The concentration of carbon dioxide (CO₂) in the atmosphere is increasing mainly due to fossil fuel combustion and industrial processes. Since the beginning of the industrial revolution at the end of the eighteenth century, its level increased from about 280 ppm and exceeded the average yearly value of 413 ppm during the year 2021 (NOAA, 2021). Being one of the major greenhouse gases, such rapid increase has severe consequences on earth’s climate (IPCC, 2021). About one third of the anthropogenic CO₂ released into the atmosphere in the past two centuries has been taken up by the ocean (Sabine et al., 2004; Bindoff et al., 2019; Gruber et al., 2019). In aquatic systems CO₂ gas dissolves, hydrates and dissociates to form weak carbonic acid (Drever, 1997), and the pH is lowered according to the following reaction:

Current CO₂ emission rates exceed the buffering capacity of the oceans and cause a shift of marine carbonate chemistry and a decrease of pH that has been quantified in 0.1 units compared to the pre-industrial period (Haugan and Drange, 1996; Doney et al., 2009; Gattuso and Hansson, 2011; Jiang et al., 2019). Depending on different emission scenarios, models predicted that further CO₂ increase would cause an additional reduction of pH between 0.3 and 0.5 units by the end of the century (Caldeira and Wicket, 2005; Joos et al., 2011; Jiang et al., 2019). Business-as-usual CO₂ emission scenarios predict that atmospheric CO₂ will reach 750 ppm and pH levels will decrease to 7.8 by the year 2100 (Jiang et al., 2019). In addition to this, both surface temperature and heat content of the ocean have increased. Specifically at the ocean surface, temperature increased by 0.88°C on average from 1850–1900 to 2011–2020. Possible future scenarios anticipate that it will arrive at 0.86°C from 1995–2014 to 2081–2100 (IPCC, 2021). Similarly, ocean heat content increased by 0.28–0.55 YJ between 1971 and 2018 and will probably continue to increase until at least 2300 (IPCC, 2021). This projection applies also for low emission scenarios due to the slow circulation of the deep ocean.

Many studies evidenced that ocean acidification (OA) will exert significant and sometimes unexpected effects on marine ecosystems (Jiang et al., 2019). Because these changes decrease the saturation state of the carbonate ion (CO₃^2–) in seawater, organisms relying on calcification for growth or protection are assumed to be most severely affected (Doney et al., 2012). On the contrary, photosynthetic organisms, such as seagrass and algae, may benefit from the increasing pCO₂ which is an essential resource for their photosynthesis and survival (Fabricius et al., 2011; Koch et al., 2012; Russell et al., 2013). It should be mentioned that even though laboratory experiments documented the benefits of OA on seagrass growth, anthropogenic stressors might counterbalance positive effects of increased CO₂ and have likely blocked potential beneficial responses of OA (Koch et al., 2012; Doo et al., 2020). To face the problem of atmospheric CO₂ increase, apart from the most logical solution remaining the strong reduction of anthropogenic CO₂ emission, one remedy proposed is the geologic carbon sequestration. CO₂ capture and storage (CCS) systems concentrate and transfer liquid CO₂ into storage sites, including sub-seabed deep geological formations such as exhausted oil or gas reservoirs. This approach is considered promising, since technically feasible (IPCC, 2005), but as with all other human technologies, it is not exempt from drawbacks. One of these drawbacks is the possibility that the chosen reservoir is not perfectly sealed and undergoes CO₂ leakage (Monastersky, 2013). If these reservoirs are offshore, CO₂ leakages from CCS can drive strong local seawater acidification (Blackford et al., 2014), exceeding the values predicted by the worst scenario of climate change. Moreover, in the case of a CO₂ leak from a storage site, the gas will also acidify the pore water in the sediments surrounding the storage site (Millero et al., 2009). This may also increase the release of harmful elements from the sediments creating an additional negative impact on the marine environment (De Orte et al., 2014; Foo et al., 2018). Flohr et al. (2021) simulated CO₂ leakage from an offshore CO₂ storage site in the British sector of the central North Sea. The CO₂ release experiment (Flohr et al., 2021; Gros et al., 2021) lasted for 1 month and the authors illustrated that different approaches can detect, attribute and quantify the release.

