Tracking lithospheric delamination and surface processes across the Messinian salinity crisis

Abstract

Lithospheric delamination involves short-lived crustal and surface responses, alkaline magmatism, high heat flow and extension. In the Western Mediterranean, delamination is hypothesized to have triggered uplift at the origin of the Messinian Salinity Crisis (MSC). But delamination as the primary cause of uplift is questioned due to the insufficient temporal resolution. We report new U-Pb ages and clumped isotope analyses from calcite veins formed in an eastern Betic intramontane basin. They reveal a brief fluid event from 8.5 to 5 Ma linked to extension and retreating delamination. After extension, shortening and uplift began at 4.5–3 Ma across the boundary between the Cabo de Gata arc basement and the Iberia margin. We show that the MSC occurred before shortening and during delamination. Slab detachment caused the demise of the MSC, the formation of a new plate boundary fault and tectonic escape between Africa and Iberia around 5 Ma.

Introduction

Delamination is a process by which the lithospheric mantle or/and lower crust are detached and lost in the asthenosphere^1,2. It provides an efficient mechanism to recycle crust and mantle into the convective mantle^3,4, and it also explains the differentiation of the continental crust in an arc setting^5,6 and in early Earth without involving two-plate subduction processes^7,8. In late-/post-collision settings, continental delamination reflects the sinking of a negatively buoyant thickened or intrinsically dense lithosphere^9,10,11 or accompanies slab rollback by decoupling the mantle lithosphere from the overlying orogen as illustrated in the Western Mediterranean, such as the Apennines^12,13 and the Betic-Rif arc^14,15,16,17. Eventually, the delaminated slab detaches (lithospheric tearing or break-off) leading to the development of new plate boundary faults, tectonic escape, and uplift, e.g., Anatolian Plateau uplift and the formation of North and East Anatolian Faults^18,19,20,21. As delamination involves the replacement of the lithosphere by the asthenosphere, it leads to complex and short-lived crustal and surface responses, including uplift, alkaline magmatism, high heat flow, and extension. The rapid succession in time of these processes is a major difficulty in distinguishing between the causes and effects of delamination.

In the Betic-Rif arc (Fig. 1), it has been suggested that delamination, witnessed in the eastern Betics through post-collisional mantle-derived magmas at 6.3 Ma, caused the uplift of the Alboran continental margin¹⁵. This triggered the Messinian Salinity Crisis (MSC, 5.97–5.33 Ma) by closing the marine gateways between the Atlantic and the Mediterranean in the Betic Corridor (Spain) and the Rifian Corridor (Morocco)^15,22. Delamination is, however, not the unique mechanism invoked for pre-MSC uplift, as it is synchronous with the onset of shortening presented as the major driver of pre-MSC late Tortonian uplift of the Betics (<7.8–7.3 Ma) and in the Rifian Corridor (7.0–7.2 Ma)^23,24. Alternatively, both mechanisms can be at play. For instance, numerical models have suggested that the westward propagation of slab tearing, and detachment following delamination, stalled by around 8 Ma, resulting in resisting forces in the mantle that can explain uplift and the current tectonic activity²⁵, notably the formation of a new convergent plate boundary between Iberia and Africa²⁶. Other numerical scenarios have linked the pre-MSC uplift to lithospheric rebound induced by lithospheric tearing modulated by the obliquity of the convergence relative to the Iberian paleomargin²⁷. Additional processes like the emplacement of a thick buoyant arc in the Eastern Alboran Basin (EAB) contributed to the emergence of the Alboran domain before the MSC from 10 to 6.2 Ma, and a land bridge might have persisted between Africa and Iberia during and after the MSC from 7 to 3 Ma²⁸.

Fig. 1: Main tectonic features, deep structure of the eastern Betics, and stratigraphy of the Tabernas basin.

a Topography with existing low-temperature thermochronology dates inferred from Apatite fission-track (AFT) and (U-Th-Sm)/He (AHe) analyses are shown in addition to crustal thickness (labeled white thin lines), location of slab tear (thick back dashed lines) main faults and Neogene intramontane basins. b GNSS-based velocities (red vectors) in southern Spain with the location of External Zone and Internal Zone (brown filled polygons) of the Betic Cordillera after ref. ³⁰. c Lithosphere-scale cross-section of the eastern Betics (see location as red dashed line in a). d Stratigraphy of the Tabernas Basin with rock samples from the Paleozoic basement of Nevado-Filábride complex (NF), the Triassic limestones of the Alpujárride complex (AJ) to the Serravallian-Tortonian siliciclastic deposits. TASZ Trans-Alboran Shear Zone.

The exact relationships between uplift, onset of indentation, and mantle-derived process have not yet reached sufficient temporal resolution to allow a clear distinction between delamination, detachment, and shortening as the main drivers of the uplift and exhumation that led to the MSC. Here, we investigate the case of the eastern Betics where the continental basement of Alboran, now accreted to the southern Iberia paleomargin and intruded by mantle-derived magmas, is indented by the magmatic arc of the EAB, represented by the Cabo de Gata Volcanic Province (CGVP), and the rotated and sheared units of the Alboran basement in the Águilas Arc (Fig. 1a, b). Geodetic data shows that indentation is accommodated by the extrusion of the Alboran block bounded to the north by a broad zone of East-West right-lateral transtensional shear in the Betics, and by the left-lateral movement of the Carboneras Fault (CF) and Palomares Fault (PF) to the south-east, which continues as transpressional left-lateral shear in the Rif^29,30. These fault segments are part of the Trans-Alboran Shear Zone (TASZ), a 400-km-long left-lateral fault network crossing the whole Alboran basin (Fig. 1b). The current tectonic context is the result of a complex geodynamic evolution of the Betic-Rif arc that began with subduction in the Eocene. Details of plate kinematics are debated but a consensus exists on the fact that from around 20 Ma, the Betic orogenic belt collapsed leading to the opening of the Alboran basin controlled by the westward retreat of the Gibraltar slab and delamination³¹. The different motion between the retreating slab domain (Alboran) and the upper plate (Iberia) caused highly oblique crustal thinning in the continental basement of the Alboran Basin throughout the Langhian to Tortonian (16-8 Ma), which led to the exhumation of E-W metamorphic domes and subsidence in intermontane basins of the Internal Betics³². The Gibraltar slab is now imaged resting at 600 km below the Alboran Sea²², and its northern edge, forming part of the Iberia continental lithosphere, is detached over 400 km (Fig. 1b) below the Betics from the Lorca basin to the region of Malaga³³.

