Salt tectonics evolution in the Provençal Basin, Western Mediterranean Sea

The study of salt tectonics has been of fundamental importance in oil exploration since its inception. Salt is considered impermeable to fluids and gases, making it an excellent seal. Furthermore, given its mobility, it deforms at geological time scales, thus forming traps and influencing the distribution of reservoirs. More recently, the characteristics of salt are in the spotlight regarding subsurface storage. Storage can be located directly within the salt or in sedimentary traps created by its deformation (e.g.Duffy et al., 2022). Another significant property of salt is its high thermal conductivity (Mello et al., 1995): salt acts as a heat pump, with obvious implications in geothermal energy and oil window productivity. Also, temperature influences salt tectonics: an increase in temperature leads to a decrease in viscosity and thus faster deformation (Carter et al., 1993). The thermal regime, jointly with the most important factors triggering salt tectonics, such as salt thickness variation, sedimentation, basin tilt, salt purity, tectonics, could potentially have a strong influence on salt mobility.

In the Mediterranean Sea, salt was deposited during the Messinian Salinity Crisis (MSC) in the deep parts of aborted basins, at a time when the drop in sea level (or beneath a deep-water saline basin, e.g.Christeleit et al., 2015) caused thick evaporites deposition within a relatively short time (∼0.64 Ma) (between 5.96 and 5.33 Ma) (Gautier et al., 1994). In all the Mediterranean Sea, salt is generally deposited above a marine sedimentary sequence and crust that spans from continental, to transitional and oceanic.

In this paper, we focus our study on the Provençal Basin, located in the Western Mediterranean Sea, because of a considerable amount of data (seismic and well data) and in-depth knowledge of the sedimentary column and deep crustal segmentation from previous studies. The youth of the salt layer and the stable tectonic setting that followed its deposition has led to less mature salt tectonic structures and an evolutionary history that is easier to decipher than in other margins around the world (e.g. Gulf of Mexico, Brazilian or Angola basins). The discussion around salt morphology and tectonics in this area began with the works of Pautot et al., 1984), followed by Le Cann 1987), Gorini 1993), Gaullier 1993), Dos Reis et al., 2005, Dos Reis et al., 2008,), Gaullier et al., 2008) and Mianaekere et al., 2020a, 2020b). Similar to other worldwide salt passive margins (e.g.Cobbold and Szatmari, 1991; Demercian et al., 1993; Letouzey et al., 1995; Vendeville, 2005; Jackson and Hudec, 2017), the authors (Gaullier, 1993; Gorini, 1993; Dos Reis et al., 2005;) describe three salt kinematic domains from the lower slope to the deep basin: (i) a proximal extensional domain, (ii) a mid-slope translation domain and (iii) a distal shortening domain. The salt structures characterising these domains are interpreted as the result of thin-skinned tectonics controlled by sedimentary spreading and gravity gliding (Dos Reis et al., 2008, Mianaekere et al., 2020a, 2020b; Granado et al., 2016; Obone-Zue-Obame et al., 2011; Geletti et al., 2014; Dal Cin et al., 2016). Other hypotheses include thick-skinned tectonics (Pautot et al., 1984; Le Cann, 1987; Maillard et al., 2003) with a significant role of the basin-scale shape of the salt layer (Gaullier et al., 2008). Several authors claim that salt tectonics started after the deposition of the Messinian Upper Unit (UU) during the Lower Pliocene, due to basinward tilting subsidence and/or sedimentary thickness (e.g.Dos Reis et al., 2005). Gaullier et al., 2018) described for the first time an early salt movement in the deep basin concomitant with the deposition of the UU (last phase of the MSC), recently confirmed by Bellucci et al., 2021a). Nevertheless, the lack of accurate dating of the Pliocene-Pleistocene sequences has so far precluded a more in-depth discussion of the timing and causes of deformation. In this work, through detailed analysis of seismic geometries and deformation, we present a detailed study of salt tectonics phases and timing in the Provençal Basin.

The Provençal Basin (Fig. 1) is a young passive margin formed after the counter-clockwise rotation of the Corso-Sardinian blocks started in the Late Eocene (Auzende et al., 1973; Olivet, 1996). A relatively short-lived rifting phase (∼9 Ma; Réhault et al., 1984; Gattacceca et al., 2007) was followed by oceanic spreading which led to the formation of a thin atypical oceanic crust in the deep basin (Afilhado et al., 2015; Moulin et al., 2015; Bache et al., 2010). Since ∼16 Ma (Leroux et al., 2019), the Provençal Basin has not been affected by any major tectonic movement. Describing the sedimentary markers and their paleobathymetric significance, Rabineau et al., 2014) observed a purely vertical subsidence in the Gulf of Lion deep basin and tilting in the continental domain, subsequently confirmed by an extensive 3D regional analysis and numerical stratigraphic modelling by Leroux et al., 2015a, 2015b). On the shelf and slope, the subsidence consists of seaward tilting while the deep basin subsides vertically (Fig. 2). The limit between tilting and purely vertical subsidence coincides with the limit between thinned continental and exhumed lower continental crust (Fig. 1A) constrained by wide-angle refraction data (Moulin et al., 2015) and deep reflection seismic lines (Bache et al., 2010).

