The onset of the Messinian salinity crisis in the deep Eastern Mediterranean basin

Astronomical tuning of the Messinian pre‐salt succession in the Levant Basin allows for the first time the reconstruction of a detailed chronology of the Messinian salinity crisis (MSC) events in deep setting and their correlation with marginal records that supports the CIESM ( ) 3‐stage model. Our main conclusions are (1) MSC events were synchronous across marginal and deep basins, (2) MSC onset in deep basins occurred at 5.97 Ma, (3) only foraminifera‐barren, evaporite‐free shales accumulated in deep settings between 5.97 and 5.60 Ma, (4) deep evaporites (anhydrite and halite) deposition started later, at 5.60 Ma and (5) new and published 87Sr/86Sr data indicate that during all stages, evaporites precipitated from the same water body in all the Mediterranean sub‐basins. The wide synchrony of events and 87Sr/86Sr homogeneity implies inter‐sub‐basin connection during the whole MSC and is not compatible with large sea‐level fall and desiccation of the Mediterranean.

For a long time, the MSC onset has been placed at the base of the evaporites, a criterion which is unsuitable, because evaporite precipitation occurred diachronously (CIESM, 2008;Dela Pierre et al., 2011;Manzi et al., 2007;Manzi et al., 2016). In shallow-water settings, primary gypsum is laterally replaced by carbonate or shale; in deeper, poorly oxygenated, settings only organic-rich shale and/or dolomitic limestone accumulated (Lugli, Manzi, Roveri, & Schreiber, 2010). Thus, the onset of the MSC does not necessarily coincide with the onset of evaporite deposition (Roveri, Flecker, et al., 2014;Roveri, et al., 2016).
The deepest record of Messinian events is much less known as it is preserved in deep offshore areas and consequently it is only based on geophysical data. The timing and the origin of this deep record are still debated (Roveri, Flecker, et al., 2014). In the Western Mediterranean, the deep evaporite unit is described with a typical threefold organization: a "flowing" salt body sandwiched between two bedded units (Hs€ u et al., 1973;Lofi et al., 2011;Roveri, Flecker, et al., 2014). In the Levant basin, the MSC deposits form a wedge-shaped unit (Bertoni & Cartwright, 2006;Gvirtzman, Reshef, Buch-Leviatan, & Ben-Avraham, 2013;Gvirtzman et al., 2017) (Figure 1) including mainly halite and subordinate terrigenous deposits (Feng, Ynkelzon, Steingberg, & Reshef, 2016) and showing: (1) a gradual basinward thickening from few tens of metres at the eastern margin up to about 1800 m in the deeper portions ; (2) a distinctive internal organization into six seismic sub-units (Gvirtzman et al., 2017); (3) an irregular erosional basal surface ("N reflector";Hs€ u et al., 1973) associated with channelized features (Bertoni & Cartwright, 2006;Gvirtzman et al., 2013) which can be traced upslope into the canyons cut on the basin margin ; (4) a top truncation surface ("M reflector";Hs€ u et al., 1973) possibly originated by the dilution of the Mediterranean water body at the end of stage 2 sealed by a thin (below seismic resolution) anhydrite unit (unit 7; Gvirtzman et al., 2017).
No stratigraphic constraints are available for these deep offshore Mediterranean successions, especially for their base. Consequently, onshore-offshore stratigraphic correlations remain highly speculative, thus hampering a full understanding of the MSC events.

| RESULTS
The unique opportunity to analyse and to date a pre-MSC deepwater succession comes from the study of four industrial boreholes in the deep Levant basin (offshore Israel; Figure 1 of the Messinian salt unit. We carried out an integrated stratigraphic study based on well logs, high-resolution seismic data, biostratigraphy (foraminifera and calcareous nannofossils), and geochemistry ( 87 Sr/ 86 Sr).

