The architecture and tectono-magmatic evolution of the lithosphere of Europe are the result of a succession of subduction, rifting and inputs from plumes that have modified the lithospheric mantle since the Neoproterozoic (750–500 Ma). These events gave birth to contrasting crust-mantle and lithosphere-asthenosphere mechanical coupling between strong, viscous, thick, cold, depleted mantle of the Archean lithosphere of the West African Craton and the East European Craton, and the weak, low viscous, thin, hot and less depleted mantle of the Phanerozoic lithosphere of Central Europe. These differences were long-lived and explain the first-order present-day stresses and topography as well as the styles of orogenic deformation. The lack of thermal relaxation needed to maintain rheological contrasts over several hundreds of millions of years requires high mantle heat flux below Central Europe since at least the last 300 Ma. A combination of edge-driven convection on craton margins and asthenospheric flow triggered by rift propagation during the Atlantic and Tethys rifting is suggested to be the main source of heat. The topography of Central Europe remained in part dynamically supported during most of the Mesozoic thinning in line with the long-term stability of thermal-mechanical structure of the lithosphere. Timing and rates of exhumation recorded across Western Europe during convergence indicate that an additional control by the architecture of Mesozoic rifted margins is required. By 50 Ma the acceleration of orogenic exhumation, from the High Atlas to the Pyrenees, occurred synchronously with the onset of extension and magmatism in the West European Rift. Extension marks the onset of distinct orogenic evolution between Western Europe (Iberia) and the Alps (Adria) in the east, heralding the opening of the Western Mediterranean. A major kinematic re-organisation occurred triggering the involvement of more buoyant and thicker portions of rifted margins resulting in widespread orogenic growth. We conclude that conceptual models of collision require to better account for the thermo-magmatic evolution of the continental lithosphere, especially the original architecture and composition of its mantle, as well as the precise knowledge of the architecture of the rifted margins to explain the timing and rates of orogenic topography.

L’architecture et l’évolution tectono-magmatique de la lithosphère européenne sont le résultat d’une succession de périodes de subduction, d’amincissement et de la contribution de panaches mantelliques qui ont modifié le manteau lithosphérique depuis le Néoprotérozoïque (750–500 Ma). Ces événements ont donné naissance à des couplages mécaniques croûte-manteau et lithosphère-asthénosphère très différents entre le manteau résistant, très visqueux, épais, froid et appauvri de la lithosphère archéenne du craton ouest-africain et du craton est-européen, et le manteau faible, faiblement visqueux, mince, chaud et moins appauvri de la lithosphère phanérozoïque de l’Europe centrale. Ces différences qui ont persisté depuis reproduisent au premier ordre bien les contraintes et la topographie actuelles, ainsi que les styles de déformation orogénique. L’absence de relaxation thermique nécessaire au maintien des contrastes rhéologiques pendant plusieurs centaines de millions d’années nécessite un flux de chaleur mantélique élevé sous l’Europe centrale depuis au moins les 300 derniers Ma. Nous suggérons que le principal moteur soit la combinaison de la convection sur les bords des cratons et d’un flux asthénosphérique déclenché par l’ouverture de l’Atlantique et de la Téthys. La topographie de l’Europe centrale est restée en partie soutenue dynamiquement pendant la majeure partie de l’amincissement mésozoïque, en accord avec à la stabilité de la structure thermo-mécanique de la lithosphère. La chronologie et les taux d’exhumation enregistrés en Europe occidentale pendant la convergence indiquent qu’un contrôle supplémentaire par l’architecture des marges continentales mésozoïques est nécessaire. Vers 50 Ma, l’accélération de l’exhumation orogénique, du Haut Atlas aux Pyrénées, s’est produite de façon synchrone avec le début de l’extension et du magmatisme dans le rift ouest-européen. L’extension marque le début d’une évolution orogénique distincte entre l’Europe occidentale (Ibérie) et les Alpes (Adria) à l’est, annonçant l’ouverture de la Méditerranée occidentale. Une réorganisation cinématique majeure s’est produite, déclenchant l’implication de parties moins denses et plus épaisses des marges riftées, ce qui a entraîné une croissance orogénique généralisée. Nous concluons que les modèles conceptuels de collision doivent davantage tenir compte de l’évolution thermo-magmatique de la lithosphère continentale, en particulier de l’architecture et de la composition initiale de son manteau, ainsi que de la connaissance précise de l’architecture des marges continentales pour expliquer la chronologie et la vitesse de construction de la topographie orogénique.

Continental collision is seen as the tectonic process that occurs at convergent boundaries by the accretion of continental rocks, after the suturing of an ocean, and detachment from the underlying lithospheric levels. Collision zones form broad regions of continental deformation, which surface expression shapes the collisional orogenic systems (Dewey et al., 1973; Molnar, 1988). Collision occupies a peculiar place in plate tectonics (Wilson, 1966; Morgan, 1968; Le Pichon, 1968) as diffuse zone of plate convergence that weakly applies to rigid plate approximation (Gordon, 1998). Although closely related during plate convergence, continental collision does not equal subduction that is defined by the loss of lithospheric materials into the asthenosphere and localized strain across the narrow plate interface. End-member types of collisional orogens can be defined based on the magnitudes of crustal thickening (e.g., Alps and Pyrenees versus Himalayas and Andes). But there is no consensus as to whether these differences reflect the mechanical properties of the continental plates (strong vs. weak lithospheres; Mouthereau et al., 2013) or the scale and type of mantle flow (large-scale/symmetric flow vs. small-scale/asymmetric flow) (Faccenna et al., 2013a). In this respect, collision zones with small asymmetric mountain belts like the Alps or the Pyrenees, transiently evolving into subduction orogens in the Mediterranean region (Jolivet, 2003; Jolivet et al., 2008; Royden and Faccenna, 2018) can reveal the deformation of weak continental lithospheres, accommodating convergence by distributed strain, or slab-pull type orogens (retreating subduction zones) related to small scale convection cells.

Despite obvious differences between collision zones they share a common element, the continental lithosphere. Due to its long-lasting tectono-magmatic evolution (> 1–0.5 Gyr), the continental lithosphere is buoyant, strongly anisotropic, both mechanically and compositionally, and structurally complex compared to the oceanic lithosphere. But these parameters, especially their interactions, are ill-defined in most orogenic belts.

Progress in modelling coupling processes between lithosphere dynamics and deep mantle flow have demonstrated that first-order indicators of plate tectonics, such as present-day surface motions, nature and magnitude of strain rates and deviatoric stresses, are best reproduced if lithospheric body forces (gravitational potential energy, GPE) arising from regional variations in topography, crustal and lithosphere thickness and upper mantle density are coupled with mantle flow (Ghosh et al., 2013). It is therefore expected that the mechanical anisotropy represented by long-lived weakness and composition of the lithosphere at cratonic margins play a dominant role on orogenic deformation (Audet and Bürgmann, 2011; Mouthereau et al., 2013), in addition to whole mantle conveyor belts and smaller scale slab-induced mantle flow (Faccenna and Becker, 2010; Becker and Faccenna, 2011).

The evolution of strain distribution, uplift and exhumation during mountain building should therefore theoretically reveal the first-order relationships between structure and inherited properties of the continents (e.g., density, temperature, thickness and anisotropy), mantle convection and the dynamics of subducted slabs. To best examine these relationships it is required to target collision domains that have well preserved the geological record of pre-convergence evolution. Only these domains can well resolve the impact of long-lasting thermal, tectonic and magmatic events on the rheology and architecture in collision zones. Here, we examine the case of Western Europe that is shaped by numerous orogenic segments of variable elevation made of Paleozoic Massifs, Mesozoic and Cenozoic Foreland and Rift Basins, bounded by low topography provinces of the Baltica and Africa shields (Figs. 1 and 2). The topographic and tectonic patterns appear to mirror the contrasting tectono-magmatic evolution between the Archean cratons and the Proterozoic and Paleozoic orogenic belts (Pan-African, Variscan and Caledonian orogenies) (Fig. 1). It has also been recently suggested from reconstructions of Iberia in the Western Mediterranean region (e.g., Angrand and Mouthereau, 2021) and evidence for amagmatic closure in the Pyrenees and the Alps (McCarthy et al., 2020) that the Alpine Tethys, which closed between Europe and Africa, was formed by rift basins overlying a strongly thinned continental lithosphere, including narrow oceanic basins instead of a single large oceanic domain.

Pioneering seismic surveys in the 90’s imaged the deep crustal structure of Europe. Combined with numerical modelling, surface data have permitted linking topography and lithosphere strength with the tectonic evolution of sedimentary basins (Ziegler et al., 1995, 1998; Cloetingh and Ziegler, 2007). Among regional studies, the tectonic reorganisation of the Alpine/Mediterranean region caused by Late Cenozoic slab retreat has received much attention (Jolivet and Faccenna, 2000; Faccenna et al., 2001, 2014). The integration of regional and global seismic tomography with geodynamic models of lithospheric deformation have provided clues on the role of subducted slabs relative to large mantle convection on topography (Faccenna and Becker, 2010) as well as the role of absolute plate motions relative to slab dynamics on crustal deformation (Spakman et al., 2018).

In comparison, although there is also an extensive body of studies on the long-term geological, geochronological and geochemical evolution of the continental crust and mantle, yet this knowledge is not exploited to resolve the temporal and spatial evolution of collisional orogeny. Improving our understanding of how properties of a continent interplay in time and space with kinematic boundary conditions to build mountain ranges was one of the main goal of the OROGEN project. This manuscript is intended to provide a review of the data and concepts on the rheology, tectono-magmatic and topographic evolution of Europe based in part on the results from the OROGEN project. It is organized in five main sections that can be read as independent analyses that together form a coherent ensemble to support our model.

In Section 2, we present the lithosphere architecture and topographic patterns of the Africa-Europe plate boundary based on geophysical constraints. In this part we make the link between the age of the lithosphere, its mechanical properties and composition of the mantle, and the topography. In Section 3, we review the long-term magmatic evolution of the European continental lithosphere through the exploitation of Re-Os and Sm-Nd depletion events. This provides us with constraints to examine further the relationships between the melting ages of the mantle and the overlying crust. This part is aimed to provide the necessary database for understanding the geodynamic processes that have modified the original structure and rheology of the cratons to build the lithosphere of Europe. Combined with geophysical constraints on the lithosphere-asthenosphere boundary (LAB) depth we establish a geological model for the architecture of the lithosphere and age-dependent composition of the mantle. This is done along a 4000 km-long transect stretching between the West African and East-European cratons. In Section 4, we analyze a large and relevant thermochronological dataset to infer spatial and temporal evolution of the Cenozoic orogenic topography. The results are then used to discuss the relationships between the formation of the orogenic topography and the tectonic and magmatic history of the lithosphere. The Section 5 aims to investigate the far-field drivers of plate motions based on a synthesis of the kinematic evolution of Africa-Europe convergence. In Section 6, we summarize the main results and present a dynamic model of the topographic evolution of Western Europe accounting for long-term interactions between lithosphere properties and plate kinematics.

Geophysical fingerprints of lithospheric evolution

The evolution of the continental lithosphere can be described by the variations of tectono-thermal or thermo-tectonic ages, which are known to be well correlated with the strength (or effective elastic thickness of the lithosphere, Te) of continents (Burov and Diament, 1995; Artemieva and Mooney, 2001; Mouthereau et al., 2013). They are defined by the timing of the last main tectonic or magmatic events resetting the lithosphere both thermally and compositionally. In Europe and north Africa, the oldest lithospheres are represented by Archean to Early Neoproterozoic basement exposed in the regions of the East-European Craton (EEC) of Scandinavia, and the Western African Craton (WAC), south of Atlas mountains (Fig. 1). Stretching between the cratonic domains, the central part of Europe has recorded deformation, metamorphism or partial melting during the Caledonian and Variscan orogenies. This newly formed Paleozoic basement was then partially overprinted by deformation and magmatism during the subsequent Mesozoic rifting stages and Cenozoic plate convergence.

The spatial resolution of seismic tomographic imaging of the deep structure of Europe lithosphere has improved in the recent years with a special attention given to resolve 1) the geometry of high-velocity anomalies of the Mediterranean slabs (Spakman, 1990; Bijwaard et al., 1998; Piromallo and Morelli, 2003; Spakman and Wortel, 2004; Meer et al., 2018; El-Sharkawy et al., 2020) and 2) temperature variations in the shallow continental mantle below geothermal fields of the Cenozoic Magmatic Province of Europe (Goes et al., 2000; Heuer et al., 2006; Geissler et al., 2010; Legendre et al., 2012; Meier et al., 2016). We present below some of the main seismic and thermal-mechanical properties the European continental lithosphere inferred from those recent geophysical models.

Crustal thickness and depth of the Lithosphere-Asthenosphere Boundary (LAB) of Figures 2a and 2b are derived from regional EPcrust model (Molinari and Morelli, 2011) and LITHO1.0 model based on global surface wave dispersion (Pasyanos et al., 2014). LAB variations show a pattern that is generally consistent with the trend of lithosphere ages. Archean to early and Neoproterozoic lithosphere (Fig. 1c) is associated with lithosphere thickness above 200 km (Fig. 2b) and thick crust of 40–45 km (Fig. 2a) in the range of crustal roots of Cenozoic mobile belts like the Alps, Dinarides, Pyrenees, or the Atlas. The lithosphere thickness model of EEC inferred from LITHO1.0 is consistent with estimates based on regional surface-wave tomography (Meier et al., 2016) and global models that assumed a direct correspondence between seismic tomography and temperature (Priestley and McKenzie, 2013; Steinberger and Becker, 2016; Hoggard et al., 2020), or other models using joint inversion of multiple data sets (Afonso et al., 2019).

South of the Hellenic trench (Ionian and Herodotus Basins) the LAB is remarkably deep, revealing an apparent lithosphere thickness up to 250 km, which is difficult to reconcile with a thermally-relaxed oceanic lithosphere. Subduction of the Tethys remnant in this domain potentially introduces biases, hence overestimating the LAB depth. Nevertheless, this domain characterized by high Vs anomalies (Fig. 2c) appears to extend below Adria, indicating a colder or/and depleted uppermost mantle and possibly more viscous sub-continental lithospheric mantle (SCLM) in agreement with the role of indenter played by Adria. The expected higher strength of Adria is however not reflected in Te values (Fig. 2d). The younger Phanerozoic lithosphere of central and southern Europe is much thinner, typically below 150 km, and has been thermally eroded, partly delaminated or mobilized during Paleozoic and Cenozoic orogenic evolution.

Differences between cratonic (e.g., EEC and WAC) and Phanerozoic lithospheres are known to reflect different degrees of depletion of the SCLM in incompatible elements and associated minerals (clinopyroxene, garnet), which is particularly important below craton because of the high degree of magmatic extraction (Poudjom Djomani et al., 2001; Carlson et al., 2005; Lee et al., 2011). Depletion of mantle lithosphere below cratons explains their buoyancy. They are also strong as indicated by high effective elastic thickness of the EEC (Te > 80 km in Fig. 2d) and cold, explaining the relatively higher shear velocities of EEC in Figure 2c. By contrast, the Phanerozoic lithosphere has denser SCLM due to refertilization by fluids and melts mobilized during Proterozoic/Paleozoic subduction events (e.g., Cadomian and Variscan subduction) and Mesozoic tectono-magmatic events during post-Variscan extension and opening of Tethys and Atlantic. They are also characterized by higher heat flow, thus limiting their maximum lithospheric thickness to about 150 km and are also much weaker (Te < 30 km). The transition between the different lithospheric provinces is well imaged in Central Europe (Fig. 2) by the abrupt reduction of LAB depth and S-wave velocities across the NW-SE-directed Trans-European Suture Zone (TESZ) (Geissler et al., 2010; Legendre et al., 2012; Schaeffer and Lebedev, 2013; Soomro et al., 2015; Meier et al., 2016).

Despite this general figure, we must distinguish the lithosphere below the Caledonides showing LAB depths close to 150–200 km, and both Vs anomalies and Te values (up to 80 km) that equate those of cratons (Figs. 2c and 2d). Low lithospheric thickness in the 50–100 km range is found in Central Europe (Pannonian Basin) and the Mediterranean Sea (Alboran, Algerian, Tyrrhenian and Aegean Basins). They are associated with attenuated crusts down to 15 km (Fig. 2a) related to back-arc thinning of the continental lithosphere (Faccenna et al., 2014), involving retreating mantle delamination in the Alboran domain (Seber et al., 1996; Calvert et al., 2000; Thurner et al., 2014; Mancilla et al., 2015a; Daudet et al., 2020), the Apennines (Chiarabba et al., 2014; Agostinetti and Faccenna, 2018), and on the northern Africa margin (Roure et al., 2012).

