Hydrogeology and geochemistry of the sulfur karst springs at Santa Cesarea Terme (Apulia, southern Italy)

This work describes the geochemical and hydrogeological characteristics of Santa Cesarea Terme, an active sulfuric acid speleogenetic system located along the Adriatic coastline (Apulia, southern Italy). It represents a very peculiar site, where rising thermal and acidic waters mix with seawater creating undersaturated solutions with respect to CaCO3, able to dissolve and corrode limestone and create caves. The Santa Cesarea Terme system is composed of four caves: Fetida, Sulfurea, Gattulla, and Solfatara. Hypogene morphologies and abundant deposits of native sulfur (especially in Gattulla Cave) and sulfate minerals are present in these caves. Fetida and Gattulla caves were investigated primarily because they are easily accessible throughout the whole year through artificial entrances, the other caves being reachable only from the sea. Geochemical analysis of water, monitoring of cave atmosphere, and measurement of the stable isotopes of S, O, and H helped to identify the main processes occurring in this complex cave system. In particular, changes in Ba2+ and Sr2+ concentration allowed for the identification of two main domains of influence, characterized by marine and rising acidic waters.


Introduction
Sulfuric acid speleogenetic (SAS) systems are hypogene caves (i.e., influenced by rising acidic waters) and, in contrast to epigene caves (i.e., formed by infiltrating waters), they typically are less involved in surface processes. Hypogene speleogenesis from a hydrogeological standpoint has been defined as "the formation of solutional-enlarged permeability structure (void-conduit systems) by fluids that recharge the cavernous zone from below, driven by hydrostatic pressure or other sources of energy, independent of direct recharge from the overlying or immediately adjacent surface" (Klimchouk 2017, p. 3).
Hypogene caves often develop along deep-rooted fractures and faults rather than bedding planes, and, being influenced by rising acidic waters and/or vapors passing through feeders (i.e. vertical and often narrow passages), produce morphologies such as cupolas, megacusps, blind chimneys, and domes (Audra et al. 2009;De Waele et al. 2016;D'Angeli et al. 2019a).
The most important reaction in SAS systems is the oxidation of H 2 S into H 2 SO 4 (Egemeier 1981), a strong acid which rapidly reacts with the carbonate (CaCO 3 ) host rock, producing gypsum by-products and releasing CO 2 into the atmosphere, which, in turn, further amplifies the carbonate dissolution.
Sulfuric acid speleogenetic caves are widespread all over the world (Galdenzi and Menichetti 1995;Hose and Pisarowicz 1999;Hill 2000;Jagnow et al. 2000;Summers Engel et al. 2004;Galdenzi et al. 2010;Klimchouk 2017), and are widely documented in Italy (De Waele et al. 2014), hosting at least 25% of the known worldwide systems in 12 regions located along the Apennine Mountain Chain, in Apulia, and in the main islands of Sicily and Sardinia (D'Angeli et al. 2019a).
The spatial and temporal concentration of H 2 S varies in both the cave water and atmosphere. In fact, at the same sample site, the variation of H 2 S concentration ranged between 17 and 750 ppm in Cueva de Villa Luz in Mexico (Hose et al. 2000), and from <0.1 to 20.5 ppm in the Frasassi caves (Jones et al. 2015). Along with H 2 S, also CO 2 and CH 4 are present in the rising fluids and intervene in some SAS reactions. As a matter of fact, rising fluids enriched in hypogene CH 4 and CO 2 may contribute to increase the concentration of greenhouse gases in the atmosphere, which is a further reason to study the CH 4 , CO 2 and H 2 S content in active SAS caves. Geological carbon sources may complicate the mass-balance measurement of the Earth's carbon cycle (Rye et al. 2012), with SAS caves behaving as methane producers, in contrast to epigene caves which are methane sinks (Fernández-Cortès et al. 2015). Important information regarding both H 2 S sources and reactions involved in the sulfur cycle (Eckardt 2001;Canfield 2001) can be inferred by sulfur stable isotopes.
The isotopic composition for sulfur is expressed in δ 34 S (‰) according to the following equation: δ 34 S is normalized to the Vienna Cañon Diablo Troilite (VCDT) standard, and the average δ 34 S values vary from −50 to +35‰. One of the most striking chemical signatures of SAS is the depletion in the heavier isotope (δ 34 S down to −25‰) in secondary gypsum ascribed to the bacterial sulfate reduction (BSR) or dissimilatory sulfate reduction (DSR). Studies on δ 34 S fractionation during sulfur oxidation have shown that the δ 34 S shift can be very small, producing no fractionation at high temperature (Ohmoto and Rye 1979) and small fractionation at low temperature (Fry et al. 1988). Nevertheless, in the Cerna Valley of southwest Romania, the wide range of δ 34 S values derive from thermal sulfate reduction (TSR) of sedimentary sulfates; the reaction of sedimentary sulfates with methane, produced by bacterial decay of nearby coal deposits, is very significant (Onac et al. 2011). These reactions are limited by methane and/or sulfate during reduction, and by oxygen during oxidation.
