Syn-depositional Mesozoic siliciclastic pathways on the Moroccan Atlantic margin linked to evaporite mobilisation

.


Introduction
In the absence of tectonic forces, the mobility of weak evaporites could lead to surface topography influences potential sediment pathways (e.g., Venus et al., 2015).Extensive fieldwork and high-resolution 3D seismic surveys document this relationship between halokinesis and sediment channelling in fluvial and shallow-marine environments (e.g., Venus et al., 2015;Rojo and Escalona, 2018).However, this relationship is poorly documented in remote onshore areas and/or in areas with scarce field data.
Evaporite mobility markedly affected the post-rift evolution of the Moroccan passive margin, in NW Africa (e.g., Pichel et al., 2018).An outstanding example is the influence of Triassic evaporites of the Eastern Diapiric Province in the post-rift sedimentary architecture of Agadir-Essaouira Basin (EAB) in the Western High Atlas (WHA, Fig. 1;e.g., Michard et al., 2008).The present-day onshore segment of the EAB is the only Atlantic coastal basin in NW Africa bearing Triassic evaporites (Hafid et al., 2008).Furthermore, thickest evaporitic bodies in the EAB platform domain, as well as their Jurassic cover, are located atop of half-grabens (Tari and Jabour, 2013).
The intense influence of halokinesis in this basin is expressed as evaporite-cored anticlines in the onshore, allochthonous evaporite bodies, diapirs and tongues offshore Essaouira, and up-right tear-drop diapirs offshore Agadir (Hafid et al., 2008;Pichel, 2018).Towards the east, in the interior of the Africa tectonic plate, evidences along segments of the Central High Atlas suggest that the former rift basin was also intensely deformed by diapirs during the Jurassic (Saura et al., 2014;Verges et al., 2017;Moragas et al., 2018).There and in the Western High Atlas, folding of the Mesozoic platform series has historically been attributed to the recent deformations of the Atlas orogeny (Ager, 1974;Frizon de Lamotte et al., 2008).
The EAB evolution is dominated by rifting events, near complete Mesozoic sedimentation, and localized to regional erosion surfaces (Ambroggi, 1963;Ager, 1974;Ellouz et al., 2003;Michard et al., 2008;Luber, 2017).Its suspected Mesozoic hinterlands -the Meseta massif to the northeast, High Atlas Massif ancient directly adjacent to the east, and the Anti-Atlas to the south and southeast -show important variations in their exhumation and burial histories (e.g., Frizon de Lamotte et al., 2009;Domenech, 2015;Ghorbal et al., 2008;Saddiqi et al., 2009;reviewed and synthetised in Charton et al., 2020).The former basin developed at the junction between the Central Atlantic and Atlasic riftings during the Triassic to Early Jurassic, and was exhumed during Alpine shortening (e.g., Michard et al., 2008).Presently part of the Western High Atlas after the Cenozoic Alpine-related exhumation, the basin exposes tens of km-long evaporite-cored anticlines roughly perpendicular to the coast line (Michard et al., 2008).It is unclear if Jurassic to Cretaceous km-scale exhumation events of the massifs surrounding the EAB (Charton et al., 2020) contributed to evaporite mobilisation and/or led to topographic growth of the anticlines (namely Tidsi, Amsittène, Imouzzer, Cap Rhir, Anklout, and Lgouz; Fig. 1c), thereby modulating former fluvial pathways delivering sediments from the Meseta, High Atlas, and Anti-Atlas domains (Luber, 2017).
This paper constrains timing and evolution of evaporite-cored anticlines in the Western High Atlas and discusses whether onshore sediment pathways and offshore sediment distribution are controlled by the topographic expression of those anticlines.We present a workflow to gather remotely geological, structural and stratigraphical data, and to derive cross-sections that are unfolded.We investigate stratigraphic variations in thicknesses, and the position and timing of the unconformities, to unravel the episodes of anticline growth, either by diapirism or by tectonic shortening (Fernández-Blanco et al., 2020).Results allow us to compare the time of evaporite mobilisation in the onshore, in the offshore and in the Central High Atlas basin, and thus, the discussion on potential links with the vertical motions across the Moroccan passive margin.

Geological history
The Western High Atlas (WHA) presently connects the High Atlas fold belt, which is an inverted Triassic-Jurassic rift basin of the Tethys, to the Atlantic passive margin (Fig. 1; Michard et al., 2008;Teixell et al., 2003).The Essaouira-Agadir Basin (EAB), now deformed and exposed in the WHA, was surrounded during its Mesozoic history by other coastal basins to the north and south (Doukkala and Souss Basins, respectively; Fig. 1a) and by a series of Variscan massifs to the northeast, east, and southeast (Meseta, Massif Ancien, and Anti-Atlas, respectively).
The Africa-Europe collision forced the Alpine inversion since Late Cretaceous (Hafid et al., 2006; see synthesis in Frizon de Lamotte et al., 2008).N-NW/S-SE directed shortening led to the reactivation of rift-related normal faults and buckling the Mesozoic sedimentary cover,  Tari et al., 2012).*Seismic lines presented and interpreted in Pichel et al. (2019).b) Mesozoic litho-stratigraphic column (after Choubert, 1957, Michard et al., 2008;Luber, 2017;Mader et al., 2017;Duval-Arnould, 2019).Red letters alongside the formation names are the references used in the present work for the unfolding of the cross-section.c) Geological map of the area (after Hollard et al., 1985) showing the Western High Atlas anticlines (numbered 1 through 6).Names and precise location of the logs are presented in Table 1.(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)ultimately resulting in the thick-skinned thrusting and folding (Teixell et al., 2003).The Alpine compressional phase overprinted evaporite-cored anticlines, thereby hindering our capacity to infer evaporite mobility during the Mesozoic (Michard et al., 2008).

