Neogene polyphase deformation related to the Alboran Basin evolution: new insights for the Beni Bousera massif (Internal Rif, Morocco)

Asmae El Bakili, Michel Corsini, Ahmed Chalouan, Philippe Münch, Adrien Romagny, Jean Marc Lardeaux and Ali Azdimousa 1 Université Côte d’Azur, CNRS, Observatoire de la Côte d’Azur, IRD, Géoazur, 250, rue Albert Einstein, 06560 Sophia Antipolis, France 2 Université Mohammed V, Faculté des Sciences Rabat, 4, avenue Ibn Batouta, B.P. 1014 Rabat, Maroc 3 Université Montpellier 2, Géosciences Montpellier, UMR 5243, CC 060, place Eugène Bataillon, 34095 Montpellier cedex 5, France 4 Centre for Lithospheric Research, Czech Geological Survey, Klárov 3, 118 21 Prague 1, Czech Republic 5 4, rue du Lavoir, 06340 Drap, France 6 Université Mohammed premier, Faculté des Sciences, BV Mohammed VI–BP 717, 60000 Oujda, Maroc


Introduction
The Gibraltar Arc is an arcuate orogenic system located in the extreme tip of the western Mediterranean Sea (Fig. 1). This orogen is composed of the Betics and the Rif mountain ranges, which consist mainly of stacked tectonic units thrust toward the Guadalquivir and the Gharb foreland Basins, in southern Spain and northern Morocco respectively (Fig. 1). The Betic-Rif Internal domain (Alboran Domain) includes the Ronda and Beni Bousera subcontinental peridotites, in the Spanish and Moroccan branches of the arc, respectively (Balanya and Garcia-Duenas, 1987).
This orogenic system is edified along the active margin of the western Mediterranean, as the final consequence of the Africa-Eurasia convergence accommodated since the Late Cretaceous by the subduction of the oceanic lithosphere (Dercourt et al., 1986;Stampfli, 2000;Michard et al., 2002;Verges and Fernandez, 2012;Platt et al., 2013).
The Alboran Basin opening was accompanied by granitic dykes intrusion within the peridotites and the underlying and overlying crustal metamorphic units from the Internal Domain (Priem et al., 1979;Zeck et al., 1989;Rossetti et al., 2010). Therefore, the understanding of the tectonic context at the time of the granitic dykes intrusion may reveal information on the timing and the mechanism of emplacement of the so-called subcontinental peridotites into the crust, a subject that remains a warm debate for decades (e.g., Kornprobst, 1976;Sánchez-Rodríguez and Gebauer, 2000;Afiri et al., 2011;Sanz de Galdeano and Ruiz Cruz, 2016;Bessière, 2019). However, the tectonic scenario related to these granitic dykes has to take into account the subsequent deformations that affected the Internal domain, such as block rotations related to the Gibraltar Arc tightening.
In the internal Rif (the southern part of the Alboran Domain), this event of granitic intrusion has been related to crustal thinning according to different tectonic scenarios Ouazzani-Touhami and Chalouan, 1995;Rossetti et al., 2013;Romagny, 2014). In these contrasted proposed scenarios, the importance of the subsequent deformations following this magmatic event has not been studied.
The present work aims at constraining the polyphase deformation that affected the Alboran Basin and the Gibraltar Arc since the Early Neogene using structural and petrological analyses, and 40 Ar/ 39 Ar geochronology in the Beni Bousera area ( Figs. 1 and 2). These new data are compared with data from equivalent geological systems in the northern branch of the Gibraltar Arc (Betics cordilleras), and new issues related to the peridotites emplacement mechanism and timing will be addressed.
The Beni Bousera massif forms an open NW-SE antiform with the peridotites in the core overlain by the metapelites units (Fig. 2). The lowermost, granulitic metapelites (kinzigites) are associated to the peridotites within the Beni Bousera unit (Kornprobst, 1974). This unit is in turn overlain by the Filali unit, which includes migmatitic gneisses at its base followed upward by mica-schists. In the NE, the antiform is crosscuted by the NW-SE trending Aaraben fault, steeply dipping to the NE (Kornprobst, 1974;Reuber et al., 1982;Chalouan et al., 1995). This fault was interpreted as a reverse fault associated with the Beni Bousera major fold (Kornprobst, 1974), then as a normal fault Romagny, 2014).The Aaraben fault crosscuts in its northwestern branch the ESE-WNW Jenane En Nich normal fault dipping 40°to 60°toward the NE . In the southeastern branch, the Aaraben fault is intercepted by the N-S Tararte-Taza fault.
Swarms of granitic dykes intrude the Beni Bousera unit (peridotites and kinzigites) and the migmatitic gneisses of the Filali unit (Kornprobst, 1974;Elbaghdadi et al., 1996;Michard et al., 2006;Rossetti et al., 2010;Fig. 2). These intrusions induced local alteration of the ultramafic rocks (Hajjar et al., 2017). Strong serpentinization of the peridotites, followed by magnesite precipitation, is linked to fluid circulations that occur preferentially along the Aaraben fault (Hajjar et al., 2015(Hajjar et al., , 2016. The emplacement and the cooling of the granitic dykes are constrained at circa 22 Ma using 40 Ar-39 Ar on biotite and muscovite together with the U-Th-Pb method on zircon and monazite from the granite dykes (Rossetti et al., 2013). The chemical signature of the dykes suggests a direct derivation from partial-melting of crustal protoliths, but with chemical compositions that differ from that of Beni Bousera and Filali units (Rossetti et al., 2013). This testifies that the peridotites rest upon a different crustal unit. The granitic intrusions have been related to the Alboran Basin opening as a response to strike-slip tectonics associated with an E-W compression (Rossetti et al., 2013), or alternatively to a radial extension event (Romagny, 2014).

