Tectonic Evolution of the Western High Atlas of Morocco: Oblique Convergence, Reactivation, and Transpression

The High Atlas of Morocco is a double‐vergent mountain belt developed by Cenozoic shortening and inversion of a Triassic‐Jurassic rift. The structural setting, the morphometric features, and the patterns of exhumation through time and space change remarkably both along and across the strike. Here we combine structural data with revised thermochronological data to unravel the kinematic and evolution of the western High Atlas. Our results show that the structural grain of the western High Atlas is defined by two main groups of faults, namely, thrust and oblique‐slip faults, which mainly strike subparallel from W‐E to NE‐SW. The slip direction of the thrust structures is NNW‐SSE to NW‐SE oriented, and the slip direction of the oblique‐slip faults is WSW‐ENE to NW‐SE oriented. Pieces of thermochronological and geological evidence indicate that in the last ~10 Ma the exhumation rate increased during the activity thrusts and oblique‐slip faults. The coexistence of these two fault systems also suggests partitioning of deformation under a transpressive regime. In the western High Atlas, we estimate a displacement of ~12 km on the frontal thrusts and of at least ~22 km on the axial oblique‐slip structures. Thrusts and oblique‐slip structures together result in a total cumulative displacement of ~25 km, which represents about half of the Africa‐Eurasia convergence.


Introduction
Orogens develop preferentially along zones of preexisting weaknesses. It is widely recognized that compressional deformation preferentially localizes on the part of the lithosphere previously weakened by rifting, reactivating, or crosscutting normal faults with a reverse motion. Examples of positive inversion of intracontinental preexisting rifts are well recognized on several orogens, such as the Pyrenees, the Eastern Cordillera of the Andes, and the Kopet-Dagh and Alborz belt in Iran, among others (Babault et al., 2013;Colletta et al., 1990;Dewey & Burke, 1973;Jackson, 1980;Mora et al., 2006;Puigdefàbregas & Souquet, 1986;Vergés & García-Senz, 2001). In these orogens, compressional deformation reworks preexisting structures and is influenced by the mechanical stratigraphy of the basin infill as, for example, in the case of synrift deposits interlayered with evaporites. The orientation of the tectonic transport direction with respect to the preexisting fabric is also relevant. If the maximum compressive stress is oblique to the preexisting fabric, the resulting deformation may be dominated by a transpressive regime and strain may be partitioned between dip and strike-slip structures (Dewey et al., 1998;Jones & Tanner, 1995;Sanderson & Marchini, 1984;Tikoff & Teyssier, 1994) or nonpartitioned, accommodated mainly by oblique-slip reverse faults (Dewey et al., 1998). Strain partitioning is well documented along oblique subduction zones, where the frontal thrusts are flanked by a strike-slip belt localized along the volcanic arcs. Analog and numerical modeling have also shown that strain partitioning along subduction zones, such as Sumatra or Hikurangi, is favored by the strength drop along the subduction faults (e.g., Chemenda et al., 2000;Fitch, 1972;McCaffrey, 1992;McCaffrey et al., 2000;Molnar, 1992). In the case of rift inversion, where shortening superimposes preexisting structures, the degree of obliquity and the strong lateral variation in rock strength may lead to a complex configuration with the coexistence of partitioned and nonpartitioned deformation.
To better understand the structural evolution of the HA and its along-strike variations, we carried out a systematic structural analysis to constrain the geometry and kinematics of the main fault systems. We unravel the evolution of the HA over time, the geometry and kinematics of the main structures, and the way deformation varies along the strike of the belt. Moreover, we used exhumation rates over time to constrain the timing of deformation. Our analysis reveals that the structural grain of the belt is largely controlled by the reactivation of preexisting features and by the presence of Triassic evaporites. In particular, we found that the specific position of the Triassic evaporites controls strain partitioning in the HA, exerting a primary control on the geometry and kinematics of the main fault systems. Field and thermochronological data outline the middle-late Miocene, in particular the last~10 Ma, as the main event that shaped the HA. This new model provides a coherent picture of the style of deformation and its partitioning and variations across and along the strike of the HA.

