From extension to compression: high geothermal gradient during the earliest Variscan phase of the Moroccan Meseta; a first structural and RSCM thermometric study

One of the present-day hot topics in the study of collisional orogens is the evaluation of the thermal and structural heritage that an orogenic belt receives at its birth from the previous rift stage (Jammes & Huismans, 2012; Vacherat et al., 2014, 2016; Jourdon et al., 2019; Saspiturry et al., 2019). In the well-studied example of the Pyrenean Belt, the high geothermal gradient linked to the hyper-extension of the European margin accounts for the ductile deformation and high-temperature metamorphism observed in the distal units of the inverted margin (Vacherat et al., 2016; Odlum & Stockli, 2019). The Pyrenees belt rose in two main orogenic phases, dated firstly at ~85–66 Ma with high-temperature (HT) metamorphism and incipient orogenic reliefs, and then between ~50 and 20 Ma with nappe emplacement and erosion (Vacherat et al., 2014; Ternois et al., 2019). In the present paper, we report on an example from the Central Paleozoic Massif of Morocco, which offers the first case of high geothermal gradient, presumably inherited from the Devonian rifting evolution of the Gondwana margin, during the earliest shortening phase of the Variscan Orogeny – that is to say, a Variscan tale which evokes the Pyrenean one.

In several parts of the Variscan Belt of Europe (Fig. 1A), syntectonic recrystallizations occurred in the higher grades of high-pressure, low-temperature (HP–LT) or high-temperature, low-pressure (HT–LP) metamorphism (Ballèvre et al., 2009; Martínez Catalán, 2012; Faure et al., 2014; Schulmann et al., 2014; Pitra et al., 2017). Likewise, the Mauritanides belt could have been affected by high-grade Variscan metamorphism (Le Goff et al., 2001; Michard et al., 2010). Contrastingly, the Variscan segment exposed in the Moroccan Meseta (Fig. 1B) only exhibits low-grade greenschist-facies metamorphic rocks, except in the Western Meseta Shear Zone where kyanite and staurolite-garnet assemblages developed (Michard et al., 2008, 2010; Chopin et al., 2014; Delchini et al., 2016). In the present paper, we consider a central part of the Meseta Orogen, known as the “Montagnes en Quartzites du Pays Zaian” (Termier, 1936), shortly the Zaian Mountains. They correspond to an anticlinorium of folded Cambrian-Ordovician quartzite-rich series unconformably overlain by Upper Visean, subaerial or shallow-water, then turbiditic sequences, which are in turn overloaded by Ordovician-Devonian nappes. Only low-grade metamorphic conditions have been recognized there. According to illite crystallinity analysis (Bouabdelli, 1989, 1994), the metamorphic grade of the Cambrian and Ordovician rocks would range from the epizone to the anchizone, and that of the Carboniferous rocks from the anchizone to diagenesis. Allary et al. (1976), in line with Huvelin (1973), concluded that an early folding phase associated with an axial-planar cleavage shaped the Zaian Mts prior to the Visean, likely by the end of the Devonian, before being superimposed by a lower-grade Late Carboniferous deformation. However, Bouabdelli (1989, 1994) and Bouabdelli & Piqué (1996) argued that K–Ar analyses (Huon et al., 1987; Bouabdelli, 1989) yield similar results between 330 and 280 Ma, regardless of the age of the rock, and concluded that the only syn-metamorphic phase was Late Carboniferous.

In a conference paper, Ouanaimi et al. (2018) argued in favour of a single, Variscan synmetamorphic event in the Zaian Mts, based on fragmentary observations (three sites). In the present work, we report on structural/microstructural and thermometric data (Raman Spectroscopy of Carbonaceous Material, RSCM) from 37 sites within and around the Zaian Mts. Both types of data support the role of a pre-Visean, Eo-Variscan folding phase associated with sub-greenschist facies recrystallizations. The T_RSCM quantification presented here for the first time in the Moroccan Central Massif shows that inversion of the Gondwana distal passive margin occurred in a high geothermal gradient presumably inherited from the Devonian rifting evolution.

The Meseta Orogen of Morocco extends north of the Anti-Atlas, which corresponds to its mildly folded foreland belt at the northern margin of the West African Craton (WAC; Fig. 1). The Meseta Orogen is currently regarded as formed during the Laurussia-Gondwana convergence at the expense of the distal margin of the WAC above the subducting Rheic oceanic lithosphere (Fig. 2). Two distinct parts are recognized in the orogen, the Western Meseta and Eastern Meseta, separated by the Mesozoic-Cenozoic Middle Atlas (Hoepffner & Houari, 2006; Michard et al., 2008, 2010; Hoepffner et al., 2017). In the Western Meseta, the Precambrian basement (Eburnean and Pan-African terrains) crops out in places (Pereira et al., 2014, 2015), included in the Mazagan escarpment (Kuiper et al., 2019). Metamorphism (Wernert et al., 2016) and granite emplacement (El Hadi et al., 2006) occurred mainly between ~330 and 275 Ma, but bimodal magmatism began at ~360 Ma in the Jebilet massif (Delchini et al., 2018). The exotic Sehoul Block displays a 367 Ma old granite (Tahiri et al., 2010). In the Eastern Meseta, folding and metamorphism have been ascribed to Late Devonian events based on Rb–Sr and K–Ar analyses (Clauer et al., 1980; Huon et al., 1987). These “Eo-Variscan” folding events are superimposed by Variscan events including granite emplacement at ~330–290 Ma and Late Pennsylvanian-Permian folding.

