Forty-five gas samples have been collected from natural gas manifestations at the island of Kos—the majority of which are found underwater along the southern coast of the island. On land, two anomalous degassing areas have been recognized. These areas are mainly characterized by the lack of vegetation and after long dry periods by the presence of sulfate salt efflorescence. Carbon dioxide is the prevailing gas species (ranging from 88 to 99%), while minor amounts of N2 (up to 7.5%) and CH4 (up to 2.1%) are also present. Significant contents of H2 (up to 0.2%) and H2S (up to 0.3%) are found in the on-land manifestations. Only one of the underwater manifestations is generally rich in N2 (up to 98.9%) with CH4 concentrations of up to 11.7% and occasionally extremely low CO2 amounts (down to 0.09%). Isotope composition of He ranges from 0.85 to 6.71 , indicating a sometimes-strong mantle contribution; the highest values measured are found in the two highly degassing areas of Paradise beach and Volcania. C-isotope composition of CO2 ranges from -20.1 to 0.64‰ vs. V-PDB, with the majority of the values being concentrated at around -1‰ and therefore proposing a mixed mantle—limestone origin. Isotope composition of CH4 ranges from -21.5 to +2.8‰ vs. V-PDB for C and from -143 to +36‰ vs. V-SMOW for H, pointing to a geothermal origin with sometimes-evident secondary oxidation processes. The dataset presented in this work consists of sites that were repeatedly sampled in the last few years, with some of which being also sampled just before and immediately after the magnitude 6.6 earthquake that occurred on the 20th of July 2017 about 15 km ENE of the island of Kos. Changes in the degassing areas along with significant variations in the geochemical parameters of the released gases were observed both before and after the seismic event; however, no coherent model explaining those changes was obtained. CO2 flux measurements showed values of up to about in both the areas of Volcania and Kokkino Nero, at Paradise beach, and at Therma spring. CO2 output estimations gave values of 24.6, 16.8, 12.7, and , respectively, for the above four areas. The total output of the island is and is comparable to those of the other active volcanic/geothermal systems of Greece (Nisyros, Nea Kameni, Milos, Methana, and Sousaki).

1. Introduction

The southern Aegean Sea is one of the most tectonically active regions of western Eurasia, where fast convergence of the Aegean microplate and the Eastern Mediterranean lithosphere (the front part of the African plate) occurs. The African plate subducts underneath the Aegean-Anatolian microplate at a rate of about 1 cm/a [1, 2], and the microplate overrides the Eastern Mediterranean [3], resulting also in the generation of a volcanic arc [4]. The South Active Aegean Volcanic Arc (SAAVA) was built on Paleozoic-Mesozoic basement rocks (Alpine basement) that were deformed during the Tertiary by the Hellenide orogeny [5]. During the Lower Pliocene, the rate of convergence between the two continental margins increased and initiated the volcanism and the development of several hydrothermal systems at the SAAVA (e.g., [4, 610]).

Kos island is part of the Kos-Nisyros-Gyali volcanic system of the SAAVA, located at the SE edge of the Aegean Sea and formed at around 5 Ma, at the beginning of the Pliocene [1, 3, 1113]. The geologic units of Kos consist of alluvial deposits with greenschists and flysch in the northern part of the island, lacustrine and terrestrial deposits of the Pliocene age in the central part with tuffs, and ignimbrites of the Quaternary age that cover the southern part of the island [14]. The most prominent volcanic formation on the island is the Kos Plateau Tuff (KPT). It is related to a caldera formed by an explosive eruption that occurred 161 ka ago, which is considered to be the largest explosive Quaternary eruption in the Eastern Mediterranean [15]. The tectonic evolution of the island is controlled by the dominant WNW-ESE and NE-SW faults, which are related to extensional processes and volcanic activity that took place during the Pleistocene and Pliocene [16, 17]. The volcanic island of Kos has been active for at least 3 million years [18] and continued to be active until recent times (e.g., [4, 610]). Some geothermal areas of particular interest have been identified in the island. The Volcania area is located 1 km northeast from the rim of Kefalos caldera [19]. It consists of a 1 km diameter basin with 14 small circular areas with evident signs of present and/or past hydrothermal alteration mainly arranged along two intersecting lineaments. These areas (5-20 m across) are devoid of vegetation and are usually covered by whitish altered deposits that contain sulfates and occasionally native sulfur [19]. Furthermore, hydrothermal activity is noticeable along the island with the most important sites being (i) the thermal spring of Therma, which is emerging on the beach close to Cape Fokas [20], (ii) the ferruginous spring of Kokkino Nero rich in CO2 [21], and (iii) the intensively degassing area of Paradise beach at Kefalos Bay.