As a result of extension, the internal intramontane basins (Fortuna-Lorca, Guadix-Baza, Huercal-Overa, and Almanzora basins) and external basins (Tabernas, Sorbas, Vera, Alicante-Cartagena, Nijar-Carboneras basins) are distinguished according to their position on the margin, proximal and distal, respectively (Fig. 1a–c). Within the more distal margin, the external Tabernas basin represents an E-W-trending basin filled by Serravallian-Tortonian continental to late Tortonian deep marine deposits³⁴ resting at ~400 m above sea level between the metamorphic domes of Sierra de Alhamilla and Sierra de los Filábres (Fig. 1a–d).

Key unknowns include the duration of extension in the Tabernas basin and its relationship to transitions with uplift, mantle-derived magmatism, slab detachment, and initiation of contraction in the eastern Betics, specifically the onset of the Carboneras plate boundary fault. To resolve these issues, we analyze new U-Pb dates tied to clumped-isotope temperatures of syn-extension calcite-filled fractures from the Tabernas basin, with a view to constraining the transition in intramontane basins from extension to compression and uplift, along with lithosphere thermal conditions. Results are compared to key transitions resolved by the analysis of the cooling evolution of the CGVP and the continental basement using apatite (U-Th-Sm)/He (AHe) thermochronology. We examine the relationships between the lithosphere structure, the thermal state in the mantle and analyze the isostatic/dynamic support of the topography. The comparison with existing age constraints on exhumation and short-lived phases of magmatism enables us to more accurately identify the mechanisms that were active before and during the Messinian Salinity Crisis, likely contributing to the onset of Mediterranean isolation.

Results

Carbonate U-Pb dating of syn-extension fractures

Carbonate samples in syn-extension fractures (veins, faults) (Supplementary Data 1) were selected along a section comprising the metamorphic basement of the Tabernas basin, exposed on the flank of the Sierra de los Filábres, and Tortonian-Messinian sediments, in the center of the basin, for in-situ Laser Ablation Inductively Coupled Plasma Mass Spectrometer (LA-ICPMS) U-Pb geochronology. Basement samples comprise high-grade schists of the Nevado-Filábride (NF) complex (TA38, TA44) and Triassic marbles of the Alpujárride complex (TA41) that yield carbonate U-Pb ages of 7.22 ± 0.79 Ma (TA38, n = 51, 2σ), 5.05 ± 0.27 Ma (TA44, n = 112, 2σ) and 8.54 ± 0.21 Ma (TA41, n = 25, 2σ) (Figs. 1d and 2). Calcite from veins collected in lower Tortonian continental deposits (TA45, TA24) yield U/Pb ages of 5.37 ± 0.25 Ma (TA45, n = 124, 2σ) and 7.23 ± 0.60 Ma (TA24, n = 84, 2σ). Note that TA24 was sampled in lower Tortonian conglomerates exposed in the Tabernas topographic ridge, a structure interpreted as a strike-slip fault (Fig. 1d). The carbonate U-Pb date in the upper Tortonian marine turbidites (TA36) is 5.47 ± 0.23 Ma (n = 75, 2σ).

Fig. 2: Tera-Wasserburg diagrams of a selection of samples.

a Sample TA38 from the Nevado-Filábride complex, b sample TA41 from the Alpujárride complex and c sample TA36 from upper Tortonian deposits of the Tabernas basin (all results are presented in Supplementary Data 1). d Cross-section shows the spatial distribution of samples across the Tabernas basin (layers have the same color code as Fig. 1d).

Clumped-isotope temperatures of fluids

Among these U-Pb dated samples, five (TA24, TA36, TA38, TA41, TA45) were jointly analyzed for clumped isotope temperatures (T₄₇) and oxygen isotopic values of the mineralizing fluids (Fig. 3) (see also Supplementary Data 2). Samples from the Nevado-Filábride (TA38) and Alpujárride complexes (TA41) show similar precipitation temperatures T₄₇ of 66.9 ± 4.4 °C (n = 3) and 63.7 ± 4.2 °C (n = 3) (°C ± 95%), corresponding to equally close values of δ¹⁸O_water of 0.7 ± 0.7‰ VSMOW and −0.9 ± 0.7‰ VSMOW, respectively. Lower Tortonian continental deposits (TA45, TA24) yield higher T₄₇ of 76.2 ± 4.8 °C and 84.5 ± 5.0 °C, and the fluid composition also show contrasted δ¹⁸O_water of −4.6 ± 0.7‰ VSMOW (TA45) and 3.1 ± 0.7‰ VSMOW (TA24). Upper Tortonian marine sediments (TA36) indicate a precipitation temperature of 76.7 ± 4.8 °C and isotopic composition of fluids δ¹⁸O_water of 0.3 ± 0.7‰ VSMOW. The TA31 sample for which we have no U-Pb date was collected in the upper Tortonian channelized turbidites, south of the Tabernas topographic ridge, in a stratigraphic level only slightly younger than TA36. It yields a T₄₇ of 56.4 ± 3.9 °C and δ¹⁸O_water of −5.2 ± 0.7‰ VSMOW.

Fig. 3: Measured oxygen isotopic composition δ¹⁸O_calcite of the U-Pb-dated vein cement (‰ VPDB) against precipitation temperature (◦C).

a Precipitation temperatures obtained from clumped-isotope analyses are depicted along with uncertainties reported as ±2σ. Oblique lines are the calculated oxygen isotopic values of the mineralizing fluids δ¹⁸O_water (see “Methods” section). b Photographs show a selection of analyzed hand samples. Numbers in parentheses are U-Pb ages and T₄₇. The composition of seawater and meteoric water composition in Tortonian and Messinian are from refs. ^87,88,89,90.