Regions with shallow LAB and thin crust are also found in the Neogene intracontinental magmatic provinces of the Bohemian Massif (Czech Republic, Poland and eastern Germany) in the Ohre Rift (or Eger Graben), in the Eifel volcanic field of the Rhenish Massif and Vogelsberg region emplaced on the Upper Rhine Rift (Goes et al., 2000; Meier et al., 2016) below which a plume has been inferred (Ritter et al., 2001; Kreemer et al., 2020). Similar plume-related volcanism accompanying uplift has been proposed based on teleseismic tomography in the Massif Central in close relationship with the Limagne Graben (Granet et al., 1995). Other Neogene volcanic centers are found in Iberia stretching from the Eastern Betics (∼ 34–2 Ma), Alboran domain (∼ 12.1–6.1 Ma), Valencia Trough (∼ 24–0.01 Ma), Calatrava Volcanic Province (∼ 9–2 Ma) and Olot–Garrotxa (∼ 10–0.01 Ma) that concentrate the most recent occurrence of alkaline lavas. They are interpreted to indicate mixing of partial melts from the lithospheric and asthenospheric mantle. The Late Cenozoic volcanic province temporally and spatially overlaps with the West European Rift system (Figs. 1 and 2).

Figure 3a presents the distribution of Vp anomalies in the deep mantle and the LAB geometry along a 4000 km-long profile stretching from Morocco to Sweden. The Gibraltar anomaly is the only well identified slab along the transect. It is dipping eastwards and rests on the 660 km discontinuity. Based on regional geophysical experiments it is inferred the slab is detached from the surrounding Rif and Betic crustal roots (Mancilla et al., 2015a). Because slab length also equals the size of the Alboran Sea, it is interpreted as part of delaminated Alboran SCLM. This is confirmed by recent geophysical studies (Thurner et al., 2014; Petit et al., 2015; Heit et al., 2017; Molina-Aguilera et al., 2019) and is indirectly supported by the uniform exhumation inferred from thermochronological study in the Flysch Complex of the western Betics (Daudet et al., 2020) and cross-section balancing (Pedrera et al., 2020a, 2020b) that argue for limited convergence between the Alboran basement and the Iberia paleomargin. The lithosphere thins gradually southward from 100–150 km below Germany to 50–100 km below Iberia (Pyrenees, Betics) and Alboran, and in northern Africa below the Rif and the Atlas (Fig. 3b). The boundaries with the thick cratonic roots of 200–250 km for the WAC and EEC are sharp, revealing localized tectono-magmatic reworking of the SCLM (Fig. 2; see discussion in Sect. 3).

It is apparent from these data that the lithosphere thickness and strength express a direct dependence on the age of the last tectonic and thermal events. The most evolved Archean cratonic (WAC and EEC) and their adjacent Proterozoic lithospheres (Figs. 1 and 2) are the strongest, whereas the Phanerozoic lithospheres modified by the succession of Variscan and Alpine orogenic cycles are the weakest.

Topographic fingerprints of European lithosphere thickness and composition

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 that could also be driven by anomalously hot asthenosphere from the deeper mantle (e.g., plume). Erosion and climate modulate these main mechanisms.

A topographic profile of Western Europe (Fig. 4a) parallel to the tomographic section of Figure 3 exhibits topographic variations at different wavelengths. The power spectral density analysis of the topography (inset in Fig. 4) reveals dominant wavelengths at 150, 210, 300 and 800 km. We then filter the topography using those selected wavelengths (Fig. 4b). Wavelengths in the range 150–300 km reveal topographic modulation associated with mountain belts (Pyrenees, Iberian Chain, Betic, Rif and Atlas), volcanic swells associated with the Cenozoic volcanic fields of Western Europe (Massif Central, Rhenish Massif), lithosphere thinning in the Alboran domain. The larger wavelength component of 800 km depicts the regional positive elevation changes ranging between 0 and 1700 m over Iberia and North Africa. Those wavelengths likely express the superposition of isostatic process in the lithosphere and dynamic topography arising from mantle flow and lithosphere strength (e.g., regional isostatic support caused by plate flexure). It clearly appears that the highest elevations are associated with the younger Phanerozoic lithosphere that was tectonically and thermally reset during the Variscan and Alpine orogenic cycles (Fig. 1c).

The non-isostatic “anomalous” component of the topography can be approximated by the residual topography obtained by subtracting the observed topography from the predicted isostatic (Airy) component of the lithosphere. It theoretically requires the knowledge of the thickness and density variations of the crust and mantle lithosphere. In practice, because the thickness variations and composition (densities) of the lithosphere are poorly resolved it is common to fix both to constant values (Faccenna et al., 2014). Assuming constant thickness of the lithosphere (115 km), constant crustal density and crustal thickness derived from updated CRUST1 model (Laske et al., 2013), Faccenna and Becker (2020) estimated negative residual topography above subduction zones (Calabria, Hellenic trench, Ionian Sea, Rif) and positive residuals of 500–800 m for the Atlas, 400 m over Iberia, up to 1000 m in the Massif Central that they identify as dynamic topography driven by mantle flow. In the Mediterranean region, subduction dynamics has indeed been shown to be a major process explaining the main features of this anomalous topography (Faccenna et al., 2014; Faccenna and Becker, 2020).

Below, we test the effect of lithosphere thickness variations on the topography of Western Europe. The approach is similar to Faccenna et al. (2014). First, we calculate the residuals using lithosphere model LITHO1.0 (Pasyanos et al., 2014) shown in Figure 2b and compare with the LAB model of Steinberger and Becker (2016). The regional crustal thickness and density model are taken from EPcrust (Molinari and Morelli, 2011). The elevation of a buoyant lithospheric column in Airy isostatic equilibrium, with respect to the average ridge height of sea level H0 ≈ 2.6 km, is given by Lachenbruch and Morgan (1990):
(1)
where hc and ρc are the crustal thickness and density after EPcrust. hl is the mantle lithosphere thickness, which is left to vary but its density is considered a constant ρl = 3.3 and the asthenosphere density is ρa = 3.226. The residuals are then calculated considering a smoothed topography required for interpolation purpose with crustal thickness, densities and LAB grids.

The predicted topography is presented in Figure 5 for models assuming constant and variable lithosphere thickness. For S&B and LITHO1.0 models that predict reduced lithosphere thickness for the Alpine-Variscan domain the elevation calculated is generally larger than the observed one. However, in many instances, in Iberia and North Africa (Betic, Rif and Atlas) these two models well reproduce the high elevation, as argued by the lower residuals than the model assuming constant thickness of 100 km (Fig. 5). The largest topographic residuals are found in the Pyrenees and in the cratonic areas, where both S&B and LITHO1.0 overestimate or underestimate the topography, respectively. In the Pyrenees, the elevation predicted by LITHO1.0 is indeed about 1 km higher than observed. In comparison, the constant lithosphere thickness model of 100 km appears more satisfactory for the central Pyrenees (Fig. 5). This result is consistent with the underthrusting of the Iberian lithosphere below Europe constrained by resistivity measurements, indicating lithosphere thickness of 110 km in the Pyrenees (Campanyà et al., 2011; Mouthereau et al., 2014). We also note for the Pyrenees that the averaged crustal density inferred from EPcrust of 2.75 (Supplementary Material Figure S1) is probably too low as it is not accounting for the exhumed dense mantle slices that are at the origin of large positive gravity Bouguer anomalies at crustal depths (Wang et al., 2016; Chevrot et al., 2018).

The large differences between observed (high) and predicted (low) topography of the WAC and EEC lithosphere for S&B and LITHO1.0 models indicate an excess of mass represented by a 200–250 km-thick lithosphere. Lower densities of stable Archean cratonic roots are expected due to their high degree of partial melting (Jordan, 1978). The assumption of constant ρl densities in the mantle is therefore certainly not valid. Assuming increasing melt depletion of the cratonic mantle such that its density decreases by Δρl = 0.02 (54 kg/m3 between lithosphere and asthenosphere) diminishes the residual topography up to 1 km where the lithosphere is the thickest (Fig. 5a). Based on these results, it appears that a careful adjustment of density contrasts in the crust and in the lithospheric mantle, informed by petrographic data and geochemical composition inferred from ophiolites, mafic rocks and xenoliths data, can provide, in addition to asthenospheric mantle flow, explanations for anomalous present-day elevation features of Western Europe. Below we provide some clues to a qualitative understanding of lateral variations of the European mantle composition based on a review of its long-term tectono-magmatic evolution.

We have seen in Section 2 that the understanding of the orogenic topography, and obviously its evolution over geological times, requires a better characterization of the architecture and the tectono-magmatic and thermal evolution of the continental lithosphere. A relevant approach for investigating the relationship between orogenic evolution, and the lithosphere rheology back in time is to focus on the evolution of the SCLM itself, as the latter is assumed to be the most resistant layer of the lithosphere (Burov and Watts, 2006). It is expected that tectonic and magmatic events that have previously acted on the lithosphere determine the mechanical coupling between crust and mantle but this is yet to be clearly established. While crustal evolution is directly accessible by studying the evolution of magmatic and sedimentary rocks at the surface, constraining the evolution of the SCLM relies on geochemical and petrographical constraints from ultramafic rocks in ophiolitic complexes, orogenic peridotites massifs, and mantle xenoliths.

In Western Europe, the lithosphere formed by a succession of continental assembly, orogenies and breakup since the Precambrian between two stable cratons, the West African Craton (Reguibat shield) and the East European Craton (Baltica shield) (Fig. 1). As it will be presented in details below, the European lithosphere has been amalgamated during a series of Proterozoic Cadomian and Fennoscandian orogenic events, Phanerozoic continent-continent collisions of Gondwana-derived terranes during the Caledonian and Variscan orogenies. Both the crust and mantle of the original EU cratonic lithosphere have been deeply remobilized during these successive tectonic and magmatic events, including magmatic arc formation and accretion, partial melting in crust and mantle, rifting between Gondwana and Laurentia. These events define a structural and mechanical background of a layered crust and mantle that was reactivated during the Tethyan rifting in the Mesozoic and during the Cenozoic convergence. In the following, we present a synthesis based in part on Re-Os and Sm-Nd depletion events inferred from mantle rocks in ophiolites and xenoliths or mantle-derived magmatism in the crust. Constraints are presented from North to South, and for cratonic and phanerozoic lithospheres, and summarized along a 4000 km cross-section of W Europe.

Europe lithosphere evolution: East European Craton and Avalonia

East European Craton (Baltica) and Avalonia

The East European Craton (Baltica) forms a 250 km-thick lithosphere (Figs. 2 and 3) made of a coherent ensemble of Precambrian continental segments (Figs. 1 and 6). These segments include Fennoscandia, exposed in the Baltica Shield, Sarmatia, exposed in the Ukrainian Shield and Volga-Uralia that were assembled at ca. 1.8–1.7 Ga (Bogdanova et al., 2008). The cratonic core of Fennoscandia is the Karelian and Kola craton (3.2–2.5 Ga) (Lahtinen et al., 2005). Re-Os isotopic composition of mantle xenoliths sampled in 600 Ma kimberlite pipes confirms the main part of the Karelian lithospheric mantle, down to 180 km, is made of harzburgitic layer that derives from a 3.3 Ga magmatic event of the same age as the overlying crust (Peltonen and Brügmann, 2006) (Fig. 7). It overlies a basal, less depleted and metasomatized, garnet-lherzolitic/harzburgitic mantle between 180 and 250 km that formed at ca. 2 Ga during the earliest stages the Svecofennian orogeny (Peltonen and Brügmann, 2006) (Figs. 6 and 7). The western craton margin recorded partial melting of an enriched mantle that result from the rise of volatile-enriched magmas from the Svecofennian subduction (1.95–1.86 Ga) at the origin of a volcanic arc that eventually collided with Fennoscandia and the Volgo-Samartian craton (Rutanen et al., 2011). A subsequent post-collisional bimodal (mafic and felsic) magmatism from the Svecofennian domain emplaced at 1.82–1.79 Ga (Fig. 7). As EEC was amalgamated to Rodinia supercontinent, during collision with Amazonia, its western margin was involved in the Sveconorwegian-Grenvillian orogeny (1.15–0.9 Ga) (Torsvik et al., 1996; Li et al., 2008; Cawood and Pisarevsky, 2017; Bingen et al., 2020). The Ediacaran-Cambrian Iapetus Ocean opened at 570 Ma by the separation of Baltica, Laurentia and Amazonia. Continental breakup was likely triggered by plume magmatism of the Central Iapetus Magmatic Province (CIMP) at 615–590 Ma on the NW margin of Baltica still preserved as dyke swarms (Fig. 7) in the overthrusted pre-Caledonian distal margin of Baltica (Svenningsen, 2001; Tegner et al., 2019a).

Peridotite lenses from the HP and UHP terrains of the Western Gneiss Region of the Scandinavian Caledonides (420–380 Ma) provide constraints on the evolution of the continental mantle of the margin of Baltica shield. An ancient Archean depletion event at 3.2 Ga, at the origin of large lenses of harzbugites and dunites, is reported in garnet mantle peridotites that were later affected by a Mid-Proterozoic melt-related refertilization event at 1.6 Ga (Fig. 7) contemporaneous with Gothian orogeny (Brueckner et al., 2002; Beyer et al., 2004; Beyer et al., 2006). Later crustal (and mantle) reworking is attested by metamorphism and magmatism during the Sveconorwegian-Grenvillian orogeny at 1.15–1.25 Ga (Lapen et al., 2005) and after at 0.95–0.94 Ga (Söderlund et al., 2002, 2005; Bogdanova et al., 2008). A more recent tectonic event dated to ∼ 600 Ma, likely related to Central Iapetus Magmatic Province (CIMP), is indicated by garnet kelyphite in mantle peridotites. It reflects decompression/exhumation to crustal depth associated with lithosphere thinning during the opening of the Iapetus (Lapen et al., 2009).

The Caledonian orogeny that resulted from the closure of Iapetus Ocean between Laurentia and Baltica initiated in the Late Cambrian-Ordovician ca. 500 Ma (Roberts, 2003) with intra-oceanic subduction and the formation of the Grampian-Taconic arc (Cawood et al., 2001; Chew and Strachan, 2014; Domeier, 2016). The final suturing in the Scandinavian Caledonides known as the Scandian phase ∼ 420–380 Ma (Silurian-Devonian) is recorded by continental subduction and underthrusting of the Baltica margin (Roberts, 2003; Hacker and Gans, 2005; Kylander-Clark et al., 2007; Corfu et al., 2014).

Laterally, south-westwards, arc-continent collision occurred between the Grampian oceanic arc and Laurentia margin in the Late Ordovician (450 Ma) while subduction was ongoing below east Avalonia (SW Britain, Brabant Massif, N Germany and Poland) (Chew and Strachan, 2014). In Scotland, collision of the southern margin of Laurentia with Baltica is dated to mid-Silurian (430–440 Ma) coeval with the Scandian phase (Goodenough et al., 2011; Domeier, 2016), but the final oblique collision of Laurentia with Avalonia is constrained by a Middle Devonian unconformity (Chew and Strachan, 2014). During the Early Ordovician, east Avalonia and Baltica margins were separated by the Tornquist Sea (Figs. 6 and 7), which then closed by subduction and collision in the Late Ordovician ∼ 450 Ma (Torsvik and Rehnström, 2003; Domeier, 2016). This narrow suture now forms the Trans-European Suture Zone (TESZ), a marked geophysical transition between the EEC (Baltica) and Caledonian/Variscan Belt of Central Europe (Figs. 1 and 2). A Late Carboniferous (Late Variscan)-Permian extension (310–245 Ma) associated with voluminous magmatism occurred on the southern margin of Fennoscandia, forming the Oslo Rift (Heeremans et al., 1996) (Figs. 1 and 6). This event is coeval with widespread post-Variscan alkaline magmatism observed elsewhere in Europe (see below). A source for these lavas is inferred to be represented by garnet-bearing peridotites that have been affected by partial melting at the lithosphere-asthenosphere boundary (Neumann et al., 2002) possibly linked to a metasomatism of the lithospheric mantle associated with emplacement of carbonatite at 600 Ma (Neumann et al., 2004) and contemporaneous with CIMP.

South of TESZ, detrital zircon from Cambrian-Silurian sediments of the Brabant Massif and Nd isotopic signature of magmatic rocks reveal the margin of East Avalonia is formed by a Pan-African (Neoproterozoic) basement recycling an older Mesoproterozoic juvenile crust (1.3–1.7 Ga), that was rifted from Gondwana during the Cambrian (Linnemann et al., 2012). The evolution of West Avalonia (Nova Scotia) is concurrent with East Avalonia since it recorded the same main Cadomian continental arc magmatic event north of Gondwana during the Neoproterozoic (630–570 Ma). Isotopic composition of magmatic rocks from Avalonia basement show it derives from recycling and melting of 1.2–1.0 Ga juvenile crust (Murphy et al., 2000; Nance et al., 2002). Most rift-related mafic rocks that crystallized between 610 and 370 Ma were derived from the same mantle enriched at 0.8–1.1 Ga (Murphy and Dostal, 2007) (Fig. 7) possibly related to global-scale high-degree melting of the mantle during Rodinia assembly (Dijkstra et al., 2016). Taken altogether these data confirm that Avalonia crust and lithospheric mantle remained coupled since rifting and drifting of Avalonia from Gondwana until collision with Laurentia.

