The use of submerged speleothems for sea level research in the Mediterranean Sea: a new perspective using glacial- and hydro-isostatic adjustment (GIA)

The investigation of submerged speleothems for sea level studies has made significant contributions to the understanding of the global and regional sea level variations during the Middle and Late Quaternary. This has been especially the case for the Mediterranean Sea, where more than 300 submerged speleothems sampled in 32 caves have been analysed so far. Here, we present a comprehensive review of the results obtained from the study of submerged speleothems since 1978. The studied speleothems cover the last 1.4 Myr and are focused mainly on Marine Isotope Stages (MIS) 1, 2, 3, 5.1, 5.3, 5.5, 7.1, 7.2, 7.3 and 7.5. Results reveal that submerged speleothems represent extraordinary archives providing accurate information on former sea level changes. New results from a stalagmite collected at Palinuro (Campania, Italy) and characterized by marine overgrowth are also reported. The measured elevations of speleothems are affected by the local response to glacialand hydro-isostatic adjustment (GIA), and thus might significantly deviate from the global eustatic signal. The comparison between the ages and altitude values of the Mediterranean speleothems and the flowstone from Bahamas with local GIA provides a new scenario for MIS 5 and 7 sea level reconstrutions.


Introduction
The study of submerged speleothems in coastal caves significantly contributed to constrain past sea level variations for the last 1.4 Myr. In particular, they have provided relevant information on the sea level for several climatically-important Marine Isotope Stages, including MIS 1, 2, 3, 5.1, 5.3, 5.5, 6.5, 7.1, 7.2, 7.3 and 7.5, especially in the Mediterranean Sea.
The importance of sea level in controlling groundwater elevation in both inland and coastal karst was first proposed by [1] but it was only in the early 1970s that the potential of using speleothems to reconstruct former sea levels was fully recognized. For example, [2] discussed Late Pleistocene sea level fluctuations based on radiocarbon ( 14 C) and U/Th ages from a submerged stalagmite retrieved from the Ben's Hole cave in Grand Bahama Island, and [3] interpreted the origin The number of speleothem studies for sea level reconstructions has significantly increased after the 80s, when the development of mass spectrometric techniques for the measurement of uranium and thorium isotopes has led to increases in the typical precision and accuracy of U-series ages compared to alpha-counting measurements and reduced sample size requirements [40,41]. Speleothems can usually provide more reliable records than, for instance, shallow-water corals given that their dense calcitic or aragonitic structure is less susceptible to post-depositional alteration, significantly reducing the isotope exchange with the surrounding environment.
Isotope-dilution mass spectrometry for submerged speleothem dating was first applied by [4] and [6] on a flowstone (DWBAH) recovered from 15 m below present sea level in Grand Bahama Island, which contains 5 hiathuses (Appendix B) and provided a record of the sea level over the past 280 kyrs. These studies reconstructed one of the first sea level change curve which, together with the pioneering works by [42,43] on fossil corals, has considerably contributed to the knowledge in the Appendix B) U/Th ages vs. sampling depth of the flowstone DWBAH (redrawn from [4]) field of sea level change.
Similarly, the use of thermal ionization mass spectrometry (TIMS) and multi-collector inductively coupled plasma mass spectrometry (MC-ICPMS) for precise U/Th dating of Mediterranean speleothems greatly improved our understanding of former sea level changes.
In addition to U/Th dating, the vadose stalagmites and stalagtites and POS can also be dated by 14

C.
However, the radiocarbon ages can be affected by the "dead carbon proportion" (DCP), which is the percentage of dead carbon incorporated in the speleothem or POS at the time of formation. The DCP is mainly the fraction of carbon within the calcium carbonate that is derived from the equilibration with the soil CO2 and the radiocarbon-free "dead carbon" from the bedrock [44]. High DCP values (e.g. > 20%), resulting from a stronger water-soil-rock interaction of the precipitating water, overestimate the radiocarbon ages and cause a large 14 C-U/Th age offset. In a recent study, [45] reported combined 14 Tables 3 and 4.