Apatite (U-Th)/He dating of volcanic and metamorphic rocks

Five AHe ages from apatites with aliquots that replicate well were obtained in CGVP rhyolite (MOJ1.3, n = 4), dacite (MOJ.1.4, n = 4) and andesite (MOJ1.1, n = 4; MOJ1.2, n = 5; MOJ1.5, n = 3) (Supplementary Data 3). Results show single-grain alpha-ejection corrected AHe ages between 10.8 ± 0.9 Ma and 3.1 ± 0.2 Ma. Mean AHe ages range in andesites from 4 ± 0.2 Ma (2σ) for MOJ1.1, 9.1 ± 0.3 Ma for MOJ1.2 and 6.1 ± 0.3 Ma for MOJ1.5, and consistent AHe ages of 7.0 ± 0.3 Ma for dacite MOJ1.4 (n = 4) and 5.9 ± 0.2 Ma for rhyolite MOJ1.3 (n = 4). West of the Carboneras fault, only one sample out of five originally collected in the NF metamorphic basement yielded enough apatite grains suitable for AHe analyses. MOJ2.1 (n = 3) sampled at 520 m in NF micaschists east of the Sierra de los Filábres yielded the youngest mean AHe age of 2.6 ± 0.1 Ma. The scarcity of apatite in NF micaschists and their low Uranium content (0.5-1.4 ppm for MOJ2.1) has already early been noticed³⁵. We note that in the region of the NF micashists (Sierra Alhamilla) where these previous apatite-fission track analyses have been successful, apatite yielded slightly older ages of 5.7 ± 1.1 Ma and 6.1 ± 1.2 Ma. An AHe age of 15.2 ± 0.8 Ma was obtained close to MOJ2.1 (elevation 2000 m) but it is more problematic considering it is similar to its apatite-fission track age of 15.6 ± 1.9 Ma³⁶. The eU-age and (U-Th)/age relationships of our AHe-dated apatite samples show a positive correlation that agrees with the expected increase of closure temperature as a function of radiation damage³⁷.

Time-temperature histories

Reconstruction of time-temperature histories of CGVP have been reconstructed based on simple thermal inputs assuming that published ⁴⁰Ar/³⁹Ar ages^16,38 faithfully represent the timing of magma cooling below 350 ± 50 °C (Tc 350 °C on biotite), and that the timing of cooling at surface temperature (20 ± 20 °C) is indicated by the deposition of Tortonian-Messinian marine limestones on top of sampled CGVP lavas (Fig. 4). These deposits attest to emersion as early as 10–9 Ma (lower Tortonian) while other stratigraphic contact suggests later emersion until the Messinian^39,40. As the sampled dacite, andesite, and rhyolite from CGVP were emplaced at the surface, cooling is expected to be instantaneous on the geological timescales. However, the large majority of AHe ages (closure temperature of 70-50 °C) in CGVP are younger than the ⁴⁰Ar/³⁹Ar ages of their parental volcanic rocks (11 ± 1 Ma) by 7 to 2 myrs. The delay between magma crystallization of calc-alkaline suites and AHe cooling ages reveals a complex cooling history caused by subsequent punctual thermal events. This may represent the effect of pyroclastic eruptions or lava flow, doming or caldera collapse, circulation of hydrothermal fluids linked to Cu–Au mineralization well documented during the CGVP formation⁴¹ and burial heating by Tortonian-Messinian to Pliocene deposits. Modeling results show that all volcanic rocks cooled below 350 ± 50 °C at different rates between 12 and 8 Ma, which marks the end of the volcanic phase. After 8 Ma, samples variably recorded the deposition of the upper Tortonian-lower Messinian carbonate platform. Samples from the southern half of CGVP MOJ1.2 to MOJ1.5 show no significant heating or cooling, or it is within the uncertainties of surface temperature, hence indicating no measurable exhumation (Fig. 4). By contrast, in the north, where lower Messinian and lower Pliocene marine deposits are preserved, MOJ1.1 thermal evolution shows burial heating below sediments followed by cooling at ca. 4.5 Ma. For thermal history modeling of the MOJ2.1 micaschist sample, we have used published AHe and AFT data from a sample collected nearby. Model shows that MOJ2.1 resided at 120-80 °C (Tc AFT) and possibly 70-50 °C (Tc AHe) at 15 ± 1 Ma, and cooled at ca. 3 Ma (Fig. 4). We infer that both CGVP volcanic rocks (MOJ1.1) and metamorphic basement (MOJ2.1) north of CF cooled between 4.5 and 3 Ma, clearly after the MSC, at a rate between 36 °C/Myr (MOJ1.1) and 48 °C/Myr (MOJ2.1).

Fig. 4: (U-Th)/He ages and time-temperature histories.

a AHe ages (circles) obtained in the Cabo de Gata Volcanic Province and the Sierra de los Filábres (metamorphic Nevado-Filábride complex) (see Supplementary Data 3). Black vectors are GNSS-based velocities. Note that NW-directed velocities diminish towards the north (e.g., between CGVP and Sierra de los Filabres), indicating shortening, whereas velocities directed towards the west, decrease towards the south (e.g., between W-Sierra de los Filábres and Sierra de Gador), indicating extension and right-lateral shear. Triangles indicate the volcanic centers filled with the same color code as for the AHe ages. Contours of exposed upper Tortonian and Messinian-Pliocene limestones are also indicated. b–e QTQt-derived thermal histories were obtained (see Supplementary Data 3) for four samples representative of the thermal evolution of CGVP (MOJ1.1, MOJ1.2, and MOJ1.5) and NF complex (MOJ2.1). Black boxes represent thermal and age constraints. Observed and predicted uncorrected AHe ages are indicated below each model T-t path. See text for abbreviations.

