© G. Manatschal et al., Published by EDP Sciences 2021

1 Introduction

Writing a paper on the “role of inheritance” in forming and reactivating rifts and rifted margins and building collisional orogens may appear ambitious if one considers the disagreements and debates that exist on the descriptions of rifts, rifted margins and orogens themselves. Therefore, if we want to think about the legacy of a “rift” in an orogenic system, or of a former orogen during the formation of the rift, we need a first order understanding of rifts, rifted margins and orogens and we need to define what we mean with “inheritance”. Here we revisit the concept of “inheritance” by distinguishing between “interface shape inheritance”, which includes transient thermal state and gravitational potential energy, and “persisting inheritance”, which encompasses long-lasting structural and compositional inheritance. This new approach allows us to define, at each stage of a Wilson Cycle, the interplay between inheritance (innate/“genetic code”) and the physical processes at play (extension/compression, magmatism, etc.). This paper aims to show that understanding “inheritance” is not an option but a prerequisite for comprehending early stages of tectonic systems. Another, related question is how can “inheritance” be implemented in our thinking of tectonic systems? Since relatively simple numerical models can reproduce structures and architectures similar to those observed at mature rifted margins or orogenic systems, two questions arise: can natural systems be described by using only simple/generic initial conditions? And, if not, when, where and how do inherited structural, thermal and compositional complexities control the evolution of tectonic systems in a Wilson Cycle?

In order to answer to these fundamental questions, it is necessary to obtain observational constraints on how inheritance may control the evolution of tectonic systems. Here we use observations from the Biscay-Pyrenean system to investigate the role of inheritance throughout an incomplete Wilson Cycle that includes a post-Variscan collapse, a Mesozoic rifting, and an Alpine reactivation. Although the Biscay-Pyrenean system represents an incomplete Wilson Cycle and may therefore not be representative of and/or comparable with global collisional orogens, it represents a world-class natural laboratory to study the impact of rift inheritance in the early stages of convergence (e.g., nucleation of a convergent plate boundary) and building of collisional orogens.

In this paper, we commence with a historical overview of the role of inheritance in tectonic systems and summarize the state of the art on the subject. We redefine some terms and present some existing and new concepts that will be tested, improved and further discussed in following parts of the paper. We then discuss the importance of rift-inherited mechanical decoupling levels, including salt and their potential role during sequential reactivation. We address the role of rift segmentation (non-cylindricity of rifts) and the contrasting tectonic histories of reactivated segment centres compared to reactivated segment boundaries. We also try to unravel what the granulites can tell us about the pre- and syn-rift thermal state and how their presence and characteristics can help choosing between the different hyperextension models proposed in the OROGEN project. In a final part we extract and discuss some of the major learnings and develop a more generic and global perspective on how to implement “inheritance” in a workflow that can be used to describe and model the evolution of tectonic systems within a Wilson Cycle.

2 Historical overview

The understanding of geological systems evolved through the last two centuries from that of a static to a mobilistic and dynamic system. The development of new analytical, imaging, and modelling tools resulted in unprecedented progress in the comprehension of both rifted margins and orogenic systems. It enabled not only to gain insights into the temporal evolution of tectonic systems, but also to explore the physical parameters that control the related processes. Important steps forward in the understanding of tectonic systems include the discovery of nappes (Bertrand, 1884), the description of remnants of former “oceanic” sequences (ophiolites) in orogens (Steinmann, 1905), and the restoration of orogenic domains (Argand, 1924a, 1924b, 1916, 1911). This evolution was in parallel with Alfred Wegener’s Continental Drift theory (Wegener, 1915) that was at the origin of a more generic and global view of the Earth evolution. Then, Arthur Holmes’s mantle convection theory (Holmes, 1931), the mapping of the oceans floor (Tharp et al., 1959), the first idea of seafloor spreading (Hess, 1962), the description of symmetric seafloor spreading anomalies (Vine and Matthews, 1963) and the understanding of the significance of transform faults (Tuzo Wilson, 1965) enabled to propose a uniform, coherent and generally accepted theory referred to as the Plate Tectonic theory (Le Pichon, 1968; Morgan, 1971; Parsons and McKenzie, 1978). Shortly after, it became generally accepted that tectonic systems are cyclic, and such a cycle was referred to as the Wilson Cycle (Dewey and Burke, 1974; Wilson, 1966). Although some of the pioneers had already understood that continental drift had happened several times (e.g., Argand, 1924a, 1924b), the Wilson Cycle theory made it explicit that each tectonic system, except oceanic systems, overprints a former/existing system (Fig. 1a). This is where “inheritance” finds its place in the overarching Plate Tectonic model that describes the evolution of our planet Earth. However, even if the notion of “inheritance” is a fundamental and integral part of the Wilson Cycle, in the early days of Plate Tectonics and modelling experiments “inheritance” was left aside. The community was more focused on the understanding of processes and the development of a physical understanding of how the Earth works, rather than on inheritance, which is neither generic nor predictable, and was therefore seen as an obstacle to understand Plate Tectonics.

In the late 1960’s and 1970’s, it became clear that several observations from the Alpine-Pyrenean, Himalayan and Zagros collisional belts were not compatible with some of the first-order predictions of the Plate Tectonic theory (Jackson, 1980). While for some scientists this was the proof that Plate Tectonics did not work, others considered that these were only details of little importance. Today we understand that the Pyrenean and Alpine system in Western Europe probably neither reached mature seafloor spreading, nor subduction. This may explain on the one hand why field geologists working in these orogens who were not aware of the new geophysical and kinematic datasets coming from the oceans did neither believe the Plate Tectonics theory, nor were able to predict it (Trümpy, 2001). Conversely, it may also explain why neither Steinman’s trinity (Steinmann, 1905), nor Decandia and Elter (1969) exhumed mantle theory, nor Ampferer’s concept of “subduction” (Ampferer and Hammer, 1911; see McCarthy et al., 2020 for a review), all of which were based on field observations, did not convince the protagonists of Plate Tectonics. The simple text-book Plate Tectonic model prevailed for a long time in the literature. Whatever did not fit into simple models was either ignored or interpreted as overprinted or subducted. Examples are the Bonloc syn-/post-rift polymictic breccias covering granulitic facies metamorphic basement, the reworked mantle rocks in syn-/post-rift sediments such as in the Basque Cantabrian Pyrenees (DeFelipe et al., 2017), Western Pyrenees, and in the Lherz and Bestiac areas of the Central Pyrenees (e.g., Fortané et al., 1986; Lagabrielle and Bodinier, 2008; Debroas et al., 2010; Masini et al., 2014; Asti et al., 2019). Another example relates to the so-called North Pyrenean Fault, which has first been introduced as a late Variscan strike-slip fault (Mattauer and Proust, 1967) and later reinterpreted as a strike-slip system to fit plate kinematic reconstructions in the Atlantic based on the work of Wilson (1965) (Choukroune and Mattauer, 1978; Le Pichon and Sibuet, 1971). The problem field geologists faced in the Pyrenees was that outcrops in the North Pyrenean Zone, the Pyrenean “suture zone”, were neither compatible with such a large strike-slip fault, nor with the proposed oceanic subduction. This may explain why most of the field geologists working in the Pyrenees were reluctant to accept some of the new Plate Tectonic interpretations (for a discussion see Canérot, 2017). Today it is clear that the Pyrenean system does not violate the Plate Tectonic theory and that all tectonic systems do not necessarily go through a complete Wilson Cycle. Thus, while the Plate Tectonic theory was a huge leap forward for the understanding of global tectonic systems, its application to orogens such as the Pyrenees or the Alps was too simplistic and often model-driven. Jackson (1980) was among the first to propose that the reactivation of former rift structures may play an important role in collisional orogens, a point that has been nicely demonstrated by Tavani et al. (2021) for the Zagros, Apennines, Oman and the Taiwan collisional belts four decades later. However, while Jackson’s rifted margin model consisted of tilted blocks, Tavani et al. (2021) built their interpretation on a complex rift architecture, showing its importance in controlling the switch from thin- to thick-skinned thrusting.

In the late 1980’s and early 1990’s, serpentinized mantle rocks were drilled offshore Iberia (Boillot et al., 1987) and detachment systems were described in the southwestern United States (Wernicke, 1985). These observations inspired new rift models whereby mantle was exhumed to the seafloor via detachment faults (e.g., Lemoine et al., 1987). The first evidence and interpretation of mantle exhumation were presented in the Apennines (Decandia and Elter, 1969). Their link to detachment faults was subsequently highlighted by Florineth and Froitzheim (1994) in the Tasna nappe in the Central Alps. Lagabrielle and Bodinier (2008) were the first to suggest that the mantle rocks exposed in the Lherz area in the Central Pyrenees were already exposed at the seafloor during Cretaceous rifting, an interpretation that was further supported by Jammes et al. (2009) and Lagabrielle et al. (2010) who suggested mantle exhumation along a detachment system in the Mauléon basin in the Western Pyrenees. Since then, many studies focused on the hyperextended rift basins exposed in the Pyrenean and Basque-Cantabrian domains. However, the application of hyperextended margin models, mainly developed in the Iberia-Newfoundland and fossil Alpine Tethys rifted margins to the Pyrenean system turned out not to be a simple copy-paste exercise. Indeed, the lack of good outcrops in the North Pyrenean Zone, the widespread occurrence of salt, the much higher sedimentation rates and the evidence for a high temperature/low pressure metamorphic event during rifting (Clerc and Lagabrielle, 2014; Ducoux et al., 2019, 2021a) makes it difficult to compare observations from the Pyrenees with the classical examples better exposed and documented in the Alps and/or drilled and seismically imaged offshore Western Iberia. In particular the interpretation of the kinematics and vergence of extensional detachment systems remain debated in the Pyrenean and Basque-Cantabrian examples and competing interpretations continue to exist (Lagabrielle et al., 2020).