The Zaian Mts and their annexes further south (J. Hadid, Bou Acila; Fig. 3A) are located in the transition zone between the Western and Eastern Meseta. Their basement comprises an undated, but likely Ediacaran “Schist-Tuff unit” associated with three types of granite dated from 625 ± 9 Ma to 552 ± 10 Ma (Pereira et al., 2014; Ouabid et al., 2017). The dolerites or gabbros that crop out in the area are not dated; they would be pre- Lower Cambrian (Morin, 1962b) or much younger (Devonian-Carboniferous; Ouabid et al., 2017). Near the base of the transgressive Paleozoic series (Fig. 3B), the Lower Cambrian is dated by archaeocyathid-bearing marbles (Morin, 1962a). This basal sequence of arkosic and carbonate beds is followed upward by greenish greywackes and pillow basalts (Bou Acila) ascribed to the Middle Cambrian, then by the ~2000 m-thick quartzite-rich series typical for the Zaian Mts. This clastic series likely begins in the Middle-Upper Cambrian, but it is mostly dated by acritarchs from the Lower-Middle Ordovician (Cailleux, 1994). The quartzite series is followed upward by the meta-psammites and metapelites of the “Schistes d’Asfar” formation where acritarchs (Drot & Morin, 1962) and chitinozoans (El Houicha et al., 2018) indicate the Upper Ordovician.

The stratigraphic record is interrupted in the Zaian Mts by a deep erosional gap spanning at least from the Upper Ordovician to the Middle-Upper Visean (Fig. 3B). However, the Silurian and Devonian series are well exposed to the west and north in the Central Massif (Fig. 1B; Charrière & Regnault, 1989; Fadli, 1994; Zahraoui, 1994a and b; Razin et al., 2001; Becker et al., 2015; Richter et al., 2016) and to the east in the allochthonous units of the Nappe Zone (Hollard & Morin, 1973; Bouabdelli, 1989, 1994; Becker et al., 2015). In the Zaian Mountains, undated subaerial conglomerates or shallow-water limestones dated from the Upper Visean (Cózar et al., 2019) overlie unconformably the Ordovician to Lower Cambrian rock units. Black shales and sandy or calci-turbidites accumulated over ~2000 m are dated from the Upper Visean-Serpukhovian (Bensaïd et al., 1979; Verset, 1988; Huvelin & Mamet, 1997). They define the Azrou-Khenifra and Fourhal basins, which merge in the north of the Zaian Mts. Gabbroic magmas emplaced as sills and lava flows all around the Zaian Mts during the Serpukhovian (Ben Abbou et al., 2001; Roddaz et al., 2002; Bamoumen et al., 2008). The basinal sedimentation ended with the emplacement of the Eastern Nappes (Nappe Zone; Fig. 1B), which are rooted west of the Tazekka – Eastern Meseta domain (Hoepffner, 1994). The lowest nappes (Ziar and Mrirt nappes) include Silurian and Devonian folded sequences, whereas the highest (Azrou units, Khenifra nappe) mainly consist of Ordovician low-grade metapelites (Bouabdelli, 1989, 1994). The Lower Carboniferous fold-and-thrust duplexes and the northern tip of the Zaian Mts are intruded by the Ment granite at 280–270 Ma (Rb–Sr whole rock; Mrini et al., 1992). Lower Permian red beds, rhyolite flows and ignimbrites (Youbi et al., 1995) overlie unconformably the folded units.

In order to check the occurrence of a former fold belt beneath the Visean unconformity, we used the classical methods of structural analysis in the rock units on both sides of the Visean unconformity. Google Earth satellite imagery proved appropriate in the moderate relief and semi-arid context of the area to prepare/exploit the field observations. Thin-section petrography allowed us to define the tectonic-metamorphic evolution beneath and above the unconformity.

The carbonaceous material (CM) present in metasedimentary rocks derives from the organic matter originally present in the corresponding protoliths. In the natural laboratory, the structure of the CM mainly evolves with increasing temperature. This thermal evolution (“graphitization” process) irreversibly changes the short and misorientated aromatic layers into better organized CM, then into perfectly stacked layers in the triperiodic structure of graphite, with microtextural, structural, and chemical modifications of the starting material (Beyssac et al., 2002). Raman spectroscopy was used in several studies to characterize qualitatively these changes in CM structure (e.g., Wopenka & Pasteris, 1993; Yui et al., 1996). Recently, Raman spectroscopy appeared to be a robust tool to quantify carbonization and graphitization processes (e.g., Beyssac et al., 2002; Lahfid et al., 2010). Consequently, a new geothermometer has been developed, namely the Raman Spectroscopy of Carbonaceous Material (RSCM) thermometry, which returns the maximum temperature reached by the rock. The first calibration of this geothermometer was established by Beyssac et al. (2002) for the temperature range of 330–650 °C. Lahfid et al. (2010) later expanded its applicability conditions to the range of 200–350 °C.

