The presence of Mg-rich calcite in foraminifera shells has impeded traditional calcite-based palaeothermometry in the eastern Mediterranean, complicating our understanding of the region's past climate. We analysed the Ca isotopic composition (⁠δ44/40Ca) in the shells of the planktonic foraminifera species Globigerinoides ruber (white), collected from 12 core-top sediment samples spanning the Aegean and Levantine seas in the eastern Mediterranean. The shells exhibited varying degrees of early diagenetic Mg-rich authigenic calcite coatings, which complicate traditional calcite-based palaeothermometry in the region. Our findings demonstrate a significant correlation between the δ44/40Ca data and the Mg/Ca ratios of the shells and suggest that the inorganic calcite overgrowth originates from seawater. Moreover, the Ca isotopic composition increases with the amount of calcite overgrowth. This relationship allowed us to quantify the maximum potential inorganic end-member. The highest amount of post-depositional calcite precipitation was observed in the warm, shallow basins of the South Aegean Sea. The diagenetic alteration of the primary biogenic δ44/40Ca geochemical signal is linked to the quantity of secondary calcite present. By combining Ca isotope and Mg/Ca measurements on isolated foraminifera tests, we provide novel insights into the early diagenetic history of carbonate-bound geochemical proxies.

Supplementary material: Mass balance calculations and Sr/Ca data are available at https://doi.org/10.6084/m9.figshare.c.7552500

Marine carbonate sediments are valuable archives of Earth history because their geochemical composition allows us to study the evolution of the Earth's climate (Morse and Mackenzie 1990; Reuning et al. 2005; Hönisch and Hall 2007; Pearson et al. 2007). However, carbonates are reactive minerals and one of the main limitations of their archive is the susceptibility of their chemistry to both post-depositional and burial diagenesis (Ahm et al. 2018; Subhas et al. 2018; Chanda et al. 2019).

Post-formational processes can alter the geochemistry of carbonate archives, causing compositional changes both during settling and between the time of deposition and the time of sampling (Chanda et al. 2019; Davis and Benitez-Nelson 2020). Inorganic calcite precipitation in marine sediments can occur due to changes in the pore water chemistry driven by various biogeochemical processes. For instance, during organic matter degradation, microbial processes such as sulfate reduction and anaerobic methane oxidation increase alkalinity, leading to conditions favourable for carbonate precipitation (James et al. 2021; Turchyn et al. 2021). As a result of their high alkalinity (Copin-Montégut 1993; Schneider et al. 2007), the waters of the eastern Mediterranean are well above the saturation state of calcite (Álvarez et al. 2014; Krasakopoulou et al. 2017) and may promote post-depositional carbonate mineral precipitation (Deyhle et al. 2003) or alter individual components of the sedimentary material, such as foraminiferal tests (Boussetta et al. 2011; Kontakiotis et al. 2011; van Raden et al. 2011) and nannofossils (Crudeli et al. 2004).

Studies of foraminifera shells with different taphonomic histories have shown that burial diagenesis can significantly impact the fidelity of the archive (Kozdon et al. 2013; Bernard et al. 2017). Inorganic carbonate minerals precipitating from seawater as coatings on foraminifera samples are known to increase the Mg/Ca ratios of the coated samples relative to pristine foraminifera (Wollast et al. 1980; Morse and Mackenzie 1990; Gehlen et al. 2004). This diagenetic alteration of the sedimentary material hinders palaeotemperature proxy calibration (e.g. Kontakiotis et al. 2011) and palaeoceanographic studies (e.g. Edgar et al. 2015; Stainbank et al. 2020) in the eastern Mediterranean region, which require pristine calcareous biogenic fossils for environmental reconstructions.