Impact of mantle-derived processes on topography

The mean elevation of the Central-Eastern Betics (Fig. 5a) region is 868 m with the highest elevation of 3467 m in the Sierra Nevada. Receiver functions and reflection seismic data⁴² reveal a bimodal distribution of crustal thicknesses ranging from 35 km in the north to 25 km in the south and east (Figs. 1a and 5a), expressing crustal and lithosphere thinning south of the Sierra Nevada, Sierra de los Filábres and Sierra de las Estancias, linked to the oblique opening of the Alboran basin⁴³. Crustal thinning onshore is reflected in the topographic decrease down to 176 m south of Sierra de los Filábres and in Cabo de Gata (Fig. 5a). The crustal thickness offshore reduces to 15 km in the EAB (Fig. 1a). To compare current topography with predictions using isostatic calculations, the lithosphere thickness across the eastern Betics was inferred from various seismic tomography models SL2013sv⁴⁴, LITHO1.0⁴⁵, and Mean model⁴⁶ (Fig. 5a). From these models, upper and lower bounds for the lithosphere thickness fall between 65 km and 50 km. The value of 65 km fits both SL2013sv and Mean models. It is also consistent with maximum depths for the lithosphere of about 60 km (1.8 GPa) independently determined from thermobarometric calculations on Neogene mantle-derived K-rich basalts (MgO > 8 wt.%, SiO₂ > 45 wt.%)⁴⁷. This value is within the range of thicknesses <100 km for fertile and hot phanerozoic sub-continental lithosphere⁴⁸, a feature characteristic of the European lithosphere⁴⁹.

Fig. 5: Lithospheric and sub-lithospheric architecture, and topography of the eastern Betics.

a Topographic profile (see location in Fig. 1a), crustal thickness variations, and S-wave velocity structure from SL2013sv⁴⁴. Values of S-wave velocities are indicated by contour lines and intervals of 0.1 km.s⁻¹. Dashed lines refer to the Lithosphere-Asthenosphere Boundary (LAB) inferred from different global seismologic models. Predicted elevation for variable lithosphere thickness and lithospheric mantle densities are shown as curves labeled in km b for crustal thickness of 25 km (crustal thickness below Tabernas Basin, TB). c Elevation as a function of mantle densities for variable lithosphere thickness, thermally and non-thermally equilibrated.

Elevation is calculated for a column of continental lithosphere with respect to the depth of a standard mid-ocean ridge, assuming pressure-temperature dependent densities and chemical depletion for the lithospheric mantle⁵⁰ (Supplementary Discussion). Results show that for a lithosphere thickness of 65 km, the calculated elevation is negative (−250 m), assuming a crustal thickness of 25 km (Moho depth below Tabernas Basin), an average crustal density of 2.7 based on EPcrust model⁵¹, and a reference density for the lithospheric mantle of 3.3 (Fig. 5b); it is null for lithosphere thickness of 50 km and −500 m for a 70 km-thick lithosphere (Fig. 5c). The lithosphere below the eastern Betics is therefore not in isostatic equilibrium unless we assume a mantle density decrease of 80-100 kg/m⁻³. Studies of rifted margins show that the formation of light mineral phases such as plagioclase during extension and shallowing of the lithospheric mantle can produce a density decrease of mantle rocks down to 3.2 at 60 km for temperatures above 800 °C⁵². This is likely to have occurred in the Betics during the Mesozoic extension as the lithospheric mantle was refertilised by tholeiitic MORB-like melts⁵³. Alternatively, elevation can also be well reproduced by the upwelling of hot asthenosphere, likely with a fraction of melt as indicated by low S-wave velocity domain of 4.2 km.s⁻¹ around 100 km (Fig. 5a). Based on the parametrization of shear wave velocity-temperature⁵⁴ we determine anomalously high sub-lithospheric temperatures of 1550 °C. Such a low viscous channel could easily explain the observed mean topography of 500 m around Tabernas Basin (see “Methods” section and Supplementary Discussion). We infer that the compositional variations in the mantle lithosphere and asthenospheric upwelling likely acted together and contributed to the support of the eastern Betic topography.

Discussion

Pre-MSC uplift and subsidence driven by mantle delamination

The timing of uplift in the eastern part of the Betics has been provided in recent research by the age of the marine-continental transition in the intermontane basins. In the internal basins of Fortuna-Lorca, Guadix-Baza Huercal-Overa, and Almanzora, the transition has been dated to the end of the Tortonian (8–7.5 Ma), indicating uplift before the MSC^{23,55,56,57,58}. It has also been suggested that the onset of N-S/NW-SE shortening occurred around 8–7 Ma in the eastern Betics^34,59,60 and in the Alboran Sea⁶¹. This timing is in line with previous T-t paths inferred from apatite fission tracks (AFT) ages ranging from 2.5 Ma to 25 Ma (mode of 8 Ma; Fig. 1a)^{35,36,62,63,64,65} that support the interpretation that exhumation in the Internal Zone was progressive and associated with orogen-parallel stretching from the Sierra de los Filábres at 13–11 Ma (Serravallian-Tortonian) to the Sierra de Nevada at 8–6 Ma (Late Tortonian-Messinian). The late Tortonian uplift and exhumation documented in the internal basins therefore appear temporally closely related to a late Tortonian stage of extension. At lower elevation, in external basins that recorded the MSC, the marine-continental transition shows a younger uplift ranging from Pliocene, ca. 5.3 Ma in Sorbas Basin, to post-mid-late Pliocene (4-3.6 Ma) for distal Vera, Alicante-Cartagena, Nijar-Carboneras basins^55,58,66. The final uplift in the external basins occurred therefore after the Zanclean reconnection between the Atlantic and Mediterranean.