Triassic
Terrestrial sediments were deposited throughout the Essaouira-Agadir Basin during Triassic rifting of the region, under dominantly arid conditions.They are exposed in the Argana Valley, where between 2500 and 5000 m thick sequence of red beds form spectacular outcrops (Fig. 1b;Brown, 1980;Piqué et al., 2002;Mader and Redfern, 2011;Ellero et al., 2012).Triassic continental clastics have also been encountered in a number of wells drilled across the basin, and the rift architecture imaged on seismic data.

Jurassic
The earliest Jurassic deposits record marine carbonates in the northern part of the study area (Jbel Amsittène and well ESS-1).These carbonates are heavily dolomitized, but part of the original fabric of the Amsittène outcrops is still preserved and contain rare ammonites, brachiopods and abundant crinoid fragments.These open marine faunas were dated Sinemurian to early Pliensbachian (Duffaud, 1960;Du Dresnay, 1988).This formation is absent in the south and east part of the basin where the first recorded Jurassic formation, namely the Amsittène Formation, is composed of Toarcian continental deposits (Duffaud, 1960;Adams et al., 1980), interpreted as braided rivers, lakes and lagoons and flood plain deposits (Ambroggi, 1963;Bouaouda, 2004;Ouajhain et al., 2011).
The base of this formation is erosive and locally cuts down to the CAMP basalt and the Triassic.The Toarcian stage records an overall transgression across the basin and the appearance of a second carbonate formation, the Tamarout Formation.This formation is composed of shallow marine to sabkha deposits, including oolites, carbonate mudstones, stromatolites and dissolution-collapse breccias or evaporites (Ambroggi, 1963;Adams et al., 1980;Peybernès et al., 1987).The Middle Jurassic record shows a second major period of siliciclastic sedimentation form the Aalenian to the Bathonian.This interval is dominated by siliciclastic deposits in most of the study area and referred to as the Ameskhoud Formation (Adams et al., 1980, Fig. 1b).It can be noted that in the north of the study area, the same period records the development of evaporites and dolomites and evaporites, similar to the lithology of the Tamarout Formation, and is referred to by some authors as the Id ou Moulid Formation (Peybernès et al., 1987).In the centre and eastern part of the EAB, the Ameskhoud Formation is dominated by red clay and siltstones with sandstones and conglomerates intercalated (Ambroggi, 1963;Adams et al., 1980;Bouaouda, 2004).The Ameskhoud Formation presents a pronounced SE/NW orientation, where the proximal siliciclastic deposits are observed in the east and south of the basin and the marine influence is more noticeable as the outcrops are further towards the NW (Ambroggi, 1963;Duval-Arnould, 2019).The depocenter of the Ameskhoud Formation would therefore be located towards the SE of the central part of the basin (Fig. 1a, Imouzzer), where the sedimentation is essentially composed of marine siliciclastics.During the end of the Middle Jurassic, the transgression initiated during the Bathonian allows the development of initially oolitic grainstones, then Brachiopod-rich floatstones and finally marls all over the EAB (Ambroggi, 1963;Ager, 1974).This widespread succession is named the Ouanamane Formation (Adams, 1979;Adams et al., 1980) and has been dated Bathonian to Middle Oxfordian (Olivier et al., 2012).The following Lalla Oujja Formation (Adams, 1979) is dominated by coral-rich deposits, locally encrusted by microbialites and dolomitized, dated middle Oxfordian to Kimmeridgian (Ambroggi, 1963;Adams, 1979;Martin-Garin et al., 2007;Olivier et al., 2012).Directly following the coral build-ups succession, the Iggui El Behar Formation is composed of mud-supported carbonates with sparse fauna including large benthic foraminifera, locally, it also comprises some stromatolites and traces of evaporites (Ambroggi, 1963;Adams, 1979).This formation has been largely interpreted lagoonal, intertidal to supratidal deposits (Adams et al., 1980;Bouaouda, 2004).
The Kimmeridgian period is dominated by fluvial deposits in the eastern part of the basin, and in the centre and west of the basin by red marls and clays, called "Marnes Chocolat", with intercalated thin marine carbonates beds (Ambroggi, 1963;Duffaud et al., 1966;Adams et al., 1980;Bouaouda, 2004).The last Jurassic stage is marked by renewed carbonate deposition, named the Tismerroura Formation (Adams et al., 1980).This Tithonian succession is subtidal to supratidal.It is composed of a thick evaporitic succession in the area of Imi'N Tanoute, in eastern-most part of the basin (Duffaud et al., 1966;Duval-Arnould, 2019), while in all other locations, it is composed of micrites, oolites, stromatolites, dissolution breccias, with some bioclastic intervals, intercalated with rare thin beds of red clay (Duval-Arnould, 2019).
Lower Cretaceous deposits mostly thicken towards the west (i.e., basinwards) and some formations disappear towards the east (proximal) as a result of coastal onlap (Rey et al., 1986).Sedimentary geometries are broad and low-relief on the shallow dipping marine shelf with limited syn-sedimentary folding observed.However, thickness variations are seen where outcrops thin and appear to pinch out towards pre-existing topographic highs inherited from the underlying structural configuration of the Jurassic (Rey et al., 1986;Luber et al., 2019).
The lowermost formation of the Cretaceous is the Agroud Ouadar Formation, dated as Lower to Upper Berriasian, and consists of silty limestones and marls rich in echinoids, oysters, bivalves and brachiopods.This is followed by the Sidi L'Housseine Formation that is constrained to Lower Valanginian to lowermost Hauterivian age and represented by marls interbedded with bioclastic limestones.Towards the top of the formation, the beds become siltier and thin sandstones are observed.Overlying this is the Tamanar Formation, of strictly a Lower Hauterivian age, made up of massive coral-rich limestones followed by R. Charton et al. sandy bioclastic limestones and laminated calcareous sandstones.The following succession, the Talmest Formation, consists of marls interbedded with bioclastic and silty to sandy limestones rich in oysters, becoming increasingly more marl-dominated vertically and is aged as Upper Hauterivian.The Lower to Upper Barremian-aged Taboulaouart Formation above is quite similar in lithology, being made up of marls interbedded with bioclastic limestones.The Bouzergoun Formation lies above this and is dated as Upper Barremian to lowermost Aptian.It starts with thick shelf-derived mudstones and minor sandstones which are then cut by a clear ravinement surface containing shallow marine sandstones and coarse fluvio-deltaic sandstones and terminates with interbedded sandstones, sandy limestones and marls.Overlying the Bouzergoun are the Tamzergout (Aptian) and Oued Tidsi (lower Albian) Formations respectively, which are dominated by marls and occasional bioclastic limestones.

