The Messinian salinity crisis: open problems and possible implications for Mediterranean petroleum systems

A general agreement on what actually happened during the Messinian salinity crisis (MSC) has been reached in the minds of most geologists but, in the deepest settings of the Mediterranean Basin, the picture is still far from being finalized and several different scenarios for the crisis have been proposed, with different significant implications for hydrocarbon exploration. The currently accepted MSC paradigm of the ‘shallow-water deep-basin’ model, which implies high-amplitude sea-level oscillations (> 1500 m) of the Mediterranean up to its desiccation, is usually considered as fact. As a consequence, it is on this model that the implications of the MSC events on the Mediterranean petroleum systems are commonly based. In fact, an alternative, deep-water, non-desiccated scenario of the MSC is possible: it (i) implies the permanence of a large water body in the Mediterranean throughout the entire Messinian salinity crisis, but with strongly reduced Atlantic connections; and (ii) envisages a genetic link between Messinian erosion of the Mediterranean margins and deep brine development. In this work, we focus on the strong implications of an assessment of the petroleum systems of the Mediterranean and adjoining areas (e.g. the Black Sea Basin) that can be based on such a non-desiccated MSC scenario. In particular, the near-full basin model delivers a more realistic definition of Messinian source-rock generation and distribution, as well as of the magnitude of water-unloading processes and their effects on hydrocarbon accumulation.

The term 'Messinian salinity crisis' (MSC) refers to the largest and geologically most-rapid set of high-amplitude environmental changes undergone by the peri-Mediterranean area during the Neogene and, possibly, the entire Phanerozoic. The sedimentary record of this event involves complex feedback between geodynamics, climate and biota, and resulted in a stratigraphy that left an indelible signature in the post-Messinian evolution of the Mediterranean Basin, in addition to important implications for hydrocarbon exploration. Up to now, a general agreement on what actually happened during the MSC, particularly in the deepest settings of the Mediterranean Basin, is still far from clear: consequently, several different scenarios of the crisis are available (see Roveri et al. 2014a).
This lack of consensus is mainly due to the difficulty in establishing a general, comprehensive, high-resolution stratigraphic framework for the upper Messinian. In fact, in this interval, owing to the lack of fossils and being fully included in the C3r chron, the classical biomagnetostratigraphic tools cannot be used (Hilgen et al. 2007;Roveri et al. 2014a). Furthermore, most data come from onshore successions, which formed in shallow-(water depth of 0-200 m) or intermediate-depth (water depth of 200-1000 m) subbasins, while the deepest Messinian settings, where the largest volume of MSC products accumulated, are buried below the present-day Mediterranean abyssal plains. These deep deposits are virtually unknown owing to the difficulties (both technical and economic) in obtaining data from scientific drillings or in accessing industry data. Moreover, since onshore and deep offshore Messinian successions are physically disconnected, a synthesis and common view of the MSC remain very difficult to obtain (Lofi et al. 2011;Roveri et al. 2014a, b).
As a consequence, owing to the need for additional deep basin data, all the different scenarios so far proposed should be considered as theories that need to be proven. However, the 'shallow-water deep-basin' model (SWDB: Hsü et al. 1973), with its highamplitude sea-level oscillations (>1500 m) up to its desiccation, is the current MSC paradigm (Roveri & Manzi 2006;Roveri et al. 2014c). This model is usually considered as a fact, with obvious implications in many related fields including hydrocarbon exploration, but such a view could lead to possible misinterpretation. This model has undergone some modifications through time, all implying that, at a certain point, the Mediterranean desiccated almost completely and its slopes underwent a phase of subaerial exposure and vigorous erosion related to the rejuvenation of an entire fluvial drainage system (Lofi et al. 2005;Ryan 2009;Bache et al. 2012). This phase of generalized exposure would have led to the formation of an erosional surface (the Messinian erosional surface -MES), which is one of the main stratigraphic features in both onshore and offshore records, and a key one for their correlation. The rapid water loading/unloading events would have caused significant pressure release and catastrophic fluid-expulsion phenomena (Ryan & Cita 1978;Bertoni et al. 2013;Sacleux et al. 2013;Bertoni & Cartwright 2015) that would have had a great impact on the pre-existing hydrocarbon migration and preservation.
We think that an alternative scenario that implies the permanence of a large and deep-water body connected with the Atlantic Ocean throughout the MSC (Schmalz 1969;Roveri et al. 2014c) is not only possible, but even more likely. In this paper, we also discuss the more general implications of our new scenario for petroleum systems.