The U-Pb ages, isotopic signatures, and temperatures of carbonate fluids in veins from the Tabernas basin are representative of the thermal conditions and fluid pathways that prevailed during extension in external basins. The narrow U-Pb age range from 8.5 to 5 Ma argues for a short-lived fluid event, for which the cessation coincides with a change in the tectonic stress field. This scenario is further argued by the differential exhumation across the CF between the CGVP (MOJ1.1) that cooled from 4.5 Ma and the NF metamorphic complex (MOJ2.1) that cooled from 3 Ma. We interpret this difference as evidence for sequential deformation from south to north associated with the movement across the CF as the CGVP and Águilas Arc collided with the continental basement. The initiation of cooling across the CF is coincident with the termination of fluid circulation and uplift of the external basins. We infer that the end of fluid circulation marks the onset of shortening across the CF at 4.5 Ma that regionally coincides with the onset of collision between the continental basement and the magmatic arc crust of EAB. The Pliocene compression is synchronous with the post-MSC uplift in external basins and is distinct from the late Tortonian, pre-MSC uplift, documented in internal basins, which is associated with extension in the Betics and transpression in the Rif.

The different timing of uplift between internal (8-7.5 Ma) and external basins (post-5 Ma) is significant and has acted at different scales. For instance, the late Tortonian uplift is described as nearly synchronous across the whole of the Betics, even farther westward of the tip of the slab tear^23,57, and in the Rif. If this late Tortonian uplift in the internal basin is compared to the age and type of magmatism in eastern Betics, it appears to coincide with a transition from post-collisional calc-alkaline volcanism (basaltic andesite, andesite, dacite, rhyolite), older than 8 Ma in CGVP (mostly 11 ± 1 Ma) and Alboran Sea, to high-K calc-alkaline, shoshonitic and ultrapotassic volcanism (Si-K-rich lamproite) dated between 8 and 6 Ma, mean ages of 7.4 Ma^16,67,68 (Fig. 6). These mantle-derived lavas are found preferentially north of CGVP and south of the slab edge (Fig. 6) where asthenosphere upwelling is imaged by low S-wave velocity anomaly (Fig. 5).

Fig. 6: Geographic and temporal relationships between the different types of volcanism in eastern Betics.

a Distribution of type and age of magmatism in the eastern Betics. b Same as (a) but with data projected relative to latitude. The timing of Messinian Salinity (MSC), delamination and crustal shortening are indicated.

Control of mantle-driven processes on late Tortonian syn-extension fluids can be tested in the Tabernas basin based on clumped-isotope water composition and temperatures. Comparison of δ¹⁸O_water inferred from samples δ¹⁸O_calcite (Fig. 3) with the oxygen isotope composition of Tortonian and Messinian seawater and freshwater supports an interpretation that the fluids originated from two distinct sources. Values of δ¹⁸O_water between −0.9 ± 0.7‰ VSMOW and +3.1 ± 0.7‰ VSMOW (T24, T36, T38, T41) reveal they precipitated in equilibrium with marine water or involved a component of formational water (i.e., fluids trapped in the sediment during its deposition and lithification), indicating fluid-rock interactions. This contrasts with the negative δ¹⁸O_water (TA45, TA31) of about -5‰ VSMOW pointing to a contribution of meteoric water. Given the temporal overlap of U-Pb-dated fluids, we infer that the circulation of seawater and meteoric water alternated in time during the 3-Myr fluid-flow event. T₄₇ estimates indicate these fluids precipitated at around 75 ± 10 °C (2σ of 5 °C) or 72 °C ± 15 °C if the undated TA31 sample is included in the calculation. Assuming they are fluids emplaced before folding of host rocks, paleodepths of 1-1.5 km are inferred for fluids that precipitated during the Tortonian (TA24, TA38, TA41) and Pliocene (TA36, TA45) (Figs. 1d and 5). It has been argued that the northern flank of the Tabernas syncline hosting NF and AJ basement samples (TA38, TA41) was folded before the Serravallian-Lower Tortonian⁴³. When associated with the lower Tortonian sample from the Tabernas ridge (TA24) characterized by identical T₄₇ and U-Pb age, we infer that these Tortonian fluids reflect a basin-scale thermal event consistent with regional gradients above 75 °C/km. We conclude that these pre-5 Ma fluids precipitated in association with regional palaeogeothermal gradients of >50 °C/km, locally rising above 75 °C/km on the flanks of the basins. Comparison with continental geotherms accounting for radiogenic crustal heat production, and transient cooling of the lithosphere, indicate that these shallow crustal gradients are consistent with steady-state temperature gradients of 55 °C/km computed for a 65-km-thick lithosphere, crustal thickness of 25 km, and a thermal-tectonic age of the lithosphere of 20 Ma (Supplementary Discussion). We predict a surface heat flow of 118 mW/m², which agrees with the measured current heat flow of 124 mW/m² in the EAB⁶⁹. Note that this value requires a high mantle heat flux of 48 mW/m² which is 50% higher than the average European mantle heat flow⁴⁹. The δ¹⁸O_water of the veins argue that fluids precipitated in equilibrium with the expected seawater and meteoric sources under a high regional continental geotherm established during lithosphere thinning and mantle upwelling.

These results combined with previous data show that basin subsidence and lithospheric thinning has continued from 8.5 and 5 Ma in the external basins of the Betics (Fig. 7), in contrast to the internal basins that recorded synchronous uplift and the closure of the marine gateway of the Betic Corridor. The distributed right-lateral strike-slip shearing (Fig. 1b) that is linked to lithospheric tearing and delamination, is likely to be the trigger of the contrasting vertical movements between internal and external basins (Fig. 7). Evidence for delamination is brought by geochemical and petrological data on the mantle sources of alkaline lavas. They show that the transition from calk-alkaline to high-K calc-alkaline, shoshonitic, and ultrapotassic volcanism occurred because of variable degrees of melting of a metasomatised lithospheric mantle prior to Paleogene collision in the Betics⁷⁰. Delamination of the Iberian mantle, accompanied by upwelling of the asthenospheric mantle (Fig. 7), provides the most reliable explanation for the observed magmatic evolution. The Na-alkaline basalts are, however, linked to low-degree melting of the upwelling sub-lithospheric mantle⁷¹. Because of the differential motion between the retreating mantle delamination zone (Alboran) towards the west and the overriding plate (Iberia), propagating strike-slip faults resulted in lithospheric tearing in the sense of Govers and Wortel⁷². In the exposed crust, the tectonic interpretation of E-W dextral transfer fault zones, vertical rotations across the Internal and External Betics⁷³, and orogen-parallel extension supports this scenario. Recent research has established that tearing is distributed rather than localized, and that extension is oblique to the Betic trend, hence resulting in the formation of a highly oblique rifted margin⁴³.