The stratigraphy of the Provençal Basin has been investigated at various levels, from the syn-rift to Pleistocene sequences (e.g.Gorini, 1993; Lofi et al., 2005; Droz et al., 2020; Rabineau et al., 2006; Leroux et al., 2017). The short-term Messinian Salinity Crisis (MSC; Hsü et al., 1973) event (∼5.96–5.33 Ma, Gautier et al., 1994) strongly impacted the stratigraphy of the whole Mediterranean region. Restriction of the connection between the Atlantic Ocean and the Mediterranean Sea (e.g.Benson et al., 1991) led to the deposition of thick evaporites, including around 0.8–1 km of halite, also called Mobile Unit (MU) (Lofi et al., 2011, Lofi et al., 2018). In the Provençal Basin, salt was deposited in a stable tectonic context, above a thick (∼2 km) marine sedimentary sequence (Fig. 2).

In this work, we use the term “Salt” for the MU described in Lofi et al., 2011, Lofi et al., 2018. We further consider an undetermined “pre-salt” sequence composed of syn- and post-rift sedimentation of Oligocene-Miocene age (Fig. 3) including the Messinian Lower Unit (LU) deposited before the salt (i.e. LU from Lofi et al., 2011 and LU1 and LU0 from Bache et al., 2009) (Fig. 5). LU sequence is considered as the expression of the first phase of the MSC, composed of detrital deposits possibly intercalated with evaporites. The thick pre-salt mega-sequence onlaps the acoustic basement, infilling earlier topography (Fig. 2). The Messinian units in the Provençal Basin mainly occupy the lower slope and the deep basin (Figs. 2, 3): the present-day salt deposit accumulated in the deep basin and lower slope, where it onlapped the pre-salt sequences (Fig. 2). The salt transparent acoustic facies is interpreted as predominantly consisting of halite (Lofi et al., 2011). The UU is the most recent Messinian unit and is composed of a set of parallel and relatively continuous reflectors of high amplitude overlying the salt (Lofi et al., 2011) (Fig. 5). In the upper slope and shelf, we observed the Messinian Erosional Surface (MES, Lofi et al., 2011), which is considered the top of our pre-salt unit or the base of the Pliocene-Pleistocene sequence (Fig. 3).

Figure 4 shows the evolution of the basin and its margins using three key ages (∼16 Ma, ∼5.6 Ma and 0 Ma). The deep crustal segmentation, geometry and nature are taken from the wide-angle refraction profiles (so in depth) interpreted in Moulin et al., 2015 and Afilhado et al., 2015. The profiles show the crustal geometry, segmentation, and nature with respect to salt deposition during the MSC and the present-day salt morphologies.

Around 16 Ma (Fig. 4a) the rotation of the Corso-Sardinian block has ceased, and the Provençal Basin assumed the shape and boundaries that are still visible today (e.g.Auzende et al., 1973;Olivet, 1996; Bache et al., 2010): no major horizontal movements have occurred from this time to the present-day. Since the formation of oceanic crust, subsidence in the deep basin (within the oceanic and transitional domains) is purely vertical while in the thinned continental crust domain, the authors observed a tilting seaward (Rabineau et al., 2014; Leroux et al., 2015a, 2020b) (Fig. 4a). Pre-Messinian sedimentation is characterised by marine deposits filling the basement roughness (Fig. 2).

Around 5.6 Ma (Fig. 4b) (Clauzon et al., 1996; CIESM, 2008; Gorini et al., 2015), a major sea-level drop (>1000 m) took place leading to the salt deposition: it thus occurred in a closed and already formed basin context, after the opening of the basin, above a thick pre-Messinian sedimentary blanket. The salt deposited in the deep basin, pinching out on the pre-salt sediments in the lower slope. The initial thickness of salt can be considered constant in the deep basin while it may decrease at the basin edges. Today, salt morphologies in the deep basin show substantial differences (Bellucci et al., 2021a), due to the evolution of salt tectonics over the last ∼5 million years (Fig. 4c).

From the lower slope to the deep basin, the Provençal margin shows salt structures outlining three different kinematic domains, as also described by previous authors (Gaullier, 1993; Gorini, 1993; Dos Reis et al., 2005; Mianaekere et al., 2020a, 2020b).

