Northern Algeria displays a complex geological setting (Fig. 1). It is mainly made of a segment of the Alpine belt that runs from Gibraltar to Calabria. In Algeria, the main units of this belt, also called Maghrebides [5], have been described by Durand-Delga [24,25] and Wildi [55]. In the south, the External Zones, also called the Tellian zones [55], have a width of a hundred kilometers. They are characterized by south- or southeast-verging Miocene folds and thrusts. The Tellian zones overthrust the Atlas foreland. Further north, two successive domains are overthrusting the External Zones from south to north, respectively: (1) the flyschs units (Massylian, Mauretanian, and Numidian), corresponding to sediments deposited in the Maghrebian Tethys that was later on subducted [12,18], and (2) the Internal Zones of the belt, made of an Hercynian or older basement and its sedimentary cover, the so-called ‘Dorsale calcaire’ [24]. Internal Zones form discontinuous massifs spread along the coast of central Algeria, namely from east to west: Lesser Kabylia, Great Kabylia, Algiers, Chenoua, and Cape Tenes Massifs (Fig. 1), from east to west. About 400 km west of Cape Tenes, they outcrop again in Morocco, in the Bokoya Massif and between Jebha and Ceuta (Fig. 1). Although the exact palaeogeography of these massifs within the AlKaPeCa domain is poorly constrained (see discussions in [12,29,39]) and part of the Tethyan ocean may have separated the AlKaPeCa from Europe (see discussion in [43], and [19]), we will assume here that they initially belonged to the European side of the Tethys, in a forearc position, close to the Balearic islands and Sardinia. The AlKaPeCa blocks separated from the European plate at the opening of the Algero-Provençal back-arc basin (Early to Middle Miocene, e.g., [17]), then drifted to the south-east, and began to collide the African passive margin between 18–15 Ma ago ([37] and [27], respectively), possibly followed by a >560-km westward migration of the Gibraltar Arc after ∼16 Ma [39]. However, the ages of these events still remain controversial, and the way the tectonic evolution of the area occurred is also highly debated. For instance, some palaeostresses reconstructions based on microfractures analysis also led to propose a major distensive period from Late Oligocene to Middle Miocene [2,3]. Other debates concern the large massifs of Kabylies, which are assumed to have undergone either an early frontal, north–south ([17,39] and references therein), or oblique, NW–SE [27,29] collision. Furthermore, large strike-slip zones [38] are believed to have segmented the Internal Zones and are supposed to extend offshore, but whether they are inherited from the palaeogeography of the northern palaeomargin of the Maghrebian Tethys [13] or whether they result from the development of a bookshelf faulting system in the Plio-Quaternary stress field [45], is unclear, due to the lack of accurate marine data.

In the following, we will assume a simplified evolution from a pre-Miocene period, where the strike-slip activity was dominant, and a post-Miocene one, characterized by compression, starting at around 8 Ma ago [6]. Nowadays, the convergence of the African plate towards the Eurasian plate is ∼5.1 mm yr⁻¹ in a N60°W direction [47] at the longitude of Algiers. Since the Quaternary, compressional structures on-land provide an approximate shortening rate of 1–2.3 mm yr⁻¹ for the Rif and Tell domains [42]. Therefore, an important part of the North African deformation (up to 50%) could be located offshore and further north on-land in the Betics. The May 21, 2003 Mw 6.8 Boumerdes earthquake [8] has demonstrated the existence of active north-verging structures offshore that are indeed accommodating part of the convergence [21,22]. They also represent an important seismic hazard for the coastal cities of Algeria, including Algiers. Several other seismicity patterns further support this assessment (e.g., [57]). These reverse faults appear to have a vergence opposite to most of those of the main Miocene thrusts of the Tell–Atlas. This pattern may arise from a major change in the tectonic style of the region through the reactivation and inversion of previous tectonic structures of the Maghrebian belt and Mediterranean margins in the present-day NW–SE compressional system [53]. Indeed, stress conditions in the region have not always been steady: important changes of the tectonic regime occurred from Late Cretaceous to Early Oligocene [48,51], and the African plate motion varied during the last few Ma [16,40]. Therefore, from our new dataset, we also aim at clarifying whether the current structural pattern is influenced by older (pre-Miocene and Miocene) inherited structures.