Geogenic carbon emissions have a critical impact on the carbon cycle [22, 23] and are regarded as one of the reasons of global climate changes on long time scales. Gas emissions from such sources strongly contribute to the increasing concentrations of greenhouse gases in the atmosphere, with methane and carbon dioxide playing a fundamental role [24, 25]. Moreover, earth degassing defines relations among flux, tectonic structures [26, 27], and volcanic activity [28, 29]. The aim of this study is to estimate the total CO2 output of Kos island and to investigate the possible relation of these manifestations and anomalous areas with the geodynamically active area of Greece. This work presents new chemical and isotope data about the main gas manifestations of the island both on land and underwater. The collected data are interpreted together with literature data in order to determine the origin of the gases and the postgenetic processes that affect them. Finally, some considerations about possible variations on gas geochemistry induced by the earthquake () on the 20th of July 2017 with an epicenter close to the island are made.

2. Methods

Bubbling gases of Kos island were sampled using an inverted funnel positioned above the emission point of the highest flux, whereas soil gases were collected by inserting a pipe in the soil at >50 cm in depth and driving the gas by a syringe and a 3-way valve. Dry gases were collected in glass flasks equipped with two stopcocks.

In the laboratory, the concentrations of He, H2, H2S, O2, N2, CO2, and CH4 on the samples were analysed by an Agilent 7890B gas chromatograph combined with a Micro GC analyser by INFICON. A single amount of gas sample is simultaneously split into the loops of the two combined systems. Concentrations of CO2 and H2S have been determined by the Micro GC analyser with He as the carrier and equipped with a PoraPLOT U column and TCD detector while all the other gases had been determined by the GC system with Ar as the carrier and equipped with a 4 m Carbosieve S II column. A TCD detector was used to measure the concentrations of He, H2, O2, and N2 and a FID detector for that of CH4. The analytical errors were less than 10% for He and less than 5% for the remaining gases.

The 13C/12C ratios of CO2 (expressed as δ13C-CO2‰ V-PDB) were measured with a Finnigan Delta S mass spectrometer after purification of the gas mixture by standard procedures using cryogenic traps (precision ). Carbon and hydrogen isotopes of CH4 were measured using a Thermo TRACE GC interfaced to a Delta Plus XP gas source mass spectrometer and equipped with a Thermo GC/C III (for carbon) and with GC/TC peripherals (for hydrogen). The 13C/12C ratios are reported as δ13C-CH4 values with respect to the V-PDB standard () while the 2H/1H ratios are reported as δ2H-CH4 values with respect to the V-SMOW standard ().

The abundance and isotope composition of He, and the 4He/20Ne ratios, were determined by separately admitting He and Ne into a split flight tube mass spectrometer (Helix SFT). Helium isotope compositions are given as , where is the (3He/4He) ratio of the sample and is the atmospheric (3He/4He) ratio (). The analytical errors were generally <1%. The 4He/20Ne ratio was used to correct the measured values for the atmospheric contamination, and the corrected values are indicated as [30]. The analytical results of the collected samples as well as their coordinates are presented in Table 1.

Flux maps were drawn according to the dataset obtained from the three field campaigns that took place in the period from October 2015 to October 2017 (Table 2). Measurements were always made during dry and stable weather conditions. CO2 flux was measured at Volcania, Kokkino Nero, Paradise beach, and Therma with the accumulation chamber method at more than 600 sites (>500 points per km2) with portable soil fluxmeters (West Systems, Italy) based on the accumulation chamber method [31]. Flux values were determined at each site from the rate of CO2 concentration increase in the chamber and are expressed in grams per square meter per day () after conversion from volumetric to mass concentrations considering atmospheric pressure and temperature values. IR spectrometers with different measurement ranges ( for the Licor LI820 and double range of and , respectively, for the Dräger Polytron) were used. This resulted in a measuring range of for the Licor LI820 spectrometer with accumulation chamber A and a range of up to for the Dräger Polytron spectrometer with accumulation chamber B and high measuring range—the reproducibility being always better than 20%. Chamber A has an area of 0.031 m2 and a volume of 0.0028 m3, while chamber B has the same area and a volume of 0.0062 m3. Particular care was taken to follow the recommendations for flux measurements in volcanic/geothermal environments made by Lewicki et al. [32].