• The Extensional domain is characterised by listric basinward-dipping faults that develop from the base of the salt in overlying units (Fig. 6). This domain is characterised by salt rollers and rollover structures (Fig. 6). The salt rollers describe low-amplitude deflections of the upper surface of a salt layer at the lower termination of normal faults in the overlying sediments (e.g.Jackson and Hudec, 2017). The growth faults are therefore syn-sedimentary. They may be still actively deforming the seafloor (Dos Reis et al., 2005; Badhani et al., 2020) or be buried (Fig. 7). Clear growth sequences in the Pliocene and Pleistocene deposits are represented in Figure 6. The Extensional domain is interpreted in the lower slope, occupying an area that extend from the upslope limit of salt up to 70 km basinward (Fig. 7), coincident with the deep hinge line separating two different crustal domains with two different modes of subsidence (Moulin et al., 2015; Leroux et al., 2015).

• The Fold domain is characterised by salt pillows, anticlines and tabular salt (Fig. 6) concordant with the overburden. The large-scale deformation associated to salt pillow affects the seafloor. Some outer-arc extension faults are locally observed over the roof of salt and mostly above the UU. An area of greater salt thickness (between latitude 41°−42°N and longitude 4°−5°E) of around 900-1100 m (with a salt velocity of 4.5 km/s) (Fig. 7) is indicated within this domain. The anticlinal axes highlight a NE-SW direction or, when involved in the Petit Rhône Fan, the axis directions follow the main sedimentary path (mainly N-S and NW-SE).

• The Large Diapir Salt Domain (LDSD) is characterised by salt walls and stocks (Fig. 6) clearly showing truncation and onlaps within the overburden. Its landward limit is coincident with the limit of two different crustal domains (exhumed lower continental crust versus proto-oceanic crust; Moulin et al., 2015) (Bellucci et al., 2021a). The salt walls show a preferential N-S direction as well as the mini-basins located between the salt structures (plot direction Fig. 7). The mini-basins (Fig. 6, Fig. 7) form preferential pathways for sediments, confirming the mutual relationship between sedimentation pathways and salt tectonics (Dos Reis, 2001). The salt structures are growing and deforming the seafloor (see inset in Fig. 6). The large salt walls and stocks become narrower and less piercing towards S-W (Fig. 6). Here, the diapirs occasionally deform the seafloor and are more connected, making it more difficult to individually identify them on salt thickness map.

We used a large dataset of reflection seismic surveys (Fig. 1) including both academic and industrial seismic lines acquired since the 1960s, coupled with several boreholes in the platform and slope. The available reflection seismic dataset, a result of collaboration between French, Spanish, Algerian and Italian research institutes, covers most of the Western Mediterranean sub-basins except for the Ligurian Basin (see also the seismic stratigraphic compilation in Bellucci et al., 2021b. All the seismic lines details can be found in Leroux, 2012 and Bellucci, 2021. In this work, we have concentrated in the Provençal Basin. The seismic interpretation were undertaken using the principles of seismic stratigraphy (Vail et al., 1977) with recognition of seismic facies, seismic unit identification based on the configuration of seismic reflectors, including reflector continuity and termination (onlaps, downlaps, toplaps). We jointly interpreted different resolution lines in time domain, from very low (e.g. ECORS survey; Gorini et al., 1993) to very high (e.g. PROGRESS survey; DOI: 10.17600/3020080). Isochron maps (TWTT) were then computed with the nearest neighbour interpolation algorithm, which assigns a weighted average value to each node that has one or more data points within a search radius (0.5 km). The radius was chosen based on the maximum average distance between lines in the dataset. Then isopach maps (in TWTT) were also calculated. The time-isopachs maps are used as a first order approximation as velocities in one unit would change according to present-day depth of the unit (with higher velocities when unit is deeper). Full time-depth conversion could not be done on all the dataset due to the limited available depth seismic data and limited information on true velocities in 3D (Leroux, 2012). However, simple time-depth conversion was applied locally (within one single unit) using average velocities (from the unit) (see Supplementary Data) to give a first approximation of thicknesses in meters. Seismic two-way travel-time (TWT) has generally been tied to formation tops in wells using velocities from sonic logs. Note that those time-depth relationships are published in Bache et al., 2015 (Fig. 2) and Leroux, 2012 (including velocities from refraction) (All the velocities information are provided in the two Supplementary Data figures).

