CO 2 Degassing in Sicily (Central Mediterranean) as Inferred from Groundwater Composition

: The circulation of groundwater is inﬂuenced by several factors, including permeability changes due to the dynamics of the stress ﬁeld acting along tectonic discontinuities. Open faults can act as preferential pathways for the escape of crustal and subcrustal gases, such as CO 2 , towards the surface, allowing their chemical interaction with meteoric ﬂuids. Here, we present the ﬁrst map of dissolved CO 2 partial pressure in the groundwater of Sicily, Italy. Based on the chemical analysis of 557 samples of groundwater, which were collected in wells and springs, we calculated the related CO 2 partial pressure ( P CO 2 ) using PHREEQC software. The spatial distribution of the calculated P CO 2 values highlights a general homogeneity at the regional scale, interrupted by positive anomalies linked to the main seismogenic or volcanic areas. Faults work as preferential escape pathways for deep CO 2 . The subsequent CO 2 –water–rock interaction determines the chemical composition of circulating water. As a consequence, groundwater composition can be successfully used to identify anomalous degassing areas.


Introduction
The chemical composition of groundwater is the result of a series of complex interaction processes occurring between natural waters, rocks, and gases, being strictly influenced by the atmosphere and biosphere. A huge number of different geochemical reactions occur in geological systems, rationalized by a finite array of reactions able to describe the sequences of states of aqueous solution composition, speciation, and mineral paragenesis over time [1]. Previous studies addressed the modeling of reactions governing silica phases and aqueous solutions [2][3][4][5][6], or carbonate aquifer systems, focusing on the interactions between minerals and aqueous species, their effects on water composition, and residence times [7][8][9].
Sicily is characterized by the presence of sedimentary, metamorphic, and volcanic rocks, mainly composed of carbonates, silicates, and alkali sulfate evaporites. Primary minerals are dissolved and/or altered during the chemical interaction with groundwater and dissolved gases.
Several processes contribute to CO 2 dissolution in groundwater. It can be generated from biodegradation of organic matter in the soil. Otherwise, metamorphic zones can release abundant CO 2 produced by high-temperature metamorphism of carbonates to groundwater [10]. Furthermore, CO 2 is the major gas species, after water vapor, released by volcano-magmatic processes, because it is the first gas to be exsolved from a magmatic melt due to its low solubility [11].

Geological Background
Sicily, located along the main Eurasia-Nubia convergent plate boundary [28][29][30], represents a segment linking the African Maghrebides with the Southern Apennines across the Calabrian accretionary wedge. Three main elements mark both the collisional complex of Sicily and its offshore continuation: (i) the Hyblean Foreland outcropping in southeastern Sicily; (ii) the Caltanissetta Basin, a dynamic foredeep basin of Late Miocene to Quaternary age; and (iii) a complex chain, thrust towards the east and southeast, consisting of the Calabrian Arc and the Maghrebian thrust belt ( [31] and references therein). The Hyblean Foreland, submerged by the Pelagian Sea and the adjacent Ionian Sea, represents a remnant of a Late-Jurassic-Early-Cretaceous passive continental margin. The Caltanissetta Basin, located along the northern side of the foreland, is presently buried by the frontal sector of the chain formed of an E and SE vergent fold and thrust belt, outcropping on land and submerged in the adjacent seas. It is made of a "European" element (Peloritani Units), a "Tethyan" element (Sicilide Units), and an African element (Maghrebian Sicilian Units) [32][33][34][35][36]. The different elements evolved in a tectonic framework characterized by two main compressional phases [33,37]: (i) the first one occurred during the Middle-Late Miocene, which caused the overthrusting of the deep-water carbonate above the not-already-deformed carbonate platform substrate; (ii) the latter is of Pliocene to Lower Pleistocene age, during which the carbonate platform units were deformed throughout deep-seated structures, reimbricating the previously emplaced deep-water carbonates. Simultaneously, thrust, backthrust, and lateral displacements [38], together with clockwise paleomagnetically detected nappe rotations (Late-Miocene-Middle-Pleistocene) created a wedge top basin filled with syntectonic deposits [34,37,39]. Details about lithostratigraphy can be found in Basilone [40].