General overview of the Mediterranean region
The Mediterranean Sea (MS) is a marine basin almost completely bordered by land and covers an area of about 2.5 million km 2 . Its geographical features were widely reviewed by [46]. The coastline extends for about 46,000 km 2 and approximately half of this is rocky, with plunging cliffs, sloping coasts, screes and shore platforms [47]. It can be divided into the Western and Eastern Mediterranean along an imaginary axis between the Straits of Sicily and Tunisia but a large scientific literature also identify an undefined "central Mediterranean area" [48].
The MS has a long and complex geological history that began about 250 Myr ago, following the break-up of the Pangea and the formation of the Tethys Ocean, the forerunner of the MS [49]. From a geological point of view, the Mediterranean borders the westernmost sector of the Alpine-Himalayan orogenic belt. Its geodynamic evolution was driven by the differential seafloor spreading along the Mid-Atlantic Ridge, which led to the Alpine orogenesis [49]. It hosts wide extensional basins and migrating tectonic arcs. Vertical and horizontal movements control the geological and geomorphological history of the area. The MS includes zones of active subduction associated with volcanic activity and older zones of quiescent subduction [49]. The coastal reliefs, coupled with seismic activity generated by geodynamics, drive most erosional processes within the area. For the aforementioned reasons, rocky coastal landforms in the MS are closely connected with sea level history [48]. These coastal landforms are spread in elevation from few metres up to more than 150 m asl, due to the relevant tectonic uplift that patchily affects the basin coastline [50].
Pleistocene highstands have mainly been responsible for the formation of stepped flights of terraces along the rocky sections of the MS.
The MS is situated at the boundary between the subtropical and mid-latitudes zones and its climate is characterized by warm and dry summers and mild and rainy winters [51]. It is a semi-enclosed and highly evaporative basin that is connected with the Atlantic Ocean through the Strait of Gibraltar (sill depth ~ 300 m) and with the Black Sea through the Strait of Dardanelles (sill depth ~100 m) and the Bosphorus Strait (sill depth ~65 m). The Atlantic water that enters the Mediterranean Sea spreads throughout the entire basin and participates in the formation of intermediate and deep waters that contribute to the Mediterranean thermohaline circulation [52]. In particular, since evaporation exceeds precipitation and river runoff, the relatively fresh (salinity ~ 36.5) surface Atlantic Water entering the Mediterranean Sea across the Strait of Gibraltar at the surface becomes progressively saltier (~ 38.5) and denser during eastward advection. Evaporation and mixing together with intense cooling and strong wind-induced heat loss in specific areas in winter (Gulf of Lion, Adriatic Sea, Levantine and Aegean Seas) result in denser waters that sink via convection and form the intermediate and deep waters in the Mediterranean Sea [53,54].
Tides vary from place to place along the coasts of the Mediterranean, depending on many parameters, such as coastal geometry and bathymetry, but in general Mediterraenan tides have lower amplitudes with respect to oceanic ones. The average tidal amplitude is about 40 cm, with the exception of the remarkably large tides observed in the Gulf of Gabes (Tunisia) and in parts of the North Adriatic Sea, where they may reach amplitudes up to 1.80 m. In other areas, such as in Greece or Sicily, tides are very small, especially near the amphidromic points where the tidal range is almost non-existent. At the Strait of Gibraltar, tide increases to 1.5 m due to the influence of the Atlantic Ocean, but it quickly decreases eastwards. Weather conditions can significantly reduce or amplify the tidal amplitude, with variations up to 1 m.

Numerical modelling of Glacial-and hydro-isostatic adjustment
We compute the GIA-driven RSL curves at the relevant Mediterranean sites by solving the gravitationally and rotationally self-consistent sea level equation [38,55,56] and including the relastic treatment of variable coastlines. We assume a spherically symmetric, deformable but incompressible, self-gravitating and rotating Earth model that is characterized by a Maxwel viscoelastic rheology.
We employ a four-layer model characterized by the VM2 mantle viscosity profile [57] and combined with a lithosphere thickness of 90 km (see Table 1). We also employ a three-layer model and test three mantle viscosity profiles (MVP 1-3), each combined with a lithosphere thickness of 100km (see Table 2). The viscosity gradient of viscosity as a function of the Earth's radius increases of one order of magnitude from MVP1 to MVP3.