Compared with the large body of knowledge on the ecological characteristics (Rohling et al. 1993; Pujol and Grazzini 1995; Giamali et al. 2021) and spatial distribution of planktonic foraminiferal species in the Mediterranean (Thunell 1978; Mallo et al. 2017; Zarkogiannis et al. 2020a), their response to diagenetic authigenesis under modern hydro-climatic conditions remains poorly studied. The diagenetic bias on foraminifera shell Mg/Ca thermometry in the eastern Mediterranean has been documented from both modern (Boussetta et al. 2011; Kısakürek et al. 2011; Sabbatini et al. 2011; van Raden et al. 2011) and past (Antonarakou et al. 2019) sedimentary archives.

Particularly for the Mediterranean Sea, several studies have highlighted the impact of secondary diagenetic depositions (overgrowths) on shell Mg/Ca ratios because the bulk foraminiferal geochemical composition combines the biogenic and authigenic calcite components of planktonic foraminifera tests (Boussetta et al. 2011; Kontakiotis et al. 2011; Sabbatini et al. 2011; Antonarakou et al. 2019). The investigation of diagenetic features, in terms of their spatial (both latitudinal and longitudinal) and temporal (up to late Miocene) patterns, indicates a difference in the magnitude of precipitated authigenic high-Mg inorganic calcite crystals as well as the nucleation sites on the interior and exterior sides of the tests (Sabbatini et al. 2011; Kontakiotis et al. 2017). Moreover, it has been shown that, due to their different test morphologies, different species (van Raden et al. 2011) and even morphotypes (Kontakiotis et al. 2017) are influenced by diagenesis to different degrees.

Foraminiferal Ca isotopes are potentially more robust to diagenetic alteration than the Mg/Ca ratios because Ca is the major cation in calcite (Henderson 2002) and is not thought to fractionate during dissolution (Fantle and DePaolo 2007). Ca isotopic fractionation is also insensitive to the parameters that influence other proxies, such as the global ice volume, evaporation and freshwater input (Skulan et al. 1997; Griffith et al. 2008). Ca isotopes have the potential to fingerprint carbonate authigenesis (Bradbury and Turchyn 2018; Fantle and Ridgwell 2020).

To date, the majority of research examining the impact of diagenesis on Ca isotopes has concentrated on alterations in bulk carbonates (Fantle and DePaolo 2007; Fantle 2015; Higgins et al. 2018; Fantle et al. 2020). Quantifying the extent of post-depositional alteration of the heterogeneous bulk sediment allows inferences to be made regarding the extent to which the individual components of bulk carbonates can be altered. However, it is also important to understand the reactivity of these individual components and the impact of secondary diagenetic depositions (overgrowths) over any timescale at the foraminiferal test level to ensure the robust interpretation of proxy archives (Chanda et al. 2019).

We explore here the idea that Ca isotopes can be used as an indicator of carbonate authigenesis (i.e. the amount of post-depositional precipitation of CaCO3) on foraminifera tests. This work demonstrates the importance of early diagenesis in shaping the chemistry of pelagic carbonate sediments. It highlights the utility of using paired measurements of Mg/Ca ratios and Ca isotopes to identify, characterize and quantify the diagenetic effects on foraminifera tests that may also help to deconvolve pristine signals for the study of deep time records in the Mediterranean.

Core-top and planktonic species selection

We targeted the subtropical spinose planktonic species Globigerinoides ruber (white) for several reasons. It is not only the most dominant species in the eastern Mediterranean, providing crucial information on sea surface properties (Pujol and Grazzini 1995; Zarkogiannis et al. 2020a), but there is also a wealth of data available for this species from both Ca isotope and Mg/Ca palaeothermometry studies (Dekens et al. 2002; Kısakürek et al. 2008, 2011). In addition, G. ruber has been well studied for its susceptibility to secondary diagenetic calcite overprints in this region (Boussetta et al. 2011; Kontakiotis et al. 2011, 2017; Sabbatini et al. 2011).