In addition to the base and top salt reflectors, we have interpreted the Messinian margin Erosional Surface (MES) in the shelf and upper slope, and the top of the UU in the lower slope and deep basin, both generally dated at 5.33 Ma (CIESM, 2008) and indicating the end of the Salinity Crisis. Considering that an exact age and duration for salt deposition are still debated (e.g.Clauzon et al., 1996; Bache et al., 2012; Meilijson et al., 2019), we assumed a salt deposition started around 5.6 Ma (Fig. 5). Some Pliocene-Pleistocene key reflectors previously interpreted on the shelf in the work of Rabineau 2001, Leroux 2012 and Leroux et al., 2017 have been extended in the deep basin. From the oldest to the youngest, the Pliocene-Pleistocene reflectors are labelled P11, Q10 and Q5 (Fig. 5). P11 is a strong erosional discontinuity dated from Autan1 borehole (location in Fig. 1B) at 2.6 ± 0.5 Ma (Fig. 5) thanks to the appearance of Neogloboquadrinaatlantica (planktonic foraminifer), used to date the base of the Gelasian (2.588 Ma; i.e. the base of Pleistocene in marine environments, Suc et al., 1992). The Q5 surface, dated at 434 ± 5 kyr (Rabineau et al., 2006; Bassetti et al., 2008; Sierro et al., 2009; Leroux et al., 2017) is part of the last five shelf erosional surfaces corresponding to the last five glacial maxima that correlates to a correlative conformity surface on the outer shelf and upper slope. The last most recent glacial maxima 20 ka) has been fully dated in its correlative conformity part using C¹⁴ dating (e.g.Rabineau et al., 2005). Q5 was interpreted as the glacial maxima related to MIS 12 (Marine isotopic Stage 12 at 434 ± 5 kyr), initially by considering architecture of deposition and numerical simulation (Rabineau et al., 2005 and 2006) (glacial Maxima). This dating is now fully proved by results from the two PROMESS drill-sites (Fig. 1B) (with nannofossils and oxygen isotopes analysis) (Bassetti et al., 2008; Sierro et al., 2009). Q10 is another high seismic amplitude erosional discontinuity on the shelf whose age is estimated at 0.9 ± 0.2 Ma based on stratigraphic correlations and numerical modelling (Fig. 5) (Rabineau, 2001; Leroux et al., 2014; Rabineau et al., 2014). The uncertainty on the Q10 surface is greater as no direct dating information are available.

The seismic dataset used in this study consists of 2D lines which, although forming a dense seismic grid with few blank zones, can lead to several out of the plane signals in the presence of complex geological structures. Correlation of interpretation within different resolution can be sometimes tricky and lead to errors. The seismic interpretation was performed on TWT lines, in an attempt to minimize errors in the correlation between the few lines available in depth domain and those in TWT. The time domain implies being more cautious in using sedimentary thicknesses between units as an argument for dating deformation of the overburden over time (e.g. exaggerated thicknesses on the flanks of sloping structures). We mainly used stratigraphic terminations (onlap, toplap..) to determine the intensity and timing of the salt deformation and the accommodation of the overburden. The Pliocene-Pleistocene reflectors often correspond to erosional surfaces and discontinuities on the shelf but were correlated as correlative conformities on the slope and deep basin. This correlation in the deep basin is also subject to some uncertainties as it is not always easy to conduct (especially when crossing faulted areas and piercing salt structures). The wells used for the verification of the seismic interpretations are mostly located on the shelf and upper slope (Fig. 1). The well-to-seismic tie therefore has an uncertainty that increases towards the deep basin, where the distance to the wells is greater and the salt deformation more discordant with the overburden. All this can lead to a locally varying uncertainty of the mapped isochrons, thickness and interpretation that we estimate in the range of a few tens of meters. Despite these uncertainties, the use of detailed markers provides an unprecedented seismo-stratigraphic background to discuss salt tectonics through time.

In this section we describe the evolution in time and space of salt tectonic structures in the Provençal Basin. We first describe the timing phases depicted on 2D profiles perpendicular to the margin and then extend our observations in space with detailed isochrons maps of the Pliocene-Pleistocene sequences. The timing phases described below were discriminated solely based on the geometry observations and not by considering the causes and drivers that led to their formation. Main parameters used are the stratigraphic terminations, the accommodation of the overburden (concordant or discordant on the salt structure) and the spatial change of the deformation style.

We have primarily focused on understanding the initial phase of salt deformation. The profile in Figure 8 (location Fig. 6) shows the evolution of salt structures from the fold to the large diapir salt domain. The UU sequence (green, Fig. 8) shows several parallel onlaps on salt within it (Fig. 8): zoom A’ displays several reflectors within the UU with clear onlaps and an unconformity that is locally faulted and regionally parallel to the Pliocene-Pleistocene layers. In the lower part of the UU, the reflectors are not parallel to each other nor to the overlying layers of the Pliocene. The onlaps terminations seem to be mainly located in the lower-middle part of the UU sequence: the two to three reflectors above the salt follow the salt top deformation while just above them the growth episodes appear. As the top salt can be locally difficult to define, we do not exclude that the deepest UU reflectors are onlapping the salt. These observations suggest an early salt deformation in the deep basin, starting during the last phase of the MSC deposition (UU sequence) or last phase of MU deposition.

Phase 1 (early salt deformation) is only well imaged on the fold domain (Fig. 8). Basinward, in the LDSD, the early salt deformation is difficult to discriminate. Here, the vertical stocks and walls inhibit small-scale deformation, and the UU could be interpreted as a pre-kinematic layer (e.g.Mianaekere & Adam, 2020a). Nevertheless, given the observed onlaps within the UU in 2D transects (Figs. 8, 9, 10 A-B) we suggest that salt began to deform during UU deposition throughout the deep basin, also within this domain. Phase 1 was thus active during UU deposition, and may have been started earlier, during salt deposition (∼5.6-5.33 Ma).