Late Cretaceous
A significant increase in subsidence in the Cenomanian (Zühlke et al., 2004) resulted in a basin-wide transgression represented by marls and limestones of an outer shelf environment (Ait Lamine Formation; Fig. 1b).The transgressive trend continued in the start of the Turonian in which organic-rich, basinal, black shales are recorded (Ferry et al.,

2007
).The latter part of the Turonian shows a return to shallower platform conditions indicated by dolomitic calcarenites (Ferry et al., 2007).This regression continued into the Coniacian (sandy dolomites, silts and evaporitic deposition) and the Santonian (shallow water limestones and marls in the west that move into deltaic silts and sands and evaporates in the east; Zühlke et al., 2004).The end of the Cretaceous is marked with a transgression between the Campanian to Maastrichtian.The Campanian consists of inner to outer shelf derived dolomites and marls and the overlying Maastrichtian is represented by shallow marine limestones and marls, which in a proximal direction shift to phosphate-enriched siliciclastic sandstones (Zühlke et al., 2004).

Mesozoic growth of the anticlines
In the literature, Jurassic to Cretaceous paleo-highs, evaporite mobilisation and/or structural folding in the EAB has been suggested and locally documented (Rey et al., 1988;Stets, 1992;Bertotti and Gouiza, 2012;Luber, 2017;Jaillard et al., 2019).In the Tidsi anticline, the presence of a diapir has been imaged on seismic (e.g., Hafid et al., 2000) and it is now well documented that halokinesis started as early as the Early Jurassic (Hafid et al., 2006(Hafid et al., , 2008)).However, the anticline lies in the basinal domain of the EAB (Frizon de Lamotte et al., 2008), as opposed to the other anticlines exposed in the WHA, mostly because of an N-NE/S-SW fault east of the Tidsi anticline (East Necnafa fault; structure directly east of the anticline in Fig. 1a; Frizon de Lamotte et al., 2008).Field evidence, such as sedimentary wedges in Lower Cretaceous rocks suggests salt accommodate horizontal compression active in the Jurassic (Bertotti and Gouiza, 2012).
A recent study in the Jbel Amsittène, one of the Western High Atlas anticlines, reveals the fold grew during the Late Jurassic and Early Cretaceous by N-NW/S-SE horizontal shortening (Fernández-Blanco et al., 2020).Their detailed structural analysis leads to the characterization of the anticline as a fold-propagation fold verging north.This, together with syn-tectonic wedges at outcrop and anticline scales suggest a minimal influence of evaporite withdrawal on the initial growth of the Amsittène, and its early development by contractional tectonics (see Fernández-Blanco et al. (2020) for additional arguments and complete discussion).They concluded that most of the folding that led to the anticline current geometry took place during the Alpine contraction.
The Imouzzer anticline, oriented SW/NE, is part of a larger E/W anticlinorium that extends from the Triassic Argana valley to the Atlantic coast and also contains the Cap Rhir and Anklout anticlines (Fig. 1c).The orientation of the Imouzzer has been attributed to basement structures, possibly rift-related (Ager, 1974).Cap Rhir has been more recently studied (Martin-Garin et al., 2007;Duval-Arnould, 2019) and analyses of Jurassic facies reveals the presence of a tilted block as a paleo-topographic constrain for the growth of a reef.Extensive facies analyses of Lower Cretaceous sediments around Cap Rhir suggests that it was forming a paleo-high during the Barremian-Aptian (Luber, 2017).
The Jurassic facies of the Imouzzer, Anklout and Lgouz anticlines were studied and added to a review of previous works in Duval-Arnould (2019).The orientation of later two anticlines is the result of that of basement structures (Ager, 1974).Shortening in the Anklout anticline, extracted from on Jurassic bed length, is estimated at ~10% (Frizon de Lamotte et al., 2000).The Lgouz and the Amsittène anticlines display an overturned southern and northern limb, respectively (Ambroggi, 1963;Ager, 1974).