In contrast, sulfate-reducing bacteria are able to produce 34 S-depleted H 2 S (Strauss 1997). BSR causes δ 34 S values to shift by as much as −30‰, and sulfides resulting from this process have an average δ 34 S value of −12‰ (Seal 2006). DSR can produce sulfides depleted in 34 S by more than −40‰ (Zerkle et al. 2016 ), and elemental sulfur (S 0 ), can induce large isotopic differences, with H 2 S depleted in 34 S (+5 to +7‰) with respect to SO 4 2− (+17 to +21‰) (Canfield and Thamdrup 1994). Furthermore, the shift of stable oxygen and hydrogen isotopes (δ 18 O and δ 2 H), normalized using the Vienna Standard Mean Ocean Water (VSMOW), from the Global Meteoric Water Line (GMWL; D'Amore and Panichi 1987) for groundwater, can help to identify meteoric sources, fluidrock interactions along flow paths and temperature gradients, including low/high temperature mineral reactions, CO 2 exchange, hydration of silicates, and H 2 S exchange (Karolytė et al. 2017;Craig 1961).
Rainfall values of the GMWL in central Italy have been documented to range between −9.02 and −4.18‰ for δ 18 O and between −58.8 and −19.8‰ for δ 2 H (Bono et al. 2005), whereas for the Apulia area, Longinelli and Selmi (2003) reported values ranging from −5.31 to −4.91‰ for δ 18 O, and from −32.0 to −29.1 for δ 2 H, at the towns of Ceglie Messapica, San Giorgio Ionico, and Santa Maria di Leuca.
In addition, the study of trace and rare earth element (REE) contents provides information on the origin and sources of rising fluids (D'Amore et al. 2005)-for example, Sr and Ba are released by rock and soil weathering (Greve et al. 2007), Mn, Fe and Al derived from Fe-bearing alumino-silicates dissolution, and Fe, Ni, Cu and Zn can testify to the presence of living systems (Ali Khan et al. 2010). REE concentration in deep-sea hydrothermal environments is generally 10-10 3 times higher than in seawater (Wang et al. 2013), and a study Fig. 1 a Map showing the location of the Santa Cesarea Terme karst system in Italy and the piezometric contours (grey lines, depth in m) in the Salento region. The grey dots show the location of cities and towns. The enlargement (b) displays the location of Fetida (PU 102), Sulfurea (PU 103), Gattulla (PU 105), and Solfatara (PU 914) caves. The sites of water sampling in Fetida Cave (SCA1, SCAfin, SCB1, SCC1), Gattulla Cave (SCG1) and the Adriatic Sea (M1) are shown as colored dots. PU refers to the regional code in the register of natural caves of Apulia. c Two geological cross-sections (A-A′ and B-B′), modified after Cotecchia (2014). Country codes according to ISO (2020) of the shells of benthic animals living close to hydrothermal springs helped in recognizing the source and evolution of fluids (Zeng et al. 2018).
The aim of this work is to describe the hydrogeological characteristics and the baseline geochemistry in the Santa Cesarea Terme (SCT) karst groundwater system (southeastern area of Apulia foreland), consisting of four SAS cavesnamely Fetida, Sulfurea, Gattulla, and Solfatara (or Sulfurara;D'Angeli et al. 2017)-which open along the Adriatic coastline at present sea level ( Fig. 1). At SCT, the groundwater flows with piezometric gradient of about 0.03% toward the Adriatic coast, where it emerges in several locations, including the four caves along the coastline. Maximum specific discharge measured in boreholes inland is over 100 L/ s/m, and is likely related to a higher level of karstification along the main discontinuity systems (Ciardo 2006).
The Salento peninsula of southern Apulia is a carbonate plateau defined by a wide, WNW-ESE trending, antiform structure, dissected by normal and strike-slip faults (Tozzi et al. 1988), and characterized by shallow-water carbonate sedimentation (Mostardini and Merlini 1986). Since the Cretaceous, the Salento platform experienced multiple phases of transgression and regression that produced an overall stratigraphic succession characterized by several unconformities (Doglioni et al. 1994).