The soil temperature was measured only at few places and only in the last campaign by means of a digital thermocouple ( in the range from -100 to 200°C).

The CO2 datasets acquired from Volcania, Kokkino Nero, and Therma were used to estimate the total CO2 flux from these areas.

To define the CO2 threshold value, CO2 flux data were processed following the Sinclair’s portioning method extracting the main populations (Table 2; [33]). 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.

Following the stochastic simulation approach, CO2 flux maps were drawn. The data were converted by normal score transformation to follow a Gaussian distribution. The normal score transformed data was used to compute omnidirectional variograms and interpolated with the sequential Gaussian simulation (sGs) method by using the executable “sgsim” of GSLIB [34] and performing 100 equiprobable realizations for each area. The grid resolution was . The final maps were produced averaging the results of the 100 realizations, using the E-type postprocessing method.

Zonal Statistics on the three CO2 flux maps, performed by using the ArcMap 10.3 (ESRI) Spatial Analyst tool, was used to estimate the total CO2 output considering only flux values above the background threshold value for each area.

3. Results

In the period from 2009 to 2017, 10 sampling campaigns took place in the island of Kos and 45 gas samples were collected. Twenty-three of them were collected underwater at various depths (<10 m; Figure 1); five are soil gases whereas the remaining ones are gases bubbling in thermal waters. Literature data were also taken into consideration for comparison [14, 35, 36]. Names, sampling date, coordinates, chemical concentrations, and isotope values are presented on Table 1. Additionally, during the last three campaigns, more than 600 CO2 flux measurements were performed covering part of the island (both on land and on the sea surface) including 4 intensively degassing areas: 135000 m2 in Volcania (445 points), 250000 m2 in Kokkino Nero (133 points), 1100 m2 in Therma (29 points), and 600 m2 in Paradise beach (12 points). Results of the flux measurements are presented in Table S1.

3.1. Gas Geochemistry

Gases collected from Kos island show that CO2 is the prevailing gas species (generally more than and up to ; Figure 2). The majority of these samples display N2/O2 ratios higher than those of air and/or air-saturated water (ASW), indicating that the atmospheric component of meteoric water has been modified by redox reactions that took place either in the subsoil or in the aquifers. However, few samples present high concentrations of atmospheric gases (O2 up to and N2 up to ), pointing to a strong contamination by an atmospheric component (Figure 2(b)). Helium ranges from 0.34 to , while CH4 ranges from 40 to . Hydrogen and H2S, typical hydrothermal gases, present significant concentrations (up to 1900 and , respectively) in the soil gases, whereas in the underwater emissions, they are mostly below detection limits.

The isotope composition of He shows values from 0.85 to 6.71 with the highest values being found in the Paradise beach samples. Regarding the 4He/20Ne ratio, values of up to 1066 are observed. Carbon isotope composition of CO2 in the island is in the range of -20.07 to +0.64‰ vs. V-PDB, although almost all samples fall within a narrower range (-3.5 and 0‰). The isotope composition of CH4 varies from -21.5 to -2.8‰ vs. V-PDB for C and from -143 to +36‰ vs. V-SMOW for H.

3.2. CO2 Fluxes and Soil Temperatures

CO2 flux measurements for the areas under investigation gave values that range from <0.1 to . The highest values were measured in Paradise beach and Therma spring (median 9540 and 15100; maximum 23100 and , respectively) with the use of the floating chamber method. The two soil degassing areas of Volcania and Kokkino Nero showed the following median values: and , respectively, whereas the maximum values were and , respectively. Due to failure of the temperature probe, the temperature was not measured in the first campaign. Only few measurements were made in the second and third campaigns, and they were mainly concentrated in the highest CO2 flux measuring points. These measurements were taken at 50 cm in depth and gave values that were approaching the annual mean atmospheric temperature (22°C) excluding significant water vapour upflow at both Volcania and Kokkino Nero.