West African Craton (Reguibat Shield) and Anti-Atlas

Positioned to the south of our transect, the West African craton (WAC) includes the Reguibat shield and the Anti-Atlas to the north, separated by the Tindouf Basin (Figs. 1, 6 and 7) where the lithosphere thickness is 200–250 km (Fig. 2). U-Pb and Sm-Nd ages from a granulitic pluton indicate the Reguibat shield formed by crustal growth events at 3.5 Ga followed by HT metamorphism between 2.98 Ga and 2.7 Ga and crustal melting at 2.7 Ga (Potrel et al., 1998). An Re-Os age of ∼ 3.4 Ga for eclogite xenolith from the Man shield, south of WAC, suggests the deep lithospheric mantle formed, or was amalgamated, during the Archean crustal growth stage (Barth et al., 2002).

The Anti-Atlas consists of a 2 Ga Eburnean basement (Thomas et al., 2002, 2004) that has been involved in the Pan-African/Cadomian, Variscan and Alpine orogenies (Ennih and Liégeois, 2001). This domain is segmented by a Neoproterozoic (686 Ma) Pan-African suture. The suture is represented by the Sirwa and Bou Azzer ophiolitic complexes dated to 759–762 Ma (Thomas et al., 2002; Samson et al., 2004; Hodel et al., 2020) that were originally emplaced in back-arc setting as a result of the intra-oceanic subduction and closure of the Pan-African Ocean between the WAC and Cadomia-Avalonia terranes (Ennih and Liégeois, 2001). Arc magmatism resulting from this subduction was long-lived, ranging from 750 to 650 Ma, and characterized by episodic accretionary stages (Triantafyllou et al., 2018). The final obduction of the back-basin and ophiolites is dated by the age of arc accretion on the northern wedge of the WAC at ∼ 630–580 Ma (Thomas et al., 2002; Gasquet et al., 2008) before the intrusion of Late Pan-African granite dated to 580 Ma (Inglis et al., 2004) possibly related to delamination and transtension (Fig. 7). To the north-east of the suture (e.g., Sirwa and Sarho Massifs), the Eburnean basement is not exposed. It is covered by Cryogenian turbidites and basalts that were deposited on the rifted margin of the Eburnean crust. This is in agreement with the formation of a juvenile magmatic arc, similar to Cadomian magmatism recognized to the North, with Nd isotopic signature from the 750–700 Ma gneiss and other igneous rocks and intrusive of the Bou Azzer inlier (D’Lemos et al., 2006; Triantafyllou et al., 2020), indicating no substantial assimilation of the Eburnean basement. To the southwest, in contrast, the Eburnean basement is exposed and forms Precambrian inliers “boutonnières”. They are positioned on the margin of the WAC that was affected by a 1.75 Ga magmatic event, likely reflecting a tentative initial breakup of the WAC lithosphere (Youbi et al., 2013).

Summary: tectono-thermal events in EEC, WAC and Avalonia

From this brief review, it is apparent that both crust and mantle of the East European Craton (EEC) have remained coupled since the Archean. Large bodies of refractory (dunite/harzburgite) Archean mantle (Beyer et al., 2006) are documented, indicating the core of Karelian craton resisted melt-related refertilization during Proterozoic Svecofennian magmatic event at 1.8 Ga and the subsequent Sveconorwegian-Grenvillian event. The western margin of Baltica, however, recorded two major magmatic and metamorphic events during plume metasomatism at 600 Ma and subduction at 500–380 Ma during opening and then closure of the Iapetus Ocean. The 600 Ma event thermally eroded and thinned the lithosphere to about 100 km (Fig. 7) as argued by geophysical data (Fig. 2). The rapid change in lithospheric thickness may have triggered small-scale edge-driven convection cells on the Baltica margin as suggested for the WAC (e.g., Kaislaniemi and van Hunen, 2014). This event promoted inversion and subduction during the Scandian orogeny and later the Permo-Carboniferous rifting along the TESZ.

The lithosphere of Avalonia bears the imprints of an older Neoproterozoic (750–540 Ma) tectono-thermal event related to subduction north of Gondwana, which resulted, during the Cambrian-Early Ordovician in rifting and drifting of Avalonia from Gondwana and the opening of the Rheic Ocean (Murphy et al., 2006; Domeier, 2016). The close genetic relationships between crustal growth and mantle depletion events since the Neoproterozoic suggests the sub-continental mantle of Avalonia remained coupled to the crust. We view Avalonia as an intermediate continental block characterized by a decratonized, relatively thin 100 km lithosphere with no significant crust-mantle decoupling (Fig. 7). Isotopic data from the Western African Craton (WAC) in the Reguibat and Man shields, although less abundant than for the EEC, indicate the crust and mantle are coupled since the Archean. The first evidence of magmatism on the margin of the WAC arguing for the removal of the cratonic root and potential decoupling between crust and mantle is dated to 1.75 Ga.

Variscan Europe and Northern Africa: a tectono-magmatic history of long-lasting lithospheric thinning

Prolonged tectono-magmatic reworking of peri-Gondwanan terranes

South of Avalonia, peri-Gondwana terranes have recorded the Cadomian/Pan-African orogeny (Fig. 6). They were then involved in the Late Paleozoic Variscan orogeny during closure of the Rheic Ocean (Martínez Catalán et al., 2007; Kroner and Romer, 2013; Ballèvre et al., 2014). These terranes include Armorica, Saxo-Thuringia, Bohemian Massif and NW Iberia (Galicia). The lithosphere with that thermal ages appears particularly thin (Figs. 2 and 3) characterized by lithospheric thickness below 100 km and weak (Te < 30 km). They are separated by Cambro-Ordovician and Devonian ophiolitic sutures (Fig. 7), with many overprinted by HP metamorphism (Berger et al., 2006). Whether these ophiolites represent sutures of large oceanic domains between pre-Gondwanan Units or thin continental crusts is still controversial (Matte, 2001; Kroner and Romer, 2013; Franke et al., 2017). They comprise the Rheic-Rheno-Hercynian suture between Avalonia and North Armorica devoid of HP metamorphism (Figs. 6 and 7), the Galicia-Southern-Brittany/Bohemia suture (also named Medio-European, Paleo-Tethys, Eo-Variscan suture or Central-Moldanubian) between Armorica and Gondwana(paleo-Adria), with an intermediate Saxo-Thuringia suture between North and South Armorica/Bohemia (Franke et al., 2017).

North Armorica exhibits a juvenile 2.06 Ga crust (Icart Gneiss), having close affinity with the 2.1 Ga Eburnean basement of the West African craton (Samson and D’Lemos, 1998). This basement was then involved in the Cadomian cordilleran evolution (620–540 Ma) to form a large zone of deformation and magmatism (Chantraine et al., 2001). Between Avalonia and Armorica, the Rheic Ocean opened by the reactivation of Neoproterozoic suture in the Late Cambrian–Early Ordovician during closure of the Iapetus Ocean (Murphy et al., 2006, 2009). The Lizard ophiolite in SW Britain is commonly viewed as the Rheic oceanic suture. However, this ophiolite hosts sheared spinel lherzolite transformed before Early Devonian (397 Ma) into low-pressure plagioclase-bearing peridotites during exhumation to depth of 25 km in a magma-poor rifted margin (Cook et al., 2000). This ophiolite therefore represents the rifted margin of short-lived Rheno-Hercynian Ocean rather than the Rheic Ocean that is inferred to be a wide oceanic domain based on paleobiogeography and paleomagnetism (Domeier, 2016; Franke et al., 2017).

South Armorica and NW Iberia display widespread Cadomian magmatism (750–550 Ma) and Early Ordovician magmatism (Martínez Catalán et al., 2007; Ballèvre et al., 2014) (Fig. 7). Geochemistry and Nd isotopic analyses of Early Ordovician lavas, including basalts from NW Iberia, reveal a mixed magmatic signature ranging from a juvenile mantle to a sub-continental mantle enriched at 1 Ga during Cadomian subduction (Murphy et al., 2008). These lavas are interpreted to indicate a continent-ocean transition on the northern Gondwana margin that developed synchronous with the opening of the Rheic Ocean. The widespread Ordovician extension between Armorica and the northern margin of Gondwana (Fig. 7) formed slow spreading centers or hyper-extended crust (Berger et al., 2006; Lardeaux, 2014). Paleobiogeographic and paleomagnetic data also indicate that the Massif Central and Gondwana were not separated by a large ocean, suggesting that these extended domains are distinct from the Rheic Ocean (Fortey and Cocks, 2003). It corresponds to the Galicia-Southern-Brittany suture (Fig. 6), south of Massif Central and in the Galicia-Cabo Ortegal complex, characterized by 495–469 Ma (Cambrian-Early Ordovician) MORB-type magmatic rocks that recorded UHP metamorphism at 412–390 Ma (Early Devonian) (Casado et al., 2001; Berger et al., 2010). Petrology and geochemistry of mantle pyroxenites within the harzburgitic massif from the Cabo Ortegal complex indicate they were produced by melt-rock reactions at the root of Cadomian arc during the Cambrian-Neoproterozoic (654 ± 72 Ma). This is before the terminal emplacement onto the rifted Gondwana margin during the Variscan orogeny (Tilhac et al., 2016, 2017).

Geochemical and geochronological analyses of detrital zircons in orthogneiss from the Massif Central indicate the pre-Variscan crust is derived from sediments deposited on the northern Gondwana margin (Melleton et al., 2010; Chelle-Michou et al., 2017). Soon after deposition in backarc setting at 550–545 Ma partial melting of these sediments produced the S-type Cadomian granites (Couzinié et al., 2017). This origin is also argued by Archean and Paleoproterozoic detrital zircons recovered from Paleoproterozoic gneissic rocks in eastern Europe that indicate the Cadomian “basement” reworked a thinned remnants of older cratonic crust (Linnemann et al., 2007). A similar Gondwanan source of detrital zircon is documented in 630–385 Ma sediments from Bohemia Massif, indicating these terranes remained close to Gondwana margin until the Devonian (Drost et al., 2011). In the Massif Central, the Early Ordovician (480 Ma) extension is attributed to the formation of peripheral marginal basins floored by thinned lithosphere with local oceanic affinity, now metamorphosed as amphibolite. A synthesis of Sm-Nd isotopic data from Neoproterozoic and Early Paleozoic mafic rocks (800–500 Ma) in the Bohemian Massif, Vosges, Armorica, Massif Central, and Pyrenees (Dostal et al., 2019), further reveals they were extracted from a juvenile mantle (i.e., depletion age of 800–500 Ma). This suggests melting during asthenospheric upwelling in a backarc basin and opening of the Rheic Ocean (Fig. 7).

Late- to post-Variscan transition: Late Carboniferous-Permian thermal event

The Variscan orogen terminates at the end of the Carboniferous-Early Permian with the transition towards extension associated with widespread volcano-sedimentary basins and mafic magmatism throughout Europe (Fig. 8; Wilson and Downes, 1992; Ménard and Molnar, 2004; Neumann et al., 2004; Timmerman, 2004; Wilson et al., 2004; McCann et al., 2006). The accumulation of Permo-Carboniferous volcanic-derived sediments (Figs. 7 and 8) reaches 1–2 km kilometers in the North-German Basin (Timmerman, 2004), in the Pyrenees (Lago et al., 2004a; Gretter et al., 2015; Rodríguez-Méndez et al., 2016; Espurt et al., 2019), Iberian Ranges (Arche and López-Gómez, 2005), Massif Central (Bruguier et al., 2003; Michel et al., 2015) and southern Alps (Cassinis et al., 2011). Magmatic events were presumably triggered by a combination of right-lateral wrenching (Fig. 6) between Gondwana and Laurussia (Matte, 2001) closely connected to the opening of the Neo-Tethys and N-directed subduction of the Paleo-Tethys (Kroner and Romer, 2013). Latest Carboniferous and Permian plutonism is widespread in the southern, central and eastern Europe, including the Pyrenees (Michard-Vitrac et al., 1980; Maurel et al., 2004; Denèle et al., 2007; Esteban et al., 2015; Kilzi et al., 2016; Vacherat et al., 2017), the Massif Central (Faure, 1995), Spain and Portugal (Dias and Leterrier, 1994; Bea et al., 1999; Fernández-Suárez et al., 2000), Germany (Romer et al., 2001), North-Sudetic Basin of Bohemian Massif (Awdankiewicz et al., 2013), Poland near the TESZ (Żelaźniewicz et al., 2016) and Dobrogea (Seghedi, 2012) (Fig. 8). The heat required by the Late Variscan magmatic event is interpreted to indicate post-collision mantle delamination (Fig. 7), which may have commenced before the latest Carboniferous, for instance, in Iberia (Fernández-Suárez et al., 2000). Indeed, there are several indications of mantle-derived sources as old as 330 Ma that also reveal melting of an enriched mantle such as for the Central Massif (Laurent et al., 2017), the Alps (Bussy et al., 2000) and central Europe (Raumer et al., 2013). Only in the external Hellenides of Crete (Zulauf et al., 2015) and in the Menderes Massif (Candan et al., 2016) the Late Carboniferous-Early Permian magmatism is related to arc magmatism that resulted from the subduction of the Paleo-Tethys.

Geochemical and petrographic data in central and southern Europe show evidence of a drastic change in the composition of the mantle. This is argued by negative Nd isotopic values and old depletion ages (1.8–0.8 Ga) from Late Carboniferous-Permian lamprophyre dykes and high Mg-K rocks (Dostal et al., 2019). An interpretation is the partial melting during post-Variscan rifting of a mantle previously enriched during the Variscan subduction (Soder and Romer, 2018). Modification of the mantle source for the same crustal regions suggest crust and mantle were both decoupled during the Variscan orogeny (Dostal et al., 2019). This is in line with inferences from the Massif Central that reveal a lack of autochthonous basement older than 550 Ma (e.g., Cadomian basement) below Gondwanan-derived sediments (Chelle-Michou et al., 2017). Mantle xenoliths (spinel lherzolites) associated with the Miocene Massif Central volcanism preserve a record of ancient depletion events at 313–377 Ma (Wittig et al., 2006; Harvey et al., 2010) in agreement with melt extraction in a spinel-facies mantle during the Variscan orogeny (Fig. 7). Despite intense metasomatic events during the Neoproterozoic (Cadomian arc), the Variscan and the Cenozoic, Re-Os isotopes reveal a 1.8 Ga depletion event, arguing for preservation of Proterozoic magmatic events (Harvey et al., 2010). These data collectively point to the heterogeneity of the Late Variscan mantle, comprising a component of juvenile mantle depletion event (0.3 Ga) in an heterogeneous, mostly Proterozoic, mantle (0.8–1.8 Ga).

Post-Variscan rift-related lithosphere evolution

Permian post-orogenic HT metamorphism of the Variscan basement (e.g., granulite) and mafic intrusions of mantle-derived magmas (gabbros, lamprophyre dykes, syenites, microdiorites) (Figs. 7 and 8) are recognized in Iberia (Orejana et al., 2008, 2009, 2020; Gutierrez-Alonso et al., 2011), in the Pyrenees (Debon and Zimmermann, 1993; Lago et al., 2004b), in the Alps (Voshage et al., 1990; Barth et al., 1993; Vavra et al., 1999; Hermann and Rubatto, 2003; Schaltegger and Brack, 2007; Schuster and Stüwe, 2008; Petri et al., 2016) and Corsica (Paquette et al., 2003; Martin et al., 2011). The magmatic event spread over a region (Fig. 8) that is now defined by a thin lithosphere (50–100 km) and low S-wave velocities (Fig. 2). This event is also documented in garnet pyroxenites of Ronda Massif (Sánchez-Rodrı́guez and Gebauer, 2000) and by anatexis found associated to the initial emplacement of the Beni-Boussera peridotites during the Permo-Carboniferous (Rossetti et al., 2020). Mafic-ultramafic xenoliths further argue for calc-alkaline Permian (ca. 257 Ma) magmatic underplating, for instance, in Massif Central (Féménias et al., 2003).