Quaternary tectonics characterize Salento terrains, with Pleistocene deposits cut by several normal faults (Martinis 1962;Moretti 1996). A model of the Apulian crust, reconstructed to identify potential detachment faults and the role of the Apennine orogeny, showed that the Moho depth in the Apulia foreland ranges between 26 and 31 km, with thickness of the Meso-Cenozoic platform increasing from Gargano (4-6 km) in the north to Salento (11 km) in the south (Amato et al. 2014). Given the widespread outcrop of carbonates, Salento terrains are extremely rich in karst landforms, with development and spatial distribution strongly controlled by tectonics, as shown by research at the surface and within cave systems (Onorato et al. 1999;Denitto et al. 2006;Festa et al. 2012;Pepe and Parise 2014). Hundreds of caves characterize both the Adriatic and Ionian coastlines of Salento, creating one of the main attractions for tourists in the area. Furthermore, the low-lying topography, with flat beaches approaching the seas, is frequently affected by sinkhole development, due to the rapid evolution of caves from the mixing of brackish and freshwater along the coasts (Delle Rose and The SCT area is located along the coast of SE Salento (Fig.  1). Here the Upper Cretaceous micritic Altamura Limestone Formation is overlain by the Eocene-Oligocene coral reef Castro Limestone Formation (Bosellini et al. 1999;Bossio et al. 2008). Both limestones are locally, in turn, unconformably overlain by the Oligocene Porto Badisco Calcarenites and the Messinian Andrano Calcarenites. Eventually, Plio-Pleistocene formations crop out, consisting of the yellow calcarenites of Uggiano La Chiesa Formation and the massive to poorly bedded and weakly cemented Salento Calcarenites. The whole carbonate succession is unconformably covered by terra rossa and talus deposits. Terra rossa may reach significant thickness (up to 4-5 m) in large topographic depressions.
Karst landforms appear to be structurally controlled in the study area, with the NW-SE fault systems playing a dominant role, as also observed in other sectors of Salento (Pepe and Parise 2014). Several normal and trans-tensional structures characterize the SCT area, mostly represented by NW-SE faults with related pull-apart features, and subordinate presence of SW-NE and N-S faults.
From a hydrogeological standpoint, the SCT area behaves as most of the Salento peninsula, a karst region with average rainfall between 500 and 800 mm/year. The surface runoff is typically limited to the occurrence of the most extreme and rapid rainfall events, or to the immediate aftermath of longlasting moderate rains. Rainwater infiltrates immediately into the subsoil in a diffuse way, or through the numerous dolines and swallow holes. Both types of contributions provide natural recharge to groundwater. There is a high degree of difficulty in delineating the boundaries of single catchments (Gunn 2007) and in evaluating the different components of the recharge, due to the prevailing flow through karst conduits. Therefore, due to widespread karst phenomena, at the surface and at depth, most of the geological formations are mainly characterized by secondary permeability, related to the presence of several structural discontinuities, with multistage development of karst processes. Interaction with marine water has to be taken into account also, with the peculiar lenticular shape characterizing the fluctuating fresh/saltwater interface.
In the hydrogeological map in Fig. 2, the geological formations have been merged on the basis of the role they play in the groundwater circulation, and also on the similarity in their hydrogeological characteristics and permeability values. Three main hydrogeological units have been thus recognized: the Pliocene-Pleistocene calcarenites, with medium permeability which is a function of grain size and lithostratigraphic conditions, with hydraulic conductivity values of about 10-5 m/s (Andriani and Walsh 2010); the Miocene calcarenites, with very low porosity, and permeability linked essentially to fracturing, leading to lower values of hydraulic conductivity (10 −6 m/s); and, finally, the set of Cretaceous-Paleogene limestones, characterized by fracturing and intense karst processes, with high permeability values (10 −3 m/s hydraulic conductivity; Cotecchia 2014, and references therein).
Notwithstanding the small groundwater flow gradient, a groundwater flow toward the sea is present (Fig. 2), as is common along most of the Apulian coastlines, where the groundwater flow is drained at many submarine springs. The presence of these springs was ascertained by infrared thermal surveys (Cotecchia 2014, and references therein), but their discharge is difficult to evaluate quantitatively, with the exception of the most significant ones .
Several faults have been recognized in the area, both along the shoreline and inland (Vigor 2014). Regarding their age, in the sectors where Plio-Pleistocene deposits crop out, some of these faults cross them, thus testifying to the fact that tectonic activity was not limited to post-Miocene time.