In the present study, all maximum temperatures determined by RSCM geothermometry, noted T_RSCM, were estimated using the RA1 parameter proposed by Lahfid et al. (2010) except for samples 15 and G034 for which the T_RSCM were determined using the R2 parameter proposed by Beyssac et al. (2002) for temperatures exceeding 350 °C. T_RSCM < 180 °C were determined qualitatively by comparing the obtained spectra to those of the Glarus reference area (Lahfid et al., 2010).

All carbonaceous material (CM) measurements by Raman spectroscopy were acquired at BRGM using a Renishaw inVia Reflex microspectrometer. Concerning acquisition parameters such as exposure time to laser, spectra accumulation number, acquisition mode and acquisition window, we followed the protocol analysis described in Delchini et al. (2016). The laser used is a Diode Pumped Solid State (DPSS) source excitation of 514.5 nm. This laser reaches the sample surface, through the ×100 objective (numerical aperture = 0.90) of a Leica DM2500 microscope, with a power lower than 1 mW. After interaction between CM and green laser, the Raman signal was dispersed using 1800 lines/mm gratings, then analyzed by a deep depletion CCD detector (1024 × 256 pixels).

In order to verify data consistency, the measurements were obtained on at least ten carbonaceous particles per sample. Renishaw Wire 4.1 was used for instrument calibration and Raman measurements. Before each measurement session, the microspectrometer was calibrated using the 520.5 cm⁻¹ line of a silicon standard.

All around the Zaian Mts and in the J. Hadid horst (Fig. 3), the Visean unconformity is well-exposed in several sites. In most of them, an angular unconformity of ~45° or more is observed above the Ordovician tilted strata, which implies the occurrence of a generalized pre-Visean folding event. Only locally (e.g., Takkat area, W of Sidi Lamine) the Upper Visean formations overlie conformably the Cambrian-Lower Ordovician formations, as a consequence of the location of these outcrops on the crest of a large pre-Visean anticline. We present hereafter the structure from two areas, southwest and northeast of the Zaian Mts, respectively, separated from each other by the NE-striking Central Zaian dextral fault (CZF).

The structure of the Zaian Mts is relatively simple in the Sidi Lamine-Kef Nsour area south of the CZF (Fig. 3A). Two regions can be distinguished there, based on the type of the earliest deposits over the Upper Visean unconformity. In the Kef Nsour region to the east of the Oulad Bacha reverse fault (Fig. 4A), the earliest unconformable deposits are shallow water limestones (Fig. 4C), whereas in the region to the west of the fault they consist of coarse sub-aerial conglomerates (Fig. 4B), followed upward by limestones similar to those of the Kef Nsour region.

The folds that affect the Upper Visean deposits (Variscan folds) above the unconformity are upright open folds with kilometre-scale wavelength (Fig. 4A). The argillaceous beds display a subvertical spaced cleavage “S2” (Fig. 4B), whose intersection with bedding results in flake or pencil structures. In contrast, the folds are much tighter in the underlying, Ordovician formations, even in the quartzite series, and they are associated with an axial-planar, subvertical to moderately west-dipping foliation “S₁” (Fig. 4C and D). Locally (Fig. 4E and F), the unconformity at the bottom of the Upper Visean limestones is superimposed by a décollement surface associated with a cataclastic foliation in the uppermost Ordovician slates.

North of the CZF, and particularly in the northeasternmost part of the Zaian Mts, the structures of the pre-Visean fold belt are more complex than those of the Sidi Lamine-Kef Nsour area. South of Aguelmous, the Bou Dobra massif exposes a succession of globally east- to northeastward-inclined or recumbent quartzite folds (e.g., Fig. 5B) thrust upon each other.

The best site to observe the Visean unconformity in the area is located at Sidi Lhoussine (Fig. 5B), north of the easternmost frontal fold of the Bou Dobra massif (Fig. 3A). There, the Upper Ordovician metapelites (“Schistes d’Asfar” Fm.) support an asymmetric, W-verging open syncline of Upper Visean formations (Fig. 3A, sites 1, 2). The western limb of the syncline displays the best exposures and shows from bottom to top: (i) Upper Ordovician sandy-pelitic beds, dipping steeply (70–80°) E to ENE-ward and affected by a pervasive slaty cleavage (S₁) dipping 30–40° ENE-ward (Fig. 5D); (ii) 10 m-thick, shallow-water sandy limestones that constitute the lowest Upper Visean unconformable deposits; thin conglomeratic beds only appear on top of these beds; (iii) flysch facies with alternating silts or pelites and thin carbonate sandstone beds. The structure of the rocks drastically changes going across the unconformity. The slaty cleavage of the Ordovician rocks does not affect the Visean rocks. A foliation slightly oblique to bedding is observed locally in the pebbly silts on top of the limestones (Fig. 5E). The lowest overlying pelites only show a bedding-parallel foliation that moulds the diagenetic carbonate nodules (Fig. 5F). The turbiditic (flysch) sequence above is affected by two types of minor folds. The variably oriented, smallest and oldest folds can be regarded as slumps, whereas the largest are related to the post-Serpukhovian (Variscan) deformation. Their axes trend N-S to N20E; some have shallow dipping axial planes, but most of them show steeply dipping axial planes and can be regarded as minor folds associated with the Sidi Lhoussine major syncline. Spaced cleavage (S₂) only appears in the tighter hinges of the latter minor folds (Fig. 5C).

