We investigate the lithospheric structure beneath the Gibraltar arc (western Mediterranean) using S-wave receiver functions (SRFs). From a dense network deployed in the Ibero-Maghrebian region during different seismic surveys, we calculated ∼11,000 SRFs that sample the upper mantle detecting the lithosphere-asthenosphere boundary (LAB). The observed seismic LAB belongs to different lithospheric domains: Iberian and African forelands, Alboran domain, and Atlantic Ocean. Common conversion point (CCP) migrated profiles show the geometrical relation among them. Under the Strait of Gibraltar, we observe a deep LAB (∼150 km). It can be associated with Jurassic-age lithosphere of ∼120 km thickness, one of the thickest ever reported in oceanic environments. There is an abrupt offset between the oceanic LAB and the shallow (80-km-deep) continental LAB of the Iberian foreland, suggesting displacement along a former transform fault. The northwestern African continental LAB is 90–100 km deep. The oceanic LAB under the Gibraltar arc continues to ∼180 km depth beneath the Alboran Sea, showing the connection between the Alboran slab and the oceanic lithosphere in the central Gulf of Cádiz. This geometry agrees with an ∼200-km-wide corridor of oceanic lithosphere between the central Atlantic and the Alpine Tethys, developed during the Middle–Late Jurassic. Our results support the proposed westward rollback of an oceanic east-dipping slab, which has continuity at least to the central Gulf of Cádiz.

The southern Iberian Peninsula and the western Maghreb include a sector of the Nubia-Eurasia plate boundary zone that is characterized by complex tectonics, widespread deformation, and the interaction between continental and oceanic domains. At the center of this region, the Strait of Gibraltar separates the Atlantic Ocean (Gulf of Cádiz) from the Mediterranean (Alboran Sea), and at the same time connects the Alpine mountain ranges in Iberia (Betics) to Africa (Rif). The present-day configuration, with the characteristic tightly curved orogenic arc (Fig. 1), results mainly from the interplay between slow, northwest-southeast convergence between Eurasia and Nubia and fast westward slab retreat since the Miocene (e.g., Lonergan and White, 1997; Spakman and Wortel, 2004; Chertova et al. 2014). The complexity of the Iberia-Maghreb plate-boundary zone has been represented with increasing detail as more and more observational data from different disciplines have become available. Yet, fundamental aspects of the geodynamics and lithospheric structure are still discussed controversially (e.g., Gutscher et al., 2012; Mancilla et al., 2013).

The Alboran slab is a key element for understanding regional geodynamics. Together with the Calabrian slab, these structures represent the present configuration of the western Mediterranean subduction system, involving the cold oceanic lithosphere of the Alpine Tethys (e.g., Lonergan and White, 1997; van Hinsbergen et al., 2014). While the Calabrian subduction remains clearly active today, with associated seismicity, active volcanism, and an active accretionary wedge (e.g., Gutscher et al., 2017; Scarfi et al., 2018), there is a lack of evidence for ongoing subduction of the Alboran slab. Deep and intermediate-depth earthquakes do not depict a continuous Wadati-Benioff zone (e.g., Buforn et al., 1991; Heit et al., 2017). Earthquake focal mechanisms, the stress field, and geodetic deformation (e.g., Stich et al., 2006) reflect Nubia-Eurasia plate motion rather than subduction dynamics. Furthermore, absolute plate motions appear to have strong influence on the current slab dynamics (Spakman et al., 2018).

The Alboran slab has been imaged clearly by recent tomographic inversions as a subvertical high-velocity anomaly beneath the western Alboran Sea and southern Spain. However, its possible connection to the oceanic lithosphere west of Gibraltar remains debated (e.g., Spakman and Wortel, 2004; Bezada et al., 2013). The variability in the geoid height in the region (Fig. 1) points to relevant topography of the lithosphere-asthenosphere boundary (LAB) in the region. In this study, we map the depth of the seismic LAB from S-wave receiver functions (SRFs), showing the conversions of teleseismic S-to-P waves at the LAB discontinuity. The analysis builds upon several seismic deployments during the last decade (Fig. 1B), which allows for broad coverage of the LAB over the region, and favors the use of stacking techniques to enhance the LAB signals. We attempt to increase frequency range and spatial resolution compared to previous SRF studies (Miller et al., 2015) in order to shed more light on the interaction of the subduction system with the surrounding lithosphere.

We analyze S-to-P conversions of teleseismic earthquakes (SRFs) recorded at a dense network of ∼300 permanent and temporary seismic stations in Iberia and Morocco (Fig. 1B). This corresponds to an average interstation distance of ∼60 km over most of the area and ∼30 km locally. SRF analysis is a particularly suitable approach to study deep lithospheric structures. S-to-P conversions arrive before the direct onset, thus avoiding the interference of LAB converted signals with crustal reverberations, common in P-to-S receiver functions (PRFs; e.g., Yuan et al., 2006). Furthermore, SRF analysis permits sampling larger areas compared to PRFs, due to the larger incidence angle of S-waves. This allows for better coverage in offshore areas because the piercing points of SRFs are located farther away from the stations (Fig. 1B).