Figures 2 and 3) immediately below the evaporites. This interval cor-
responds to the Non-Distinctive Zone (NDZ) marking the MSC onset in onshore settings Gennari et al., 2018;Iaccarino et al., 2007;Manzi et al., 2013). More detailed analyses were carried out on Aphrodite-2 borehole, where this interval is thicker.
This age definition is further confirmed by the distribution of calcareous nannofossil (Figure 2; Figure S1, Table S2). In Aphrodite-2, we recognized a prominent peak of Sphenolitus abies at 3961 m (Figure S1), closely followed by a decrease of the number of species of calcareous nannofossil at 3958 m ( Figure 2; Figure S1).
Between the depth of 3996 m and the evaporites base, we found 33 cycles, which is the expected number of precessional cycles between bioevents 3 and 11, roughly corresponding to a 700ka interval. Moreover, the GR log shows eight intervals with a smaller variability (marked by arrows in Figure 2) separating longer (4-5 cycles) intervals with larger GR variability. This pattern is commonly considered as the result of the eccentricity (red curve in Figure 2) interference on the precession signal (Ochoa et al., 2015;Sierro, Ledesma, Flores, Torrescusa, & Martinez del Olmo, 2000). The low GR variability intervals match the expected position of eccentricity minima based on our biostratigraphic results, supporting an astronomical origin of the lithological cyclicity inferred from the GR-RES log. Consequently, we anchored the interval with low GR variability with the eight eccentricity minima and tuned the GR curve with the insolation curve; in this way, we were able to define a high-resolution age calibration of the succession below the salt. The calculated sedimentation rate of~90 mm/ka is in good agreement with those of other pre-MSC peri-Mediterranean successions ( Figure S2). The highest occurrence (HO) of foraminifera, 6 m above the HO of T. multiloba, corresponds to the em5 eccentricity minimum and marks the MSC onset in the onshore successions  at 5.971 Ma.

| Spectral analysis
In order to support our cyclostratigraphic interpretation, we performed a spectral analysis on the age-constrained pre-MSC GR record of Aphrodite-2 borehole (3,997-3,958 m depth interval) and found a dominant frequency peak of 2.46 m (Figure 4b) F I G U R E 4 Spectral analyses on the gamma ray (GR) record of the Aphrodite 2 borehole. a) Age model and time constraints used for the tuning of the gamma ray curve. b) Power spectrum and evolutive spectrum of the GR in depth domain. Notice the good persistence of the peak with a period of 2.46 m. c) Results of spectral analysis: in the upper part, the power spectra for GR (orange) and insolation 65°N (grey) for the investigated time window; in the lower part, the coherence calculated after the spectral analysis. Non-zero coherence is higher than: 0.38413 (80%); 0.477452 (90%); 0.550595 (95%). The precession-related periodicities (19 and 23 Ka) have a very high coherence, whereas the obliquity-related ones have a slightly lower coherence [Colour figure can be viewed at wileyonlinelibrary.com] corresponding to~28 ka; this is partially overlapped by two less prominent peaks at 3.2 and 1.89 m corresponding to~36 and 21 ka respectively.
The cross spectral analysis of GR and insolation (Figure 4c) shows high coherence peaks (over 95% and around 90%) linked to precession, at 19.3 ka and 24.8 ka. Obliquity is expressed by a single peak, at 43.1 ka, showing coherence lower than precession. These results suggest a good control of the precession component with a minor control of the obliquity signal. The 27-m-thick foraminiferabarren interval just below the evaporites contains 16 GR-based lithological cycles. Assuming a precessional origin for these cycles, duration of~350 ka can be estimated for the FBI.

| 87 Sr/ 86 Sr isotopic analysis
In Aphrodite-2, the FBI is overlain by a thin (5.5 m) anhydrite/shale unit (unit 0) and, in turn, by very thick (up to 1,800 m) salt body separable into six seismic sub-units, which are tilted basinward and trun-   Figure S4) shows some 87 Sr/ 86 Sr values higher than the global ocean field, but is mainly characterized by values typical of stages 1 and 2 (e.g. 0.7088-0.70906; Gvirtzman et al., 2017); values <0.7088, which are typical of stage 3, were not found.

