This paper is based on the analysis of a total of 534 entries (i.e., ‘items’, see below) retrieved from the OA-ICC database. These included: 86% peer-reviewed journal articles, 4.5% PhD theses, 4.5% book chapters, 3% conference papers, and 2% other documents (i.e., gray-literature reports, etc.) (Fig. 2A). This reflects the fact that the scientific community is more interested in producing scientific material (peer-reviewed articles), while communication products dedicated to disseminate OA results to a broader audience, such as conference papers, books, reports and targeted policy-briefs, are not a priority.

On average, around 1 to 4 OA items were published annually before 2008. After this year, the number increased to 20 OA items per year, mainly consisting of book chapters, in parallel with the launching of the European Project on Ocean Acidification (EPOCA project - 2008-2012) and the publication of the CIESM report containing multiple book chapters tackling the topic of OA (CIESM, 2008). The number of items continued to increase over the years during the MedSeA project (2011-2014), eventually leading to a peak of 60 OA items in 2017 alone. This was followed by a remarkable decrease in OA-related items up till present ( Figure 2A ).

Data show that most of the OA publications are dedicated towards the understanding of the biological response to OA through experimental studies (33%) and research conducted in natural CO₂ gradients, particularly in CO₂ vent sites (29%), followed by biogeochemical studies intended to quantify the carbonate system and OA trends (19%). This is followed by studies in the field of biological observations (7.5%) studying the in-situ responses of various biological groups, followed by reviews (4%), modeling (3%) and socio-economic studies, as well as paleontological, technological, policy-aspect studies, and carbon dioxide removal known as CDR (2%). There is only one recent article in 2021 tackling the carbon dioxide removal, reflecting that the Mediterranean OA community is still far behind in evaluating and assessing CDR techniques that might be applied in this sea ( Figure 2C ).

The interest of Mediterranean countries in the topic of OA has clearly blossomed thanks to large, trans-boundary projects that integrated multiple Mediterranean countries, starting in 2008 and revamped in 2010 ( Figure 2B ). This study shows that over the years OA studies were mainly implemented in the Western basin of the Mediterranean with more than 65% of the total studies, while only 20% were conducted in the Eastern basin, followed by 15% at the Mediterranean scale ( Figure 2B ).

Specifically, Italy, France, and Spain are the countries with the highest number of OA studies in the region, with 143, 91, and 67 publications, respectively ( Figure 3A ). This explains the large number of OA studies in the Western basin, predominantly conducted by these countries ( Figure 3D ), mainly in the Tyrrhenian (32%) sub-basin where CO₂ vents are well studied, followed by the Liguro-Provencal (~ 26%) sub-basin where mesocosm studies have been implemented. These two sub-basins are then followed by studies conducted in the Adriatic (8%), Aegean (7%), Levantine (4%), Alboran (3%), Ionian (1.5%), and the Algero-Provencal (1%) sub-basins. The reason why the Algero-Provencal Sub-basin is the least studied can be explained by the almost non-existing OA items produced in the neighboring Mediterranean countries with a total of three in Tunisia, two in Algeria, and zero OA items from Morocco ( Figure 3 ).

More than 90% of all OA-related studies were performed by scientists that come from the Northern region of the Mediterranean ( Figure 4 ). Thus, current OA conclusions are predominantly derived from European institutions. In fact, there were only a total of seven joint OA studies between North and South (N-S) Mediterranean countries (~ 1%) found in the database, reflecting a significant lack of collaboration. However, OA items published between Northern Mediterranean institutions and research centers outside the Mediterranean were 2.5-fold higher than the ones conducted between N-S Mediterranean institutions (Med. & beyond in Figures 3 , 4 ). Similar to studies originated in the South, these few collaborative N-S studies were focusing on chemical, biological and modeling OA aspects and are sparse in time with a better productivity after 2010. Studies originating in the North were covering most aspects of OA ( Figure 3D ).

These observations are supported by the results of the survey made for this study, where participants (n = 47) expressed the lower availability of financial and technical resources for institutions located in the Southern shores of the Mediterranean, as well as the lack of cooperation with institutions from the Northern shores as the main reasons behind this difference. In contrast, the lack of expertise in the South, the poor research infrastructure in the South, and the low consideration of proposals from the South are considered somehow relevant reasons ( Table S6 ).

Surprisingly, OA items from authors not affiliated to Mediterranean countries are the second most productive group, with about 20% of all OA-related studies ( Figure 4 ). This is likely due to the classification of the Mediterranean Sea as a climate-change impacts’ hotspot (Giorgi, 2006), the uniqueness and vulnerability of its marine ecosystems (Balzan et al., 2020), and the presence of natural CO₂ vents and seeps used as proxies to evaluate the resilience of marine species towards OA. All of these factors have attracted the attention of scientists from outside the Mediterranean region, such as from the UK, USA, and Germany, as well as Argentina, Chile, Brazil, and China ( Figure 4 ). This may explain why CO₂ vents and experimental biology studies are the main type of studies tackled by non-Mediterranean countries ( Figure 3D ).