Notwithstanding the increasing number of studies on the ecological consequences of OA and CCS leakage, many issues remain unexplored and, until now, the vast majority of them have been performed in laboratories mainly as short-term and univariate experiments (Cornwall and Hurd, 2016). To have a more realistic picture, experiments should be made on marine organisms in their natural ecosystems. In this sense, areas with natural CO₂ vents represent useful experimental locations to investigate the impact of OA on entire ecosystems (Hall-Spencer et al., 2008). Natural underwater vents of volcanic origin release gases composed mainly of CO₂ and may therefore represent a natural analog to study the impact of seawater acidification. The CO₂ vent areas are also perfect natural laboratories to study the impact of CO₂ leakage from CCS systems.

Few of these “natural laboratories” have already been used to study the effects of elevated CO₂ on ecosystems (Vizzini et al., 2010; Lauritano et al., 2015; Linares et al., 2015) sometimes evidencing the adaptation of complex ecosystems such as coral reefs (Golbuu et al., 2016; Teixidó et al., 2020).

These vent sites allow to study different habitats, including shallow coral reefs in Papua New Guinea, Japan, and Northern Mariana Islands (Enochs et al., 2016; Golbuu et al., 2016); seagrass meadows, macroalgae stands, and coralligenous in the Mediterranean Sea (Columbretes Islands, Spain—Linares et al., 2015; Ischia, Italy—Hall-Spencer et al., 2008; Vulcano, Italy—Boatta et al., 2013; Panarea, Italy—Rogelja et al., 2016; Methana, Greece—Baggini et al., 2014); as well as in the subtropical North East Atlantic reefs (La Palma, Canary Islands—Hernández et al., 2016). However, a larger representation of environments is needed to predict the biological and ecological consequences of OA.

Our study will give a first catalog of the gas vents within the Aegean Sea comprising a description of the areas. It will provide important information to researchers who study the impact of enhanced seawater CO₂ concentrations related to increasing CO₂ levels in the atmosphere or even to leaking CCS systems like (i) extension and morphology of the exhaling area; (ii) preliminary gas flux estimations and geochemical characterization of the gases; (iii) presence of possible confounding factors as for example emission of thermal waters and/or hydrogen sulfide, iron oxi-hydroxide flocculation. The geochemical characterization is based almost exclusively on literature data that are gathered together with some new results and are made available to the reader in Supplementary Table 1, while a description of the degassing sites and a rough estimation of the gas fluxes are available in Table 1. We also present unpublished data on pH and temperature measurements of surface seawaters in four areas affected by the submarine degassing (Supplementary Table 2).

The Aegean Sea (Figure 1) is located in the eastern Mediterranean and is a rift formed in a “backarc” setting. It is situated in the upper plate of the Hellenic subduction zone and west of Anatolia, where active tectonics is observed. In fact, the northern Aegean Sea is a part of the Eurasian plate and the boundary with the Aegean microplate is called the North Anatolian Trough (NAT). The latter is the continuation of the North Anatolian Fault Zone (NAFZ) and is a ∼300 km long system of tectonically active marine basins, up to 1,000 m deep (Le Pichon et al., 1987; Taymaz et al., 1991; Kreemer et al., 2004).

The thinning of various tectonic units mainly emplaced during the Upper Cretaceous– Paleocene convergence–collision processes has resulted in the creation of the basin (Boccaletti et al., 1974; Robertson et al., 1991). It should be noted that the Hellenic subduction system was active since at least the Late Cretaceous, while the “backarc” rift was developed during Eocene-Early Miocene (Agostini et al., 2010; Jolivet et al., 2013). Despite the long-lasting formation of the Aegean basin (∼40 Ma), the extension rate is relatively low, so that the oceanic crust was not generated (Agostini et al., 2010).

