Glacioeustatic control on the origin and cessation of the Messinian salinity crisis

The desiccation of the Mediterranean during the Messinian salinity crisis (MSC) is one of the most intriguing geological events of recent Earth history. However, the timing of its onset and end, as well as the mechanisms involved remain controversial. We present a novel approach to these questions by examining the MSC from the AC C EP TE D M AN U SC R IP T ACCEPTED MANUSCRIPT 2 Atlantic, but close to the Gibraltar Arc, analysing the complete Messinian record of the Montemayor-1 core of the Guadalquivir Basin (SW Spain). Flexural backstripping analysis shows a tectonic uplift trend that would have reduced the depth of the Rifian Corridors considerably. Nonetheless, the rate of tectonic uplift was insufficient to account for the close up of the corridors. At 5.97 Ma, a global cooling and associated glacioeustatic sea-level drop, estimated in 60 m, is observed. This would have been sufficient to restrict the Rifian corridors and to trigger the MSC. The later flooding of the Mediterranean occurred during a sea-level rise associated with global warming during a stable tectonic period. We postulate a two-step flooding event: 1) A glacioeustatic sea-level rise during interglacial stage TG 11 (5.52 Ma) led to subtropical Atlantic waters entering the west-central Mediterranean through pathways south of the Gibraltar Strait, probably the Rifian corridors. 2) A global sea-level drop at 5.4 Ma, that might have favoured intensification of regressive fluvial erosion in the Gibraltar threshold, along with the subsequent global sea-level rise would have generated the Gibraltar Strait leading to complete Mediterranean refilling during the earliest Pliocene.


Introduction
The Messinian salinity crisis (MSC) in the Mediterranean has attracted the interest of Earth scientists ever since the seminal publication by Hsü et al. (1973). The general consensus is that the Mediterranean was isolated from the Atlantic due to the closure of the Betic and the Rifian corridors and that this led to deposition of the Lower and Upper Evaporites. There are still key issues under intense debate. One of these concerns the main mechanisms that led to the isolation of the Mediterranean. Global glacioeustatic sea level lowering, regional tectonic uplift in the Gibraltar area, or a combination of both processes have been invoked as trigger mechanisms (Weijermars,  Furthermore, the orbital obliquity and eccentricity were constructed using the Laskar orbital solutions La2004 and La2010 respectively (Laskar et al., 2004(Laskar et al., , 2011 following the recommendations of Laskar et al. (2011). To generate the ETP (eccentricity, obliquity, precession) curve the normalised eccentricity La2010 orbital solution, the normalised obliquity La2004 orbital solution and the negative normalised precession La2004 orbital solution were summed.
In addition, quantitative sea-level changes were estimated using a transfer function based on benthic foraminifera developed by Hohenegger (2005)  Global sea-level curves of Hardenbol el al. (1998) and Miller et al. (2005), as well as the regional sea-level curve of 4th order eustatic cycles of Esteban et al. (1996) were used for comparison with the sea-level curve of the Montemayor-1 core.
Flexural backstripping analysis was carried out following the methodology proposed by Allen and Allen (1990). This method is used to reconstruct the vertical movement of the basin floor. Sediments are flexurally unloaded from the basement by applying different corrections for the paleobathymetric changes, compaction, loading effects of the sediments and eustatic sea-level fluctuations (Watts, 1988). According to the formula of Allen and Allen (1990), the vertical movement of the basement (Z) is: where σ m is mantle density (3,300 kg×m -3 ), σ s is sediment density (1,400 kg×m -3 , average density for pelagic clay in marine environments (Hamilton, 1970;Orsi, 1991)), σ w is water density (1,000 kg×m -3 ), Wdi is paleobathymetry of the Montemayor-1 core estimated with the transfer function, ΔSli is the eustatic sea-level change derived from the curve of Miller et al. (2005), S is the original thickness before compaction, and Ç is the isostatic compensation function.
The isostatic compensation function (Ç) was calculated with the following equation: where g is gravitational acceleration (9.81 m×s -2 ), D is flexural rigidity (4.80×10 22 N×m for a effective elastic thickness (Te) of 20 km in the Betic Cordillera (van der Beek and Cloetingh, 1992)), π is the mathematical constant, and λ is the wavelength of the periodic load which is 90 km for the western sector of the Guadalquivir Basin.
The original thickness before compaction (S) was estimated calculating the compaction of clays since the Montemayor-1 core consists of mostly bluish-greenish clays from the Arcillas de Gibraleón Formation (Fig. 2). For that purpose, the compaction of clays was calculated using the formula of Einsele (1992): where h sl is the original thickness before compaction, h sp is the compacted thickness, n l is the original mean porosity, and n p is the final mean porosity after compaction. An average n l of 80%, and n p of 20% for clays was used (Leeder, 1982;Velde, 1996;Boggs, 2009). Using this formula, the compaction of the Montemayor-1 core is 75%.

Results
The paleodepth curve derived from the transfer function based on benthic foraminifera shows a long-term decreasing trend with 3 significant sea-level drops ( The benthic oxygen isotope record has a fluctuating trend and shows a gradual increase from 6.17 to 5.