The bibliometric assessment clearly shows that a considerable effort has been made over the last decade to characterize the carbonate system in order to estimate the concentration of anthropogenic CO₂ that has already been absorbed, and to evaluate the corresponding pH trend in the Mediterranean. Most of the studies (51%) were conducted in the Western Mediterranean basin, followed by 36% in the Eastern basin, and 13% at trans-Mediterranean level. The Liguro-Provencal sub-basin is the most studied compartment of the Western basin, while the Adriatic sub-basin is the most studied in the Eastern basin ( Figure 5 ). Articles dealing with water exchanges between the Mediterranean Sea and the adjacent oceanic compartments (Atlantic Ocean, Marmara Sea and Red Sea), as well as between the Mediterranean sub-basins, correspond to 15% of the working dataset, with a special focus on the Strait of Gibraltar (12%).

To constrain the carbonate chemistry in seawater, two out of four carbonate system parameters should be measured: total alkalinity (A_T), pH, partial pressure of CO₂ (pCO₂) and/or total dissolved inorganic carbon (C_T) (Touratier and Goyet, 2011; Hassoun et al., 2015; Krasakopoulou et al., 2017). The vast majority of research dealing with OA chemistry includes two different combinations of the seawater carbonate system measurable parameters (pH and A_T: 34%, C_T and A_T: 26%).

In 16% of the scrutinized studies, three out of the four widely-measured carbonate system parameters were measured to quantify the CO₂-carbonate system ( Figure 6 ). In three studies, in addition to the pair pH-A_T, the individual concentrations of Ca²⁺, CO₃ ^2- and the stable carbon isotope ratio of dissolved inorganic carbon (δ¹³C_DIC) were determined. A single parameter was measured in 17% of the studies, thus providing limited information about the carbonate chemistry. Studies based on continuous measurements of surface water pCO₂ were less common (10%). Moreover, almost one fourth (24%) of the considered articles provided estimations of anthropogenic CO₂ penetration in the Mediterranean Sea mainly using the TrOCA approach (Touratier and Goyet, 2004; Touratier et al., 2007). Interestingly, in some of the studies more than one of the methods calculating anthropogenic CO₂ have been applied [e.g. Flecha et al., 2012, TrOCA, back-calculation technique (ΔC*) and φCT° method; Keraghel et al., 2020, TrOCA and back-calculation technique of Chen and Millero (1979)].

The overall analysis revealed a paucity and sparseness of measurements of the carbonate system parameters, since most studies rely on discrete water samples acquired across different periods of time, covering particular regions of the Mediterranean ( Table S4 ). This spatio-temporal scatter of the marine carbonate system observations is mainly attributed to resources’ availability (both EU and national). The present study reveals a gap in research conducted in the Aegean (4 articles) compared to the Adriatic (10 articles), although in both areas Eastern Mediterranean deep water mass formations take place, playing thus an important role in the sequestration of anthropogenic CO₂. Important studies on the Rhodes cyclonic gyre (Northern Levantine), where open-sea convection occurs in Winter, leading mostly to the formation of the LIW (Levantine Intermediate Water) but also episodically to deep waters (Nittis and Lascaratos, 1998), are completely missing. This is a critical gap since LIW is the most voluminous water type of the Mediterranean that spreads westwards, outflows into the Atlantic through the Strait of Gibraltar forming the warm and salty tongue of the Mediterranean waters that contribute to the global ocean circulation.

There are relatively few stations with time-series to monitor the carbonate system in the Mediterranean (GOA-ON Portal : Home), with the vast majority located in the Northern sector of the Mediterranean: DYFAMED and W1-M3A stations (offshore), SOLEMIO (Wimart-Rousseau et al., 2020) and Point B (Kapsenberg et al., 2017) stations (coastal) in the NW Mediterranean Sea, E2-M3A (offshore; E2-M3A Southern Adriatic Observatory - OGS), C1-Miramare (coastal; http://nettuno.ogs.trieste.it/ilter/GoTTs/en_index.html, Ingrosso et al., 2016a) and PALOMA (coastal; Cantoni et al., 2012) stations in the Adriatic, B1 (coastal) and B2 and A3 (offshore) stations in front of Lebanon-Levantine (Hassoun et al., 2019; Hassoun et al., 2022), G1, G2 and G3 (GIFT) stations in the Gibraltar Strait (Flecha et al., 2015; Flecha et al., 2019)]. However, datasets originating from these time-series stations have fueled only one-fourth of the published articles. High-frequency measurements of the seawater carbon chemistry provided by sensors on moored observatories, volunteer observing ships, floats and gliders seem to be relatively scarce in the Mediterranean contributing to around 10% of journal articles in the database, which possibly reveal two major challenges: 1- the high cost, and 2- the complexity of use, thus preventing the widespread utilization of in situ instrumentations (Byrne et al., 2010). In most of these studies, only pCO₂ in surface waters was measured (Hood and Merlivat, 2001; Turk et al., 2010; Turk et al., 2013; Merlivat et al., 2018), whilst in two studies the pH variability was recorded with a SENSORLAB-pH system located at a fixed mooring (González-Dávila et al., 2016) and with an ISFET sensor on an underwater glider (Hemming et al., 2017). In another study, both SAMI-pH and SAMI-CO₂ Submersible Autonomous Moored Instruments were used (Flecha et al., 2015).