The portioning method of Sinclair [33] was applied to extract data populations from the dataset. Three main populations, i.e., “background (A),” “intermediate (B),” and “hydrothermal (C)” (Table 2), were identified from the CO2 flux datasets. The “background” population includes values from at Volcania and values from 0.44 to at Kokkino Nero; background population was not identified in the dataset acquired from Therma while at Paradise beach, only the “hydrothermal” population was present. The “intermediate” population includes values from ~6 to at Volcania, values from 23 to at Kokkino Nero, and values from 13 to at Therma. The “hydrothermal” population includes values from 2500 to at Volcania and Kokkino Nero and values of up to 53100 and for Paradise beach and Therma, respectively.

Table 2 summarizes the number of points contributing to each population of the dataset and the statistical parameters.

According to the identification of background population from the probability plot, the threshold values used for estimation of the CO2 flux from the Volcania area were and for Kokkino Nero, while no background value was used for Paradise beach and Therma.

4. Discussion

4.1. Origin of the Gases and Secondary Processes Affecting Their Composition

Carbon dioxide is the carrier phase for mantle-derived fluids. The combination of C and He isotope ratios is a useful tool to evaluate the origin of fluids. Helium isotope ratios are used to distinguish between crustal and upper mantle-derived components. In order to identify general controls on the CO2 characteristics of Kos island, we plotted the data on a CO2-3He-4He ternary diagram (Figure 3; [40]). The binary mixing trajectories between MORB-type mantle fluids (, ) and various crustal volatile endmembers (0.01 , -1015), as well as the general trends expected from addition and/or loss of a particular volatile phase, are also plotted in the diagram. Results propose a mantle origin for He that at points arrives close to the MORB endmember. Samples from Agia Irini and Therma represent products of variable amounts of mixing between mantle-derived and crustal volatiles with a preferential CO2 addition and/or He loss. On the other hand, Agia Irini 2 is found on the base of the triangle with CO2/3He ratios lower than those of the MORB, indicating CO2 removal, possibly caused by the higher solubility of CO2 with respect to He in aquatic environments (Figure 3(b)).

For the evaluation of the geologic processes’ effects, the CO2-He data are plotted on the vs. 4He/20Ne and CO2/3He vs. δ13C-CO2 binary diagrams (Figure 4). In the aforementioned diagrams, the binary mixing curves that display the trends drawn by mixtures of the atmospheric component with different mantle and crustal sources [30], as well as the typical values of sediment, limestone, and mantle-derived CO2 [41], are also plotted. Literature data of gas manifestations along the SAAVA [14, 35, 36, 4244] are plotted for comparison.

Only few samples show an important atmospheric contribution for He, as they plot close to the atmospheric endmember (Figure 4(a)). Most of the samples display a mixed crustal-mantle contribution always within the range of the SAAVA samples (up to 85% of mantle contribution). In particular, samples from Therma, Kokkino Nero, and Agia Irini present medium to low mantle contributions for He (up to 35% considering a MORB-type source), whereas data of Paradise beach, Kefalos, and Volcania display a relatively higher range (75-80%) with respect to the aforementioned areas.

The distribution of the values seems to follow a geographical distribution with the highest values collected in the western part of the island, while the lower ones are concentrated in the eastern part (Figure 5). Many volcanic systems show an approximately regular radial distribution of the values with the highest values being found close to the main volcanic or geothermal vents. Examples can be found at Nevado del Ruiz [45], Mt. Ontake [46], Lesser Antilles islands [47], Cascades [48], and Mt. Elbrus [49]. Such geographical pattern is generally explained with an increase in the contribution of both crustal (radiogenic) and atmospheric He components when the aquifer water reequilibrates with air, going from the main magmatic feeding system towards the peripheral areas [50]. However, in the present case, no active or recent volcanic conduit can be recognized on the island. Nevertheless, it is worth noting that the westernmost sampled gas manifestations are found along the supposed margin of the caldera that formed after the KPT explosive eruption. Along the margin of this structure, many other volcanic systems have been grown since that eruption (i.e., Nisyros, Strongyli, Pacheia, and Pergousa). It may be therefore hypothesized that the westernmost gas manifestations of Kos are connected to a deep mantle source by the presence of the ring faults of the caldera. At present, in the area, there is no surface sign of thermal anomalies that could point to the presence of a geothermal system possibly fed by an ascending magma batch. On the contrary, the easternmost sampling sites are found away from the caldera margins, mostly along the main tectonic structures, and are, at least in the case of Therma, related to thermal water circulation. Deep-rooted faults frequently constitute channels of high permeability that facilitate the migration of mantle fluids [51]. The strong difference in He isotope composition between the two groups of sampling sites may be explained also with the different geologic substrate of the two areas. In the western part of the island, young volcanic formations prevail, while in the eastern part, mostly older metamorphic rocks crop out. The latter could be the source of the crustal component that lowers the values of the gases collected in the eastern part of Kos.