The Extensional domain is characterized by basinward-dipping listric faults (Figs. 6, 8) typical of thin-skinned salt tectonics, i.e. affecting only the salt and the overlying deposits (e.g.Cobbold and Szatmari, 1991). To reconstruct fault-evolution timing, it is necessary to recognize the growth strata, induced in this case by fault activity. The parallel UU reflectors involved in the salt rollers (black arrows, zoom B’, Fig. 9) and the almost total absence of onlaps suggest an activation of the listric faults after its deposition, thus after or from the Lower Pliocene onwards. The UU sequence was already fully deposited at the time of fault initiation, whose age therefore spans from the Lower Pliocene to the present-day, as confirmed by seafloor deformations (also according to Badhani et al., 2020; Droz et al., 2020) and several post-Messinian growth strata (red arrows, fig. 9-zoom B’). The beginning of listric fault activity in the lower slope (Early Pliocene) is therefore subsequent to the salt deformation of phase 1 in the deep basin, which occurred in the Late Messinian. Phase 2 began in the Early Pliocene and ended around the Pliocene-Pleistocene boundary (5.33 − ∼2.6 Ma) (Figs. 8, 9, 10 A-B).

In the N-E sector of the fold domain (Figs. 8, 9, 10 A-B), the growth of pillows continued during all the Phase 2. In the yellow unit, the number of reflectors decreasing in the anticline limbs compared to the synclinal part (from about 15 to 9) and some onlaps within it attest a growth rate that seems to be more intense in the deep basin (basinward part of the fold domain) (Fig. 9–zoom B”; Fig. 8–zoom A”).

In the LDSD, phase 2 reflectors onlap the salt diapirs, gently folding following the upward deformation (Figs. 8, 9, 10 A-B). The overlying sequences are discordant and syn-diapirism. The LDSD structures probably underwent a passive evolution during phase 2. The structures grew towards the surface (currently lacking or having a thin overlying sedimentary sequence) at the same time as the mini-basins on the diapir side were sinking.

Phase 3 is the most significant deformation phase in the Provençal Basin (Figs. 8, 9, 10 A-B). It started at the end of the Pliocene, Early Pleistocene (∼2.6 Ma) and is still active.

The Extensional domain is characterised by basinward-dipped listric faults, salt rollers and rollover structures. The salt glides along its detachment surface, concomitant with the growing of the listric faults. Regionally, the still active faults are located in the basinward part of the Extensional domain, while those buried (inactive) are localised in its landward sector (Fig. 6), as also mentioned by Dos Reis 2001. This observation suggests that the salt gliding and subsequent fault activity started in the Early Pliocene (phase 2) but are still active today. The progressive basinward gliding of the salt led to the welding of the primary salt layer and cessation of fault movement upslope the Extensional domain. Instead, downslope of this domain, the faults are still active testifying to a gravitational gliding of the salt until the present-day. As previously mentioned, the deformation is mostly thin-skinned. However, an important base of salt displacement is observed in figure 10B, where the salt is thickened in correspondence of a volcanic structure (Maillard et al., 2020). The compressional salt thickening in the updip side caused an increase in gliding velocity in the downdip sector, resulting in the formation of listric faults and growth strata (see also Dooley et al., 2017). In the updip part, the Pliocene and Pleistocene units show syn-deformation deposition, which may be caused by progressive thickening of salt (given by the continuous gliding from the mid-slope) with consequent deformation above the volcano. An almost constant UU thickness indicates predominantly active gliding from the Lower Pliocene onwards.

The salt-cored folds continued growing throughout phase 3 within the entire domain, as attested by several onlaps (Figs. 8, 9, 10 A-B) observed within it. In the central sector of the Fold domain, the pillows are more deformed, especially at the limit with the LDSD (Figs. 8, 9, 10 A-B).

In the LDSD domain, as in phase 2 (yellow colour sequence), phase 3 reflectors (blue sequence) are discordant with the salt diapirs, onlapping them. The salt deformation rate increases at this phase, especially in the fold domain.

Figure 11A represents the sedimentary isopach map (TWTT) of the Pliocene unit whose age ranges from the end of the MSC (5.33 Ma) to the P11 reflector (∼2.6 Ma) (corresponding to phase 2). During the Early Pliocene, the sedimentary inputs filled the subaerial canyons formed during Messinian erosion (up to 1000 m deep; Clauzon, 1982), both on land and up to the present-day outer shelf location, which also emerged at that time (Leroux, 2012). Once the sediments had filled the Messinian canyons, they spread into the deep basin via a slope bypass (Fig. 11A). The sedimentary thickness is greatest in the north-eastern sector of the fold domain (Figs. 8, 9, 11A), while it decreases towards the S-W (Figs. 10 A-B, 11 A). The mapped sedimentary unit (MSC-P11; 5.33-2.6 Ma) corresponds to phase 2 of the salt deformation described in this work. The salt tectonics phase 2 is characterised by basinward-dipping listric fault formation in the Extensional domain, more quiescent deformation phase in the present-day fold domain and active, intense deformation within the LDSD. The constant thickness in the N-E sector of the fold domain (Fig. 11A) confirms a period of relative salt tectonic quiescence in this area while, as observed in profile 10B, in the S-W sector, deformation is stronger since the Lower Pliocene. In the LDSD sedimentary depocenters within the mini-basins suggest active deformation throughout phase 2 (Fig. 11A). In the fold domain, because the significant input of Pliocene sedimentation, the salt does not appear to have deformed as intensely and rapidly as it has in the LDSD. The large sedimentary input at this stage may have slowed the deformation. In the S-W sector, characterised by a much lower sedimentary thickness (Fig. 11A), the salt deformed prematurely.

