The ocean plays a significant role as a sink for anthropogenic carbon dioxide (CO₂) (Sarmiento and Gruber, 2002; Le Quéré et al., 2015), absorbing approximately a quarter of the CO₂ emitted by humans to the atmosphere (Friedlingstein et al., 2022; Gruber et al., 2023). This marine contribution to CO₂ withdrawal mitigates climate change, albeit at a cost to the ecosystem, causing significant alterations in seawater chemistry (Orr et al., 2005; Doney et al., 2009; Gledhill et al., 2015).

When the CO₂ molecule enters the ocean, it hydrates and forms carbonic acid, which is a weak acid that tends to dissociate water producing hydrogen ions. As a result, the concentration of hydrogen ions ([H⁺]) in the seawater increases steadily and slowly, leading to ocean acidification (OA) over a long period of time, i.e. a decline in the pH (Raven et al., 2005; Impacts, 2014; Williamson and Widdicombe, 2017; Doney et al., 2020). In addition, part of these [H⁺] bind to the carbonate ions (CO₃ ^2-) to generate bicarbonate (HCO₃ ^-), causing a decrease in the concentration of CO₃ ^2- and, consequently, their availability in the medium. Carbonate ions are essential for marine calcifying organisms to form their skeletons and shells, and their decline severely affects the biogenic calcification process (Riebesell et al., 2000; Orr et al., 2005; Feely et al., 2012; Kroeker et al., 2013; Pörtner et al., 2014; Mostofa et al., 2016; Doney et al., 2020), as marine organisms must combine Ca²⁺ and CO₃ ^2- to precipitate CaCO₃ for calcareous shells formation (Silverstein, 2022) (Equation 1):

Calcification ability can be assessed through the estimation of the saturation state (Ω) of calcium carbonate minerals, which is defined as shown in Equation 2:

where [Ca²⁺] and [CO₃ ^2-] are the concentrations of calcium and carbonate ions, respectively, and K'_sp is the apparent solubility product for aragonite or calcite, the polymorphic mineral forms of CaCO₃ in the ocean. This dimensionless parameter largely depends on temperature (T), salinity (S), and pressure (P) in the medium. Values of Ω for calcite (Ω_Calcite) and aragonite (Ω_Aragonite) higher than 1 indicate supersaturation, i.e. favourable conditions for calcification to proceed whereas, values of Ω lower than 1 denote undersaturation and thus, corrosive conditions for calcifiers (Doney et al., 2009).

Nevertheless, biogenic calcification is also affected by other factors, such as the ability of certain organisms to regulate their internal chemistry or the presence of protective organic covers (Miller et al., 2009; Pan et al., 2015; Melzner et al., 2020). Several studies have reported reductions in calcification rates of certain organisms even when Ω exceeds 1; for instance, in some corals when Ω is comprised within the range 2-3.5 (Guinotte et al., 2003; Yamamoto et al., 2012; Eyre et al., 2018). In fact, a value equal to 1.5 for Ω in both minerals, has been considered to define negative consequences for marine ecosystems even at planetary scale (Gruber et al., 2012; Broadgate et al., 2013; Ekstrom et al., 2015; Zhai, 2018) and it has been adopted as a critical threshold for calcification to occur properly.