The MARADJA cruise took place on board the R/V Le Suroît (IFREMER) from 21 August to 18 September 2003. It has been scheduled in order to identify accurately the bathymetry and offshore structures from Oran to Dellys (Fig. 2). The area covered by the cruise was divided into three zones: from Dellys to Mt Chenoua, from Mt Chenoua to El Marsa, and finally from El Marsa to Oran (Fig. 2). We gathered more than 5000 km of seismics and continuous seafloor imagery, along 600 km of coastline, from Oran, to the west, to Dellys, to the east. The bathymetric and reflectivity data were obtained using a Kongsberg EM300 Simrad multibeam echosounder (and EM1000 for the continental platform). Simrad EM300 is a 32-kHz multibeam system that allows for overall swath coverage of ∼6 times water depth, increasing with depth to a maximum width of 5000 m at 1000 m. We have obtained a resolution of $15\times 35\text{m}$ at 1000-m depth with a vertical accuracy from 2 m (central beam) to 10 m (lateral beam). Sound speed is estimated from regularly spaced velocity profiles for accurate depth conversions. The bathymetric and backscattering data have been processed with the CARAIBES^® software (IFREMER). We produced a digital elevation model for the seafloor topography, with a resolution of 50 m for the regions of Oran and Tenes, and 25–35 m (depending on depth) for the Algiers zone. Two types of seismic reflection data were obtained, i.e. 6- or 24-channel seismics (4169 and 802 km of sections, respectively). The SU (Seismic Unix) CWP/SU software (Center of Wave Phenomena, Colorado School of Mines) was used for the stack and migration of the seismic data. In addition, a 2–5-kHz CHIRP sonar was used during the entire cruise, as well as a gravimeter and a magnetometer, and eight cores (56 m in total) were successfully obtained by a Küllenberg piston corer. For the purpose of studying the morphology of the seafloor and the structure of the margin, we will mostly use the results obtained from the bathymetric and seismic data, and describe the margin from east to west by dividing it into 3 main zones (Algiers, Tenes, and Oran, Fig. 2).

Between Algiers and Dellys (Fig. 3), the continental platform is narrow (especially close to the Algiers massif where the slope is as steep as 15%), except in the bays, where it widens. Along the margin, our new bathymetric map reveals numerous canyons, generally presenting a network of tributaries. The two main ones are the Algiers canyon and the Dellys canyon [10]. The first one is a meandering canyon, formerly draining the Isser River, according to the palaeo-location of the river [10], and presents signs of deviations (Fig. 3). The second one is rather straight, with several tributaries, and probably drains the Sebaou River. Some limited areas on the slope present a smoother topography, without any canyons (see s in Fig. 3). The continental slope east of Algiers is strongly irregular, with an intermediate flat of various width and a rather sinuous slope break (east–west, NE–SW and NW–SE segments). In this area, the margin does not exceed 20 km in width, and is depicting NE–SW slope breaks striking in a N70° mean direction, a direction already known on-land, through folds in Great Kabylia, for example, the ones in the Tizi-Ouzou basin (i.e. [50]). Two different types of slope breaks were identified: a rollover basin on the slope, formed on top of a listric normal fault, and a piggyback basin further down in the deep basin ([22]; Fig. 3), a simple sedimentary filling behind a blind thrust. These slope breaks result mostly from the Plio-Quaternary activity of deeper thrust ramps [22], one of them striking ∼N70° being assumed to connect at depth to the Boumerdes earthquake rupture (this strike corresponds to the one found by Delouis et al. [21], from seismological and geodetical observations). Folding affects the layers up to the top of the Quaternary layers (Fig. 3, seismic line 1), regardless of salt diapirism effects. We also identify east–west lineaments on the bathymetry in the deep basin. They may underline the presence of tenuous accidents, not necessarily active anymore, along which the salt extruded and formed salt walls (Fig. 3). West of Algiers, the slope changes to a N115° direction (again, a well known direction in Great Kabylia, [2]) as it encounters a large submarine bank (Khair al Din). As a consequence, the margin width increases up to 45 km north of Tipaza (Fig. 3).