Fig. 7: Deep lithospheric structure and tectono-magmatic evolution of the eastern Betics before, during, and after the MSC.

a Present-day tectono-magmatic model of the eastern Betics along profile shown in Figs. 1 and 5. Note that the mantle slab is still attached below the western Betics (see Fig. 1b). b Tectonic scenario of the evolution of the eastern Betics from extension to onset of shortening and presenting key elements of the evolution of the mantle from retreating delamination to slab detachment before and surface processes before and after the MSC.

Delamination and the origin and demise of the Messinian Salinity Crisis

We propose that the association of oblique extension in the distributed right-lateral strike-slip shear zone and delamination retreat, explains the contrast between the uplifted internal basins that developed on the thick proximal margin, and the subsidence of external basins that underwent larger rift-related extension (Fig. 7). The extension triggered by westward retreat of the slab that we constrain between 8 and 5 Ma in the eastern Betics is synchronous with the pre-MSC closure of marine corridors in Spain and in Morocco. In Morocco, uplift is attributed to transpression, which likely reflects the differential movement of the sinking slab relative to the motion of Africa²⁵. Importantly, the mid-late Miocene stage of formation of the strike-slip transfer zone in the Betics is temporally distinct from the onset of slab tearing, which is a localized lithospheric feature that does not extend over the whole Betics, as imaged by seismic data (Fig. 1b). According to our interpretation, this tectonic configuration lasted until 5 Ma when the delaminated slab detached and shortening initiated in the Betics (Fig. 7). Extrusion of the Alboran block and onset of the CF developed at this time. It has been suggested that the transition from oceanic subduction to continental delamination was responsible for the change between the pre-MSC high-K calc-alkaline, shoshonitic, and ultrapotassic volcanism and the emplacement of the most primitive lavas, comprising the isolated emission of post-MSC Na-alkaline basalts (hawaiite) (2.9-2.3 Ma)⁷¹. In contrast, we argue that the 3 Myrs gap between pre-MSC and post-MSC magmatism observed in the eastern Betics (Fig. 6) is the consequence of shortening. The same timing of pre-MSC high-K calc-alkaline volcanism at ~6.6 Ma relative to post-MSC Si-poor volcanism at 4 Ma in Morrocco^16,67,68 indicates that slab detachment and mantle upwelling, which are coincident with the shift to shortening, took place regionally.

The closure of the marine gateways in Spain and Morocco between 7.8 and 7.0 Ma occurred during the latest stages of continental delamination retreat initiated as early as 11 ± 1 Ma (onset of EAB magmatic arc) that was accompanied by extension in the Betics (Fig. 7) and shortening in the Rif. The remaining marine gateway supplying salt to the Mediterranean at the origin of the MSC is the Gibraltar gateway. We link the MSC to the progressive uplift of this marine corridor caused by a combination of lithospheric delamination and magmatism. This model agrees with the occurrence of an Alboran volcanic archipelago that permitted faunal exchanges before and after the MSC²⁸. While the role of geodynamic processes in the Gibraltar region is central, the onset of the MSC further indicates the additional impact of global sea-level lowering and paleoclimate⁷⁴. Following the rapid migration of the slab detachment in Spain and Morocco towards Gibraltar, slab pull forces increased in this region causing the subsidence and widening of the Gibraltar gateway. The slab detachment event is therefore considered the cause of the demise of the MSC rather than its trigger. This post-5 Ma event was coincident with the formation of a new plate boundary fault between Africa and Iberia has so far produced limited regional effects on topography compared to deep mantle processes.

Methods

Samples

Calcite samples were collected in veins predominantly striking NW-SE and WNW-ESE opened perpendicular to the regional extension-oriented NE-SW determined from paleostress data⁴³ and in calcite steps on faults consistent with the same regional extension. Calcite veins are present in all the stratigraphic sections both in the basement and the overlying sedimentary series (Fig. 1d). Most veins are predominantly vertical (TA31, TA36, TA45, TA38) and locally at moderate angle (TA44) with respect to bedding, and their widths range from millimeters to centimeters (Figs. 2 and 3). The calcite crystals form generally elongated blocky syntaxial to fibrous textures (Supplementary Data 1). Fourteen slabs were prepared and polished thin sections were examined by conventional and cold cathodoluminescence microscopes to detect the different generations of crystallization/recrystallization and select the in-situ rock sample for further analyses. In most veins, one continuous single calcite phase precipitation (TA31, TA24, TA45, TA38, TA44) is observed indicating coeval opening and mineralization (Supplementary Data 1). Samples TA36 and TA41 have two generations of calcite (consistent with U-Pb analyses showing secondary late U-Pb ages, Supplementary Data 1).

Sampling for (U-Th)/He analyses has focused on the most felsic lavas in CGMP (andesite, dacite, rhyolite) (MOJ1.1, MOJ1.2, MOJ1.3, MOJ1.4, MOJ1.5) and in NF metamorphic complex (MOJ2.1) for comparing cooling histories across the Carboneras fault zone (see sample information and analyses in Supplementary Data 3). An attempt was made to select CGVP samples near marine carbonates of Tortonian-Messinian and lower Pliocene ages that were deposited after magma emplacement to place thermal constraints for modeling the post-Tortonian uplift and exhumation.