Moreover, gases collected in Kos along with those of SAAVA are found in the mixing line between mantle and limestone endmembers, while the contribution of the organic sediments is trivial (Figure 4(b)). Samples of Paradise beach, Volcania, and Kefalos are those with the highest mantle component showing CO2/3He ratios similar to those of the MOR gases. Samples collected in Agia Irini 2 present CO2/3He ratios lower than those of the MOR range indicating a relative CO2 loss (Figures 3 and 4(b); [5255]).

Hydrothermal hydrocarbon production can be described by two main mechanisms that deal with the biotic and abiotic origins of methane [56]. Considering this, the origin of CH4 can be investigated using the classification diagram of Schoell [57, 58]. Thermogenic CH4 has been reported to exhibit δ13C-CH4 values that range from -50 to -30‰ and δ2H-CH4 (e.g., [5761]), whereas microbial CH4 usually has δ13C-CH4 (e.g., [6062]).

Samples collected in Kos island as well as samples of the SAAVA [43] plot in the field of volcanic geothermal systems and thus, a geothermal origin is suggested for CH4 (Figure 6). Exceptions are the samples collected at Kefalos, Agia Irini 2, and Therma, which sometimes present extremely (sample no. 2 of Therma) positive isotope values (for both C and H), pointing to CH4 oxidation processes. Inorganic oxidation of CH4 [63] in some samples cannot be ruled out. However, it is noticeable that the isotope fractionations of organic oxidation and inorganic oxidation of CH4 follow different fractionation paths. The former follows slopes ranging from 5.9 to 13 ([64] and references therein) and the latter a slope of 21 [63]. Daskalopoulou et al. [43] considered and values that cluster the majority of the samples as the most probable values of the isotope composition of geothermal CH4 in the Greek geothermal systems before oxidation. Based on that, the values, comprised between 3.8 and 13.6, are mostly overlapping the typical range of biogenic oxidation processes pointing towards the consumption by methanotrophic microorganisms. It is worth noting that the sites showing signs of methane oxidation are those presenting the lowest gas emission fluxes. The slower uprise of the gases allows a longer interaction with the methanotrophic microorganisms before bubble emission and therefore a higher consumption and a consequent fractionation of methane. This justifies the sometimes-strongly positive values that, considering the abovementioned starting δ13C-CH4 value, indicate a residual fraction of CH4 of about 0.4 [61].

4.2. Possible Influence of Seismic Activity on Fluid Geochemistry

Fluids play an important role in earthquake generation by reducing the friction between the fault blocks [6668] and transporting upper mantle energy with geochemical anomalies that occur before, during, and after earthquakes [69]. Therefore, heat flow and tectonics are related to both active faults and geothermal anomalies, and for that reason, many earthquake epicenters occur worldwide in areas with elevated heat flow [66, 67]. Many researchers have studied the relations between seismicity and geochemistry and have noticed changes in the physicochemical parameters, the ground deformation, the gas flow rate, and the isotope composition of the gases (e.g., Kobe, Japan [70]; Kamchatka, Russia [71]; El Salvador, Central America [72]; Nisyros, Greece [73]; and Campi Flegrei, Italy [74]).

An earthquake of occurred in the Gulf of Gökova between the areas of Bodrum and Kos on the 20th of July 2017. Heavy damages were noticed in both areas with the strongest intensities being recorded in the latter (United States Geological Survey (USGS)). Karasözen et al. [38] attributed this event to a normal fault gently dipping (~37°) northwards. This fault reached the sea bottom along a more than 10 km long E-W trending line generating tsunami waves that added further damages in Kara Ada island, Bodrum, and Kos island [75, 76].

Our campaigns included the systematic gas collection in the period from 2009 to 2018, thus including the seismic event. Three of the sampling sites were specifically resampled on the 26th of July 2017, 6 days after the main shock. In almost each area, results indicate a decrease in CO2 and an increase in both He and CH4 concentrations postseismically (Figure 7). This can be explained by a possible CO2 loss and a consequent relative enrichment of He and CH4; the difference in solubility may lead to extreme enrichments in the less soluble gases when a gas mixture rises through nonsaturated waters, especially when the gas/water ratio is very low [77]. This is probably the case of the gas collected 10 months before the seismic event at Agia Irini 2.