The village of Santa Cesarea Terme is located on a structural high of the Altamura and Castro Limestone formations, bounded by NW-SE structures that may be followed offshore (Zuffianò et al. 2013). The importance of the NW-SE fault system is further shown by the control it exerts on the spatial development of the four caves, and the presence of the thermal springs therein.
The emergence of deep upwelling water at the sea surface is visible as white plumes during quiescent marine conditions and winds. In the caves, degassing H 2 S from rising thermal waters is rapidly oxidized into H 2 SO 4 in condensation waters on the cave walls and ceilings, reacting with the host carbonate and replacing calcite with gypsum (D' Angeli et al. 2018). Dissolution rate monitoring in the innermost portion of Fetida Cave (SCB1 and SCC1 in Fig. 1) is still ongoing, but the first results showed that dissolution-corrosion is greater at the water-air interface than in subaqueous and aerated conditions (D'Angeli et al. 2019c). Several hypotheses have been presented for the origin of the sulfidic springs. In one scenario, elevated temperatures, several degrees higher than in shallow groundwater, are due to exothermic reactions of sulfate in the seawater interacting with organic matter in the Miocene calcarenites (Zezza 1980;Calò et al. 1983;Calò and Tinelli 1995). According to researchers, thermal waters could derive from connate fluids, confined at depth and associated with hydrocarbons in the Otranto Channel, rising along fault systems in the Dinaric foredeep bedrock, with elevated temperatures due to the regional geothermal gradient (Maggiore and Pagliarulo 2003).
For reference, sulfur-rich thermal water in the Salento karst region is not limited only to the SCT area, since it has been identified at other sites such as the Ionian coast near Ugento, where a borehole intersected sulfur-rich water below a thick layer of lower Pleistocene clays (Calò and Tinelli 2004).
The thermal properties of these caves have been known and described by scholars since the nineteenth century (Mauget 1875;De Giorgi 1882, 1901, 1922, and three have been used as spas since the 1930s (Corti 1992). These caves are a significant tourist attraction and economic resource in Apulia. All four caves are relatively short (30-588 m), low and wide, developed at modern sea level (maximum depth of 17 m in Fetida Cave; Table 1), and follow NW-SE discontinuities.

Methods
The hydrogeological units have been defined based upon the stratigraphic data from the official geological maps, integrated by the outcomes of the Vigor Project (Vigor 2014).
The piezometric contours at Salento, and in the SCT area as well, were reconstructed from a variety of sources, starting from the Water Safeguard Plan of Apulia Region, updated after Cotecchia (2014) and the Vigor Project. All sources used thousands of wells, where the water level measurements were carried out using a piezometric probe with centimeter accuracy. At SCT, for the detail shown in Fig. 2, more than 200 wells (Irrigation Authority of Apulia and Basilicata, Consorzio di Bonifica Ugento Li Foggi, private wells, etc.) were surveyed. The topographic elevations, to which the depths of the piezometric level were referred, were obtained from the Technical Regional Topographic Maps (scale 1: 5000).
Geochemical analyses of water and monitoring of the cave atmosphere were performed at Gattulla and Fetida caves, the only ones of the SCT karst system that are easily accessible during the whole year.
From October 2015 to November 2018, 27 waters samples were collected from four sample sites in Fetida Cave (SCA1, SCAfen, SCB1, SCC1), 7 from Gattulla Cave (SCG1), and four samples in the Adriatic Sea (M1)- Fig. 1, Table 2, Table S1 of the electronic supplementary material (ESM). At the sample sites close to the entrance of Fetida Cave (SCA1 and SCAfen), because of a storm and adverse conditions, it was impossible to collect samples on March 7, 2016. Water parameters such as pH, temperature (T), electrical conductivity (EC), and total dissolved solids (TDS) were measured using a Hanna HI991001 portable probe at the time of sampling (relative accuracies at 25°C: ±0.02 pH, ±0.5°C, ± 2% EC/TDS). Water samples were collected in 250-ml HDPE bottles, one acidified with 65% HNO 3 for cations, and stored at 4°C. The concentration of dissolved HS − was measured in situ using a Hach DR/2010 spectrophotometer. Geochemical analyses were performed at Politecnico of Torino University: cations were analyzed using an atomic absorption spectrophotometer Thermo S (AAS), anions with ion chromatography, NH + by colorimetry, and alkalinity (HCO 3 − ) through titration with HCl and methylorange as indicator. Saturation indices (SI) for calcite, dolomite and gypsum were calculated using the Debye-Hückel Eq. (SI = 0 equilibrium condition, SI < 0 undersaturation, SI > 0 oversaturation).