Geological Background
The HA of Morocco runs WSW-ENE for approximately 600 km and rises up to 4 km a.s.l. The HA includes three geographic domains: the eastern HA located east of 4°W, the CHA from 4°W to 7°W, and the WHA from 7°W to 8°W (Figure 1). Here we also define the far WHA (FWHA) as the region from 8°W to 9°W ( Figure 1).
The HA is parallel to the Paleozoic Anti-Atlas (AA) belt, which is located further south and rises up to 2.5 km a. s.l. These two reliefs are linked by the Miocene-Pliocene Siroua volcanic field (Berrahma & Delaloye, 1989). Another Miocene-Pliocene volcanic field is located in the eastern AA in the Saghro massif (Berrahma et al., 1993). Together, these two volcanic fields split the southern foreland of the HA into two basins: the Souss basin to the west, which opens toward the Atlantic Ocean, and the intramontane Ouarzazate basin to the east ( Figure 1). The northern foreland of the HA also consists of two basins: the Haouz basin, which flanks the FWHA and the WHA and is bordered by the Jebilet thrust front (JTF in Figure 1), and the Tadla basin, which flanks the CHA to the south and the Middle Atlas (MA) to the east (Figure 1). The FWHA and the WHA rise to higher elevations and are narrower with respect to the CHA (Figure 1). They also consist of different lithologies ( Figure 2): The FWHA and WHA are largely composed of Paleozoic and Precambrian rocks, whereas the CHA exposes mostly Mesozoic rocks. In the FWHA and WHA, the Precambrian and Paleozoic units are represented by schists, gneisses, and granites that are locally directly overlain by Cretaceous sediments (Taroudant log of Figure 2), whose thickness and distribution are extremely variable. In the FWHA, Triassic sandstones and evaporites with variable thicknesses of up to 1.3 km are confined within narrow grabens (Ourika, Telouet and Taroudant logs of Figure 2; Baudon et al., 2009;Domènech, 2015;Domènech et al., 2016;El Arabi et al., 2003). Detrital zircons U-Pb ages indicate the southern AA region as the dominant source for the HA Triassic sediments (Domènech et al., 2018). In the HA, the Jurassic and Cretaceous strata are also confined within basins controlled by the extensional structures of the Mesozoic rift. In the CHA, up to 10 km thick Jurassic carbonates are preserved together with minor Cretaceous and Paleogene deposits (Arboleya et al., 2004;Teixell et al., 2003). The HA was partly involved in the Variscan compressional phase, during which deformation was mostly focused along the Tizi n'Test right-lateral strike-slip fault (Mattauer et al., 1972;Petit, 1976;Proust et al., 1977;Qarbous et al., 2003). In the AA, the Variscan orogeny formed a fold-and-thrust belt, reactivating late Proterozoic-to-Precambrian structures (Helg et al., 2004). Since the Permian in the HA and MA, extension related to the breakup of Africa caused the opening of a rift basin developed primarily during the Triassic and Jurassic (Arboleya et al., 2004;Beauchamp et al., 1996;Choubert & Faure-Muret, 1962;Gomez et al., 2000;Jacobshagen et al., 1988;Laville & Pique, 1991;Manspeizer et al., 1978;Mattauer et al., 1977;Teixell et al., 2003;Van Houten, 1977). The NW-SE oriented extension generated N-S trending normal faults (Arboleya et al., 2004;Baudon et al., 2009;El Arabi et al., 2003;El Kochri & Chorowicz, 1996;Qarbous et al., 2003;Van Houten, 1977) and reactivated preexisting NE-SW to ENE-WSW Variscan structures. An example of these structures is the Tizi n'Test Fault, which is reactivated as a left-lateral strike-slip fault (Beauchamps, 1988;Laville et al., 2004;Laville & Petit, 1984;Laville & Pique, 1991;Mattauer et al., 1977Mattauer et al., , 1972Ouanaimi & Petit, 1992;Proust et al., 1977). In the Late Cretaceous, the HA and MA rifts is inverted by the overall compression field related to Africa-Eurasia convergence (Frizon de Lamotte et al., 2000;Froitzheim et al., 1988;Gomez et al., 2000;Hafid, 2000;Piqué et al., 2002;Teixell et al., 2003). The timing of the inversion is constrained by both field observation and thermochronology data (Domènech et al., 2016;Froitzheim et al., 1988;Hafid, 2000). It is not clear whether after the onset of inversion, deformation and exhumation continued at a constant pace (Arboleya et al., 2008; Balestrieri et al., 2009;Barbero et al., 2007;Missenard et al., 2008;Tesòn & Teixell, 2008) or whether they occurred in pulses starting from the late Eocene (Frizon de Lamotte et al., 2000, 2009Leprêtre et al., 2015Leprêtre et al., , 2018. However, the Cenozoic inversion in the HA resulted in higher exhumation rates in the axial portion of the FWHA and WHA, where the youngest ages are the late Miocene-Pliocene and where cooling below~200°C is recorded to have occurred since the beginning of the Cenozoic (Balestrieri et al., 2009;Bertotti & Gouiza, 2012;Domènech, 2015;Domènech et al., 2016;Ghorbal, 2009;Leprêtre et al., 2018;Missenard et al., 2008).

