At large scale, the Atlas system is comprised between the Tell and South-Atlas Fronts (Fig. 1a). In Algeria, the Atlas system comprises the Atlas fold-thrust belt (Sahara Atlas and Aurès) and the High Plateau domain, characterized by a thin and tabular Mesozoic sequence. Along our cross-section (Fig. 2), the High Plateau domain does not exist and the contact between the Tell and Atlas systems is realized through the ‘Pre-Atlas’ domain, a region characterized by spaced anticlines developed above Upper Triassic evaporite or shallower décollements (Fig. 2).

The contact between the Pre-Atlas and the Sahara Atlas is underlined by a north-vergent thrust fault: the so-called North-Atlas Fault. As the South-Atlas Front, the North-Atlas Fault is a composite feature involving shallow thin-skinned deformation within the sedimentary cover [10,11,21]. However, the basement appears also involved at different steps of the deformation history [49].

The inversion and related uplift of the Atlas system of Algeria result from two main tectonic events. The first one, responsible for major deformation and the development of large NE–SW trending folds, occurred between the Middle Eocene and the Aquitanian [10,44]. The second event generates east–west-trending folds and thrusts during the Pleistocene [1,24,27,33]. Between these two compressive pulses, the Miocene was a period of relative quiescence, characterized by the deposition of a relatively thick flexural sequence deposited in continental or marine environment. Normal faulting developed in some places during this period [10]. The Miocene sequence is preserved along the cross-section (Fig. 2) in the Miocene (Burdigalian to Tortonian) M'sila basin, which rests unconformably on previously folded and faulted Cretaceous rocks belonging to the Atlas system (Fig. 3).

The Tell system (Figs. 1 and 2) is located between the Flysch domain and the Atlas. It is formed by an accretionary complex deriving from the African palaeomargin of the Maghrebian Tethys. This sedimentary cover (Triassic to Neogene) is composed of thick sedimentary sequences of dominantly marly character.

The Tell system overthrusts the Atlas and the overlying Miocene sequences of the M'sila basin. This basin supports a thick pile of allochthonous terranes, the so-called ‘Tellian nappes’, which are gravity-driven nappes forming a quite complex imbricated fan [32]. From a lithostratigraphic point of view, these nappes are formed by Upper Cretaceous to Lower Miocene series [50].

To the north, the décollement flooring the Tell nappes rests upon the large east-trending imbricated stack of south-directed thrusts made up of Jurassic to Albian strata cropping out in the Bibans tectonic window (Fig. 3). In our interpretation, these imbricated strata constitute the former substratum of the Tellian nappes that were detached from it along under-compacted and over-pressured horizons located at the top of the Albian strata. Accordingly, the Bibans window is interpreted as a large duplex with a roof thrust within Albian strata and a floor thrust along Triassic evaporites.

As in the Atlas, basement rocks do not crop out in the External Tell system, at least along our cross-section (Fig. 2), and the involved cover thrust units show a geometry (i.e. folds) indicating that the basement beneath them is nearly flat. This indicates that the basement is not involved beneath the Tell thrust system (and therefore that basement deformation took place more to the north) or that it is deformed forming a duplex with a relatively flat roof thrust (hypothesis chosen in Figs. 2 and 4).

The Jurassic to Albian rocks cropping out in the Bibans window are affected by a synfolding cleavage, which is moreover expressed everywhere in the ‘external massifs’ of the Tell–Rif system [3,19,21,41]. Due to the lack of such a cleavage in the Upper Cretaceous rocks of the Tellian nappes, it has been interpreted by some authors [41] as evidence for Upper Cretaceous compressive tectonic events. On the contrary, in our interpretation, the difference in tectonic style between Lower and Upper Cretaceous levels is related to a strong de-coupling between the hanging wall of the duplex roof thrust and the horses of the duplex. As a consequence, the roof layers were detached to form the Tellian nappes whereas internal deformation, expressed by cleavage, was concentrated in the horses of the duplex.

North of the Bibans tectonic window, the Flyschs domain (Fig. 1) is interpreted as the sedimentary cover of the Maghrebian Tethys [18,25,50]. This domain is presently exposed in large thrust sheets (nappes), which are sandwiched between the Bibans imbricated thrust stack and the Kabylides but also superimposed to them. This last position had generated vigorous debates about its initial position (see discussion in [50]). However, following the stratigraphic and sedimentologic arguments developed in [50], it is presently acknowledged that its initial position was between the ‘Dorsale calcaire’, which represents the Mesozoic cover of the Kabylides, and the External Tell. So, the present-day position of the Flysch nappes, either beneath or above the Kabylides, should result from partitioning in front of the backstop formed by the Kabylian massifs.