| DISCUSSION
The data from Aphrodite-2 indicate that the FBI found below the evaporites could record the whole MSC stage 1 and that no significant hiatus is present below the evaporites in this area. Moving eastward towards the Israeli margin the base of the evaporites becomes progressively erosive and can be traced within the Ashdod canyon, where clastic evaporites have been found . It follows that the base of the evaporites can be identified as the Messinian erosional surface (MES) in the margins and by its correlative conformity (MES-cc) in the deep part of the basin. This situation is similar to what documented in the Northern Apennines (Manzi et al., 2007) where a 60-m-thick organic-rich barren shale unit, that has been identified as the deep-water counterpart of the PLG (Lugli et al., 2010;Manzi et al., 2007) of stage 1, is overlain by a unit made of clastic evaporites (Manzi, Lugli, Ricci Lucchi, & Roveri, 2005),  Lugli et al., 2013;G17, Gvirtzman et al., 2017). The main salt unit show a signature typical of the stage 1 + 2 of the MSC and are in agreement with those obtained from the clastic evaporite onshore (Ashdod-2; Lugli et al., 2013). This suggests that the halite precipitated from a water body connected with the Western Mediterranean and with the global ocean. It is worth noting that Sr values obtained from unit 0 to unit 5 are markedly different from those obtained from the stage 3 evaporites of unit 7 (Gvirtzman et al., 2017) [Colour figure can be viewed at wileyonlinelibrary.com] foraminifera-barren deposits in relatively deep settings, where PLG did not accumulate (Manzi et al., 2007;Manzi et al., 2011;Roveri et al., 2016); in Sicily, Calabria and Cyprus, these deposits are associated with halite (CIESM, 2008;Manzi et al., 2016).
The Sr isotope stratigraphic significance for the salinity crisis interval has been assessed in various papers for sulphates and carbonates (Flecker, de Villiers, & Ellam, 2002;M€ uller & Mueller, 1991;Roveri, Flecker, et al., 2014 and references therein). The Sr isotope composition presented here for halite are in the range of stage 1 and 2, but show some anomalous values higher than the global ocean curve (Figure 6). Compared to sulphates, halite may incorporate an extremely lower proportion of Sr and may show a depositional rate up to one order of magnitude higher. It follows that halite isotope composition may be very sensitive for local diverse, short-term Sr input (Data S1). Despite the anomalous values, the deposition of the halite body during stage 1 or 2 is supported by the range of values obtained for the sulphates found in the salt body, which are within the field of stage 1 or 2 ( Figure 6; Table S5).
The recognition below the salt in Aphrodite-2 of the FBI which is recording the entire stage 1 implies that the main halite body (Units 1-6) was not precipitated during stage 1.
On the other hand, the clastic-rich evaporites of Unit 7, which is capping the salt body, yielded Sr isotope values typical of stage 3 (Or-South-1 borehole, Gvirtzman et al., 2017) and is overlain in the Israeli margin by an evaporite-free unit containing Lagomare fossil assemblages (Derin, 2000).
The inescapable conclusion is that the salt unit, being sandwiched between the FBI (stage 1) and Unit 7 (stage 3), must have been accumulated during stage 2. This interpretation is also supported by the recovery of Discoaster quinqueramus ( Figure S1; Table S2) within unit 5 of the halite complex, which went extinct towards the end of stage 2.
Further considerations on the duration of halite deposition in the Levan Basin can be deduced from seismic facies and well logs, both showing alternation from nearly pure halite units (seismically transparent) to well bedded units (reflection-rich) containing thin layers of clays Feng et al., 2016; Figure S4). According to  and Manzi et al. (2016), we suggest that these seismic facies may reflect precessional-controlled alternations of relatively arid/humid climate marked by low/high terrigenous supplies.
The 87 Sr/ 86 Sr data suggest that Levant basin was not isolated from the Global Ocean (McArthur, Howarth, & Shields, 2012) before, F I G U R E 6 Sr isotope data from the Levant basin (this work) compared to the other Mediterranean areas  implemented with published data from the Levant basin (L13, Lugli et al., 2013;G17, Gvirtzman et al., 2017) [Colour figure can be viewed at wileyonlinelibrary.com] during or after the deposition of the main halite unit (Gvirtzman et al., 2017); thus, implying the persistence of the Mediterranean Sea level at a relatively high-stand conditions (at least higher than intra-basinal sills) in order to allow the Atlantic inflow to reach the Eastern Mediterranean.
It follows that our data do not support the hypothesis of a complete desiccation of the Mediterranean Sea during the salinity crisis (Hs€ u et al., 1973).

| CONCLUSIONS
The MSC successions in the deep Eastern Mediterranean are characterized by the following features:  (Roveri, Flecker, et al., 2014;Roveri et al., 2016). This suggests that all the Mediterranean sub-basins, regardless of their water depth, remained hydrologically connected also during the acme of the crisis. An obvious implication is that the usually envisaged high-amplitude sea-level oscillations and the desiccation of the Mediterranean Sea are not supported by these data, thus suggesting that alternative scenarios of the MSC are possible Roveri et al., 2016). Data S1. Description of the different methodologies adopted for this work. Figure S1. Aphrodite-2 borehole calcareous nannofossils distribution.