An alternative scenario: stratigraphic framework
Our scenario is based on a recently established chronology of the main MSC events mainly built on onshore data Hilgen et al. 2007;Manzi et al. 2013) that includes both outcrop and subsurface data. A major consensus has been reached on this stratigraphic framework, which includes three evolutionary stages (Clauzon et al. 1996;CIESM 2008;Roveri et al. 2014a) (Fig. 1), each of them characterized by a particular evaporite association recording significant hydrological changes in the Mediterranean Basin. The latter are well documented by the 87 Sr/ 86 Sr Mediterranean curve (Fig. 1), which shows a significant, stepwise detachment from the global ocean curve during the MSC, suggesting a progressive hydrological isolation and/or an increase in the relative proportion of continental waters over that of the ocean ones (see Flecker et al. 2002;Roveri et al. 2014b). Each one of the three stages of the crisis shows a distinct range of 87 Sr/ 86 Sr values: > 0.708900 for stage 1; between 0.708800 and 0.708900 for stage 2; and < 0.708800 for stage 3. In the following section, we briefly summarize the main characteristics of each MSC stage.
MSC onset and stage 1 (5.97-5.60 Ma) The onset of the MSC occurred synchronously at 5.97 Ma ) (i.e. well after the base of the Messinian stage: 7.246 Ma), following a long phase of progressive reduction in Atlantic connections and the consequent restriction in Mediterranean circulation and water-column stratification witnessed by the widespread cyclical deposition in deep-marine settings of organic and opal-rich sediments (the pre-MSC stage, e.g. the Tripoli Formation of Sicily: . The onset of the crisis is not necessarily coincident with the base of the lowermost evaporite bed, as sometimes erroneously envisaged in the literature (e.g. see Ochoa et al. 2015), but with a dramatic decrease in the normal marine biota followed by their disappearance (Manzi et al. 2007(Manzi et al. , 2015. In fact, while the biological record of the onset of the crisis is synchronous throughout the Mediterranean Basin and at any depth, the onset of the bottom-grown evaporites of stage 1 (selenite gypsum of the Primary Lower Gypsum unit -PLG) is diachronous (Roveri et al. 2014a;Manzi et al. 2016). PLG evaporites started to form from about 5.97 Ma, but only in shallow-water, semiclosed, silled sub-basins developed along the Mediterranean continental margins ): moving to a deeper setting, however, the onset of the PLG is progressively younger Dela Pierre et al. 2011;Roveri et al. 2014a). The water depth limiting the deposition of the bottom-grown gypsum (<200 m, including areas beyond the shelf break) is suggested by the common occurrence of photosynthetic microorganism communities trapped within primary gypsum crystals (mainly cyanobacteria: Panieri et al. 2010). MSC onshore records clearly show that in deeper and/or unsilled sub-basins, evaporite-free deposits accumulated, mainly consisting of organic-rich shales and dolostones barren of normal marine fossils (Manzi et al. 2007;Lugli et al. 2010;Dela Pierre et al. 2011;Ghielmi et al. 2013;Rossi et al. 2015).
Evaporite deposition was modulated by precession-controlled climatic oscillations inducing changes in the Mediterranean hydrological budget (Vai 1997;Krijgsman et al. 1999;Van der Laan et al. 2005;Hilgen et al. 2007). Up to 16 gypsum-shale couplets recording dry-wet precessional cycles formed in stage 1, allowing the end of this phase to be dated at 5.60 Ma Roveri et al. 2014a). The lithology of these cycles shows an impressive similarity