The survey results validated the database analysis. Around 70% of respondents measure the carbonate system parameters in the following order: pH (86%), A_T (77%), C_T (49%), pCO₂ (43%), followed by carbonate ions (23%). Moreover, more than 75% of the respondents encounter problems in: (1) securing Certified Reference Materials (CRMs) mostly from the USA Dickson’s lab (> 85%), (2) assessing the quality of the carbonate system measurements mainly due to limited or no funds (64%), (3) shipment (18%), and other reasons such as limited CRMs production, difficulty in paying in foreign currency, and institutional administrative issues. Although 78% of survey respondents are able to share their OA chemistry data, only 30% are submitting it to the IOC-UNESCO for the reporting of the SDG indicator 14.3.1., while only few respondents are uploading their data in different databases such as PANGAEA (via OA-ICC), GOA-ON data portal, SOCAT and GLODAP. This study therefore reveals a serious issue related to data access and availability within the OA research community.

Ocean acidification trends also vary spatially across the Mediterranean Sea. The pH trends calculated since the pre-industrial era are mainly derived from two combinations of the carbonate system parameters (A_T-C_T) and (A_T-pH) based on three main approaches used to estimate the amount of anthropogenic carbon (C_ANT), a variable that cannot be measured directly (Álvarez et al., 2009; Vázquez‐Rodriguez et al., 2009; Álvarez et al., 2014; Hassoun et al., 2015; Touratier et al., 2016; Krasakopoulou et al., 2017). These approaches were applied in various parts of the Mediterranean, and are: TrOCA (Touratier and Goyet, 2004; Touratier et al., 2007), Transit Time Distribution (Waugh et al., 2006), and the Back-calculation (Vázquez-Rodríguez et al., 2009). Since the pre-industrial era, the acidification trend has varied between -0.055 and -0.157 pH units in both basins with slight differences between intermediate and deep water masses ( Table S3 ).

For the yearly pH trends, the Eastern basin presented homogenous trends varying from -0.0021 to -0.0024 pH units y^-1 in the upper 80 m, while high annual acidification rate was reported in shallow waters (< 5m) of the South-Eastern coastal area, namely offshore Lebanon (-0.009±0.004 pH unit y^-1; Table S3 ). The Western basin, however, showed wider annual pH ranges, varying between -0.0013 and -0.0015 pH.y^-1 in the upper 75 m, and from -0.0017 to -0.003 pH unit y^-1 in surface waters ( Table S3 ). The pH trends of coastal and offshore time-series do not appear to be significantly different, so far.

When compared to the Atlantic and the global ocean (-0.001 to -0.0026 pH units y^-1; Takahashi et al., 2014 and references therein, Bates et al., 2014), the Mediterranean Sea displays a wider range of acidification rates (-0.001 to -0.009 pH units y^-1; Table S3 ). The relatively higher estimates have been associated to a specific capability of this sea to absorb larger amount of C_ANT that is efficiently transferred to deep layers, compared to the global ocean (Schneider et al., 2010; Hassoun et al., 2015). In fact, C_ANT has already penetrated all water masses of the Mediterranean Sea (Touratier and Goyet, 2011; Hassoun et al., 2015); in its various sub-basins: North-Western Mediterranean (Touratier and Goyet, 2009), Algerian (Keraghel et al., 2020), Adriatic (Krasakopolou et al., 2011; Ingrosso et al., 2017), Aegean (Krasakopoulou et al., 2017), and the Levantine (Schneider et al., 2010; Touratier and Goyet, 2011; Sisma-Ventura et al., 2016; Hassoun et al., 2019). This is due predominantly to the following reasons: i) high A_T favoring the dissolution of atmospheric CO₂ (Álvarez et al., 2014), ii) low Revelle factor ranging between 9.4-11.2 (Álvarez et al., 2014), iii) deep water formation processes and the relatively short ventilation time scale of deep waters (Schneider et al., 2014; Stöven and Tanhua, 2014) and, iv) constant inputs of anthropogenic CO₂ at the Strait of Gibraltar (Schneider et al., 2010; Flecha et al., 2012). Similarly, A_T in the Mediterranean Sea (>2500 µmol kg^-1) is higher than the global ocean due to the high evaporation rates, and the inputs from rivers and the Black Sea (Copin-Montégut, 1993; Schneider et al., 2007). This peculiarity leads to an increase of the buffering capacity of the carbonate system, allowing to uptake a larger amount of CO₂ from the atmosphere (Álvarez et al., 2014; Hassoun et al., 2015), as discussed also by Palmiéri et al. (2015) who, in a modeling experiment, showed a net increase of CO₂ uptake (10%) in response to an increase of 10% of A_T. In addition, the overturning time of the Mediterranean waters is very fast and allows a complete renewal of its water in a short range of time (60 to 220 years; Stöven and Tanhua, 2014) in comparison to the global ocean (more than 1000 years; Khatiwala et al., 2012). As a consequence, surface waters affected by the increase of anthropogenic carbon in the atmosphere can transfer this signature to deep layers in a shorter time frame, making the whole sea quickly invaded by C_ANT.