Gas monitoring was done during fieldwork on June 8, 2017, and on May 7, 2018, using a portable MSA Altair 4X multigas detector able to measure O 2 (0-30 ± 0.7 vol%), SO 2 (0-20 ± 1 ppm SO 2 ), H 2 S (0-200 ± 2 ppm), and temperature (−20 to 60 ± 0.5°C). Fig. 3 Cave entrances along the Santa Cesarea Terme coastline: a-b view from the sea, and toward the sea, of Sulfurea Cave; c the entrance of Fetida Cave as seen from the sea. Note the presence of a vertical discontinuity, NW-SE oriented, at the Fetida Cave entrance Two samples of water in Gattulla Cave (at SCG1 and at the natural entrance) were collected for sulfate δ 34 S analysis on December 10, 2016, using 1-L HDPE containers. Samples were acidified by adding 10 ml of a 0.25 M BaCl 2 solution to precipitate BaSO 4 , and the resulting powders were collected in Eppendorf tubes and analyzed at the ETH Zurich, Switzerland. The precipitates were wrapped in tin capsules with V 2 O 5 and converted into SO 2 in a Thermo Fisher Flash-A 1112 analyzer coupled with a ConFlo IV, interfaced to a Thermo Scientific Fisher Delta V isotope ratio mass spectrometer (IRMS). Isotope ratios are reported in the conventional δ-notation with respect to VCDT. The method was calibrated with the reference materials NBS-127 (δ 34 S = + 21.1‰), IAEA-SO-5 (δ 34 S = +0.5‰) and IAEA-SO-6 (δ 34 S = −34.1‰). Measurement reproducibility based on the repeated analysis of an internal standard was better than 0.3‰.
During the same field campaign in December 2016, waters in Gattulla (SCG1) and Fetida caves (SCAfen, SCC1), and waters from ceilings in Fetida Cave (SCAfin, SCC1) were sampled in 2-ml borosilicate vials to study δ 2 H and δ 18 O. These samples were analyzed at the Indiana University-Purdue University Indianapolis water isotope laboratory using a Picarro L2130-i analyzer coupled to an auto sampler and high-precision water vaporizer unit. Measurements were corrected for memory and drift following the methodology of Geldern and Barth (2012). Final values were corrected to the VSMOW scale using calibrated standards from Los Gatos Research. Precision for δ 18 O and δ 2 H was 0.1 and 0.6‰, respectively. In Table 2 all the sampling and monitoring campaigns carried out in Gattulla, Fetida and water collection of Adriatic Sea are summarized.

Results
Ion concentrations for all the six sample sites are reported in Table S1 of the ESM and show charge balances lower than 1.09% for all sites. The geochemistry of both sea and cave samples belongs to the Na-Cl-SO 4 type of waters (Fig. 4).
In the innermost zone of Fetida Cave (SCC1 and SCB1) temperatures higher than those close to the entrance (SCA1, SCAfen) are observed, and particularly in SCB1 where a clear site of rising sulfur-rich water was found that reached 29.1°C in December 2016. Nevertheless, the whole system is influenced by the sea; in March 2016, during a storm, temperatures of 14.6 (SCB1) and 15.3°C (SCC1) were recorded.
In Gattulla Cave (SCG1) temperatures range between 15.3 and 25.4°C and are slightly higher than seawater (M1) which vary from 15.7 to 21.1°C (Table S1 of the ESM).   Table 4) such as Cu and V, are more abundant in seawater (M1), whereas Ba, Sr and Zn concentrations are higher in the innermost portion (SCB1 and SCC1) of Fetida Cave. SCG1 in Gattulla Cave shows the highest concentration of rare earth elements (Fig. 5e, f; Table S2 of the ESM) such as Y, Er and Yb, whilst in the innermost portion (SCC1) of Fetida Cave, La, Nd and Eu are more abundant.
Saturation indices (denoted C-SI for calcite, D-SI for dolomite, and G-SI for gypsum) at the sample sites ( Fig. 6) showed that waters are mostly oversaturated with respect to calcite and dolomite, and undersaturated with respect to gypsum during the whole monitoring period (October 2015 to November 2018). Only sample sites SCAfen, SCB1, and SCC1 showed some samples to be undersaturated and/or close to the equilibrium line with calcite. In addition, in the two caves, some samples were slightly oversaturated or at equilibrium with gypsum. Calcite (C-SI) and dolomite (D-SI) saturation indices exhibit a clear correlation, whereas both gypsum (G-SI) versus calcite (C-SI) and gypsum (G-SI) versus dolomite (D-SI) do not appear to be correlated.