Calcite LA-ICP-MS U-Pb geochronology

Calcite U-Pb analyses were performed at the Geochronology & Tracers Facility, British Geological Survey (Nottingham, UK) using a Nu Instruments Attom single collector inductively coupled plasma mass spectrometer coupled to a NWR193UC laser ablation system fitted with a TV2 cell. Laser parameters were 100 µm spots, ablated at 10 Hz for 30 s with a fluence of ~4 J cm² (see Supplementary Data 1 for full instrumental conditions). Normalization used WC1⁷⁵ as a primary reference material for Pb/U ratios and NIST614 for Pb/Pb ratios⁷⁶; ASH15D⁷⁷ was run as a secondary reference material and the pooled result across the analytical sessions is within 2% accuracy of the accepted age. Data reduction uses the Time-Resolved-Analysis function in the Attolab software, followed by normalization offline in Excel. Age calculations and plotting use IsoplotR⁷⁸. All ages are derived from lower intercept ages on Tera-Wasserburg plots (see Fig. 2 and Supplementary Data 1), quoted at 2σ and shown as ±α/β, where α and β are without and with systematic uncertainties, respectively. Several samples were analyzed but were not used in our study due to low U/Pb ratios hindering a robust age calculation, or to highly scattered analyses indicating age mixing and/or U mobility. All data are presented in Supplementary Data 1.

Clumped isotope analysis

All analyses were performed at the Laboratoire des Sciences du Climat et de l’Environnement (LSCE, Gif sur Yvette). Six samples (TA24, TA31, TA36, TA38, TA41, TA45) of homogenized carbonate powder were converted to CO2 by phosphoric acid reaction at 90 °C in a common, stirred acid bath for 15 min. The initial phosphoric acid concentration was 103 % (1.91 g/cm³) and each batch of acid was used for 7 days. After cryogenic removal of water, the evolved CO₂ was helium-flushed at 25 mL/mn through a purification column packed with Porapak Q (50/80 mesh, 1 m length, 2.1 mm ID) and held at −20 °C, then quantitatively recollected by cryogenic trapping and transferred into an Isoprime 100 dual-inlet mass spectrometer equipped with six Faraday collectors (m/z 44–49). Each analysis took about 2.5 h, during which analyte gas and working reference gas were allowed to flow from matching, 10 mL reservoirs into the source through deactivated fused silica capillaries (65 cm length, 110 μm ID). Every 20 minutes, gas pressures were adjusted to achieve m/z = 44 current of 80 nA, with differences between analyte gas and working gas generally below 0.1 nA. Pressure-dependent background current corrections were measured 12 times for each analysis. All background measurements from a given session within ±6 hours of any given analysis were used to determine a mass-specific relationship for that analysis, linking background intensity (Zm), total m/z = 44 intensity (I44), and time (t): Zm = aI44 + P(t), with P being a polynomial of degree 2 to 4. Background-corrected ion current ratios (δ45 to δ49) were converted to δ¹³C, δ¹⁸O, and “raw” Δ₄₇ values as described by Daëron et al.⁷⁹, using the IUPAC oxygen-17 correction parameters. The isotopic composition (δ13C, δ18O) of our working reference gas was computed based on the nominal isotopic composition of all ETH carbonate standards⁸⁰ and an oxygen-18 acid fractionation factor of 1.00813. Raw Δ₄₇ values were then converted to the I-CDES reference frame⁸¹ using a pooled regression approach as implemented by the Δ47crunch Python library. Full analytical errors are derived from the external reproducibility of unknowns and standards (Nf = 39) and conservatively account for the uncertainties in raw Δ₄₇ measurements as well as those associated with the conversion to the I-CDES reference frame. T₄₇ reconstructions are based on the OGLS23 calibration (see Supplementary Data 2). Temperature confidence intervals account for analytical error in Δ₄₇ but not for calibration uncertainties, which remain below ±1 °C (95% CL) in the range 0–50 °C. The δ¹⁸O values of the mineralizing fluids (δ¹⁸O_water) are derived from T₄₇ and the δ¹⁸O_CaCO3 of the mineral based on the fractionation equation.

Apatite (U-Th)/He dating

Separation and selection of apatites were carried out at the GET laboratory. Each sample was crushed to obtain grains smaller than 500 µm. Two successive steps of dense liquor separation followed to recover grains with a density between 2.97 and 3.38 (apatite density 3.2). This fraction was then magnetically sorted by Frantz. The remaining fraction, non-magnetic, was finally sorted by optical microscope (picking). Apatite crystals for (U-Th-Sm)/He dating were handpicked to select inclusion-free crystals with a minimum of three aliquots. Crystal dimension and geometry were measured along two axes, considering the apatite crystals as regular hexagonal prisms, grains had in general a minimum width of 60 μm to minimize the effect of alpha particle ejection. After this first step, (U-Th)/He analyses on apatite were performed at the Noble Gas Laboratory of the Geosciences Montpellier laboratory. Each grain, packed in Pt tubes, was placed under vacuum and heated for 5 minutes with a 1090 nm fiber laser operating at 4.0 W (900 °C). After spiking (addition of a known amount to be able to quantify the real content in the minerals) with 3He, gas purification was performed by a cryogenic trap and two AP-10-N SAES getters, and the helium content was measured on a PrismaPlus QMG 220 quadrupole. The 4He content was determined by the peak height method which is 10 to 10,000 times higher than typical blank levels. A second heating cycle using the same analytical procedure was routinely performed to verify that more than 99% of the 4He was extracted in the first cycle. After helium extraction, the Pt tubes were retrieved from the sample chamber and transferred to a 2 ml polypropylene conical tube. The samples were triple spiked (149Sm, 230^Th, and 233U) and dissolved following the procedures described by Wu et al.⁸². The resulting solutions were diluted and U (233U and 238U), Th (230Th and 232Th), and Sm (149Sm and 147 Sm) were measured using ICPMS isotope dilution. For the age calculation, it is necessary to perform the alpha (4He) particle ejection correction. This correction was calculated using Ft software considering the size, and shape of each apatite^83,84. Analyses are presented in Supplementary Data 3.