284
M. Roveri et al. in terms of types of gypsum sedimentary facies, stacking patterns and overall trend, permitting pan-Mediterranean bed-by-bed correlation . The gypsum in these beds formed subaqueously: each precessional evaporitic cycle is characterized by a facies sequence recording a progressive increase in brine saturation, followed by a phase of relative dilution (Roveri et al. 2008a). It is worth noting that evidence of subaerial exposure and/or erosion are not observed within these cycles, but only at the top of the PLG unit. This PLG unit may locally consist of less than 16 gypsum cycles, owing to the absence of the basal members (replaced by their laterally equivalent evaporite-free deposits: Manzi et al. 2007;Dela Pierre et al. 2011;Gennari et al. 2013) or because of erosion and resedimentation during the subsequent stage 2 (Roveri et al. 2001(Roveri et al. , 2014aManzi et al. 2005).
Unlike what is claimed by some authors (e.g. Ochoa et al. 2015), where outcrop observations and complete subsurface data (seismics and boreholes) are available, it has been documented that the PLG evaporites are absent in deep-water and/or unsilled settings during MSC stage 1 (Manzi et al. 2007;Lugli et al. 2010;Dela Pierre et al. 2011;Ghielmi et al. 2013;Rossi et al. 2015). Two different models have been proposed so far to explain this fact. Lugli et al. (2010) suggested that bottom-grown gypsum only developed in shallow (<200 m), silled sub-basins acting as bottom brine traps; and De Lange &  suggested that a sill is not necessary and that the main controlling factor is the rate of sulphate consumption due to degradation of organic matter, which in deep water (below 200 m) would be greater than the supply rate of sulphate, thus hampering gypsum precipitation and preservation. Both models recognize the absence of primary evaporites of the first stage in deepwater settings, and this fact is a fundamental observation for the correlation of deposits from shallow to deeper parts of the basin.
Stage 2 (5.60-5.54 Ma) The second stage of the crisis is characterized by a period of strong erosion of the Mediterranean continental margins (MES) and by the concurrent deposition of huge volumes of highly soluble primary evaporites (halite and K-Mg salts), as well as by resedimented PLG evaporites (i.e. as a clastic facies) in deeper sub-basins (Apennines, Sicily, Calabria, Tuscany, Cyprus). The resulting unit observed in onshore successions has been named Resedimented Lower Gypsum (RLG), and shows very rapid and significant lateral changes in terms of lithology and thickness, which is also related to tectonic activity affecting several Mediterranean areas in this stage. The clastic component of the RLG unit mainly consists of gypsum turbidites, giant PLG olistoliths (Roveri et al. 2001(Roveri et al. , 2008bManzi et al. 2005) and microbially derived brecciated limestones (i.e. the Calcare di Base of Sicily: Manzi et al. 2011). Locally, the RLG unit may mainly consist of terrigenous sediments (i.e. turbiditic sandstones of the Apennine foreland system depocentresthe Laga pro parte and Fusignano formations; Roveri et al. 2001;Manzi et al. 2005;Rossi et al. 2015). This stage, which is considered the acme of the crisis, encompasses a very short time window, according to cyclostratigraphic considerations based on stage 1 (Roveri & Manzi 2006) and stage 3 cyclic patterns (Manzi et al. 2009). Thus, the RLG unit would end at 5.54 Ma, spanning no more than 60 ka.
Subaerial erosion of stage 1 evaporites (PLG) is commonly observed in onshore successions, suggesting a relative base-level fall, the amplitude of which, however, cannot be clearly defined (see Lugli et al. 2013Lugli et al. , 2015Roveri et al. 2014c). Onshore, the MES can be traced downbasin in deeper settings at the base of the RLG unit (Manzi et al. 2005(Manzi et al. , 2007. It is worth noting that, in such settings, the deep-water equivalent of the PLG evaporites does not show any evidence of subaerial exposure; in some places, these deposits are eroded at the top and only partially preserved, thus suggesting subaqueous erosional processes. Stage 3 (5. 54-5.33 Ma) The last stage of the MSC is probably the most enigmatic phase. Onshore successions consist of both shallow and relatively deeper water deposits. Sr isotope data (<0.708800) and fossils (mollusks and ostracods) suggest that surface waters underwent a significant dilution, with the development of brackish environments throughout the Mediterranean. Deeper successions are usually barren of fossils, thus hampering palaeoenvironmental reconstruction. Despite this general signal of more diluted waters, primary evaporites (sulphates) are also formed in this stage, but only in the southern and easternmost sectors of the Mediterranean (e.g. Sicily, Calabria, Cyprus). These evaporites, named the Upper Gypsum (UG), bear some lithological similarities with stage 1 PLG evaporites, but can be easily distinguished based on their facies characteristics and, particularly, their Sr isotope values (Manzi et al. 2009). Like the PLG, the UG unit has a well-developed cyclical pattern induced by precession, which allows its accurate chronostratigraphic calibration. Stage 3 can be subdivided into substages 3.1 and 3.2, based on the sudden increase in terrigenous sediment input at around 5.42 Ma, especially in the northern and western Mediterranean sectors. Substage 3.2 is also characterized by the greatest development in and the diffusion of the inclusion of Lagomare faunal assemblages, which have been classically considered to derive from the Paratethys (e.g. Orszag-Sperber 2006; Roveri et al. 2008a). In this low-salinity environment, some evidence of the permanence of the Atlantic connections is given by the occurrence of marine fish (Carnevale et al. 2008) and alkenons (Mezger 2012). The return to normal marine conditions is sudden and marks the base of the Zanclean at 5.33 Ma, and is usually interpreted as being related to a catastrophic re-opening of the Atlantic connections.