The sample collected for δ 34 S (December 10, 2016) at the entrance of Gattulla Cave returned values in the range of marine water, whereas in its innermost zone the values are more depleted in the heavier isotope (Table 3). For δ 18 O and δ 2 H, cave waters returned values similar to ocean water. Waters from ceilings in Fetida Cave are more depleted in the heavier isotope (Table 3).
The minimum, maximum, median and standard deviation values of O 2 , H 2 S, SO 2 concentration in the cave air during the fieldwork in June 2017 and May 2018 both in Gattulla and Fetida caves are reported in Table 4. Oxygen concentrations are at standard atmospheric values (20.8%). The concentration of H 2 S was quite variable at all sites during the monitoring time. The highest concentration of H 2 S was observed at SCB1 (15.40 ppm lasting for 30 s). The recorded temperature was always above 23°C.

Discussion
The rising groundwater in the caves of SCT shows a mean temperature of 23°C, 7 C°higher than the local average annual air temperature. Since Schoeller (1962) and White (1969) suggested that a mean groundwater temperature of at least 4-5°C higher than the regional annual mean may be considered slightly thermal, and Ford (1995) identified groundwaters elevated by 20 C°above the mean as a threshold for hydrothermal environments, the SCT cave system can be considered slightly thermal. Santaloia et al. (2016) describe SCT as a peculiar low-temperature geothermal system, located in a zone of low heat flux and great lithospheric and crustal thickness (120 and 45 km, respectively).
In general, the SCT system is influenced both by rising sulfidic fluids and seawaters, which tend to mix. Seasonal effects, modest water head variations (i.e. tides), and meteomarine conditions (smooth, rough, long-term rough water) strongly affect the SCT caves, causing geochemical variations, in particular at the cave entrances.
The innermost portion of Fetida Cave (in SCB1 and SCC1) is dominated by rising sulfidic waters where temperatures and pH can reach values of 29.0°C and 6.76, respectively, whereas its entrance (SCA1) and Gattulla Cave (SCG1) are controlled by mixing of rising sulfidic fluids and seawaters producing temperatures and pH ranging from 15.3 to 26.50°C, and from 7.01 to 8.16, respectively. Air temperature in the innermost portion of Fetida Cave is more stable with respect to the entrance of Fetida (SCA1) and Gattulla Caves (SCG1), which are more influenced by outside variations. However, the H 2 S concentrations recorded in the atmosphere and water of SCA1 and SCG1 suggest that sulfidic rising fluids characterize the underground environment of SCT. Sulfide-rich waters show a milky appearance (Hose et al. 2000;Macalady et al. 2006;D'Angeli et al. 2017) and, when conditions are favorable, they are visible as white plumes along the coastline.
Based on these results, the waters collected inside the caves (SCA1, SCAfen, SCB1, SCC1, SCG1) and along the coastline (M1) can be classified as Na-Cl-SO 4 waters (Fig. 4), with detailed geochemical composition (major and minor, trace and rare earth elements) reported in Tables S1 and S2 of the ESM. Saturation indices (Fig. 6) for calcite (C-SI), dolomite (D-SI) and gypsum (G-SI) showed these waters to be undersaturated with respect to gypsum, slightly oversaturated with calcite and oversaturated with dolomite. Only a few samples from SCB1 and SCAfen are at the equilibrium condition for calcite, whereas in SCC1 some are slightly undersaturated with respect to calcite.
The concentration of [HS − ] is quite variable over time and seems to be linked to episodic activities, as documented in Cueva de Villa Luz (Hose et al. 2000) and in Frasassi caves (Jones et al. 2015). Indeed, in SCT it was observed that the H 2 S concentration can suddenly change both in water and air (Table S1 of the ESM and Table 4).
In the water, [HS − ] tends to increase with higher temperatures and lower pH, both associated with warmer sulfidic  Taylor and McLennan (1985) inputs in the system (Fig. 7). The correlation between temperature and [HS − ] is a bit stronger (R 2 = 0.46) than that related to decreasing pH (R 2 = 0.36).