Volcanism has occurred since the Late Triassic times in western Sicily ( [41] and references therein) and in the Iblean area [42,43]. Etna is a prominent recent example of active basaltic volcanism in Sicily ( [44] and references therein).
Major earthquakes presently occur in eastern Sicily ( [30] and references therein), along the NNW-trending offshore fault system located between Siracusa and Catania, parallel to the Malta Escarpment and to the transtensional fault zone separating Sicily from the Ionian abyssal plain [45,46]. Another zone of medium-to high-magnitude earthquakes is along the E-W trending compressive belt in the southern Tyrrhenian Sea ( [47] and references therein). Moreover, contractional to transpressional mechanisms have been recorded in southwestern Sicily [48]. Finally, in the northeastern Sicily belt (Madonie-Peloritani) and beneath the Caltanissetta Basin, extensional to transtensional mechanisms have been recorded.

Materials and Methods
Data used in this work refer to physicochemical parameters and concentrations of dissolved ions taken from different literature sources, namely, the "Piano Tutela Acque Regione Siciliana" ( [10,49,50], integrated with previously unreleased data for western Sicily (courtesy of Paolo Madonia, INGV, Sezione di Palermo)), for a total of 618 groundwater samples.
Gaseous CO 2 dissolution in water occurs due to the following reaction: Since the equilibrium of this reaction is shifted towards the left side and CO 2(gas) can be expressed as PCO 2 , its constant can be written as CO 2(aq) reacts with water (hydration), forming H 2 CO 3 following the reaction We can define H 2 CO 3 * as the sum of dissolved CO 2(aq) and H 2 CO 3 [51], and since H 2 CO 3 is only 0.16% of CO 2(aq) and the reaction rates are so fast with respect to the average residence times of groundwater, we can neglect H 2 CO 3 and write the following expression: Following Equation (4), the protolysis of carbonic acid can be expressed as the equilibrium of which constant is given by Combining Expressions (2) and (6), we have Referring to (7), the logarithmic expressions is as follows: Hence, it is possible to calculate the partial pressure of CO 2 starting from pH and alkalinity using the constants K H and K 1 once the temperature is known.
Based on the above-described approach, we calculated the partial pressure of CO 2 . We processed our data using PHREEQC software [52]. The PCO 2 contour map was realized using Golden Software Surfer (release 13), with the kriging interpolation algorithm. In order to avoid pseudofeatures due to missing data, we separated the sampling points into 4 groups (Western Sicily, Madonie Mountains, Piazza Armerina area, and Eastern Sicily), computing the contouring separately for each area and plotting the results on the same base map; points were separated using the spatial queries options implemented in Qgis software (release 2.18). In order to investigate the potential relationship between the spatial pattern of dissolved CO 2 anomalies and tectonics, we used data from the "Catalogo Parametrico dei Terremoti Italiani" (CPTI15; [53]) reporting locations of high-magnitude historical earthquakes (1000-2014 A.D.) (see cited reference for details). Moreover, we used the most advanced seismogenic source model, namely, the "Database of Individual Seismogenic Sources" (DISS Working Group, 2018), which lists active faults that can generate Mw 5.5 and larger earthquakes, classified as Individual (ISS) or Composite (CSS) Seismogenic Sources. In particular, we looked for the potential role of active faults in releasing CO 2 to aquifers by analyzing Locations of historical earthquakes and seismogenic sources.

Physicochemical Composition of Groundwater
Due to the huge amount of data, the main statistical parameters for pH, total alkalinity, PCO 2 , major dissolved ions, and SI calcite were computed for all samples and are reported in Table 1 (the full datasets are available in [49]).
A normal probability plot (PP) was used to identify populations of data. This type of plot expresses the probabilities of cumulative frequencies of the measured major elements dissolved in the groundwater. Data that are normally distributed plot along a straight line. Therefore, the presence of multiple straight lines with different slopes indicates the presence of various populations of normally distributed data with diverse origins of the element considered [18]. They show up as extreme points at either end of a probability plot. Figure 1 shows PPs for all major elements, while Table 2 summarizes all identified populations for every major element with the respective threshold values and the percentage of populations recognized. The graphs also indicate the threshold values deduced from the inflection points.