We use teleseismic events with magnitude Mb >5.8 within the epicentral distance range of 65°–85°. We include in the analysis all of the available seismograms with signal-to-noise ratio >2.5, and rotate the three-component records into the ZRT (vertical, radial, transverse) coordinate system using theoretical back-azimuth angles. Following, a rotation into ray-based longitudinal, radial, transverse (LQT) coordinates is done, minimizing the SV-wave energy on the Q component at the theoretical S-wave onset. We time-reverse the traces about the S-wave onset in order to analyze the SRF in the precursor time window, only affected by the primary S-to-P conversions.

Prior to deconvolution, we use a band-pass filter from 1 to 30 s. We apply a time-domain deconvolution method of the SV component (Q component) from the vertical component (L component) using the iterative deconvolution method developed by Ligorría and Ammon (1999) with a 2-s-wide Gaussian basis function. A 6 s low-pass filter is applied after deconvolution to enhance the LAB conversion. We obtain ∼11,000 individual SRFs (Fig. DR1 in the GSA Data Repository1). The LAB can be identified by the prominent negative signal (blue color) after the Moho conversion, at 15–20 s travel time.

Piercing points at 100 km depth show a dense distribution, including offshore areas in the Alboran Sea and Gulf of Cádiz (Fig. 1B). This enables us to build cross-sections by stacking SRF amplitudes applying a common conversion point (CCP) approach (e.g., Mancilla et al., 2015). We consider single scattering, taking into account the first Fresnel zone, for back-projection of SRFs in the IASP91 velocity model (Kennett and Engdahl, 1991). We build north-south and west-east cross-sections by stacking all receiver-function amplitudes with piercing points within 1° distance at both sides of each profile (Fig. 2). We apply a binning of 5 × 5 km2; bins with fewer than 10 samples are not taken into account.

Migrated profiles show clear and coherent signals. Two principal signals can be recognized in all cross-sections. The shallower signal corresponds to a positive conversion (increasing velocity with depth) within the first 60 km, attributed to the Moho discontinuity. Within the resolution of SRFs (>10 km), the lateral variations of the Moho topography in this study agree with previous PRF observations (e.g., Mancilla and Diaz, 2015). The deeper signal corresponds to a negative velocity contrast at depths between 50 and 180 km. This signal is interpreted as the seismic LAB discontinuity.

Common conversion point (CCP) sections across the Iberian massif show a relatively flat LAB located at 80–90 km depth (IBL in Fig. 2, profile W-I). Similar lithospheric thickness has already been reported by previous tomographic (e.g., Palomeras et al., 2017), PRF (Mancilla et al., 2015), and SRF (Miller et al., 2015) studies. This shallow LAB discontinuity has been attributed to a large-scale delamination of thickened continental lithosphere of the Variscan belts in Europe during middle Permian time (e.g., Gutiérrez-Alonso et al., 2011).

In northern Morocco, a west-east profile shows relatively large topographic variations in the LAB (Fig. 2, profile W-III). An offset of ∼40 km occurs near 4°W longitude from 100 km depth under the Rif mountains to 60–50 km depth near the Nekor fault to the east. This observation agrees with previous studies that reported strong crustal and lithospheric thickness contrasts in the area, coincident with the intersection between the Trans-Alboran shear zone and the Moroccan margin (e.g., Gil et al., 2014; Miller et al., 2015).

The largest variations in the topography of the LAB discontinuity are observed beneath the Strait of Gibraltar and Gulf of Cádiz. The CCP profiles reveal two different negative converters (Fig. 2, profiles W-II, N-I, and N-II). The shallower one has depths of 80–90 km and represents the Iberian LAB (IBL in Fig. 2). The deeper signal emerges at 8°W at ∼150 km depth (Fig. 2, profile W-II), starts deepening eastward under the Strait of Gibraltar (∼6.5°W), reaching ∼180 km depth at ∼5°W beneath the Alboran Sea (JOL in Fig. 2, profile W-II). By 4°W, the oceanic LAB (JOL) has disappeared completely, and a shallow LAB at 50 km depth emerges eastern side of the profile W-II (Fig. 2), corresponding to the thinned continental Alboran lithosphere, already imaged in previous studies (ALL in Fig. 2, profile N-III; e.g., Mancilla et al., 2015; Heit et al., 2017). The western termination of the oceanic LAB signal (JOL) can be attributed to the lack of data coverage beyond 8°W (Fig. 1).