In the Tanayfout area northeast of Aguelmous (Fig. 3A), the tilted Upper Ordovician metapelitic beds (“Schistes d’Asfar”) are affected by a penetrative foliation almost normal to bedding (Fig. 5G). The overlying, coarse conglomerates exhibit only brittle tectonic structures (fractured pebbles), except in some matrix-rich beds within which a foliation occurs, almost parallel to bedding and then similar to that observed at Sidi Lhoussine (Fig. 5E).

Around Goaida centre (Fig. 3A), the Ediacaran Schist-Tuff Fm. and granites (Pan-African basement; Ouabid et al., 2017) are exposed beneath the Lower and Middle Cambrian unconformable formations, which in turn are surrounded by the folded Cambrian-Ordovician quartzitic series. The post Pan-African structures first include a shallow dipping thrust fault that carries a mylonitic granite sliver from west to east on top of the main granite massif (Cailleux, 1994; Ouabid et al., 2017); the thrust fault roots in the NE-striking major fault that bounds the Goaida massif to the west (Fig. 3A).

Second, ductile shear structures are conspicuous in the Lower Cambrian conglomeratic marbles overlying the main granitic massif at Goaida centre. The kinematic indicators such as sigmoidal boudinaged pebbles, boudinaged early quartz veins and back-rotation of boudins (Fig. 6A) all point to an intense eastward shearing within the Lower Cambrian beds. To the NE, the Lower Cambrian marbles and overlying Middle Cambrian metagreywackes disappear beneath the large folds of the Cambrian-Ordovician quartzite series (Ouerdane-Mtourzgane unit; Fig. 3A). The contact between the Lower-Middle Cambrian shallow dipping formations and the folded quartzite series is sinuous in map view, and truncation of the thick quartzite beds on the contact can be observed on the satellite imagery (Google Earth), suggesting a shallow dipping tectonic contact between the Goaida massif and the Ouerdane-Mtourzgane unit. Similar features characterize the southwestern boundary of the Goaida massif (Fig. 3A).

The Bou Issardene recumbent folds (Fig. 6B) offer another illustration of the structures formed in the deepest parts of the Zaian Mts. First described by Cailleux et al. (1983), these folds occur southwest of the Mtourzgane summit and ESE-trending synclinal axis (Fig. 3A). Their geometry suggests a SW-verging shear strain in ductile conditions in the lowest levels of the western limb of the major Mtourzgane syncline.

Examination of thin sections of pelitic lithologies (Fig. 7) from both sides of the Visean unconformity highlights the contrasted metamorphic evolutions of the Cambrian-Ordovician and Upper Visean rocks, respectively. The Upper Visean rocks exhibit spaced cleavage (SC), frequently lacking and developed in the tightest parts of the folds (Fig. 7A). This cleavage basically results from crenulation of the bedding-parallel burial foliation, and from pressure solution processes that affect the carbonate grains. As this cleavage is linked to the second regional compressive phase (post-Serpukhovian Variscan event), it is referred to as S₂ cleavage.

On the contrary, the pre-Visean rocks show a tight, penetrative foliation axial-planar to the micro-folds, and crosscutting even the impure sandstones beds (Fig. 7B and C). This well-marked foliation is associated with coeval shear zones (“C/S” structures; Fig. 7B). The large detrital mica grains are plastically twisted (1, Fig. 7C) or elongated and recrystallized (Fig. 7D). These clastic micas have been commonly converted into shuttle-shaped, mixed-layered phyllosilicates, consisting of an intergrowth of chlorite and mica (?). The quartz grains are boudinaged and affected by dissolution-recrystallization processes (grains 2, Fig. 7C). Tenuous phyllitic grains, at least partly recrystallized, extend in the tectonic foliation domains.

The T_RSCM measurements have been performed successfully in 25 samples from sites located around (samples 1–29, 37) and within (samples 30–35) the Zaian Mts (Fig. 3A). Table 1 shows the T_RSCM results obtained from all these samples, together with the type of cleavage observed in the field and at the microscope scale.

The T_RSCM results are plotted on the geological map (Fig. 8), without their error bar. The eleven sites located within the Zaian Mts and Jebel Hadid yield T_RSCM values ranging from 290 ± 9 °C (KN37, Upper Ordovician) to 365 ± 7 °C (GO34, Lower Cambrian), with a mean value of 323 ± 23 °C, excluding #samples 12 and 13 from the Bouhssoussene shear zone across the Fourhal Basin (this strongly folded basin is outside the Zaian Mts; see Sect. 6).

Upper Visean samples have been collected from eleven sites around or within (sample GO35) the Zaian Mts. Leaving aside sample #15 (530 ± 32 °C) collected in the Ment granite aureole and sample #10 (298 ± 6 °C) from the Bouhssoussene shear zone inside the Fourhal Basin, the T_RSCM values range from 160 ± 20 °C to 260 ± 10 °C, with a mean value of 209 ± 37 °C.

Sample #16 was collected for comparison in the frontal part of the Khenifra nappe (Allary et al., 1976; Bouabdelli, 1989, 1994). The T_RSCM measured in this sample (296 ± 14 °C) fits with the low-grade greenschist facies evolution described in the Ordovician series of the Khenifra nappe by the cited authors.