During the Pleistocene (Phase 3), strong turbidite sedimentation occurred and led to aggradation and progradation of thick turbidite systems linked to the main canyons on the Gulf of Lion (e.g.Dos Reis, 2001; Droz et al., 2020; Badhani et al., 2020). The main source is the Rhône, which mainly feeds the canyons of Sète, Marti, Petit-Rhône and Grand Rhône. The doubling of sedimentary volume at 0.9 Ma, coinciding with the Mid-Pleistocene Climatic Revolution, is characterised by a large amount of continental terrigenous input, a change in cyclicity and higher sea-level amplitude variations that facilitate connections between onshore rivers and offshore canyons (Leroux, 2012) (Fig. 11B). The isochron map in Figure 11B illustrates the resulting sediment deposits on the youngest part (0.9 Ma-0) of phase 3 of salt deformation (estimated in this work between 2.6 and the present-day) and aims to illustrate in detail the sedimentary arrival caused by the Rhône at 0.9 Ma. Phase 3 is characterised by intense salt deformation both in the fold domain, with the acceleration of salt rising in pillow structures, and in the LDSD, where diapirs deformed vertically with a passive character (see Rowan and Giles, 2021; Jackson and Hudec, 2017). In the LDSD, the diapirs were subject to passive diapirism from the earliest stages of deformation, except for the Early Pliocene phase 2 in which the salt probably broke through the overburden. After that, the diapirs grew near-surface, syn-depositional, with a thin roof top. The Extensional domain is characterised by the evolution of the basinward-dipped listric faults, already active during phase 2. Pillow sizes do not appear to show significant differences between the different sectors (S-W and N-E, fig. 6). Salt thickness increases towards the S-W (Fig. 6a), as shown in the perpendicular profiles (salt thickness in figs. 8, 9, 10 A-B). The present-day directions of the salt structures and associated mini-basins in the LDSD show the same direction as the Petit-Rhône fan, aligned in an N-S and NW-SE (Figs. 8; 11B).

In the Provençal Basin, the geometrical correlation between the Ocean-Continent transition and the variation in salt structures was already observed by Pautot et al., 1984 and Le Cann 1987. Bellucci et al., 2021a have recently confirmed that salt structures changing morphologies at the boundary between different crustal natures can be observed both in the Western Mediterranean but also in other salt-bearing passive margins. One of the objectives of this work is to further investigate this observation, detailing the relationship between the evolution of salt structural styles in time and space and the nature of underlying crust.

Three main salt tectonic phases were dated using previous stratigraphic markers (P11, Q10, Q5) whose dating and suitability were constrained by well data and numerical modelling. These phases are then geometrically related to the underlying crustal nature, showing different salt structure features and the growth deformation rate with respect to their position above the crustal segments.

Figure 12 summarizes the salt structure deformation in the study area within the deep regional crustal evolution previously highlighted (Fig. 4). It shows the salt deformation evolution from the lower slope to the deep basin. We considered an almost constant initial salt thickness deposited above a flat Miocene surface. Salt pinches out on pre-salt Miocene sediments, mainly formed by detrital deposits resulting from erosion due to rapid desiccation (Clauzon et al., 1996; Lofi et al., 2005, 2011,). On the lower slope, salt is considered less thick than in the deep basin, pinching out on the edge of the basin. The thick (up to 2.5 km, fig. 2), unfaulted pre-salt sequence and the thin-skinned salt tectonics recorded over the entire margin seem to rule out, for its entire history, a regional tectonic influence on salt deformation, unless locally (Fig. 10B).

The hinge line highlighted in Figure 12 is located between the domain of tilting subsidence on the slope and the purely vertical subsidence in the deep basin (Rabineau et al., 2014; Leroux et al., 2015a). This hinge line also corresponds to the basinward boundary of the crustal necking domain, at the edge of the transitional domain (Fig. 12) (Moulin et al., 2015; Leroux et al., 2015a).