Structural analyses
Structural data (faults with slickensides, ductile shear planes with the stretching lineation, fold axes, and cleavage planes) are collected from stations across three main crosssections (two are orthogonal and one is parallel to the Beni Bousera antiform) (Fig. 2). A total amount of 77 faults and ductile shear zones, 19 cleavages and 4 folds axes were used in this work. The data were handled using the stereonet program (Allmendinger et al., 2012;Cardozo and Allmendinger 2013, updated version 10.2.9, http://www.geo. cornell.edu/geology/faculty/RWA/programs/stereonet.html) in order to obtain their statistical distribution, using their strike and dip.  The determination of the principal stress axes (strike and plunge) and the stress ratio R (R = s2 À s3/s1 À s3) is performed using the right dihedron method (Angelier and Mechler, 1977;Delvaux and Sperner, 2003). The stress axes determination was carried out using scarce but efficient data providing all geometrical faults parameters and good kinematic observations, using (1) fault plane, slickenside and kinematics for faults and (2) shear planes, stretching lineation and kinematics for the ductile shear zones. We considered that brittle-ductile shear zones were suitable for palaeostress analyses by the right dihedron method because we are in conditions where the stretching lineations and the sense of movement can be measured directly on the shear plane and thus the movement vector can be resolved (Eisbacher, 1970;Srivastava et al., 1995;Blewett and Czarnota, 2007).