Each element shows a polymodal distribution ( Figure 1). Four populations are recognized for the main ions, except for sodium with five populations, and magnesium with three. Generally, the anomalous populations (population IV) concern about 1% of the data; only sodium shows two anomalous populations (IV and V). Population I, defined by lowest concentrations, represents less than 20% of the samples. For magnesium and chlorine, population I is characterized by 96% and 50%, respectively. Population II (intermediate groups) represents more than 40% of the complete dataset. Table 1. Main statistical parameters of CO 2 partial pressure (PCO 2 ), pH, saturation index, and major element concentrations.  The PP for total alkalinity shows three inflection points, reflecting four populations of data with normal distribution: a low concentration group (population I), an intermediate group (population II), and two groups with high concentration values (populations III and IV). Most of the samples belong to population II, which represents 89.5% of the samples, while the highest concentration groups represent  The PP for total alkalinity shows three inflection points, reflecting four populations of data with normal distribution: a low concentration group (population I), an intermediate group (population II), and two groups with high concentration values (populations III and IV). Most of the samples belong to population II, which represents 89.5% of the samples, while the highest concentration groups represent only 5%. Most of the waters falling in population IV also show the highest levels of PCO 2 . They are located in the volcanic area of Mount Etna, around the volcanic edifice or in the Catania Plain (Naftia, Acqua Grassa, Acque Difesa). Other points fall in areas where hydrothermal activity is recognized, such as the Peloritani mountain range (Ali Terme, Terme Vigliatore). These considerations suggest a link between the total alkalinity of groundwater and the input of deep inorganic CO 2 , as also proposed by Grassa et al. [10]. Calcium (Ca) and total alkalinity (Alk) show a similar distribution of various populations of data. As shown in Figure 2, there is a direct relationship between Ca 2+ + Mg 2+ and the total alkalinity. High Ca 2+ and Mg 2+ levels in groundwater are mainly present where carbonatic aquifers occur, pointing to the water-rock interaction as the source of these dissolved ions. Most of the samples fall around the line 1:1, indicating that CO 2 plays a main role in determining the amount of dissolved ions. Some samples from hydrothermal areas (Termini Imerese, Terme Selinuntine) fall with a cation excess far from the calcite-water equilibrium line, indicating other source processes responsible for the dissolved ions. Accordingly, for these samples, a clear seawater (SW) contribution has been recognized [10,25].  PPs for sodium (Na) and chlorine (Cl) show a similar distribution for low concentrations with three inflection points, reflecting four populations of data: about 47% of samples show an intermediate concentration group (population II), and a reduced group (about 1%) shows higher concentrations (population IV). Only sodium shows an additional population with strongly anomalous values (population V), the sampling points of which (Naftia, Terme Selinuntine, Terme Vigliatore, T. Imerese, Sclafani Bagni, G. Cassibile) are in some cases located where volcanic or hydrothermal activity has been identified [10,24,25]. In the diagram of Na-Cl (Figure 3), samples containing seawater contributions can be clearly identified. Moreover, according to Grassa [10], the water-silicate minerals interaction coupled with an inflow of CO2-rich gas to groundwater is responsible for the removal of alkaline earth elements due to the formation of carbonate or silicate alteration minerals in Naftia and Terme Vigliatore. This  PPs for sodium (Na) and chlorine (Cl) show a similar distribution for low concentrations with three inflection points, reflecting four populations of data: about 47% of samples show an intermediate concentration group (population II), and a reduced group (about 1%) shows higher concentrations (population IV). Only sodium shows an additional population with strongly anomalous values (population V), the sampling points of which (Naftia, Terme Selinuntine, Terme Vigliatore, T. Imerese, Sclafani Bagni, G. Cassibile) are in some cases located where volcanic or hydrothermal activity has been identified [10,24,25]. In the diagram of Na-Cl (Figure 3), samples containing seawater contributions can be clearly identified. Moreover, according to Grassa [10], the water-silicate minerals interaction coupled with an inflow of CO 2 -rich gas to groundwater is responsible for the removal of alkaline earth elements due to the formation of carbonate or silicate alteration minerals in Naftia and Terme Vigliatore. This was also evident in most of the waters located in the active volcanic systems of Mount Etna. However, the Na/Cl molar ratio for most of the samples falls along the typical seawater molar ratio line, with contributions from both maritime aerosols and direct seawater, particularly in sampling points close to the coast (Figure 3). Data of rainwater used in Figure 3 are from [54], in particular, the OVS sample from Mount Vesuvius, representative of rainwater not influenced by volcanic activity and with a low contribution of anthropogenic sources.  