In the Flyschs domain, a major décollement is situated between the siliciclastic levels of Lower Cretaceous age and their Jurassic substratum (Achaïches series), which is only known in a small tectonic window below the Lesser Kabylie massif [7]. This décollement allowed the development of a major north-directed backthrust involving only the youngest flysch sequences (i.e., Lower Cretaceous and younger sequences), whereas older Mesozoic basinal sequence were underthrust below the Kabylies massifs. The Oligocene–Lower Miocene levels are also detached from their substratum, forming the Numidian thrust sheet. This upper nappe is primarily composed of an up to 2000-m-thick turbiditic sequence interpreted as a flexural sequence deposited in front of the moving Kabylides and equivalent to the one known in the southern Apennines and in Sicily [13] as well as in the Rif and western Betics [18]. A late out-of-sequence thrusting [39] is required to explain the current position of the Numidian thrust sheet and to achieve the chain building.

The Kabylides form an upper crustal wedge, initially situated north of the Maghrebian Tethys [8], that overthrusts the African margin of the Maghrebian Tethys along a north-dipping major thrust fault that corresponds to the Miocene suture of this ocean. It comprises different nappe complexes equivalent to the ones known in the internal Betics and Rif as well as in northern Sicily and Calabria (see [36] for a review).

The internal structure of the Kabylies is complex, with Alpine superposition of extensional and compressive faults. It includes different nappe complexes, cut by Miocene low-angle, north-dipping extensional faults and active north-dipping [6] and south-dipping (as the fault responsible for the Boumerdès 2003 earthquake [52]) reverse faults. At the moment, the Kabylides look like a giant crustal pop-up structure.

The Kabylides comprise from top to bottom the ‘Dorsale kabyle’ and the Kabylie massifs s.s. The ‘Dorsale kabyle’ complex consists of a pile of thrust sheets involving a thick Lower Jurassic platform carbonate resting on Upper Palaeozoic and Triassic detrital sediments and supporting a thin cover of Jurassic–Cretaceous pelagic and Cenozoic siliciclastic sediments. The ‘Dorsale calcaire’ is considered as the former northern passive margin of the Maghrebian Tethys [8,14]. Below the ‘Dorsale calcaire’, the Kabylie massifs comprise greenschist facies phyllites of Palaeozoic age resting on amphibolite facies rocks constituting the Kabylian basement. Most authors have considered this basement as rigid blocks weakly deformed during alpine orogeny. However, more recent work emphasises the importance of Alpine events (see discussion in [36,44]).

The Kabylides are covered by Chattian–Middle Burdigalian molasses [23]. These so-called ‘Oligo-Miocène kabyle’ molasses of the French authors may be regarded as synextension sediment deposited during the rifting that preceded the opening of the Algerian basin [2,25]. The ‘Oligo-Miocène kabyle’ molasses support tectonic slices pertaining to the Flysch nappes. The tectonic position of these slices is the product of a major backthrust over the Kabylies domain. Finally, the molasses and the nappes are overlain by marine sediments that are Langhian-Early Serravallian in age (i.e., Tizi Ouzou basin) [2,25].

Due to the important uncertainties on the geometry at depth, the Flyschs and Kabylies domains cannot be restore with sufficient confidence. So, we have limited our analysis to the Atlas and Tell systems. For these domains, we have restored the regional cross-section to its original predeformational length using the classical line length restoration technique with the help of the Locace software [40] (Fig. 2). Restoration of the entire section (External Tell and Atlas only) yields a minimum value of 39 km (17%) of horizontal shortening within the Mesozoic cover. This total shortening is distributed as follows: 13 km of shortening (11%) in the Atlas and 26 km (24%) in the External Tell. Restoration of the basement structures in the whole section yields approximately 25 km of shortening corresponding to an 11% ratio, which is significantly lower than in the cover. This implies an important decoupling between the sedimentary cover and the basement everywhere along the section (Fig. 2).