Seismic and well log expression of evaporitic units
When its 16 lithological cycles are largely preserved, the PLG unit may attain a total thickness ranging between 100 and 300 m , typically around 150-200 m in the best outcrops of the Apennines, Sicily and southern Spain. This unit does not have a peculiar seismic facies allowing it to be distinguished from the evaporitic units of the other MSC stages, especially where only commercial, low-resolution seismic profiles are available. In this case, it appears as a thin seismic unit consisting of one or two parallel, high-amplitude reflectors, similar to the RLG and UG units, where the latter are thinner. However, the PLG unit can be well identified in well logs owing to its peculiar blocky pattern, as documented in many offshore and onshore boreholes Lugli et al. 2010;Rossi et al. 2015). Where high-resolution seismic profiles are available, PLG evaporites appear as a horizontal bedded unit (BU of Lofi et al. 2011) with a conformable base and an erosional top (e.g. the MES: see Maillard et al. 2014;Driussi et al. 2015); the erosional v. non-erosional character of the bounding surfaces is probably the most useful criteria for distinguishing it from other evaporite-bearing units (i.e. the suggested equivalents of RLG and UG units, the MU and UU units, respectively, of Lofi et al. 2011). RLG, in particular, is usually thicker (up to 2 km in the deepest Mediterranean basins) and mainly consists of halite, thus appearing as an acoustically transparent seismic unit. Locally, the RLG unit may include chaotic seismic facies related to slump and/or debris-flow deposits. Its base, the MES, is commonly unconformable at the margins, and becomes conformable in the centre of the basin.

The onshore lesson: clues for shallow to deep correlations
This three-stage Messinian stratigraphic framework is based on the onshore record of the MSC, but has a large potential for also being applied to deep offshore successions because of its robust physical stratigraphic architecture, constrained by key surfaces and time lines (e.g. the MES, Zanclean base), that can be easily recognized from 285 The Messinian salinity crisis: petroleum systems available seismic and borehole data, and which allows a sequence stratigraphic approach (Roveri et al. 2008c;Rossi et al. 2015). The three stages of the model are characterized by distinctive Sr isotope values that can be easily obtained by analysing the evaporite rocks and the fossils from cores or cuttings. A first attempt at suggesting an onshore-offshore correlation was provided by Roveri et al. (2014b) (Fig. 2), based on the recognition that onshore successions also include intermediate-depth (up to 1000 m) depocentres that show continuous subaqueous deposition. These relatively deep settings are the key stratigraphic link between the shallow-and the deep-basin records.
According to this correlation, a possibility exists that the largest part of the evaporitic deposits lying in the deepest basins could have formed during MSC stage 2. The evaporitic unit in the western Mediterranean and in the Ionian Basin is a tripartite seismic unit (the famous 'Messinian trilogy'): the three seismic units (LU, Lower Unit; MU, Mobile Unit; UU, Upper Unit: Lofi et al. 2011) have been classically considered to be the offshore equivalent, from the bottom to the top, of the Lower Gypsum (i.e. the PLG), of the Sicilian salt and of the Upper Gypsum. In the Levantine Basin, the evaporitic unit is not tripartite and only the MU is recognized (Lofi et al. 2011).
In fact, the MES, marking the boundary between stages 1 and 2, can be traced downbasin to deep offshore areas, where it corresponds to the basal surface of the deep canyons ) that possibly continued into the erosional features imaged at the base of the deep Levantine evaporites (Bertoni & Cartwright 2007). The MES also represents a pervasive erosional surface of the Mediterranean slopes that can be followed into the deeper basins (the basal erosional surface (BES): Lofi et al. 2011) and which progressively smooths out in the basin plain settings, becoming a correlative conformity (BS: Lofi et al. 2011) marking the base of the Lower Evaporites (Lower Unit: Lofi et al. 2005Lofi et al. , 2011. Because of these characteristics, the MSC stages 1 and 3 would be represented, respectively, by: (i) a thin, evaporite-free layer below the base of the Lower Evaporites; and (ii) a relatively thin unit, mostly below seismic resolution, composed of shales with minor evaporites lying above the Mobile Unit. As for the uppermost evaporites of the MU recovered from the DSDP-ODP cruises, Sr isotope signatures suggest that they still belong to MSC stage 2 (Roveri et al. 2014b).