As OA is strengthening and it will keep increasing over the 21^st century at rates dependent on future CO₂ emissions, biogenic calcification will continue being at risk. However, the magnitude of OA is not homogeneous worldwide and certain areas are more affected than others (Gattuso et al., 2015). In particular, some regions in the North Atlantic and the Mediterranean Sea (MedSea) have been found to be highly sensitive to OA (Fontela et al., 2020) and noticeable gradual decreases in seawater pH have been measured over time in Atlantic and Mediterranean water masses surrounding the southern Iberian Peninsula (Flecha et al., 2015; Flecha et al., 2019). Therefore, it can be anticipated that the saturation state of calcium carbonate minerals in this area is being affected by the progressive OA process already documented. This assumption can be tested by assessing the temporal trend of Ω in waters of the Strait of Gibraltar (SoG), the only connection between the Atlantic Ocean and the MedSea. The SoG holds a two-way exchange, with a surface inflow of Atlantic water (AI) that moves eastwards, and a compensating westward deep outflow of Mediterranean water (MOW) underneath. The AI is primarily formed by the North Atlantic Central Water (NACW) whereas the colder and saltier MOW consists of, mainly, two water masses, the Levantine Intermediate Water (LIW) and the Western Mediterranean Deep Water (WMDW). The former is the result of the transformation process undertaken by the Atlantic surface water as it moves eastward, which makes it saltier, until it sinks to intermediate depths in the Levantine basin, the easternmost basin of the MedSea (Lascaratos, 1993). In contrast, the WMDW is formed in the western Mediterranean basin, as a result of the winter cooling of highly modified surface waters in the Gulf of Lions (MEDOC Group, 1970; Fourrier et al., 2022). Both water masses flow at depth towards the SoG and leave the MedSea, merging in the MOW in the Gulf of Cádiz. Recent estimates indicate that the median transit time of the WMDW from the Gulf of Lions to the SoG is 5 years whereas the LIW takes roughly 8 years from the Strait of Sicily to the SoG (Vecchioni et al., 2023). Furthermore, the transit time of the LIW from its formation basin to the Sicily channel ranges between 8 and 13 years, depending on the approach (Roether et al., 1998; Gačić et al., 2013). Similarly, Gačić et al. (2013) based on salinity anomalies estimated that the total time interval needed for the signal propagating from the Levantine basin to reach the deep mixed layers of the Algero-Provençal sub-basin is about 25 years. Thus, the LIW identified in the SoG has crossed a larger distance and, consequently, is older and biogeochemically more stable than the WMDW also present in the area (Flecha et al., 2019 and referenes therein). Therefore, the three water masses (NACW, LIW and WMDW) found in the SoG are characterized by well-differentiated thermohaline and biogeochemical properties and history, making the channel a very suitable spot for the assessment of Ω in them.

In this study, we used periodic high-quality measurements of the carbonate system parameters collected in the SoG during 17 years of measurements (2005-2021) to accurately determine the temporal evolution of Ω in Atlantic and Mediterranean water masses. Furthermore, as the projected changes in availability of CO₃ ^2- in the ocean due to OA would be essentially irreversible on centennial time scales (Mathesius et al., 2015; Heinze et al., 2021), long-term trends of Ω were also assessed under the IPCC AR6 scenario of fossil-fuel-rich development (SSP5-8.5) (IPCC et al., 2021). The calculated projections are finally used to infer approximately when communities of calcifying organisms would be exposed to waters undersaturated for calcium carbonate minerals in this key oceanographic region.

Results shown here show clear evidence that the saturation state of calcium carbonate minerals in the water column of the SoG is decreasing due to the decline in carbonate ions in response to the gradual diminution of seawater pH in the area. Our findings are coherent with a previous study reporting that water masses in the SoG were markedly affected by the uptake and storage of anthropogenic CO₂ during the 2005-2015 decade, leading to an evident OA in them (Flecha et al., 2019). In this work, when the time series of measurements spanned six additional years (up to 2021), the temporal decline in pH_T25 in Mediterranean and Atlantic water masses is maintained at nearly the same annual rates as those provided by Flecha et al. (2019). Furthermore, our analysis also indicates that even though the LIW is experiencing a certain increase in its CO₂ content over time, it is less affected by the process of OA in relation to its Mediterranean counterpart and the NACW. As suggested by several authors, its age and intermediate position in the water column prevent a high penetration of atmospheric CO₂, which makes it more biogeochemically stable (Hassoun et al., 2015 and references therein). Nevertheless, it still contains a large natural component of CO₂ due to remineralization and/or mixing with water masses having higher anthropogenic carbon content during the long transit from its origin basin towards the Strait (Flecha et al., 2019; Vecchioni et al., 2023), which impacts its pH. OA rates have been measured in a number of marine eco-regions from coastal to open ocean regions, resulting in a broad range of values. Carstensen and Duarte (2019) provided an average acidification rate for the coastal ocean equivalent to ±0.023 pH units yr^-1 versus a range of -0.0004 pH units yr^-1 and -0.0026 pH units yr^-1 in open ocean areas. If the bulk of pH data available in the water column at the SoG are considered, the average value for pH_T25 decline in the region equals to -0.002 pH units yr^-1, falling within the range found for open ocean sites (Carstensen and Duarte, 2019). Certain areas are characterized by a strong pH decline tendency, such as the coastal ocean in Korea, the continental shelf of the Gulf of Cádiz, and surface waters of the Mediterranean Sea. In contrast, lower acidification rates have been reported in the North Pacific Ocean, the Munida and the Iceland Sea time series, and the upper water column (∼75 m) of the Western Mediterranean. In the Atlantic time series ESTOC, BATS, and CARIACO, the rates are of the same order of magnitude as those observed in several sections of the MedSea ( Table 6 ). Acidification rates calculated here in the water masses exchanging at the SoG (-0.0030, -0.0019 and -0.0012 pH units yr^-1 for the NACW, the WMDW and the LIW, respectively) are in agreement with those provided in other Atlantic and Mediterranean regions. However, pH decline in the NACW seems to be slighter higher than others, which is the outcome of the conjunction between the anthropogenic CO₂ concentration contained by this water mass (Flecha et al., 2019) and the accumulation of CO₂ resulting from organic matter degradation in the productive Gulf of Cádiz (Navarro et al., 2006; Flecha et al., 2019; Álvarez-Salgado et al., 2020).