The absorbed C_ANT is estimated to range between 29 and 137 μmol kg⁻¹, depending on the depth and type of the water mass ( Table S3 ). Moreover, the Mediterranean Sea imports from the Atlantic Ocean a large amount of C_ANT (3.5 Tg yr^-1, Schneider et al., 2010) that is dissolved in the surface waters entering the Strait of Gibraltar (Flecha et al., 2012). The intensity of the pH decreasing trend has been estimated by a few observational studies conducted on multi-annual and decadal scales (Yao et al., 2016; Flecha et al., 2019; Hassoun et al., 2019; Coppola et al., 2020 and references therein; Table S3 ). Along the water column, the pH decrease varies differently, with maximum and minimum decreasing trends in the surface and the intermediate layers, respectively ( Table S3 ). Contrary to what is reported for the global ocean (Santana-Casiano et al., 2007; Dore et al., 2009), deep layers of the Mediterranean also display a noticeable decrease of pH, slightly lower than the surface waters’ values. This vertical pattern is attributed to water column processes and dynamics. Surface water, being in direct contact with the atmosphere, is more affected by an increase in atmospheric CO₂. Deep waters also display an evident pH decreasing trend due to dense water formation processes that transfer surface water, enriched in anthropogenic CO₂ to intermediate and deep layers. In the global ocean, as the intermediate and deep waters are much older than the surface water, the anthropogenic carbon signal is not particularly evident. However, in the Mediterranean Sea, the intermediate and deep waters are much younger, therefore showing anthropogenic CO₂ invasion and consequently clearer acidification signals.

Acidification trends have been assessed also through model re-analysis. Cossarini et al. (2021) focused on the reconstruction of changes of pH and spatio-temporal dynamics of the major carbonate system parameters in the Mediterranean Sea during the last 2 decades, as a part of the Copernicus model-based reanalysis of biogeochemical parameters in the Mediterranean, now available as Copernicus Marine Service product (Data | Copernicus Marine). In this study, model simulations are constrained by qualified available information through data-assimilation techniques, and the final result -named reanalysis- can be considered as the best possible spatio-temporal interpolation of available information, since it is performed by using all the physical and ecological knowledge on ecosystem dynamic that is embedded in the model. The reanalysis suggests that, in the last 30 years, there were significant trends in all carbonate variables, with an increase of A_T (1.7 and 0.7 μmol kg^-1 y^-1 in the Western and Eastern Mediterranean basins, respectively) as well as in C_T (2 and 0.6 μmol kg^-1 y^-1), and a related decrease in pH of approximately -0.0006 to -0.0012 units y^–1 (higher in the Eastern sub-basins). Since those trends are computed over the last 30 years period, they cannot be linearly extrapolated over the last two centuries, which have been characterized by a constant increase in atmospheric CO₂. In fact, such an exercise would return higher pH changes than those computed in Palmiéri et al. (2015), which is coherent with the different assumption in the two studies.

Even though they are arguably not directly at risk from OA, ~ 29% of the studies evaluating the effects of OA on marine organisms focus on autotrophs (algae, phanerogams, phytoplankton; Figure 7 ). This is due to their presence at most investigated field sites, like CO₂ vent sites that represent 38.5% of biological study approaches and 71% of all field studies ( Figure 8 ). The vast majority of field studies using the natural pH gradients provided by CO₂ vents have been conducted in the Tyrrhenian sub-basin (90%), specifically in the volcanic Island Ischia, and to a lesser extent in Vulcano and Panarea islands (all in Italy). A smaller part (8%) has been performed in the Aegean sub-basin, namely in Milos and Methana islands (Greece). Proper site selection or experiments at a wide range of CO₂ vent sites would be necessary to assure the comparability of results with future conditions. This is notable because site-specific trace metals accumulation that can affect seagrass performance are often associated with CO₂ seeps. Currently, it is robustly clear that phanerogams (seagrasses) will be able to actually benefit from future elevated CO₂ levels, although many recent studies are showing positive to no effects in specific species (Koch et al., 2013; Hendriks et al., 2017b; Guerrero-Meseguer et al., 2020). Meta-analyses show neutral effects of OA (Kroeker et al., 2013) or slightly positive effects (Hendriks et al., 2010; Guilini et al., 2017) on macrophytes. The effects of increased CO₂ on autotrophs are difficult to predict because some species utilize carbon concentrating mechanisms that buffer their sensitivity to ambient CO₂ levels and require variable energy investments. However, it seems that photosynthetic responses to OA are highly variable throughout taxa and might be relatively small for most species. The photosynthetic benefits of high CO₂ could be minor relative to the cell's overall energy and material balances, or other negative effects, such as possible respiratory costs from low pH outweigh the benefits of photosynthesis at high pCO₂ levels (Mackey et al., 2015).

A major advantage of using pH gradients in CO₂ vent sites is the possibility of studying effects at a community level, under natural light, nutrients’ concentrations, and current conditions, which explains why 38% of OA biological studies adopted the field vent approach ( Figure 8 ). In many laboratory studies (>40% of biological study approaches, Figure 8 ), stable pH conditions are used to infer effects of OA, which might have limited comparability to realistic conditions in natural environments. Due to the complexity of studying whole communities, only ~11% of the studies (42 in total) have tackled the community-level ( Figure 7 ), with more than half (57%) of community studies implemented in field settings and the majority (45% of community level studies) in field vents. Laboratory settings were used in 14% of the studies, while 26% of the studies used a review of existing literature to unravel the effects of OA on communities.