VM2
LT ( Table 2 Elastic lithosphere thickness and viscosity of the three-layer MVP 1-3 profiles.
We employ the following ice sheets models as forcing functions: -ICE-5G [57], where two consecutive chronologies are combined in series to cover the last 250 ka. -ICE-6G [58], where two consecutive chronologies are combined in series to cover the last 250 ka. -ANICE-SELEN [38,59,60], where a ~ 2.5 m eustatic highstand occurs during MIS 5.5, mostly from the Greenland Ice Sheet melting.

Speleothem samples
The Mediterranean speleothems reported in this study were collected in coastal caves in the Balearic Islands and along the Croatian, Maltese and Italian coasts ( Fig. 1; Table 1

Speleothem from Palinuro
A ~ 10 cm stalagmite was collected in a fossil tidal notch along the carbonate coast of the Palinuro promontory in the Campania region (South Italy) at 2.1 m above the sea level. A variety of living organisms, including the crustose coralline algae Lithophyllum, the acorn barnacle Chthamalus and the limpet Patella, were obsersed below the tidal notch, exposed to strong wave conditions. The external surface of the speleothem was completely covered by a blackish patina that is likely the result of the precipitation of Fe-Mn oxides when the rising sea level approached the speleothem ( Fig.   5 (4,7)). The top portion of the stalagmite is encrusted by a marine overgrowth that consists of several specimens of the barnacle Chthamalus stellatus that typically lives in the intertidal zone on high energy rocky shores and can be found above the highest tidal level [66].

14 C and U-series dating
Three carbonate fragments were sub-sampled from the upper portion of the stalagmite from Palinuro ( Fig.5 (4,7,8)) and prepare for radiometric dating. In particular, sub-sample 1 (Palinuro-1) was collected within the marine overgrowth (barnacle) and analysed for AMS-14 C. The fragment was carefully cleaned using a small diamond blade to remove any visible speleothem calcite and retrive only the encrusted overgrowth. The sample was further processed and analysed for 14  were separated using UTEVA resin (Eichrom Technologies, USA) and analysed using a ThermoScientific Neptune Plus multi-collector inductively coupled plasma-mass spectrometer following the protocol developed at LSCE [69]. The 230 Th/U ages were calculated from measured atomic ratios through iterative age estimation [70], using the 230 Th, 234 U and 238 U decay constants of [71] and [72]. The ages were corrected for the non-radiogenic detrital 230 Th fraction using an initial 230 Th/ 232 Th activity ratio of 0.85 ± 0.36, which corresponds to the mean upper crust value [73].
The U/Th ages of the DWBAH fownstone reported in [6] have been re-calculated using the most recent 230 Th, 234 U and 238 U decay constants of [71] and [72] and corrected for the detrital 230 Th fraction using a initial 230 Th/ 232 Th activity ratio of 0.8 ± 0.8 (Supplementary table 1), as suggested by [74] for the Bahamas speleothems. The difference between the original U/Th ages [6] and the re-calculated and corrected ages is minimal (< 3.5 kyr for ages younger than 270 kyr).

Results
Tables 3 and 4 report the relevant details of the sampling locations and the published U/Th and 14 C ages of the vadose speleothems, POS and marine biogenic overgrowths used for the present Mediterranean submerged speleothems, which were published between 1992 and 2020: 33 caves ( Figure 1, Table 3) carved on carbonate lithology that preserved vadose speleothems, POS and marine biogenic overgrowth that were used to reconstruct former sea level variations. An essential condition for the conservation of submerged speleothems is that the caves entrance are small in size so that the energy of the waves can be almost completely attenuated.
On the Adriatic coasts of Croatia, speleothems were sampled in 7 submerged caves distributed along ~ 300 km of coast ( Figure 1) and dated by 14 C and U/Th [19,24,65,75]. Their ages range between 7 and 217 kyr (MIS 1 and MIS 7), and results are summarized in [21]. Finally, a submerged speleothem was sampled at -14.5 m in the southwestern coast of Malta and dated by radiocarbon. Results showed that Malta remained tectonically stable during the Holocene. dots are the areas where X-ray diffraction analyses were carried by [21] to identify the hiathus between 87 kyr and 82 kyr.  Table 4 Location, depth, sample,U\Th, corrected age of the speleothem studied in Croatia (see also Fig. 2). 14 C ages of speleothems L-1 and L-10 measured by a liquid scintillationcounter, and of speleothems Z-41, B-38, B-36, B-34, B-28, P-23 and R-21 measured by agas proportional counter. The youngest parts are marked with A and S, while B regards theoldest parts. 14 C ages are expressed as conventional 14 C corrected for A0 = 85% andmeasured δ 13 C (-8‰ when not measured), and as calibrated ages.