Structural Analysis
We carried out a structural analysis mainly in the FWHA and locally in the WHA to define the geometry, kinematics, and, where possible, the timing of deformation of the main structures. A field survey was conducted using available geological maps of Morocco at different scales (Choubert et al., 1970;Hindermeyer et al., 1974;Hollard et al., 1985;Jenny & Couvrer, 1988;Le Marrec, 1985;Tixeront, 1974), revising the main structural features and constructing geological cross sections. We systematically measured the strike and dip of the faults, the pitch or azimuth of the slickenlines, and the sense of shear along the main structures. This approach differs partly from that of Angelier (1984), as we limited our analysis along the main faults and shear zones. We adopted common criteria to define the fault kinematics, such as lunate and S/C structures, Riedel shears, calcite steps, bedding drag along faults, and offset of geological markers (Petit, 1987). Growth strata, synsedimentary deposits, and thermochronology data were used to define the age of faults. In the case of multiple slickenlines, crosscutting relationships were used to determine the relative chronology. We preferentially collected data on faults that crosscut the Mesozoic and Cenozoic units to unambiguously obtain information on Tertiary deformation. Data collected at different locations along the same fault were grouped to obtain a statistically meaningful indication of the sense of motion.
The data were plotted and analyzed using the Daisy software (Salvini, 2004). On the maps (Figures 3 and 4), we indicate the average slip direction of the fault hanging wall with respect to the footwall.

Exhumation Rates
We use the results of the inversion of cooling ages to erosion rates from Lanari et al. (2020) to derive erosion rate maps of the HA at different times. These results include exhumation rates and closure depths and were obtained from a compilation of 100 cooling ages from previous studies (Tables S2-S4 in the supporting information) using Agetoedot (Willett & Brandon, 2013). This method inverts individual ages to erosion rates, solving for the closure temperatures based on the Dodson (1973) equation and taking into account the effects of heat advection with erosion. The basic assumption of this method is that the cooling rate is constant during erosion as this is implicit in the Dodson solution to the closure temperature. The resulting erosion rates are constant from the time of closure to the surface, whereas the geothermal gradient increases with erosion.
With the Agetoedot inversion, a sample dated with two or more methods provides independent erosion rates for each age, and therefore, changes of erosion rate during the interval between ages from the same sample are not resolved. In order to highlight these possible variations, we used the available data to calculate erosion rates also with a different approach on samples dated with fission-track and (U-Th-Sm)/He methods. For the time interval between two ages of double-dated samples (28 samples), we used the closure depth output from Agetoedot and we derived the erosion rates from the ratio of the differences between closure depths and between ages. With this approach, heat advection during the considered time interval is underestimated in the case of accelerating erosion through time, and this, in turn, will result in an estimated erosion rate higher than the true rate. Similarly, in the case of decelerating erosion rates, heat advection will be overestimated, and the resulting erosion rate will be lower than the true rate. Thus, our approach is not likely to improve the uncertainty of the rate estimates, but it improves the resolution on possible changes of erosion through time. A comparison between erosion rates derived from individual ages and from coupled ages is shown in Figure S11 (supporting information).
We focus on the period from the Eocene to the present, when most of the compressive deformation of the HA occurred. Within this period, we chose four time steps (35,25,15,and 5 Ma), and for each of the time step, we plot the erosion rates at those sample locations where the erosion rates integrate over the period from or beyond the chosen time steps to the present. For instance, Figure 12a shows

Structural Data
We collected 483 structural data in 59 sites in the FWHA and WHA (Table 1 and Figure 3). Data are clustered along 13 main faults, where we define the direction of motion of the hanging wall with respect to the footwall (Table 2 and Figure 4). Data have been also analyzed in order to define the strike, the dip of the faults, and the azimuth of the slickenlines over the region ( Figure 5).
Overall, the main strike of the faults is ENE-WSW (N253°and N85°; Figure 5a). Locally, secondary conjugate faults strike NW-SE, and they do not contribute to define the structural grain of the belt. The fault dip angles cluster in two groups at 68°and 40°dipping mainly toward NNW and SSE (Figure 5b), and the fault slickenlines cluster in three groups with pitch angles of 77°, 124°, and 28°( Figure 5d). The slip directions (all projected on the northern half of the stereonet) cluster at NW-SE, E-W, NE-SW, and NNW-SSE ( Figure 5c). These directions are related to the following kinematics: right-lateral oblique slip or normal (e.g., S-20 or S-17, Figure 3) for the NW-SE cluster; right-lateral oblique slip to strike slip (e.g., S-10, Figure 3) for the E-W cluster; sinistral oblique slip (e.g., S-53, Figure 3) for the NE-SW, and thrust (e.g., S-46, Figure 3) for the NNW-SSE cluster.