Fig. 1 (a) Main stages (rifting, seafloor spreading, subduction, and late orogenic collapse) within a Wilson Cycle are illustrated by the shapes of the three main interfaces, which are the Earth surface, the Moho and the base of the lithosphere. (b) Idealized lithosphere made of a thermally equilibrated layer cake versus a “real” post-orogenic lithosphere showing inherited structural and compositional complexity (modified after Manatschal et al., 2015). (c) Schematic evolution of the thermal state and gravitational potential energy depicted by the base of the lithosphere (1300 °C isotherm) and the surface topography within a Wilson Cycle. p.-o. ext.: post-orogenic extension.

3 Inheritance: concepts, definitions and methods

3.1 Inheritance: where do we come from and why is it important?

Tectonics evolved from a mainly descriptive to a more processes-oriented science focusing on a physical understanding of large-scale geodynamic systems. This evolution was paired with the development of new quantitative geological techniques, as well as imaging, analytical and modelling approaches. However, this evolution did not overcome the fact that all tectonic systems, except oceanic systems, build on their own and specific inheritance that is controlled by their geological history. Understanding how to define the initial conditions and what and how to implement it in numerical models is one of the big challenges in developing more predictive and quantitative models. To implement “inheritance” in numerical models, inheritance needs first to be defined and mapped. Secondly, as all inherited structures will not be reactivated, what and under what circumstances inheritance is reactivated need to be determined. Finally, it is necessary to identify the typical “conditions” prevailing at each tectonic stage of a Wilson cycle. These can be defined by the sum of “interface shape inheritance” and “persisting inheritance”. “Interface shape inheritance” encompasses the transient thermal state and gravitational potential energy, which can be approximated by the three main interfaces defining the lithosphere, namely the earth surface, the Moho and the base of the lithosphere. “Persisting inheritance” includes lasting lithospheric structural and compositional heterogeneities (for details see Sects. 3.2–3.3).

3.2 Interface shape inheritance: the role of potential energy during a Wilson Cycle

During a Wilson cycle, tectonic systems evolve through convergent and divergent stages in which their thermal state and gravitational potential energy, expressed by the surface topography and position of the base lithosphere, are changing (Fig. 1c). In the convergent part of the cycle, excess gravitational potential energy forms as a consequence of crustal and lithospheric thickening (slab underthrusting, crustal thickening and orogenic topography), while in the divergent part of the cycle, gravitational potential energy is lost due to crustal/lithospheric thinning. The decrease in gravitational potential energy goes hand in hand with an increase in the geothermal gradient, which is largely linked to the thinning of the lithosphere and the thermal equilibration of the continental crust. This first-order evolution can be outlined in a simple way by the shape of three main interfaces, which are the Earth surface, the Moho and the base of the lithosphere (Fig. 1a, c). While in the convergent part of the cycle these interfaces diverge (surface topography and lithospheric thickness), they converge in the divergent cycle. In this intentionally first-order view, there is only one stage at which a tectonic system may reach a gravitational and thermal equilibrium (i.e., all interfaces are roughly sub-horizontal), namely during the transition between a convergent and a divergent stage. Note, however, that it takes at least tens of millions of years to reach a gravitational and thermal equilibrium, and thus equilibrium may be reached only when the convergence-divergence transition is achieved on such a long (or longer) period (Fig. 1c). As each stage of a Wilson Cycle is characterized by a specific inherited gravitational potential energy and a thermal state, a key in understanding and defining the initial conditions of a tectonic system is to identify its position within the Wilson Cycle (Fig. 1a, c) and to define the extent of equilibration since the last tectonic event. This 2D view is, of course, a simplification since this problematic needs to be understood in 3D.

3.3 Persisting inheritance: the role of compositional and structural heterogeneities

While gravitational and/or thermal states are transient and may be equilibrated through a Wilson Cycle, structural or compositional inheritance recorded in rocks (Fig. 1b) can last and leave a legacy for subsequent events. Compositional inheritance results from the juxtaposition of rocks with different properties (density, velocity, permeability, radiogenic heat production, etc.), and thus different rheologies. Yet the bulk rheological architecture and composition of the crust/lithosphere are important to determine and predict the distribution of decoupling levels, buttresses, as well as dense materials that are “subductable” versus buoyant materials that are difficult to subduct. Compositional inheritance is often linked to igneous and metamorphic processes or hydration reactions. In contrast, structural inheritance results either from a strong anisotropy linked to crystal or preferred shaped orientations (e.g., foliation or alignments of grains or aggregates), or from mechanical discontinuities (e.g., faults). Examples of systems leaving a strong inheritance (compositional and structural) in convergent systems are subduction systems with arcs/fore-arcs/back-arcs, and collisional orogens with suture zones or fold-and-thrust belts (Fig. 1b). In divergent systems, igneous processes are often linked to mafic underplating and mantle depletion and/or enrichment. Deposition and formation of low frictional materials (e.g., salt, clays, or serpentine) or thick sedimentary sequences are also often linked to extensional systems. Structural and compositional inheritance cannot always be dissociated and often depend on boundary conditions (stress field, thermal state, pressure, etc.). Compositional and structural inheritance can exert a major control on strain localisation and/or distribution during subsequent reactivation.

Under some circumstances, structural and/or compositional inheritance can be wiped out. Examples are prograde metamorphic events that can dehydrate and change mineralogical compositions and/or thermally anneal microstructures (Braun et al., 1999; Yamasaki et al., 2006). Processes that can modify the composition of the subcontinental mantle lithosphere have been reported by Picazo et al. (2016). They may for instance relate to slab breakoff, which results in a lithospheric thinning and depletion linked to intense magmatic activity (e.g., Permian evolution in W-Europe; Petri et al., 2017). Magma infiltration and enrichment of the subcontinental mantle lithosphere during hyperextension (Müntener et al., 2004) is another process that can modify the mantle lithosphere through a Wilson cycle (e.g., Chenin et al., 2018).

3.4 Proposed concepts to map inheritance and investigate its role within a Wilson cycle

Two geological mapping approaches have been developed in the past to investigate the role of inheritance. A first mapping approach was proposed by  Tugend et al. (2014, 2015) who used seismic data, potential field methods and field observations to identify and locate the limits of structural entities referred to as rift domains. This approach was first applied to the Biscay-Pyrenean domain, where it enabled to map off- and onshore rift domains and to describe their reactivation during the Alpine convergence. The type of map that resulted from this approach is shown in Figure 2b. It shows the distribution of rift domains offshore and onshore the Pyrenean-Cantabrian domain. The method, developed by Tugend et al. (2014, 2015), enables to recognize the position of a domain in both present-day margins and remnants of former margins exposed on land, relying on first-order characteristics like the thickness of the underlying crust and the available accommodation space (offshore), and/or the lithology and type of sediments. Such information is not visible in more classical maps as that shown in Figure 2a.

Building on this approach, Chenin et al. (2015) undertook to assess the impact of orogenic inheritance on the architecture, timing and magmatic budget of the North Atlantic rifting. These authors defined a series of mappable first-order characteristics of orogens (suture zones, fold-and-thrust belts, arcs) and compared their distribution with that of rift domains (Fig. 3a, b; for details on the mapping methodology, see Chenin et al., 2015 and Chenin, 2016). Based on the mapping results they proposed that the Variscan and Caledonian orogenic domains were reactivated in a different way. While Mesozoic rifting reactivated and finally separated parts of the former Caledonian orogen, apart from few exceptions, it circumvented the Variscan domain (Fig. 3b, c). As most of the Variscan domain in Western Europe is floored by a layered and reflective lower crust (Fig. 3e) that has been interpreted as mafic underplating or granulite-facies metamorphic lower crust (e.g., Rey, 1993; Bois et al., 1989; for details see Sect. 5.1), Chenin et al. (2018) hypothesized that the different characteristics of the inherited mantle/lower crust beneath the Variscan domain, namely the large-scale underplating by mafic bodies and underlying depleted mantle had a strong impact on future rift phases, including on their magmatic budget. The impact of mafic underplating and mantle depletion on subsequent rifting was further explored using a numerical modelling approach (Chenin et al., 2019). A key result of this study was that the distribution of mafic underplating and depleted mantle may have a strong impact on the strain distribution during subsequent rifting.

Thus, apart from defining and mapping inheritance, it is also important to evaluate what type of inheritance controls subsequent tectonic events. In the case of Western Europe, the inheritance that had the biggest impact on subsequent rifting may have been located in the lower lithospheric mantle. It is therefore important to understand the nature of the subcontinental mantle and its evolution through the Wilson cycle in order to evaluate its role in controlling rift distribution and/or orogenic reactivation (Chenin et al., 2018, 2019). A map of the mantle domains in Western Europe was published by Picazo et al. (2016) (Fig. 3d) and will be further discussed later in this study.