Furthermore, the collected gases evidenced variations in the ratios (Figure 8). In particular, all sites but one (Paradise beach), where He isotopes were measured, showed increased ratios at about 10 months before the earthquake. On the contrary, Paradise beach shows at that time a relative minimum, which is subsequently increased to the highest measured value 6 days after the earthquake. Helium isotopes at Therma present a slightly lower value with respect to the previous and subsequent samples. It is worth mentioning that Therma is the site closest to the epicenter and also the second site presented in this work, where He isotopes were measured 6 days after the seismic event. All seven localities along the island present ratios greater than those typical for crustal production (; [78]), revealing the presence of mantle-derived He throughout the fault zones (Table 1; Figures 3 and 4(a)). It is worth noting that all the sampling sites can be related to tectonic structures (Figure 1) that represent a preferential pathway for geogenic degassing. The strain induced by both the impending earthquake and the subsequent aftershock sequence may either induce variations in the permeability of these tectonic structures [79] or induce release of gases from magmatic or geothermal systems [36, 73].

The δ13C-CO2 values (Figure 8(a)) show also important variations that may be attributed to different processes like (a)degassing of CO2 from the geothermal waters that results in both the decrease of the CO2 contents in the geothermal water and the increase of the δ13C values of the residual dissolved fraction(b)addition/dissolution of CO2 into groundwater, which leads to partial dissolution (CO2(aq)), hydration (HCO3-(aq)), and dissociation into HCO3- and then CO32-. The fractionation factor between HCO3-(aq) and CO2(g) () is temperature dependent [8082], and therefore, at temperatures encountered at the sampling localities (<100°C), gaseous CO2 progresses towards lower δ13C values with increased dissolution (i.e., decreasing CO2)(c)dissolution/precipitation of the mineral calcite, where C isotope values become higher in the solid phase relative to the CO2 in the geothermal water [83] and, consequently, also in this case, δ13C values and CO2 contents decrease in the residual gas phase(d)mixing of gases with isotopically different CO2

The variation of δ13C-CO2 values in Figure 8(a) is significantly stronger than the variation of values. A strong decrease in δ13C-CO2 values is observed at Agia Irini 2, 10 months before the seismic event and, coherently with the variations in the chemical composition, may be attributed to CO2 dissolution processes. A reduction in gas upflow rate induced by the impending earthquake can be the cause. Relatively lower δ13C-CO2 values can be noticed also in the samples taken at Paradise beach just before and immediately after the earthquake. Conversely, at other sites, the δ13C values are increasing either slightly and constantly like in the case of Agia Irini and Kokkino Nero or more spike like as seen at Kefalos (10 months before) and Therma (6 days after). None of these variations could be definitely linked to a particular process. Mixing of heavier carbon may derive from fracturing and dissolution of carbonate minerals of the limestones in the sedimentary series or marbles in the metamorphic sequence.

Along with the variations in the chemical and isotope compositions, changes in the degassing areas were also witnessed. In particular, in the submarine manifestations of Paradise beach, the degassing area became wider and a remarkable increase in the flux was observed. Raised water temperatures (at least at Therma) and gas fluxes were also recorded in the areas closer to the epicenter (Therma, Agia Irini, and Agia Irini 2). However, due to the lack of repeated flux measurements and the sometimes-incoherent temporal variations in chemical composition, and δ13C-CO2, no conclusions regarding the pre- and postseismic changes can be reached in the present work.

4.3. Total CO2 Output Estimation

The CO2 total output estimation is performed following a stocatistic approach obtaining the most probable CO2 output value for each of the three investigated areas (Volcania, Kokkino Nero, and Therma). Such data processing is used to produce the CO2 distribution maps for the three areas (Figure 9).

Analysing the “background” populations extracted from the data, some differences between the datasets are noticed. The background threshold of Volcania is nearly one order of magnitude lower than that obtained from Kokkino Nero. This discrepancy can be referred to the soil assemblage. In fact, the soil at Volcania is more altered and less covered by vegetation even in the low-flux areas with respect to the soil at Kokkino Nero. The strong alteration of the soils in the Volcania area is probably due to past fumarolic activity [19]. The amount of CO2 produced by the biomass at Volcania is lower than that produced at Kokkino Nero, and, consequently, the CO2 threshold value is higher at Kokkino Nero.