An alternative scenario: erosion and deposition in a nondesiccated deep Mediterranean basin
The 'shallow-water deep-basin' model (SWDB) is mainly based on: (i) the interpretation of the erosional features of the margins as being due to mostly subaerial processes; and (2) the supposed shallowwater to subaerial nature of the deep evaporites. As for the second point, recent studies (Hardie & Lowenstein 2004;Lugli et al. 2015) have demonstrated that the evaporites at the top of the MU do not show evidence of subaerial exposure and could have precipitated at any water depth. As for the first point, it was possible to document that only the shallower parts of the onshore successions underwent subaerial erosion during stage 2, while the intermediate-depth depocentres experienced continuous subaqueous deposition throughout the crisis and/or subaqueous erosion mainly by gravity flows and related slope-failure processes.
Starting from these considerations, Roveri et al. (2014c) suggested a genetic link between the deposition of salt in the deepest basins and the erosion along the basin slopes due to the downslope flow of hypersaline, dense waters that led to the formation of deep-water brines. This process is similar to the present-day cascading of dense shelf waters along the Mediterranean margins (Canals et al. 2006); together with sediment gravity flows (i.e. turbidites and hyperpycnal fluvial floods), these processes work together to shape submarine slopes, and to cut gullies and canyons (Roveri et al. 2014c). We infer that the Messinian slopes were profoundly reshaped during the MSC, and particularly during stage 2, by forming new erosional features or by rejuvenating pre-existing ones, as documented along both the western (Lofi & Berné 2008) and eastern  Mediterranean margins. However, the MES was not generated exclusively by subaqueous processes, since a moderate relative sealevel fall, ranging in amplitude between 200 m (Roveri et al. 2014c) and 550 m (Rossi et al. 2015), also promoted the subaerial exposure and erosion of the basin margins. It follows that the MES is a polygenic erosional surface with both subaerial and subaqueous tracts, mainly developed during the peak of the MSC and commonly superimposed on older features. Stage 2 was then characterized by two high-amplitude glacial episodes (TG12 and TG14: Fig. 1) and by an acceleration of active tectonic processes along the entire Africa-Eurasian margin, as clearly shown by the angular unconformity commonly associated with the MES. Thus, in our opinion, a number of geodynamic and climatic causes acted simultaneously to modify the water and the atmospheric circulation within the Mediterranean during the Messinian. These causes were likely to have been linked to complex feedback mechanisms, leading to an extreme amplification of processes still acting today along the Mediterranean margins. The Black Sea slopes are characterized by a widespread erosional surface of Messinian age, the origin of which has usually been interpreted as being related to desiccation, similar to that of the Mediterranean (Hsü