Subsequent to deposition (phase 0), the salt tectonic phase 1 is concomitant with UU deposition (∼5.6–5.33 Ma) (or probably started during late MU deposition) and the initiation of deformation in the form of large pillows in the deep basin, away from the margin, above the proto-oceanic domain and the seaward sector of the transitional crust domain (Fig. 12b). The first salt tectonic phase is the most difficult to investigate because it is the oldest and mildest of the three observed. Early deformation (since the first phase of salt deposition) has been discussed in the Western Mediterranean by several authors (e.g. Gaullier et al., 2014; Soto et al., 2022; Blondel et al., 2022). Since the deformation started in the flat, horizontal deep basin and the listric faults are only active later (since the Lower Pliocene), we exclude that the gravitational gliding is the main trigger cause of the onset of salt deformation. We do not observe any thrusting of the salt base or underlying/overlying units that would suggest regional shortening that may cause early salt deformation, as for example described and quantified by Soto et al., 2022;) in the Algerian Basin. The Provençal Basin, unlike the Algerian Basin, did not undergo regional shortening during the Messinian until the present-day. If we consider an initial constant salt and UU thicknesses, differential sedimentary loading also seems unlikely. Nevertheless, an almost imperceptible difference in UU thickness could trigger the formation of passive diapirism as long as the sedimentation rate is not great enough to inhibit the process (see for instance Rowan and Giles, 2021 and Blondel et al., 2022). The first phase of deformation is clearly observed only in the transitional crust domain. However, above oceanic crust, the UU pinching out onto diapiric structures suggests a syn-depositional subsidence and diapiric rise also in this domain. Here, early onlaps are currently hidden by a more intense evolutionary history.

Phase 2 (5.33–2.6/1.8 Ma) is characterised by the formation of listric faults in the lower slope (Fig. 12c). In the lower slope (present-day Extensional domain), the tilting subsidence has increased the base salt slope towards the deep basin, aiding both gliding of salt (leaving small remnant salt pillows on the slope), subsequent extension and formation of basinward-listric faults that dislocate UU sequences forming growth strata since the Lower Pliocene. During the same deformation phase, little deformation occurred above the fold domain (above the Transitional crust) while in the LDSD (above the proto-oceanic crust) the salt deformed more rapidly, passively and probably driven by updip extension. The formation of listric faults in the lower slope during the Phase 2 is therefore subsequent to the formation of the first diapirs in the deep basin observed in Phase 1.

According to several authors, a few inclination degrees of the base salt (and top salt) would be sufficient to establish a gravity gliding regime (e.g.Duval et al., 1992; Ings et al., 2004; Brun and Fort, 2011). A depth profile perpendicularly crossing the Provençal Basin (Fig. 2) highlights a slope base of salt of around 0.1° basinward of hinge line 3. Landward, it currently shows a slope of 0.75° in the distal lower slope and 1° if we consider the whole Extensional domain. Although the angle of the base salt is not an absolute criterion to generate deformation − other characteristics may come into play (presence of fluids, overlying sedimentary thickness, initial salt thickness, lithology...) − active salt gliding seems to be possible in the Extensional domain only. The base salt slope angle in this salt domain has increased throughout its history due to the progressive tilting of the margin and the value recorded at the present-day is therefore the highest slope it has ever reached. On the contrary, in the deep basin, its slope has not changed, as the subsidence is still purely vertical today. Phase 2 is characterized by active diapirism in the transitional crust domain showing pillow-like structures similar to those observed today. Here, the large sedimentary input could be the cause of the slowdown in the evolution of salt structures. The morphologies transition is sharp in oceanic crust, where salt diapirism mechanism is passive as early as Phase 2. Above the oceanic crust, salt broke early (probably as soon as Early Pliocene) through its UU roof and continued to grow until the present-day, possibly related to down-dip gliding associated to updip extension. The diapirs, characterized by a relatively fine drape-folded roof, were not at this stage arranged in salt walls but rather as individual salt stocks. From this phase onwards (from the beginning of the Pliocene), concordant and non-concordant deformations in the transitional and oceanic crust domains, respectively, become more distinct. Nevertheless, no structural variation that could explain this salt deformation difference has been observed in the available data.

During phase 3 (Fig. 12d), concomitant with major sedimentary input from the Rhône River (Fig. 11B), salt is actively deforming in the transitional crustal domain (pillow-like structures) and in the oceanic domain (near-surface passive diapirism). The important Rhodanian sediment supply, especially when it doubled after 0.9 Ma, may have influenced the salt structures growing directions, probably orienting them in the present-day thalweg direction (N-S and NW-SE) (Inset bathymetry Fig. 6). We suggest that salt, hitherto organized in salt stocks, forms salt walls that follow the main directions of the Rhône thalwegs. A more in-depth analysis of the spatial and temporal evolution focused on salt stocks in the deep basin would be necessary to confirm the relationship with Rhône sedimentary input. During phase 3, listric faults in the Extensional domain are locally still active and thus deforming the seafloor.

