Hellenic karst waters: geogenic and anthropogenic processes affecting their geochemistry and quality

Karst hydrosystems represent one of the largest global drinking water resources, but they are extremely vulnerable to pollution. Climate change, high population density, intensive industrial, and agricultural activities are the principal causes of deterioration, both in terms of quality and quantity, of these resources. Samples from 172 natural karst springs were collected in the whole territory of Greece. To identify any geogenic contamination and/or anthropogenic pollution, analyses of their chemical compositions, in terms of major ions and trace elements, were performed and compared to the EU limits for drinking water. Based on chloride content, the collected karst springs were divided into two groups: low-chloride (< 100 mg L−1) and high-chloride content (> 100 mg L−1). An additional group of springs with calcium-sulfate composition was recognised. Nitrate concentrations were always below the EU limit (50 mg L−1), although some springs presented elevated concentrations. High contents in terms of trace elements, such as B, Sr, As, and Pb, sometimes exceeding the limits, were rarely found. The Greek karst waters can still be considered a good quality resource both for human consumption and for agriculture. The main issues derive from seawater intrusion in the aquifers along the coasts. Moreover, the main anthropogenic pollutant is nitrate, found in higher concentrations mostly in the same coastal areas where human activities are concentrated. Finally, high levels of potentially harmful trace elements (e.g. As, Se) are very limited and of natural origin (geothermal activity, ore deposits, etc.).


Materials and methods
From May 2016 to October 2022, 172 karst water samples were collected along the Hellenic territory ( Fig. 1) with their water chemistry being analysed at the laboratories of Istituto Nazionale di Geofisica e Vulcanologia (INGV-Palermo).
Samples were taken almost exclusively from natural springs. Only two samples (Mavrosoulava and Kaissarianis) were taken from drillings tapping karst aquifers. Sampling sites were selected mostly basing on the spring mean flow rates (> 50 L s −1 ). Only 13 samples were collected from springs with mean flow rate between 20 and 50 L s −1 . Approximate position and flow rate data were taken prevailingly from the catalogue of Hellenic karst springs made by HSGME (Hellenic Survey of Geology and Mineral Exploration (former IGME) in the 1970s and 1980s [25][26][27][28][29][30][31][32][33][34] . Unfortunately, this catalogue does not cover the whole Greek territory, not comprising a few important areas (Attica, Epirus, Central Macedonia, Chalkidiki, and the Aegean islands). For these areas information was obtained from different publications (e.g. 20,[35][36][37][38][39]. It is worth mentioning that mean flow rates may have changed significantly since the time that the original measurements took place. Indeed, three of the springs included in the catalogues with high measured flows at the time of the compilation, were at the time of our visit completely dry. Even though this work has covered the vast majority of karst springs with the highest water flow on the mainland, it should be noted that many big springs have not been sampled due to either imprecise geographic indications, or inaccessibility. Nevertheless, more than 80% of the catalogued springs with flow rates > 50 L s −1 have been analysed in the present study. Sampling date, geographical coordinates and mean flow rate are taken from literature 20,[25][26][27][28][29][30][31][32][33][34][35][36][37][38][39] of the collected springs and can be found in Table SM1. Physico-chemical parameters, (water temperature, pH, redox potential (Eh), and electric conductivity (EC)) were measured in situ by portable instruments, whilst total alkalinity was determined by titration with 0.1 M HCl on unfiltered samples, expressed as mg(HCO 3 − ) L −1 . Water samples were filtered (0.45 µm MF-Millipore cellulose acetate filters) and stored in LDPE bottles for anion and isotope determinations, while an aliquot for the determination of cation contents was stored in Polypropylene (PP) bottles and acidified with ultrapure concentrated HNO 3 . Water chemistry was analysed using standard methods 40 . Major anions (F − , Cl − , NO 3 − and SO 4 2− ) and major cations (Na + , K + , Mg 2+ and Ca 2+ ) were determined by ionic chromatography (IC; Dionex ICS 1100). Silica (SiO 2 ) was determined with Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES; Jobin Yvon Ultima 2).