Material and methods
The geologic horizons and markers were mapped out using previous cartographic information around the Amsittène and Imouzzer anticlines, and their geospatial position defined using Google Earth integrated with satellite images from the Shuttle Radar Topographic Mission (STRM) and its Digital Elevation Model (DEM), of 90 m spatial resolution and c. 16 m vertical error (Hoffmann and Winde, 2010;Rusli et al., 2014).Dip angles and azimuths were reconstructed at km-scale using a 3D plane-solver Google Earth plugin, developed by J. Jamieson and G. Herman (service was interrupted; http://www.impacttectonics.org/GeoTools/3pphelp.html).Stratigraphic boundaries are defined on remote sensing images on the basis of colour changes and interpolating (or combining) the geological contacts extracted from geo-referenced 1:50,000 geological maps (e.g., Jaidi et al., 1970).Selected stratigraphic logs (Table 1), serve as the basis for the defining main siliciclastic fluxes in the basin for the Jurassic and Cretaceous (Duval-Arnould, 2019; Bryers, in progress; respectively).To check the quality of this data set and complement it, field measurements (contact coordinates and attitudes) were taken locally from both anticlines studied (Tables 2 and 3).This allowed correction of the DEM in Google Earth, which has been noted by previous authors to not always yield accurate results for steep slopes (Richard and Ogba, 2016).Geometrical consistency of structural cross-sections was tested using two dimensional restorations (Dahlstrom, 1969;Elliot, 1983;Mitra and Namson, 1989;Durand-Riard et al., 2010), using the Move 2D software from Midland Valley.This study preserves the thickness of marine-dominated rocks and uses flexural slip unfolding (Moretti, 2008) to rotate fold limbs to a horizontal datum or an assumed regional surface.Flexural slip is removed during rotation of fold limbs by layer-parallel shear and according to a defined fold axis (based on the orientation of the beds in each of the fold flanks).During flexural slip unfolding, line length is preserved and true bed thickness is constant, leading fold area conservation in the section direction (Moretti, 2008), which is likely different but somewhat similar than the unfolding direction.However, it is important to note that the unfolded horizons and related sediments are not uncompacted.
For the sedimentary logs collected from and around the anticlines, the identification of the carbonate facies is based on the Embry and Klovan (1971), extension to the Dunham (1962) classification.The siliciclastic descriptions are based on textural classification, the grain size grades follow the scale defined by Wentworth (1922) and the textural classes from Folk (1980).Finally, a correlation is constructed between the Jurassic sediments of the Imouzzer and Amsittène anticlines by using two exploration wells (Fig. 1; TMS-1 and TMR-1).These two wells are located at the crest of an anticline of lesser magnitude, which is part of the larger Haha syncline (e.g., Piqué et al., 1998).(Duffaud, 1964).See Table 2 for dip measurements of the Mesozoic sedimentary rocks.The stereoplot displays the field measurements in purple and the digital ones in black.Photographs from the marked location are displayed in Fig. 3 2 and Fig. 4, but only their average, given their close proximity, was used (022/19).

Amsittène
The ENE-WSW-striking Amsittène Anticline is a c. 40 km-long asymmetric fold (Fernández-Blanco et al., 2020, Fig. 2).The Oxfordian beds in the fold northern limb are overturned in the west of the fold, with S-SE dips up to 40 • , and dip 20 • -30 • to the N-NW in the east of the structure (Table 2).In combination with the constant dip angles of 25-30 • S-SE along the southern flank, we constrain the axial plane that dips c. 85 • towards the S-SE.Halokinetic effects are only observed in the west of the anticline, where a diapiric Triassic evaporite outcrops leads Jurassic layers to dip away from it.Thickness variations in the Middle Jurassic (also called Dogger in the area, although it only refers to Aalenian to Bathonian stratigraphy) to the Tithonian formations along lines following the scalped anticline show an overall thickness decrease from the northern flank towards the east (Fig. 2c; already published in Fernández-Blanco et al., 2020).From the 2D restoration, we recover a tectonic shortening of 7.5-13% along the cross-section direction (perpendicular to fold axis) that is consistent with previous estimates (Fernández-Blanco et al., 2020).
In the southern limb, all formations progressively increase in thickness to the west (Fig. 2), which is explained by the fact that the EAB develops as an overall carbonate ramp (e.g., Jaillard et al., 2019).The Callovian, the Oxfordian and the Kimmeridgian thicken westward along the southern limb to (with thicknesses changing from c. 50 to 140-200 m, c. 140 to over 300 m, and c. 80-100 m respectively).Profiles A and B show northward thickening for all horizons, when restored to the Oxfordian (Fig. 2a).
On the northeastern side of the anticline at the marked location (Fig. 3; Fig. 2), a cliff of c.15 m of Aalenian to Callovian limestones, shows discontinuous layers forming onlaps (Fig. 3).Onlaps occur towards the south, i.e., the centre of the anticline.These onlaps result in a thickening of more than 5 m over a distance of about 50 m towards the north (Fig. 3b).This results in a change in dip from 25 • in the upper section to 15 • in the lower section.

Imouzzer
The NE/SW striking Imouzzer periclinal anticline is c. 15 km long with two closures (in the SW and NE; Fig. 4), with Jurassic and Cretaceous rocks folded asymmetrically (Table 3).Thick layers of evaporite (gypsum) are deforming or locally interbedded in carbonate layers of the Toarcian (Fig. 5; Ambroggi, 1963).Syn-sedimentary onlap structures, normal faulting and asymmetric sedimentary thinning, as observed in the field, suggest active evaporite diapirism during the Early and Middle  (Ambroggi, 1963;Adams et al., 1980).Jurassic.Thicknesses in the Aalenian-Bathonian sediments vary slightly along the anticline axis and range from c. 150-330 m (Fig. 4c).The thickness of Callovian rocks average 80 m and remains relatively constant throughout the profiles, except for thickening near evaporite outcrops.Thus, this horizon does not follow the thickness trends observed for Aalenian to Bathonian rocks (Fig. 4c).No thickness variations were calculated for Oxfordian and Kimmeridgian stratigraphic units, since their data points are too few and far apart to provide meaningful results.
Results of the structural geometry show a clear difference in the deformation on the Lower Jurassic unit, especially when compared to that of the Aalenian-Bathonian (Fig. 6).Unfolding of NW/SE crosssections show that shortening varies between ~9.5 and 6%.Evaporite layers are dipping NE following the morphology of the fold core.Deformation of the Toarcian carbonate layers interbedded in the evaporite are found all along the breached valley, showing multiple folding structures (Fig. 5a and b).The thickness of some evaporite bodies can be assumed to amount at least to 50 m where the base of the evaporite could not be located.
The Callovian carbonates, fairly constant in thickness, show little horizontal facies variation where exposed in Imouzzer, and a homogeneous evolution vertically.This suggests that Callovian-Oxfordian rocks were deposited horizontally atop a peneplained Lower Jurassic anticline, however no field evidences of emersion or erosion were observed.In the Northern part (Fig. 5c), Aalenian-Bathonian sedimentary rocks dip 68 • N-NW, forming a conformable contact with the Callovian.The limestone layers within the Middle Jurassic are at an angle of approximately 30 • with the Callovian to the north, meaning that the Aalenian-Bathonian siliciclastics between the Middle Jurassic carbonates might have 'corrected' this structural dip when the Callovian was deposited, potentially due to a pause in the evaporite mobilisation.
The marly to sandy deposits of the Middle Jurassic, with thicknesses of 200-250 m, are strongly deformed.Folding is observed at the base whereas towards the top deformation features are of a faulted nature, showing normal faulting and smaller fold structures (Fig. 5d).Subtle changes (cms) in thicknesses, from the layers shown in Fig. 5d, from S to N across a normal fault might indicate syn-sedimentary deformation.