Figure 3. Na vs. Cl ratio as mmol L −1 in log scale. Solid line is Na/Cl ratio for seawater. White triangle is Na/Cl ratio in meteoric water (data from [54]). Figure 4 shows the total alkalinity plotted against the sum of the cation equivalents (ΣCat). Since CO2-driven weathering should give a direct positive correlation between the two parameters (e.g., [55]), the graph allows one to recognize the role of the weathering process as a source of metals dissolved in groundwater. However, all of the samples deviate from the 1:1 line, exhibiting an excess of cations and suggesting the influence of other processes determining the chemical composition of groundwater. This is particularly evident in the waters from hydrothermal and volcanic areas.  . Na vs. Cl ratio as mmol·L −1 in log scale. Solid line is Na/Cl ratio for seawater. White triangle is Na/Cl ratio in meteoric water (data from [54]). Figure 4 shows the total alkalinity plotted against the sum of the cation equivalents (ΣCat). Since CO 2 -driven weathering should give a direct positive correlation between the two parameters (e.g., [55]), the graph allows one to recognize the role of the weathering process as a source of metals dissolved in groundwater. However, all of the samples deviate from the 1:1 line, exhibiting an excess of cations and suggesting the influence of other processes determining the chemical composition of groundwater. This is particularly evident in the waters from hydrothermal and volcanic areas.
Saturation indexes of calcite ( Figure 5) for these waters indicate that most of the samples are close to the equilibrium saturation or oversaturated with respect to calcite, while the remaining waters are undersaturated to strongly undersaturated (−4.47 < SI < −1.2), such as in some of the springs located in the Peloritani area. Dongarrà et al. [56] also found low saturation indexes of calcite up to −3.3 in some waters of the southeastern sector of the Peloritani area, associated with a low degree of mineralization of these waters, suggested by the authors as the result of short contact time with soil and host rocks. Water samples from the volcanic area of Mount Etna (around the volcanic edifice or in the Catania Plain) with highest computed dissolved PCO 2 are characterized by a volcanogenic CO 2 -anomalous degassing area [18] and show slight undersaturation in view of calcite. For two samples belonging to the Etnean aquifer, Grassa et al. [10] also suggested a weak water-rock interaction due to short residence time. Figure 4 shows the total alkalinity plotted against the sum of the cation equivalents (ΣCat). Since CO2-driven weathering should give a direct positive correlation between the two parameters (e.g., [55]), the graph allows one to recognize the role of the weathering process as a source of metals dissolved in groundwater. However, all of the samples deviate from the 1:1 line, exhibiting an excess of cations and suggesting the influence of other processes determining the chemical composition of groundwater. This is particularly evident in the waters from hydrothermal and volcanic areas.  In spite of the wide variability of the chemical composition of the groundwater, the CO2 partial pressure solely depends on gaseous CO2 interacting with groundwater, and its spatial distribution can provide useful information to identify anomalous degassing areas.

Distribution of Dissolved CO2 Partial Pressure
Apart from minor anomalies related to a limited number of measures deprived of adequate statistical reliability, it is possible to distinguish different areas ( Figure 6) where dissolved CO2 partial pressure shows levels above the regional background. These are (i) an area aligned N-S between the Castellammare Gulf, in the north, and Sciacca, in the south; (ii) two areas, one on the Tyrrhenian and the other on the Ionian sides of the Peloritani mountains; (iii) the Etna area and Catania Plain that, albeit with a certain articulation, extends to the northern extremity of the Hyblean Plateau; (iv) the area located close to the Augusta harbor; and (v), finally, an area with two anomalies in the southern side of the Hyblean Plateau.
In terms of amplitude, the strongest anomalies are observed in eastern Sicily and, in particular, along the Ionian side of the Peloritani mountains, in an area of the Catania Plain and in the northern portion of the Hyblean Plateau.