Pressure-temperature estimates in Late Variscan gabbros or diorites and post-Variscan Permian alkaline magmatism (lamprophyre dykes) in Iberia reflect increasing depths of magma production from the garnet-spinel lherzolites field in the mantle lithosphere to the asthenospheric mantle (Orejana et al., 2020). This late magmatic event could mark the transition towards the CAMP at ca. 200 Ma (Figs. 7 and 8) recognized in the same region of Spain (Cebriá et al., 2003), announcing the onset of break-up of Pangea and the opening of the Central Atlantic at 190–175 Ma (Labails et al., 2010; Olyphant et al., 2017). The imprint of the Permian magmatic event in the mantle peridotites is particularly well recorded in the Alps and Apennines. Nd isotopic data in Ligurian ophiolites reveal a Permian 275–285 Ma depletion event in a Proterozoic (2–1 Ga) subcontinental mantle of Gondwanan (African) affinity (Rampone et al., 1995, 1996; McCarthy and Müntener, 2015). The depleted Permian mantle formed in regions of partially melted asthenosphere, which after cooling became part of the Permian sub-continental mantle (McCarthy and Müntener, 2015; Picazo et al., 2016). Asthenospheric upwelling related to extension and thinning during the Jurassic ca. 160 Ma is reflected in co-genetic gabbros and basalts with typical MORB composition (Kaczmarek et al., 2007). A similar relationship is found in Corsica at ca. 162 Ma (Rampone et al., 2009). In addition, the ascending melt can react with the old and Permian fertile spinel lherzolites, forming domains of refertilized plagioclase lherzolites (Müntener and Piccardo, 2003; Piccardo et al., 2004). Grain-size reduction associated with melt-rock reaction is a potential softening mechanism, through grain-size sensitive (GSS) creep (Dijkstra et al., 2002), during syn-magmatic extensional shearing (Kaczmarek and Müntener, 2008). The Pyrenean peridotites appear to be in a different situation. They are spinel lherzolites formed by asthenospheric melt-related refertilization of previously depleted harzburgites of an Archean (> 2 Ga) mantle (Reisberg and Lorand, 1995; Roux et al., 2007). The refertilization event is suggested to reflect the Late Variscan magmatic event (Fig. 7), contemporaneous with the emplacement of lower crustal granulite and occurred in spinel stability field at about 50 km. After thermal re-equilibration, the late amphibole-pyroxenite dykes show no evidence for widespread refertilization and rapid cooling at 104 Ma, contemporaneous with alkali basaltic lavas emplaced during the Cretaceous rifting (Fabries et al., 1991; Henry et al., 1998). Refertilization at the base of a 1.3–1.4 Ga subcontinental mantle (Reisberg et al., 1989; Pearson and Nowell, 2004), possibly as old as 1.6–1.8 Ga (Sánchez-Rodrı́guez and Gebauer, 2000; Marchesi et al., 2010), by melt-rock reactions is also identified in Ronda lherzolites (Garrido and Bodinier, 1999) and the pyroxenites of Beni Bousera (Varas-Reus et al., 2017) (Fig. 7). They have preserved the imprint of successive stages of exhumation from depths > 140 to 85 km by decompression and cooling from diamond field to graphite-garnet lherzolite facies (Garrido et al., 2011). This occurred during the Jurassic (Wal and Vissers, 1993) as indicated by Early Jurassic to Late Jurassic-Early Cretaceous U-Pb zircon ages in garnet pyroxenites and eclogite (Sánchez-Rodrı́guez and Gebauer, 2000). This initial event was followed by cooling and exhumation, leading to garnet-spinel mylonite, high heat flow and partial melting at 21–22 Ma (Garrido et al., 2011) (Figs. 7 and 8), likely linked to delamination of the Alboran slab seen in Figure 3, and responsible for the observed anomalous topography (Fig. 4) and thin lithosphere (Fig. 2). Despite recent melting, Re-Os dates on sulfides document preservation of melting events of the European-North African mantle at 1.6–1.8, 1.2–1.4 and 0.7–0.8 Ga (Marchesi et al., 2010). Some similar Re-depletion ages from peridotite xenoliths of the Calatrava volcanic field (2 Ma) (González-Jiménez et al., 2013) report the oldest mantle-crust magmatic events at ca. 1.6–2.0 Ga. Those ages are consistent with the original Paleoproterozoic mantle below W-Europe and N-Africa, related to Columbia supercontinent assembly. The crust and mantle were then remobilized at 600 Ma during the Cadomian orogeny and at 300 Ma during a melting event following the Hercynian orogeny. This pattern of Re-Os ages is shared by most Cenozoic magmatic provinces throughout Europe and northern Africa (e.g., Massif Central, Rhenish Massif, Bohemian Massif, Pyrenees, Azrou) (González-Jiménez et al., 2013). The eastern Betics and Beni Boussera Massif, Sicily and Adria, in contrast, appear to preserve an older Archean (2.2–2.6 Ga) lithospheric mantle.

Variscan to post-Variscan magmatic evolution on the northern margin of WAC

The Moroccan Variscan Belt driven by moderate convergence resulted from the inversion of the NW Gondwana margin. It was then displaced relative to its Iberian counterpart by large transcurrent movement during the Late Carboniferous (Mattauer et al., 1972; Houari and Hoepffner, 2003) and Permian (Hoepffner et al., 2006; Michard et al., 2010) (Fig. 6), which significance and age are greatly debated for Pangea reconstruction (Domeier et al., 2020). The Pan-African crust of Anti-Atlas was weakly overprinted by Late Variscan shortening (Helg et al., 2004; Séverine et al., 2004; Burkhard et al., 2006). Pan-African granitoids (625–552 Ma) involved in the Variscan Belt (Moroccan Meseta) (Fig. 7) originally formed at the end of the Cadomian-Pan African orogeny. In the Central Meseta, Nd model ages of 0.8–1 Ga of Early Paleozoic gabbro, dolerites and rhyolites dated to Cambrian-Early Ordovician (519–479 Ma) reveal they have a relatively primitive signature (Ouabid et al., 2020). This part of the Moroccan Meseta basement together with the Anti-Atlas formed the NW Gondwana rifted margin. In the Western Meseta, depleted Nd model ages from syntectonic Variscan ultramafic-granitoid magmatism, and calc-alkaline granitoids of the Jebilet Massif (330–300 Ma) reveal contribution from Proterozoic (1.7–0.8 Ga) and Eburnean crust (1.9 Ga), in addition to a juvenile mantle component (younger Nd model ages of 1.3–1.5 Ga). This evolution marks the progressive cessation of subduction and lithosphere delamination (Fig. 7).

Magmatism that occurred during Late Variscan transpressional stage produced increasing heat flux that caused partial-melting of the Proterozoic metasediments (Essaifi et al., 2014). Carboniferous-Early Permian zircon U-Pb ages recovered from crustal xenoliths sampled in mafic Triassic (∼ 235 Ma) lamprophyre dykes indicate that HT metamorphism may have been responsible for granite emplacement (Dostal et al., 2005). Old zircons dated at ∼ 2 Ga point to the presence of the Eburnean basement below the Jebilet Massif, indicating this domain was part of the NW Gondwana margin. Nd isotopic data from the same Triassic dykes yielded model ages of 0.6–0.8 Ga consistent with Pan-African/Cadomian magmatism. This Triassic magmatism correlates with other Late Permian (264–255 Ma) lamprophyres dykes, dolerites, and gabbros documented in the Tafilalt Basin of the Anti-Atlas sourced from a deep metasomatized mantle that heralds the CAMP event (Najih et al., 2019).

The northern margin of the Anti-Atlas was reactivated during the Atlantic and Tethys rifting, as recorded in the High and Middle Atlas Rift systems in the Late Permian-Early Triassic to the Early Jurassic (Frizon de Lamotte et al., 2008). Rifting was accompanied by widespread lava flows, sills and dykes of the CAMP (201 Ma), encompassing the Anti-Atlas, Central and High Atlas, and Western and Eastern Meseta (Marzoli et al., 2019) (Fig. 8). Partial melting associated with Moroccan CAMP lavas occurred in the asthenospheric mantle between 110 and 55 km (Fig. 7), which is the range of lithosphere thickness inferred from geophysics (Fig. 2). It is characterized by relatively higher potential temperature (1430 °C) possibly triggered by a plume (Tegner et al., 2019b). The combined effects of tectono-magmatic events are responsible for the sharp thickness difference observed on the northern margin of the African craton. This geometry in turn triggered small-scale edge-driven convection at the origin of episodic Cenozoic alkaline magmatism and high topography (Missenard and Cadoux, 2012; Kaislaniemi and van Hunen, 2014) (Fig. 7).

Younger Upper Jurassic to Lower Cretaceous and Cenozoic magmatic events are recognized in the High Atlas (Fig. 7). They are interpreted to reflect melting of the lithospheric mantle in areas previously affected by rifting and CAMP magmatism (Bensalah et al., 2013). The region is marked by anorogenic Miocene to Quaternary alkaline volcanism that developed in a narrow SW-NE-trending corridor of thinned lithosphere and high heat flux dubbed the “Moroccan Hot Line” (Missenard et al., 2006; Frizon de Lamotte et al., 2009) (Figs. 6b, 7 and 8). Mantle xenoliths from Middle Atlas lavas show evidence for a metasomatized mantle (Wittig et al., 2010a, 2010b; Malarkey et al., 2011), indicating lavas derive from a 1.7 Ga residual mantle (Wittig et al., 2010a). This age is consistent with Nd isotopic data from lherzolite and wehrlite xenoliths that reveal a Neo- to Paleo-Proterozoic depletion event (Raffone et al., 2009). Detailed isotopic and trace elements analyses confirm these lavas formed by partial melting of a metasomatized mantle near the lithosphere-asthenosphere boundary at 65 km (Bosch et al., 2014), a depth similar to the current lithosphere thickness. Refertilization of the lithosphere by alkaline melts is interpreted to indicate a flux of hot asthenospheric mantle, induced by edge-driven convection in combination with enriched plume materials. This mechanism was probably effective since the Late Cretaceous and was enhanced during the Cenozoic by the retreat of the Alboran slab.

Summary: plume-related, subduction and rifting events in peri-Gondwana terranes

The succession of metasomatic and tectonic events in northern Gondwana, during the Neo-Proterozoic convergence (Cadomian arc), Ordovician rifting (back-arc extension), Late Paleozoic subduction and orogeny (Variscan collision) and Permian extension (Pangea breakup/Tethys opening) caused a marked thermo-chemical thinning of the SCLM, and the formation of a juvenile arc crust (Fig. 7). One can distinguish two main tectono-magmatic events that have profoundly thinned the SCLM. The first one is the Cadomian subduction and backarc rifting at 500–600 Ma, the second one is Pangea assembly and breakup in the Permian at about 270 Ma, which is the last main magmatic event and therefore should be considered as the tectono-thermal age of the lithosphere in Central Europe (Fig. 8). Chemical thinning has additionally been enhanced by mechanical removal of the SCLM during delamination (Fig. 7). In addition to these events, it has been proposed that the thermal insulation below Pangea supercontinent produced an increase of the potential temperature (up to 100 °C) of the convective mantle responsible for the CAMP event on the Atlantic margins and initial Pangea breakup (Wilson, 1997; Marzoli et al., 1999a, 199b, 2004) at 200 Ma (McHone, 2000; Coltice et al., 2007, 2009; Ganne et al., 2016; Peace et al., 2019), leading to final Central Atlantic breakup at ca. 185 Ma. The contribution of the Late Cenozoic subduction and delamination in the Mediterranean region to the current SCLM thickness cannot be ignored but is yet to be estimated.

Implications for lithosphere-asthenosphere and crust-mantle coupling

Strong plate-mantle and crust-mantle coupling of cratonized lithosphere

The structure of Baltica conforms well with the typical structure of a craton (Fig. 7). Based on Griffin and O’Reilly (2007), we follow Afonso et al. (2008) to suggest that Baltica is composed of two layers: 1) one shallow depleted Archean mantle layer (dunite/harzburgite) topped by spinel lherzolite (not represented in Fig. 7), and 2) a deeper Proterozoic refertilized layer, composed of depleted lherzolites in the garnet field, possibly overlying a melt-metasomatised layer in contact with the asthenosphere. Such depleted mantle is expected to be stable due to its buoyancy and high viscosities (1023–1024 Pa.s) relative to the asthenosphere (Ghosh et al., 2013; Wang et al., 2015). Such high viscosities should result in larger plate-mantle coupling that, in addition to lower viscosities of Phanerozoic lithosphere, is required to explain plate motions and lithospheric stresses (Ghosh et al., 2013). Such cratonic lithospheres are also cold (900° C on average, Vs = 4.7; see also Fig. 2c) and thick (250 km or more; see Fig. 2b) according to global average (Poudjom Djomani et al., 2001; Lee et al., 2001; Carlson et al., 2005).

In Figure 9, we show the compositional layering of the lithosphere of Europe inferred from data presented previously for both a stable Archean lithosphere of 3 Ga and a Phanerozoic lithosphere of 300 Ma (right panels in Figs. 9A and 9B). The depth-dependent rheology (viscosity) and temperature profiles calculated for these two lithospheres are also presented (left panels in Figs. 9A and 9B) (see Mouthereau et al., 2013 and Supplementary File S1 for calculation of the temperature profiles). Based on the continental geotherm calculated, we derive the corresponding yield stress profiles. A homogeneous brittle property of rocks is assumed for the upper part of the lithosphere (Byerlee’s law of rock failure) and the ductile-viscous term is represented by non-linear power law with three sets of material parameters that correspond to the properties of four lithological layers: upper crust (wet quartzite), lower crust (diabase), mantle and asthenosphere (olivine) (see Supplementary File S1 for the governing equations and rheological parameters material properties). The yield stress profile is then converted to an effective viscous profile to compare with viscosity values inferred from global models that include convection (e.g., Ghosh et al., 2013) using the relation ηeff=Δσϵ˙forumla where Δσ is the deviatoric stress and ϵ˙forumla the background strain rates fixed to 10−15 s−1.

The continental geotherm for the Archean lithosphere assumes a lithosphere thickness of 250 km. Note that thermal models of the continental lithosphere suggest that 400–700 Myrs are required for the thermal structure to become stationary (e.g., Burov and Diament, 1995; using conductive cooling for an equilibrium thermal thickness of 250 km). So we must consider the Archean cratonic lithosphere as thermally equilibrated. We predict a Moho temperature of ∼ 400 °C at 30 km (Fig. 9A) consistent with TMoho of 350–500 °C estimated based on modelling of heat flow in the East European Craton (Čermák and Bodri, 1986). The calculated mantle heat flow is 13 mW/m2 in the range of values derived from xenoliths (Kukkonen and Peltonen, 1999) for Fennoscandia, and more generally in agreement with heat flow studies showing that the mantle heat flux is 15 mW/m2 or lower for Archean cratons (Jaupart et al., 2015). The large viscosity of 1024 Pa.s estimated at 100–110 km (Figs. 9A and 9C) is consistent with geophysical constraints for the Archean-Proterozoic lithosphere of Baltica that is defined by a stiff lithosphere with high effective thickness larger than 80 km (Fig. 2) (Pérez-Gussinyé et al., 2004).

Baltica has been thinned on its margin by thermal erosion and addition of alkaline plume-related magmas at ca. 615 Ma. This event triggered rifting between Laurentia and Baltica, leading to the opening of the Iapetus (Tegner et al., 2019a) and Tornquist Sea. These refertilization events increase the density and weaken the cratonic lithospheric mantle by addition of water or other volatile elements and melts. The newly formed lithosphere on the edge of Baltica became prone to strain localization during subsequent tectonic stages. Plume-related magmatism provided the necessary conditions for subduction initiation and building of the Caledonian orogen. Together with Baltica margin, Avalonia is represented by a thinner and younger lithospheric mantle (Fig. 7). We infer it is formed by less depleted Proterozoic mantle represented by garnet-bearing lherzolite topped by spinel-bearing lherzolite. The thinner lithospheric mantle is assumed less viscous at its base, implying a weaker plate-mantle coupling for Avalonia compared to Baltica. The lack of crustal deformation however suggests that crust and mantle remained coupled during the Mesozoic and the Cenozoic, which is arguably related to low Moho temperatures of 400–500 °C (Fig. 9A).

Variscan and post-Variscan lithosphere: weak plate-mantle coupling and crust-mantle coupling

It is expected that the cumulative effect of tectonics, metamorphism and magmatism from 1–2 Ga to ca. 300 Ma and locally as young as 21–22 Ma resulted in considerable mechanical weakening of the lithosphere of Central Europe (Fig. 9B). The addition of volatiles during subduction, the incorporation of weak and hydrous minerals (e.g., plagioclase or serpentines) during Paleozoic and the Mesozoic rifting combined with the transient inputs of heat from the mantle together contributed to reduce the strength of the lithosphere. These successive events left a lithologically anisotropic crust and mantle with a thin layer of old and refractory mantle preserved in the uppermost mantle (Figs. 7 and 9B). Thermomagmatic thinning during the Early Jurassic CAMP magmatic event (Fig. 8), then from Late Cretaceous to Quaternary associated with alkaline volcanism from Europe to High Atlas, suggest a continuous rising of sub-alkaline melts in the hot asthenosphere. These heating events acted collectively to increase the mantle heat flux. We therefore anticipate the European lithosphere was not appreciably thermally relaxed after the Variscan orogeny (300 Ma) and remained weak until present, as shown by the low elastic thickness of about 20 km (Fig. 2d).