The evolution of Raman spectra of carbonaceous material from selected Cambrian-Ordovician and Upper Visean samples with increasing peak temperature is illustrated in Fig. 9. Raman parameters and associated T for all spectra from the studied samples are given in Table S1, freely available as Supplementary Material linked to this article on the GSW website of the journal: https://pubs.geoscienceworld.org/eurjmin.

The most important feature of the Zaian Mts is the occurrence of a generally high-angle unconformity between the Cambrian-Ordovician siliciclastic series and the overlying Upper Visean-Serpukhovian carbonate turbiditic series. Both juxtaposed series are folded, but fold axes trend differently in each one, except along the NW boundary of the massif, i.e., along the Fourhal basin. Fold axes dominantly trend N-S in the Zaian Massif itself, and NE-SW in the adjoining Carboniferous basins (Fig. 3A). The lowermost Upper Visean deposits, which include subaerial conglomerates and shallow-water limestones, overlie unconformably the folds of the Cambrian-Ordovician massif (Fig. 4A–C). Hence we assume that two main folding episodes shaped the Zaian area, (i) a post-Ordovician, pre-Upper Visean episode and, after a period of subaerial erosion, (ii) a post-Serpukhovian, pre-Mid Permian episode sealed by the intrusion of the 280–270 Ma-old granite of Ment. The latter folding episode corresponds to the major Variscan phase of Morocco and Western Europe (Matte, 2001; Michard et al., 2010). The previous folding may have occurred during the Silurian-Devonian (Acadian orogeny of Van Staal & Barr, 2012; early Variscan tectonism of Matte, 2001; Eo-Variscan cycle of Faure et al., 2014) or even during the Late Devonian-Early Visean transition, thus being a late Eo-Variscan event.

In map view, the Zaian Eo-Variscan fold axes exhibit virgations. Some (e.g., Sidi Hassine area) are spatially linked to the dextral throw of the CZF major fault, which was probably active during folding. However, the Mtourzgane-Bou Dobra major virgation defines a tectonic arc whose “arrow” (Elliott, 1976) points to a NE direction of transport.

The profile across the southern Zaian Mts (Fig. 10A) makes visible the moderate eastward vergence of the folds, and suggests that the folded Paleozoic series are detached from the Precambrian (Pan-African) basement. This detachment would occur in the Lower Cambrian carbonate levels, converted into calc-mylonites as observed at Goaida (Fig. 6A). The basement itself is shortened through brittle faulting, as illustrated along the western border of the Goaida massif (Fig. 3A; Cailleux, 1994; Ouabid et al., 2017). The occurrence of repeated, shallow dipping shear zones in the basement formations has been described in the J. Hadid by Bouabdelli (1994), suggesting that the basement also underwent distributed, brittle-ductile deformation during the synmetamorphic Eo-Variscan folding episode.

The cross-section (Fig. 10A) also illustrates the effects of the Variscan shortening on the Eo-Variscan structures. The arcuate CZF (Fig. 3A) was reactivated as a reverse-dextral fault, although with a limited strike-slip throw according to the available maps (Allary et al., 1976; Bouabdelli, 1989). The SSW-NNE Oulad Bacha reverse fault separates a western unit (Oued Grou Elbow unit; Fig. 10A) characterized by thick subaerial conglomerates at the base of the Upper Visean deposits from an eastern unit (J. Tighermine-Kef Nsour unit) within which the Upper Visean series begins by shallow-water limestones. This suggests that the Oulad Bacha fault results from the inversion of a former west dipping normal fault active during the subaerial erosion of the Eo-Variscan belt and subsequently sealed by the carbonate, then argillaceous-clastic sedimentation. The Visean unconformity is also deformed by large-wavelength folds, which causes shearing of the unconformity surface. This is well illustrated between the Oulad Bacha Upper Visean limestones and the underlying Ordovician terrains of the Tighermine unit (Fig. 4E and F). The parallel-to-bedding foliation that occurs at the bottom of the Sidi Lhoussine syncline (Fig. 5E) and at Tanayfout would also result from shear activation of the mechanical discontinuity linked to the unconformity during the Variscan folding event.

Variscan re-folding of the Eo-Variscan folds is well-illustrated in the J. Timekhdoudine-Kef Nsour unit, where a large, NE-trending Variscan fold deforms the Eo-Variscan NNW-trending folds (Figs. 3A and 10A), and in the Bou Mahraz-Sidi Salah anticline along the Fourhal-Zaian Mts boundary (Fig. 10B). Such re-folding is necessarily associated with small-scale deformation of the Cambrian-Ordovician terrains. Actually, Allary et al. (1976) describe the zonal development of steeply dipping spaced cleavage and kink-folds associated to the post-Serpukhovian folding of the Timekhdoudine massif. We observed such small-scale Variscan structures superimposed to the S1 Eo-Variscan cleavage (Fig. 11) in the Bou Mahraz-Sidi Salah thrust fold. However, it seems that the most important reworking of the Eo-Variscan structures during the Variscan shortening occurred through reverse faulting. The Bou Dobra Eo-Variscan thrust faults may have been significantly reactivated during the Variscan shortening, which formed the juxtaposed Sidi Lhoussine syncline (Fig. 10C).