40 Ar-39 Ar analyses
Two samples of granitic dykes were collected; a sample from a dyke crosscutting the peridotites of the Beni Bousera unit (BB17-60) and a sample from a dyke crosscutting the gneiss of the Filali unit (BB17-116) (Fig. 2). Both biotite and muscovite grains were used for 40 Ar-39 Ar dating the sample BB17-116 and muscovite for the sample BB17-60. Four samples from the Aaraben normal ductile shear zones were collected along the coastal road within small shear zones in the micaschists of the Filali unit (BB15-08; BB15-09; BB15-10; BB15-11) (Fig. 2), from which only muscovite was used for 40 Ar-39 Ar dating. All analyzed muscovites sampled in the core of the shear zones are synkinematic (see Fig. 8).
Samples were crushed and sieved and the selected grain size for the crystals was in the order of 100-200 mm, then they were cleaned and dried. Muscovite and biotite were finally selected under a binocular microscope. Samples were packed in aluminum foil for irradiation in the core of the Triga Mark II nuclear reactor of Pavia (Italia) with several aliquots of the Taylor Creek sanidine standard (28.34 ± 0.08 Ma, Renne et al., 1998) as flux monitor. Argon isotopic interferences on K and Ca were determined by irradiation of KF and CaF2 pure salts from which the following correction factors were obtained: ( 40 Ar/ 39 Ar) K = 0.00969 ± 0.00038, ( 38 Ar/ 39 Ar) K = 0.01297 ± 0.00045, ( 39 Ar/ 37 Ar) Ca = 0.0007474 ± 0.000021 and ( 36 Ar/ 37 Ar) Ca = 0.000288 ± 0.000016. Argon analyses were performed at Géosciences Montpellier Laboratory (France). The gas extraction and purification line consists of (a) an IR-CO2 laser of 100 kHz used at 3-15% power to heat samples during 60 s, (b) a lenses system for beam focusing, (c) a steel chamber, maintained at 10-8-10-9 bar, with a copper holder  in which 2 mm-diameter blind holes were milled, (d) two Zr-Al getters for purification of gases. Two different mass spectrometers were used: a MAP 215-50 noble gas mass spectrometer and a multi-collector mass spectrometer (Argus VI from Thermo-Fisher).
Aliquots of 40 to 50 grains of biotite and white micas were distributed as micropopulation five to ten grain deep in one and two holes of the copper holder, respectively, and were step heated. Blank analyses were performed every three sample analyses. Raw data of each step and blank were processed and ages were calculated using the ArArCALC-software (Koppers, 2002). The criteria for defining plateau ages are: (1) plateau steps should contain at least 70% of released 39 Ar, (2) there should be at least three successive steps in the plateau and (3) the integrated age of the plateau should agree with each apparent age of the plateau within a 2s confidence interval. All the subsequent quote uncertainties are at the 2s level including the error on the irradiation factor parameter J. With the MAP spectrometer the atmospheric contribution was difficult to determine precisely because of a high 36 Ar background on blanks. Thus, errors on individual ages are large (up to 5%) but we choose to present these results because they are complementary to those obtained on the Argus. Raw data can be downloaded from Supplementary Materials.

Granitic dykes
At the outcrop scale, granitic dykes and veins have diverse sizes with plurimetric length and centimetric to decimetric thickness (Figs. 3a-3e). They are mostly emplaced into major fractures, networks of faults and shear zones. Some of them are branched and may isolate blocks from the hosting rocks. A thin alteration envelope often occurs at the contact with the peridotites (Fig. 3). These alterations products consist mainly of magnesite, which offen occurs as veins (3 to 12 cm thick) filing fractures and fault zones, parallel to the granitic dykes (see the paragraph below).
Locally, in the metapelites, granitic dykes or veins were emplaced within normal ductile shear zones as evidenced by asymmetric bending of the foliation on both sides of the dykes (Figs. 3a and 3b). In the peridotites, dykes show evidence of deformation at the brittle-ductile transition, and at the dyke's walls, well-developed S/C planes are again indicative of normal shearing (e.g., Fig. 3c). After their emplacement and cooling, the granitic dykes are affected by several deformation phases (Figs. 3d-3f) and they could have been significantly reoriented. Therefore, we performed analyses only of granitic dykes preserved from late deformation effects, at least at the local scale. Granitic dykes are globally E-W trending (Fig. 5a), with N70°to N140°distribution, and intermediate to steeply dip (40°to 80°). Few N-S and NNE-SSW dykes were observed at the scale of Beni Bousera massif and located mainly in the Oued Amter zone as previously mapped by Elbaghdadi et al. (1996) (Fig. 2) and studied by Rossetti et al. (2010Rossetti et al. ( , 2013. Fractures or faults in which magnesite veins are concentrated were also measured (Fig. 5a), they are trending parallel to granitic dykes with intermediate to steeply dip (35°to 90°). As this alteration product are concentrated in faults parallel to the granite dyke, and they provided the same kinematics, they are consequently considered as generated from the same deformation event. We determine the stress axes strike and plunge using structural elements obtained from these materials  edge and/or from the host rock contact. Results give a N111°, 69°plunging maximum stress (s1), a N308°, 20°plunging intermediate stress (s2) and a N216°, 5°plunging minimum stress (s3), with a stress ration R = 0.44 (s1 vertical, 0.25 < R < 0.75), reflecting a pure extensional regime (Fig. 6a).
Concerning the late deformation events, the granitic dykes are affected by numerous late reverse and normal strike-slip faults (Figs. 3d and 3e; see Romagny, 2014). In the peridotite massif, a nearly NE-SW oriented penetrative spaced disjunctive cleavage (Powell, 1979;Passchier and Trouw, 2005) is observed in the granitic dykes and the peridotites (Figs. 3e and 5b). Afterward, this cleavage is affected by nearly E-W oriented folds (Figs. 3f and 5b).