To provide a first-order picture of the 300 Ma-lithosphere rheology, we adopt in Figure 9B the same approach as for cratons to obtain the geotherm and the viscosity profile but for a lithosphere thickness of 100 km (Supplementary File S1), which corresponds to the average LAB depth deduced from geophysical constraints (Fig. 2b). This implies that the lithosphere did not change thickness over the past 300 Ma at the regional scale. Because this also means the lithosphere did not cool, the mantle heat flow must increase to adjust to the new boundary conditions. A greater output from the mantle is indeed expected due to the relatively recent tectonic and magmatic activity. We predict a present-day TMoho of about 650 °C, and mantle heat flow of 32 mW/m2. Such high temperatures agree with Moho temperature of ∼ 550–650 °C (Müntener et al., 2000) estimated in the Variscan basement during the Jurassic. There is also an abundant literature on mantle mylonites formed at 800–900 °C in the spinel and plagioclase field (30–80 km), including those found at Lherz, Ronda and Lanzo (Precigout et al., 2007). These results indicate a mobile mantle and a higher temperature gradients relative to Archean cratonic mantle. Our predicted temperatures in the uppermost mantle (30–80 km) vary between 650 and 1100 °C. They are in the range of temperatures defined in mantle mylonites that are thought to have developed during the Jurassic-Cretaceous rifting or earlier during the Permian. These data collectively lend support to the stationary thermal structure of the European lithosphere over the past 300 Myr.

One main implication of our modelling is the requirement of long-lived contrasting mantle heat flux between the Phanerozoic lithosphere of Europe and the adjacent cratons. In our view this reflects the contributions of 1) small-scale convection cells triggered by lateral temperature/buoyancy contrasts on craton margins similar to edge-driven convection (e.g., King and Anderson, 1998), 2) asthenospheric flow triggered by rift propagation (e.g., Mondy et al., 2017) during the Mesozoic (Atlantic, Tethys) combined with thermal insulation below Pangea at the origin of plume-related inputs (e.g., CAMP), and 3) slab-induced mantle flow for the most recent tectonic evolution of the region (e.g., Faccenna and Becker, 2010). In addition, the positive feedbacks between short scale convection cells and the whole mantle convection could have also contributed to produce the higher heat flow below Central Europe.

Based on inferences from global models of lithosphere-mantle coupling (Ghosh et al., 2013; Wang et al., 2019) and in accordance with our viscosity profiles (Fig. 9) this type of relatively young lithosphere appears low viscous (e.g., 1023 Pa.s) relative to cratons (1024 Pa.s) and weakly coupled to the low viscosity asthenospheric channel (1020 Pa.s). We infer that both buoyant cratonic lithospheres (WAC and cratonic core of Baltica, Karelia) are acting as lithospheric buttresses and strongly coupled to the asthenosphere. They build-up stresses and localize strain in regions formed by the lithosphere of peri-Gondwanan terranes that is hotter and weaker (Fig. 9C).

These results strengthen the view that the margins of cratons are long-lived zones of weakness that localize deformation during the Wilson cycle (Audet and Bürgmann, 2011). Here we further demonstrate that such weak lithospheres are made of a tectonically sheared, ductile and compositionally heterogenous mantle (Fig. 9B). Because of the weak coupling between crust and mantle, and their weak mantle roots, the Phanerozoic lithospheres preferably deform during compression by thick-skinned tectonics (Mouthereau et al., 2013). This explains the widespread occurrences of Variscan basement involvement in the external domains of mountain belts like in the Alps or the Pyrenees (Lacombe and Mouthereau, 2002; Mouthereau et al., 2013, 2014; Bellahsen et al., 2014; Jourdon et al., 2019).

lithospheric cross-section of Europe accounting for mantle composition and architecture

To provide a precise structural framework for crust and mantle evolution and architecture summarized in Figures 7 and 9, a lithosphere-scale geological section has been established (Figs. 10 and 11). This novel 4000 km-long structural cross-section stretches across Europe and North Africa. We have adopted interpretations from seismic reflection programs that imaged the crustal structure of Europe such as the IBERSEIS in Spain (Simancas et al., 2003; Carbonell et al., 2004), ECORS in France (Bois and Party, 1990), DEKORP in Germany (Heinrichs et al., 1994; Dekorp Basin Research Group, 1999), and international initiative like EUROPROBE and BABEL experiments in different countries across the Trans-European Suture Zone (TESZ) and the Baltic Sea (Babel Working Group, 1993). This was accompanied with more local seismic campaigns and geophysical acquisition that were integrated in geological interpretations labeled “a” to “m” in Figure 10. Moho depths and LAB depths are inferred from data presented in Figure 2. These regional sections were then projected onto our transect to form a composite lithospheric section at the scale of Europe. For instance, we have projected the ECORS seismic sections obtained across the Armorican Massif “i” and “j” and merged it with section “k” and “l” of the Rhenish Massif to connect with DEKORP section from the Harz Mountains and German Basin to the TESZ and Baltic Sea. Where offsets still persist after merging, our section was completed based on our own interpretations. By the addition of constraints on the laterally variable mantle depletion ages, crustal and lithosphere thicknesses, our section does not overlap, and rather complement, with earlier lithosphere scale sections published across the Mediterranean region in the TRANSMED Atlas (Cavazza et al., 2004).

Figure 10 first depicts lithosphere thinning below peri-Gondwanan (Cadomia) terranes documented by geophysical constraints (Fig. 2) and our synthesis of the tectonic and magmatic evolution (Fig. 7). Mantle thinning correlates with the depletion ages younger than 1 Ga, locally falling below 0.3 Ga, reflecting the significant thermal erosion and chemical overprint during the successive Cadomian subduction, Variscan subduction and Late Variscan-Permian post-Pangean heating event (see also Fig. 9). It should also be noted that the lithosphere thinning most affects lithospheres that were involved in the Variscan orogeny. Avalonia, in contrast, preserves an older and variably thicker mantle, about 100 km close to Armorica but is nearly 200 km when approaching the TESZ. Northern Africa depicts strong lithospheric attenuation across the transition from Anti-Atlas to High-Atlas. This limit seems to overlap with the position of CAMP and more recent Cenozoic magmatism. This is responsible for drastic mantle thinning from 250 km below the Reguibat Shield to less than 100 km below the High Atlas (Fig. 2b). These observations suggest an age- and depth-dependence on mantle thinning controlled by melt-enhanced weakening during successive subduction, extension and delamination. The Late Variscan dextral strike-slip faulting (Fig. 6) appears to have been preferentially localized in these weak domains leading to the segmentation of peri-Gondwanan terranes in the Atlas, Rif and the Betics, in Iberia (Iberian Range, Pyrenees) and further north along the northern Variscan front near the TESZ. The addition of plume-related volatile-enriched magma (CAMP) in the Early Jurassic enhanced or maintained this process of lithosphere thinning initiated in the Late Variscan. Magmatism was active until recently as indicated by Cenozoic magmatic from Morocco (“Moroccan Hot Line”) in the south, through Spain (Calatrava, Olot), Massif Central, up North associated with the Cenozoic volcanic province in the Germany (Vogelsberg, Eger Massif) and Bohemian Massif.

From a tectonic perspective, Figure 11 reveals that lithosphere deformation involves decoupling levels in the crustal basement and mantle. This is indicated by crustal thrust ramps rooting into the middle and lower crust and below in mantle ductile shear zones. The observation of vertically distributed strain between crust and mantle is expected for Phanerozoic lithosphere (Mouthereau et al., 2013) and is consistent with the weak crust-mantle coupling triggered by the thin and hot mantle lithosphere of Central Europe (Fig. 9).

Despite peri-Gondwanan terranes may be considered to first-order rheologically similar, orogenic deformation does not appear to be accommodated by a continuous mid- or lower crustal detachment. Instead the orogenic systems are discontinuous, reflecting additional weakening processes and strain localization in Mesozoic Rift domains (Atlas, Rif-Betics, Iberian Ranges, Pyrenees) possibly triggered by Late Variscan strike-slip faulting that accompanied the earliest stages of opening of the Tethys at the initiation of the Alpine cycle (e.g., Angrand et al., 2020). Crustal thinning and mantle exhumation during Alpine Tethys and Atlantic Rift evolution were also determinant. Among the important weakening processes is the transformation of the exhumed mantle in the plagioclase facies by melt-rock interactions (Chenin et al., 2018).

We have established in Sections 2 and 3 the relationships that exist between the structure of the lithosphere and its tectonic and magmatic evolution (e.g., Figs. 2 and 7). In addition, the present-day topography of Europe has been suggested to reflect the long-term evolution of the continental lithosphere. Indeed, low-elevation and low-relief regions (Fig. 4) correspond to the stable Archean craton characterized by low mantle heat flux and strong coupling between and crust and mantle (Fig. 9). On the opposite, the Phanerozoic lithosphere, involved in the Variscan and Alpine orogenic cycles, is defined by high heat flow and weak crust-mantle coupling (Fig. 9) and shows the highest elevations and reliefs (Fig. 4). In this section, our main objective is to decipher the topographic evolution of Europe during the Cenozoic in order to further discuss the relationships with the lithosphere architecture.

Strategy

Low-temperature thermochronology resolves cooling histories of upper crustal rocks from which exhumation in orogenic systems can be inferred. Although exhumation does not equal surface uplift, the rates at which exhumation/cooling occur may be used as a proxy for the timing of uplift in orogens. Here, we use the onset of acceleration of cooling rates to define the periods of topographic uplift. The apatite fission-track (AFT) thermochronometer that has closure temperature of 80–120 °C is less sensitive to deep tectonic processes and hence appears suitable to resolve cooling/erosion of rocks in the upper 3–4 km (assuming thermal gradients of 25–30 °C/km) and therefore uplift.

It must be reminded that cooling in orogens may be complex due to the architecture and the thermal regime inherited from the rifted margin thermal evolution, especially during the first stages of collision. In their coupled thermo-mechanical and thermochronological models of inverted rifted margin, Ternois et al. (2021) have distinguished two periods of cooling (Fig. 12). The first period corresponds to the onset of the tectonic inversion on the hyper-extended distal margin that is controlled by processes other than erosion, including post-rift thermal relaxation accentuated by the inversion-induced downward deflection of the isotherm. In their study, Mesalles et al. (2014) indeed demonstrated cooling of mid-crustal rocks in incipient submarine accretionary prism. In case post-rift cooling is compensated by heating caused by sediment burial, as recorded in the Pyrenees, the initial cooling is not observed. An isothermal cooling stage is recorded instead (Vacherat et al., 2014). Note that erosion caused by submarine current is documented in young emergent orogens like Taiwan and are associated with dynamic reorganization of the drainage system (Giletycz et al., 2015; Das et al., 2021) but these processes do not result in recordable exhumation by low-temperature data. In practice, we have searched to isolate the period of orogenic growth when topography is compensated by a crustal root and its evolution coupled with surface erosion, by determining the timing of first acceleration of cooling after plate convergence onset (i.e., 84 Ma). We selected sampling sites along our 4000 km-long transect so as to make a homogeneous dataset based on sites sharing a common thermochronometer, that is AFT (some sites also contains zircon fission-track and U-Th/He data), which T-t paths are computed using the same numerical models QTQt or HeFTy (Ketcham, 2005; Gallagher et al., 2009; Gallagher, 2012). Timing of first major cooling events and cooling rates were then calculated for each path (Fig. 13). Strong variations may be observed along a single orogenic segment due to model resolution and specific factor related to the sequence of thrusting. To account for these variations while keeping the main picture of the time-integrated orogenic evolution, we plot our results as boxplot graphs by calculating minimum, quartiles Q1(25%) and Q3(75%), median and maximum ages of maximum cooling rates (Fig. 13; see also Supplementary Material Figure S2). In practice, we have focused our study in regions of Variscan basement that present AFT ages younger than 80 Ma. We have excluded time-temperature histories that were obtained for regions too far from our transect, like the Alps, the Apennines and the Dinarides. Independent information on regional and local tectonics and sedimentary record have allowed to further identify and discuss the potential modulation of the cooling record by climate, fluid circulation and non-isostatic component on crustal cooling and exhumation. We first review the AFT ages record in each tectonic provinces of Europe (Figs. 14 and 15). Published time-temperature paths are then analysed to obtain the timing of increased erosion from which a topographic evolution can be defined.

Apatite fission-track data across Europe and North Africa

To examine the spatial and temporal topographic evolution of the European lithosphere, we have compiled 5440 apatite-fission track (AFT) ages from magmatic rocks or reset sedimentary rocks from Western Europe and Northern Africa (Fig. 14a and Supplementary File S2). The thermochronogical dataset builds on a previous global compilation (Herman et al., 2014), complemented by recent studies and a novel compilation made for the OROGEN project.

AFT ages range between 16 Ma (Q1 = 25%, first lower quartile) and 160 Ma (Q3 = 75%, third upper quartile), with the lowest value of 1.7 Ma recovered from gneiss of the Aar-Gotthard Massifs (Michalski and Soom, 1990) and upper bounds at 1027 Ma are found in gabbro of Fennoscandia shield from southern Finland (Kohn et al., 2009). To limit our discussion to the lithosphere tectono-thermal events we have interpolated AFT data within a 1000 km search radius (Fig. 14b) and contoured data according to geologically meaningful ages of 250, 150, 80 and 30 Ma.

Rocks carrying AFT ages older than or close to 250 Ma indicate they preserve Paleozoic and even earlier Precambrian cooling events. These old AFT ages are documented close to or within the cratonic areas of the WAC in eastern Reguibat Shield (Leprêtre et al., 2017) and the Baltic shield, in southern Sweden (Cederbom et al., 2000) or southern Finland (Kohn et al., 2009) (Figs. 14 and 15). They are also recognized in northern Scotland and northern Ireland (Green, 1989; McCulloch, 1993; Allen et al., 2002) of the Laurentian and Caledonian domains. Those ages are therefore found preferentially in regions of Europe defined by thick lithosphere and low thermal flux (see Figs. 2 and 9).

They reveal very slow cooling in continental domains that remained poorly affected by exhumation (below 3 km) or magmatism during to Mesozoic rifting and Cenozoic collision phases, but with some possible imprints by the Late Carboniferous-Permian magmatism (Fig. 8).

AFT dataset ranging between 250 Ma and 80 Ma indicates cooling histories related to periods of Mesozoic rifting in Western Europe (Figs. 14 and 15). We have recast our data relative to 150 Ma to allow distinction between the thermal evolution associated with Central Atlantic/Alpine Tethys rifting and the younger events related to the southern North Atlantic opening and Pyrenean rifting.

Rocks with ages between 250 Ma and 150 Ma are widespread in Europe. They are present from UK, Ireland, Southern Norway and Sweden to western France (e.g., Armorica Massif, Massif Central, Ardennes), Spain (Galicia, Central system) and in the domain extending between WAC and the High Atlas (Fig. 14). These ages reflect tectono-thermal events that temporally overlap with the timing of initial rifting in the north-central Atlantic and in the Alpine Tethys (Ligurian Tethys) (Fig. 15). The attribution of AFT ages to either of these two rift phases is challenging. In Morocco, for instance, the distinction between the different tectonic events may be particularly difficult as the Moroccan Atlas and Rif also recorded CAMP (201 Ma) magmatic event that overlaps with Late Permian-Early Triassic and the Early Jurassic rifting (Frizon de Lamotte et al., 2008; Marzoli et al., 2019), thus leading to controversial interpretations (Gimeno-Vives et al., 2019, 2020; Michard et al., 2020). This problem has also been raised in the Iberia peninsula based on zircon fission-track thermochronology (Rat et al., 2019).

Besides, the thermal imprint of the Alpine collision (Alps, Apennines, Dinarides) is so large that the pre-collisional cooling ages from the Tethyan paleomargin are generally not preserved. Some studies, however, have reported AFT ages consistent with the Alpine Tethys evolution, such as in Sardinia (Malusà et al., 2016) or Catalan Coastal Range (Juez-Larré and Andriessen, 2006). In summary, the distribution of AFT ages in the 250–150 Ma range shows affinity with regions surrounding Avalonia and peri-Gondwana terranes (Cadomia-Iberia, Armorica), and in particular with the Variscan Belt in France and Spain (Fig. 14). They are therefore correlated with thin lithospheric domains characterized by LAB shallower 150–100 km, higher mantle heat flow and a metasomatism mantle (Figs. 2, 7 and 9).

The domain comprising AFT ages between 150 and 80 Ma includes rocks that recorded cooling during a transitional period between the Early Cretaceous rifting and the onset of convergence between Africa and Europe (Figs. 14 and 15). They are found associated with: 1) the major Alpine orogenic systems in North Africa, western Iberia and Pyrenees; 2) the Massif Central (Peyaud et al., 2005; Barbarand et al., 2020), the Rhine graben (Timar-Geng et al., 2005; Dresmann et al., 2008), and are found spatially correlated with the mid-Cretaceous Vocontian and south Provence Basins (Graciansky and Lemoine, 1988) and the Helvetic Alps (Cardello and Mancktelow, 2014; Tavani et al., 2018); 3) Ireland and Central West Britain where these ages reveal cooling associated with the emplacement of the Icelandic plume and magmatism (Cogné et al., 2016).