Onset of the MSC
The Pacific oceans (Shackleton et al., 1995, Shackleton andHall, 1997;Vidal et al., 2002) ( Fig. 4). All these evidences point to a glacioeustatic origin for the sea-level drop related to the onset of the MSC. This cooling was probably associated with ice-sheet expansions on western Antarctica and the Arctic taking place during the late Miocene (Zachos et al., 2001). Furthermore, it could be estimated quantitatively assuming the in an anomalously high tectonic uplift rates. Therefore, the tectonic uplifting pulses at 5.99 and 5.76 Ma appear not reliable. Considering these two pulses as outliers, the resulting average uplifting rate is 0.6 mm/yr, well within the ranges estimated for the Late Neogene uplift of the Betic Cordillera: from 0.2 to 0.7 mm/yr (Weijermars et al., 1985;Braga et al., 2003). Mediterranean outflow in order to produce kilometre-thick evaporites (Meijer and Krijgsman, 2005;Meijer, 2006;Krijgsman and Meijer, 2008). This has been also proven by geochemical data that suggest a marine origin for the Upper Evaporites in the Sorbas and Níjar basins (Playà et al., 1997;Lu et al., 2001Lu et al., , 2002. Moreover, fully marine post-evaporitic sediments were deposited in marginal basins of SE Spain during the late Messinian as consequence of a sea-level rise during the late Messinian (Riding et al., 1998;Aguirre and Sánchez-Almazo, 2004;Braga et al., 2006). These deposits contain a plethora of fully marine fauna, including hermatypic corals, coralline algae, echinoderms, demosponges, fish, bryozoans, foraminifera and bivalves, that inhabited the western and central Mediterranean during the late Messinian (Hsü et al., 1977(Hsü et al., , 1978Cita et al., 1978;Riding et al., 1991;Martín et al., 1993;Martín and Braga, 1994; A C C E P T E D M A N U S C R I P T
The same trend is recorded in the Rifian Corridors, Atlantic and Pacific oceans (Shackleton et al., 1995;Vidal et al., 2002;van der Laan et al., 2006) (Fig. 4), and is related to a period of global warming that persisted until the mid Pliocene (Vidal et al., 2002). ). In addition, global sea level rise during the TG 12-TG 11 transition is ~75 m (Miller et al., 2005) (Fig. 4).
Backstripping analysis shows a tectonic stable period during the late Messinian ( Fig. 5), suggesting that tectonism in the vicinity of the Gibraltar Arc was insignificant.
Therefore, we propose that the glacioeustatic sea level rise at TG 11 (5.52 Ma) could have triggered the first reflooding step of the western and central Mediterranean shortly before the Miocene/Pliocene boundary (Fig. 6). This agrees with the view that a small connection to the Atlantic was sufficient to quickly reinundate the Mediterranean (Meijer and Krijgsman, 2005). This conclusion is also corroborated by the presence of late Messinian marine deposits in western and central Mediterranean marginal basins (Martín and Braga, 1994;Riding et al., 1998;Braga et al., 2006;Carnevale et al., 2006aCarnevale et al., , 2006bCarnevale et al., , 2008Bourillot et al., 2010aBourillot et al., , 2010b.

A C C E P T E D M A N U S C R I P T
It is generally assumed that Atlantic waters flooded into the Mediterranean through the Gibraltar Strait (Blanc, 2002;Meijer and Krijgsman, 2005;Estrada et al., 2011;Bache et al., 2012). In such a case, cold Atlantic waters would enter the Mediterranean (Martín et al., 2010). However, photozoan carbonates developed at the uppermost Messinian deposits in SE Spain (Martín and Braga, 1994;Braga et al., 2006;Martín et al., 2010). These data indicate the inflow of warm subtropical Atlantic waters into the Mediterranean during the first reflooding step from areas south of the Gibraltar Strait, probably through Rifian Corridors, promoting the formation of coral reefs in SE Spain (Martín et al., 2010). Messinian canyons excavated at NW Morocco shelf might have accounted for this first inundation (Loget and van den Driessche, 2006).
Furthermore, the Rifian Corridors were not completely closed during the MSC because a continuous inflow of Atlantic water is crucial to develop kilometre-thick evaporite deposits formed in the Mediterranean basin during the MSC (Meijer and Krijgsman, 2005;Meijer, 2006;Krijgsman and Meijer, 2008). Hence, a reflooding via the Rifian Corridors during the global sea-level rise related to TG 11 is very likely.
In the Sorbas Basin (SE Spain), marine post-evaporitic deposits are overlain by continental sediments of the Zorreras Member (Martín and Braga, 1994;Martín-Suárez et al., 2000). This indicates a sea-level drop at the very end of the Messinian (Martín and Braga, 1994;Martín and Braga, 1996). Mediterranean are improbable since they are not concurrent with the available evidence.

Conclusions
Our      from Site 846 in the Pacific Ocean (Shackleton et al., 1995). D: Global sea-level curve (Miller et al., 2005)   and obliquity curve (grey) from La2004 orbital solution versus age (based on Jiménez-