Regarding Figure 6, good agreement is observed between the spatial distribution of seismogenic In spite of the wide variability of the chemical composition of the groundwater, the CO 2 partial pressure solely depends on gaseous CO 2 interacting with groundwater, and its spatial distribution can provide useful information to identify anomalous degassing areas.

Distribution of Dissolved CO 2 Partial Pressure
Apart from minor anomalies related to a limited number of measures deprived of adequate statistical reliability, it is possible to distinguish different areas ( Figure 6) where dissolved CO 2 partial pressure shows levels above the regional background. These are (i) an area aligned N-S between the Castellammare Gulf, in the north, and Sciacca, in the south; (ii) two areas, one on the Tyrrhenian and the other on the Ionian sides of the Peloritani mountains; (iii) the Etna area and Catania Plain that, albeit with a certain articulation, extends to the northern extremity of the Hyblean Plateau; (iv) the area located close to the Augusta harbor; and (v), finally, an area with two anomalies in the southern side of the Hyblean Plateau.
In terms of amplitude, the strongest anomalies are observed in eastern Sicily and, in particular, along the Ionian side of the Peloritani mountains, in an area of the Catania Plain and in the northern portion of the Hyblean Plateau.
Regarding Figure 6, good agreement is observed between the spatial distribution of seismogenic structures and main dissolved CO 2 partial pressure anomalies.
Moving from west to east, the area hit by the 1968 Belice earthquake (and related seismogenic structures) falls in the middle of the main dissolved CO 2 anomaly of western Sicily. Here, Caracausi et al. [22] recognized three main thermal systems by using 3 He/ 4 He ratios and heat flux data: Alcamo, Sciacca, and Montevago. These are linked to tectonic discontinuities-cutting the continental crust into a regime of differential shortening, oblique thrusting, and clockwise rotations [57]-which convey mantle volatiles to the surface.
The second area is composed of two coupled anomalies lying along the Tyrrhenian and Ionian sides of the Peloritani mountains, respectively. The northern anomaly is delimited northwards by the offshore segment of the Aeolian-Tindari-Letojanni Fault System (ATLFS, [58]). The southern anomaly is delimited southwards by the onshore segment of the ATLFS and eastwards by the seismogenic structure responsible for the 1908 Stretto di Messina earthquake.
The area surrounding Mount Etna shows a wide CO 2 anomaly, which extends southwards reaching the SSW-NNE tectonic structure coinciding with the front of the Sicilian chain-foredeep system. A secondary anomaly is found at the intersection between this front and a N-S active tectonic lineament (Scicli Line, [59] and references therein); this anomaly corresponds to an area interpreted by some authors [60,61] as a southward extension of the wide degassing system of Mount Etna. The high dissolved CO 2 partial pressures recorded here, following this interpretation, are related to volatiles carried by the magmatic gas phase continuously dissolving into shallow aquifers.
Water 2020, 12, x FOR PEER REVIEW 9 of 14 Moving from west to east, the area hit by the 1968 Belice earthquake (and related seismogenic structures) falls in the middle of the main dissolved CO2 anomaly of western Sicily. Here, Caracausi et al. [22] recognized three main thermal systems by using 3 He/ 4 He ratios and heat flux data: Alcamo, Sciacca, and Montevago. These are linked to tectonic discontinuities-cutting the continental crust into a regime of differential shortening, oblique thrusting, and clockwise rotations [57]-which convey mantle volatiles to the surface.
The second area is composed of two coupled anomalies lying along the Tyrrhenian and Ionian sides of the Peloritani mountains, respectively. The northern anomaly is delimited northwards by the offshore segment of the Aeolian-Tindari-Letojanni Fault System (ATLFS, [58]). The southern anomaly is delimited southwards by the onshore segment of the ATLFS and eastwards by the seismogenic structure responsible for the 1908 Stretto di Messina earthquake.