Nowadays, the extension is seemingly localized around the Corinth-Patras rift (southern Greece), however; it was widespread during the Miocene (Sébrier, 1977; Mercier et al., 1979). Oligo-Miocene extensional metamorphic complexes outcrop in the Cyclades archipelago and the northern Aegean Sea (Lister et al., 1984; Gautier et al., 1993). The extension has proceeded from north to south, while the subduction front was retreating southward (Lauritano et al., 2015).

This geodynamically active regime is also characterized by intense seismic activity (Taymaz et al., 2007), by the presence of the south Aegean active volcanic arc (SAAVA) (Fytikas et al., 1984) and anomalous geothermal gradients (Fytikas and Kolios, 1979). Similar to other regions of intense geodynamic activity, extensive geogenic degassing takes place (Daskalopoulou et al., 2018a, 2019a) with gas manifestations being widespread both on land and underwater.

A total of 10 areas characterized by submarine degassing were documented and sampled along the Aegean Sea. Table 1 summarizes the general characteristics of the underwater sampling sites. A brief description of the degassing sites and in situ observations documented during the field campaigns are presented in this paragraph. Where possible, underwater filming allowed us to document and describe the degassing areas, and estimate the gas fluxes (as described in Supplementary Material).

In the current study, a total of 122 data from submarine gas manifestations are presented. This dataset comprises both literature (Chiodini et al., 1998; Shimizu et al., 2005; Kyriakopoulos, 2010; Carey et al., 2013; Tassi et al., 2013; Rizzo et al., 2016, 2019; Daskalopoulou et al., 2018a, b, 2019b, 2021a; Tarchini et al., 2019) and 7 unpublished results from submarine degassing areas distributed in ten areas of the Aegean Sea; seven of which belong to the SAAVA. Gases were collected using the inverted funnel method (Figure 5). Sample IDs, coordinates, references, gas flux estimations and their chemical and isotopic content are found in Supplementary Table 1. pH and temperature data of seawater affected by the submarine degassing are also presented in Supplementary Table 2. Details on the sampling techniques and the laboratory methods used are found in Methods section in Supplementary Data Sheet 1.

No samples plot close to the atmospheric point (Kipfer et al., 2002) in the CO₂-N₂-O₂ ternary diagram (air in Figure 6A), thus excluding important air contaminations. The vast majority of the samples present N₂/O₂ ratios higher than the ratios of air saturated waters (ASW; Kipfer et al., 2002), suggesting that the atmospheric component deriving from meteoric recharge has been modified by microbial or inorganic redox reactions that took place in the subterranean circuit. In their majority, gases are rich in CO₂, while few samples have N₂ or CH₄ as the prevailing gas species (Figure 6B). Geographically, gases rich in N₂ (Santorini, Nisyros, Kos, Methana) are distributed in the SAAVA and just after SAAVA to the north (Euboea), while CH₄ prevails in the gas vents of north eastern Aegean Sea (Samothraki) (Daskalopoulou et al., 2018a, 2019a). Figures 6A,B show the occurrence of dissolution processes, with the latter being likely responsible for the CO₂ loss and the enrichment of the less soluble gases (CH₄, O₂, and N₂).