We associate the deeper negative velocity contrast (JOL in Fig. 2, profile W-II), under the Alboran Sea, to the LAB discontinuity of the Alboran slab observed in tomography studies (e.g., Spakman and Wortel, 2004; Bezada et al., 2013, Palomeras et al., 2017). The eastern termination of the LAB signal (∼5°W; Fig. 2, profile W-II) agrees with the location where the Alboran slab starts subducting nearly vertically, impeding detection by S-to-P conversions (Li et al., 2011). Our results show the continuity of the LAB discontinuity from Alboran slab oceanic lithosphere across the Strait of Gibraltar to the Gulf of Cádiz (JOL in Fig. 2, profile W-II). This allows for connecting the Alboran slab with the oceanic crust detected in the Gulf of Cádiz (Sallarès et al., 2011), providing direct evidence for the presence of a corridor of oceanic lithosphere beneath the Strait of Gibraltar. This corridor has been inferred by paleogeographic reconstructions to be Jurassic-age oceanic materials connecting the Alpine Tethys with the central Atlantic Ocean (Frizon de Lamotte et al., 2011; Stampfli and Borel, 2002). The extent of the oceanic LAB along cross-section N-I (Fig. 2) shows a good match with the geoid low under the Strait of Gibraltar (Fig. 1).

The apparent width of the oceanic corridor can be estimated as ∼200 km in the north-south profiles (Fig. 2, profile N-I). To the north, the flat oceanic LAB is limited by an offset of ∼70 km with respect to the Iberian continental LAB at ∼37°N. This lithospheric offset may be related to a prior transform fault, in agreement with the sharp transition between continental and oceanic crust observed by Sallarès et al. (2011) and with the location and focal mechanisms of subcrustal earthquakes (Stich et al., 2005) possibly associated with this structure. This observation is supported, too, by CCP images of PRFs along a similar north-south profile (Mancilla et al., 2015). The PRF profiles image Moho offsets of roughly 6–9 km under the Strait of Gibraltar, in agreement with the expected thickness of oceanic crust (Mancilla et al., 2015). The apparent overlap of the Iberian continental and oceanic LABs in profile N-II (Fig. 2) is likely an artifact due to the projection of independent structures within 1° at both sides of the profile. To the south, the transition to the African foreland LAB is imaged less abruptly at ∼35°N.

The observed LAB depth of the oceanic corridor is unusually large in a global comparison (e.g., Rychert and Harmon, 2018). The thickness of oceanic lithosphere depends on age and is currently explained by two competing models: a half-space cooling model (HSC; Stein and Stein, 1992) and a plate model (PM; Grose and Afonso, 2013). They differ for ages >70 Ma. The oceanic lithosphere in the Gulf of Cádiz is from the Middle to Late Jurassic (>145 Ma, e.g., Stampfli and Borel, 2002), among the oldest not yet consumed by subduction. For these ages, PM predicts a lithospheric thickness of slightly less than 100 km, and HSC predicts values of ∼110–130 km (Rychert and Harmon, 2018). The observed LAB of the oceanic corridor supports the HSC model. The remaining difference can be attributed to the presence of ∼30 km of overriding Alboran domain crust on top of the oceanic lithosphere at the Strait of Gibraltar.

The analysis of a large number of SRFs (∼11,000) has revealed the topography of the seismic LAB discontinuity beneath the complex plate-boundary zone between Iberia and northern Africa. Four different domains can be clearly distinguished, some of them limited by sharp vertical offsets. We confirm previous LAB depth estimates for the Iberian and Moroccan forelands. Under the Alboran domain, we find a thin lithosphere of 50–60 km thickness. The largest LAB depths are observed under the Strait of Gibraltar, outlining an ∼200-km-wide corridor of Middle to Late Jurassic lithosphere with ∼120 km thickness, being one of the thickest oceanic lithospheres ever detected. The main conclusion that emerged from the CCP images is the clear spatial connection between the Alboran slab under the Alboran Sea and the Atlantic oceanic crust through this oceanic corridor below the Gulf of Cádiz.

We are grateful to the staff involved in the Topo-Iberia, Picasso, Indalo, and Hire projects and the Instituto Andaluz de Geofísica and Instituto Geográfico Nacional permanent networks. A. Molina-Aguilera acknowledges Felix M. Schneider for helping him at the beginning with SRF methodology. We thank W. Spakman and the editor D. Brown for their comments. This work was supported by the Spanish national projects CGL2015-67130-C2-2-R/FEDER and CGL2012-31472. We acknowledge work on free software SAC and GMT.

1GSA Data Repository item 2019084, raw S-wave receiver functions, is available online at http://www.geosociety.org/datarepository/2019/, or on request from editing@geosociety.org.
Gold Open Access: This paper is published under the terms of the CC-BY license.