For trace element analysis, filtered samples were stored in 50 ml PP bottles and then acidified to a pH of ~ 2 with ultrapure concentrated HNO 3 . Twenty-five trace elements (Li, Be, B, Al, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn,

Results
Major ions. Minimum, maximum and median values of the physico-chemical parameters as well as chemical compositions and saturation indexes of the main minerals of the collected karst water samples are reported in Table 1. The complete dataset can be downloaded from the Earthchem repository 44 .
Temperature values range between 5.6 and 33.5 °C. The highest values were measured in Glyfa, Gouvo, and Kalamos Tsirloneri karst springs; sampling of these springs occurred during the summer period at the first accessible point, far from the main stream. Indeed, the strongest emission points in these springs are surrounded by reeds that make them inaccessible to sampling. For these springs the measured temperature was not considered representative of the groundwater conditions before emergence. The pH values vary from 6.5 to 8.5, whereas EC ranges from 174 to 31,400 μS cm −1 and Eh from − 38 to 399 mV.
The major ions show a large range of concentrations, sometimes four orders of magnitude. According to the median value, the concentration of, respectively, cations and anions decreases in the following order of abundance Ca 2+ > Mg 2+ > Na + > K + and HCO 3 − > SO 4 2− > Cl − > NO 3 − > F − (Fig. 2a). www.nature.com/scientificreports/ All water samples were plotted in a Langelier-Ludwig diagram (Fig. 3), where three groups with different geochemical compositions can be recognised: (group a) characterized by Ca-HCO 3 composition and low Table 1. Statistical values of physico-chemical parameters, the chemical composition of major ions and trace elements, and of saturation index of the main minerals of the collected karst springs.  Trace elements. The collected karst waters were analysed also for trace elements determination (Table 1).
They show a wide range of concentration, generally, more than two orders of magnitude (Fig. 2b). Trace elements can be subdivided into elements never exceeding 1 µg L −1 (Be, Co, Cd, Sb, Tl), 10 µg L −1 (Ti, V, Cr, Ni, Cs, Pb, U), 100 µg L −1 (Al, Mn, Fe, Cu, Zn, As, Se, Rb, Mo), 1000 µg L −1 (Li, Ba), and 10,000 µg L −1 (B, Sr). Sometimes the concentration was below the detection limit. Considering the concentration below the detection limit as a missing value, only four elements (Rb, Sr, Ba, U) show no missing data, whilst five elements (Mn, Cu, Cd, Tl, Pb) had less than 60% of determined values. Beryllium shows values always below the detection limit.

Discussion
Geogenic processes. To better discriminate saline and non-saline karst springs, samples were subdivided into two groups, according to low (< 100 mg L −1 ) and high (> 100 mg L −1 ) chloride content. Furthermore, based on their calcium-sulfate composition, an additional group, that includes 10 karst springs was recognised (Acheron, Bobos, Doliana, Gorgogouvli, Gouvo, Mana Nerou, Nelles, Rogozi, Vathy, and Zavarina Limni). According to 24 , low chloride waters (group a) show the typical bicarbonate-alkaline-earth composition of groundwater circulating in carbonate aquifers. Carbonate dissolution process within the aquifers is confirmed by a good positive correlation between Ca 2+ + Mg 2+ and HCO 3 − along the 1:1 equivalent ratio line (Fig. 4a). Saline and sulfate waters (groups b and c), instead, have an excess of Ca 2+ and Mg 2+ respect to the 1:1 equivalent ratio line; for the former, the excess can be explained with a seawater influence and is generally associated with a higher Mg 2+ /Ca 2+ ratio tending towards that of seawater (Fig. 4b). The low salinity waters display on the same   (Fig. 4c), the most saline waters fall along the seawater dilution line, confirming significant marine contamination of the aquifers. This is consistent with their coastal location. In the same diagram, some low saline waters show an excess of Na + with respect to the seawater ratio line, suggesting that water-rock interactions within the aquifer may modify this ratio. The ionic exchange process between Ca 2+ in water and Na + in clay minerals may justify this pattern 46 .