The difference in salt deformation intensity and morphologies between the transitional and oceanic crust domains (Tab. 1) is hard to explain when considering conventional driving forces like sedimentary loading, gravity or tectonic influence. No remarkable post-Messinian sedimentary thickness difference is recorded between the two domains, nor any difference in base salt inclination that could have generated differences in the evolution of salt tectonics. A tectonic component is unlikely as a thick sedimentary blanket separates the basement from the salt deposits and no significant Oligocene-Miocene fault activity displacing the base salt is observed. The first hypothesis we consider plausible is a temperature difference given by the underlying crustal nature. The viscosity of the salt is directly related to temperature: at higher temperatures, its viscosity decreases, thus favoring its movement. Heat flow measurements in the Provençal Basin (Burrus and Foucher, 1986; Poort et al., 2020) have shown a warmer proto-oceanic crust (>100 mW/m²) than transitional crust (50–60 mW/m²). This heat flow trend is not surprising considering the young age of the basin and is consistent with other world areas where a young oceanic crust shows high heat flow values. However, it is still unclear how temperature would influence the evolution of salt structures, which are characterized by a thermal conductivity two to three times higher than the sediments they are surrounded by. Basal heat is channeled into the diapirs up to the surface. Thus, as the diapir grows, it channels additional heat from the crust, further decreasing the salt viscosity and enhancing its mobility. In addition, fluids are well known to transfer heat. Upper oceanic crust is more porous and permeable than the continental crust, allowing active fluid circulation to the sedimentary column, often associated with high heat flow (e.g.Bellucci et al., 2024; Fisher and Becker, 2000). Although a thick sedimentary blanket above the basement, fluids could reach the sub-salt, generating overpressure (Dale et al., 2021) and affecting the salt deformation history to some extent. These fluid leaks may also affect the composition of the salt layer and thus its rheology. Salt tectonics may not only be related to structural mechanisms but also to halite rheological response to temperature changes and fluid activity. Further heat flow measurements coupled with numerical models may elucidate part of these hypotheses.

The Provençal Basin proves to be a useful area in understanding salt tectonic mechanisms in a passive margin context. The structural and salt deformation characteristics differ substantially from the deposition and evolution of salt structures in other well-known salt passive margins. This work has established a detailed evolution of salt structures in space and time. The main conclusions are:

• The salt precipitated in the deep basin and lower slope: the initial salt thickness can be considered constant in the deep basin while it decreases at the basin edges, where it pinches out on pre-salt sediments. The salt was deposited long after the formation of the basin, in a tectonically stable context. The present-day variations in salt structures are therefore not structurally conditioned.

• The salt has been deforming since its deposition to the present-day, in a relatively short period (∼5 Ma). Salt tectonics started very early, during the UU deposition (phase 1) and/or during the last phases of salt deposition. The Pliocene and Pleistocene salt movement can be divided into two more main phases (phases 2 and 3), resulting in morphologies distributed differently in space.

• The present-day salt walls direction in the deep basin were likely influenced by the dynamics of the Rhône sedimentary fan, strongly active since 0.9 Ma. Major sedimentary input could also explain the quiescence phase above the Transitional crust during phase 2 but hardly explain the intense and rapid deformation above the oceanic crust during this phase.

• Regardless the deformation phase, salt deformation appears to be more rapid and intense above the oceanic crust than in the continental or transitional crust domains. The salt deformation strain rate varies over time and space, showing an acceleration above the oceanic crust. We suggest that the salt morphologies − crustal segmentation relation could be explained by differences in temperature associated with different crustal natures. Salt above the oceanic crust, subject to higher temperature and/or potential water leaks associated with crustal nature that may impact the rheology and nature of the salt layer, deforms more rapidly. This could explain the development of more evolved and discordant structures when compared to salt located above transitional crust.

The authors acknowledge the constructive English editing advice and corrections by Alison Chalm. The authors wish to thank the Editor-in-Chief Laurent Jolivet and the Associate Editor Jean-Pierre Suc for giving us sufficient time to review this paper. The authors acknowledge the reviewers, Juan I. Soto and anonymous, for their advice and comments on the manuscript which greatly improved its quality. We would like to thank Adriano R. Viana for his support. We thank Bruno Vendeville, for his kind help during the first phases of this work and for his contribution to the scientific community. This work was supported by Labex MER (ANR10-LABX-19) and the ISblue project, Interdisciplinary graduate school for the blue planet (ANR-17-EURE-0015) co-funded by a grant from the French government under the program “Investissements d’Avenir" and by a PHD grant awarded to M. Bellucci by the Regional Council of Brittany and IFREMER and prepared at the EDSML doctoral School at UBO (University of Brest) and the University of Trieste. It was further supported by COSTCA15103-MEDSALT program and “Universite Franco-Italienne, Programme VINCI 2019”. We acknowledge the “Archivo Técnico de Hidrocarburos of the Spanish Ministerio de Industria Comercio” for having provided the SPBAL01 seismic data as well as Petroceltic International PLC and TGS (original acquirers and processors of migrated seismic line RM01). Seismic interpretation was integrated into Kingdom Suite. Maps are made with Global Mapping Tool (Wessel et al., 2019).