The Khair al Din bank is an elongated high (500-m depth), previously roughly known (e.g., [38]), overhanging the deep basin as deep as 2700 m. On the eastern flank of the bank, the canyons are narrow, dense and straight, and look more like linear gullies. On its northwestern side, the slope break is striking N70°E and is very sharp. We identify there a deep-sea fan and an area without canyons, probably smoothed by sedimentary processes like slope instabilities. On the seismic line 2 (Fig. 3), active thrust faulting is suggested by the folding of the Plio-Quaternary sediments and numerous reverse faults distributed above the acoustic basement. In the deep basin, the main features are salt domes organized as walls of several kilometre length.

This main change of morphology between the western and eastern sides of Algiers may indicate that the Boumerdes area is composite: Kabylian Oligo-Miocene sediments or Flyschs units and some volcanic deposits can compose it, taking into account the continuity that should exist between known onshore outcrops in the Boumerdes-Dellys area [55] and offshore rocks. Conversely, the Khair al Din bank corresponds to a relic of the Kabylian basement, like the nearby Algiers massif (Fig. 6). The difference of lithology could also explain the difference of drainage pattern and faulting style along this part of the margin (i.e. several successive thrusts striking northeast and southwest across the margin east of Algiers, but only one major north-verging thrust striking N115° between Chenoua and Algiers).

On the southwestern end of Khair al Din, we identify the prolongation of a major tectonic feature, known on-land as the Chenoua anticline. This south-verging thrust fold (Fig. 3, Inset 3) has been seismically activated on both sides of the shoreline during the 29 October 1989 $M_{\mathrm{S}}$: 6.0 reverse faulting event [41]. Thanks to four seismic lines crossing this structure (Fig. 2), we are able to map for the first time the eastern offshore prolongation of the Chenoua thrust fold (Fig. 3).

The main thrusts identified offshore the Algiers zone on the slope or in the deep basin (Fig. 3) have a northern vergence, therefore opposite to most tectonic features found inland ([27,52] and references therein). For instance, the Sahel anticline and the Kabylies thrust front are controlled by north-dipping faults (e.g., [11,41]). The offshore thrusts may represent south-dipping backthrusts recently formed (Plio-Quaternary) behind the main south-verging suture south of the Kabylies [22,52]. It is believed that this suture zone formed at the end of the Kabylian collision with the African plate, during the Miocene. However, the position and geometry of the Sahel and Chenoua folds and thrusts relative to the main active thrust suggest that they represent backthrusts of the main south-dipping thrust found at the foot of the Khair Al Din Bank (Fig. 3).

This zone is characterized by a linear and east–west-oriented continental slope (Fig. 4), which is also quite steep (10%) compared to other Western Mediterranean margins (e.g., [1,4,49]): in Cape Tenes, the cliffs almost overhang the abyssal plain. On the eastern side of the zone, we observe the westward end of the Khair al Din bank, also called the perched valley of Dahra [26], which progressively deepens along strike, and finally vanishes into the deep basin. This causes a sharp shrinkage of the margin length (37 km at E2°to 20 km at E1°45′). On the bank, the few canyons are shallow, and two large slide areas (Fig. 4) are located on the flanks of the perched valley. It appears that the entire Khair Al Din block is tilted, and that tilting is controlled by a blind thrust in the front, and accommodated by a normal fault in its back (Fig. 4, seismic line 1).