Aaraben ductile shear zones
The Aaraben Fault was also the object of structural analyses (Fig. 2). The Aaraben fault is a N160°E northeastward dipping fault, marked by a few meters large cataclasites band developed under brittle conditions (Fig. 4a). The mirrors of the lensoid elements broadly parallel to the fault plane bear sub-vertical striae and displays kinematic indicators consistent with normal faulting.
In the north-east of the Aaraben fault (Fig. 2), numerous ductile shear zones, decimeter to a meter wide (Figs. 4b-d) crosscut the Ghomarides and the Sebtides units. All these shear zones constitute a deformation corridor of several hundred meters wide (Fig. 2).    At the outcrop scale, the ductile shear zones are marked by the development of mylonitic foliation into centimeter to meter size deformation bands. The shear zones are also underlined by quartz veinlets (Fig. 4b) with aggregates of chlorite and white micas, reflecting pervasive fluid circulation during the development of the shear zone under greenschist facies conditions.
The ductile shear zones are globally trending NW-SE and moderately to steeply dipping toward the northeast (34°to 70°), (Fig. 5c). The stretching lineation is steeply dipping and shear-sense criteria such as asymmetric rotation of the foliation plane, asymmetric boudins, and S/C planes indicate a northeastward collapse of the hanging wall, compatible with normal ductile shearing (Figs. 4b-4d).

Granitic dykes petrography and chemistry
Petrographic analyses are performed at the thin section scale from granitic dykes crosscutting the peridotites and the migmatitic gneiss. All granitic dykes have a coarse-grained texture, two assemblages are observed; the first assemblage is composed of Kfeldspar, quartz, plagioclase, white mica and tourmaline (Fig. 7a). Additionally, cordierite or biotite are also observed in some samples (e.g. Fig. 7b). The secondary assemblage consists of chlorite after biotite (Fig. 7b), pinite and sericite resulting from cordierite and feldspars alteration respectively. Accessory minerals are zircon and/or monazite and opaques minerals. According to Rossetti et al. (2013), these dykes are classified as biotite-cordierite bearing granodiorite to monzogranite or as tourmaline-muscovite bearing alkali-felspar granite.

Aaraben ductile shear zones
Petrographic investigations performed on the mica-schists from the shear zones allow to precise the mineralogical  assemblages associated with ductile deformation. Microstructures were observed from thin sections perpendicular to the foliation and parallel to the lineation to obtain kinematic constraints. The foliation plane is underlined by the planar disposition of muscovite and biotite as well as polycrystalline quartz ribbons (Figs. 8a and 8b). Biotite and staurolite porphyroclasts are destabilized and wrapped by muscovite and chlorite (Fig. 8d). Numerous shear criteria like microfolded quartz ribbons (Fig. 8a), S/C structures marked by deflections of quartz ribbons, chlorite and muscovites (Figs. 8b-8d) and recrystallized quartz grains in asymmetric pressure shadows around staurolite porphyroclasts (Fig. 8d) are typical of mylonitic rocks deformed by dominant simple shear and the kinematic indicators systematically provide a normal sense of shear.
A biotite single grain (BB17-116B) yields a weighted age at 20.88 ± 0.21 Ma corresponding to 69% Ar released and to eight steps (Fig. 9). The staircase shape of the spectrum may suggest a reheating event younger than 20 Ma that may have slightly affected the plateau age. The inverse isochron for the plateau steps gives a concordant age at 21.31 ± 0.30 Ma (MSWD = 1.93; initial 40 Ar/ 36 Ar ratio of 280.0 ± 8.9) (see Supplementary Material 3). The plateauage at 20.88 ± 0.21 Ma is considered as the best age estimate.