In a Ca 2+ vs. SO 4 2− binary diagram (Fig. 5a) group c samples plot along the 1:1 equivalent ratio line suggesting gypsum or anhydrite dissolution within their aquifer. To discriminate which sulfate-minerals have undergone dissolution, the saturation index of gypsum (Fig. 5b) and anhydrite were calculated; karst waters circulating in carbonate aquifers are all undersaturated both in gypsum and anhydrite, whilst the most sulfate-rich waters of group c reach saturation in gypsum but remain undersaturated in anhydrite ( Table 1). The sulfate-composition of these waters is consistent with their geological environment. Springs with high sulfate content are located in Epirus (Fig. SM2). Their waters circulate within the Triassic pre-rift sequence of the Ionian zone 47 , which mainly consist of alternating gypsum formations and carbonate breccias cropping out close to major faults. According to previous studies 20, 39 , the composition of sulfate springs is consistent with gypsum dissolution. On the other hand, for those samples collected near the shoreline, SO 4 2− content mostly derives from seawater intrusion although some contribution from gypsum dissolution may in some cases not be excluded (Fig. 5a,b).
Water quality. Water resources, in particular karst systems, are essential for the development of life. Thus, their management and protection are of crucial importance not only for human health but also for the correct balance of all terrestrial and marine ecosystems. Unfortunately, overexploitation and human activities (industry, agriculture, tourism) are the major responsible for the deterioration of water resources, in terms of quantity and, especially, quality.
Higher trace element concentrations are related to both natural and anthropogenic sources. In this respect, to recognize any anthropogenic impact on the aquifer systems, the knowledge of the hydrogeological setting together with the geological and structural features of the region is essential to discriminate the natural baseline 48 .
Salinity. The very great length of the Greek coastline, both absolute and relative to the extension of its territory, explains the high percentage of coastal karst aquifers feeding springs both at and under the sea level. The peculiarity of the Mediterranean area, which explains also the frequent occurrence of submarine karstic springs, derives from the important sea level drop that took place from 5.9 to 5.3 Ma before present during the Messinian Salinity Crisis 49 . If not isolated from the sea by impermeable sediments, these karst systems, which extend deep below sea level, may represent an easy inland access to contaminating seawater 50 . According to 24 , about 5% of the Greek karst hydrosystems is of poor quality due to seawater intrusion in coastal aquifers as a consequence of over-exploitation. Most of the karst springs sampled for this study have water of good quality for human consumption and are often used as drinking water. Only 15% of sampled springs have a high concentration  The distribution map of the sampled springs, which is subdivided in two classes (Cl − above or below 100 mg L −1 ), is shown in Fig. SM4 (in supplementary material). All but one of the springs with high chloride content are located along the coast of Greece.
To better discriminate the possible sources of chloride, the Cl − /Br − ratio was calculated (Fig. 6a). Most of the samples show a narrow range of values (200-500) similar to the seawater ratio (291-52 ). The most saline ones (Cl − > 100 mg L −1 ), as evidenced before, denote a clear influence from seawater intrusion, while those with lower salinity reflect the Cl − /Br − ratio of the meteoric recharge. Some low-salinity karst samples located in the Epirus area show higher chloride content than other nearby karst springs. Due to the great distance from the coast, the elevated chloride concentration cannot be related to present seawater intrusion. These samples show Cl − /Br − ratio > 800 and belong to waters circulating within the Ionian zone sequences, where evaporite outcrops are present. In this respect, these Cl − /Br − values may be ascribed to the dissolution of halite salt domes in the cores of anticlines 39 . Samples of Krya and Perama (IDs 100 and 101 in Fig. SM4), situated near Ioannina city, show the highest values of Cl − /Br − ratio, 2642 and 8329 respectively: these values may be alternatively ascribed to the use of road salts (Cl − /Br − > 5000 53,54 ). Indeed, these springs are located in an area where winter temperatures often drop below zero and are very close to main roads where salts are used for de-icing purposes. Resampling of the spring of Perama at the end of the summer season revealed a similar high Cl − /Br − ratio (6707) indicating that a natural origin from evaporite dissolution is the most probable explanation.