Moving from the innermost portion of Fetida Cave towards the Adriatic Sea, a general increase of water salinity (Na + , Cl − , and TDS; Fig. 8a (Fig. 8b). This would indicate [Ca 2+ ] in solution linked to an increase of [HS − ]. [HS − ] inputs from rising acidic fluids would also change the pH and temperature. Moving from the innermost portion of Fetida Cave towards the sea, pH tends to increase whilst T and [HS − ] decrease. Trace elements such as Sr and Ba, as shown in Fig. 9, are more abundant in the innermost portion of Fetida than in the seawater. It is well known that [Sr 2+ ] and [Ba 2+ ] can substitute [Ca 2+ ] ions (Lucas et al. 1990;Gabitov et al. 2013), and thus, the further dissolution-corrosion of carbonate host rock induced by rising acidic waters (D'Angeli et al. 2018) in contact with deep-  seated Ca-sulfates, would increase their concentration in solution. In contrast, V and especially Cu ions, which show a constant concentration inside the cave, tend to increase in seawater. Cu is essential for marine biogeochemical processes such as photosynthesis, respiration, N-cycling and Fe-assimilation. Actually, a low concentration of free Cu (lower than 10 −11 ppm) can limit phytoplankton production (Maldonado et al. 2006;Thompson 2015). Na + and Cl − are well correlated (R 2 = 0.95497) as shown in Fig. 10a. As expected, [HS − ] has an influence on [Ca 2+ ] (Fig. 10c), but not on [Na + ]/[Cl − ] (Fig. 10b). Figure 10d-f document a clear two-field distribution of the [Ba 2+ ]/[Sr 2+ ] ratio corresponding to: (1) rising acidic fluids; and (2) marine waters. The samples collected in the Adriatic Sea are always in the domain of marine influence, whilst all the other sample sites, from both Gattulla and Fetida caves, depending on sampling conditions, can fall in both the domains. These distributions would testify that, even if strongly dominated by rising acidic waters, the innermost portion of Fetida Cave is still subjected to marine influence.
[Cl − ] and [Br − ] are conservative and highly soluble ions, and their ratio can be used as a natural tracer to understand the origin and evolution of groundwater (Freeman 2007;Alcalá and Custodio 2008), and also to elucidate the source of salinity in coastal aquifers (Alcalá and Custodio 2004). [F − ] has been considered an important tracer both for seawaters and geothermal active areas (Edmund and Smedley 2013). Geothermal waters can constitute a source of natural fluorine contamination (Nordstrom and Jenne 1977).
In SCT, using [Cl − ], [Br − ], and [F − ], unusual ratios were observed ( Table 5) Almost all REEs show positive anomaly trends (Fig. 5e, f). As described by Lawrence and Kamber (2007), shalenormalized seawaters are characterized by various anomalies, The sample sites for gas detection are the same as for the water sampling with La as the most abundant element, followed by Y/Ho, Eu, Gd, and Lu. Since leaching of RREs La, Nd and Y increases with the consequent rise of acidity and temperature, the concentration of La, Nd and Y, shown in Fig. 5e, f, might be linked with rising sulfidic thermal fluids and sulfuric acid interaction (Kim et al. 2020).
Values of δ 34 S at the entrance in Gattulla Cave are close to modern Mediterranean seawater (+21‰; Böttcher et al. 1999), whereas waters sampled in the cave interior (SCG1) are more negative, likely due to BSR. Indeed, waters in both Gattulla and Fetida caves during calm seawater conditions exhibit abundant white floating and rock-attached filaments, which in Fetida Cave have been recently described by D'Angeli et al. (2019b) to be dominated by bacterial taxa involved in sulfur oxidation and reduction in aquatic, aphotic, microaerophilic/anoxic environments (Campylobacterales, Thiotrichales, Arenicellales, Desulfobacterales, Desulforomonadales).
Stable isotopes of O and H from Gattulla in SCG1 (1.19‰ for δ 18 O, and 7.79‰ for δ 2 H, respectively) are different from those observed in the waters of Fetida Cave (Table 3) collected at SCAfen (0.17‰ for δ 18 O and 1.85 for δ 2 H) and SCC1 2− (on the secondary y axis); c Mean values of pH, HS − (on the primary y axis), temperature (on the secondary y axis). The dashed line is for the elements on the secondary y axis (0.11 for δ 18 O and 1.96‰ for δ 2 H), which may indicate differences in recharge source inputs, water circulation, residence time, and/or degree of mixing between rising fluids and seawaters (Karolytė et al. 2017).
