by the higher specific surface area of its leaflets with respect to Erica's. Additionally, equally important factors that should be taken into consideration are the solar radiation and the humidity. These two factors along with the specific surface area of the leaflets may possibly explain the elevated Hg concentration of the local background sample that was noticed at Erica. It is worth mentioning that the local background samples of both species were collected at a place located close to the sea, at noontime during summer period. On the other hand, sulfur is an essential macronutrient for plant growth. Its uptake and distribution are tightly controlled by the environmentally induced changes in nutrient demand [66,67]. Therefore, the high concentrations in S as well as the good correlation between the soil and plant samples in both species can be regarded as mainly caused by its uptake by the roots through the metabolic processes of the plants (Figures 8(c) and 8(d)). Nevertheless, the contribution of the transpiration process via the stomata of the leaves may also be important.

Environmental and Human Health Issues.
Mercury is considered to be among the most toxic metals that could be taken by the human body through different pathways [13], especially concerning methylated species. Furthermore, its deleterious effects can be enhanced through its biomagnification along the trophic chain [11]. One of the primary uptake paths of Hg is through the inhalation of GEM. For the World Health Organization [68], atmospheric concentrations of Hg in the range of 15,000-30,000 ng/m 3 may have adverse effects on humans (tremors, renal tubular effects, change in plasma enzymes, and others). However, using an uncertainty factor of 20, the same organization proposed a guideline value for Hg concentration in air of 1000 ng/m 3 [68].
At the same time, the US Occupational Safety and Health Administration considered a permissible occupational exposure limit for GEM of 100,000 ng/m 3 in the air [69], while the National (US) Institute for Occupational Safety and Health (NIOSH) established a recommended exposure limit for GEM of 50,000 ng/m 3 as a time-weighted average (TWA) for up to a 10 h workday and a 40 h workweek [69]. The American Conference of Governmental Industrial Hygienists assigned to GEM a threshold limit value of 25,000 ng/m 3 as a TWA for a normal 8 h workday and a 40 h workweek [69]. The minimum risk level (MRL) for chronic inhalation of GEM is 200 ng/m 3 [69,70]. The MRL is an estimate of the daily human exposure to a hazardous substance that is likely to be without appreciable risk of adverse health effects over a specified duration of exposure. The US EPA reference concentration for inhalation is calculated to be 300 ng/m 3 (TWA) [69].
Taking into consideration the above limits and thresholds, it may be deduced that, although the atmospheric GEM concentrations within the Nisyros Caldera are sometimes many orders of magnitude above the global background, they do not represent a general hazard for human health. The area is yearly visited by many tens of thousands of tourists, but the zones where they arrive show GEM concentrations rarely exceeding the MRL. Areas exceeding the WHO guideline value of 1000 ng/m 3 are very close to the main fumarolic vents where tourists are not allowed to go and where other toxic gases of higher danger are present (i.e., H 2 S [19]). Moreover, people who work all day in the area (ticket operators, owner, and employees of the Volcano Café) spend most of their time in areas with atmospheric GEM concentrations well below the occupational limits and generally also below the MRL. Probably, only volcanologists that take gas samples from the main fumaroles may be exposed to atmospheric GEM levels of a few thousands of ng/m 3 for some hour, which has still to be considered a low exposition. Even though results propose no particular risk, a more complete survey is highly suggested to have a more accurate picture. 5.6. Total Output from the Hydrothermal System of Nisyros. It has long been established that the contribution of volcanic activity to the total natural emissions of Hg to the atmosphere is substantial [2,4,47]. However, due to the limited number of data available, significant uncertainties on the annual emissions of volcanic Hg still remain. Estimates range 13 Geofluids between 0.6 and 1000 t/a representing a proportion that varies from <1% up to 50% of total natural emissions [2, 4, and references therein]. The strongest contribution (90%) of the total output of volcanic Hg derives from explosive eruptions while the rest derives from passive degassing [2]. Emissions of single open-conduit volcanoes like Etna (Italy), Ambrim (Vanuatu), or Masaya (Nicaragua) are in the order of units to tens of t/a [4], pointing to strong underestimation of global volcanic outputs lower than 50 t/a. The contribution of fumarolic emissions from closed-conduit volcanoes is instead very limited as it is found in the range from 0.2 to 167 kg/a (0.0002-0.167 t/a (Table 5)).
Most of the Hg flux estimations from volcanic systems have been indirectly obtained cross-correlating the measured SO 2 fluxes with the Hg/SO 2 ratios measured in the volcanic plume [4]. Such method cannot be used for lowtemperature fumarolic areas because magmatic SO 2 is strongly scrubbed by the hydrothermal system. To obtain output estimates for such areas, different methods have been proposed. One of which is the measurement of Hg fluxes from the soils with accumulation chambers (dynamic flux chambers or static closed chambers) and the consequent integration of the fluxes over the whole hydrothermal area. Such method was applied only few times [49,53,54]. Another method is the cross-correlation of the total CO 2 release of the fumarolic area with the Hg/CO 2 ratio being measured either in the fumarolic fluids or in the air close to the vents [4]. Both methods give the order of magnitude of 14 Geofluids the Hg output, but they coherently confirm much lower outputs of fumarolic areas with respect to open-conduit volcanoes. Following the second approach, a rough estimation of the Hg output of Nisyros can be obtained considering the total CO 2 output, as determined by Bagnato et al. [53], which is 84 t/d, and the Hg/CO 2 ratios measured in the fumaroles that range from 0 8 × 10 −8 to 3 5 × 10 −8 (median value = 2 3 × 10 −8 ) or in the air ranging from 1 5 × 10 −8 to 8 2 × 10 −8 (median value = 6 3 × 10 −8 ), corresponding to the total Hg output of 0.7 and 1.9 kg/a, respectively. Such figures fall in the lower range of the outlet values for fumarolic areas (Table 5).

Conclusions
Mercury and its compounds are highly toxic for humans and ecosystems. Volcanic and hydrothermal emissions are major natural sources of Hg in the atmosphere. At Nisyros, a potentially active volcano with intense and widespread degassing activity, real-time measurements of GEM showed concentrations up to 7132 ng/m 3 within the Lakki Plain. The good correlation between Hg and the main fumarolic gases (H 2 O, CO 2 , and H 2 S) confirms the hydrothermal origin of the former. In the fumarolic gases, Hg was estimated in the range 10,500-46,300 ng/m 3 ; these values should be considered the lower limit due to the plausible Hg loss within the condensing fumarolic vapour. Relatively high Hg concentrations were also identified in the soils; the accumulation of Hg in the soil matrix is dependent on both the amount carried by the upflowing hydrothermal gases and the soil ability to fix a part of it. The lack of vegetation at the crater area maybe responsible of the poor correlation between elemental C and Hg in the soil. No bioavailability through the roots was noticed in the plants collected at the Lakki Plain. The slightly high concentrations of the vegetation samples could have therefore been caused by the transpiration process that takes place in the stomata of the leaves, making Cistus a better candidate for biomonitoring investigations with respect to Erica due to the greater specific area of its leaves. The aforementioned synoptic analysis of the results highlights that more than one matrix can be affected by hydrothermal Hg from the degassing activity. Furthermore, it underscores that Hg concentrations are positively correlated with the distance of the sample/measurement from the emission point. Even though the measured Hg concentrations were enhanced and at cases exceed the WHO limits in terms of inhalation, they seem to be of minor risk for human health as the exposure is for a limited time and the access for nonvolcanologists is prohibited. However, an uptake originating from the trophic chain should not be disregarded.