Comparatively, fewer water samples show Cl − /Br − ratios below that of seawater. These are generally waters with low salinity (Cl − generally below 10 mg L −1 ) and their Cl − /Br − ratios may in some cases reflect lower values of their meteoric recharge. Alcalà and Custodio 56 found that rainwater at high altitudes and/or inland areas are characterised by Cl − /Br − ratios lower than that of seawater (down to < 100). Indeed, most of these samples were taken at high altitudes and far from the coast. This situation does not apply to the sample of Gorgogouvli (low altitude and close to the coast; ID 129 in Fig. SM4) where Br − excess may be ascribed to anthropogenic sources, e.g., the use of agrochemicals 56 .
According to 13 , 80% of the total consumption of water resources in Greece is due to agricultural activities. Growth and yield of crops are strongly related to the quantity of dissolved salt in the soil waters. Although some plant species can grow in highly saline soils (e.g. halophytes), chloride, sodium, and boron have generally toxic effects on the growth of plants and may reduce the permeability of the soil 55 . In order to evaluate the salinity effects of soil water, salinity and sodium hazard index are used. The former considers the electrical conductivity of waters, indicating the value of 2250 µS cm −1 as the maximum salinity level in water for use in irrigation 55 . The potential sodium hazard is quantified using the sodium adsorption ratio (SAR), defined by the following equation:  55 . Twelve samples with EC > 10,000 µS cm −1 fall outside the graphs. C1 low, C2 medium, C3 high, C4 very high salinity hazard, S1 low, S2 medium, S3 high, S4 very high sodicity hazard. www.nature.com/scientificreports/ where the ion concentrations are expressed in milliequivalents per litre 57 .
The EC values of karst springs of this study range from 174 to 31,400 μS cm −1 , with 20% of the samples, almost all belonging to the saline group, showing values falling in the high or very high salinity hazard fields. The highest EC values are associated with the karst springs located in coastal areas and affected by seawater intrusion into the aquifers. Moreover, nearly all of these waters show the highest SAR values, indicating an elevated potential for toxicity to plants (Fig. 6b). On the contrary, most of the saline waters do not create soil permeability problems due to the high sodium contents (Fig. 6c). Only the spring of Perama (Epirus), whose salinity is likely derived from evaporite dissolution, shows a moderate risk of soil permeability reduction (Fig. 6c). The low chloride karst springs group fall almost all in the low and medium salinity hazard classes (Fig. 6b) with a low sodicity hazard (< 2). Only few of the low-chloride karst springs fall in the high salinity hazard class, while almost all of the sulfate karst springs are included in this class (Fig. 6b); their salinity derives from the dissolution of evaporite rocks, mainly gypsum, but their sodicity hazard remains negligible. All these waters, falling in the high salinity hazard field, can still be used for irrigation provided that the irrigated soils are well drained preventing salt accumulation 55 . Moreover, the cultivation of salt-tolerant plant varieties allows often the use of waters belonging to the class of very high salinity hazard if no salt accumulation occurs in the soil. Especially in the areas of Greece characterised by a semi-arid climate, the use of salt-tolerant varieties has long been introduced, allowing the cultivation of vegetables sometimes with water conductivity up to nearly 10,000 µS cm −158 .