An important atmospheric contribution for He is noticed for the gases of Samothraki Island as they plot close to the atmospheric point (Figure 7A, after Sano and Wakita, 1985). As expected, the vents located at the SAAVA present an enhanced MORB-type mantle contribution arriving up to ∼90% (Kolumbo), while the gases of non-volcanic areas show a more crustal origin for He (up to ∼95%). Figure 7B (Sano and Marty, 1995) reveals a mixed mantle-limestone origin for C for the great majority of the gases. The contribution of the organic sediment is relatively negligible. Some gases present CO₂/³He ratios that fall below the Mantle field, indicating CO₂ loss. This is likely due to the dissolution of CO₂ in water or to the precipitation of carbonates (Kanellopoulos, 2012; Winkel et al., 2013; Stefánsson et al., 2016, 2017; Kanellopoulos et al., 2017). A change in the flow as observed from Daskalopoulou et al. (2019b) for the sample of Kos presenting a strong decrease in δ¹³C-CO₂ may have also contributed to the dissolution processes. Methane for most submarine gases plots in the field ascribed to abiogenic hydrocarbons emitted from volcanic-geothermal systems (Figure 7C, after Etiope and Schoell, 2014). Contribution from biogenic sources cannot be excluded due to the wide range (from ∼4 to 244) of CH₄/(C₂H₆+C₃H₈) ratios (Bernard et al., 1978). Some gases exhibit high δ¹³C-CH₄ and δ²H-CH₄ values. This points to either organic or inorganic CH₄ oxidation processes. It is worth mentioning that isotope fractionation for organic and inorganic processes follow different fractionation paths (details in Daskalopoulou et al., 2018a, 2019b). Low δ²H-CH₄ values of Milos can be explained by non-equilibrium fractionation of CH₄-H with either H₂O or H₂ (Botz et al., 1996). Biogenic origin is attributed to CH₄ for the area of Samothraki. In particular, low δ¹³C-CH₄ and δ²H-CH₄ values as well as intermediate CH₄/(C₂H₆+C₃H₈) ratio indicate mixing between thermogenic and microbially-derived gases. The latter has been attributed to CO₂-reduction by Daskalopoulou et al. (2018a) on the basis of carbon isotope fractionation factor between coexisting CO₂ and CH₄ (Whiticar et al., 1986).

The CO₂ loss is more evident in the binary plots of Figure 8. The positive correlation between He and N₂ indicates the impact of the CO₂ dissolution on the gases (Figure 8A). He and N₂, as well as CH₄ (Figure 8B), are less soluble respect to CO₂, hence the strong solubility difference between the gases in the marine environment may has resulted in CO₂ loss (Reid et al., 1987). In fact, D’Alessandro et al. (2014) showed that when a gas mixture ascends through non-saturated waters, solubility contrasts might enrich the less soluble gases. This is specifically applicable in reduced gas upflow conditions (Daskalopoulou et al., 2021b).

Figure 9 further evidences the impact of solubility-related processes on gas composition. The elevated CO₂/H₂S ratios of the gases in Kos, Gyali and Santorini demonstrate the interaction between magmatic gases and hydrothermal systems (Supplementary Table 1). This process, known as magmatic scrubbing (Symonds et al., 2001), occurs when ascending gases encounter any aquifer interposed between the source magma stored at depth and the surface (Di Napoli et al., 2016). The more water-soluble gas species dissolve due to the gas-water-rock interactions, thus modifying the composition of the primary magmatic gas phase. In their great majority, gases collected along SAAVA are poor in CH₄. This scarcity evidences the lack of underlying organic-rich source rocks in these areas. Few gases of Milos, Kos and Santorini are virtually enriched in CH₄ due to loss of CO₂ by dissolution as evidenced also in Figure 8B. Finally, some gases of Chalkidiki, Milos, and Euboea plot close to the atmospheric point, further demonstrating the CO₂ loss due to dissolution processes, but also indicating some contribution of the atmospheric component.