The T_RSCM measured in the Cambrian-Ordovician samples are consistent with the sub-greenschist facies of these rocks (Fig. 7), in the sense of Bishop (1972) and Meere (1995). Illite crystallinity (IC) in the Zaian Mts rocks ranges from the lower epizone to the anchizone (Bouabdelli, 1989, 1994), which compares with the IC results from the external Variscides of SW Ireland and points to the same temperature range (~275–325 °C; Meere, 1995).

All samples of the Zaian Mts and J. Hadid display T_RSCM values higher than those of the surrounding Upper Visean samples (Table 1 and Fig. 8). Mean values of 323 ± 23 °C and 209 ± 37 °C, respectively, can be calculated leaving aside samples #15 from the Ment granite aureole and #10–13 from the Bouhssoussene-Aguelmous fault zone. As a matter of fact, sample #15 underwent a late hydrothermal/metamorphic re-heating, and samples #10–13 are localized along a shear zone cutting across the Fourhal Basin and potentially affected by shear heating. So, these samples are not representative of the Zaian Mts/J. Hadid Lower Paleozoic series and closely related Upper Visean deposits.

The T_RSCM measured in samples from the same stratigraphic formation are generally equal within error. Significant differences are observed locally in the T_RSCM of the Upper Visean samples around Sidi Lamine (Fig. 8). They could results from fault-controlled hydrothermal advection during the Upper Visean rifting, comparable with examples described in the Suez Rift by Chandrasekharama et al. (2018).

The T_RSCM mean value obtained for each stratigraphic formation has been plotted in relation to the position of the samples in the stratigraphic column, taking into account the estimated thickness of the formations (Fig. 12). The diagram highlights that the mean T_RSCM of the Visean samples do not plot on the same geotherm as that of the Cambrian-Upper Ordovician (except Hirnantian) samples. This strongly supports the conclusion of the structural study in favour of a polyphase, Eo-Variscan then Variscan evolution of the Zaian Mts.

The diagram (Fig. 12A) suggests that a geothermal gradient of ~35 °C/km occurred inside the Cambrian-Ordovician pile. This relatively hot gradient would correspond to an early, pre-folding equilibration of the geotherm. However, another hypothesis must be considered, i.e., that of a late equilibration of the geotherm by the end of the pre-Visean folding event. As the shortening of the Cambrian-Ordovician section might be estimated from the present-day cross-section (Fig. 4A) at ~50%, the measured thermal gradient would have been established across a 3 km thick folded Lower Paleozoic pile (Fig. 12B). In that case, the gradient in the deeper section of the pile (“deep” gradient in the following) would be only 22 °C/km.

Such a “deep” gradient of 22 °C/km is not a particularly hot gradient by itself, considering the gradient below the “shallow” 1–2 km of the rock section. For example, most of the stable areas of present-day Morocco display geothermal “deep” gradient in the range 20–25 °C/km (Zarhloule, 2004). However, the T_RSCM mean value of 310 °C in the Upper Ordovician strata indicates that a hot geothermal gradient occurred in the overlying (Hirnantian)-Silurian-Devonian folded package. In order to estimate this “shallow” gradient, hypotheses have to be done concerning the age of the folding event and the thickness of the sedimentary sequences deposited before folding occurred. We may consider two extreme hypotheses as follows:

• An early folding event, referable to an Acadian phase of late Early Devonian-Middle Devonian age (Fig. 12B). This phase has been repeatedly evoked in the Western Meseta to account for conglomeratic formations (cf. Rehamna and Jebilet massifs, Michard et al., 2008, 2010; Imouzzer du Kandar outcrops, Charrière & Regnault, 1989). The corresponding sedimentary sequence on top of the future Zaian Mts would have hardly reached 400–500 m, by comparison with the adjoining areas where the sequence is preserved (Zahraoui, 1994a and b; Baudin et al., 2001; Razin et al., 2001; Hoepffner & Houari, 2006). Folding and duplexing would have thickened this cover up to ca. 1 km at a maximum, and the shallow geothermal gradient should have been close to 300 °C/km.

  Such an extremely hot, shallow geothermal gradient is unlikely. Moreover, the occurrence of an Acadian folding phase in southwestern Meseta is controversial, and the Lower Devonian conglomerates could as well result from block tilting in an extensional regime (Cornée, 1989).