The area surrounding Mount Etna shows a wide CO2 anomaly, which extends southwards reaching the SSW-NNE tectonic structure coinciding with the front of the Sicilian chain-foredeep system. A secondary anomaly is found at the intersection between this front and a N-S active tectonic lineament (Scicli Line, [59] and references therein); this anomaly corresponds to an area interpreted by some authors [60,61] as a southward extension of the wide degassing system of Mount Etna. The high dissolved CO2 partial pressures recorded here, following this interpretation, are related to volatiles carried by the magmatic gas phase continuously dissolving into shallow aquifers. Moving eastward, we found another anomaly at the NE border of the 1693 Val di Noto earthquake seismogenic source, located in the immediate hinterland of the Augusta Gulf, close to the onshore prolongation of the Hyblean-Malta escarpment. A final anomalous, hourglass-shaped area is found in the south of the Hyblean Foreland around another tectonic structure connecting the Val di Noto seismogenic source with the N-S lineament previously described (Scicli Line, [59]). Grassa [10] found concentrations and isotope ratios of helium dissolved in groundwater from these areas compatible Moving eastward, we found another anomaly at the NE border of the 1693 Val di Noto earthquake seismogenic source, located in the immediate hinterland of the Augusta Gulf, close to the onshore prolongation of the Hyblean-Malta escarpment. A final anomalous, hourglass-shaped area is found in the south of the Hyblean Foreland around another tectonic structure connecting the Val di Noto seismogenic source with the N-S lineament previously described (Scicli Line, [59]). Grassa [10] found concentrations and isotope ratios of helium dissolved in groundwater from these areas compatible with a mixing between crustal and MORB fluids, which he indicated to be the result of the uprising of deep fluids along local active tectonic structures.

Conclusions
CO 2 emissions are usually determined through the direct measurement of gaseous CO 2 fluxes at the soil-atmosphere interface. However, due to its high solubility in water [64], CO 2 easily dissolves in water and favors water-rock interactions. Consequently, deep CO 2 inputs can be recognized in groundwater composition.
Starting from the chemical composition of groundwater, we calculated the related PCO 2 , extending our analysis to all the sectors of Sicily where sampling points were available (557 samples collected from both wells and springs). We presented our results as a matched map of PCO 2 contour lines, locations of historical earthquakes, tectonic lineaments, and seismogenic sources.
The spatial relationship between these elements highlighted anomalous PCO 2 areas, emerging from the regional background, linked to the main seismogenic or volcanic structures. This indicates that the observed anomalies can be ascribed to deep CO 2 degassing sustained by magmatic and hydrothermal activities.
These anomalous areas are: (i) an area aligned N-S between the Castellammare Gulf, in the north, and Sciacca, in the south; (ii) two areas, one on the Tyrrhenian and the other on the Ionian sides of the Peloritani mountains; (iii) the Etna area and Catania Plain extending to the northern extremity of the Hyblean Plateau; (iv) the area located close to the Augusta harbor; and (v) the southern side of the Hyblean Plateau.
In seismogenic areas, groundwater circulation can be affected by permeability variations driven by the dynamics of the stress field. Open faults work as preferential escape pathways for deep CO 2 , fostering its chemical interaction with shallower meteoric waters and generating the observed anomalies.
Our results point out that the sampling and chemical analysis of groundwater and subsequent data processing is an efficient and fast method to detect anomalous degassing areas and their relationship with tectonics.
This also implies that the long-term monitoring of groundwater composition could provide some useful information on CO 2 emission variations over time. However, since the groundwater composition also depends on other exogenous processes (e.g., rainfall amount, seawater intrusion, volcanic activity, etc.), data from long-term monitoring programs have to be carefully evaluated in order to avoid overinterpretation of the data.
A step forward after the identification of anomalous CO 2 degassing areas is a deep geochemical investigation based on the chemical and isotope composition of dissolved gases, which would allow for better constraining of the CO 2 source. Furthermore, strontium isotope investigations can provide useful scientific evidence to evaluate the origin of deep brines (e.g., [65]).
Here, we highlight that the proposed approach does not require time-consuming activity in the field and can be carried out with the most common analytical instruments used to determine the chemical composition of groundwater.
Finally, we believe that such an approach can provide a useful tool not only for the scientific community but also for government decision makers. Indeed, the understanding of the geological processes occurring around us represents a good opportunity to make responsible decisions. We suggest that similar geochemical explorations should be extended to the entire national territory.