The shortening obtained for the Atlas system is comparable with the ones proposed by Frizon de Lamotte et al. [22] for other sections crossing the Sahara Atlas or the Aurès in Algeria and by [48] for sections through the Moroccan High Atlas. Concerning the External Tell, our estimate (24%) is the first one proposed for this orogenic system in particular and for the Tell–Rif system in general. It could appear as relatively low for a region where large nappes are identified since a long time (see review in [18]), but we must keep in mind that these nappes result from the gravity unroofing of para-autochthonous domains (such as the imbricate stack of thrusts cropping out in the Bibans tectonic window) along shallow décollements. They are consequently quite different from the classical Alpine nappes in which at least the whole sedimentary cover is involved. As a matter of fact, in the Tell, the displacement of some individual Tellian nappes can be larger (>40 km) than the total shortening calculated for the whole orogenic system.

Despite these estimated shortenings in the Atlas and Tell systems as well as the presence of significant compressive deformation in the Flyschs and Kabylides domains at upper crustal levels, the crust and lithosphere do not show evidence of significant thickening south of the Algerian basin. Gravity and heat-flow models [37] show fairly constant crustal thickness of 35 km that decrease rapidly near the coastline to reach the low crustal values of the offshore Algerian basin. Accordingly, the earthquakes are superficial along the Algerian margin as well as in the Tell (see review in [52]). Such lack of thickening at both deep crustal and lithospheric mantle levels could be explained by the presence of a detachment at the base of the upper crust and by the subduction of the underlying levels beneath the Iberian plate. Moreover, seismic tomography provides evidence for a deep detached slab below the Tell [12,42,46,47]. Such a process is also strongly supported by the petrological features of the Neogene magmatism [15,35]. This detached slab would represent the remnant of the subduction responsible for the formation of the whole Maghrebides chain, once connected through the Gibraltar Arc to the Betics and through Sicily to the Apenninic slabs [46].

Previous work done in the Tell and Atlas systems rarely focused on the timing and kinematics of the orogenic processes. Using the THRUSTPACK software [45], a four-step kinematic scenario is proposed in this work, which is the first one for the whole Tell and Atlas systems (Fig. 4). We acknowledge that our scenario remains oversimplified due to the lack of accurate bracketing of tectonic events.

The initial stage (Fig. 4), corresponding to our restored cross-section in Fig. 2, is supposed to represent the state of the region at the end of the Eocene times, just before the beginning of the inversion process related to the convergence between the Africa and Europe plates. It is known that this convergence, which was active since the Late Cretaceous, has been already recorded by some deformation observed in Upper Cretaceous and Palaeocene–Lower Eocene sediments (see review in [26]). However, the effects of these early tectonic pulses remain small and can be neglected at the scale of our section.

The second stage (Fig. 4) corresponds to the Eocene–Oligocene boundary (35 Ma). In the Atlas system, the widely observed angular unconformity between the Oligocene and older sediments is indicative of an important pulse of deformation (Atlas phase) occurring before the emplacement of the Tell units. The same unconformity is observed in the Tell below the M'sila basin [10], but seems absent in the Tell nappes where the passage from Upper Eocene to Oligocene is transitional [51]. Accordingly, Late Eocene deformation has been considered negligible in the External Tell and, therefore, only significant in the Atlas. The lack of compressive deformation in the External Tell, the predominant south-directed vergence of Atlas cover structures and the lack of basement structures beneath the Pre-Atlas and M'sila basin suggest that the boundary between the Atlas and Tell belongs to a major north-dipping basement fault, which accommodated most of the shortening observed in the Pre-Atlas cover structures.

The third stage (Fig. 4) illustrates the development of the thick clastic wedge of Lower to Middle Miocene sediments as a response to the progressive flexure of the African plate towards the north in front of the moving Kabylides. This clastic wedge covered the whole region and lies unconformably, in the central and southern parts of the section (Fig. 2), over the deformed Mesozoic cover of the Atlas. From a tectonic point of view, the Late Oligocene-Middle Miocene is characterized by non-significant tectonic activity in the Atlas and External Tell: only some normal faulting is reported in the Atlas system (see review in [10]). However, more to the north an accretionary prism involving the Flyschs units was already developed in front of the Kabylies on the subducting Maghrebian Tethys.

The final stage corresponds to the present-day cross-section (Fig. 4). To achieve this geometry, at least two deformation events are required. The first event (Middle Miocene) corresponds to the development of the Tell system with the emplacement of the Tellian nappes and the coeval development of the Bibans duplex. The second one belongs to a compressive deformation affecting all the section and expressed by late out-of-sequence thrusts that cut through the previous tectonic pile (Fig. 3). This second step is post-Miocene and already active.