286
M. Roveri et al. & Giovanoli 1979): however, Messinian evaporites are absent in the Black Sea (Tari et al. 2016). As occurs in modern times (Flood et al. 2008), we argue that during Messinian cascading of hypersaline, dense waters, together with sediment gravity flows, could also have produced the Black Sea erosional surface but in a different way from the Mediterranean. Black Sea deep brines might not have reached high saturation values, thus explaining the lack of evaporitic deposits there. In our model, the Mediterranean was a persistent water body characterized by reduced connections with the Atlantic, and by a hydrological budget controlled by regional climate oscillations and by exchanges with the freshwater reservoir of the Paratethys Basin (s) . This general setting could also well explain the last portion of the salinity crisis, which is considered to be characterized by an empty Mediterranean Basin with several isolated freshwater or brackish lakes. According to our model, this phase (MSC stage 3) was, instead, probably characterized by an overall positive hydrological budget and a high base level, punctuated by cyclical episodes of relative base-level fall (Roveri et al. 2008a(Roveri et al. , 2014aManzi et al. 2009;Rossi et al. 2015). Also in this late stage, the Mediterranean was a single, permanent water body, as suggested by the ostracod assemblages (Stoica et al. 2016) and uniform Sr isotope values (Roveri et al. 2014a, b, c).

An alternative scenario: implications for hydrocarbon exploration
The Messinian successions are characterized by an extreme lithological variability expressed in a complex stratigraphy that resulted in a diverse array of potential source rocks, reservoirs and seals. For these reasons, besides their own potential as a petroleum system, the Messinian sediments also played a substantial role in the other Mediterranean petroleum systems, especially for the pre-Messinian ones (see Pawlewicz 2004;Belopolsky et al. 2012;Al-Belushi et al. 2013;Bertoni & Cartwright 2015). We think that our chronostratigraphic framework and sequence stratigraphic approach may help in better identifying and characterizing these potentials in a coherent scenario.
Here, we will focus on two elements directly derived from our model that should be considered for their potential implications in hydrocarbon exploration: the Messinian source rocks and the effects of base-level changes throughout the crisis.

The Messinian source rocks
Source rocks originate from zones of high organic productivity and where organic-rich sediments are deposited in a low-oxygen environment allowing their preservation. During the Messinian, the intermediate-and deep-water settings were characterized by water stratification throughout the salinity crisis, and even before, owing to restricted exchanges with the Atlantic, which eventually led to the formation of deep brines. This resulted in the development of conditions favouring the accumulation and preservation of organic-rich deposits.
The source-rock potential of MSC deposits has been documented in several areas: the Chelif Basin (northern Algeria: Arab et al. 2015), the Prinos-Kavala Basin (northern Aegean: Kiomourtzi et al. 2008), the island of Zakynthos and the Hellenic Trench (Greece: Maravelis et al. 2013Maravelis et al. , 2015, the Ionian Basin (Deroo et al. 1978), the southern Apennines (Guido et al. 2007), and the northern Apennines (Manzi et al. 2007). However, a full knowledge of the Messinian source rocks is lacking, mainly due to the difficulty in organizing the available scattered data and observations into a comprehensive and detailed stratigraphic framework.
In this respect, our MSC scenario offers some clues for a better definition of the source-rock potential. We show here a first attempt to systematically organize the available data concerning organic matter in order to assess their areal and temporal distribution and characteristics. We collected Rock-Eval pyrolysis data (Espitalie et al. 1977) from the literature and also included a set of unpublished data, mainly from northern Italy and Sicily (Fig. 3a). After age recalibration of the available samples, we plotted the S 2 v. TOC (hydrocarbons generated by the pyrolysis v. total organic content: Fig. 3b-f ) and the HI v. OI (hydrogen index v. oxygen index: Fig. 3j-k) values of sediments belonging to the same stage from different areas. The compilation of these organic matter data into our three-stage chronostratigraphic model provides some revealing trends and features (see below).
Substantially, every sample considered in this work is immature (T max < 435°C), particularly those collected in exposed successions. However, local-and regional-scale geological reconstructions document that, in the main depocentres, Messinian organic-rich units may have reached burial depths sufficient for the attainment of thermal maturity. S 2 v. TOC values (Fig. 3b-f) provide an estimate of the petroleum potential, and show that most stage 1 and 2 values plot in an overall good potential field. Conversely, MSC stage 3 is generally characterized by a reduced organic carbon content and reduced S 2 values.
The different kerogen types were defined mainly based on the HI; in Figure 3e-k, we display modified van Krevelen diagrams for samples with a TOC of > 0.5% (note that c. 20% of these are not represented because no S 3 data were available). HI values show that organic matter of deep-water pre-MSC sediments and stage 1 range between type II and II-III kerogens (sensu Peters & Cassa 1994); stage 2 organic matter plots between kerogen types II and I; conversely, stage 3 records a progressive shift towards kerogens types III and IV, possibly representing an increased influence of continental input in the latest stage of the MSC (see also the organic matter composition of the northern Apennines Messinian units in Fig. 4).
Our results show that the deep-water equivalents of the evaporite deposits of stage 1 have a very good source-rock potential, also considering that this unit, where it is preserved below the resedimented evaporite deposits, may have thicknesses of the order of several tens of metres in the outcrop (c. 60 m in the northern Apennines; Fanantello borehole: Manzi et al. 2007) and up to 400 m in the subsurface (Po Plain foredeep basin: Rossi et al. 2015). Furthermore, in the deeper basins, anhydrite derived from the transformation of clastic gypsum due to lithostatic loading (Sinninghe Damsté et al. 1995;Manzi et al. 2005;Lugli et al. 2013) may represent an efficient early seal, so preventing migration of Messinian hydrocarbons.
In this scenario, the close association in deep settings of the potential source rock (i.e. deep-water stages 1 and 2 deposits) directly overlain by or interlayered with clastic evaporite deposits may be of great importance in the reconstruction of hydrocarbon migration pathways and for the recognition of potential reservoirs. In this respect, it is worth noting that frequently (e.g. northern Apennines and Sicily) large-scale zones of sulphur mineralization are associated with RLG clastic evaporites. Sulphur formed after the bacterial sulphate reduction of the Messinian evaporites favoured by hydrocarbon migration and led to the transformation of the parent rock into sulphur-bearing limestone (Dessau et al. 1962;Manzi et al. 2011). Although the hydrocarbons involved in these processes may be from older sources, the close association of sulphate and organic-rich rocks may point to a Messinian source rock.