As a result of the different OA rates observed, the current saturation state of CaCO₃ minerals in water masses at the SoG markedly differs. It is well known that Ω varies regionally and distinct values of this parameter have been provided in the Atlantic Ocean (Fernández-Guallart, 2015; Jiang et al., 2015), the Gulf of Cádiz (Jiménez-López et al., 2021), the MedSea (Hassoun et al., 2015), the southeastern Yellow Sea (Choi et al., 2020), the northern Gulf of Alaska (Evans et al., 2013), and others (Feely et al., 2009). Our Ω estimates are of the same order of magnitude as those shown in these studies, and similar to previous calculations provided for several Mediterranean water masses (Hassoun et al., 2015; Hassoun et al., 2019). Data collected in an earlier survey (2013) carried out in the SoG, provided values of Ω_Aragonite and Ω_Calcite of ~2.6-3 and ~4.2-4.6, respectively (Hassoun et al., 2015), in agreement with our results. Similarly, Ω_Aragonite of approximately 2.5 has been given for the NACW_T (North Atlantic Central Water of subtropical origin) at the Gulf of Cádiz using data taken from 2014 to 2016 (Jiménez-López et al., 2021), which is not far from our calculations based on 17 continuous years of measurements in the SoG ( Figure 2 ). This study also reported Ω_Aragonite values of 1.56 for the NACW_P (North Atlantic Central Water of subpolar origin) and 2.22 for the MOW, which is much lower than omega shown here for the Mediterranean outflow at the Strait. Moreover, in the Levantine Basin, average values of Ω_Aragonite and Ω_Calcite equivalent to 3.4 and 5.3 have been obtained elsewhere, corresponding to a recently formed LIW (Hassoun et al., 2015).

Our assessment also shows a gradual temporal reduction of Ω in the three water masses over the time series. Overall, Ω_Aragonite decreased at annual rates of -0.0107 yr^-1 in the whole water column of the Strait, -0.0091 yr^-1 in the MOW and at specific rates of -0.0139 yr^-1, -0.0102 yr^-1 and -0.0080 yr^-1, in the NACW, WMDW and LIW, respectively ( Table 4S ). These rates are much lower than those reported in the continental shelf of the Gulf of Cádiz (-0.0552 yr^-1, Jiménez-López et al., 2021) using disperse measurements from 2006 to 2016 and in 2013 in the Levantine basin (-0.07 yr^-1, Hassoun et al., 2019). Our results are, however, in line with several studies conducted across the Atlantic that provide values comprised between -0.010 yr^-1 and -0.0140 yr^-1 (González-Dávila et al., 2010; Bates et al., 2012; Takahashi et al., 2014; Fontela et al., 2021). In the case of calcite, the rates of decline (-0.0168 yr^-1, -0.0145 yr^-1, -0.0215 yr^-1, -0.0161 yr^-1 and -0.0129 yr^-1 in the water column, the MOW, NACW, WMDW and LIW, respectively) are also similar to those computed in Atlantic time series (values between -0.010 yr^-1 and -0.0209 yr^-1, Bates and Peters, 2007; González-Dávila et al., 2010; Takahashi et al., 2014; Bates and Johnson, 2020) but lower than that in the Levantine basin (-0.1 yr^-1, Hassoun et al., 2019).