In order to further demonstrate the impact of water-gas interactions on the gas content, two samples for each site were collected at the degassing centers of Agia Paraskevi (Chalkidiki) and Agia Irini 2 (Kos). One of the two samples was taken in the usual manner from the emission site at sea bottom (for details see “Material and Methods” section in Supplementary Data Sheet 1), while the second at the sea surface after the gas bubbles have risen through the entire water column. Results evidence that sea bottom samples have CO₂ as the major component. In one case they show the presence of some H₂S. The superficial samples have much lower CO₂ concentrations, H₂S always below the detection limit, and become enriched in N₂, O₂, He, and CH₄ (Figure 10 and Supplementary Table 1). These, sometimes very strong changes, can be explained by two processes that drive the gas exchanges between the rising bubbles and the seawater: mixing between two end-members and fractionation due to different solubility. The first process accounts for the virtual enrichment of the less soluble gases of hydrothermal origin (He and CH₄) with respect to CO₂ (Figure 10A). The second process is responsible for the decrease of CO₂ and H₂S (hydrothermal end-member) and the increase in O₂ and N₂ (Air-saturated seawater end-member) (Figure 10B). The extent of the changes suffered by the ascending gases depends on many conditions, which are mainly temperature, area of the interaction surface and interaction time (i.e. distance to be covered from the sea bottom to the surface). In the case of dry gases (no water vapor) the temperature is generally that of seawater because even if the emitted gases are hot they rapidly equilibrate with the seawater temperature due to the water/gas mass ratio and the thermal inertia of water. Both the interaction surface area and interaction time strongly depend on gas flux, bubble dimension, and depth of the water column. Higher gas fluxes, greater bubble dimensions, and lower emission depths all reduce gas exchange between bubbles and water, limiting the changes in gas composition. In the case of the above mentioned sites, the high gas flux and shallow depth of Agia Paraskevi prevents strong compositional changes like those registered in the site of Agia Irini 2. For example, while in the first case 78% of the initial CO₂ content arrives at the sea surface in the second case less than 1% does (Supplementary Table 1).

Changes on water characteristics were documented for the areas of Kanavas and Paleochori (Milos), Therma (Kos), and Ilion (Euboea) (Figures 11A–D). These areas comprise widespread degassing vents that are characterized by elevated CO₂ contents (>90%) and presence of H₂S (Supplementary Table 1). pH transects that were performed along the coast revealed pH values lower than the value of average seawater (Supplementary Table 2). Such lowering of the pH is driven by the dissolution of CO₂ in seawater.

Daskalopoulou et al. (2018b) identified areas (on-land and in the sea) of intense degassing and anomalous CO₂ flux at both Kanavas and Paleochori that correspond to the sites where the lowest pH values are found (Figures 11A,B). These areas are connected to the main fault structures recognized in the island (Stewart and McPhie, 2006) that have likely taken part in the volcanic activity of Milos, acting as pathways for the ascending magma (Kokkalas and Aydin, 2013) and the uprising gases (Dando et al., 1995; Daskalopoulou et al., 2018b). Especially for Paleochori Bay, Aliani et al. (2004) and Khimasia et al. (2021) mapped minor fault structures in the bay with the latter using microbial mats and high temperatures (Khimasia et al., 2020) for fault identification. It is worth mentioning that the temperature profiles by Khimasia et al. (2021) took place at hydrothermal vents within the bacterial mats (Figure 11A). Intense on land and submarine degassing is also identified in the island of Kos at Therma with the CO₂ upflow being completely addressed to the hydrothermal component (Daskalopoulou et al., 2019b). Megalovasilis (2020) suggested that the low pH values (Figure 11C) are the result of either fluid mixing with seawater in the substrate or water-gas interactions. It is important to note that the lowest pH values in Milos and Kos are found in emission points characterized by high temperatures (Figure 12). This, combined with the waters’ enriched SO₄-Cl content (Li Vigni et al., 2021), evidences the impact of hydrothermal activity on the systems.

In the area of Ilion, the low pH values are concentrated in front and east of a thermal (>62°C; D’Alessandro et al., 2014; Li Vigni et al., 2021) spring at the road (Figure 11D). The spring, as well as many of the thermal water emissions in Euboea, is connected to one of the tectonic structures in the border with the Sperchios Basin-Evoikos Gulf graben. The spring water ascents from the deep and hot geothermal system of the area (Li Vigni et al., 2021), and according to D’Alessandro et al. (2014) is affected by the Quaternary volcanic system. The volcanic impact though is also evident at the eastern side of the coast, where the degassing is diffuse. There, rocks present a rusty color, which is indicative of the emission of iron-rich (probably thermal) groundwater in the area. It is worth mentioning that the mobility of Fe and other trace metals (e.g., Cd, As, Pb) is enhanced by reducing conditions (Tarasov et al., 2005). The main driver is the dissolution of reactive gases (e.g., CO₂, H₂S, H₂) that results in intense rock leaching (Aiuppa et al., 2000).