AFT ages between 80 and 30 Ma are distributed in regions that recorded syn-orogenic exhumation during the AF/EU plate convergence. They are found in the High Atlas, Iberian Range, Pyrenees and Provence, and Corsica-Sardinia and the Alps (Fig. 14). Variscan Massifs preserve Late Cretaceous to Early Cenozoic AFT ages like the Black Forest and Vosges (Timar-Geng et al., 2005), Harz Mountains (Eynatten et al., 2019), Holy-Cross Mountains (Botor et al., 2018), Sudetes/Bohemian Massif (Botor et al., 2019), south Carpathians (Fügenschuh and Schmid, 2005) and eastern Carpathians (Gröger et al., 2008).

We have also distinguished orogenic domains with AFT ages younger than 30 Ma. This is a period of major tectonic reorganisation in western Mediterranean region that resulted in the formation of the Mediterranean arcs, including the Apennines, Betic-Rif, Maghrebides, Carpathians, southern Dinarides and external Hellenides controlled by the dynamics of retreating slabs (Royden and Faccenna, 2018). AFT younger than 30 Ma are found also in older collisional orogens like the Pyrenees, the High Atlas, the Alps and sporadically in the Cantabrians, the Iberian Central System. Northern Europe is not affected by this late cooling.

The distribution of syn-collision (80–30 Ma) to late-post-orogenic (< 30 Ma) AFT ages and to a lesser extent the ones in the 150–80 Ma range depict along our transect (Fig. 15) a good match with the thinnest, youngest and less depleted lithosphere (Figs. 1 and 2), stretching between the southern Massif Central and the Rif-Betic region. Outside of that domain (see also Figs. 6 and 14), the lithosphere of Avalonia north of the Variscan front (Fig. 15) shows a majority of old, pre-Alpine AFT ages. Below we focus on cooling and exhumation events in the region containing the syn-collision AFT ages that predominantly correlates with the peri-Gondwanan lithosphere.

Late Cretaceous-Paleogene cooling and exhumation in Harz Mountains, Sudetes and Vosges-Black Forest

Late and post-Variscan basement (Carboniferous, Permian) exposed in the Bohemian Massif and Harz Mountains are covered by thin and discontinuous Jurassic and Early Cretaceous rocks. After a period of erosion and weathering, subsidence resumed during the Late Cretaceous forming the Subhercynian Cretaceous Basin (Voigt et al., 2006) in Germany and the Bohemian Cretaceous Basin. In the Sudetes, low-temperature thermochronological data indicate that the Late Carboniferous-Permian basement underwent a burial of ca. 6.5 km during the Late Cretaceous followed by exhumation at 85–70 Ma (Danišík et al., 2012). This exhumation is associated with the Late Cretaceous-Paleogene far-field compression and intraplate tectonic inversion of Mesozoic Rift Basins recognized in central Europe (Mazur et al., 2005; Kley and Voigt, 2008). To precise the timing of cooling we have compiled time-temperature paths published from the Harz Mountains, Sudetes and Vosges Massif (Fig. 16).

Time-temperature paths in the Harz Mountains show cooling from 85–86 Ma contemporaneous with the deposition of syn-tectonic Coniacian-Santonian strata in the footwall of the inverted Harz Mountains (Voigt et al., 2006). The maximum cooling is recorded between 83 and 70 Ma at rates of 4–12 °C/Ma (Eynatten et al., 2019) equivalent to exhumation of 0.1–0.4 km/Ma (assuming thermal gradient of 30 °C/km).

Carboniferous gneiss dome from NE Bohemian Massif in the Sudetes recorded mid-Triassic to mid-Cretaceous erosion and burial in the Late Cretaceous that formed the Bohemian Basin and the intra-Sudetic Basin. Time-temperature paths indicate maximum cooling occurred between 58 and 55.5 Ma (1–6 °C/Myr or 0.03–0.2 km/Myrs) although exhumation started as early as 75 Ma (Sobczyk et al., 2020). Thermal modelling of AFT data from the intra-Sudetic Basin also documents initial cooling in the Late Cretaceous at 89–72 Ma (Botor et al., 2019) with maximum cooling between 60 and 50 Ma with 3–4 km of cumulated erosion since the Cenomanian. An earlier stage of exhumation at 85–75 Ma is indicated by a phase of rapid cooling from 120 °C in the northern Saxo-Bohemian Massif (Mid-German Crystalline rise) (Thomson and Zeh, 2000). The Lusatian block located north of the ENE-WSW-directed Eger graben also reveals an acceleration of exhumation between 95 and 70 Ma (Ventura et al., 2009). This Upper Cretaceous event slightly predates an early magmatic pulse at 68–59 Ma centered in the Eger graben (Ulrych et al., 2008), which is part of the Late Cretaceous to Cenozoic Central European Volcanic Province. The region then recorded the Eocene extensional phase and magmatism of the West-European Rift from ca. 42 Ma to 16 Ma (Dèzes et al., 2004; Ulrych et al., 2011). Volcanic activity has continued until the Quaternary in the Eifel volcanic field of the Rhenish Massif (Goes et al., 2000; Schmincke, 2007; Meier et al., 2016).

Cooling history of the Variscan basement in the Vosges and Black Forest was impacted by high geothermal gradients (Timar-Geng et al., 2005). During Late Jurassic, the temperature evolution of the Vosges and Black Forest Massif is interpreted to reflect hydrothermalism, uplift and tectonic inversion of Variscan structures (Dresmann et al., 2008). Two more phases of cooling dated to the Late Cretaceous-Paleogene and the Bartonian-Priabonian are also reported (Timar-Geng et al., 2005). The latter Late Eocene phase of cooling is interpreted to result from denudation associated with rift flank uplift during the formation of the Upper Rhine Rift (Danišík et al., 2010). An earlier episode of heating related to magmatism and initial subsidence is also reported during the Lutetian. Ages of maximum cooling from the first Late Cretaceous-Paleogene event are constrained to range between 67 and 54 Ma (Fig. 16). These ages are comparable with the earliest magmatic event dated to 60 Ma (Keller et al., 2002), that is attributed to rifting (Hinsken et al., 2007). Time-temperature paths show onset of cooling as early as 80–70 Ma, with some cooling paths indicating acceleration at 75 Ma (Fig. 16). Low-temperature constraints from the Ardennes Massif (Barbarand et al., 2018) document a Late Jurassic-Early Cretaceous exhumation event but Cenozoic exhumation remained very limited.

In our view, there is no significant difference in the timing of initial fast cooling recorded in the Harz Mountains, the Bohemian Massif and the Vosges Massif. Exhumation started at 80–70 Ma then increased in the Paleogene at ca. 55–50 Ma. The topographic fingerprint of the AF/EU convergence in Western-Central Europe likely reached a climax in the Paleocene-Eocene, that is slightly before, or synchronous with, the onset of extension and magmatism associated with the Central European Volcanic Province. A terminal Miocene (post-Burdigalian) topographic uplift in the Vosges-Black Forest is indicated by the occurrence of Burdigalian shallow-marine sediments of the Upper Marine Molasse (Giamboni et al., 2004; Hinsken et al., 2007). It temporally correlates with the establishment of the Kaiserstuhl volcanic complex at 18–16 Ma. At a larger scale it coincides with the Miocene plume-related volcanic event documented in Western Europe that is closely related to the previous Eocene Rift system.

Cooling histories in Massif Central, Pyrenees, Cameros, Sardinia

The Massif Central is characterized by contrasting crustal cooling histories that have been synthesized in a number of recent papers (Barbarand et al., 2001, 2013, 2020; François et al., 2020; Olivetti et al., 2020). These studies shows that while the western Massif Central (Limousin, Vendée, Rouerge) depicts AFT ages generally much older than 100 Ma, its eastern and southern margins (Morvan, Cévennes, Montagne Noire) are defined by ages younger than 100 Ma, often Late Cretaceous (Olivetti et al., 2020).

AFT data analyses from the Variscan basement exposed in the Morvan Massif, south of Paris Basin, reveal a Late Cretaceous-Paleocene exhumation event (65–50 Ma) (Barbarand et al., 2013). Time-temperature constraints from the eastern Massif Central (Olivetti et al., 2020) reflect an interval of maximum cooling ranging between 95 to 75 Ma at rates 3–5 °C/Myrs (0.1–0.15 km/Myrs), which is transitional between the mid-Cretaceous rifting and the onset of tectonic inversion of the margin during the Pyrenean-Provençal orogenic event (Fig. 17). Despite the Late Cretaceous exhumation event, topographic reconstruction on the southern margin of the Massif Central (Cévennes) argued for a post-34 Ma uplift based on preservation of marine sediments until Late Eocene (Lettéron et al., 2018; Olivetti et al., 2020) or Early Oligocene in the Velay region (Turland et al., 1994), which is not well resolved by AFT data (Barbarand et al., 2001). As for the Vosges-Black Forest Massif and other regions of the Central Europe Magmatic Province, magmatism and extension-related topography affected the Massif Central. Pre-extension magmatism occurred as early as 62–57 Ma, then increased during the Late Eocene-Lower Miocene rifting episode that formed the Limagne graben (37–22 Ma) and renewed mostly between 9–6 Ma and 3.5–0.5 Ma with the emplacement of the Chaîne des Puys (Michon and Merle, 2001). Plume-like volcanism has been suggested to result from shallow mantle upwelling triggered by decompressional melting pockets ahead and around Ionian retreating slab (Faccenna et al., 2010). Post-Miocene (post 13–11 Ma) surface uplift and erosion are documented in the Massif Central (Olivetti et al., 2016; Fauquette et al., 2020) likely associated with late-stage volcanism and ongoing plate convergence.

Along the transect, the Pyrenean Belt has been the target of numerous low-temperature thermochronology studies with only a few of them providing AFT-based time-temperatures paths in Variscan basement modeled with QTQt or HeFty (Bosch et al., 2016; Vacherat et al., 2016; Fillon et al., 2021). Cooling histories from the Axial Zone reveal a period of maximum exhumation between 44 and 26 Ma at rates of 10–30 °C/Myr (0.5–1 km/Myr), punctually reaching 100 °C/Myr (3 km/Myr), one of the fastest erosion rates measured along our transect (Fig. 17).

Different paleoaltitude estimates confirm that the topography of the Pyrenees formed in the Bartonian-Priabonian 41–34 Ma (Huyghe et al., 2012; Curry et al., 2019). The growth of orogenic topography therefore postdates a long-lasting phase of progressive accretion initiated at 70–75 Ma in Central Pyrenees (Mouthereau et al., 2014). Modelling of the inversion of a salt-rich hyper-extended margins reproduces well such long delays between the onset of collision and the rise of orogenic topography (Mouthereau et al., 2014; Jourdon et al., 2020a). Alternatively, the time delay could mirror the duration of the deformation event in the most distal part of the margin until the inversion of the necking and proximal domains occurs (e.g., Mouthereau et al., 2014; Mesalles et al., 2014; Vacherat et al., 2014; Ternois et al., 2021). A late post-orogenic uplift is well identified in the last 10 Myr synchronous with an uplift of 500 m detected in the eastern Pyrenees (Suc and Fauquette, 2012; Huyghe et al., 2020), and the late post-orogenic exhumation of about 1 km linked to reactivation of Mesozoic rift-related crustal architecture in the West (Fillon et al., 2021). But this post-10 Ma uplift seems to have been large-scale as inferred by post-tectonic piedmont erosion of nearly 2 km in the southern Pyrenees in the last 9 Ma (Fillon and Beek, 2012; Fillon et al., 2013) and based on ca. 600 m of fluvial incision of the Ebro Basin (Garcia-Castellanos and Larrasoaña, 2015). The drivers of this large-scale uplift are debated as they may include different ingredients such as asthenospheric temperature anomaly below Iberia (Conway-Jones et al., 2019), long-term dynamic support in eastern Pyrenees caused by lithosphere thinning and magmatism trapped in the lithosphere during the opening of Gulf of Lion (Huyghe et al., 2020). Another mechanism not specific to the Pyrenees may alternatively explain a dynamic support such as the subduction-related plume-like magmatism during Ionian slab retreat (Faccenna et al., 2010).

Time-temperature constraints in the Iberian Range are limited to the Cameros Basin because only the syn-rift Late Jurassic-Early Cretaceous sediments were sufficiently heated (up to 300 °C) for AFT data to record Cenozoic cooling (Río et al., 2009; Rat et al., 2019). Exhumation started at 60–50 Ma, and rates of exhumation reached a climax between 48.5 and 31.5 Ma (Fig. 17). These results are consistent with cooling history of the Pyrenees although maximum cooling rates of 2–12 °C/Ma (0.2–0.4 km/Myr) are noticeably lower. Movements along the Cameros main thrust may have lasted until the Tortonian. After ∼ 9 Ma post-tectonic erosion occurred (Rat et al., 2019). A larger-scale mantle contribution to uplift has been suggested for Iberia possibly explaining the post-30 Ma exhumation in mountain belt including the Iberian Ranges and the Pyrenees (Conway-Jones et al., 2019). The long-lasting shortening inferred from geological data also argues that crustal orogenic processes are at play, perhaps combined with lithospheric and sublithospheric processes.

We have also included results from Sardinia. Based on published cooling histories we infer that acceleration of exhumation occurred between 70 and 51 Ma at rates between 4 and 8 °C/Ma (0.1–0.2 km/Ma) (Fig. 17). This is consistent with a Late Paleocene to Middle Eocene phase of contraction recognized in Sardinia and Corsica (Dieni et al., 2008). These cooling ages further reveal that Sardinia preserves exhumation stages related to the onset of convergence and has not been affected by the rapid post-50 Ma exhumation documented in the Pyrenees. A counterclockwise rotation of Sardinia of 45° relative to Europe has been proposed between 50 and 30 Ma based on paleomagnetic data (Advokaat et al., 2014) that could explain this difference. Alternatively, the preservation of Sardinia from Pyrenean crustal thickening and exhumation could be related to the southern paleoposition of Sardinia at distance from the mid-Cretaceous Pyrenean Rift (e.g., Angrand and Mouthereau, 2021). Magmatism in Sardinia is dated from 38 to 12 Ma (Lustrino et al., 2009) and was contemporaneous with early extension in the Mediterranean Sea and opening of the Gulf of Lion.

Cooling histories in the Betics, High-Atlas and Anti-Atlas

The Betic-Rif arc formed in response to Neogene tectonic reorganization along the margins of Africa and Iberia imposed by west-directed delaminated slab retreat, tearing and slab detachment. AFT ages are much younger than the surrounding domains of North Africa (Atlas) and northward in the Iberian Range (Fig. 18). Thermal modelling of low-temperature data from the Betic-Rif reveals increased exhumation of the Variscan basement from 20 Ma (Sierra de Gador; (Janowski et al., 2017)), 15 Ma in the Rif (Romagny et al., 2014) and 6 Ma (Sierra Nevada; (Vázquez et al., 2011)) at rates up to 40–54 °C/km (equivalent to 1.3–1.8 km/Ma). This timing agrees with other geochronological constraints from higher temperature thermochronometers that reveal a high-temperature event ca. 20 Ma and rapid cooling of 200 °C/Ma (Platt et al., 2003). The Miocene tectonic event has overprinted older, subduction-related Late Eocene-Early Oligocene events (Monié and Chopin, 1991; Bessière et al., 2021). An older contractional event has been reported ca. 50 Ma based on thermal evolution of accreted Cretaceous flysch sediments (Daudet et al., 2020). A Paleocene-Eocene cooling event is also in good agreement with the formation of a Paleogene Foreland Basin onto the Iberian margin, reflecting the development of an orogenic topography. A widespread pre-Oligocene erosional surface is recognized in many places of the Western Mediterranean, including the Balearic promontory, Valencia Basin, Sardinia and Corsica. This large-scale uplift event and erosion is arguably related to the growth of pre-Alboran topography when the region recorded increasing shortening (see Sect. 5 below). This proto-Betic orogenic stage ended before the Burdigalian ca. 20 Ma which marks the onset of widespread extension and HT metamorphism in the region.

Post-20 Ma exhumation of the Betics reported from thermochronology is independently inferred based on the occurrence of Late Tortonian marine platform, 7.2 Ma in age, now at an elevation of 1600 m in the eastern Betics (Braga et al., 2003; Janowski et al., 2017). This late uplift is contemporaneous with the onset of contraction both onshore (Weijermars et al., 2007; Do Couto et al., 2014; Giaconia et al., 2014), and offshore in the Alboran domain (Martínez-García et al., 2017). Quaternary erosion rates of 0.14–0.4 km/Myr inferred from river terraces (Geach et al., 2015; Larrey et al., 2020) reveal that uplift is still ongoing and likely associated with active strike-faulting along the Trans-Alboran Shear Zone (Fig. 18). The Tortonian uplift is synchronous with slab detachment under the Iberia margin that triggered anorogenic alkaline magmatism (11–7 Ma and up to 18.5 Ma) (Duggen et al., 2004). This type of volcanism is well recorded in whole Western Europe and includes in Iberia the Valencia Trough (∼ 24–0.01 Ma), Calatrava Volcanic Province (∼ 9–2 Ma) and Olot–Garrotxa (∼ 10–0.01 Ma) in Catalonia. Older 16–29 Ma carbonate-rich alkaline magmas are also reported for Calatrava (Villaseca et al., 2018).