West of this block, the continental shelf is particularly narrow: the margin width changes from only 20 to 40 km from east to west. The seismic lines offshore Tenes (Fig. 2) depict an acoustically rough basement, very chaotic, sometimes locally folded, on the continental slope (Fig. 4, line 2). The numerous canyons strike generally north–south and depict linear and narrow paths. Furthermore, the crests between the canyons become sharp and the roughness of the topography increases. We interpret this morphology as an evidence for a resistant lithology. Many of the canyons present short along-strike shifts or steps in their bed topography that underline the presence of east–west faults (Fig. 4). On the lower part of the slope, the Pliocene reflectors are tilted towards the north (Fig. 4, seismic line 3). This tilting may be a consequence of several tectonic and or gravity-driven events. Firstly, the salt may have slid downslope: this retreat towards the deep basin could have induced the southward tilting of Plio-Quaternary layers. Secondly, canyon-related sedimentary deposition would have been responsible for the fan architecture of the reflectors. Secondly, the margin has probably been uplifted (as described in the Tenes area [56]), which made the sedimentary layers slide toward the north, and may have triggered the opening of east–west fractures underlined by the east–west lineaments leaving imprints in the sediment layers above the bedrock (Fig. 4, Inset 3). Finally, the tilting of the Plio-Quaternary sediments is not sealed by an unconformity, so that the deformation must still be active. Indeed, a broad zone of north–verging faults and flexures has been predicted offshore along the margin, between El Marsa and Chenoua [56] (Fig. 4).

Finally, northwest of El Marsa, a large deep-sea fan with typical sedimentary waves (e.g., [44]) is found to mark the western end of this part of the margin. It is limited to the south by a linear east–west structure, ∼22 km long, which controls the geometry of the Quaternary deep-sea fan, allowing for the northern levee to become dominant. As shown on the seismic line 4 (Fig. 4), the Plio-Quaternary canyon path lies upon a network of subvertical faults that could be associated with a strike-slip zone. We can follow the canyon almost up to the coast (Inset 5, Fig. 4). Furthermore, we observe a drastic change in the acoustic basement facies below the deep-sea fan, on both sides of the fault system: north of it, we identify a chaotic and highly folded basement (similar to the basement facies found in front of Tenes), whereas to the south the basement is organized as discontinuous, weak reflectors (Inset 4, Fig. 4). This is an indication that the east–west fault zone limits two different geological units: (1) the smooth topography located south of the accident could be the surficial expression of the soft material from the Tellian units; (2) the rough basement, north-east of the accident, could represent the Palaeozoic Kabylian basement or the ‘Dorsale kabyle’ sediments, composed by carbonates and also found onshore at the southern flank of the Kabylian basement [14].

Further west, south of the previously described east–west accident, in the area extending from Oran to El Marsa (Fig. 5), a striking observation is the change of structural direction, from east–west to NE–SW. The continental platform appears to deepen continuously from Arzew (200 m deep) to north of Mostaganem, where is reaches 800 m depth, on its edge. It is narrow in front of the capes (less than 10 km) and widens in the bays (up to 40 km). Between Oran and El Marsa, the continental slope is very steep (10 to 16%) and remarkably linear and narrow (in the area of Arzew, the bathymetry starts at 200 m upslope and suddenly falls at 2600-m depth about 10 km further downslope). It is cut by smooth and shallow canyons perpendicular to the coast, and remains rather linear. According to the smooth topography of the slope, we hypothesize that this part of the margin is made of the same geological units found nearby onland, i.e. the Tellian units composed mostly by soft marls or flyschs nappes [55]. Therefore we propose that we enter here another geological domain belonging to the External Zones.

Between Oran and El Marsa, we observe some lineaments with a N10°trend and, at the foot of the Arzew-El Marsa linear slope, en echelon N45°trending lineaments. Both families of lineaments are clear on a slope map (Inset 3, Fig. 5). The N45° and N10° trending lineaments show the same right-stepping en echelon segmentation. They could be associated to strike-slip movement. However, we do not see any evidence for active faulting on the seismic lines (Inset 1, Fig. 5, shows very horizontal reflectors, except where the salt forms domes, far in the deep basin) and this area is also seismically almost inactive. So it may have played a role only in the past and could be locked in the present-day stress field. The very linear shape of the margin in this area may be due to a former faulting activity along a palaeo-transform margin.