Nitrate. Nitrate is the most abundant nutrient, but it is considered also the most widespread pollutant. Although it may have a natural origin, such as atmospheric deposition or decay of organic matter, the main contribution derives from the increase of anthropogenic activities. The main anthropogenic sources are N-based fertilizers, untreated domestic and industrial wastewater, old septic systems, or leachate from landfill sites 3,59 . Nitrate is, often, added in excess to the soil to increase its productivity and most of it is leached to the aquifers below. Although there is no general consensus on the danger to human health represented by nitrate itself 60 , it remains an undesirable constituent generally accompanied by dangerous components (i.e., toxic agrochemicals or harmful microorganisms).
According to 12 , nitrate pollution is the second major source of groundwater degradation in Greece Many aquifers in Greece display high nitrate content, exceeding the European maximum admissible concentration (50 mg L −1 ) for drinking water 12,61 , making them non-suitable for human consumption. The most affected aquifers, with values exceeding the European limits, are the Boeoticos Cephissos hydrosystems in Central Greece 17 , the Vocha plain in Korinthos prefecture 62 , Thessaly district 12 . The main source of nitrate is the excessive application of fertilizers (NH 4 NO 3 , (NH 4 ) 2 SO 4 , and nitrogen phosphate potassium) in intensively cultivated lands (such as for cotton, tobacco, and olive). Other sources of nitrate are septic tanks and untreated domestic effluent from abandoned wells in urban areas 62 .
In this study, nitrate concentration in karst springs never exceeds the European limit for drinking water (50 mg L −143 ), suggesting a good quality for the majority of the sampled karst groundwater. Nevertheless, high NO 3 − content was found in some springs, with values up to 47.6 mg L −1 (Fig. 7a). Three main log-normal populations were recognised from the probability plot using the partition procedure proposed by 63 (Fig. 7a). Population A is characterised by an average value of ~ 0.6 mg L −1 and is mainly represented by waters with the lowest NO 3 − concentrations (< 1 mg L −1 ). These populations can be considered representative of un-polluted water, representing the natural background conditions. Population B is characterised by a mean NO 3 − of ~ 4.8 mg L −1 , suggesting only limited input from anthropogenic sources. The third population (C) has the highest average www.nature.com/scientificreports/ NO 3 − , ~ 24.5 mg L −1 , and a 95th percentile of ~ 47 mg L −1 . The most representative springs of this population (evidenced in red in Fig. SM5) are located close to urban centres, farmland, or highly touristic coastal areas, clearly evidencing pollution issues. The distribution map of nitrate concentrations (Fig. SM5) can be compared to the maps of the main agricultural areas and the population densities in Fig. SM6 (supplementary material).
In a correlation with the altitude (Fig. 7b), the lowest nitrate concentrations were found in mountain areas, whilst the most polluted springs are located in coastal areas, which are heavily exploited for agriculture and tourism and where urban centres are widely present. Exceptionally, some springs (Petres, Kefalovryso Karpenisi, Santovou, Xino Nero, and Tria Piagadia-IDs 11, 33, 125, 146 and 171 in Fig. SM5), although located at about 600-800 m of altitude, display an elevated nitrate content (up to 23 mg L −1 ). These derive from higher altitude, urbanised, and intensively cultivated intramountain basins.
Arsenic. Arsenic is considered a highly toxic metalloid, which has harmful effects on human health, classified as a class1 carcinogen by the International Agency for Research on Cancer 64 . The European Council has established the value of 10 µg L −1 as the maximum contaminant level for drinking water. Arsenic contamination may have both natural and anthropogenic origins: it can be naturally derived from the chemical weathering of sulfide ore deposits or transported by geothermal waters, whilst the main anthropogenic sources are mining activity, coal combustion, and As-pesticides 65,66 .
Many regions of Greece, especially in the northern part, are affected by elevated concentrations in groundwater 67 . The highest As concentration was found in the groundwater of the geothermal area of Chalkidiki in northern Greece, with values up to 1000 µg L −168 . The main sources of As in Greece are geothermal fluids arising from active tectonic and volcanic areas 67 .