The results based on the OA-ICC database are in consonance with the survey outcomes, showing that 40% of participants are doing only field biological observations, 20% are doing only experimental biological studies, and 40% are using both biological approaches. Although field biological observations can realistically reflect the feedback of organisms and ecosystems towards OA, there are significant obstacles related to limited-funding needed to sustain long-term observations (81%), the necessity of long-term monitoring to derive conclusions (69%), and the difficulty in distinguishing the main drivers of change (41%). Whereas for experimental studies, difficulties are mainly attributed to sustaining long-term experiments (66%), the complexity of factors that should be taken into consideration (47%), and the absence of unified methodologies (25%), according to the Mediterranean OA community (survey; Table S6 ).

The main studied biological groups are algae (36%), phytoplankton (36%), mollusks (36%), zooplankton (32%) and corals (27%) with lesser to non-existent interest in other groups, in harmony with the results deduced from the database analysis. Moreover, the main studied processes are the abundance (73%) of organisms, followed by community composition (40%), primary production (37%), calcification and growth (37%), then mortality (33%), morphology (33%) and photosynthesis (30%), with less interest in other biological processes. These results reflect the gap in assessing the potential effects of OA on some biological groups (i.e., nematodes, vertebrate organisms, worms, etc.), and key processes (i.e., performance, die ecology, cellular and sub-cellular responses, production of biotoxins, bioaccumulation of chemicals, etc.).

Of all studies looking at the effects of OA on marine organisms, ~78% (337 studies, Figure 9 ) looked at the conventional OA single driver (pH or CO₂). Nevertheless, as future scenarios not only predict decreasing pH (increasing CO₂), but also forecast changes in other key drivers, it is becoming crucial to include all relevant drivers when evaluating organismal responses. Temperature is especially relevant as the Mediterranean Sea ranks among the ocean regions warming up the fastest (Marbà et al., 2015) and, together with habitat degradation, eutrophication, hypoxia, ocean acidification and biological invasions, warming has been identified as a major driver of change for the Mediterranean biodiversity since the last century (Coll et al., 2010; Marbà et al., 2014; Marbà et al., 2015; MedECC, 2020; Hassoun et al., 2021).

Coupled atmosphere-ocean general and regional circulation models project rapid warming in the Mediterranean region and an increase of the frequency and intensity of heat waves for the 21^st century that could affect coastal marine ecosystems (Jordà et al., 2012). Because of the semi-enclosed nature of the Mediterranean Sea, marine species, in particular benthic organisms, have limited scope to avoid new thermal conditions resulting from global warming by keeping pace with poleward migration of isotherms as occurs elsewhere (Poloczanska et al., 2016). Temperature is also the most used parameter in studies using more than one driver, with 14% of all experimental studies on OA ( Figure 9 ) including temperature, which represents 58% of the studies taking into consideration multiple drivers. Other drivers are also included in multiple-stressor studies, such as nutrients (27% of studies using multiple drivers), light (11% of studies looking at multiple drivers), salinity, and UV (2% each of studies looking at multiple drivers). Manipulation of multiple drivers is even more logistically challenging in field conditions, as reflected by the high percentage of single driver studies (92, 80 and 93% respectively for field, field mesocosm and field vent studies) compared to 56 and 38% of single driver studies in laboratory and laboratory mesocosm settings, respectively. Only 3% of studies looked at more than two drivers, and only one laboratory study included four drivers (0.2%), OA, light, nutrients and UV when evaluating phytoplankton responses.

These results also reflect the community’s perception ( Table S6 ), since in addition to controlling pH/CO₂, the main factors manipulated during multi-stressors experiments are temperature (81%), oxygen (42%), salinity (38.5%) and light (35%), while pathogens and bioturbation are almost not considered.

Heightened anthropogenic pressures that affect both coastal and marine ecosystems of the Mediterranean Sea basin were recognized by all surrounding countries as they adopted the first-ever Regional Seas Programme under the auspices of UNEP in 1975. Through this Programme, several basin-wide monitoring initiatives and programmes were subsequently implemented to provide baseline information on the state of the health of marine ecosystems as well as to establish monitoring guidelines to address marine pollution in the Mediterranean Sea (Galdies, 1999; Galdies, 2008).

The Mediterranean Sea’s unique resilience with respect to a continuous increase in OA deserved special attention (see section 3), as this will in turn help inform and direct decision- and policy-makers to focus on targeted, realistic measures. Unfortunately, there is still a general lack of understanding at the scientific and public level regarding best policy options, which can be attributed to the complex management of both direct and indirect impacts (Riebesell and Gattuso, 2015; Osborne et al., 2017). These barriers are leading to inadequate access and sharing of data and information, and are pushing back the much-needed progress towards effective decision-making at the regional level (Cramer et al., 2018).