Aside from the local/regional-scale inherited features that may survive tectonic events, various types of inheritance may be created during a tectonic event depending on the size/maturity of the system involved. For instance, Șengör (1991), Chenin et al. (2017) and McCarthy et al. (2020) proposed that the closure of wide/mature oceans differs from that of narrow/immature “oceans” not only by the existence of a “mature” oceanic crust and the type of mantle (depleted in the case of oceans, enriched or inherited in the case of margins), but also by the occurrence or absence of a magmatic arc and associated LP/HT (low pressure/high temperature) metamorphism, mantle depletion, HP/LP (high pressure/low temperature) slab metamorphism related to subduction. It appears that the closure of narrow oceans and failed hyperextended rift systems is essentially mechanical and that little or no mantle melting/depletion occurs meanwhile. These results are of particular importance for the OROGEN project because the Biscay-Pyrenean system is a type-example of a narrow/immature hyperextended rift system.

Fig. 2 (a) Map of the Mesozoic rift basins between France and Spain along the Iberian-European plate boundary. Modified after Lescoutre et al. (2021) and Masini et al. (2014). (b) Rift domain map of the Pyrenean-Cantabrian area. STS: Santander Transfer System, W: Western section, C: Central section; E: Eastern section. Modified from Tugend et al. (2014) and Cadenas et al. (2020). CC: continental crust; OC: oceanic crust; OCT: ocean-continent transition.

Fig. 3 (a) Major orogenic domains and structures in the North Atlantic realm (modified from Chenin et al., 2015 and Chenin 2016). (b) Map showing (i) the rift domains and corresponding rifting duration (i.e., the time lag between necking and breakup), as well as the distribution of breakup-related magmatic additions for the North Atlantic rift system; and (ii) orogenic inherited structures and magmatic additions onshore Western Europe and eastern N-America (from Chenin 2016). (c) Map showing the correlation of parts of the Variscan orogen across the Biscay-Pyrenean system (from Martínez Catalán et al., 2007). (d) Distribution of mantle types in Europe at 110 Ma (Late Aptian) (from Picazo et al., 2016). (e) Interpreted seismic section through the Variscan orogenic area (from Bois et al., 1989; see map in panel b for location). VT = Variscan Thrust; VF = Variscan Front; the brown shading highlights the layered and highly reflective lower crust that is interpreted as either magmatic underplate or granulite-facies metamorphosed rocks (from Chenin et al., 2019).

4 The Biscay-Pyrenean system: a natural laboratory to study inheritance

The Biscay-Pyrenean domain, located between the southern North Atlantic and Alpine Tethys, includes one of the best-investigated, partly reactivated magma-poor extensional systems worldwide. Moreover, it shows a transition from a slightly to completely reactivated pair of rifted margins/rift system that is well imaged by industry and academic surveys and that has been studied by several academic consortiums, including the recent OROGEN project.

In the OROGEN project, the Biscay-Pyrenean domain was chosen as a natural laboratory to investigate the role of inheritance within an incomplete Wilson Cycle. This choice was justified by: (1) the position of the Biscay-Pyrenean system between the southern N-Atlantic and Alpine Tethys systems, both belonging to the best-investigated magma-poor rift systems worldwide; (2) the access to a unique dataset that includes geophysical and geological data from both academia and industry, with a scientific community that has been investigating these sites over generations; (3) the possibility to work on one and the same tectonic system onshore and offshore; and (4) the access to different genetic stages of an incomplete Wilson Cycle that initiates after a Paleozoic orogenic cycle (Variscan), goes through a Mesozoic extensional cycle, before undergoing an Alpine reactivation during Late Cretaceous to Cenozoic.

Several studies within the OROGEN project were directly or indirectly linked to the study of inheritance and its consequences. This includes the description of multistage rift systems and the use, further development and refinement of mapping approaches that resulted in new maps and sections across the North Iberian margin and the Basque-Cantabrian and Western Pyrenean domains (Cadenas et al., 2020; Lescoutre et al., 2021; Miró et al., 2021). It also encompasses the study of lower crustal and mantle rocks by Asti et al. (2019) and Saspiturry et al. (2019), the thermal structure of the lithosphere (Clerc and Lagabrielle, 2014; Ducoux et al., 2019, 2021a; Lescoutre et al., 2019; Saspiturry et al., 2020b), and the role of salt (Issautier et al., 2020; Saspiturry et al., 2021; Ford and Vergés, 2021). A major effort was devoted to the development of new passive seismic imaging methods (Wang et al., 2016; Polychronopoulou et al., 2018, 2019; Chevrot et al., 2015, 2018) that enabled to gain insight into the deep structure of the Pyrenean domain, as well as to resolve the local 3D architecture of a hyperextended system (Maupasacq experiment, Lehujeur et al., 2021). The results of all these studies enabled to define and map the size and maturity of the Biscay-Pyrenean extensional system and to understand the role and importance of inheritance during the reactivation of the initial rift system and subsequent collision.

Despite all these benefits, the Biscay-Pyrenean system is far from being understood as demonstrated by the numerous ongoing debates that persist even after intense investigation. Lively debates concern: (1) the competing and partly diverging kinematic reconstructions between the Iberia and European plates and the ages of the related tectonic events; (2) the lithological architecture of the crust and lithosphere prior to rifting (e.g., position of granulites referred in this paper to as the “granulite conundrum”); (3) the role of Triassic salt during the extensional and convergent cycles; (4) the architecture of the hyperextended rift basins, including the vergence of their main fault systems, the existence and importance of rift segmentation for subsequent reactivation; and (5) the thermal state during rifting and its importance in controlling the localization of subsequent shortening phases.

In the following, we aim to show how the Biscay-Pyrenean system can be used as a laboratory to study inheritance. We do neither aim to present a detailed description of the geological history of the Biscay-Pyrenean domain, which is the subject of other papers in this special volume, nor to list all results of the OROGEN project that are too numerous and partly already published or presented in this special volume. We rather focus on key elements that enabled us to identify compositional and structural inheritance, and on the learnings that enabled us to develop a more generic understanding of the role of inheritance in rift and orogenic systems.

5 The Biscay-Pyrenean system: evolution of an incomplete Wilson Cycle

The Biscay-Pyrenean domain (Fig. 2a) experienced a complex tectonic evolution that started in the Paleozoic with the Variscan orogeny (Martínez Catalán et al., 2007; Matte, 1991). After the post-Variscan extensional collapse, multistage Triassic to Cretaceous rift events linked to the opening of the Central and North Atlantic resulted in the opening of the Bay of Biscay and extensional basins along the Iberian-European plate boundary (Roest and Srivastava, 1991; Ziegler, 1988). These basins were reactivated during the subsequent Alpine convergence that initiated in Campanian/Santonian time. In the following, we describe the geological evolution of the Biscay-Pyrenean system, focusing on the role of inheritance. For more detailed descriptions of the evolution described below see Cadenas et al. (2020), Issautier et al. (2020), Saspiturry et al. (2021), Lescoutre et al. (2021) and Miró et al. (2021).

5.1 Paleozoic orogenic evolution and late- to post-orogenic collapse

The Paleozoic orogenic evolution is intimately linked to the amalgamation of the Pangea supercontinent, which is documented along the N-Atlantic/Tethys system in W-Europe, and along the conjugate NW-African and N-American continent (Aguilar et al., 2015; Arthaud and Matte, 1975; Cochelin, 2016; Cochelin et al., 2018). The formation of the Caledonian and Variscan orogens followed the closure of the Iapetus and Rheic oceans, respectively. While the main suture zones, arcs, fold-and-thrust belts and flexural foreland basins of the Caledonides have been mapped at the scale of the N-Atlantic, the existence and location of arcs and the nature of suture zones in the Variscan system remain debated. For instance, it is still unclear how to correlate parts of the Variscan orogen across the Biscay-Pyrenean system. This uncertainty may be linked to the fact that present-day restorations of the Variscan system in the Biscay-Pyrenean system are based on kinematic restorations that are outdated and not coherent with the current knowledge of the Iberian geology. Thus, correlations of the Variscan domains across the Iberia-European plate boundaries, as shown in Figure 3c, strongly depend on the plate kinematic models that are at present still debated.

A key observation is that none of the first-order Variscan structures (sutures, arcs) correlate and/or coincide with the Mesozoic rift trends mapped in the hyperextended Biscay and Pyrenean rift systems (Figs. 2a and 3a, b). Indeed, Mesozoic rift systems are highly oblique and/or truncate most of first order Variscan structures. At a more local scale, correlations between inherited Variscan structures and younger, either Mesozoic rift structures, or Alpine structures have been proposed, but these remain often debated. Examples are the Pamplona, Leiza or Louhossoa faults (for location of the latter two see Fig. 2b), for which different interpretations still co-exist (e.g., Lescoutre et al., 2021; Saspiturry et al., 2019) and discussion in Sect. 5.2). Depending on the authors, these structures are interpreted either as inherited and reactivated, or as rift-related, or, in the case of the Pamplona fault, as fictional. The fact that the Variscan orogen has been disrupted during the late Carboniferous to Permian orogenic collapse, the multistage Mesozoic rifting, and the subsequent Alpine reactivation makes it often difficult to assign structures to specific events. Therefore, rather than focusing on single structures, here we try to establish a first order understanding of the conditions that prevailed at the end of the Variscan phase, i.e., to establish the initial conditions for the subsequent Mesozoic rifting. This requires determining whether or not the system went to thermal and gravitational equilibration after the Variscan orogeny (for a discussion see Sect. 7.1).