Structural evolution
Structural analyses underlined that granitic dykes are emplaced within major N70°to N140°faults, resulted from an NNE-SSW extension (N216°). This deformation is coeval (21-22 Ma) with the N70°to N150°ductile normal shear zones development which resulted from a nearly N-S extension (N179°). Thus, the stress axes orientation obtained from the shear zones and those obtained from faults related to the granitic dykes form an acute angle of 37° (Fig. 6). This variation could be explained either by different rheological response of the peridotites relative to the metapelites or either by a late effect of the Aaraben fault kinematics.
Indeed, as suggested by Romagny et al. (2014), faults acting under brittle conditions with similar direction as the Aaraben fault are responsible for the present-day uplift of the internal Rif. Moreover, Chalouan et al. (1995) described a vertical displacement of 4-5 km since the Neogene along the Aaraben fault. Our work highlights that most of this vertical movement was accommodated by a large NW-SE trending ductile normal shear zones system acting under greenschist facies metamorphic conditions. Such P/T environment implies an exhumation of the peridotites from at least the ductile-brittle transition. The paleomagnetic data available in the vicinity of the Beni Bousera massif indicate a 76°± 13° (Saddiqi et al., 1995), 72 ± 12° (Berndt et al., 2015) counter-clockwise rotation since the Miocene. Hence the initial orientation of extension was comprised between NE-SW and ESE-WNW at 21 Ma. At that time, the restored subduction trench in front of the Alboran Domain was ∼ N-S (e.g. Spakman and Wortel, 2004;Verges and Fernandez, 2012). Hence the restored direction of extension was roughly normal to the restored subduction trench (Fig. 10a).
Regarding the tectonic context, our interpretation with an E-W extensional regime associated to a non-coaxial deformation contrasts with radial extension associated to the Fig. 9. 40 Ar/ 39 Ar age spectra as a function of 39 Ar released. The error boxes of each step are at the 2s level. The error of the plateau ages (P) is given at the 2s level. Ages were calculated using the Ar-Ar Calc software (Koppers, 2002). Raw data is presented in Supplementary Data. Fig. 9. Spectres d'âge 40 Ar/ 39 Ar en fonction du 39 Ar libéré.
The major extensional event that caused the exhumation of the peridotites and the emplacement of the granite dykes at 20-22 Ma was followed by late compressional events mainly evidenced by NE-SW cleavage and E-W folding.

Geochronological evolution
The muscovites from the two samples of granitic dykes crosscutting the peridotites of the Beni Bousera unit and the gneiss of the Filali unit yielded concordant ages at 21.22 ± 0.42 Ma and 21.55 ± 0.13 Ma respectively. The weighted average at 21.38 Ma is considered as the best age estimate. The biotite yielded an age at 20.88 ± 0.21 Ma which is slightly younger than the muscovites ages, which could be explained by a very rapid emplacement and cooling of the granitic dykes from 21.68 Ma to 20.67 Ma. The structural data highlight that the granitic dykes emplaced in normal faults, reflecting an extensional context (Fig. 3). On the other hand, within the normal ductile shear zones in the Filali unit, all the white micas yielded concordant ages, with a weighted average age at 21.68 Ma. Muscovites from the mica-schists sampled in the core of these shear zones are clearly synkinematic prophyroblasts that have grown or at least recrystallized during the development of the normal shear zones under greenschist facies metamorphic conditions (Fig. 8). Thus, as granitic dykes emplacement and normal ductile shear zones give the same age at circa 21 Ma, we consider that the ages around 21 Ma correspond to an extensional event.
This extensional event occurred during the opening of the Late Oligocene to Early Miocene sedimentary basins of Fnideq and Sidi Abdeslam (Feinberg et al., 1990;Ouazzani-Touhami and Chalouan, 1995;El Kadiri et al., 2006;Serrano and Guerra-Merchán, 2007;Hlila et al., 2008). Biostratigraphic analyses in these basins established the duration of sedimentation between the Late Oligocene and the Middle Burdigalian (Feinberg et al., 1990;Hlila et al., 2008). The deposits unconformably overlay the Ghomarides nappes contact, attesting that (i) nappes stacking was prior to Aquitanian and (ii) the metamorphic basement was exposed at the surface at that time (Michard et al., 2006;Serrano et al., 2006).
In the Internal Rif, the timing of the late compressional events is poorly constrained, due to the lack of precise sedimentary markers and geochronological data. The time range for these events is post Burdigalian (age for the youngest folded sediments in the Fnideq and Sidi Abdeslam Basins Hlila et al., 2008) and prior to Early Zanclean (age of the oldest unfolded sediments in the Tirinesse Basin, Saji and Chalouan, 1995;Cornée et al., 2014).
However thermal modeling using (U-Th)/He apatite data, 40 Ar/ 39 Ar and K/Ar data on biotite, zircon, and apatite fission track highlights that the rocks suffered a rapid to moderate cooling between 22.5 and 18 Ma, a re-heating between 18 and 15 Ma, then a rapid cooling and an exhumation to the surface at around 13 Ma (Azdimousa et al., 2014;Romagny et al., 2014). These authors interpreted the re-heating during the late Burdigalian-Early Langhian as a renewal in thrusting and burying of the internal units during a compressional event (back-thrusting).