Geological Cross Section
The HA of Morocco shows an overall sigmoidal "S" shape, as the main tectonic features change eastward from W-E directed toward SW-NE in the WHA and CHA and then toward WSW-ENE in the eastern part of the CHA (Figure 1). This along-strike variation corresponds to a change in the geometry and kinematics of the main fault zones. This change is best illustrated by the four N-S to NW-SE cross sections shown in Figure 6.
The AA′ section runs across the FWHA ( Figure 6) and shows a large-scale double-vergent structure. Along the southern HA front, Cretaceous rocks thrust over Neogene deposits of the Souss basin (Domènech et al., 2016;Fekkak et al., 2018;Missenard et al., 2007;Sébrier et al., 2006). Approximately 5 km to the north, a reverse fault places the Paleozoic rocks over the Cretaceous layers (Ellero et al., 2012). Along this fault, we observed a positive flower structure with vertical Triassic layers ( To the north, the basement is bounded by a reverse fault reactivated as a right-lateral oblique-slip fault (F-XII). Farther north, a south verging thrust fault runs E-W for several kilometers, deforming the Triassic layers (S-52; Figure 6). Along the northern front of the WHA, the Triassic basalts and sandstones thrust over the Jurassic-to-Paleogene layers that are deformed in a syncline (Missenard et al., 2007). This syncline has steep flanks as a box-fold syncline, which suggests the presence of a shallow detachment at the base of the syncline. The main frontal thrust places Jurassic layers on top of Neogene 10.1029/2019TC005563 Tectonics LANARI ET AL.  sediments (S-29, Figure 6). Farther to the north, there is the Jebilet thrust front, which shows two sets of slickenlines ( Figure 10, S-44).
The DD′ section crosscuts the eastern sector of the WHA and the Ouarzazate basin. The main faults trend ENE-WSW at a low angle, and they cut through the Paleozoic, Triassic, and Jurassic sequences, acting mostly as reverse faults (Teixell et al., 2003). Along the northern front, Jurassic layers thrust over the Neogene deposits, crosscutting Quaternary deposits (Figure 11). At the center of the section, we also observed minor strike-slip structures, which border synclines in the Triassic and Jurassic sequence (S-49 and S-48, Figure 6). Along the southern front, Paleozoic units overthrust Paleogene deposits (F-V; Ellero et al., 2012;Teixell et al., 2003;Tesòn & Teixell, 2008). Along the southern rim of the Ouarzazate basin, Quaternary strata lie unconformably over tilted Paleogene and Cretaceous layers (Guimerà et al., 2011).

Exhumation Rates
The distribution of the available cooling ages in the HA is strongly influenced by the rock fertility in apatites and zircons, resulting in a high data density in the axial region of the HA where the Precambrian granites are exposed. Thus, this region is where the data best resolve erosion rates in space, and this is also where the youngest ages are found. Given that the obtained erosion rates integrate from the time of closure to the present, the temporal resolution of our results is higher for younger cooling ages. This is best explained if we consider, for example, a cooling age of 30 Ma: This age resolves an average erosion rate between 30 and 0 Ma, but it does not resolve any change in erosion rate during that time, whereas a 5 Ma age indicates a relatively high erosion rate from 5 to 0 Ma. Thus, the axial region of the HA is where erosion rates are best resolved both spatially and temporally. In this region, a few exceptions apart, maximum rates commonly do not exceed 0.3 km/Ma, and the highest rates occur during the middle-late Miocene and from the end of the

Structural Grain and Kinematics of the HA
The orientation of the HA changes from west to east: The FWHA is W-E directed, the WHA and the CHA run NE-SW, and the eastern CHA runs ENE-WSW (Figure 1). These three segments are parallel to the structural grain within each region and altogether define a dextral sigmoidal architecture of the HA. Exceptions to these general trends are the frontal structures, such as the Jebilet thrust front and the South Atlas fault, which preserve the same trend across different regions of the HA (Figure 1). Two types of faults occur in the FWHA and WHA: low-angle thrust faults that strike W-E to NE-SW and steep oblique-slip faults that strike mainly ENE-WSW to NE-SW (Figure 3). The spatial distribution of these fault systems is not uniform along or across the strike of the HA. The oblique-slip faults (F-IX, F-X, F-XI, and F-XIII) are located only in the inner portion of the WHA (Figure 4). Thrust faults are dominant in the inner portions of the CHA (Arboleya et al., 2004;Teixell et al., 2003) and define the frontal structures along the entire HA (F-VI, F-VII, and F-VIII) (Figure 4). Exceptions to this pattern are W-E high-angle oblique-slip faults across the Siroua volcanic field (Figure 4; F-I and F-II).