National policies could provide incentives to penalize the use of fossil fuels and at the same time encourage a move towards a greener economy. However, despite increasing knowledge of OA, the global community does not have any legal provisions that specifically address OA mitigation strategies. According to Harrould-Kolieb and Hoegh-Guldberg, 2019, this has led to “significant gap in the global governance of this issue with no multilateral agreement understood as having jurisdiction over the mitigation of rising ocean acidity”. One possible reaction to this was the suggestion to formulate a number of proposals ranging from the elaboration of a specific international agreement focusing on combating OA (Kim, 2012), to the inclusion of the mitigation of OA within the core obligations of UNFCCC (Harrould-Kolieb and Herr, 2012), or even proposing a new treaty on OA (Lamirande, 2011). Currently, the only comprehensive solution that the global community has to address OA is to drastically reduce, and as soon as possible, global anthropogenic CO₂ emissions and the resulting uptake of this gas by the ocean. At the Mediterranean Sea scale, multi-level policy tools targeting CO₂ emissions and adaptation strategies do exist in Europe, such as the EU Climate Action and the European Green Deal. However, these are not designed to address OA specifically, but can only be marginally adapted to do so (Billé et al., 2013; Jagers et al., 2019).

Until this is achieved, Mediterranean Sea countries need to continue to refer to currently fragmented legislations which can mitigate, albeit indirectly, OA. Galdies et al. (2020) showed how within the Mediterranean Sea basin, the OA problem is at best uncoordinated even at the European Union level, which spans much of the northern waters of the Mediterranean Sea. Although OA is acknowledged at the higher levels of governance, its status as an environmental challenge is greatly diluted at the European Union Member State level. All major EU measures in these areas take the form of directives rather than directly applicable regulations, and they are transposed into binding measures at the national level. However, a quick review of reports submitted by EU states as part of the MSFD to achieve Good Environmental Status (GES) in European waters (Dupont et al., 2014) shows that consistent EU-wide monitoring of the degree of OA and its impacts in European waters is still not in place. For example, a quick look at the monitoring activities conducted by EU member states for the descriptor D7 (Hydrographical changes) as part of the EU’s Marine Strategy Framework Directive to achieve Good Environmental Status (GES) shows that only Italy has marine acidification as part of its monitoring programme within its territorial waters (MSFD, 2014). Other EU Mediterranean countries are either not participating in such monitoring exercises or opting out from measuring marine acidity in their monitoring programmes. This reflects how even at the EU level, with its available funding and advanced environmental legislation, European OA-related policy and legislation must urgently close this gap by improving their strategies to support OA mitigation, primarily through continuous and systematic monitoring. It is interesting to note that the measurement of pH constitutes perhaps the easiest metric to use for the determination of the success of mitigation measures (Bernie et al., 2010). This metric should be adopted by UNEP/MAP and introduced as part of marine pollution monitoring that should be conducted by all Mediterranean countries in order to ensure a Mediterranean wide action to assess and monitor OA. Indeed, OA is recognized among the list of pressures threatening the coralligenous reefs in the Mediterranean Action Plan on climate change impact on marine biodiversity in the marine protected areas. However, direct measures of OA are not included among the list of monitoring indicators. Two of the four Regional Seas Programs in Europe have set specific assessing and monitoring strategies for OA, such as setting the OSPAR acidification study working group (ICES, 2014), and defining the acidification indicators, as in the Baltic program in 2018.

Many studies show how the impacts of OA can have serious social and environmental consequences that are in turn rooted in a highly fragmented and complex institutional setting (Jagers et al., 2019). This complex nature is even more pronounced in the Mediterranean because a good number of national institutions are faced by poor political, administrative and technical constraints, resulting in lengthy socio-economic and environmental progress. Within all of this, OA is indeed an important thematic, the management of which can become exacerbated within such a multi-faceted governance structure characterized by painfully slow progress. It must be made clear that this sluggishness of the political and administrative structures in the Mediterranean cannot be solely attributed to types of politics such as in those where democracy is still a contested process. Indeed, Jagers et al. (2019) explains how even democratic institutions can fail to provide public goods such as the case for Good Environmental Status of European and Mediterranean waters, as discussed in the previous section. Gorjanc et al. (2020), for example, showed how the ambitions of the Programmes of Measures within the MSFD framework are deemed to be low, characterized by non-existent coordination between EU-member states at sub-regional and regional levels. One can argue that the EU leaves it up to its individual States to choose their own instruments, methods and political preferences in order to reach the final common goal. At the same time, one should be aware of the fact that EU-Member states bordering the Mediterranean Sea are amongst the most advanced and rich Mediterranean nations with adequate finances to support such a public good (e.g. GES).

Taking into consideration the current gap in OA legislations and policies in the Mediterranean Sea, the organizations that can best address this issue and make it more mainstream at the Mediterranean political level were classified by the survey participants (n=47; Table S6 ), from the most relevant as follows: the United Nations Environment Programme (UNEP), Mediterranean Action Plan-Barcelona Convention and IOC-UNESCO (Intergovernmental Oceanographic Commission-UNESCO), followed by CIESM (The Mediterranean Science Commission), Governments, national, regional and local entities, then the IAEA (The International Atomic Energy Agency), UfM (Union for the Mediterranean), Plan Bleu, and non-governmental organizations (e.g., WWF, Greenpeace, etc.). The level of OA-related discussion by the most relevant parties, favorably perceived to address OA by the Mediterranean community, is mostly acceptable (60%), while 17.5% refer to it as an avenue for scientific collaboration and exchange of management practices, and 15% see that they are not interested so much in OA issue, whereas only 7.5% of participants mention that this discussion is highly mature and ambitious. These various perceptions reflect the different influence of the listed parties in the Mediterranean countries, and the need to enforce the bridges of communication to better address OA within these local, regional, and international organizations.