Based on the size variability and biogeographical distribution of G. ruber morphotypes in response to the environmental parameters of the study area (Zarkogiannis et al. 2020a), we analysed G. ruber specimens from the restricted size fraction of 250–350 μm. This approach was chosen to minimize any possible ontogenetic effects (Elderfield et al. 2002) and therefore to increase the robustness of the geochemical analyses. We incorporated Ca isotope measurements on carefully selected core-top samples to explore the potential of Ca isotopes in constraining diagenesis. These Ca isotope results are interpreted alongside Mg/Ca data from previous studies (Kontakiotis et al. 2011; Sabbatini et al. 2011).

A total of 12 modern sediment core-top samples were selected from the eastern Mediterranean basins (Fig. 1) to explore the effects of both the initial chemical composition and early diagenesis on the δ44/40Ca values and Mg/Ca ratios of foraminifera carbonate sediments in a geological context. The impact of diagenesis (e.g. on Mg/Ca thermometry) is significant in both the Aegean and Levantine basins (Boussetta et al. 2011; Kontakiotis et al. 2011; Sabbatini et al. 2011), especially in the South Aegean Sea (Kontakiotis et al. 2017). Our sample set covers all the representative sub-basins of the eastern Mediterranean (e.g. the North, Central and South Aegean Sea and the Levantine Sea), which may be of different diagenetic potential due to a substantial environmental gradient in the sea surface temperature (SST; 18–23°C), salinity (35–40‰) and calcite saturation state from north to south. The samples were carefully selected to check the impact of diagenesis on both Ca isotopes and the Mg/Ca ratios measured on planktonic foraminifera. For additional information related to the scientific expeditions and sampling methods, we refer to previously published studies (Kontakiotis et al. 2011; Zarkogiannis et al. 2020a).

Analytical methods

Mg/Ca analyses

The previously published geochemical data for the Mg/Ca ratios referred to throughout this paper were reported by Kontakiotis et al. (2011) and Sabbatini et al. (2011). The G. ruber specimens from the 250–350 μm sieve fraction in both studies were cleaned following the procedure of Barker et al. (2003). This protocol omits the reductive hydrazine step of Boyle and Keigwin (1985) and includes: (1) multiple ultrasonic cleaning steps in water and then alcohol to remove clays and other fine material; (2) the elimination of organic matter through a hydrogen peroxide treatment at 100°C; and (3) rapid leaching with 0.001 M nitric acid to eliminate any adsorbed contaminants from the surface of the test fragments. The samples were then dissolved in 400 μL of 0.1 M nitric acid and centrifuged to remove any remaining insoluble particles.

δ44/40Ca analyses

The coarse fraction of the 12 samples considered in this study was cleaned using the HyPerCal cleaning protocol (Zarkogiannis et al. 2020b). On average, c. 15 G. ruber specimens from the 250–350 μm sieve fraction of each sample, resulting in c. 200–300 μg of CaCO3, were dissolved in distilled 2 M HNO3. The Ca was purified using an automated Ca–Sr separation method (PrepFAST MC, Elemental Scientific, Omaha, NE, USA). This process separates Ca from Sr, Mg and other matrix elements to avoid MgO+ and Sr2+ isobaric interferences during multi-collector inductively coupled mass spectrometry (MC-ICP-MS). Ca was eluted from the Ca–Sr column with 5 mL of 0.1 M HCl, dried down in Teflon beakers and redissolved in distilled 2% nitric acid.

The Ca isotope ratios were determined at the University of Oxford using a Nu Instruments multi-collector inductively coupled mass spectrometer with a desolvating nebulizer sample introduction system, following the method of Reynard et al. (2011). All solutions were at 10 ± 1 ppm concentration and the samples were measured with standard–sample bracketing using NIST SRM915a as the standard reference material. Each sample was analysed four times. We derived the δ44/42Ca and δ43/42Ca values from MC-ICP-MS measurements of 42Ca, 43Ca and 44Ca. On the basis that [K] << [Ca] and any isotopic variation was caused by natural mass-dependent fractionation, δ44/40Ca was calculated using δ44/40Ca = δ44/42Ca × [(43.956–39.963)/(43.956–41.959)] (Hippler et al. 2003).