Tectonics
The lateral changes of the faults pattern and style of deformation correspond to variations in the width of the mountain belt. The WHA belt, excluding the Jebilet thrust, is narrower than the CHA and is cut by right-lateral oblique-slip faults, whereas the CHA mainly shows thrust-and-fold structures (Figures 1 and  3). The lateral changes of the deformation style, geometry, and orientation of the faults along the HA are probably controlled by the inherited structures active during the Early Mesozoic extension, which, in turn, locally reworked Variscan structures (Mattauer et al., 1972;Petit, 1976;Proust et al., 1977; Qarbous et al.,

10.1029/2019TC005563
Tectonics 2003). This is well verified by the along-strike thickness variations of the Mesozoic sequences (Figure 2). In the FWHA and WHA, relevant Triassic and Jurassic synrift deposits are present only in the external region of the orogenic belt, whereas in the axial region, they are present in minor amounts within confined grabens (e.g., Tizi n'Test graben; Domènech et al., 2015). The thickness of the Jurassic and Triassic deposits varies along the strike from 1.5 km in the FWHA to 2 km in the WHA and up to 10 km in the CHA (Baudon et al., 2009;Domènech et al., 2015;El Arabi et al., 2003;Teixell et al., 2003). Based on the Mesozoic stratigraphy, it is thus possible to reconstruct the geometry of the rift (Baudon et al., 2009;Domènech, 2015;Domènech et al., 2015;El Arabi et al., 2003): Horsts and grabens formed in the western regions of the HA (Domènech et al., 2016), and a large and deep graben formed the CHA (Arboleya et al., 2004;Teixell et al., 2003).

Timing of Deformation
Fault kinematic analysis shows four main slip directions (WSW-ENE, NW-SE, NNW-SSE, and NE-SW; Figure 5c) and three main kinematics. The WSW-ENE and NW-SE slip directions are related mainly to right-lateral/reverse oblique slip, the NNW-SSE slip direction is related mainly to thrust, and the NE-SW slip direction is related to sinistral-reverse oblique slip. A few exceptions are the NW-SE slickenlines azimuths in S-16 and S-17, which relate to normal faulting ( Figure 3) and rarely to thrust faults (F-XII, Figure 4).
In the WHA, these slip directions can be attributed to at least two episodes of fault activity, as indicated by the superposition of slickenlines on the same fault plane. Evidence of the first episode of fault activity is found along the Jebilet thrust front (S-44 and S-43) and along the western portion of the Tizi n'Test fault (S-16 and S-17), where a dextral sense of motion locally overprints a sinistral-normal one. The older sinistral-normal kinematics may be compatible with the transtensional regime related to the Mesozoic rifting phase (Beauchamp, 1988;Laville et al., 2004;Laville & Petit, 1984;Laville & Pique, 1991;Mattauer et al., 1977Mattauer et al., , 1972Ouanaimi & Petit, 1992;Proust et al., 1977) ( Figure S3 in the supporting information, S-26). In the axial region of the WHA, the kinematic of high-angle structures shows right-lateral/reverse oblique-slip motion superimposed over sinistral one (F-IX, F-X, F-XI, and F-XIII; Figure 4). The sinistral NE-SW oriented motion is on high-angle fault planes that have been tilted around the horizontal axes of about 20-40°s o that the Triassic sediments, originally deposited on the hanging wall of this fault system, currently lie on the footwall (e.g., S-16, Figure 3). The dip reversal of the fault implies that the NE-SW present orientation of the slickensides was originally NW-SE (as evident on other untilted faults such as S-16).
The subsequent right-lateral/reverse oblique-slip fault episode is evident in the axial region of the WHA, along the Jebilet thrust and in the Siroua volcanic field. Along the northern front of the FWHA (F-IV), we found evidence of syndepositional growth structures and progressive unconformities within clastic Neogene continental deposits (Figure 7). This has also been described in other locations along the same front by Missenard et al. (2007). The northern front of the WHA (F-VIII) crosscuts Quaternary river terraces, indicating recent deformation (Figure 11). Evidence of recent deformation along the South Atlas fault (F-V) is constrained by Holocene deposits and deformation (Leprêtre et al., 2015;Pastor et al., 2012;Teixell et al., 2003) and by the presence of a river knickpoint (Boulton et al., 2014;Stokes et al., 2017). This finding is also confirmed by tilted Quaternary terraces along the northern rim of the Ouarzazate basin ( Figure S8 in the