Correlation between the anticlines
Unfolded sections in the WHA anticlines points towards synsedimentary thickness variations of Jurassic sediments at these locations.To shed light on the evaporite activity in the platform domain of Fig. 7. | South-north well correlations between Amsittène and Imouzzer anticlines.The anticline core series flattened to the top-Callovian as it is a clear geological marker in all studied areas and with present-day elevation profile (inset).The two wells used were drilled on a smaller anticline where the Jurassic is partially exposed (Fig. 1c).
the EAB, it is important to understand the regional thickness changes, especially away from, and in between, the anticlinal structures (Fig. 1c).A strong increase in thickness in between anticlines could indicate the formation of mini-basins due to active evaporite mobilisation during forming anticlines.It is noteworthy that the wells were drilled atop a smaller anticline than Amsittène and Imouzzer.However, assuming that the size of the presently exposed anticlines is an indication of their importance in the Jurassic, there could be some substantial thickness variations.
Therefore, two wells located in the northern portion of the 50 km wide Haha syncline have been used to create a ~N/S correlation for the Triassic to Kimmeridgian (Fig. 7).The stratigraphic logs of the two wells, namely Tamanar 1 (TMR-1) and Timsiline 1 (TMS-1), were provided to the present authors by the OHNYM.The correlation panel shows no large thickness gradients, indicating that there was no substantial diapir forming the two anticlines, i.e., ones that would be comparable to the EAB offshore or the Tidsi anticline (Fig. 1), nor minibasins forming in between the above studied anticlines.Instead, the Aalenian-Bathonian, Callovian and Oxfordian stratigraphic units are surprisingly constant in thickness.Only the Lower Jurassic (also referred to as Lias) is substantially thickening away from the Amsittène anticline towards the south (from ca.350-800 m), although information on the Lower Jurassic thickness is missing from the Imouzzer anticline.Still, Early to Middle Jurassic stratigraphic units thin towards the anticlines (where they locally have a fairly consistent thickness; see previous parts), which is in line with a pre-Alpine growth of the folds.

During the middle and Upper Jurassic
Along the Imouzzer and Amsittène anticlines, variations in the amount of siliciclastic flux have been recorded (Duval-Arnould, 2019, Fig. 8).The Aalenian to Bajocian/Bathonian period is marked by the deposition of thick continental and marine siliciclastics in the East and South of the EAB (Jurassic "Sa" in ASH; Safor Siliciclastic Ameskhoud Fm; see Table 1 for coordinates).The north-western part of the basin (Amsittène Anticline) in the same period is dominated by carbonates and evaporites.These lateral facies variations indicate an E/W or SE/NW siliciclastic system orientation during this period.During the Callovian, the siliciclastics flux in the basin decrease, and the sedimentation is dominated by carbonates until the Middle/Upper Oxfordian.
anticline.Potential evaporite movements might have occurred prior to the Callovian stage to create this high.The high could also have acted as a topographic barrier to siliciclastic deposits coming from the north during the same period (Jurassic "S1" in AME).The Callovian siliciclastics, associated to the siliciclastic pulse "S1"would have deposited in this shallower domain, and the topographic barrier could have prevented them to reach the other side of the anticline.This would have induced a re-routing of the Callovian siliciclastic sediments coming from the north east of the basin (Jurassic 'S1' in well NDK-2) towards the offshore to the west, guided by the northern flank of the Amsittène anticline.Dating of dinoflagellates cysts place the siliciclastic pulse "S1" during the Upper Callovian (Jaydawi et al., 2016).We can note that the well MKL10, initially dated Callovian to Upper Oxfordian has been reattributed to the Lower Callovian (Jaydawi et al., 2016) and the absence of siliciclastics in this particular well is therefore expected (Jurassic "S1" in MKL10).
The north and south of the Imouzzer anticline present fairly similar deposits thicknesses during the Callovian (Duval-Arnould, 2019).The absence of the siliciclastic influx "S1" along this anticline can be noted in locations, ASH, IMZ and TID (Fig. 8).The Callovian facies are homogeneous along the anticline and followed by coral build-ups during the Oxfordian.This coral-rich unit is then interrupted in the north of the anticline by cross-bedded siliciclastic deposits.
Along the Imouzzer Anticline, the NW flank records siliciclastic deposits during the Middle to Upper Oxfordian (Fig. 8, Jurassic "S2" in ASH), which are absent along all the eastern flank of the anticline and in Fig. 10.| Conceptual model for the evolution of the a) Amsittène and b) Imouzzer anticlines since the Jurassic.Note that 1) Triassic evaporite presence in the Imouzzer anticline is inferred, 2) this inferred Triassic evaporite was mobile during the Lower and Middle Jurassic, and would either stop before the Callovian or the influx of siliciclatic of the Ameskhoud Fm was so important that it covered completely its topographic expression, and 3) Toarcian evaporites in the Imouzzer anticline were heavily deformed during the initial fold growth.
the SW part of the anticline (TID and IMZ).In the Amsittène anticline, a comparable Oxfordian siliciclastic influx is observable along the northern and southern flanks (Jurassic "S2" in AME and AMCA).This siliciclastic influx can be traced further to the North and East, in the wells NDK-2 and NDK-3 (Fig. 8b, Jurassic "S2" in NDK2 and NDK3).This siliciclastic influx "S2" takes place at the same stratigraphic position in the Amsittène and Imouzzer anticlines and have been therefore interpreted to be due to the same event linked to the exhumation of the Meseta (Charton et al., 2020).
The siliciclastic deposits on the north of the Imouzzer anticline and in the Amsittène anticline indicate high energy settings, potentially linked to their location on topographic highs.In the locations AMCA and ASH, the higher energy settings persist after the siliciclastic influx, and carbonate grainstones progressively replace the sandstones, before the onset of the generalized regression and the development of tidal-flat deposits all over the basin.In other locations, the transition is more abrupt and there is no record of high-energy transition facies between the deeper coral-rich unit and the tidal-flat deposits.