At a local level, more than 60% of participants stated that OA is irrelevant or neutrally perceived by their country’s environmental authorities/government, while only 36% mention that these local authorities are perceiving OA as an important environmental issue. This supports their negative replies (>52% said no) when asked if they are aware of any project or initiative in their countries to mitigate OA. And when such projects are taking place, they are predominantly targeting mitigation of anthropogenic CO₂ emissions (67%), and equally (9.5% each) for mitigation of other greenhouse gas emissions, adaptation of ecosystems and human activities that build resilience to OA, restoration activity, lesser (5%) for mitigation of other local/regional factors that contribute to (or compound the effect of) OA (such as pollution, eutrophication, etc.), and none for establishing OA refugia. Besides, for non-EU member states, participants are not aware of any national legislation that is relevant to address OA.

Moreover, survey respondents identified the following activities that are taking place in their respective countries: vulnerability assessments of the resource(s) as a function of sensitivity, exposure and adaptive capacity (39%), clear management goals, including selection of the organism(s) and biological processes of interest for conservation (31%), potential refugia are being identified and/or determined (25%), whereas the majority (42%) confirmed that none of the above activities are taking place in their respective countries.

Upon closer inspection of the research and knowledge equity in the region, the majority of the survey respondents from non-EU Mediterranean countries sense a big difference when it comes to the number of OA-related studies, which they consider to be a relevant matter. Equally relevant (and concerning) is their perception of the current state of funding to manage OA, of the current lack of expertise in the South and to the poor research infrastructure. If true, such a situation is able to manifest itself in lack of equity when it comes to the level of participation and cooperation from the North; such a scenario was in fact rated by the respondents as very relevant.

There is an urgent need for higher resolution, integrated (physical-biogeochemical-ecological) and spatially-nested models which would provide new perspectives on where OA impacts could be more pronounced. Models should be able to address the cumulative effect of OA and other stressors, also considering synergistic effects and non-linear interactions. This is crucial to better understand the multiple effects of changes in stressors and to explore adaptation/mitigation measures that can be applied at different spatial and temporal scales (Duarte et al., 2013; Zunino et al., 2021; Cappelletto et al., 2021). Furthermore, it would be important to at least try to incorporate organism plasticity and long-term adaptation in modeling ecosystem response, possibly also considering different life time-span of different organisms and therefore turnover rates (Solidoro et al., 2010).

Focus should be given to forecast direct consequences of OA on habitat structuring species and other ecologically relevant species, as well as on cascading consequences on the ecosystem network, its functioning and related services. Moreover, a quantitative assessment of uncertainties in model projection would be needed, in order to properly inform research priorities and management choices.

Models should be improved to provide answers to fundamental questions, such as: what are the impact of OA on marine functional biodiversity and possible consequences of OA-induced biodiversity loss on ecosystem functioning and resilience? and more applied questions, including: What are the local time-response of mitigation measures applied to global emissions? Can a model assess the effects of alternative mitigation solutions?

Biological aspects of OA research should address:

(1) multiple environmental stressors. It is more realistic to study OA with different combinations of abiotic and biotic parameters (e.g., temperature, light, salinity, nutrients, grazing, respiration, hypoxia, etc.) in both in situ and experimental studies (Riebesell and Gattuso, 2015). Organismal response to OA is not homogeneous nor easily predictable and appears to depend on biological coping mechanisms to help offset high ‘costs’ of physiological processes (i.e., calcification) under reduced pH conditions. It is thus important to assess multiple stressors under future scenarios, such as natural diurnal pH fluctuations, food availability, including additional potential drivers [e.g., deoxygenation, warming, salinification, etc.] (Boyd et al., 2018). Also, as multiple-drivers studies are relatively scarce, especially on large scales or using larger organisms, cooperation between Mediterranean countries in sharing and enabling larger/more developed mesocosm facilities would be highly beneficial and cost effective.

(2) biotic response at different organizational levels from genes to food webs and ecosystems to assess its potential adaptation strategies in specific ecosystems, and the genes responsible for such plasticity (Riebesell and Gattuso, 2015). This work should take into account the different life stages of species through ecophysiological studies that might yield further insight into detailed physiological processes that are impacted by OA (not only calcification and respiration processes).

(3) key and most vulnerable groups/ecosystems. For example, calcifiers that live in conditions where there is no temporary refuge for calcification (do not live within canopies), or do not have any possibility of compensation for the high costs of physiological processes (i.e., calcification) like in conditions with limited food availability, would be most at risk from OA. The limitations to biological compensation of the negative effects of OA would be a necessary future direction for research, and could elucidate when organisms would be most vulnerable. An interesting case is the group of coralline algae, able to use CO₂ as a substrate but also potentially affected negatively through calcification processes. Also, future OA studies should pay more attention to habitat-forming species, various larval-stages of commercial species, and potential OA effects on harmful/toxic and invasive species as well.