10.1029/2019TC005563
Tectonics supporting information). Seismic activity in the Agadir region and along the mountain belt also provides evidence of ongoing tectonic activity along the frontal thrusts (Meghraoui et al., 1999;Sébrier et al., 2006).
The timing of the right-lateral oblique-slip kinematic activity in the axial part of the belt is difficult to constrain because syndeformational deposits are not present there. Low-temperature thermochronology data indicate that exhumation started at least in the Eocene and accelerated during the Miocene (Balestrieri et al., 2009;Bertotti & Gouiza, 2012;Domènech, 2015;Domènech et al., 2016;Froitzheim et al., 1988;Ghorbal, 2009;Hafid, 2000;Lanari et al., 2020;Leprêtre et al., 2018;Missenard et al., 2008). Exhumation rates obtained from the inversion of cooling ages show low initial values, most commonly ≤0.1 km/Ma (Figures 12a and 12b), and no significant changes before the Miocene. In contrast, from the middle-late Miocene to the present, relatively high exhumation rates, up to 0.3 km/Ma are common especially in the axial part of the HA (Figures 12c and 12d). The fault activity in the axial part of the belt controlled the Neogene exhumation pattern, as demonstrated for the South Atlas fault (Lanari et al., 2020). Thus, faster exhumation from the middle-late Miocene to the present suggests intense fault activity during that time. Geomorphologic observations, such as knickpoints along bedrock river channels in the southern HA (Boulton et al., 2014) and changes in the drainage patterns (Delcaillau et al., 2011), also suggest that the fault activity might have influenced the topographic evolution.
The timing of deformation along the structures in the AA is constrained by the presence of Neogene volcanism, as also documented by previous studies (Malusa et al., 2007). In the Siroua massif, faults crosscut both a recent fan (F-I) and late Miocene volcanic deposits, indicating synvolcanism/postvolcanism activity (F-II) (Figures 4 and 8d). Furthermore, along the northern rim of the AA, Quaternary deposits lie unconformably over Paleogene rocks (Figure 6, section DD′), indicating late Miocene uplift of the AA (Guimerà et al., 2011) that produced up to 1.5 km of exhumation (Lanari et al., 2020).

Style of Strain Partitioning in the Western Region of the HA
Structural analysis indicates that the FWHA and WHA build up under a transpressional regime (Ellero et al., 2012). Moreover, a dominant strike slip to transpressional regime has also been inferred along the entire Atlasic front based on focal mechanism and fault kinematics (Soumaya et al., 2018).
In transpression, different degrees and styles of strain partitioning may occur ( Figure S10 in the supporting information; Dewey et al., 1998). In the case of simple partitioning, strain is accommodated along coexisting and parallel thrusts and strike-slip faults, whereas if strain is not partitioned, slip occurs parallel to the tectonic transport direction ( Figure S10 in the supporting material; e.g., Chemenda et al., 2000;Dewey et al., 1998;Geoff & Jones, 1984;Jones & Tanner, 1995). Moreover, partitioning may be continuous or discontinuous depending on the way the transition from strike slip to thrust occurs (e.g., Dewey et al., 1998).
In the FWHA and WHA, the last deformation phase is related to two main fault systems: thrusting with NNW-SSE slip direction and right-lateral/reverse oblique slip with WSW-ENE to NW-SE slip direction. Those fault systems are active during the Neogene-Quaternary, suggesting a certain degree of strain partitioning. In particular, our study area resembles the case of continuous partitioning of Dewey et al. (1998) with right-lateral/reverse oblique-slip faults localized in the axial region and thrust faults along the fronts. For this case, Dewey et al. (1998) predict that in transpression the particle path changes with respect to the main structures from orthogonal to parallel with the degree and style of partitioning. In our case, the slip direction can be considered as a proxy for the particle path, and this changes across the mountain belt from orthogonal to parallel with respect to the main structures as shown in Figure 13a.
Strain partitioning in the FWHA and WHA may be facilitated by (a) reactivation of preexisting faults with a certain obliquity with respect to the current tectonic transport direction and (b) across-/along-strike changes of the Mesozoic sequences combined with discontinuous presence of salt layers (Figure 2).
1. The present-day tectonic transport direction, as deduced based on the geodetic strain of North Africa, is WNW-ESE, (Figure 1; Serpelloni et al., 2007), and the direction of motion of Africa with respect to Eurasia during the last 15 Ma (Dewey et al., 1989), is approximately NW-SE. In the WHA, this tectonic transport direction strikes at an angle of 20°to 40°with respect to the preexisting fabric. 2. The remarkable difference in thickness and facies distribution, especially for the evaporites layers in the frontal part of the WHA (Figure 2; Baudon et al., 2009;Domènech et al., 2015;Missenard et al., 2007), exerts a primary structural control in the deformation (Missenard et al., 2007)

10.1029/2019TC005563
Tectonics salt tectonics features such as box-fold synclines ( Figure 6; CC′ section; Missenard et al., 2007). These observations suggest that the presence of décollement layers, such as salt levels, may locally perturb the strain field (e.g., Lohr et al., 2007) and relate to the occurrence of strain partitioning in the FWHA and WHA.
In summary, we suggest that during the late Cenozoic our study area deformed under transpression with continuous partitioning and localization of oblique-slip faulting in the axial region of the mountain belt, which is facilitated by the inherited structures and mechanical stratigraphy.