We report the δ44/40Ca values normalized to NIST SRM 915a. Two SRM 915b solutions were purified alongside the samples to provide a combined column chemistry and analytical accuracy assessment. The measured values for our purified SRM 915b sample were δ44/40Ca = 0.72 ± 0.08‰ (2 SE, n = 10), which match the values obtained by thermal ionization mass spectrometry of δ44/40Ca = 0.72 ± 0.04‰ (2 SE; Heuser and Eisenhauer (2008). The uncertainty on the Ca isotope data is quoted as the t-distribution-derived 95% confidence interval on the mean of repeat measurements calculated using either the standard deviation on all repeat measurements on each sample or the standard deviation on all secondary standard analyses, whichever is greater. All previous data discussed herein have been recalculated and presented relative to the standard NIST SRM 915a and in terms of 44Ca/40Ca.

Scanning electron microscopy

Selected samples were carefully inspected using scanning electron microscopy (SEM) to assess the surface ultrastructure of G. ruber tests. The specimens were ultrasonically cleaned with ultrapure Milli-Q water for 30 s. Some specimens were gently crushed to expose the inner surfaces of the chambers. Both the crushed and uncrushed specimens from the various samples were mounted on SEM stubs and then coated with gold. High-resolution SEM analyses were performed at the National and Kapodistrian University of Athens, Department of Historical Geology-Paleontology, using a JEOL JSM 6360 scanning electron microscope.

Table 1 presents the sampling station details, including the sampling depth, SST, sea bottom temperature (SBT) and geochemical results. The water depth increases towards the southern stations, with corresponding decreases in the SBT, whereas the SST increases towards the south, reaching its highest in the Levantine Basin. Consequently, the SST–SBT temperature difference is significantly larger in the south (c. 7.5°C in the Herodotus Basin) than in the north (c. 2.5°C in the North-Central Aegean Sea), resulting in a larger temperature gradient between G. ruber calcite production and deposition in the southern regions.

A strong linear correlation (R2 = 0.62, P < 0.01) was observed between δ44/40Ca and Mg/Ca (Fig. 2), with the linear fit slightly better than a logarithmic fit (R2 = 0.60, P < 0.01). Excluding the Central Aegean Sea samples, the G. ruber Mg/Ca ratios range from 3.51 to 10.39 mmol mol−1, which is consistent with the variability previously reported for the Mediterranean (Ferguson et al. 2008; Kontakiotis et al. 2011; Sabbatini et al. 2011). However, the absolute Mg/Ca ratios are high considering the measured SSTs at these locations (c. 19–22°C, Table 1), where the expected Mg/Ca range would be c. 2.9–4.7 mmol mol−1 (Dämmer et al. 2020). The Central Aegean Sea samples exhibit particularly high Mg/Ca ratios (up to 15.28 mmol mol−1 and correspondingly high δ44/40Ca values, ranging from 0.90 to 1.01‰ (Fig. 2). Samples 1 (M51–3 #602) and 4 (SC2012-1) from the North Aegean Sea have the lowest Mg/Ca ratios (c. 3.7 mmol mol−1), which fall within the expected range for G. ruber (white) at these SSTs (Anand et al. 2003).

The δ44/40Ca composition of the 12 studied eastern Mediterranean samples varies by c. 0.60‰, with values ranging from +0.41 to +1.01‰ (see Table 1 and Fig. 2). These values are in line with previously measured δ44/40Ca values for G. ruber from Atlantic Ocean and Red Sea sediments, water columns or cultured samples, which typically range from +0.6 to +0.9‰ (Sime et al. 2005; Griffith et al. 2008; Kasemann et al. 2008; Kısakürek et al. 2011). However, the δ44/40Ca variation (c. 0.6‰) in this study is about double that observed in core-top and net-caught samples of this species from other regions (Sime et al. 2005; Griffith et al. 2008; Kasemann et al. 2008; Kısakürek et al. 2011). The lower δ44/40Ca range (0.41–0.47‰) observed in some samples is low compared with modern planktonic foraminifera, where the average Holocene δ44/40Ca composition is c. +0.8‰ (Kasemann et al. 2008).