Submarine degassing may have an impact on marine environments through ocean acidification. Hence, there is a necessity to study and better understand the ocean and its components. The current work reviews all known submarine gas manifestations of the Aegean Sea and summarizes the geochemical processes taking place in the individual areas.

All in all, degassing occurs in both volcanic and non-volcanic areas and is associated with the complex tectonics of the individual systems. Carbon dioxide is the dominant gas species for most vents and is often related to volcanism, geothermal energy, and elevated heat flow (Fytikas and Kolios, 1979). On the other hand, sites where CH₄ is the dominant gas species are likely related to hydrocarbon reservoirs (Rigakis et al., 2001).

The isotope signatures of He for gases found in SAAVA yield an important mantle contribution, while a dominant crustal origin characterizes gases in non-volcanic areas. Carbon dioxide derives from mixed mantle-limestone sources for most samples and in cases exhibits unimportant contributions from organic sediment sources. Methane is attributed to abiogenic hydrocarbons discharged from volcanic-geothermal systems. Inorganic and organic CH₄ oxidation processes resulting in isotope fractionation have also been identified (Daskalopoulou et al., 2018a). Only at Samothraki, where it is the main gas species, CH₄ is of biogenic origin.

The impact of water-gas-rock interactions on the initial gas phase is evident as soluble gas species dissolve in the water. This results in their depletion and the consequent enrichment of less soluble gas species. This phenomenon was also noticeable while comparing the composition of the gases at the emission point on the seafloor and the sea surface after its rising through the entire sea column. In addition to gas content variations, pH transects were performed in 4 sites. These are characterized by volcanic/geothermal activity, have CO₂ as the dominant gas species, and presented lower pH respect to the average marine value.

Even though the impact of gases on marine flora and fauna was not investigated in the current study, it shouldn’t be disregarded. Various researchers (e.g., Boatta et al., 2013; Price and Giovannelli, 2017; Aiuppa et al., 2021; Caramanna et al., 2021) have already highlighted that shallow marine vents can provide us with an accessible and economic way to investigate the effects of CO₂ on the whole marine ecosystems. In addition to this, parameters like the presence of light, wave action, tides, the input of meteoric water, salinity variations, etc., can significantly influence the geochemistry of the vents and the microbial diversity and distribution (Giovannelli and Price, 2018).

It is important to note that this is a preliminary catalog of shallow submarine vents found in the Aegean Sea. Springs found in tectonic structures on-land close to the coast (Li Vigni et al., 2021) characterized by strong degassing (Daskalopoulou et al., 2019a) and soil alterations (D’Alessandro et al., 2020) can be good indications of nearby submarine degassing underscoring that the catalog has still to be completed. Nevertheless, we aim that this study will initiate further research on the OA in Greece and in other countries. Following the identification of new emission sites and the quantification of gas flow, research should move towards a bio-, hydro-, and geochemical monitoring direction. As a next step, the anthropogenic input has to be taken into consideration. Understanding and defining the impact of both geogenic and anthropogenic processes affecting the various spheres in systems such as the Mediterranean, will result in the improvement of not only the carbon cycle knowledge but also of the dynamics and vulnerability of individual systems. This research and our group invite researchers from related disciplines to use multiple approaches and investigate other aspects of this problem. Hence, we make results accessible as electronic supplements and we aim to publish them as stand-alone datasets with their own doi.

SC and KD contributed to the conception and design of this study. ML, GP, and SC collected the samples by diving. ML contributed to the laboratory analyses. KD and WD’A wrote the first draft while all authors contributed to manuscript revision, read and approved the submitted version. All authors were involved in the sampling campaigns.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.