Amounts of Shortening in the WHA
Previous shortening estimates range from approximately 14 km in the FWHA (17-20%; Domènech et al., 2016;Fekkak et al., 2018) to 13-15 km in the WHA (15-21%; Domènech et al., 2016;Missenard et al., 2007 ;Teixell et al., 2003). We estimate the Cenozoic shortening in the WHA along a NNW-SSE section (Figures 13a and 13b; details in the supporting information). As we discussed in section 5.2, before the deformation, the axial structures were dipping toward the basement, in according also with prevous restoration (Domènech et al., 2016). In fact, during the Cenozoic deformation, those axial oblique-slip faults should have been tilted along the horizontal axis up to changing the dip direction. Thus, the Triassic synrift deposits that were deposited in the hanging wall of the faults are currently located in the footwall. Our estimation of Cenozoic shortening is~12 km, in good agreement with previous ones (Domènech et al., 2016;Missenard et al., 2007;Teixell et al., 2003). These values represent a minimum bound of the total displacement as they do not account for the oblique-/strike-slip component. We then estimate the displacement along the major oblique-slip structures integrating structural/kinematics data with the vertical offsets derived from thermochronological data and stratigraphy (details in the supporting information). For example, along the South Atlas fault (F-X, S-10), thermochronological data constraint a vertical offset of at least 2.8 km (Lanari et al., 2020), which combined with the geometry and kinematics of this fault, indicates a total throw of 17.1 km ( Figure S1 in the supporting information). With the same approach and on the basis of stratigraphic data, along the Tizi n'Test fault (F-XIII), we estimate~5.5 km of strike-slip throw (see details in the supporting information). A summary of the displacement pattern along the main fault systems is shown in Figure 13. Altogether, accounting for the strike-slip component along the WSW-ENE direction, we obtain a lateral displacement of~22.5 km. Combining the dip and strike displacements gives a total cumulative displacement of~25.5 km, oriented WNW-ESE (Figure 13c). This direction is similar to the one derived from horizontal GPS velocities (Serpelloni et al., 2007) and to the direction of motion of Africa relative to Europe during the last 15 Ma (Figure 13d; Dewey et al., 1989;DeMets et al., 1994;).
Based on thermochronological data (Balestrieri et al., 2009;Barbero et al., 2007;Domènech, 2015;Domènech et al., 2016;Ghorbal, 2009;Lanari et al., 2020;Leprêtre et al., 2018;Missenard et al., 2008), which indicate a major pulse of exhumation from the middle/late Miocene to the present, and based on our structural analysis, which suggests this pulse of exhumation is due to deformation and faulting (Lanari et al., 2020), we infer that most of the total displacement occurred during the last~10 Ma. This results in a shortening rate in the WHA of approximately 2.6 mm/year, which is about half of the Africa-Eurasia relative plate motion rate (DeMets et al., 1994;Dewey et al., 1989;Serpelloni et al., 2007).
The style of deformation in the WHA is strongly different from that in the CHA where deformation occurred mostly by folding and thrusting ( Figure 14; Arboleya et al., 2004;Teixell et al., 2003) and where the upper crust tectonic evolution might be significantly influenced by a dominant present of salt diapirs (Calvín et al., 2018;Teixell et al., 2017). In addition, the amount of shortening of the CHA is more than double that of the WHA (approximately 30-34 km; Beauchamp et al., 1999;Gomez et al., 2000;Teixell et al., 2003). Further studies are required to understand the along-strike relevant difference between these two regions of the HA.