The Permo-Triassic sedimentary succession in the Biscay-Pyrenean domain shows that the first sedimentary basins over the Variscan orogenic crust formed during latest Carboniferous to Permian and remained sub-aerial. This suggests a relatively fast disappearance of the “orogenic” topography and the establishment of a Basin and Range-type morpho-tectonic evolution with the local occurrence of core complexes (López-Gómez et al., 2019; Saspiturry et al., 2019). The first regional marine transgression is recorded during Triassic time and is largely diachronous from east to west. Based on isostatic considerations, marine conditions can only occur over continental crust that is less than 30 ± 5 km thick in a thermally equilibrated lithosphere. Beside the lack of significant accommodation space creation until late Jurassic time, i.e., over almost 100 My, and the widespread occurrence of shallow marine to sub-aerial conditions tip the scale against significant crustal thinning (i.e., less than 30 ± 5 km thick). These observations indicate a general thermal and gravitational equilibrated crust and lithosphere during Triassic time. Thus, we assume that equilibrium was reached before the onset of the Mesozoic multistage rifting.

Although the tectonic system was equilibrated at a lithospheric scale, the crust and underlying mantle were complex and contained significant structural and compositional inheritance. To establish and characterize the nature of inheritance, it is necessary to reconstruct the continental crust and its underlying subcontinental mantle at Triassic time, i.e., at the transition from the orogenic to the extensional cycle. Such a reconstruction is difficult to establish and needs to rely on the combination of structural and petrological data from tectonically exhumed rocks and xenoliths, as well as on geophysical data. The upper crust shows a structurally and compositionally complex “patchwork” of Cambrian to Carboniferous meta-sediments, the latter bearing the flysch-type Culm facies reminiscent of the Variscan foreland basin (Delvolvé et al., 1998) that is coeval with the emplacement of granitic magmatism and HT/MP-LP metamorphism in the hinterland dated 330 to 295 Ma (Cochelin, 2016; Martínez Catalán et al., 2007; Cochelin et al., 2018). A significant change in magmatism occurred during the Late Permian, when alkaline magmatism became dominant (Lago et al., 2004; Denèle et al., 2012;). Granulite facies metamorphism affected lower structural levels of the crust (Ribeiro et al., 2019). Metamorphism was coeval with a dextral strike-slip tectonic setting and with the onset of the Variscan orogen extensional collapse, and was contemporaneous to or followed by calc-alkaline plutonic and volcanic magmatism (Aguilar et al., 2015; Denèle et al., 2012, Petri et al., 2017). At present, no basement rocks exposed in the Pyrenean belt, including granulites, show P values higher than 6 kbar (Vielzeuf and Kornprobst, 1984; Vielzeuf and Pin, 1989). Thus, the nature of the pre-rift lower crust deeper than 18 km depth is not constrained by outcrop data. However, reflection seismic data show that the present-day Iberian and European Variscan crusts are floored by a reflective, stratified and high velocity lower crust (Bois et al., 1989; Alvarez-Marrón et al., 1996) (Fig. 3e). ODP Sites 900 and 1067 offshore West Iberia and xenoliths in volcanic rocks over Iberia and Western Europe suggest that the lowermost crust is likely made of Permian underplated mafic bodies (Petri et al., 2017). Thus, the pre-rift lower crust of the Biscay-Pyrenees realm may have been similar to the one described from the Ivrea zone in the Alpine domain (Petri et al., 2019). For both the understanding of rifting and the question of pre-rift and rift inheritance, the granulites that occur throughout the Biscay-Pyrenean system are of particular importance and will be discussed in Section 10.

Mantle rocks are exposed throughout the Biscay-Pyrenean domain (Fabriès et al., 1991, 1998). Lagabrielle et al. (2010) described two types of occurrences, namely reworked within Cretaceous sediments, and exhumed along extensional detachment faults. The best exposed examples are at Lherz and Urdach, both showing relations to surrounding crustal rocks. All mantle rocks outcropping in the Biscay-Pyrenean domain have been classified as inherited shallow subcontinental mantle (mantle type 1 of Picazo et al., 2016; Fig. 3d). Infiltrated or oceanic mantle rocks (types 2 and 3 of Picazo et al., 2016) have not been found, which is compatible with the lack of Cretaceous MORB basalts throughout the Pyrenean system. Thus, in contrast to the Biscay system, there is no evidence that the Pyrenean domain went to lithospheric breakup, which means that it may have never evolved into a pair of conjugate rifted margins.

5.2 Multistage Mesozoic rifting at the junction between N-Atlantic and Alpine Tethys

Rifting, defined here as the extensional process that stretches and thins the crust below 35 ± 5 km, and to less than 10 km in hyperextended examples, occurred in the southern N-Atlantic and Alpine Tethys systems from Late Triassic to late Early Cretaceous (Frizon de Lamotte et al., 2015; Le Pichon et al., 1977; Leleu et al., 2016). During the Late Triassic and Early Jurassic, the main locus of rifting was localized along the future Central Atlantic margins and in the different branches of the Alpine Tethys realm (Fig. 4). It formed a plate boundary separating Africa/Gondwana from North America and Europe, so that Iberia was part of the European plate at that time. Around and inside Iberia, the importance of Triassic-Lower Jurassic rifting remains debated (Angrand et al., 2020). Nevertheless, the occurrence of massive bodies of diabase with ophitic structures (Rossi et al., 2003) throughout the Pyrenean domain suggests a significant and large-scale magmatic event. Deposition of salt occurred during the Late Triassic-Hettangian rifting, and as evaporites are efficient decoupling material, understanding their paleogeographic distribution is important to evaluate their role in controlling subsequent tectonic events (for more details see Sect. 8). In the following we refer to this Late Triassic-Hettangian extension phase as a stretching phase sensu Lavier and Manatschal (2006). During the late Early to Middle Jurassic, rifting culminated in the Central Atlantic with breakup and subsequent seafloor spreading, coeval with the formation of the Alpine Tethys margins separating Adria from Iberia and Europe (Angrand et al., 2020). During this stage, the former Biscay-Pyrenean realm was little affected by extension as indicated by the lack of major depocenters of this age.

During the Late Jurassic to Aptian, the Atlantic rifting propagated in the southern N-Atlantic realm and resulted in hyperextension and mantle exhumation along the Iberia-Newfoundland margins (Alves et al., 2009; Alves and Cunha, 2018; Soares et al., 2012), as well as in the Orphan and Porcupine basins (Nirrengarten et al., 2018) (Fig. 4). Several competing kinematic models exist for this time for the Biscay-Pyrenean domain. Based on the restoration of the J magnetic anomaly as an isochron, Sibuet et al. (2007) and Vissers et al. (2016) suggested the existence of an oceanic domain in the Pyrenean realm during the Aptian, which was closed by subduction during the Albian. However, neither Aptian seafloor spreading, nor Albian subduction are supported by data. Another group of kinematic models is based on the assumption that the J magnetic anomaly does not represent an oceanic magnetic anomaly and can therefore not be used as an isochron for plate kinematic reconstructions (Nirrengarten et al., 2017; Angrand et al., 2020; Frasca et al., 2021). These kinematic models propose tight fit reconstructions for the southern N-Atlantic, which imply sinistral trans-tension along the Iberian/European plate boundary during Late Jurassic to Albian time. Angrand et al. (2020) and Frasca et al. (2021) suggested that sinistral movements were accommodated in the different branches of the Central Iberian Range, separating the Iberian plate from the European/Ebro plate (Fig. 4c). The precise location, kinematics and offset along this plate boundary remains, however, disputed. In the model of Angrand et al. (2020) and Frasca et al. (2021), it is suggested to be localized along a strike-slip corridor that linked the Asturian, Cabuérniga, Polientes, Cameros, Columbrets (for location see Fig. 2a) and other rift basins interpreted as pull-apart basins and strike-slip corridors. Such an interpretation is supported by the size, shape, bounding structures and main trend of these basins. These structures are aligned along a kinematic direction that plots on the same small circle as the Newfoundland or Flemish Pass strike-slip corridors, both of which are Late Jurassic to Aptian in age (Frasca et al., 2021). Therefore, we propose that all structures may be contemporaneous and originate from the same kinematic framework. Further constraints supporting these new plate kinematic models come from Lescoutre et al. (2021), who showed that the Ebro block lying between the Central Iberian Range and the Pyrenees was part of the European plate until the Late Aptian. Besides, Cadenas et al. (2020) showed that the Asturian basin formed prior to, and independent from, the Late Aptian to Cenomanian rift event, incompatible with the models previously proposed by Sibuet et al. (2004) and Vissers et al. (2016). Thus, the results of the OROGEN project suggest that the Late Jurassic to Barremian extension phase was a distinct rift event during which most of the sinistral strike-slip movement between the Iberian and European/Ebro plates occurred, in line with the plate kinematic model proposed by Frasca et al. (2021).