During the Lower Cretaceous
The location and timing of Lower Cretaceous siliciclastic influx and their pathways are presented in the work of Luber (2017;Fig. 9).The first period of clastic influx is documented several kilometers inland of the Cap Rhir anticline (Fig. 1c; Cretaceous "S3"; see AH), and dated as Early Valanginian by recent biostratigraphy (Bryers, pers comm.), part of the Sidi L'Housseine Fm.The sandstones are interpreted as marine shoreface deposits with outcrop data indicating a N-NW/S-SE palaeoflow direction.The interval varies in thickness from 6 to 40 m (Ain Hammouch to Tamzargout localities).This sandstone interval is not documented north of Cap Rhir or on either side of the Amsittène anticline.Approximately 10 km northeast of Cap Rhir, a very thin interval (several 10s of cm thick beds) of calcareous sandstones is recorded interbedded with marls.They are very fine to fine grained, have flute mark scours on their bases and show no internal bedding structures.These sandstones are interpreted as minor confined shelfal turbidite flows, and the interval dated as Upper Valanginian (based on superposition amongst ammonite dated beds) is not recorded elsewhere in the EAB.
A second influx of clastics is recorded during the Lower Hauterivian (Cretaceous "S4"; constrained by ammonite biostratigraphy; Bryers, pers comm.).This incursion is found more frequently across the EAB than the Lower Valanginian episode, documented 2-3 km southeast of the Cap Rhir anticline (AH, Tamzargout Fm) and slightly north of it (see log IG; Fig. 9) and is also found adjacent to the Amsittène anticline (AMS-1) although strata preservation is quite poor due to alpine deformation.The facies is interpreted as a marine shoreface environment with a N-NW/S-SE palaeoflow and interval thickness varies from less than 2 m (AMS-1 log) to ca. 10 m (Ain Hammouch Fm).
A third, and particularly notable period of clastic input occurred during the Upper Barremian to Lower Aptian (Sbfor Siliciclastic Bouzergoun Fm; dated by recent biostratigraphy Luber et al., 2019).These are interpreted as fluvio-deltaic to shallow marine deposits that incise into the underlying Lower Cretaceous stratigraphy, forming a regional erosive unconformity as a result of a forced regression of the shoreline (Luber et al., 2019).Thickness variations of the Bouzergoun Fm (ca.10-80 m) and variabilities in measured palaeocurrent directions suggest the influence of palaeohighs.For example, palaeocurrent data north and south of Cap Rhir indicate the local palaeohigh may have diverged sediments moving towards the basin into multiple transport pathways as well as providing additional local sediment input (Luber, 2017).