The Ca isotopes show no significant correlation with either the SST (R2 = 0.04, P > 0.1) or SBT (R2 = 0.15, P > 0.1). Similarly, no linear (R2 = 0.03) or exponential (R2 = 0.03) relationship was found between the Mg/Ca ratios and the mean annual SSTs (Table 1). However, the correlation between the Mg/Ca ratios and SBTs strengthens significantly (R2 = 0.83, P < 0.01) when the two shallow North Aegean basin samples that precipitated in cooler waters with the lowest Mg/Ca values are excluded (Fig. 3). These two North Aegean Sea samples also have the lowest δ44/40Ca values.

Geochemical signatures of early diagenesis in eastern Mediterranean foraminifera

The high alkalinity waters and the ultra-oligotrophic nature of the eastern Mediterranean (Schneider et al. 2007; Siokou-Frangou et al. 2010) make this region ideal for studying forms of foraminiferal carbonate diagenesis other than dissolution or supra-lysoclinal dissolution. This study presents the first Ca isotope data for modern Mediterranean foraminifera, combining δ44/40Ca and Mg/Ca ratios from species-specific planktonic shells to assess the potential of different eastern Mediterranean basins to promote secondary diagenetic deposition on biogenic carbonates.

The anomalously high Mg/Ca ratios in most of the samples have been reported to be the result of early diagenesis, specifically caused by the inorganic precipitation of high-Mg calcite (Boussetta et al. 2011; Kontakiotis et al. 2011; Sabbatini et al. 2011) and were obtained principally from core-tops located in the central part of the Aegean Sea. Here, foraminifera shells not only have the highest Mg/Ca ratios (up to 15.28 mmol mol−1), but also display the highest δ44/40Ca values, ranging from 0.90 to 1.01‰ (Fig. 2), highlighting the effect of early diagenesis (CaCO3 overgrowths) in the central Aegean basin. Furthermore, we found that δ44/40Ca neither co-varies with the SST (R2 = 0.04, P > 0.1) nor with the SBT (R2 = 0.15, P > 0.1), suggesting that temperature is not the dominant control on the Ca isotopes, either in pristine foraminifera or in the foraminifera with inorganic CaCO3 overgrowths, in agreement with previous studies (Skulan et al. 1997; Griffith et al. 2008).

The observed variation in G. ruber δ44/40Ca values is noteworthy considering the narrow range of environmental conditions in the study area. Given the very small (c. 0.02‰ for each degree centigrade) expected temperature dependence of Ca isotopic fractionation in G. ruber and most foraminifera species (Skulan et al. 1997; Gussone et al. 2003; Sime et al. 2005), especially within this very narrow range of SSTs (c. 3°C), the recorded variability in δ44/40Ca rather points to a second, inorganic carbonate source. This is further supported by the correlation between the Mg/Ca ratios and the SBTs (Fig. 3), which is characteristic of calcite overgrowths that precipitate inorganically from seawater (Mucci 1987). However, given that the change in the Mg/Ca ratios of our samples (c. 60% for each degree centigrade) significantly exceeds the 3% for each degree centigrade predicted by thermodynamic Mg uptake into calcite (Rosenthal et al. 1997; Lea et al. 1999), this strongly suggests that the overgrowths act as a secondary calcite source, elevating the Mg content of our samples. The correlation between SBT and the Mg/Ca ratios became significant when samples 1 (M51-3 #602) and 4 (SC2012-1) from the North Aegean Sea were excluded. These two samples had both low δ44/40Ca and Mg/Ca values and are considered here to be pristine compared with the other samples that are affected to some degree by secondary diagenetic overgrowths. This conclusion is further supported by inspection of the specimens under SEM (Fig. 4).