Data on arsenic in Hellenic karst water are very scarce 19,[21][22][23] . However, in the present study values exceeding the European limit were found in three karst springs (Fig. 2b): Tempi (ID 39; up to 17.0 µg L −1 ), Potamos (ID 63; 12.1 µg L −1 ), and Paleomylos (ID 51; 12.0 µg L −1 ). The three springs (shown in red in Fig. SM7 with their IDs) are found in the eastern part of Thessaly and Central Greece and the arsenic contamination can be related to the geological settings. Indeed, the involved karstic systems were formed within carbonate formations at the contact with metamorphic and metavolcanic formations of the Ampelakia Unit (Blueschist unit) and Pelagonian Unit 15 . Many occurrences of As-rich mineralisations have been found in the area mainly related to the metamorphic rocks 67,69 . In some cases, the As contamination can be related to the presence within the aquifer of As-rich Karst-Type Bauxites 70 . For comparison, the distribution of the main industrial areas and the main mineralizations in Greece are shown in Fig. SM8 (supplementary material).
Although not showing extreme As concentrations like that of many geothermal waters in Greece 71 , the impact of these waters should not be disregarded. Because these As-rich karstic waters were sampled from springs with large flows (up to more than 2000 L s −1 ), even concentrations not strongly exceeding the maximum allowed level correspond to large As fluxes that may have an adverse influence on the ecosystems fed by these waters.
Further discussion about other trace elements (Sr, Cr, Ni and Pb) in the karstic waters of Greece can be found in the supplementary material.

Conclusion
The main hydrogeochemical types of karst water in Greece are calcium-bicarbonate for hinterland springs and sodium-chloride for coastal karst aquifers. Furthermore, a third hydrogeochemical group of waters whose calcium-sulfate composition derives from the dissolution of gypsum within their aquifers has been recognised. Trace elements contents are generally low except for elements associated with carbonate or sulfate minerals dissolution (B, Sr and Ba). Drinking water limits are rarely exceeded except for parameters related to seawater contamination in the coastal aquifers (EC, Na, Cl, B). In these areas most of the human population and activities are concentrated and, therefore, also the highest nitrate levels are found, though always below the drinking water limit. Among the remaining elements only As and Se exceed in few cases their maximum admitted contaminant limits. Such exceedance could not be related to anthropogenic activities and probably derives from present or past hydrothermal activities.
Excluding those waters with EC > 10 mS cm −1 , most of the waters unsuitable for drinking purposes due to high salinity may still be used in agriculture to irrigate salinity and boron-tolerant plant varieties on well-drained soils that do not allow salt accumulation.
The present study, while not covering the totality of the Greek big karstic springs, represents a first attempt to give a homogeneous dataset on the geochemistry of the waters circulating in the karst hydrosystems of the country. This dataset gives precious information about the quality status of these waters, even though it considers only the main ionic species and a large set of trace elements. On this basis, further studies should investigate also possible microbiological contaminations, the presence of organic pollutants or other potentially harmful trace elements (e.g. technological critical elements) and should also define the origin of the few trace element contaminations found in this study. Furthermore, among these springs, those representing the most important water resources should be chosen to follow up in time the most important quality indexes, in order to correctly manage this precious asset.
Notwithstanding the above limitations, this study shows that at present the Greek karstic hydrosystems, at least those located far from the coast, have to be considered a still intact water resource of national interest. These systems, being generally prone to contaminant infiltration, have to be carefully protected. Most of the population and human activities are concentrated in coastal areas where the karstic aquifers are often naturally contaminated by seawater intrusion, although sometimes salinity is increased by overpumping. In recent times human activities, that have the potential to contaminate precious water resources, are extending also towards mountainous areas. Luckily many recharge areas of important karstic aquifers are included within natural reserve areas, but it is of www.nature.com/scientificreports/ utmost importance to preserve also those outside these areas. In order to succeed, different administrative policy is applied in the numerous hydrologic basins of Greece 72 , where strategic (principles and planning) and functional (implementation of measures and actions until the final user) management of the water resources takes place 73 .

Data availability
The datasets generated during the current study can be obtained from the Earthchem Repository 44 .