The Late Aptian to Cenomanian rift event was, in the past, not disassociated from the Late Jurassic to Early Aptian event. Choukroune and Mattauer (1978) suggested that the Biscay-Pyrenean domain was formed by a succession of Albian pull-apart basins linked to sinistral strike-slip movements along the North Pyrenean Fault (NPF), an interpretation that was largely accepted prior to the discovery of hyperextended basins (Lagabrielle and Bodinier, 2008; Jammes et al., 2009). However, neither the precise location of the NPF, nor the kinematics of this fault system were constrained in a satisfactory way. This was particularly true for the West Pyrenean-Cantabrian junction (Hall and Johnson, 1986). Classical models linked the NPF with the Leiza fault by an offset accommodated by the Pamplona transform fault (Engeser and Schwentke, 1986; Tavani et al., 2018; Lescoutre et al., 2021). This correlation was mainly based on the occurrence of mantle rocks, granulites and HT/LP pre-rift sediments all along this Pyrenean-Cantabrian system. However, new studies of the Pyrenean-Cantabrian junction do not support the existence of a NPF sensu Choukroune and Mattauer (1978). For instance, Saspiturry et al. (2019) demonstrated through the analysis of the Bidarray basin the impossibility to consider a large-scale sinistral strike-slip after Permian. In an alternative interpretation, Lescoutre et al. (2021) suggested the existence of two overlapping en-échelon hyperextended rift segments rather than two basins separated by a sharp transform fault. Further westwards, at the central North Iberian margin, Cadenas et al. (2020) mapped two en-échelon rift segments and described related extensional detachment faults similar to those described by Masini et al. (2014) from the Mauléon basin at the western termination of the Pyrenees. All these detachment systems are Aptian to Cenomanian in age are interpreted to be responsible for the exhumation of mid-crustal granulites and mantle rocks, and to show N-S directed kinematics (Cadenas et al., submitted). Further evidence for a segmentation of the rift system comes from tomographic imaging by Chevrot et al. (2018), who suggested a deep-seated along-strike segmentation of the Pyrenean system. Evidence for N-S directed extension comes also from kinematic indicators determined from exposed exhumation faults (e.g., Northern Mauléon Detachment fault; Masini et al., 2014, see also Tavani et al., 2018 for further examples), from the arrangement of the overlapping segments (Lescoutre et al., 2021; Cadenas et al., 2020), and from the occurrence of roughly N-S directed lateral ramps surface (Masini et al., 2014). All these observations cannot be satisfactorily explained by localized E–W-directed Aptian-Albian strike-slip movements in the Western Pyrenean (Canérot, 2017) and Basque-Cantabrian domains, and offshore in the North Iberian margin. Yet it is important to keep in mind that kinematic data showing E–W movements were reported from the granulites of the Basque massifs (Boissonnas et al., 1974). E–W-directed lineations in these rocks formed at conditions higher than 350 °C and need therefore to be older than the lowermost Jurassic Ar/Ar ages on biotite and muscovite (Masini et al., 2014). Indeed, U-Pb ages from the Ursuya massif indicate that granulites and paragneisses crystallized and were deformed between 295 and 274 Ma (Hart et al., 2016; Vacherat et al., 2016). Based on a structural analysis of the Ursuya granulites, Saspiturry et al. (2019) proposed that they underwent E-W orogen-parallel mid-crustal flow and were exhumed at the base of the upper crust, beneath Paleozoic meta-sediments. A more detailed discussion of the exhumation of the granulites and their importance in establishing the kinematics and timing of detachment faulting and related thermal structure can be found in Section 10. Kinematic data from the supra-salt cover have also been interpreted to show E-W directed movements of the supra-salt cover (Oliva-Urcia et al., 2010; Ford and Vergés, 2021). However, whether the latter data can actually be used to determine the local pre-salt basement deformation as well as the larger scale plate kinematic transport directions, or are only related to salt tectonics, remains questionable.

The results of the OROGEN project allowed to establish a well-constrained kinematic framework, at least for the Western Pyrenean and Bay of Biscay domains. Moreover, it allowed to describe the rift template, including the nature of rift segments and segment boundaries, and to understand the temporal and spatial evolution of a multistage rift system that includes a Late Triassic-lowermost Jurassic stretching system, a Late Jurassic-Barremian transtensional system, and a Late Aptian to Cenomanian hyperextended system (Cadenas et al., 2020). Each of these events is linked to a different kinematic setting, displays its own nature and architecture of rift basins, and its own subsidence and thermal evolution. Interestingly, although the Asturian and Cameros Late Jurassic-Barremian basins developed a significant tectonic subsidence (up to 8 km thick), no major thermal subsidence was recovered from these basins, in contrast to the later Aptian-Cenomanian rift domains (e.g., Angrand et al., 2018). Such an observation suggests that the lithosphere underlying these basins did not significantly thin. There is also an agreement that Iberia was not necessarily one single plate (Fig. 4). The new kinematic interpretations by Frasca et al. (2021) and Angrand et al. (2020) have important consequences for the evolution of the Biscay Pyrenean domain. If their models are correct, the main Late Jurassic to Barremian plate boundary was linking the North Atlantic with the Tethyan system through the Bay of Biscay-Central Iberian Range (i.e., south of the Ebro micro-continent) and not, as previously suggested, along the Pyrenees sensu stricto. However, more work is necessary to confirm this assumption.

Fig. 4 Restoration of the southern North Atlantic, with North America fixed on present-day coordinates after Nirrengarten et al. (2018) with modifications for the Biscay Pyrenean domain based on Frasca et al. (2021). (a) Santonian/Campanian (83 Ma), (b) Late Aptian/Albian (112 Ma), (c) Aptian (125 Ma), (d) Late Jurassic/Early Cretaceous (145 Ma), (e) end of Triassic (200 Ma). RECC: Reconstructed Edge of the Continental Crust. d.: domain; MOR: mid-oceanic ridge; EOB: East Orphan Basin; WOB: West Orphan Basin; PB: Porcupine Basin.

5.3 Onset of convergence

The age and the location of the onset of convergence in the Biscay-Pyrenean system are poorly constrained (Mattauer and Proust, 1967; Thinon et al., 2001) but are commonly assumed to be Latest Santonian to Campanian time (anomaly 34; 84 Ma) (Andrieu et al., 2021). Most evidence for onset of convergence is indirect. The problem in dating the onset of convergence is that the areas where shortening initiated are in the internal parts of the orogen (e.g., Ternois et al., 2019), therefore, they are either overprinted by later deformation, subducted and thus not accessible to observations, or located offshore and not drilled. Compression-like structures were interpreted to form already during the lower Cretaceous. However, the related structures were more recently re-interpreted as being linked to the interaction between normal faulting and salt tectonics (Gómez-Romeu et al., 2019; Ducoux et al., 2019; Izquierdo-Llavall et al., 2020; Ford and Vergés, 2021). At present, many authors proposed that initial shortening was accommodated in the hyperextended and/or exhumed mantle domains. The problem is that subduction of exhumed mantle does not lead to uplift and topography, and thus is difficult to date and locate with classical methods. Although evidence for convergence is difficult to access within the highly deformed internal domains, former studies (Muñoz, 1992; Teixell, 1996) and recent work in the Pyrenees (Mencos et al., 2015), on the Iberian margin (Andrieu et al., 2021) and within the Aquitaine basins (Rougier et al., 2016) demonstrated a change in Late Santonian time with the onset of flexural basins development. Andrieu et al. (2021), show a Coniacian uplift of the Iberian margin and significant subsidence similar to what is observed at the European margin. What controls this deformation is debated (far-field versus convergence), but a start of convergence causing lithosphere buckling cannot be excluded.

A reorganisation of the main orogenic structures occurred between 50 and 40 Ma (Eocene; Muñoz, 1992; Vergés and García-Senz, 2001; Vacherat et al., 2016) and was associated with the formation of the present fold-and-thrust belt well exposed in the southern Pyrenees. The onset of this reorganization is linked to the main phase of mountain building that started when hyperextended rift basins closed and the necking domains of the former rift basins collided. Macchiavelli et al. (2017) reconstructed the movement of the Iberian plate relative to the European plate using the well-imaged and unquestioned magnetic anomaly 34 and younger anomalies in the southern North Atlantic. Their interpretation may be considered as the best constrained kinematic description of the convergence between the Iberian and European plates. A new compilation of the magnetic anomalies from the Bay of Biscay-West-Pyrenean domain by Le Maire et al. (2021) may provide further constrains for future kinematic reconstructions.

6 Restoration of crustal sections back to pre-rift stage and implications for rift models

6.1 Kinematic restoration of sections back to the pre-rift stage

To evaluate the role of inheritance during rifting and reactivation, it is critical to kinematically restore sections back to their initial pre-orogenic (i.e., post-rift) and even pre-rift stages (Fig. 5). In the OROGEN project, Gómez-Romeu et al. (2019), Miró et al. (2021) and Lescoutre et al. (2021) proposed fully restored sections that fulfill area conservations as well as pre-rift marker horizon length conservation. The methodology of such restorations has been formulated in Chao (2021). Here we present a slightly modified version of the methodology. Requirements for an accurate kinematic restoration are that the restored section is parallel to shortening and extension direction and that the crustal area and the length of a pre-rift horizon are conserved during restoration (Dahlstrom, 1969). Restorations can be done using software such as ImageJ (freely available on the Internet: https://imagej.nih.gov/ij/) or 3DMove.