Eastern Mediterranean G. ruber (white) shells from the North Aegean Sea (Fig. 2) were found to have the lowest δ44/40Ca values compared with those published for various oceanic basins (Sime et al. 2005; Griffith et al. 2008; Kasemann et al. 2008; Kısakürek et al. 2011). The two North Aegean Sea samples that also had the lowest Mg/Ca ratios of our dataset appeared to be unaffected by diagenetic calcite overgrowths (Fig. 4). The higher growth rates of foraminifers under optimum conditions in some species are correlated with changes in Ca isotope fractionation, resulting in an isotopic composition closer to that of seawater, indicating rapid precipitation with less fractionation (Kasemann et al. 2008). Given that the isotopic composition of Mediterranean seawater (⁠δ44/40Casw) is 1.96‰ (Sime et al. 2005), the light isotopic composition of the pristine G. ruber shells with δ44/40Ca values as low as c. 0.4‰ from the North Aegean Sea reveals slow foraminifera growth rates in eastern Mediterranean seawaters, which is in line with the ultra-oligotrophic nature of the basin (Ignatiades 2005).

The G. ruber shell δ44/40Ca isotopes in this study co-varied with the shell Mg/Ca ratios (Fig. 2). Such a relationship between the δ44/40Ca isotopes and the incorporation of Mg2+ in biogenic calcite has been reported for Neogloboquadrina pachyderma (Kozdon et al. 2009) in core-top sample material from the polar regions, as well as for fossil Globigerinoides sacculifer shells from the equatorial regions of the Atlantic Ocean (Nägler et al. 2000). Figure 5 compares our results with those of previous studies that measured both δ44/40Ca and Mg/Ca values in foraminifera and shows that there is a clear distinction between pristine calcite (from the Atlantic Ocean and the Nordic seas) and diagenetically altered calcite (from the present study in the eastern Mediterranean).

Some of the scatter in our results may stem from the fact that the δ44/40Ca and trace element analyses were performed on different aliquots of foraminifera from the same core-top sediments. However, in contrast with the biogenic foraminiferal carbonate of the previous two species for which the slopes between δ44/40Ca and Mg/Ca are c. 0.6 and c. 0.7, respectively, the rate of change in the Mg/Ca ratio of the diagenetically altered shell with the changes in its Ca isotopic composition is c. 14 (Fig. 5). A four-fold change in the Mg/Ca ratios of our samples cannot be explained by Rayleigh fractionation applied to the overgrowth CaCO3 either. Any such Rayleigh fractionation (as [Ca] and [Mg] in the pore solution become depleted) would at most double the overgrowth Mg/Ca ratio, as per calculations with the carbonate modelling tool of Owen et al. (2018). Instead, a simpler explanation is that the data form a mixing line (Fig. 5) between pristine foraminifera (Mg/Ca c. 3 mmol mol−1, δ44/40Ca c. 0.4‰) and pure inorganic CaCO3 (Mg/Ca c. 72 mmol  mol−1, δ44/40Ca c. 1.96‰) precipitated at equilibrium from eastern Mediterranean seawater. The Mg/Ca ratio of the inorganic CaCO3 was calculated from the Mg/Ca of Mediterranean seawater (c. 5.1 mol mol−1, Lebrato et al. 2020) by applying an experimental inorganic partition coefficient (Day and Henderson 2013) to derive the Mg/Ca ratio of the CaCO3.

The inorganic CaCO₃ (Mg/Ca c. 72 mmol mol−1) points to high-Mg calcite (>4 mol% Mg; Morse 2003), which aligns with previous experimental findings on the nature of the overgrowths (Boussetta et al. 2011). The δ44/40Ca of the inorganic CaCO3 end-member was calculated using Mediterranean seawater δ44/40Ca value of 1.96‰ (Sime et al. 2005) assuming no fractionation between seawater and inorganic calcite. This simple mixing model is further supported by the additional comparison of Ca isotopes with the Sr/Ca ratio in our samples (see Supplementary Materials for additional mixing line plots and for full details of underlying calculations).