Croatia (2005 -2010)
In particular, 16 submerged speleothems were collected at depths between -41.4 and -1.5 m in 7 caves (out of 235 visited) located along the Croatian coast over a distance of ~ 300 km. The age model of these samples was constrained by 15 radiocarbon measurements and 36 U/Th analyses. Among these, the U/Th ages obtained from the stalagmites K-14 and K-18 sampled in U vode Pit (Krk Island,  remaining of bryozoans and sponges [17]. This encrustation plays a significant role because it protects the continental portion of the speleothem from degradation, bioerosion or dissolution. The spelethems from the Argentarola and Rumena caves in Italy [17,23] represent the only cases in the world where more than one continuous deposition of continental and marine layers were observed in the same speleothem (for subsequent marine and continental highstands).

Malta
[64] studied a submerged spelothem collected at −14.5 m depth in a recently discovered submerged cave at Gebel Ciantar, Malta island Table 1, Figure 5 (1). Since the cave was mainly formed in a subaerial karst environment, the radiocarbon dating of the speleothem with serpulid encrustations Figure 5 (6) enabled reconstructing the sea level during the mid-Holocene when the cave was fully flooded. In particular, the mean radiocarbon age of 7.6 kyr perfectly aligns with the sea level curve predicted by [78] for Malta. [64] concluded that the Maltese islands were tectonically stable during the mid-Holocene, and this tectonic behaviour still persists nowadays.

Mallorca (1974 -2020)
Appendix C: Tide amplitude and the POS accretion, redrawn from [27] The results of more than 30 years of work carried out in 10 different caves of Mallorca, all located along the southern coast of the island (Figure 1) have been published in numerous papers [26,28,35,45,61,76]. POS have been extensively studied and radiometrically dated, both modern samples formed few cm above and below the water   Figure 6. As also mentioned by the authors, the older ages (> 650 kyr), which were obtained using the 234 U/ 238 U disequilibrium method, suffer of large uncertainties attributable mainly to the errors associate to the initial  234 U estimates needed to calculate  234 U ages.

Palinuro
The

Predicted RSL curves
The predicted RSL curves for ICE-5G, ICE-6G and ANICE-SELEN and VM2 mantle viscosity profile (grey curves in Fig. 8  The relatively limited regional RSL variability during the interglacials, is characterized by a maximum deviation from the eustatic of ~2.5 m for the three ice sheets models (Fig. 8 a-

C-age (yrs BP) Median probability (yrs BP)
This also supports the results of [39] for Bahamas, where the RSL gradient as a function of latitude vanishes during the MIS 5.5. Our predict RSL curves for Bahamas (red curves in Fig. 8 a-c) [37,38], and shows that the ice-loading term causes strong regional RSL gradients also in the Mediterranean Sea.
Also at Bahamas, the deviation from the eustatic increases towards the interstadials and glacials [39],   Fig. 8 d-f). Again, the variability increases towards the interstadials and glacials, implying the viscosity gradient is a significant factor in the regional RSL variability in the Mediterranean Sea. The predicted variability at Bahamas is significantly larger than in the Mediterranean (red, green and blue curves in Fig. 8 d-f) and fully bounds the predictions at the Mediterranean sites. Increasing the viscosity gradient, i.e. moving from MVP 1 to MVP 3, results in an upward shift of the RSL curves between MIS 5.5 and LGM. The predicted total RSL glacial-interglacial excursions (glacial-interglacial) for MVP 1are larger than for MVP3, indicating that a larger viscosity gradient (and a larger viscosity value for the lower mantle) results in a significant delay of the solid Earth response. This can be appreciated during the MIS 5.5, when the maximum peak of transgression appears at the end of the interglacial and during the Holocene (i.e. steeper RSL rise curves).