Evaporite mobilisation and mechanism
Early Mesozoic evaporite mobility is interpreted to have been instrumental in the formation of both anticlines from the Early and into the Middle Jurassic (Fig. 10).These evaporites were overlain by Late Jurassic carbonates which show systematic N-S and E-W thickness variations (Figs. 2 and 4).In the Amsittène anticline, Triassic evaporite mobilisation occurs from the Early to Middle Jurassic (Fig. 10a), and in the Imouzzer anticline (Fig. 10b) the Triassic and Toarcian (?) evaporites formed a bulge during the Early and Middle Jurassic (Fig. 11a), which strongly deformed the two sedimentary units and possibly led to crestal erosion with carbonate shedding.
In the Agadir-Essaouira Basin (EAB), folds expressed on the sea-floor led to the formation of large carbonate reefs and reef foresets (Fig. 11a; Duval-Arnould, 2019).Relatively-thin evaporites of Toarcian age interbedded in carbonate layers are in good agreement with the observations of thin evaporites causing structures dominated by narrow box-fold anticlines (Hudec and Jackson, 2007).The chaotic/tight folding patterns of evaporite deformation in the core of the Imouzzer anticline (Fig. 5a and b) could suggests that deeper, potentially Triassic, evaporite were mobile in the Lower to Middle Jurassic, especially during the deposition of the thick Aalenian-Bathonian Ameskhoud Fm.
This paper constrains the evolution of the Amsittène in the Early to Middle Jurassic and support that its growth is associated with halokinesis.Fernández-Blanco et al. (2020) constrains the evolution of the Amsittène in the Late Jurassic to Early Cretaceous, and support that its growth is associated with shortening tectonics.Irrespective of the triggering mechanism(s) for the growth of the folds, their effect is to partly compensate for the regional subsidence experienced by the EAB.This is justified by the fact that none of the stratigraphic intervals reviewed here document emersion, meaning the anticline tops never came above sea level in the Mesozoic.In other words, these regions only experience less subsidence than their surroundings.
Early and Middle Jurassic evaporite mobilisation in the Central High Atlas has been inferred from sedimentary onlaps described by Saura et al. (2014), Martin-Martin et al. (2017), and later reviewed in Moragas et al. (2018).While the evaporite mobilisation mechanisms in the offshore EAB may differ from those onshore, timing of early evaporite mobilisation is similar (Neumaier et al., 2016;Pichel et al., 2019, Fig. 11b).N-NW/S-SE to N-NE/S-SW regional shortening is suggested for the Tidsi and Amsittène anticlines (Bertotti and Gouiza, 2012;Fernández-Blanco et al., 2020;respectively) to explain subsequent Middle Jurassic to Early Cretaceous syn-sedimentary deformations and anticline growth in the EAB and the exhumation in the basin hinterlands.
In the Lusitanian Basin (Portugal), salt diapirism has been identified during the upper Jurassic, which induced the formation of mini-basins.The structure of these mini-basins have been constrained through seismic stratigraphy and show important lateral thickness variations and onlaps against the salt structure during the Upper Jurassic, and minor evidences during the Early Cretaceous (Alves et al., 2002(Alves et al., , 2003)).Similarly, to the EAB, onlap geometries and quick lateral bedding dip changes have also been pointed out along some diapirs as markers of halokinetic movements (Davidson and Barreto, 2020).The Lusitanian Basin underwent similar evolution to the offshore EAB segment, which recorded halokinetics movements during the Middle Jurassic to Lower Cretaceous (Pichel et al., 2019).
During the Jurassic, coeval vertical motions resulted in subsidence of the EAB (e.g., Bertotti and Gouiza, 2012) and erosional exhumation of surrounding massifs, especially in the Anti-Atlas domain (Fig. 11b; e.g., Ghorbal, 2009;Gouiza et al., 2017;Charton et al., 2018), thereby leading to a topographic gradient (assuming exhumation is linked to uplift), and possibly a hydraulic head (salt/evaporite considered here as a fluid) gradient within the Triassic evaporites.Initially, mobile evaporite would then migrate N-NW, and traveling upward above structural features such as E/W Atlas rift related faults.Consequent evaporite lateral flow in the basinal domain of the EAB is interpreted to be one of the mechanisms that initiated the concomitant Jurassic folding and the migration of evaporite diapirs distally from east to west (Hafid et al., 2006;Tari and Jabour, 2013).Superimposed to the passive margin thermal subsidence, this is reinforcing the thickening of Jurassic successions towards the south-west, from near zero in the Argana Valley to ~1 km in the continental shelf break, (Zühlke et al., 2004).The Cretaceous series similarly thickens westward, from ~750 m east of the EAB to ~2.5 km (Zühlke et al., 2004).Westward thickening of Mesozoic successions corroborates the westward (High Atlas uplift; Stets, 1992) slope that may thus have led to lateral evaporite mobility and fold growth.Conversely, a potential northward slope caused by a Jurassic uplift in the Anti-Atlas (e.g., Charton et al., 2018), south of the basin, would better explain the E/W fold growths that are observed here.
Alternatively, and although we favour the above-presented mechanism, evaporite mobilisation may be due to substantial temperature gradients in the evaporite bodies (i.e., thermal loading; Hudec and Jackson, 2007).Processes that could lead to such gradient include: i) extensive magmatic activity in the Late Triassic (Central Atlantic Magmatic Province; e.g., Davies et al., 2017) following the end of rifting, ii) by reactivation of ~NE/SW basement normal faults (e.g., Hafid, 2000;Piqué and Carpenter, 2001), as described for the offshore areas further north in Morocco (Neumaier et al., 2016), iii) significant changes in the supra-evaporite thickness, and/or vi) different lithosphere structure within the basin.

Relations with source-to-sink systems
Sediment transport, channel feeder system location and depocenter location (Fig. 11a) are likely influenced by the surficial expression of syn-depositional folding and generation of paleo-topography in the EAB.Potential source-to-sink systems in the EAB are illustrated in Figs. 12 and 13, assuming that i) evaporite-cored anticlines were acting as siliciclastic flux barriers, ii) the EAB was characterised by an overall carbonate ramp topography, and iii) the few control points available for some time intervals are sufficient to draw tentative source-to-sink system (e.g., Fig. 12d and e).Figs. 12 and 13 are similar in nature and differ only for the scale at which they were constructed and for the illustrated periods (exploration well data used in Fig. 13 are presented in Table 4).
In the Central High Atlas, Lower and Middle Jurassic diapirism has been extensively studied (Fig. 11;Verges et al., 2017;Teixell et al., 2017;Moragas et al., 2018), where the structures observed show a strong influence of the diapirism on the sedimentation in the corresponding mini-basins.Compared to this area, the diapirism impact on the sedimentation in the EAB is restricted and imply more subtle salt movements during this period.
In the onshore EAB, the end of the Middle Jurassic (Bathonian-Callovian) seems to have recorded the first important halokinetics movements.This is observable in the Imouzzer anticline (Fig. 10b), with the angles between the Middle Jurassic carbonate layers (Fig. 5c); and in the Amsttene Anticline (Fig. 10a), with the onlap structures observed in the north of the anticline.Evidence of halokinetics movements during the Upper Jurassic and Lower cretaceous are harder to identify or disprove due to the constant erosion of these units along the anticline flanks onshore the EAB.
Middle Jurassic-Early Cretaceous evaporite mobility (Fig. 11), marked by thickness variations in the up to 10 km wide mini-basins, suggests deep-water sinks may have forced differential loading, thus driving evaporite mobility (Fig. 13).Most eroded material was transported over the rifted margin, south into the offshore EAB (Agadir segment; Pichel et al., 2019), where turbiditic deep-water facies were deposited in the Middle Jurassic and Early Cretaceous (Fig. 13b and c).The Early Cretaceous fluvial systems that sourced detritus and lead to a sedimentary succession of c. 7 km atop the Triassic evaporite (Pichel et al., 2019) are preserved and exposed in the onshore EAB (Luber, 2017;Luber et al., 2019).