Spatial variations in the ranges of calcium carbonate mineral decline can be attributed to the fact that every region and each water mass has specific characteristics that directly influence the rate of change of Ω. For instance, coastal areas and marginal seas generally have a higher tendency to acidify as a result of stronger anthropogenic impacts (Carstensen and Duarte, 2019). Additionally, a significant increase in temperature and biological factors are also associated (Jiménez-López et al., 2021). In contrast, in most open ocean areas, the influence of the biological uptake of CO₂ decreases and the change in saturation of calcium carbonate minerals is mostly caused by the uptake of atmospheric CO₂ (García-Ibáñez et al., 2021).

Considering the rates of Ω_Aragonite decline in the SoG and taking the IPCC AR6 SSP5-8.5 scenario for the CO₂ emissions progression, a high vulnerability of calcifiers to ocean acidification is expected in the area before the end of the 21^st century. In particular, critical conditions for biogenic calcification will be reached in less than 60 years in the upper and bottom layers of the SoG (51 and 56 years for the NACW and WMDW respectively). Furthermore, even though the LIW seems to be less strongly perturbated than the other two water masses, an unfavourable scenario for calcification would also occur in approximately 75 years from now. Due to the lower solubility of calcite in seawater (Sulpis et al., 2022), unfavourable conditions with respect to this mineral will arise in the region beyond 2100. Critical conditions have been defined here by a threshold of Ω=1.5 for both aragonite and calcite, as this value has been adopted in several studies to assess the vulnerability of marine calcifiers to OA and its repercussion for marine ecosystems (Gruber et al., 2012; Broadgate et al., 2013; Ekstrom et al., 2015; Zhai, 2018). Moreover, evident reductions in calcification rates of marine organisms have been reported at Ω values of 2-3.5 (Guinotte et al., 2003; Yamamoto et al., 2012; Eyre et al., 2018). Therefore, the range of current estimations of Ω_Aragonite in the SoG along with its projected rates of decline in all the water masses under a high CO₂ emission pathway depict a catastrophic scenario for the marine ecosystem in the region. Calcifiers are essential to maintain the ecological status of the pelagic and benthic ecosystems (and their living resources) in the sub-basins connected by the Strait, the Gulf of Cádiz and the Mediterranean Alboran Sea (Vertino et al., 2010; Movilla, 2015; Lozano et al., 2020), where a significant number of planktonic calcifiers (coccolithophores, foraminifera), pelagic mollusks (pteropods), bivalves, calcareous algae and corals thrive (Reguera et al., 2009; Bautista-Chamizo et al., 2016; Jiménez-López et al., 2021). According to our findings, regional calcifying communities are at high risk due to the gradual OA that is proceeding in the area, which will restrict the levels of carbonate ions required for a proper calcification well before the end of the 21^st century.

It has been predicted that by 2100, 91% of the whole ocean will be undersaturated to aragonite (Gattuso et al., 2015) including the Southern Ocean and parts of the North Pacific (Feely et al., 2009), the Irminger and Iceland Basin (García-Ibáñez et al., 2021), the Nordic Seas (Fransner et al., 2022), the polar Surface water (Matear and Lenton, 2014) and the lower NACW in the Western Atlantic (Fernández-Guallart, 2015). This scenario would co-occur with critical conditions in certain eco-regions, such as the tropical oceans, where survival of coral reefs will be compromised by the availability of carbonate to form aragonite (Matear and Lenton, 2014). Although undersaturation of calcite in the surface ocean is expected beyond the end of the 21^st century (Feely et al., 2009), levels of this mineral will be greatly reduced with respect to current concentrations. Therefore, the trends found in a choke oceanographic spot, such as the SoG, after almost two decades of observations, support the concern about the impact of the pH decline on marine ecosystems. Maintenance of ocean observing systems is then essential to establish baselines required to document and predict the vulnerability of marine ecosystems to climate change and provide tools for efficiently developing and implementing adaptation and mitigation strategies.

The original contributions presented in the study are publicly available. This data can be found here: https://doi.org/10.20350/digitalCSIC/14556.

SA-V: Data curation, Formal analysis, Investigation, Methodology, Software, Writing – original draft; SF: Data curation, Formal analysis, Methodology, Software, Supervision, Validation, Writing - review & editing; FP: Conceptualization, Methodology, Formal analysis, Writing - review & editing; GN: Methodology, Funding acquisition; JG-L: Funding acquisition, Data curation, Methodology, Writing - review & editing; AM: Data curation, Resources; IEH: Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Resources, Supervision, Validation, Writing – review & editing.