Discussions
The use of submerged speleothems provided remarkable results for the sea level change history during the Holocene and for the long-term sea level change reconstruction during the Late Pleistocene, in particular for MIS 1, 7.1, 7.2, 7.3 and 7.5 highstands (Argentarola stalagmites), MIS 5.1 highstand (Croatia and Mallorca), MIS 5.5 highstand (Mallorca) and MIS 6.5 highstand (Nettuno cave in Italy) (Tables 1, 2). Unfortunately, marine overgrowth layers encrusting Italian or Croatian speleothems with ages beyond the radiocarbon dating range (i.e. > 40-50 kyr) cannot be analysed by the U-series disequilibrium method due to diagenetic overprints that cause exchange of uranium [86].

Palinuro speleothem
For the interpretation of the 14 C and U/Th results obtained from the Palinuro sample, a study of the tidal area was conducted along the promontory of Palinuro over a distance of ~3 km from the sampling site. The survey revealed the presence of a fossil tidal notch, below which we found specimens of the crustose coralline algae Lithophyllum, the acorn barnacle Chthamalus and the limpet Patella. Based on [78], the sea level was 4.4 m, 3.3 m and 0.73 m lower that the present level at 5.5 kyr, 4.5 kyr and 1.6 kyr, respectively (Table 3, Figure 7). Therefore, the barnacles started to colonize the stalagmite at 1.6 kyr when the rising sea reached a level that exposed the sample to strong wave conditions, favourable to the barnacle growth. However, no living barnacles were observed during the sampling of the stalagmite and this might be due to changes of the environmental conditions since the deposition of the fossil barnacles.  [17,18]. Based on the  18 O results obtained from the serpulid overgrowth on a stalagmite collected at -18.5 m from the same cave (Figure 4 (6)) and the correlation with foraminifera  18 O values from the Mediterranean ODP site 975 [87] , [16] identified as MIS 5 the timing of the marine overgrowth deposition, without any visible interruption from MIS 5.1 to MIS 5.5 (Figure 9). It is worth mentioning that the timing of the MIS 7.3 and 7.2 highstands identified by U/Th dating on the speleothem agrees with the global sea level curve reconstructed by [88]. In the Nettuno cave, the deepest speleothem sampled in the Mediterranean at -52 m was dated by the alpha-counting U-series method (see Table 1 Figure 4), which provided an age of 165 ± 12 kyr, corresponding to MIS 6.5 and in agreement with [88].

Speleothems from Italy
Finally, as regard the speleothem Plemm A (Table 3) in the Plemmirio cave [62] collected the sample at -20.2 which is corrected to -35.3 m when considering the tectonic uplift (0.2 mm/yr, [62]) .

Speleothems from Croatia
The speleothems retrieved along the Croatian coast and studied by [20,21,24,65,75,89,90] greatly contributed to the sea level reconstruction from 1.5 kyr to 220 kyr ( Figure 11). In particular, high-precision U/Th. dating of the speleothems K-14 and K-18 sampled at -18.8 m and -14.5 m in the U vode Pit cave (Krk Island) provided significant data to constrain the relative sea level curve during MIS 5.1 highstand. Regarding the presence of 2 hiatuses in the K-18 stalagmite, on the basis of the lack of XRD analyzes that invoked the presence of a hiatus between 90.8 ka and 82.9 ka, in correspondence with a change in color, we suppose that only the second hiatus is present between 77.7 ka and 64.5 kyr Fig. 3. The ~13-18 m difference between the elevation of the two speleothems (-18.8 m and -14.5 m) with the ice-volume-equivalent global sea level curve by [91] has been justified by [21] invoking a long-term regional tectonic uplift with an average rate of 0.15-0.25 mm/yr during the last 75-85 kyr. However, the vertical tectonic movements along the Croatian coast are still a matter of debate. For example, [90] suggested a generalized downlift in Istria and uplift along the southern cost of Croatia. [7] investigated the carbonate coast stretching from Trieste to southern Croatia and showed that the tidal notch is always submerged between 1.8 and 0.5 m below mean sea level. Furthermore, archaeological studies along the Croatian coast revealed the presence of Roman remains (2 kyr BP), including pier and fishtanks, at ~1.5-1.8 m below sea level as a result of tectonic subsidence or coseismic downlift [92,93]. Finally, none of the fossil outcrops related to the MIS 5.5 highstand that are usually found at ~6 m a.s.l. in tectonically stable regions [36,94] have been observed so far in the north-eastern Adriatic Sea. Therefore, we hypothesize that most of this area is affected by tectonic subsidence (a minimum of 6 m since MIS 5.5), with possibly local uplift movements [21]. Accordingly, we decided to correct the elevation of the stalagmites K-14 and K-18 by 4 m, to -10.4 and -14.7 m, respectively, assuming a constant subsidence of 0.048 mm/yr for the last 80 kyr.