• A recent folding event referable to a late Eo-Variscan phase (Fig. 12C); this phase would have occurred during the Late Devonian-Tournaisian (Early Visean?). A late Eo-Variscan phase is attested by field evidence at short distance to the northeast of the Zaian Mts, in the Bou Khadra allochthon (20 km north of Mrirt, Fig. 1B). There, an unconformable, upper Tournaisian conglomerate overlies folded Upper Devonian terrains (Bouabdelli, 1994). More to the east, a tectonic-metamorphic event has been dated around 360 Ma in the Debdou-Mekkam and Midelt massifs (Fig. 1B) by Rb–Sr (Clauer et al., 1980) and K–Ar (Huon et al., 1987) analyses of <2 μm mica fractions. The pre-folding cover sequence in the Zaian area could have reached a great thickness, assuming it was comparable to the sequences preserved in the adjoining areas. For example, the Hirnantian to Famennian-Tournaisian series occur in the western and northern parts of Western Meseta with a thickness of ~1 km (Fadli, 1994; Zahraoui, 1994a and b; Baudin et al., 2001; Razin et al., 2001). In the western Tazekka units, Hirnantian periglacial deposits, Silurian and Devonian formations occur with a total thickness of several hundred metres (Hoepffner, 1994; Ghienne et al., 2018). The Zaian Mts are located in the axis of the “Marrakech-Oujda trough” (Piqué & Michard, 1989; Hoepffner et al., 2005) that extended northeastward to the Oran Meseta (Bougara et al., 2017) and was characterized by a pelagic-turbiditic sedimentation from the Lower-Middle Devonian up to the Famennian-Tournaisian. The Givetian-Frasnian turbiditic facies alone reached ~1 km thickness in the Debdou-Mekkam (Marhoumi et al., 1983). Then, the thickness of the Hirnantian-Late Devonian series in the Zaian area could have been close to 2 km before tectonic shortening and close to 3 km after tectonic thickening. In that case, the “shallow” gradient would have been about 100 °C/km within the Zaian uppermost package. The origin of this reasonably hot gradient is discussed below (Sect. 6.3).

Two more questions must be addressed here: (i) what was the geothermal gradient within the Lower Carboniferous terrains accounting for the T_RSCM values measured in the Upper Visean beds close to the regional unconformity, and (ii) what were the thermal effects of the Variscan events in the Cambrian-Ordovician units beneath the unconformity?

The T_RSCM values reported above from Upper Visean samples are too scarce and grouped close to the unconformity for defining a geothermal gradient in the Azrou-Khenifra or Fourhal basins. However, the thickness of the fold-thrust Lower Carboniferous prism would have hardly exceeded 5 km before erosion (Ben Abbou et al., 2001). Then, a “shallow” geothermal gradient of ~40 °C/km can be inferred (Fig. 12D). This is a hot gradient, consistent with the synsedimentary bimodal magmatism observed in the area (Roddaz et al., 2002). If this gradient is extrapolated from the Lower Carboniferous to the 3 km-thick Cambrian-Ordovician package of the Zaian Mts, the temperatures then would have reached ~220 °C in the Upper Ordovician and ~340 °C in the Lower Cambrian at a maximum (Fig. 12D). A lower, but still hot “deep” gradient (e.g., 30 °C/km), which is more likely at depth, would lead to only 310 °C in the Cambrian rocks. In both cases, the Variscan temperatures of the Cambrian-Ordovician beds remained lower than the measured T_RSCM. Then the latter must be definitely ascribed to a pre-Visean, Eo-Variscan thermal event.

On the other hand, these Variscan temperatures are high enough to account for the K/Ar dates of 335 ± 6, 283 ± 5 and 318 ± 6 Ma reported by Bouabdelli (1989) from <2 μm white mica fractions of the Zaian Mts and J. Hadid pre-Visean rocks. Consistently, these ages are in the same range as those from the adjoining Fourhal and Azrou-Khenifra basins (289–330 Ma; Huon et al., 1987).

According to the above discussion, we assume that a hot “shallow” geothermal gradient (~100 °C/km) coexisted with a normal “deep” gradient (~22 °C/km) within the Paleozoic series by the end of their Eo-Variscan (early Tournaisian?) folding (Fig. 12C). The hot shallow gradient could result from a dominant conduction regime in the Silurian-Devonian package, consistent with its richness in impermeable, argillaceous layers. In contrast, the lower gradient in the Cambrian-Ordovician package, which is rich in brittle quartzite beds, would be linked to dominant advection. This is consistent with the importance of the dissolution-crystallization processes in the latter rocks (Fig. 7B–D), and the abundant quartz veins in the quartzite beds. The limit between the two Zaian packages would correspond to the classical décollement level of the Silurian shales (Cailleux, 1978; Baudin et al., 2001).

The bulk, linear geothermal gradient between the surface and the Cambrian décollement level defined above (Fig. 10A) has been estimated at ~60 °C/km (Fig. 12C) by the end of the Eo-Variscan folding event (360–350 Ma). The origin of such a hot geothermal gradient must now be discussed.

Magmatic outpours of Devonian-Tournaisian age are unknown in the Zaian Mts yet. However, magmatism is widespread around the Zaian area during the Famennian-Tournaisian: (i) mafic or bimodal magmatism began at 358–345 Ma in the Jebilet massif (Delchini et al., 2018); (ii) mafic to andesitic lava flows are intercalated in the Famennian-Tournaisian turbidites along the WMSZ and RTF fault zones (Piqué, 1981; Fadli, 1994; Kharbouch, 1994); (iii) volcanic fragments are recorded in the Tournaisian olistostromes north of the Central Massif (Tahiri, 1991); (iv) a pyroclastic intercalation occurs in the Lower-Middle Devonian series of the Eastern Nappes (Bohrmann & Fischer, 1985), and (v) the Debdou-Mekkam massifs of Eastern Meseta yielded a well-defined Late Devonian (~380 Ma) population of detrital zircons (Accotto et al., 2018). These early records of magmatic activity around the Zaian Mts support the idea of heating of the upper crustal levels due to mafic or bimodal magmatic activity and fluid advection related to lithospheric extension during the Devonian. Moreover, outgassing of hot mantle volatiles may occur away from volcanic outpours (Caracausi and Sulli, 2019), and such outgassing might have affected the Zaian Mts.