The Therma dataset was mainly acquired along the shore with most of the data acquired on the water surface. The biomass producing CO2 in this area was almost absent, and all the CO2 upflow can be addressed to the hydrothermal component. The intermediate population in the Volcania and Kokkino Nero areas represents the mixing of the background component, the hydrothermal component (higher values), and/or the air mixing/dilution (lower values); the intermediate population obtained from the Therma dataset can be addressed to the hydrothermal component mixed with air. As in this site, the CO2 flux is highly sustained reaching the extraordinary value of and the CO2 upflow rate is so high to reduce and in some way prevent the air dilution. Almost all flux measurements included in the hydrothermal population were made with the floating chamber on the water of the artificial pool created for thermal bath purposes. On the contrary, almost all measurements referring to the intermediate population were made on the shores surrounding the pool. The shores are made of highly permeable coarse gravel favouring air circulation. This may explain also the low amount of data included in the intermediate population at Therma and its high upper limit ().

The sGs produced an E-type map with the mean expected value for each cell. The total CO2 output was obtained, for each area, summing all cell values above the threshold multiplied by the surface covered by each cell. The total CO2 outputs estimated by sGs for Volcania, Kokkino Nero, and Therma are 24.6, 16.8, and , respectively (Table 3). At Paradise beach, the number of flux measuring points was not enough to produce a map with the sGs method. Therefore, to estimate the CO2 output of this area, we multiplied the output area by the average of the flux values obtaining a value of . All the four investigated areas sum up an estimated total CO2 flux for the island of Kos of about . Such value falls within the range reported for the other volcanic/geothermal systems along the SAAVA (Table 4) where outputs as low as 2.6 (Methana) and up to 84  (Nisyros) have been obtained. All these values fall at the lower end of the output estimations of volcanic systems worldwide [84, 85] and typical of quiescent volcanic systems.

5. Conclusions

Gas emissions are spread over the volcanic island of Kos and are expressed as hot springs, underwater bubbling, and sulfate salt efflorescence. Results have proposed a mixed mantle-limestone origin for CO2, which is the prevailing gas species, whereas an up to 85% mantle contribution has been recognized for He. Gas components such as H2, H2S, and CH4 have indicated a geothermal origin within a still not recognized reservoir beneath the island. Gas manifestations as well as anomalous degassing areas can be related to important tectonic structures. The active tectonics of the area often results in seismic activity with the recent earthquake (20th July 2017—) causing a tsunami phenomenon and heavy damages on the island. Variations in both the chemical and isotope compositions and changes in the degassing areas and the physicochemical parameters of the systems have been observed. These have been possibly caused due to changes in mixing ratios of fluids of different origins and in the flux of uprising gases, although it has not been made possible to interpret all variations univocally.

The four main degassing areas (Volcania, Paradise beach, Kokkino Nero, and Therma) that have been recognized present sometimes very high flux values (). The total CO2 output for Kos has been estimated in . This estimation has been in the range of values typical for the volcanic geothermal systems along the SAAVA and in the lower end of the worldwide volcanic output estimations

Data Availability

The data on gas geochemistry used to support the findings of this study are included within the article. The CO2 flux data are included within the supplementary information file.

Conflicts of Interest

The authors declare that they have no conflicts of interest.


We kindly acknowledge the help of the Marine warrant officer Ilias Simadakis who gave us the permission to collect samples, the owner of “Kardamena Watersports Center” Mr. Nikos Nikolakopoulos who gave us one of his boats, and Mr. Yannis Limperis who accompanied us to the points of interest. CO2 flux measurements would have not been possible without the friendly help of Salvatore Inguaggiato who gave us the device that he manages. All analyses were made at the laboratories of the INGV of Palermo, and we are grateful to all the following laboratories responsibles and technicians: G. Capasso, F. Grassa, M. Martelli, Y. Oliveri, A. Rizzo, F. Salerno, A. Sollami, and M. Tantillo.

Supplementary Materials

MS Office Excel file “Table S1” containing the coordinates and the values of all the CO2 flux measurements made on the island of Kos. (Supplementary Materials)