QTQt thermal modelling

Thermal histories were retrieved from AHe data analyses using thermal inversion QTQt modelling software⁸⁵. For AHe, QTQt incorporates kinetic models for helium diffusion in apatite crystals that account for radiation damage production and annealing (see Supplementary Data 3 for reference). These two diffusion models may produce different results, depending on the integrated thermal history of each grain. The inversion procedure uses a Bayesian “Markov chain Monte Carlo” algorithm to determine the best time-temperature path reproducing our data through a large number (between 100,000 and 200,000 on average) of randomly tested solutions. In our case, we have run nearly 170,000 models. The final model is represented by a set of time-temperature histories for which we included the expected model (weighted mean model) and the maximum likelihood model (best data fitting model). A table summarizing the thermal history model input parameters and all thermal models is provided in Supplementary Data 3.

Isostatic calculation, Vs-temperature conversion, and dynamic topography

The topography of continents reflects the combined effects of (1) isostasy that results from the variation of age-dependent lithosphere thickness and densities and (2) a non-isostatic (dynamic) component caused by plate strength, subduction dynamics, or more broadly mantle flow driven by anomalously hot asthenosphere. The elevation, e, of a buoyant lithospheric column in Airy isostatic equilibrium, with respect to the average mid-ocean ridge depth, is given by

$$e={h}_{c}\left(\frac{{\rho }_{a}-{\rho }_{c}}{{\rho }_{a}}\right)-{h}_{{oc}}\left(\frac{{\rho }_{a}-{\rho }_{{oc}}}{{\rho }_{a}}\right)-{h}_{w}\left(\frac{{\rho }_{a}-{\rho }_{w}}{{\rho }_{a}}\right)+\left({h}_{l}-{h}_{c}\right)\left(\frac{{\rho }_{a}-{\rho }_{m}}{{\rho }_{a}}\right)$$

(1)

where ρ_c, ρ_oc, ρ_w, ρ_m, and h_c, h_oc, h_w, h_l are the densities and thickness of the crust, oceanic crust, water column above the mid-ocean ridge, and continental lithosphere, respectively. The calculation assumes the density of the asthenosphere ρ_a accounts for an adiabatic temperature gradient and for the effect of compressibility. ρ_m varies with depth according to a linear geotherm and accounts for the effect of depletion and compressibility. Elevation is calculated for the case of thermal equilibration and the case of the instantaneous removal of the lithospheric material by delamination, assuming no thermal equilibration at the base of the original lithosphere⁵⁰. An original lithosphere thickness of 100 km is consistent with the average thickness of Phanerozoic lithosphere. Equations and parameters are presented in Supplementary Discussion.

S-wave velocities from model SL2013sv are then used to calculate temperatures in the mantle using the parametrization of ref. ⁵⁴. V_s (P, Θ, a) is dependent on pressure P (depth, z), temperature Θ in °C and a which is a variable describing the activation process. V_s is given by

$${{V}_{s}}=\left(\right.1-{b}_{v}\left(z-50\right)\left(m \Theta+{\rm{c}}+{\rm{A}} \exp \left(\frac{-E-P{{V}_{a}}}{{RT}}\right)\right)$$

(2)

where empirical constants are b_v = 3.84 × 10⁻⁴ km⁻¹, m = −0.28 m.s⁻¹ $\oint$C⁻¹, c = 4720 m.s⁻¹, A = −1.8 × 10¹⁶ m.s⁻¹, E = 409 × 10³ J.mol⁻¹, V_a = 10 × 10⁻⁶ m³.mol⁻¹ and R the gas constant is 8.3145 (see Supplementary Discussion). The highest temperature obtained is 1550 °C at 100–125 km below the Tabernas basin and to the south where V_s decrease to 4.2 km.s⁻¹.

Dynamic uplift U created by the temperature anomaly is then calculated assuming thermal expansivity from background temperature and Airy isostasy⁸⁶ using the following equation:

$${\rm{U}}=\frac{H\alpha \bar{T}}{1-\alpha {T}_{0}}$$

(3)

where H is the layer thickness, α is the thermal expansivity, ${T}_{0}$ is the background temperature, and $\bar{T}$ is the average temperature in the low viscous asthenospheric channel produced by the excess in temperature (parameters are provided in Supplementary Discussion). For an average thermal anomaly of 100 and 200 km the predicted uplift ranges between 500 m and 1 km. Considering the elevation of −250 m predicted by isostasy assuming pressure-temperature dependent densities and chemical depletion for the lithospheric mantle, the range of dynamic uplift can produce the elevation of 500 m observed around Tabernas Basin.

Temperature of continental lithosphere and heat flux estimates

The continental geotherms T_cont is given by

$${T}_{{cont}}\left(z,{age},\,{H}_{s}\right)={T}_{{std}}\left(z,\,{H}_{s}\right)+T\left({age}\right)$$

(4)

It is calculated using the stationary part of the geotherm T_std and contribution due to the radiogenic heat production H_s in the crust, and correction T(age) due to transient cooling of the lithosphere that depends on the age of the last major plate-scale thermal event.

Temperature T_m at Moho depth, h_c, is used for calculation of temperature at depths below the Moho and is given by:

$${T}_{m}={T}_{0}+\frac{{q}_{m}}{{k}_{c}}{h}_{c}{+T}_{r}$$

(5)

where T₀ and q_m correspond to the temperature at the surface and the mantle heat flux calculated at the Moho depth. T_r is the radiogenic contribution that is dependent on the thickness of the crust h_c, density ρ_c, radiogenic production H_s, radiogenic production decay depth h_r, and thermal conductivity coefficient k_c according to the following equation:

$${T}_{r}=\frac{{\rho }_{c}{H}_{s}{h}_{r}^{2}}{{k}_{c}}\left(1-{e}^{\frac{{h}_{c}}{{h}_{r}}}\right)$$

(6)

Details on the calculation are given in Supplementary Discussion.