Assessment of diagenetic calcite overgrowths through Ca isotope quantification

Sime et al. (2005) reported the δ44/40Ca of the Mediterranean seawater (⁠δ44/40Casw) to be 1.96‰ (2σ = 0.14‰). Foraminifera carbonate is isotopically depleted in 44Ca relative to ambient seawater (Griffith et al. 2008). If the temperature-dependent fractionation of Ca isotopes in the study area is minimal (c. 3°C range of SSTs), then any inorganic calcite precipitated in equilibrium with seawater (assuming a 0‰ fractionation) will have a Ca isotope composition close to that of seawater. By solving the isotopic mass balance equation for the sample with the highest δ44/40Ca (sample 5, C38)
(1)
(1)
where X is the amount of pristine biogenic calcite (with the lowest δ44/40Ca) and (1 − X) is the amount of secondary authigenic calcite, we find that the authigenic calcite can represent up to 38% of the total calcite mass (2σ = 31–41% when considering the uncertainty in δ44/40Casw).

The Ca isotopic composition of our sample set can be categorized into three distinct groups. The first group consists of samples with δ44/40Ca values between 0.41 and 0.47‰, including samples 1 (M51-3 #602) and 4 (SC2012-1), which also have the lowest Mg/Ca ratios. These samples represent locations unaffected by early diagenesis. The second group exhibits δ44/40Ca values of c. 0.58–0.73‰, which fall within the range published for modern G. ruber from the global ocean (Sime et al. 2005; Griffith et al. 2008; Kasemann et al. 2008; Kısakürek et al. 2011). However, their elevated Mg/Ca ratios are known to be caused by secondary calcite overgrowths (Kontakiotis et al. 2011, 2017; Sabbatini et al. 2011), with the amount of added/precipitated calcite overgrowths calculated here to be c. 12% (2σ = 3–19%) using the simple mixing model of Equation (1).

Sample 7 (M51-3 #578), despite its low to normal Mg/Ca ratio, and sample 10 (M71-3 #Ier01), despite its low to normal δ44/40Ca values, are included in the second group due to their relatively high δ44/40Ca and Mg/Ca values, respectively. The discrepancy between their Mg/Ca and δ44/40Ca values may from the different analyses being performed on separate sets of specimens, with one set potentially being more affected by diagenesis than the other. The third group consists of samples with δ44/40Ca values as high as 0.90–1.01‰ and Mg/Ca ratios severely increased due to c. 33% (2σ = 27–39%) in situ calcite precipitation.

The higher Ca isotope values in the two groups are likely to have been driven by early diagenesis, specifically the formation of calcite overgrowths. Other potential factors for δ44/40Ca variation, such as gametogenic calcite (Kasemann et al. 2008), dissolution or the ambient salinity, seem less plausible. G. ruber does not deposit a gametogenic crust of significant thickness (Caron et al. 1990) and its primary test has a relatively homogeneous elemental composition (Benway et al. 2003; Eggins et al. 2003). Moreover, neither the ambient ion strength (Tang et al. 2012) nor dissolution (Hönisch 2002; Fantle and DePaolo 2007) has been shown to significantly impact the Ca composition.

The group of samples that is heavily affected by secondary calcite overgrowths comes from the shallow and warm southern Aegean basins, whereas the amount of overgrowth is less (c. 15%) in the samples from the deeper and colder basins (Fig. 3). A possible reason for this could be the change in the calcite saturation state (⁠Ωcc), such that sediment from shallower basins has an increased potential for diagenesis due to higher Ωcc values near the sea surface (Álvarez et al. 2014). The warm, highly alkaline waters of the eastern Mediterranean are generally supersaturated with respect to calcite, with Ωcc values ranging from 4–6 at the surface to 2–3 at the bottom (Schneider et al. 2007; Álvarez et al. 2014). This supersaturation likely promotes in situ calcite precipitation at the sediment–water interface or within pore waters.