Area conservation, i.e., neither addition nor loss of crust out of the section can be formulated as: $$\text{A}{\text{c}}_{\text{initial}}=\text{A}{\text{c}}_{\text{final}}\text{,}$$(1)

where Ac_initial is the area of the crust at the pre-rift stage and Ac_final the area at the end of shortening. Such a reconstruction must be done at a crustal scale and needs to be based on the definition of top and base crust, which requires access to geophysical and geological data. More detailed restorations in which both the brittle and ductile crusts are restored bear the problem that the thermal evolution and/or hydration reactions may change the position of the brittle ductile transition (e.g., embrittlement concept of the crust in extensional systems; Pérez-Gussinyé et al., 2003). Therefore, the brittle and ductile crust areas do not remain constant and cannot be restored using the present method. Changes of the brittle and/or ductile crust area can be due to flow of ductile crust in and out of the section, magmatic additions and erosion that is mainly linked to convergence. Thus, if one of these processes occurs, they need to be quantified and can be formulated as $${\text{Ac}}_{\text{initial}}\pm {\text{Ac}}_{\text{volume}\text{change}}={\text{Ac}}_{\text{final}}\text{,}$$(2) A second requirement is that the length of a hypothetical pre-rift top basement marker horizon is conserved during extension and subsequent shortening. Such a pre-rift top basement marker corresponds, in an ideal case, to a continuous pre-rift stratigraphic marker horizon (e.g., Buntsandstein Lower Triassic deposits in Western Europe). The length of such an idealized pre-rift Marker Horizon (MH) forming the initial top basement (Tb_{MH_initial}) cannot change during the subsequent tectonic events, which can be formulated as: $$\text{T}{\text{b}}_{\text{MH}\_\text{initial}}=\text{T}{\text{b}}_{\text{MH}\_\text{final}}\text{,}$$(3) $$\text{T}{\text{b}}_{\text{final}}=\text{T}{\text{b}}_{\text{MH}\_\text{initial}}+\sum \text{T}{\text{b}}_{\text{f}}+\sum \text{T}{\text{b}}_{\text{m}}\text{,}$$(4)where Tb_final is the length of top basement at the end of rifting (or convergence), Tb_f is the length of newly created top basement owing to fault play, and Tb_m is the length of new top basement made of newly formed (accreted) oceanic crust. Establishing the nature of top basement in a section through a rifted margin is challenging, in particular through the most distal parts, since it requires distinguishing between Tb_{MH_initial}, Tb_f and Tb_m, which is often difficult and interpretative. A relation between Tb_{MH_initial} and Ac_final can be established through the following equation: $${\text{Z}}_{\text{initial}}\times \text{T}{\text{b}}_{\text{MH}\_\text{initial}}=\text{A}{\text{c}}_{\text{initial}}=\text{A}{\text{c}}_{\text{final}}\iff \text{T}{\text{b}}_{\text{MH}\_\text{initial}}=\text{A}{\text{c}}_{\text{final}}/{\text{Z}}_{\text{initial}}\text{,}$$(5)

where Z_initial is the thickness of the crust before the onset of rifting, which can be approximated by the thickness of the crust outside the areas affected by either extension or compression on either side of the restored section. Erroneous interpretations of top basement, such as the misinterpretation of an exhumation fault (Tb_f) and/or Tb_{MH_initial} and the location of first oceanic crust (Tb_m) results in gain or loss of crust during the extensional process (see Eq. (5)). Thus, for the restoration of 2D sections back to the pre-rift stage an accurate interpretation of top basement is essential.

The restoration of compressional systems is more difficult for internal parts of orogens, where imaging methods can often not resolve the nature and location of the top and base of the crust. Moreover, collisional systems form topography that is submitted to erosion and thus changes in crustal area, in contrast to extensional systems that are largely below sea level. Thus, restorations of collisional systems are commonly less constrained than those of extensional systems. While restorations of thrust faults heaves are relatively simple in external fold and thrust belts (especially into systems without salt tectonics), such methods fail in heavily pre-structured internal parts, in particular if thrust faults reactivate former extensional detachment faults. However, in kinematic restored sections, Equation (5) needs to be fulfilled if one assumes that crust has been neither added nor removed from the section.

Fig. 5 Kinematic restoration of a crustal section back to the pre-rift stage with area conservation as well as pre-rift marker horizon length conservation (see text for further explanations). Tb: top basement; Tbm: magmatic top basement (neww real estate); Tbf: fault top basement (new real estate); TbMH: stratigraphic marker horizon top basement.

6.2 Implications for the rift models proposed for the Pyrenean-Cantabrian domains

Among the restorations of sections from the Basque-Cantabrian and Pyrenean domains back to the pre-rift stage presented in Lescoutre et al. (2021), Miró et al. (2021), and Saspiturry et al. (2020a) (see also Lagabrielle et al., 2020, and sections in there), two groups of mutually exclusive interpretations can be distinguished: The first group of interpretations shows highly extended, asymmetric basins floored by detachment faults that dip to the north. In these interpretations the basin is dominantly floored by Tb_f. with some extensional allochthons (e.g., Ducoux et al., 2019; Jammes et al., 2009; Masini et al., 2014; Teixell et al., 2016; Lescoutre et al., 2021; Miró et al., 2021). The second group of interpretations shows more symmetric hyperextended basins, bounded by south- and north-dipping faults and floored mainly by pre-rift sediments. In these interpretations the basin is dominantly floored by Tb_{MH_initial} (Corre et al., 2016; DeFelipe et al., 2017; Pedrera et al., 2017; Saspiturry et al., 2021, 2020b). While the first group of interpretations fulfils Equations (1) and (3), the second group of restorations assumes a loss of crustal area, i.e., it meets Equation (2). Two hypotheses were proposed to account for this crustal loss: Corre et al. (2016), Lagabrielle et al. (2016) suggested a hot rift with a ductile crust that flowed during rifting out of the section. Asti et al. (2019) proposed that the crust was already thinned to 20 km in Permian time, suggesting that the crust was not equilibrated at rift onset. Thus, the two groups of interpretations require different initial conditions for the Mesozoic rifting observed in the Pyrenean-Basque-Cantabrian domain with far-reaching implications for the thermal evolution and the related rheology of the extending crust. These different models and their implications are further discussed in Section 9.

6.3 Implications for section restorations across the Pyrenean-Cantabrian domains

In Figure 6, we present three fully restored crustal-scale sections, one through the Cantabrian basin, one through the central segment of the Basque-Cantabrian basin and a last one through the en-échelon overstepping termination of the Basque-Cantabrian and Mauléon basins located in the Basque massifs, at the Pyrenean-Cantabrian junction (for original sections and explanations see Lescoutre et al., 2021 and Miró et al., 2021). All were interpreted based on field observations and are consistent with published geophysical and drill hole data. The three sections distinguish between Tb_{MH_initial} and Tb_f and fulfil Equations (1) and (5), assuming that no crust was gained or lost out of the section. Thus, they suggest that erosion of basement and the volume of magmatic additions in the crust were minor. Note also that they show substantial differences to previously published lines from DeFelipe et al. (2017), Pedrera et al. (2017, 2021). The restorations shown in Figure 6 assume that the main left-lateral strike-slip movement of Iberia relative to Europe was accommodated along a strike-slip corridor located along the Internal Iberian Range, as proposed by Angrand et al. (2020) and Frasca et al. (2021) and shown in Figure 4c. Moreover, they assume that the kinematic transport direction during Late Aptian to Cenomanian extension and subsequent reactivation were both approximately N-S directed (see discussion in Sect. 5.2). Thus, even if the interpretation of sections at a crustal scale can never be considered as a unique solution, the three restored sections shown in Figure 6 can at least be considered as coherent interpretations respecting a well-defined restoration method and consistent with all available geological and geophysical data. Note also that all three sections provide coherent values for crustal extension, between 60 and 90 km. Shortening in the sections is between ca. 100 km in the eastern section, >105 km in the central one and >65 km in the western one. Note however, that this method does not consider the amount of extension or shortening in the exhumed mantle domain. Yet, since the eastern section across the Mauléon-Basque/Cantabrian junction (see Fig. 6) does not show any evidence for exhumed mantle, the values obtained by its restoration of about 60 km of extension and 100 km of north-south shortening can be considered as the best constrained values for the Pyrenean system. Much more extension appears to have been accommodated west of the Santander transform system, where extension also included late Jurassic to Albian sinistral strike-slip movements of the Iberian plate relative to Europe (see Fig. 4). This further supports that plate movements were partitioned during the Mesozoic rifting and that Iberia cannot be considered as one single plate (Canérot, 2017; for more details see Frasca et al., 2021).

Fig. 6 Sections through the Cantabrian (Western transect) and the Basque Cantabrian Basin (Central transect) (after Miró et al., 2021) and through the Mauléon-Basque massifs (Eastern section) (after Lescoutre et al., 2021). Sections show the present-day situation and restorations at middle Cenomanian (96 Ma) and at Barremian (124 Ma) times. These sections are restored respecting area conservation as well as pre-rift marker horizon length conservation. See Fig. 2b for section location. B.: basin; H.: high; BCB: Basque-Cantabrian Basin; SPBT: South Pyrenean Basal Thrust; NPFT: North Pyrenean Frontal Thrust.