(4) not only short-time scale physiological responses but also generational population responses over years and even longer time spans. OA requires a high resolution and long-term monitoring strategy to detect the variability at short-term and -spatial scales and to enable disentangling long-term (multidecadal) anthropogenic signals from short- and medium-term spatial and temporal variability (Doney et al., 2020). There is a clear need to implement long-term monitoring of biological ecosystems to track any potential responses based on near-real-time observations. This is feasible through merging the chemical and biological monitoring in key areas (e.g., highly urbanized coastal areas, MPAs, seagrass meadows, etc.), while focusing on specific biological indicators (i.e., coccolithophore, pteropods, coralligenous, etc.) to simplify the long-term monitoring. Long-term OA observations are crucial to understand the plasticity, micro-evolution and adaptation of key species.

Robust socio-economic studies should be developed to better assess impacts of the prolonged OA impacts in the Mediterranean Sea. In order to do so, there is a need to invest in research that can address the interactions among drivers and processes at different spatial and temporal scales. Such research should focus on species/habitats/ecosystems of commercial and cultural values and possibly be supported by modeling and assessment performed at the relevant (high) spatial and temporal scales. Quantification of uncertainties across broad scales would be very important (Solidoro, 2021).

Moreover, there is a necessity to assess the loss of ecosystem services due to OA, and to value the costs/benefits of adaptation/mitigation measures using projected future scenarios based on global SSP, and IPCC AR6, and local downscaling scenarios. Such studies will help to bridge the knowledge gaps between policymakers and the OA community by reflecting on the importance of supporting OA research and implementing OA mitigation strategies to sustain the resources and avoid projected losses.

Mediterranean Sea countries are particularly vulnerable to OA, specially those in the Middle East and North Africa (MENA) (Hilmi et al., 2017), thus regional budget and governmental resources should be directed to capacity building of these countries’ institutions to ensure quality and sustainability of the obtained data. Moreover, OA research, particularly on regional key economic species, in the Southern Mediterranean countries should be encouraged and systematically supported in order to conserve their economic resources and save threatened habitats, and therefore the livelihoods of coastal vulnerable communities.

Furthermore, it is important to encourage research to fill gaps in understanding chemical, biological, ecological and socio-economic impacts of OA at regional and interregional scales (see Tilbrook et al., 2019), especially in the MENA region. This can be widely recognized and adopted in national monitoring programs of the Mediterranean countries in support of legislations aimed at mitigating OA (similar to Barcelona convention).

According to the survey’s respondents, to reduce the discrepancy related to OA-related research, knowledge and governance within the Mediterranean the following additional measures should be implemented (from the most relevant to the least): 1) Promote more scientific exchanges (training, capacity building efforts, etc.), 2) Involve researchers from the South in Mediterranean oceanographic cruises and experimental studies, 3) Increase the regional and North-South projects, 4) Support more PhD students from South to North Mediterranean countries, 5) Fix a quota (minimum number of partners) from the Southern countries for the project to be eligible. More North-South and East-West projects and programmes can be supported through including the diversity of involved countries and researchers as a basic criteria in projects’ selection procedures to encourage researchers to cooperate with peers from under-represented areas and to fortify scientific exchanges, inter-comparison exercises and experimental collaborations. This should not only be conducted through short-term capacity-building efforts, but also via common trans-Mediterranean oceanographic cruises, and by supporting associated graduate and postgraduate projects.

In parallel with the need to further fine-tune research related to the understanding of OA trends and consequences in the Mediterranean Sea (see the above recommendations), the available governance and legal opportunities remain the most relevant tools to address OA. It is time to re-enforce coordination and harmony of global, regional and national governance bodies so that they are tailored for effective governance of OA in the Mediterranean Sea (see Mossler et al., 2017 for more information about the effects of OA and other carbon emissions frames). Such structures would include existing EU and regional multilateral environmental laws, conventions and protocols that are pertinent to the protection of the Mediterranean Sea. For example, Mediterranean countries can use Marine Spatial Planning (MSP) techniques to improve their location of mariculture activities on the basis of OA impacts and survival thresholds of economically important species (Meaden et al., 2016).

Tools to monitor, model, analyze and make informed decisions should be developed together with innovative ocean literacy approaches, such as direct discussions with marine scientists (Fauville, 2017) and Virtual Reality (VR) techniques (Fauville, 2016; Fauville et al., 2021), to disseminate concomitantly key knowledge that might help in creating a public momentum about OA. Ocean literacy is needed to improve the awareness among practitioners, institutions and the Mediterranean wide and diverse public. This is feasible through re-elaborating the results of key trans-Mediterranean OA-related projects (such as EPOCA and MedSeA), and some key findings published in peer-reviewed articles and OA-related networks (OA Med-Hub and GOA-ON) using public-friendly terminologies. Although such literacy-awareness efforts already took place in some countries, the public is still only vaguely aware of OA and its threats in the Mediterranean Sea. OA literacy should not be exclusively restricted to posters, flyers, and a few videos on social media platforms, but also dovetailed with national ministries (ministries of education/research/environment), municipalities and NGOs to publicize this knowledge efficiently at a national-level, in educational curriculums and via national platforms (TVs, journals, radio stations, etc.). Only through this suggested approach, can OA become an environmental public cause that may trigger national representatives into adopting actions and long-term adaptation/mitigation strategies.