West of Oran, the margin depicts again a major change in strike and recovers a general east–west direction. Furthermore, variations in the slope morphology might indicate the presence of volcanic material, as it was described onshore in this region and in some offshore samplings [23,36]. West of Oran, on the bathymetry, we observe long NW–SE lineaments, with some of them presenting en echelon patterns. The study of our seismic lines shows that each lineament is correlated to a small fault. These faults form a network of fractures typical of flower structures (a negative, symmetric one, and another one, asymmetric). These two structures are probably rooted together on a deeper main fault (Inset 2, Fig. 5). This pattern indicates a dextral strike-slip motion with a normal component, probably corresponding to the eastern end of the Yusuf ridge [20]. Miocene volcanic activity [20,38] also underlines the presence of a faulted zone. Furthermore, there is evidence onland for dextral strike-slip motion [54] exactly in the prolongation of offshore structures.

As a whole, these observations are in agreement with structural observations [20], seismicity pattern and several fault plane solutions of earthquakes [9,31] indicative of a large strike-slip zone west of Oran, associated with the Yusuf ridge [20].

7.1 Limits of the Internal Zones

7.2 Consequences on the tectonic evolution of northern Algeria

7.3 Reactivation of structures

The Maradja'2003 cruise provides for the first time a detailed morphostructural database of the ∼500-km-long Algerian margin from Dellys to Oran. From a preliminary interpretation of the new bathymetric maps and seismic sections obtained, we identify several tectonic features related to the main geological suture of North Africa in the Miocene (the ‘Alboran-Kabylia collage’) and to the coeval opening of the Algerian basin. We find several offshore markers of the boundary between the Internal Zones deriving from the AlKaPeCa microcontinent and the External and flysch domains, and propose a transition to the newly formed deep basin using bathymetry and magnetic anomalies. This allows us to infer that Internal Zones constitute the basement of most of the margin from Dellys to 0°10′E, getting narrower from east to west (from ∼100 km to ∼30 km), and then disappearing at the place where the margin changes strike (Fig. 6). Our observations in the Oran area are in line with several tectonic models [27,30,43] favouring a strong westward translation of the Alboran terrane, shaping the northern and southern margins of the East Alboran sea as transform segments through a westward rollback process. Active folding and faulting is widespread mostly at the outer and inner limits of the Internal Zones. Some old structures, inherited from the Miocene, certainly contributed to the current morphology of the margin. The steep margin between Arzew and El Marsa illustrates this former activity. In the present-day NW–SE stress field, during the Plio-Quaternary, new structures were formed, like the important east–west or NE–SW faults and folds around Algiers. The location of the seismic activity in the process of tectonic inversion of the margin is not necessarily around the old sutures. For instance, in the case of the former transform margin in the Oran region, the present deformation is located in the Cheliff area, further inland. Furthermore, the Plio-Quaternary tectonic style is totally different from the one during the Miocene. For instance, strike-slip activity along NW–SE or east–west structures appears to be weak or absent, whereas reverse dip-slip faulting dominates.

The interpretation of the geological nature of the basement proposed in this paper is only preliminary. We acknowledge that several questions remain open concerning the exact nature, extent and structure of the offshore domain: for instance, whether or not large strike-slip zones (part or not of large pull-apart basins [38]) torn the Internal Zones apart during their drift, or bookshelf faulting [42] plays a role in the last stages of deformation, cannot be assessed from our observations alone. Some other uncertainties still remain: for instance, the exact Internal/External Zones boundary between Tenes and Chenoua probably lies in the coastal zone, but is not properly depicted because not fully covered during our survey. Further cruises are therefore needed to complement this mapping, especially near the coastline. Sampling of the basement through coring, dredging and diving is also necessary in order to better assess our interpretation in most areas. It could for instance lead to a better understanding of the role played by the Yusuf ridge in the tectonic evolution of the area of Oran. Finally, future studies of the gravimetry and magnetic data, together with deep seismic sections, should also provide important insights on the transition from Internal Zones to the basement of the Algerian basin.