Conclusions
This study sheds new light onto the controversial question about whether evaporite-cored folds in the EAB are shaped by Mesozoic evaporite mobilisation or Cenozoic compressional forces.The results show that anticline growth was already active during the Early to Middle Jurassic and, before the Atlasic orogeny, and therefore likely due to evaporite mobilisation.Mobilisation of Triassic evaporites layers was most likely initiated by Jurassic exhumation in the hinterland, developing a topographic gradient, and associated differential subsidence in the EAB onshore and offshore (e.g., Ameskhoud and Bouzergoun Fms for which depocenters were dependant on the sediment entry point(s)), causing a hydraulic head gradient that led to lateral evaporite flow).
The observed E-W folding is interpreted to be the results of a gravitational gradient caused by the Jurassic uplift of the Anti-Atlas to the south of the basin.In the Berriasian-Aptian, the Cap Rhir N/S trending paleohigh (depicted in Luber, 2017) could be linked to uplift in the High Atlas massifs to the east.No supporting evidence can be found for the proposed regional shortening in the Jurassic, postulated in previous studies.
As a consequence of early fold growth, the syn-sedimentary structures controlled sediment discharge pathways, and the route of fluvial systems coming from the exhumed Variscan massifs in the hinterland, being shed towards the offshore basin.We postulate that the location of depocenters that may contain deep-water fan sandstones of Jurassic to Early Cretaceous age are in part dependent on the location of the fluvial entry points along the coastal margin, that were controlled by the evaporite-cored anticlines in the EAB that diverted paleoflows.

Fig. 1 .
Fig. 1. | a) Map of evaporite structures in the Agadir-Essaouira Basin (EAB) in relation with faults and folds (afterTari et al., 2012).*Seismic lines presented and interpreted inPichel et al. (2019).b) Mesozoic litho-stratigraphic column (afterChoubert, 1957, Michard et al., 2008;Luber, 2017;Mader et al., 2017;Duval-Arnould, 2019).Red letters alongside the formation names are the references used in the present work for the unfolding of the cross-section.c) Geological map of the area (afterHollard et al., 1985) showing the Western High Atlas anticlines (numbered 1 through 6).Names and precise location of the logs are presented in Table1.(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 2 .
Fig. 2. | a) Geological map of the Amsittène Anticline based on Google Earth mapping and Tamanar geological map(Duffaud, 1964).See Table2for dip measurements of the Mesozoic sedimentary rocks.The stereoplot displays the field measurements in purple and the digital ones in black.Photographs from the marked location are displayed in Fig.3.* Field observations in Fernández-Blanco et al. (2020).b) Cross-sections A, B, and C, running perpendicular to the fold axis, without vertical exaggeration.c) Resulting thickness maps for Middle and Upper Jurassic units using the Paradigm Gocad® 'kine3d-1' tool (maps are published in Fernandez-Blanco et al., 2020; see details therein).The resolution of the thickness maps suggests a precision of around 20 m and lies below the thinnest formations modelled in the anticline.Note that each map displays the thickness along a line where the formation is outcropping with a similar line thickness for visualisation purposes only.(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Fig. 2. | a) Geological map of the Amsittène Anticline based on Google Earth mapping and Tamanar geological map(Duffaud, 1964).See Table2for dip measurements of the Mesozoic sedimentary rocks.The stereoplot displays the field measurements in purple and the digital ones in black.Photographs from the marked location are displayed in Fig.3.* Field observations in Fernández-Blanco et al. (2020).b) Cross-sections A, B, and C, running perpendicular to the fold axis, without vertical exaggeration.c) Resulting thickness maps for Middle and Upper Jurassic units using the Paradigm Gocad® 'kine3d-1' tool (maps are published in Fernandez-Blanco et al., 2020; see details therein).The resolution of the thickness maps suggests a precision of around 20 m and lies below the thinnest formations modelled in the anticline.Note that each map displays the thickness along a line where the formation is outcropping with a similar line thickness for visualisation purposes only.(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 4 .
Fig. 4. | a) Geological map of the Imouzzer Anticline based on Google Earth mapping and field observations.See Table 3 for dip measurements of the Mesozoic sedimentary rocks.The stereoplot displays the field measurements in purple and the digital ones in black.Photographs from the marked locations are displayed in Fig. 5 b) Cross-sections D, E, and F, running perpendicular to the fold axis.C) Resulting thickness maps for Middle Jurassic and Callovian units using the Paradigm Gocad® 'kine3d-1' tool.(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 11 .
Fig. 11.| a) Highly idealized channelization of sediment transport in a shallow marine mixed siliciclastic and carbonate environment due to fold growth and b) timing of evaporite mobilisation and exhumation in surrounding massifs (see references therein).

Table 1 |
Location of the stratigraphic logs used in this work(after Duval-Arnould, 2019 and Bryers, pers comm.).

Table 2 |
Digital (Google Earth) and field dip measurements of beddings from the Amsittène anticline.Dogger: Aalenian-Bathonian.The elevation has been obtained using the GPS coordinates and combined with NASA SRTM3 DEM.Overturned strata are highlighted in grey.*average of four field dip measurements from the location shown in Fig.3a.Stratigraphy Dip angle [ • ] Dip azimuth [ • ] Limb Latitude [dec • ] Longitude [dec • ] Elevation [m] Acquisition R.Charton et al.

Table 3 |
Digital (Google Earth) and field dip measurements of beddings from the Imouzzer anticline.The elevation has been obtained using the GPS coordinates and combined with NASA SRTM3 DEM.