The High Atlas results from the inversion of a Late Permian-Early Triassic to Early Jurassic Rift. A first local contractional episode is documented in the Late Cretaceous (post-Turonian) (Guiraud et al., 2005) but orogenic topographic did not develop before the Bartonian-Priabonian (41–35 Ma) (Froitzheim et al., 1988; Frizon de Lamotte et al., 2008, 2009). Ages of 51 Ma and 36 Ma argue for increasing cooling/exhumation (Fig. 18). In addition to initial Eocene cooling, young dates of 8–5 Ma resolved by both AFT and apatite (U-Th)/He data reveal a second stage of exhumation. This second episode of uplift is contemporaneous with mid-Miocene to Pliocene alkaline magmatism (15–6 Ma) (Harmand and Cantagrel, 1984) related to thermal erosion of the lithosphere (Missenard et al., 2006). About the possible drivers, a combination of Canary plume-like magmatism and delaminated slab below Alboran has been proposed (Duggen et al., 2003, 2009). We note that the magmatic influx in northern Morocco also occurred at 35 Ma at Azrou in Middle Atlas (Raffone et al., 2009) and 44–33 Ma at Tamazert in High Atlas (Bouabdellah et al., 2010), which is temporally synchronous with the early magmas emplaced during the formation of the Western European Rift. This suggests that part of the older exhumation signal might be related to this older magmatism. Cooling rates are on average close to 2–3 °C/Ma (< 0.1 km/Ma) but can reach up to 15–20 °C/Ma (about 0.5 km/Ma).

AFT ages in the Anti-Atlas are often older than in the High Atlas. Thermal evolution of the western Anti-Atlas and Tarfaya Basin indicates a component of heating and cooling during the Atlantic rifting (Ruiz et al., 2011; Sehrt et al., 2017, 2018). To the east, AFT analyses by Malusà et al. (2007) supplemented by apatite (U-Th)/He data from Ruiz et al. (2011) allow to resolve a recent Neogene-Pliocene cooling event at 12 and 5 Ma. Cooling paths further support an onset of Paleogene cooling 65–55 Ma (Sehrt et al., 2017) but peaks in exhumation are not achieved before 49 Ma. Acceleration of exhumation occurred in Late Eocene-Early Oligocene contemporaneous with exhumation in the High Atlas. Cooling rates are 3–4 °C/Ma (about 0.1 km/Ma) throughout the Anti-Atlas indicating limited cumulated total exhumation of 1–3 km throughout the Late Cretaceous and the Cenozoic. The Reguibat Shield to the south shows Paleozoic to Jurassic and Cretaceous AFT ages. It has been suggested based on AHe data that this part of the WAC has experienced uplift and slow erosion related to Atlantic rifting and CAMP event (Leprêtre et al., 2014; Leprêtre et al., 2015).

Timeline of Cenozoic exhumation and tectonic events in peri-Gondwanan terranes

Figure 19 presents a synthesis of the exhumation events recognized in the young Phanerozoic lithosphere and their relationships with the Cenozoic tectonic and magmatic evolution of Western Europe and northern Africa. Geological data first confirm that contractional deformation spread rapidly after the onset of AF/EU plate convergence over Western and Central Europe and Northern Africa, between 80 and 60 Ma. One exception is the Massif Central, where the fastest exhumation occurred at 105–95 Ma. During mid-/Late Cretaceous the interactions of Europe with the east-moving Iberian blocks and northward drift of Adria may explain sporadic contraction and extension in the region (Angrand and Mouthereau, 2021).

One of the most striking results is depicted by the difference in the timing of fast exhumation between northern and southern Europe. We distinguish a north-eastern Europe domain, including Sardinia, defined by rapid exhumation synchronous with, or 10 Myrs after, onset of convergence and contraction. To the south-west, in a domain encompassing the Pyrenees (but excluding the eastern Pyrenees), Iberia and North Africa, fast exhumation is distinctively younger. In High Atlas and the Pyrenees, for instance, rapid exhumation starts from 50 Ma, which is ∼ 30 Myrs after the onset of contraction. The acceleration of Africa/Eurasia convergence is arguably one main driver of this large-scale evolution. Alternatively, the onset of fast exhumation may reveal the inversion of buoyant and thicker proximal domains of rifted margin combined with subaerial erosion as for the Pyrenees (Fig. 12; Jourdon et al., 2019; Ternois et al., 2021). The shortening of thin and presumably hot crust of hyper-extended rift domains that did not lead to subaerial topography is not recorded by low-temperature thermochronology data in the basement. This period has been resolved successfully by thermochronological analyses conducted on inverted Late Jurassic to Cretaceous Rift Basins (Whitchurch et al., 2011; Mouthereau et al., 2014; Vacherat et al., 2014; Rat et al., 2019; Daudet et al., 2020). Western and eastern domains can be tectonically distinguished during the Eocene-Oligocene time (Fig. 19). The eastern domain is affected by extension and magmatism related to the opening the Western European Rift, whereas in the western domain the signature of extension is limited to sporadic magmatic pulses. The contrasting evolution appears to be geographically coincident with the transition between Alps (Adria) and Pyrenees (Iberia) which separation was effective after 35–30 Ma.

We have shown in the previous sections that the architecture of the European lithosphere linked to the thermo-chemical evolution of its mantle exerts a first-order control on the topographic evolution of the collision between Africa and Europe (Figs. 14 and 15). In addition, exhumation patterns indicate this is the crustal architecture of the young, weak and mobile Phanerozoic lithosphere, inherited from episodes of rifting, that determine details of the topographic evolution during the earlier stages of convergence (Fig. 19). The latest stages of mountain building were however perturbed by the onset of the Western European Rift and the opening of the Mediterranean Sea. We have also demonstrated that plate-mantle coupling differs in Europe according to the thickness and the degree of depletion of the lithospheric mantle (Fig. 9). In this section, we explore the kinematic evolution of the AF/EU plate boundary from rifting to collision in order to discuss the respective roles of the lithosphere architecture and variations in plate-mantle coupling on the topographic evolution.

Current kinematics and stress patterns across Africa-Europe plate boundary: linking lithosphere evolution to plate-mantle coupling

The present-day absolute motions calculated in Western Europe from a No-Net-Rotation kinematic model, MORVEL56 (DeMets et al., 2010; Argus et al., 2011) show Africa (AF) or Nubia moving in the N40°E direction at a rate of 2.4–3 cm/yr and Eurasia (EU) in the N40-60°E direction at 2.1–2.3 cm/yr (Fig. 20). Using the same MORVEL model, inferred velocities of Africa relative to Europe illustrated in Figure 20a reveal the N-S convergence are the largest, up to 9.5 mm/yr, in the eastern Mediterranean, reducing to ∼ 5 mm/yr in western Mediterranean. Along the Hellenic trench, where the Mesozoic Neo-Tethys Ocean is subducting, the fast Nubia plate subducting velocities of 4 cm/yr (Serpelloni et al., 2013; Faccenna et al., 2014) produces deep earthquakes down to 200 km. In the western Mediterranean, seismogenic deformation is concentrated on continental margins in the Maghreb and in Betic-Rif regions, south of Iberia, associated with NNW-directed AF/EU velocities that progressively align westwards parallel to the Gloria–Azores transform zone, in the Atlantic.

GPS velocities further reveal Anatolia-Aegea, Adria and Alboran, positioned along the Africa(Nubia)-Eurasia plate boundary, are moving independently from Africa (Fig. 20b). Surface movements of Adria and Anatolia-Aegea microplates in particular are well reproduced by a combination of plate convergence and uppermost mantle flow model, emphasizing the role of slab-induced flow in addition to long-wavelength mantle density anomalies (Faccenna and Becker, 2010). West-directed movements of Alboran Basin relative to Africa and Iberia indicate extrusion of an Alboran block (Koulali et al., 2011; Nocquet, 2012; Palano et al., 2015), along strike-slip shear domains in the Betics and the Trans-Alboran Shear Zone. Away from the plate boundary zone that accumulates strain and generates large earthquakes on tectonically-loaded faults, there is a much wider domain encompassing most of Europe, characterized by no significant GPS displacements (below 0.2 mm/yr) and devoid of earthquakes. This large domain defines a relatively stable Europe continental interior, where tectonic activity is not ruled by strain accumulation on faults and rapid stress increase (Calais et al., 2016). Current crustal maximum horizontal stress (σHmax) from the WSM2016 database shows a homogeneous compressional pattern oriented NNW-SSE to NW-SE (Fig. 20c), N145°E on average, parallel to shortening defined by stress inversion of focal mechanisms and GPS data (Heidbach et al., 2018; Martínez-Garzón et al., 2019). They are in good agreement with the relative plate AF/EU motion, and GPS data but at 90° to the APM (Fig. 20a). For comparison, stresses calculated from dynamic models combining plate convection and gravitational potential energy (GPE) calculated from lateral density variations and lithosphere thickness (Ghosh et al., 2013) (see also Fig. 9) define a most compressional stress oriented parallel to σHmax (Fig. 20c). It has been inferred that the European lithosphere can support statically significant deviatoric stresses over long geological intervals independent of transient stress loading/unloading at AF/EU plate boundary (Calais et al., 2016). These results support the fact this is the coupling between age-dependent properties (composition and rheology) of the continental lithosphere (Fig. 9) and the vigor of convection that modulates stress and strain distribution in collision zones.

Kinematic reconstruction across the Africa-Europe plate boundary

new reconstruction accounting for Iberia plate fragmentation

The kinematic reconstruction of the Africa-Eurasia plate boundary in the Mesozoic-Cenozoic period is debated, especially regarding 1) the evolution of fragmented Adria and Iberia microcontinents, which moved independently from Africa, and 2) the nature, size and lateral extent of the Alpine Tethys between Africa, Iberia, Adria and Europe (Handy et al., 2010, 2015; Le Breton et al., 2017; van Hinsbergen et al., 2019a, 2019b; Müller et al., 2019; Angrand et al., 2020; Romagny et al., 2020). We present below the evolution of the Africa-Eurasia convergence based on the reconstructions of Angrand et al. (2020) and Angrand and Mouthereau (2021). This reconstruction has several specific features with respect to previous kinematic models: 1) the geometry of Iberia is revised and now involves the Ebro, West Iberia and South Iberia blocks, moving separately between Europe and Africa since the Jurassic, 2) a reviewed chronology and kinematics, including the Paleogene “Pyrenean” collision onset in the Betics and the timing of intraplate deformation in the Iberian Ranges between South Iberia and Ebro blocks. The solution for Apulia and Africa is based on the model of Müller et al. (2019) that implements a reconstruction of Le Breton et al. (2017), and we adopted the solution of van Hinsbergen et al. (2019a, 2019b) for Eastern Europe. Along our transect the total AF/EU plate convergence to be accommodated is about 300 km. The reconstruction of Alboran (Internal Betics) accounts for maximum 150 km of shortening estimated from section balancing in the past 20 Ma across the Africa-Iberia plate boundary that overall agree with stratigraphic, structural and thermochronological constraints, indicating Alboran should be close to the Iberia possibly since the Paleogene (Daudet et al., 2020; Pedrera et al., 2020a, 2020b). Accordingly, we have restored Alboran 150 km to the south-south-east of its present position at 20 Ma, and about 240 km at 60 Ma (Angrand and Mouthereau, 2021), which amount falls in the limit permitted by convergence estimated between South Iberia and Africa (Macchiavelli et al., 2017). We infer from these constraints that the westward movement of the Gibraltar Arc should be maximum 260 km since the Eocene. Larger displacement of 400 km has been inferred in another recent reconstruction based the reconstruction of the original thickness of the Betic orogenic wedge (Romagny et al., 2020). In any case, the westward motion appears significantly smaller than the 600 km-long Alboran slab imaged beneath the Gibraltar arc (Bezada et al., 2013; Mancilla et al., 2015b; Villasenor et al., 2015; Palomeras et al., 2017), which we suggest reflects backarc extension by retreating delamination from below Alboran terrane.

In this reconstruction Ebro forms with Sardinia-Corsica a continuous continental block tectonically distinct from Iberia and Europe, individualized along Early Carboniferous-Late Permian right-lateral shear zones. During the Late Jurassic-Early Cretaceous, the eastward drift of Iberia, Ebro-Corsica-Sardinia, and other smaller terranes (AlKaPeCa) is accommodated in a broad zone of left-lateral transtensional deformation. The fact that oblique extension is only slightly partitioned between strike-slip and pure extension explains our difficulty to identify unambiguously the strike-slip movement in the Pyrenees and Iberian Ranges. The presence of Triassic evaporites that decouple basement deformation from the sedimentary cover, further argues for the apparent lack of evidence of strike-slip faulting. When accounting for Adria motions, the opening of Alpine Tethys during the Lower Jurassic is seen to be characterized by oblique extension, between the Atlantic and the Neo-Tethys, and did not resulted in a large oceanic basin. Being segmented and floored only partially by oceanic crust, we interpret the Alpine Tethys as the result of the evolution of two rift segments. The western rift segment (Betic Ocean/West Ligurian Ocean) is positioned to the east of the Iberia paleomargin and extends between the Betics and the eastern Pyrenees. The eastern branch (East Ligurian Basin/eastern Maghrebian Basin) was the largest oceanized domain between Adria/Africa (Sicilia) later localizing retreating subduction zones on the western border of Adria.

From 35 Ma onward, the kinematics of the West Mediterranean is based on Romagny et al. (2020) with the noticeable exception for Alboran that we have placed farther west at 35 Ma. The Kabylides is interpreted to have moved southward with respect to Alboran in the Early Miocene. We infer that by the Late Eocene all Mesozoic oceanic domains and rifted margins were closed, resulting in an orogenic belt that will then be reworked during the opening of the West Mediterranean Sea. As convergence continued slab retreated toward the Ionian Basin, triggering the opening of the Ligurian-Provençal Basin and Gulf of Lion. In this model, the formation of the Alboran Sea is suggested to have occurred by subduction and retreating delamination.

Patterns of Africa/Eurasia kinematics and driving mechanisms

Pre-collision kinematic evolution

The main quantitative results of our kinematic model are summarized in Figures 21 and 22 for Africa, Adria, and peri-Iberian blocks in both fixed Europe reference frame and mantle reference frame. The direction of absolute plate motion when integrated over the Mesozoic-Cenozoic shows a primary N-S motion for both Europe and Africa (Fig. 21). The continental drift of Africa reveals control by upwelling of buoyant mantle from the African LLSVP (Large Low Shear Velocity Province) region associated with LIP eruptions (e.g., Afar, CAMP, Etendeka, Karoo, Sierra Leone; see numbers labeled in Fig. 22d) and hot spots (Torsvik et al., 2006; Burke et al., 2008). The African LLSVP is also proven to be the main driver of the dynamic support of Africa (Moucha and Forte, 2011; Moore et al., 2017). In association with mantle downwellings in subduction zones they form the convection conveyor belt driving the movement of Africa (Becker and Faccenna, 2011).

The opening of Central Atlantic resulted in SSE-to-E-directed motion of Africa, with respect to Europe, during the Early Jurassic (200–150 Ma). The migration of oceanic spreading into the southern North Atlantic during the Late Jurassic-Early Cretaceous (150–120 Ma) triggered the eastward drift of Iberia (Figs. 21 and 22). Between Africa and Europe, the movement of south Iberia occurred faster than, and oblique to, Adria after 140 Ma (Angrand et al., 2020; Angrand and Mouthereau, 2021) which led to moderate contraction in the Alpine Tethys. The Bay of Biscay eventually opened in the Late Aptian (118 Ma), reactivating Jurassic Basins and fragmenting Iberia along inherited post-Variscan shear zones in the Pyrenees and Iberian Ranges. From 130–120 Ma, the absolute motion of Europe, Africa, Adria and other peri-Gondwanan terranes (Ebro, S-Iberia and W-Iberia) increased (Figs. 21 and 22). This is coincident with the acceleration of continental rifting in South Atlantic, after emplacement of the Etendeka igneous province (labeled 3 in Fig. 22d), a precursor event of oceanic spreading that weakened the southern Africa plate boundary (Brune et al., 2016).