Speleothems from Mallorca
A large number of precise U and Th isotope measurements have been performed on submerged speleothems sampled from 10 coastal caves on the island of Mallorca. Data published by [35] agree with previous studies reporting MIS 5.5 fossil deposits in Mallorca [95,96]. These deposits contain the Tyrrhenian Senegalese fauna assemblage that was analysed by the amino-acid dating technique [95] and remainings of the coral Cladocora caespitosa that were dated by U/Th [96].
The elevation of MIS 5.1 POS samples studied by [28] in Mallorca still represents a scientific debate.
The reconstructions obtained by [88] and [97] indicate that sea level during MIS 5.1 was 21. • The finding of some deposits from the same age between +2 and +1.5 m near Gibraltar [98]; but disagree with: • All the eustatic curves (in particular [75] and [85]); • The U/Th ages reported by [84] for Mallorca; • The speleothem from Plemmirio located at -20.2 m [62]; • The speleothems K-18 and K-14 from Croatia at -18.8 m and -14.5 m; • The DWBAH flowstone from Bahamas at -15 m; • The morphology of the MIS 5.5 tidal notches  and the lack of younger tidal notches; • The mushroom-like landforms with tidal notches at -25 m found at Tavolara island (Sardinia) and interpreted by [36] as being MIS 5.1 in age; • The results of glacial and hydro-isostatic adjustment (GIA) of the Mediterranean sea (See Figures 8,9).
We decided to compare the vertical position of most of the submerged speleothems described in this review with the DWBAH flowstone [4,6] (Fig. 15, 16 and 17). This exceptional flowstone contain 5 hiatuses identified by the deposition of a thin layer of red or yellow mud (Fig. 12), and was extensively dated by the U-series disequilibrium method, providing precise ages that unravealed for  mm/yr, as evaluated by [63].
In Croatia, the elevation of the stalagmites K-14 and K-18 was corrected considering a subsidence of ~ 4 m in 81 kyr. The correction derives from the fact that MIS 5.5 at 6 m a.s.l., or at higher elevations, has been never found in Croatia. This means that MIS 5.5 is below the present-day sea level and 4 m is extrapolated from a subsidence of the coast of at least 6 m in 125 kyr, corresponding to MIS 5.5.
The elevation of the other speleothems considered in the present review (i.e. Argentarola cave, Mallorca, Malta and Sardinia) was not corrected for uplift or subsidence owing to the tectonic stability of these areas.