7 Assessing the role of rift inheritance in the Biscay-Pyrenean system

This section focuses on examples that have been investigated during the OROGEN project and have been either already published or are about to be published. The aim of the current section is to put the results of these studies into perspective and to discuss the potential new emerging concepts they enabled.

7.1 The role of Variscan inheritance in the Biscay-Pyrenean system

In this study, we introduced the concept of “interface shape inheritance” to describe the Biscay-Pyrenean system at the end of the Variscan cycle. Its application requires evaluating the thermal state and the residual gravitational potential energy of the system at this stage. Based on subsidence analysis, studies of depositional environments and taking isostatic considerations into account, we can assume that the thermal and gravitational potential energy was equilibrated before the onset of the Mesozoic rift events in the study area. This means that the main interfaces, i.e., top basement, base of the crust and base of the lithosphere were largely parallel. In contrast, compositional and structural heterogeneities inherited from the Variscan and the post-Variscan cycle persisted in the crust and the underlying mantle. Although the detailed nature and rheological architecture of the crust, in particular of the lower crust, and the distribution of major inherited structures remain debated, it is generally accepted that main rift structures truncate Variscan structures and that, at a larger scale, the Mesozoic rift system circumvented the Variscan system (Chenin et al., 2015). Thus, inherited Variscan structures appear not to have significantly controlled the subsequent Mesozoic rift system. This does not mean that inherited structures did not control extensional faults at a local scale, only that a general pattern cannot be recognized. For instance, the change in the composition of the upper crust along strike does not seem, at a first order, to control the distribution and architecture of the extensional systems. At the scale of the N-Atlantic, Chenin et al. (2019) suggested that the mantle composition (depleted versus inherited) and the occurrence or absence of underplated mafic crust formed during orogenic collapse are the main factors controlling subsequent rift location. Thus, it can be hypothesized that the nature of the mantle inherited from the orogenic evolution may have controlled the rift system at a large scale. In W-Europe, the lower lithospheric mantle inherited from the post-Variscan collapse was presumably depleted during the post-orogenic collapse (Picazo et al., 2016). The existence of such a depleted mantle that was likely thermally equilibrated at the onset of rifting may explain the low volume of magma produced during Mesozoic rifting over rifted Variscan lithosphere. It could also explain that most Mesozoic rift systems circumvented the Variscan lithosphere, and that oceans formed in the southeast (Alpine Tethys) and west (North Atlantic) of the orogenic domain.

7.2 Assessing the initial conditions for the Mesozoic rifting in the Biscay-Pyrenean system

As stated in Section 6.2, two competing and mutually exclusive models have been proposed to explain hyperextension in the Pyrenean-Cantabrian junction. Here the aim is not to exclude one of the two, but to discuss the implications of these models on the “interface shape inheritance” at onset of rifting. The first model assumes an equilibrated crust and lithosphere without anomalously hot geothermal gradients at the end of the Variscan cycle, in line with the conclusions from the previous chapter and with the model of Duretz et al. (2019). The second model assumes either high geothermal gradients and ductile crust flow during rifting (Clerc and Lagabrielle, 2014) with the presence of salt playing an important role for controlling the high thermal gradient and ductile deformation in the crust (Lagabrielle et al., 2020; Duretz et al., 2019). An alternative model assumed that rifting started with an already thinned continental crust (Asti et al., 2019). The latter hypothesis implies that the crust was thinned to 20 km already during Permian time. If that would have been the case, major accommodation space should have been created during Triassic and Jurassic time, except in the case of an anomalously high geothermal gradient that would have provided the necessary thermal support. Yet, on the one hand the stratigraphic record observed in the Biscay-Pyrenean domain does not display evidence for kilometre-thick Triassic or Early Jurassic depocenters; and on the other hand, massive thermal support would imply an anomalous high geothermal gradient that persisted from the Permian until the onset of major rifting in Late Jurassic time (i.e., over more than 100 My). Although such a long-lasting high thermal anomaly could explain the high syn-rift thermal gradient proposed by Clerc et al. (2015), Lagabrielle et al. (2016) and Corre et al. (2016), it remains unclear what process(es) could prevent thermal equilibration over such a long time.

7.3 Assessing the role of rift inheritance during reactivation and collision

From a mapping point of view, there is a good correlation between rift domains and orogenic structures, as already suggested by Tugend et al. (2014). The results of the OROGEN project allow to distinguish between three situations: (1) reactivation of rift segment centres, (2) reactivation of rift segment boundaries, and (3) localization and subsequent stepping of shortening aside.

In all examples shown in Figure 6, reactivation initiates with soft collision in the exhumed or hyperextended domain along weak, low frictional serpentinized mantle and/or reactivation of rift inherited extensional detachments. This suggests that compositional weakening due to serpentinization may be a dominant factor controlling reactivation and may be even more important than thermal weakening for subduction initiation.

Shortening that follows mantle subduction can take different avenues. At segment centres (e.g., Central section across the Basque-Cantabrian basin; Fig. 6) the collision shifted from soft to hard collision when the Coupling Point (CP) began to subduct. Here the CP refers to the location separating hyperextended, brittle, <10 km-thick crust from >10 km-thick crust that preserves inherited ductile crustal levels (i.e., the necking zone and proximal domain). At that time, shortening was accommodated by ductile mid-crustal levels in the necking zone (thick-skin) and/or by low friction layers (salt or clay) at shallow levels (thin-skin). In such systems the CP corresponds to the tip of the buttress, which corresponds to the oceanward limit of the former necking zone, and becomes the S-Point (sensu Willett et al., 1993) of the collisional system. The presence of salt can result in a complete decoupling between sub- and the supra salt levels and may create a large-scale and relatively well-preserved pop-up structure consisting of former syn- and post-rift sequences. Remnants of the former footwall may be accreted during the subduction of the former hyperextended and exhumed domains, in which case they occur in the internal parts of the orogen as shortcut tectonic slivers. However, most of the hyperextended and exhumed mantle domain is usually subducted/underthrust, so that the hyperextended crustal and exhumed mantle sections are not preserved in the central part of former rift segments.

At segment boundaries (e.g., Eastern section across the Basque massifs area; Fig. 6), thrust faults that reactivate former detachment systems are forced to create shortcuts to make lateral linkage with adjacent segments. Therefore, many former rift structures escape reactivation and thus remain preserved in the orogenic wedge.

A third behaviour (e.g., Western section across the Cantabrian margin; Fig. 6) shows a stop of convergence at the site of initial subduction followed by a reorganisation and localization of shortening within the proximal margin, resulting in out-of-sequence structures. This observation not only questions how shortening in the continent can compete energetically and mechanically with mantle subduction, but also what may control the stepping aside of shortening and why does the continent become the locus of shortening. A possible reason could be the onset of hard collision in the neighbouring segment to the east, which may have triggered a reorganisation of shortening.

Thus, a segmented rift template can have important consequences during shortening and may explain out-of-sequence reactivation and/or the spatial re-organisation of shortening. This shows that reactivation cannot be studied using 2D sections only. Where complex multi-stage, segmented rift systems exist, it is important to map rift domains in 3D and to analyse reactivation with a map view in mind. Indeed, rift segmentation can explain the along-strike non-cylindricity of the internal parts of the Pyrenean-Cantabrian belt. The complex structure of the internal parts contrasts with the rather continuous along strike extent of the fold-and-thrust belt in the more external parts that developed during the final stage of collision. Observations from the OROGEN project suggest that rift inheritance controls mainly the initial stages of reactivation, while later stages can be explained at a first order by Coulomb wedge theory (Willett et al., 1993).

8 The role of pre-rift salt in formation and reactivation of hyperextended systems

8.1 Distribution of pre-rift salt in hyperextended systems

The behaviour of syn-rift salt during the formation of rifted margins has been widely investigated in the past (Rowan, 2014 and references in there). In the Biscay-Pyrenean domain Biteau et al. (2006), Canérot et al. (2005), Jammes et al. (2010), Ferrer et al. (2012), Issautier et al. (2020) and Lagabrielle et al. (2020) investigated the link between salt, rifting/hyperextension and the early stages of reactivation. The results of the OROGEN project shed new light on the importance of pre-rift salt during the evolution of the Biscay-Pyrenean rift system, and on its importance during subsequent compressional reactivation. Indeed, in contrast to many rift systems where salt is deposited during hyperextension (e.g., Gulf of Mexico, S-Atlantic, Rowan, 2020), the main salt deposition is of Late Triassic age in the southern North Atlantic and in the Tethyan domain in Western Europe, and thus predated the main rifting event. One of the key questions relates to the presence of salt over exhumed crust and mantle: if salt was present prior to hyperextension, when and how was it emplaced over these exhumed surfaces that did not exist at the time of salt deposition (for a discussion of this problem see Jammes et al., 2010; Rowan, 2014)? Answering to this question requires to understand the distribution of the Triassic mother salt (see discussion in Angrand et al., 2020) and its redistribution (gravitational versus tectonic) during subsequent rifting and reactivation. The extent and thickness of Triassic salt appear to increase from west to east in the Biscay-Pyrenean system (Vargas et al., 2009). Jammes et al. (2010) and Lagabrielle et al. (2020 and references therein) proposed that salt may have moved gravitationally during hyperextension. However, how much salt was present in the most extended and exhumed parts of the Biscay-Pyrenean hyperextended rift domain and what was its control during onset of convergence remains unclear.