Tectonic Evolution of the HA: From the Rifting to the Modern Configuration
We propose a reconstruction of the tectonic evolution of the HA and AA (Figure 15), which merges structural observations with exhumation rates observed at different time intervals.
The Triassic-Jurassic rift phase in the HA was characterized by a left-lateral transtensive regime oriented NW-SE (Figure 15a Mattauer et al., 1977Mattauer et al., , 1972Ouanaimi & Petit, 1992;Proust et al., 1977). The rift phase was followed during the Cretaceous by extensive deposition of limestone (Arboleya et al., 2004;Domènech et al., 2016;Leprêtre et al., 2018;Teixell et al., 2003). In the AA, the extent of the Cretaceous deposit is disputed. Some authors propose that the Cretaceous sediments covered the majority of the AA (Guimerà et al., 2011;Soulaimani & Burkhard, 2008), whereas others suggest that they were restricted to seaways (Frizon de Lamotte et al., 2009). Our cartoon in Figure 15b is inspired by this latter view and indicates that some relief was already present in the AA.
Low-temperature thermochronology and field observations constrained the onset of the compressional deformation in the HA to the Late Cretaceous (Domènech et al., 2016;Froitzheim et al., 1988;Hafid, 2000) with limited exhumation during the Paleogene (Frizon de Lamotte et al., 2009;Leprêtre et al., 2018) ( Figure 15c). Our exhumation rates also show a slow exhumation during the Eocene-Oligocene. Progressive unconformities and syntectonic deposits, together with the acceleration of exhumation from shallow depths (Figures 12c and 12d), indicate that the main compressional and exhumation episode occurred from the middle/late Miocene (Figure 15d).
Here we first document a quantitative linkage between high exhumation rates and timing of deformation in the HA and AA, both coeval with the late Miocene volcanism of the Siroua and Saghro (Berrahma & Delaloye, 1989;Mhamed Berrahma et al., 1993), which so far was only inferred (Frizon de Lamotte et al., 2009;Guimerà et al., 2011;Malusa et al., 2007). Fast exhumation in the last~10 Ma (Figure 12d) is strongly controlled by faulting, which in turn is tied with the AA volcanism, as we documented in F-I (Figure 8d). Moreover, at this time, the North Africa relative plate motion, with respect to Eurasia, changes from N-S to WNW-ESE (Dewey et al., 1989). This later direction is the one that favors the strain partitioning we observed.

10.1029/2019TC005563
Tectonics Since the middle/late Miocene, deformation has developed heterogeneously in the FWHA-WHA and the CHA. In the FWHA-WHA, the right-lateral oblique-slip fault reactivation of inherited high-angle structures and the sparse distribution of the Mesozoic covers favored the localization of shortening largely along the boundaries of an inherited structural horst (the Toubakl massif). The localization also resulted in the partial inversion of the rift structures, as suggested by the small shortening and uplift along the Jebilet thrust, which could represent the partly reworked northern rift-shoulder fault in the WHA (Hafid et al., 2006;Sebti et al., 2009). Thus, in the WHA, localization of shortening resulted in the uplift of a narrow mountain belt that scarcely expanded in the N-S direction only (BB′ section; Figure 15d) and that is characterized by high relief incised by steep rivers (Boulton et al., 2014;Stokes et al., 2017). In the CHA, the thick and extensive Mesozoic sequence may have favored the wider fold-and-thrust belt (AA′ section; Figure 15d) with moderate relief and a lower exhumation with respect to WHA, producing a relevant along-strike change in the morphotectonic and kinematic.

Conclusions
Based on the structural analysis of the FWHA, WHA, and AA, combined with the construction of geological cross sections and exhumation rates, we conclude the following: • The stratigraphy and thickness of the Mesozoic synrift sediments and the presence of preexisting faults exerted a primary control on the style of the structural inversion. The structural grain of the WHA is defined by two main families of structures with two main slip directions and kinematics. These fault families consist of NNW-SSE slip-direction-oriented thrusts, which are common in the frontal regions of the mountain belt where the Mesozoic cover is present, and of WSW-ENE slip-direction-oriented, right-lateral oblique-slip faults, which characterize the axial regions where the Mesozoic cover is absent or sparsely distributed. • Field, thermochronological, and geophysical data indicate that these two fault systems were active at the same time, suggesting for a partitioning of the strain, induced by the oblique convergence and locally by the occurrence of Triassic salt layers. • The localized deformation produced rapid uplift and exhumation in the axial region of the WHA and resulted in high topographic growth limited over a narrow belt. Across the WHA, the deformation extended northward in the Jebilet thrust front, reactivating the preexisting northern rift shoulder. Conversely, in the CHA, the presence of salt layers and a thick sedimentary pile favored the development of a thrust-and-fold belt that resulted in a wider thrust-and-fold belt. • Based on thermochronological and structural data, we estimate that the amount of displacement accommodated by the NNW-SSE thrusts is approximately~12 km, while the ones accommodated by the WSW-ENE right-lateral oblique-slip structure is at least~22 km. Combining these values, we estimate the amount of total shortening, which is approximately~25 km oriented approximately WNW-ESE. • Exhumation rates indicate acceleration in the exhumation from the Oligocene with maximum rates during the last~10 Ma, coeval with main faults and with late Miocene AA volcanism. This suggests that exhumation of the HA, volcanism, and tectonics are strictly link.