The Rif belt
The exhumation of the Beni Bousera peridotites mechanism and timing are strongly debated and different scenarios are proposed: (1) a pre-Alpine exhumation by crustal thinning followed by thrusting during the Alpine orogeny (Reuber et al., 1982;Chalouan et al., 2001;Chalouan and Michard, 2004) and (2) an Early Miocene exhumation due to an extensional regime in a back-arc context accommodated by low angle lithospheric shear zone located on top of the peridotites (Afiri et al., 2011;Álvarez-Valero et al., 2014;Frets et al., 2014;Bessière, 2019). In the Ceuta Peninsula, Homonnay et al. (2018) demonstrated that the tectonic coupling between mantle peridotites and crustal metamorphic rocks occurred at a depth of 20-30 km under high-temperature conditions during crustal thickening at ∼ 30 Ma. The peridotites were subsequently exhumed during the extensional event at ∼ 21 Ma. Our data show that the peridotites were located somehow upon a partially melted continental crust during dyke's intrusion, and were exhumed by the Aaraben shear zone from at least the ductile-brittle transition at 21 Ma, which is in line with the second exhumation model cited above.
Following this extension event, major compressional events were recognized in the Beni Bousera area: a shortening phase associated to the NW-SE Beni Bousera antiform (Reuber et al., 1982) and a nearly N-S shortening phase, evidenced by E-W folds (this work) and by NE/SW sinistral and NW/SE dextral conjugate strike-slip faults (Romagny, 2014).These compressional events were also described in the northern part of the Internal Rif (Chalouan, 1986;Vitale et al., 2014Vitale et al., , 2015Homonnay et al., 2018;Homonnay, 2019), where they folded the Fnideq Basin (Feinberg et al., 1990) and are therefore post-Burdigalian. At the scale of the Internal Rif, the first compressional event produced regional-scale open folds like the Beni Mezala and Beni Bousera antiforms parallel to the belt. The timing of this compressional event is difficult to determine precisely due to the lack of biostratigraphic markers associated with sedimentary basins during this period. However, it could be related to the re-heating recorded by low-temperature thermochronometry between 18 and 15 Ma ascribed to back-thrusting and burying of the Internal Zones .
In Western Betics, the exhumation of the Alpujarride complex occurred during the Early Miocene (22 to 18 Ma) in a ≈ N-S to NNE-SSW (present coordinates) extensional setting (i.e., Monié et al., 1994;Crespo-Blanc, 1995;Platt et al., 2006;Esteban et al., 2013). The late exhumation of the Alpujarrides is also coeval with the development of Aquitano-Burdigalian extensional basins lying unconformably on the Malaguide and the Alpujarride complex (Serrano et al., 2006, Serrano andGuerra-Merchán, 2007) and with the extensional development of the Alboran Basin (e.g., Do Couto et al., 2016). Paleomagnetic data and kinematic reconstructions suggest that the Western Betics were submitted to a ≈ 53°clockwise rotation since the Late Miocene (Lonergan et al., 1993;Crespo-Blanc et al., 2016). In this way, the Early Miocene extension was oriented ≈ WNW to NW-SE and, like in the Beni Bousera area, was orthogonal to the subduction trench.
In the Central and Eastern Betics, the Nevado-Filabride complex exhumation occurred during the Early to Late Miocene (i.e., de Jong, 1991;Monié et al., 1991;Platt et al., 2005;Augier et al., 2005a;Vazquez et al., 2011) and was controlled by a ≈ E-W (present coordinates) regional-scale extension (Jabaloy et al., 1992;Martínez-Martínez, 2006) evolving progressively to a NW-SE, then N-S (present coordinates) at the end of the Tortonian (Augier et al., 2013). This extension also controlled the formation of extensional intra-mountain basins in the Eastern Betics, at the Nevado-Filabrides area (Augier et al., 2013). Paleomagnetic data and kinematic reconstructions suggest that the Central Betics were submitted to a nearly 12°clockwise rotation since the upper Miocene (Crespo-Blanc et al., 2016). This extension was then first oriented nearly ENE-WSW (parallel to the subduction trench) and progressively shifted to nearly NNW/SSE. This E-W to ENE-WSW extension was widespread in the Gibraltar Arc during the Early-Middle Miocene and is considered as related to the westward migration of the Alboran Domain (Balanya et al., 2012;Jolivet et al., 2008). This strengthens the hypothesis proposing an important strain partitioning within the whole Gibraltar Arc during the Miocene   (Balanya et al., 2012) with concomitant arc-perpendicular extension in the Western part of the Alboran Domain, arcparallel extension in its Eastern part and arc-perpendicular compression within the orogenic wedge.
This simultaneous arc-perpendicular and arc-parallel extension in the Western and Eastern Betics respectively could reflect the particular crustal/mantle evolution of the eastern part of the Gibraltar Arc where the subduction is inactive since the Early Miocene and the slab tearing processes have probably occurred (Villaseñor et al., 2015;Mancilla et al., 2018).