On the other hand, the formation of depocenters contrasting with uplifted blocks and the generalized reworking of platform deposits toward the basins are well documented during the Middle-Late Devonian and suggest a regime of extension/transtension in the Meseta domain (Piqué, 1981; Piqué & Michard, 1989; Hoepffner et al., 2005; Becker et al., 2015). Likewise, extension characterized the Anti-Atlas domain during the same span of time (Baidder et al., 2008), and more generally the whole northern margin of Gondwana (Frizon de Lamotte et al., 2013).

The concept of thermal setting of a compression fold belt inherited from the previous extensional setting has been developed recently in the Pyrenees (Vacherat et al., 2014, 2016; Jourdon et al., 2019; Odlum & Stockli, 2019; Bellahsen et al., 2019). Here, we propose that the high linear geothermal gradient (~60 °C/km) reconstructed in the Zaian Paleozoic cover during its Eo-Variscan folding is inherited from the Late Devonian, which corresponds to the climax of the lithospheric extension in the Meseta and Anti-Atlas domain. The model suggested here for the building of the Zaian Mts (Fig. 13) includes thermal inheritance as a key process to explain the high T_RSCM values measured in the Cambrian-Ordovician rocks. The Gondwana margin in the Meseta transect during the Late Devonian would correspond to the “hot magma-poor passive margin” type of Clerc et al. (2018).

The crustal segment shown in this qualitative crustal model is located above the subducting Rheic slab (not in view; see Fig. 2). During the climax of the Devonian extension (Fig. 13A), and because of the hot mantle anomaly, the 300 °C isotherm climbs above the Cambrian-Ordovician formations of the Zaian area of the Marrakech-Oujda trough. By comparison with some Pyrenean models (Lagabrielle et al., 2016; Asti et al., 2019), detachment of the Cambrian-Ordovician series and part of their folding could have already occurred at this stage. At the onset of the Eo-Variscan compression, i.e., during the Famennian-Tournaisian transition (Fig. 13B, 360–350 Ma), inversion of the crustal normal faults causes the shortening of the most stretched part of the crust and triggers (or increases) the folding of the sedimentary series of the Zaian belt. The Ordovician duplexes shown further to the NW in the Central Massif could result from this phase (Cailleux, 1975, 1987). The Zaian belt emerges, as well as large parts of the Meseta domain (Coastal Block, Sehoul, Marrakech Atlas, Tazekka and Eastern Meseta; Michard et al., 2008, 2010). At about 340 Ma, the emerged areas are eroded and prone to be covered by the unconformable Upper Visean deposits.

The subsequent Upper Visean-Serpukhovian subsidence occurred in a fore arc syntectonic setting in front of the Eo-Variscan, Eastern Meseta-Tazekka thrust-fold prism (Bouabdelli, 1989, 1994; Bouabdelli & Piqué, 1996; Ben Abbou et al., 2001). Calc-alkaline pillow basalts and dolerite sills emplaced in the turbiditic deposits, and would result from wet melting of the Moroccan metasomatized mantle (Roddaz et al., 2002). This geodynamic context accounts for the still hot geotherm suggested above for the Zaian belt by the end of the main Variscan folding event (Fig. 1D).

Based on a new structural analysis, we deduced that the Zaian Mts expose a pre-Visean fold belt embedded in the Late Carboniferous-Early Permian Variscan Orogen. Using the RSCM method, we gave here the first quantitative geothermic characterization of the pre-Visean and Variscan events in the Zaian Mts area.

The early, pre-Visean fold belt developed in relatively high-temperature conditions (290 ± 9 to 365 ± 7 °C) corresponding to the sub-greenschist facies under a ~3 km sedimentary-tectonic burial. By contrast, the Variscan folds around and above the early fold belt formed under low-temperature conditions, 160 ± 20 to 260 ± 10 °C. We propose that the Eo-Variscan high geotherm is inherited from the Late Devonian extensional setting. The Paleozoic series possibly detached from the extended crust and suffered part of their folding during the climax of crustal extension. However, the Zaian Eo-Variscan fold belt was only achieved through a compressional event, which resulted in its emersion and subaerial erosion before the Upper Visean. This folding event would have occurred during the Devonian-Tournaisian transition or early Tournaisian.

The early fold belt subsided, together with the whole Meseta, during the Upper Visean-Serpukhovian before being shortened again by the Variscan tectonics of compression-transpression. The geothermal gradient was still hot, but less than during the Eo-Variscan event, which permitted the preservation of the T_RSCM reached earlier in the Cambrian-Ordovician rocks.

Hence, the Zaian Mts appear as a fine example of thermal inheritance during the extension-to-compression transition, as well as a key area for the geodynamic interpretation of the Moroccan Meseta orogen. The next frontier is to understand which phenomenon triggered the onset of the Eo-Variscan compressional event and caused the Upper Visean unconformity – an issue that must be considered at the scale of entire Morocco.

Logistic support from the Faculty of Sciences Semlalia and the Ecole Normale Supérieure, Cadi Ayyad University, Marrakech, and from the Faculty of Sciences Ain Chock, Hassan II University, Casablanca, are gratefully acknowledged. We warmly thank our anonymous reviewers and guest editor Hans-Peter Schertl for their remarks and recommendations.