GIA-modulated RSL curves
The GIA modulated regional RSL variability in the Mediterranean is minimal during the Interglacials (~2.5 at MIS 5.5) but increases during the Interstadials (~15 m at MIS 5.1-5.3), thus complicating the task of finding the correct eustatic value (grey curves in Fig. 8). In fact, the GIA signal is strongly dependent on the solid Earth rheological profile, which is itself an unknown. In  (Fig. 9 b), implying that the local signal is almost eustatic between MIS 5.1 and MIS 5.3, and the role of the mantle viscosity profile is negligible (see also Figure 16). The MIS 5.1-5.3 RSL variability increases slightly in the other Mediterranean sites and is larger for ANICE-SELEN (blue curves in Fig. 9 c-e; see also Fig. 14).
When compared to the observations, the predictions are generally significantly lower, with the MVP 3 scenarios reducing the vertical gap between model predictions and observation (Fig. 9 a-f  The predicted MIS 5.1 regional RSL variability, according to ANICE-SELEN and MVP 3, is characterized by higher-than-eustatic values towards East-North East and around the continental margins (Fig. 17). On the other hand,the predicted values are lower than eustatic at the center of the basins towards South East (where the ocean loading term dominates). Interestingly, Plemmirio and Mallorca are very close to the eustatic (Fig. 17).
The use of submerged speleothems significantly contributed to reconstruct the short-term (Holocene) and long-term (Middle and mostly Late Pleistocene) sea level changes in the Mediterranean. Even though there are still some open questions, the different groups that have been working on the speleothems from Mallorca, Croatia and Italy since many years, have carried out an extensive and valuable work, which significantly contributed to improve our knowledge on the sea level changes during the Mid-and Late Pleistocene.
The present review reports the elevations and radiometric ages ( 14 C and U/Th) of the published submerged speleothems in the Mediterranean. Elevation data are corrected for local tectonics (uplift or subsidence) and compared with results from the DWBAH flowstone in Grand Bahama Island.
Holocene: Most of the data on the Holocene sea level curve in the Mediterranean result from the investigation of continental portions of submerged speleothems, marine overgrowth (serpulids) and Lithophaga boring mussel in samples collected in Italy and Croatia; MIS 3: The K-14 stalagmite from Croatia shows that sea level during MIS 3 never reached 15 m b.s.l. Sea level during MIS 3 was lower than 21.7 m in Argentarola, lower than 25 m in Marettimo and lower than 45 m in Sicily. However, this latter value was corrected for vertical tectonics. . However, other fossil evidences suggest sea level between 7 m and 8 m a.s.l. [36]; [35] reported a duration of 11 kyr for MIS 5.5, from 116 kyr and 127 kyr; MIS 6.5: The deepest stalagmite collected so far in the MS from the Nettuno cave (Sardinia) at -52 m and dated to ~ 165 kyr [81] suggests that the sea level was lower than the elevation of the submerged sample; MIS 7.1: Three speleothems from the Argentarola cave ( Figure 13) clearly show a serpulid layer between MIS 6 and MIS 7.2 [17,18]. The sea level was therefore higher than 18 m b.s.l.. Based on the results from the DWBAH flowstone, lacking a hiathus at 197 kyr [6], the sea level was lower than 12 m b.s.l., in agreement with the global sea level [88]. The duration of the sea level highstand was between 201.5 kyr and 198.7 kyr [18]; MIS 7.2: The highstand is represented in the Argentarola cave. The sea level stillstand between 18.5 m b.s.l. and 21 m b.s.l. is quite similar to the global curve that reaches 27 m b.s.l.; MIS 7.3: Based on the results from the stalagmites of the Argentarola cave (Fig. 13), the sea level was slighty lower than 18 m b.s.l. In fact, speleothems record the presence of the sea at 21.3 m and 18.5 m b.s.l. Therefore, resuts from Mediterranean speleothems reduce the elevation of the MIS 7.3 highstand, reconstructed from the global sea level curve [88]. However, the DWBAH flowstone shows a hiathus during MIS 7.3, which seems to be older than the marine layers from the Argentarola stalagmites. The duration of the sea level highstand was between 217.2 kyr and 201.6 kyr [18]; MIS 7.5: The results from the stalagmites of the Argentarola cave and DWBAH flowstone indicate that the sea level was between 18 m and 12 m b.s.l., in agreement again with the global sea level curve [88]. The duration of the sea level highstand was between 248.9 kyr and 231.0 kyr [18]. Some GIA models have been published for the Mediterranean basin [35,36,38]. The latter two models were tested in the field with the comparison of coastal fossil deposits aged MIS 5.5 and fossil tidal notches of the same age. The maximum reported differences for the entire Mediterranean basin for MIS 5.5 are less than 2.5 m in elevation. This value is consistent with the predicted GIA-driven RSL variability within the Mediterranean basin during the MIS 5.5.
Here we show, for the first time, that the regional RSL variability increases during Interstadials MIS 5.1-5.3 and Glacials (MIS 2), most likely as a consequence of the ice-loading-induced spatial RSL gradients. However, because the GIA signal also depends on Earth rheology, this further complicates the search for the eustatic constraints in the Mediterranean Basin.
Overall, we argue that the observations call for a revised eustatic (i.e. ice sheets volume) during MIS 5.1-5.3 and MIS 7.1-3. In particular, we speculate that a reduction of ice sheets volume, hence an increase of the eustatic, is needed during MIS 5. 1-5.3. This is consistent with the recent findings of [99], which are based on a novel ice sheet modelling technique.