The high Ωcc values in the deep Mediterranean waters are unusual compared with the open ocean, where most bottom waters are either undersaturated or only slightly supersaturated (⁠Ωcc slightly >1). Given that Ωcc is higher in shallower waters (Table 1), the shallow core-top samples from stations 2 (C6), 3 (C18) and 5 (C38) (c. 250 m water depth) are more diagenetically altered, leading to higher Mg/Ca ratios and δ44/40Ca values than samples from greater depths.

There are at least two processes likely to lead to early diagenetic alterations under fluid-buffered conditions for Ca: (1) a reduction in the sedimentation rate or a depositional hiatus that keeps shallow sediments at or near the seafloor for prolonged periods of time; and (2) fluid flow in shallow sediments driven by various external processes (e.g. changes in eustatic sea-level or density gradients due to evaporation). Both of these processes seem to take place in the region from which samples 2 (C6), 3 (C18) and 5 (C38) were retrieved. The waters of the South Aegean Sea are the densest in the eastern Mediterranean due to their salinity and have lower temperatures than the Levantine Sea (Klein et al. 1999; Zarkogiannis et al. 2020a). Furthermore, despite the similarity in the collection depths of samples 2 (C6), 3 (C18) and 4 (SC2012-1), the difference in the sedimentation rates between the east and west of the southern Aegean basins (higher in the east), along with the different sediment provenance (Leontopoulou et al. 2019, 2022), could explain why the samples from the western region are more affected by secondary calcite overgrowths.

The changes in the Ca isotopic composition of sedimentary foraminifera specimens, resulting from variable in situ carbonate mineral authigenesis, have been used in this study to quantify these overgrowths, constituting almost 40% of the total calcite volume. Estimates for the Aegean Sea surpass those of Boussetta et al. (2011) from the Levantine Sea (c. 20%), suggesting the northeastern Mediterranean as a highly reactive region for carbonate mineral authigenesis. However, it is essential to note that the mass balance calculations in this study should be considered preliminary. Further research is necessary to determine the Ca isotopic composition of modern eastern Mediterranean waters and foraminifera, thus establishing the mass balance end-members. The elemental mass balance suggests that the precipitating minerals are low-Mg2+ calcites (<4 mol% Mg2+; Morse 2003) rather than aragonite (or high-Mg2+ calcite) as would be expected from direct precipitation from seawater. This observation hints that the calcareous foraminifera shell may serve as a nucleation site for precipitation, also exerting control over the mineralogy of the precipitates.

Our data show that the Mg/Ca ratios and δ44/40Ca of G. ruber shells retrieved from core-tops of the semi-enclosed eastern Mediterranean Sea are correlated and they both increase as a function of the degree of early diagenesis due to post-depositional secondary calcite overgrowths. The mass balance calculations suggest that a considerable amount of authigenic calcite (up to c. 40% by mass), particularly within the shallow central Aegean sub-basins, can cover the surface of the foraminifera shells after deposition, although more information is needed for precise estimates of the δ44/40Ca values in the eastern Mediterranean foraminifera and seawater. Nevertheless, the combined Mg/Ca and δ44/40Ca analyses of foraminifera shells offer a promising method to quantify systematic offsets caused by calcite authigenesis. Studies such as these can provide important mechanistic insights that may help to correct deeper time proxy records for diagenesis, leading to more accurate palaeotemperature reconstructions in the eastern Mediterranean Sea. Much of the scatter between the published Mg/Ca data and our δ44/40Ca values likely originates from the use of different sample splits. Future research on using combined Mg/Ca ratios and δ44/40Ca values as an indicator of diagenesis should include simultaneous measurements of aliquots from the same samples, along with size-restricted foraminifera specimens.

The authors thank Elizabeth Griffith and the anonymous reviewer for their valuable and constructive comments, which significantly improved the quality of this paper.

SZ: conceptualization (lead), investigation (lead), methodology (equal), visualization (lead), writing – original draft (lead); CD: data curation (lead), methodology (equal), writing – review and editing (equal); GK: resources (equal), writing – review and editing (equal).

This research received no external funding.

The authors declare no conflicts of interest.

All the data required for the reproduction of the study are given in Table 1 and the Supplementary Material.