About this time, the Iberian blocks (Ebro and South Iberia) moved in the ESE direction (∼ 120°E) at maximum rates of 1.5 cm/yr, while Adria positioned along the same Euler latitude but moving faster had a North-directed (N60°E) motion. This kinematics resulted in the closure of the eastern branch of the Alpine Tethys (Angrand and Mouthereau, 2021), and to the North of Adria, the subduction in the Alps (Eo-Alpine event) (Handy et al., 2010; van Hinsbergen et al., 2019a, 2019b). Differential movements between Iberia and Adria at the northern edge of Africa seems to have governed the complex pattern of deformation in the Mediterranean, including episodic extension and contraction during the middle-Late Cretaceous. Velocities of Africa relative to Europe reached a maximum during the 120–80 Ma interval, in part due to a combination of mantle upwelling, active spreading in the southern Atlantic and slab pull along Tethyan subduction zones. Between 110 and 100 Ma absolute velocities of Africa and Adria appear to have slightly decreased (Fig. 22) consistent with the slowing of plate velocities globally (Müller et al., 2019). Africa, Adria and Iberia then recorded in less than 20 Myr (100–80 Ma) an acceleration from ∼ 3 cm/yr to above 6 cm/yr (North Africa) (Fig. 22).

Onset of collision: weak Western Europe caught in between two cratons

Both AF/EU convergence and absolute velocities decreased at 80–65 Ma as the direction of convergence changed to N-S orthogonal to plate boundary (Figs. 21 and 22). In Europe and northern Africa, the Late Cretaceous stage marks a transition from a tectonic regime moderately compressional around Adria to broadly contractional regime from East (Adria) to West (Iberia), and from Morocco in the south to northern Europe along the TESZ (Kley and Voigt, 2008). The reduction of Europe motion resulted in the increase of AF/EU convergence. Several mechanisms have been proposed to explain the Late Cretaceous (Santonian, 84 Ma) rapid slowing. The emplacement of Morondava LIP at 92–84 Ma (Torsvik et al., 2000) between Madagascar and Seychelles would have induced the acceleration of the northward drift of Africa (labeled 4 in Fig. 22d) followed by collision with Europe and reduction of Africa motion. Such acceleration is also proposed to have induced Tethyan slab penetration into the denser lower mantle, in turn explaining the emplacement of several south Tethyan ophiolites between 100 and 75 Ma (Jolivet et al., 2016). Alternatively, the emplacement of south Tethyan ophiolites could result from subduction initiation by pre-Deccan plume at about 100 Ma, hence unrelated to Neotethyan subduction dynamics to the north (Rodriguez et al., 2021). In our view, the rapid spread of contraction over 4000 km is an indication of the first-order role played by the strong plate-mantle coupling below the viscous cratonic lithosphere of WAC and EEC (Fig. 23a). The arrival of the strong, buoyant and thick cratonic lithosphere of African/Gondwana against the strong Baltica craton, closed the Alpine Tethys and resulted in the built-up of resistive forces, thus limiting further indentation of Africa into the weak lithosphere of Europe (Fig. 23a).

Paleogene kinematic reorganisation: increasing role of North Atlantic

A general decrease of absolute velocities and clockwise rotation of 20–30° is recorded at 62–40 Ma for Europe, and at 57–35 Ma for North Africa (Fig. 22). This is combined with the onset of WNW-directed movement of Iberia over the 57–42 Ma time interval. The decrease of Eurasia velocity coupled to the less important reduction of Adria and North Africa velocities appear responsible for a second period of acceleration of AF/EU plate convergence at 55–50 Ma. This resulted in the Eocene phase of mountain building documented throughout Iberia all the way to the Atlas (Fig. 19). The slowing of Europe and Africa motions during the Paleocene-Eocene requires large-scale processes likely combining the effects of increasing gravitational potential in the Africa-Eurasia collision zone coupled to the development of 1) spreading ridges in the Labrador Sea at about 62 Ma (Abdelmalak et al., 2019), North Atlantic at 54 Ma (Mosar et al., 2002) and the rise of icelandic plume of the North Atlantic Igneous Province (Torsvik et al., 2014) (Fig. 23), 2) slowing of convection cell below Africa at about 65 Ma linked to the emplacement of the Deccan LIP by Réunion plume (labeled 5 in Fig. 22d; van Hinsbergen et al., 2011; Cande and Stegman, 2012), and 3) the penetration of Tethyan slabs in the lower mantle at 65–55 Ma (Faccenna et al., 2013a).

This episode was followed by the increase of AF/EU convergence until 32 Ma in eastern Africa, 43 Ma in Adria and 50–40 Ma in northern Africa (Fig. 22a). For Iberia (South Iberia, Ebro blocks), in contrast, velocities relative to Eurasia dropped below 0.5 cm/yr after 40 Ma (Fig. 22b). Velocity reduction is connected to the progressive accretion of Iberian blocks to Europe after 40 Ma. After 35 Ma, the whole of AF/EU convergence dropped to 0.9–0.5 cm/yr and was maintained at this level or with a slight increase from 18–15 Ma to present (Fig. 22a). This period was coeval with back-arc extension in the Mediterranean region, corresponding to the Mediterranean Tectonics stage spanning from 35 to 8 Ma (Jolivet et al., 2021), which also coincides with the counterclockwise rotation of Africa convergence. Remnant of Africa convergence to be accommodated in Iberia was transferred to the south in the Betics. We note that the model of Müller et al. (2019) adopted in our reconstruction and that of van Hinsbergen et al. (2019a, 2019b) shows relatively similar AF/EU velocities evolution in the past-30 Ma (Fig. 22a) but they noticeably differ in Africa absolute velocities (Fig. 22d). The slowing of Africa has been suggested to be synchronous with the initiation of continental rifting along the Red Sea at 25 Ma (McQuarrie et al., 2003; Reilinger and McClusky, 2011) triggered by the arrival of Afar plume (Fig. 23) that emplaced from 45 to 30 Ma (Faccenna et al., 2013b). The breakup of Arabia disconnected Africa from cold spots in the Tethyan subduction and led to progressive kinematic reorganisation of the region, including the opening of Gulf of Aden (Leroy et al., 2004; ArRajehi et al., 2010) and collision in the Zagros (Mouthereau et al., 2012). The complex post-35 Ma kinematic evolution in the east was related in Europe to the reactivation of Late Cretaceous-Paleogene collision belts e.g., in the Betics and the Alps caused by slab retreat and delamination in the Mediterranean (Fig. 23).

Long-term thermal and rheological stability of the lithosphere and topography

The orogenic evolution of Western Europe cannot be understood without considering the profound and long-lived transformation of the continental lithosphere initiated at 750–500 Ma. During Rodinia assembly at 1.2 Ga, the protagonists of the Africa-Eurasia collision were part of resistant cratonic domains of Gondwana (West Africa Craton) and Baltica. Thermal erosion, thinning and refertilization by alkaline melts of the Archean SCLM during plume emplacement at about 600 Ma, Pan-African orogeny/Cadomian subduction and back-arc rifting at 570 Ma removed the Archean cratonic root from northern Gondwana (Fig. 24). The strength of the continental lithosphere of the newly formed peri-Gondwana terranes, comprising Avalonia and Armorica, was drastically reduced compared to remnants of Archean cratons. We suggest that Early Paleozoic Europe and northern Africa were in a thermal and mechanical configuration relatively similar to the one described at present (Fig. 9). This implies that the peri-Gondwanan lithosphere of Europe and North Africa have not been thermally relaxed for several hundred million years. Also required is the limited convergence, here estimated to 300 km along our transect of Western Europe, between the pre-Gondwana terranes and the adjacent cold and buoyant cratons in order to maintain these conditions. The nature of the subsequent tectono-magmatic events shows these assumptions are correct.

First, partial melting and magmatism during the Variscan orogeny argue for a combination of high mantle heat flux and high crustal heat flux related to the young radiogenic plutons (e.g., Vanderhaeghe et al., 2020). There are also indications that the Variscan Belt did not form an orogenic Tibet-like plateau (Franke, 2014), which would require a mechanically strong cratonic lithosphere to maintain its high topography, as for the current India-Eurasia collision.

Late Carboniferous elevation of ∼ 3 km estimated for the Variscan Belt (Massif Central) (Dusséaux et al., 2021) and evidence of the lack of marked rain shadow that suggests moderate elevation of 2000 m during Late Carboniferous-Permian (Roscher and Schneider, 2006) supports this model. We have estimated that the mantle heat flow established during the Late Carboniferous-Permian (300–270 Ma) was maintained to high values of 32 mW/m2 (Fig. 9). Thermal erosion and delamination are indicated at this time by calc-alkaline magmatism characterized by an increase of mantle source in the lava flows. This also reflects the transition towards increasing juvenile asthenospheric inputs related to the propagating Neo-Tethys ridge (Fig. 25) to the south of the closing Paleo-Tethys Ocean (e.g., Angrand et al., 2020). We have suggested that edge-driven convection induced by density contrasts with craton margins and the asthenospheric flow associated with rift propagation (Mondy et al., 2017; Jourdon et al., 2020b) are the principal mechanisms to explain a stable mantle heat flux. The lack of thermal relaxation and lithosphere thermal thickening following the Late Carboniferous-Permian event is also suggested by the lack of marine incursion until the Triassic (Bourquin et al., 2011). The Early Jurassic CAMP event recognized in the northern margin of West African craton, Iberia and Brittany reveals high temperature in the asthenospheric mantle caused by deeply rooted plume, thermal insulation below Pangea or a combination of rifting and plume magmatism. Subsequent lithosphere thinning and inputs of heat from the mantle during the episodes of Jurassic-Early Cretaceous Atlantic and Alpine Tethys rifting contributed to keep the Europe lithosphere weak. The fact that these rifting events were essentially magma-poor is not an indication of a lower mantle heat flux. This reveals instead that rifting occurred in a depleted Permian mantle. There are also indications that the mid-Cretaceous, especially the Albian, was a period of topographic uplift, weathering and exhumation as revealed on the formation of paleosoils and karsts in Massif Central (e.g., Marchand et al., 2021), and possibly over most of the Variscan massif (Thiry et al., 2006). The predominance of mature nearshore to continental sands deposited in Iberia (Rat, 1988; Rat et al., 2019) and northern Africa (Lefranc and Guiraud, 1990) document the occurrence of large continental surfaces exposed to subaerial erosion during the mid-Cretaceous. In addition, the large majority of AFT ages over peri-Gondwanan domains are essentially younger than 300 Ma with 75% younger than 160 Ma. Those post-Variscan but pre-Alpine AFT ages suggest transient inputs of heat and burial caused by the kinematics of rifting during the Mesozoic. Taken together these data argue for a model in which the topography of Central Europe remained in part dynamically supported during most of the Mesozoic in line with the first-order long-term stability of thermal structure and strength of the lithosphere, at least over the past 300 Ma. This point is essential because this is the contrast in lithosphere strengths, between cratonic lithosphere and the weak Europe, and the degree of coupling with large-scale convection that explain far-field stresses and deformation in the collision zone.

Impact of rift maturity on topographic evolution and late collision evolution

Soon after the onset of Africa-Europe collision, in the Santonian, contraction spreads rapidly over the whole Western Europe and northern Africa margin (Fig. 26). We have suggested that this is the far-field stresses exerted by the convergence of buoyant, thick and strong cratonic lithospheres of Africa and Baltica that explain the strain distribution in the weak European lithosphere. Secondary effects related to the high strength of hyper-extended rift domains at the initiation of inversion in the Pyrenees (Dielforder et al., 2019), the Betics and possibly in many other parts of the Mediterranean region further facilitated strain transfer throughout Europe. Crustal domains that were uplifted first (Atlas, Harz, Vosges, Sudetes) are also those with lower initial rift-related extension during the Mesozoic (Fig. 26). This transient Laramide-like stage of contraction in central Europe can therefore reflect the initial rift architecture. In contrast, extension associated with rifting along the Atlantic-Alpine Tethys segment led to much larger hyper-extended rift systems and oceanic basins in the Betic-Rif and in the Pyrenees. In these regions the rifted margins were also associated with thick pre-rift salt layer (Flinch et al., 1996; Carola et al., 2015; Flinch and Soto, 2017; Ducoux et al., 2019; Daudet et al., 2020; Izquierdo-Llavall et al., 2020; Jourdon et al., 2020a, 2020b; Pedrera et al., 2020a, 2020b; Roca et al., 2020). As expected, the onset of inversion in these domains did not result in subaerial topography as observed in Central Europe and were still largely submerged (Fig. 26). At 50 Ma, a major kinematic change occurred corresponding to accrued Africa/Eurasia convergence driven by plume activity in east Africa and below Iceland, and the opening of the North Atlantic. This led to closure of hyper-extended domains in the Pyrenees and the Betics that resulted in the onset of accretion of necking domains marked by rapid growth of orogenic topography. This is recorded by the increasing subaerial erosion and well-developed foreland basins. The change in the internal dynamics of orogenic systems appears synchronous with the opening of West European Rift (WER). From 40 Ma, our reconstruction indicates that Iberia and Ebro block were accreted to Europe and the proto-Betic domain was topographically mature in the south and coupled to mountain building in the Atlas (Fig. 26). By the Late-Eocene-Oligocene, the WER separates two distinct tectonic domains. In the West, the accretion of Iberia and Ebro blocks resulted in orogenic belts from the High Atlas to the Pyrenees. In the East, however, around Adria, the Ionian Basin and the European margin in the Alps are not yet closed (Fig. 27). From 35–30 Ma, the retreat of subduction front towards the East along the active plate boundary with the Ionian Basin opened the Western Mediterranean in the former mature “Pyrenean” orogenic domain (Fig. 27). However, considering the size of the Ionian Basin and the fact the Alpine Tethys region was lacking a large oceanic domain, it must be envisaged that the WER might have been the trigger of the West Mediterranean opening. In this reconstruction (Fig. 26) this is the size and the architecture of the rifted domains that dictate the timing and the vigor of exhumation in the collision zone. As rifted basins were progressively inverted to form mature orogenic belts, this is the motion of Iberia and Adria independently from Africa and large-scale lithospheric heterogeneities that controlled the post-collision evolution of in Europe.

In this paper our goal was to resolve how inherited properties of the continental lithosphere explain the patterns of mountain building in Europe. In this aim we have synthesized and analyzed constraints on the tectonic and thermal evolution of the Europe lithosphere, plate kinematics and temporal evolution of the orogenic topography. We show that the thermal and mechanical conditions necessary to form orogenic belts of Western Europe lie in the perennial contrast between a weak, thin and dense lithosphere of Central Europe made of a refertilized mantle and the strong, thick and buoyant cratonic depleted mantle of East European Craton and West African Craton. A pivotal trigger of mantle weakening is the thermal erosion and refertilization event of the continental lithosphere during the Neoproterozoic (750–500 Ma) when Western Europe lithosphere was positioned north of Gondwana and its evolution controlled by the subduction of Iapetus. The second key event was the Late Carboniferous and Permian (310–270 Ma) magmatic event as Pangea broke up and got heated at the end of the Variscan orogen. To maintain the contrasting rheologies between the Phanerozoic and Archean lithosphere of Europe, thermal relaxation beneath the Phanerozoic lithosphere must be compensated by an increase of mantle heat flux. We find that edge-driven convection on craton margins and the asthenospheric flow below Central Europe associated with Tethyan rifting are the main drivers. These events are suggested to have resulted in anomalous topography over Europe. Stress built-up and strain distribution during AF/EU collision are in turn explained by the different coupling with convection in the asthenosphere between the Phanerozoic and Archean cratonic lithospheres. The patterns of exhumation across Western Europe are inferred to be linked to the crustal architecture inherited from the evolution of rifted margins associated with the Neo-Tethys, Atlantic and Alpine Tethys opening. The marked topographic change detected at around 50 Ma, from the High Atlas to the Pyrenees, emphasizes the acceleration of orogenic exhumation in southern Europe synchronous with onset of extension in the West European Rift. We show this period of widespread orogenic growth was caused by the onset of accretion of more buoyant and thicker portions of rifted margins triggered by the kinematic reorganization related to opening of the North Atlantic and onset of the Icelandic plume.

This paper is a contribution of Orogen, a tripartite project funded by TOTAL, BRGM and CNRS. It benefited from discussions along the course of the project with the Orogen Community. Alexandra Robert is thanked for her help in handling with lithosphere thickness models. We are also grateful to Olivier Vanderhaeghe, Yoann Denèle and Michel Grégoire for discussions on an earlier version of the manuscript and their feedbacks regarding the evolution of the Variscan orogen and interpretation of xenoliths data. We gratefully acknowledge Laetitia Le Pourhiet and an anonymous reviewer for their careful reading, questions and comments that significantly improved the quality of the manuscript. Oliver Lacombe is thanked for his editorial handling and suggestions of improvements.

Cite this article as: Mouthereau F, Angrand P, Jourdon A, Ternois S, Fillon C, Calassou S, Chevrot S, Ford M, Jolivet L, Manatschal G, Masini E, Thinon I, Vidal O, Baudin T. 2021. Cenozoic mountain building and topographic evolution in Western Europe: impact of billions of years of lithosphere evolution and plate kinematics, BSGF - Earth Sciences Bulletin 192: 56.

Supplementary data