Amplitude of Mediterranean base-level changes
In the scenario of Roveri et al. (2014c), the amplitude of the base-level changes during the salinity crisis was much less pronounced than usually envisaged; we think that the Mediterranean Sea experienced only a moderate relative base-level fall (Christeleit et al. 2015) and that desiccation, as well as a catastrophic refill (Hsü et al. 1973), did 287 The Messinian salinity crisis: petroleum systems not occur. The lower slopes and the deep-water settings did not undergo subaerial exposure and erosion. It follows that the organic matter in the pre-MSC and in the stage 1 deposits was far better preserved than expected for a complete basin desiccation.
Besides these more obvious aspects, the desiccation scenario implies rapid and massive water loading/unloading of the order of thousands of metres (Ryan & Cita 1978;Govers et al. 2009;Sacleux et al. 2013). These changes and their isostatic effects would cause overpressure and catastrophic fluid expulsions, even through the thick Messinian evaporitic unit that is usually considered an ideal seal (Bertoni & Cartwright 2015). Another aspect would be the degradation and/or remigration of pre-existing hydrocarbons (Al-Belushi et al. 2013;Iadanza et al. 2015), with important implications for assessing the overall quality of the Mediterranean petroleum systems. In our model, the magnitude of these processes would be considerably lower, translating to a significantly lower exploration risk for pre-Messinian targets (cf. the Black Sea: Tari et al. 2016).

Conclusions
Far from being a 'fact', as commonly considered, the desiccation model of the Messinian salinity crisis is only one of several possible scenarios. We suggest that an alternative, deep-water, nondesiccated model for the MSC is not only possible, but also even more likely. This model has several important implications in the assessment of the Mediterranean petroleum systems, as well as of the adjoining area (e.g. Black Sea: Tari et al. 2016). We think that the impact of the model for source-rock generation and distribution, as well as for the effects of water unloading for breaching preexisting hydrocarbon accumulations, should be carefully considered and evaluated. Our new data and a reconsideration of all available data suggest that the pre-salinity crisis sediments and the stage 1 source rock have a greater potential than previously thought. In addition, the stage 2 resedimented deposits may provide an excellent seal, especially in deep Mediterranean settings.
The onshore and offshore perspectives of the MSC will be reconciled only when deep drilling hopefully reaches the pre-salt unit, and core data made available to the scientific community, especially from sediments on the ocean floor of the Mediterranean. This obviously requires a great joint effort between academia and industry.

289
The Messinian salinity crisis: petroleum systems