Consequences for the Gibraltar Arc and the Alboran Basin formation
The Early Miocene extension event is broadly described in the Rif and the Betics (García-Dueñas et al., 1992;Galindo-Zaldívar et al., 1993;Saji and Chalouan, 1995;Romagny, 2014). This event is contemporaneous with the opening of the Alboran back-arc Basin developed since the Oligocene in response to slab retreat towards the west (e.g., Comas and Soto, 1999;Martínez-García et al., 2011). In this context, crustal thinning (Fig. 10a), decompression and heat transfer from the hot asthenosphere mantle triggered partial melting of the crustal formations located under the peridotites and subsequent magmatic intrusion within the peridotites and the overlying crustal units (Rossetti et al., 2013). Crustal thinning associated with ductile shear zones and normal faults was also responsible for the late exhumation of the Sebtides-Alpujarrides.
From the Late Early to the Middle Miocene the Gibraltar Arc was subjected to a period of shortening which resulted in the formation of an orogenic wedge with thrusting of the Internal zones upon the External zones toward the Rharb and the Guadalquivir Basins and back-thrusting in the internal zones toward the Alboran Basin   (Fig.  10b). In this scenario, the arc-perpendicular shortening is the consequence of the collision between the Alboran Domain, assigned to the upper plate of the subduction complex, the Flysch nappes, and the External domain, corresponding to the accretionary prism and to lower plate paleomargin respectively (Fig. 10b).
During the Late Miocene, a N-S to NNW-SSE compressional event induced a tightening of the Gilbraltar Arc controlled by the Europe-Africa convergence that drastically altered its geometry (Crespo-Blanc et al., 2016) (Fig. 10c).

Conclusion
The main outcomes of this study are as follows. In the Beni Bousera area, the granitic dykes were emplaced within nearly E-W normal faults at 21 Ma, and the NW-SE oriented Aaraben normal shear zones are also dated between 22 to 20 Ma. Granitic dykes and normal shear zones were generated in response to NNE-SSW extension (actual coordinate). Based on available paleomagnetic data, the tectonic setting which affected the Alboran Domain during the Early Miocene was a nearly E-W extension related to the opening of the back-arc Alboran Basin. At the time of the extension, the subcontinental peridotites were located upon a partially melted continental crust at least from the brittle-ductile transition. After this extension, two compressional events are evidenced successively by NW-SE and E-W folds in the framework of Africa-Eurasia convergence.