The waters from ceilings collected in two sample sites of Fetida Cave showed negative values −3.15‰ for δ 18 O and −3.50‰ for δ 2 H (SCC1) and −3.64 δ 18 O and −6.86 for δ 2 H (SCAfen), and plot above the GMWL in the area where H 2 S exchange might occur (Karolytė et al. 2017;Fig. 11). They likely derived from evaporation of rising sulfidic water mixed with seawater, which rapidly condensate on ceilings. The relative absence of stalactites, soda straws and stalagmites would testify to the limited amount of surface-water infiltration, as previously stated in the description of the hydrogeological setting. Nevertheless, further investigations on stable S isotope signatures will help to understand the source of H 2 S, thought to be associated with the reduction of sulfates contained in the seawaters (Zezza 1980) and/or Triassic deep-seated gypsum (Galdenzi and Menichetti 1995). Fig. 9 Graph displaying the tendency of trace elements moving from the innermost portion of Fetida Cave toward the entrance and the Adriatic Sea; mean values of Ba, Ni, Zn, Ni, V, Cu, (on the primary y axis) and Sr (on the secondary y axis) Fig. 10 a The correlation between Na + and Cl − , which is close to 95%; b The concentration of Na/Cl, which is shown not to be influenced by the increase of HS − ; c The concentration of Ca 2+ , as influenced by HS − . d-f The Ba 2+ /Sr 2+ ratio, compared with Mg 2+ , Ca 2+ /HCO 3 − , and Ca 2+ /SO 4 2 − , respectively; in these three graphs it is possible to see two main domains: (1) rising acidic fluids; (2) marine. The legend is displayed in f These results, together with previous works (Santaloia et al. 2016;Zuffianò et al. 2018), suggest the SCT coastal karst represents a localized geothermal anomaly, where moderately faulted zones, acting as the most relevant permeability sectors, alternate with less permeable areas.
As demonstrated by Calò and Tinelli (1995), rising sulfidic waters can reach the surface through deep tectonic fractures and karst flowpaths, generating bell-shaped upwelling of groundwater. Nevertheless, their behavior is extremely sensitive to both (1) modest water head variations (tidal fluctuations, increase of colder freshwater thickness), and (2) seawater conditions (smooth, rough, long-term rough water). Hydrogeological monitoring of thermal water flowpaths helped to assert the important influence of sulfidic fluids on the SCT karst system, especially during low and calm seawater conditions, showing evident zonation (Calò and Tinelli 1995). Rough sea conditions create different hydrodynamic situations (see Fig. 9 in Calò and Tinelli 1995), inducing the loss of water zonation, with more homogeneous values of temperature and salinity. The concentration of hydrogen sulfide is variable as a consequence of tides (Visintin 1945).

Conclusions
In Santa Cesarea Terme, rising sulfidic fluids mixing with seawater along the coastline produced a system composed of four small-sized caves-Fetida, Sulfurea, Gattulla, and Solfatara-formed along the main regional, NW-SE oriented, geological structures.
This study describes the hydrogeological and geochemical investigations performed during several field campaigns from October 2015 to November 2018. The hydrogeology of the system is complex, and still needs further efforts in order to reach substantial conclusions about the origin of the sulfur springs. The present state of knowledge can surely confirm the important influence of seawater movements on the local hydrogeology at SCT; this can significantly modify the elemental contents, temperature and pH of the overall solution. During a storm, in the sample sites where it was possible to collect samples, lower temperatures and higher values of pH were recorded, which are surely linked with seawater intrusion inside the caves.
Nevertheless, geochemical analysis showed the presence of a clear rising sulfidic fluid spot, recognized in the Cave showed values typical of modern Mediterranean seawater (+21‰) with a clear imprint of bacterial sulfate reduction (+15‰) likely induced by the sulfur-reducers found inside the white filaments coming towards the surface with rising waters. δ 18 O and δ 2 H of waters upwelling in Gattulla and Fetida caves may have differences in recharge source inputs, water circulation, residence time and/ or degree of mixing between rising and seawaters (Karolytė et al. 2017), whilst those collected from ceilings in Fetida Cave exhibit values typical of H 2 S exchange, and would derive from the evaporation of rising sulfidic fluids mixed with seawater and their subsequence condensation on ceilings. The variable concentration of Ba 2+ and Sr 2+ in the same sample site over time allowed for identification of both marine and rising acidic domains. In addition, the higher concentrations of La, Nd, and Y with respect to the other REEs would testify to a significant influence of sulfuric acid in their mobilization. The lack of common epigene speleothems such as soda straws, stalactites, and stalagmites reflects the scarce surface seepage, thus indicating minimal contribution by freshwater in the dissolution. Santa Cesarea Terme, already classified as a sulfuric acid speleogenetic system (D'Angeli et al. 2019a), can be also considered a great example for the study of karst evolution due to the mixing of sea water and deep rising sulfidic waters.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.