8.2 Pre-rift salt in the northern Iberian hyperextended basins

Cadenas et al. (2020) show that evaporites and related salt tectonics occurred in Triassic basins while evidence for salt in the distal parts of the North Iberian margin are rare. Besides, the occurrence of mantle clasts in salt diapirs in the Mauléon Basin (Jammes et al., 2010) and in the Basque-Cantabrian Basin (DeFelipe et al., 2017) supports the idea that salt was present in the exhumed and hyperextended domains. This assumption is strengthened by recent studies for instance in the Chaînons Béarnais, where authors document not only the presence of salt onto the exhumed mantle (Asti et al., 2019; Corre et al., 2016), but also important metasomatism related to evaporite dissolution and remobilisation and mineralization in the Jurassic to Cretaceous sedimentary cover (Corre et al., 2018).

In their section through the Basque-Cantabrian Basin, Miró et al. (2021) proposed that salt acted as a major decoupling level between the pre- and post-salt units (Fig. 7c). They suggested that salt was present in the hyperextended domain below the more than 12 km thick sedimentary infill, including thick pre-hyperextension Upper Jurassic-to Albian sediments. In contrast, in the Cantabrian section (Fig. 7b), salt does not occur in the distal hyperextended margin and only occurs in the more proximal basins and in the necking zone. Thus, while salt may occur in hyperextended domains east of the Santander system, it is not observed further to the west, suggesting that the distribution of salt may have changed along strike in the hyperextended domains. Two main processes may explain why salt does not exist along the distal western North Iberian margin and is more common in the Basque-Cantabrian basin: (1) the amount of extension, and therefore the amount of new real estate (i.e., newly exhumed basement (Tb_f)), was significantly higher west of the Santander zone compared to the east (see Fig. 6b, c); and (2) a general decrease in salt westwards. However, these hypotheses do not explain the process by which salt can occur over hyperextended domains.

The presence of salt along the Biscay-Pyrenean system is often linked to the presence of thick Late Jurassic to Early Aptian depocenters (Fig. 7a). Cadenas et al. (2020) and Miró et al. (2021) showed that, while the Late Jurassic to Barremian Asturian and Cabuérniga basins were not the locus of hyperextension, the thick Late Jurassic to Barremian depocenters in the Basque Cantabrian Basin occur over hyperextended crust. Thus, two situations can be envisaged: (1) no interference/reactivation between salt-bearing basins and hyperextended systems (e.g., Cantabrian section) (Fig. 7b); or (2) full interference/overprint of salt-bearing basins during later hyperextension (e.g., Basque-Cantabrian example) (Fig. 7c). While in the first case the presence of salt over exhumed domains is unlikely, in the latter case pre-rift salt can occur as allochthons in the hyperextended and exhumed domains. A likely interpretation is therefore that Triassic salt was linked to allochthons of thick pre-rift sections from Triassic and Jurassic basins that were caught up during Aptian to Cenomanian hyperextension.

Fig. 7 (a) Schematic view of the two possible fates for Triassic (pre-hyperextension) salt in a multistage rift system: the top-left panel shows no interference/reactivation between salt bearing basins (blue) and subsequent hyperextended systems (green); the top-right panel shows full interference/overprint of salt bearing basins during later hyperextension (modified from Miró et al., 2021). (b and c) Role of pre-rift salt during rifting and Alpine reactivation in the Asturias (b) and Basque-Cantabrian (c) transects (from Miró et al., 2021); Note that these sections are zooms on the sections displayed in Figure 6.

9 The thermal state of hyperextended rift basins in the Biscay-Pyrenean domain

9.1 Thermal structure of hyperextended and exhumed domains

Defining the thermal structure of hyperextended and exhumed mantle domains remains one of the most challenging endeavours in extensional tectonics since, in contrast to the structural or stratigraphic record, the thermal state is transient and equilibrates after rifting. Heat flow data is usually used to assess the thermal state of a (equilibrated) margin. However, these data do not allow to retrieve the thermal state during the (transient) rifting process. Moreover, there is only little public data where heat flow is calibrated by drill holes. Studies from young, not yet thermally equilibrated rifted margins such as the Gulf of Aden (Lucazeau et al., 2010) show generally high heat flow values that are, however, highly variable across the ocean continent transition. Similar observations were reported from the Red Sea (Agulles et al., 2020), the Gulf of California (Neumann et al., 2017) and the South China Sea. For the latter example, Nirrengarten et al. (2020) showed based on IODP drill hole data that pre-rift sediments were heated to temperatures as high as 200^∘C during the early stages of continental/lithospheric breakup, which requires a high heat flow during rifting given the relatively small sedimentary burial of the margin. In contrast, post-rift sediments have lower thermal maturities, which is likely due to a combination of a lower geothermal gradient, limited burial, and absence of late post-rift magmatism.

Data from sediment-starved Mid Ocean Ridges at magma-poor slow spreading systems show that the thermal structure is mainly controlled by hydrothermal convection linked to active faults and magma emplacement (Kelley and Shank, 2010). Here again, the thermal state is heterogeneous and very local. Measured T_max values at fossil rift systems can be linked to hydrothermal activity or magma emplacement, and thus can be very local and do not necessarily document an equilibrated thermal state (Lescoutre et al., 2019). Therefore, such data must be combined with thermochronological data that need to be sampled in a systematic way across a basin. Both techniques have been extensively used in the Pyrenean-Cantabrian system and the results have been reviewed by Clerc and Lagabrielle (2014), Clerc et al. (2015), Vacherat et al. (2016), Ducoux et al. (2019) and Saspiturry et al. (2020b). The main result is that a HT/LP event can be linked to the hyperextension and occurred between 110 and 90 Ma. However, T_max values are highly variable along the Pyrenean-Basque-Cantabrian system and can be >550 °C (Clerc et al., 2018). Such high T_max values occur, however in sediments that often occur together with granulites and/or mantle rocks and are overlain by thick syn- and post-rift sedimentary successions. Evidence for hydrothermal fluids, and to a lesser extent for magmatic activity, correlate with the distribution of high T_max values that occur along the so called Internal Metamorphic Zone (IMZ) in the Northern Pyrenean zone, as well as the Nappe de Marbres in the Basque-Cantabrian Basin.

9.2 The hyperextended basins in the Pyrenean-Basque-Cantabrian domain: anomalously hot?

Clerc and Lagabrielle (2014) suggested the existence of an anomalously “hot” rift segment in the eastern Pyrenees, while the lower T_max values in the Western Pyrenees were interpreted as a “cold” rift segment. However, Ducoux et al. (2019) showed that T_max values in the Nappe de Marbres in the west are similar to those observed in the eastern Pyrenees, questioning the postulated decreasing thermal gradient from east to west. Lescoutre et al. (2019), Saspiturry et al. (2020b) and Ducoux et al. (2021b) show that very high temperatures and thermal gradients can also be retrieved from the Mauléon basin, suggesting a similar thermal evolution as in the other basins further to the east and to the west, clearly arguing against the initial “hot” and “cold” rifts model. Saspiturry et al. (2020b) published a comprehensive summary of the thermal data from the Mauléon basin, showing a change in the thermal gradient across the basin (Fig. 8). These authors suggested that the thermal structure equilibrated quickly (i.e., the isotherms in the basin were subparallel; see Figure 13 in Saspiturry et al., 2021), which contrasts with the interpretation from Vacherat et al. (2014) who assumed that equilibration was slow and lasted until about 50 Ma. Lescoutre et al. (2019) proposed an alternative model that was able to predict the asymmetric thermal gradients published by Saspiturry et al. (2020b) and Ducoux et al. (2021b). In their model shown in Figure 8, the thermal state is mainly controlled by the final rise of the asthenosphere during the asymmetric hyperextension and mantle exhumation phases, resulting in a local, asymmetric thermal anomaly instead of the more regional, basin-wide thermal anomaly suggested by Saspiturry et al. (2020b). Thus, from the available interpretations three questions arise: (1) was the thermal event basin-wide or was it localized? (2) was it anomalous relative to other rifted margins? and (3) did the thermal state control the crustal rheology during rifting? A last question, not further discussed in this paper concerns the contradictory record of the post-rift thermal relaxation, which seems to be mid-Cenomanian from a sedimentary point of view (Issautier et al., 2020; Saspiturry et al., 2019; Razin, 1989), and Late Cretaceous to Eocene from a thermal point of view (Saspiturry et al., 2020b; Vacherat et al., 2014).

The thermochronological data presented by Masini et al. (2014), Hart et al. (2016, 2017) and Odlum and Stockli (2020) showed that basement rocks derived from the hyperextended domains were not reset during rifting, suggesting that the high geothermal gradient proposed by Lagabrielle et al. (2016) and Corre et al. (2016) was late and/or local, or that these authors overestimated the thermal gradient (see Figure 11 in Lescoutre et al., 2019 for an explanation). Vacherat et al. (2014) showed a late thermal equilibration for the most distal parts of the margin, suggesting that the thermal structure did not begin to equilibrate before 50 Ma and that the hottest geothermal gradient was localized in the hyperextended basins. Further support of a localized thermal event is the very local occurrence of the HT/LP values. However, it remains unclear whether they originate from the upper-lower plate transition as proposed by Lescoutre et al. (2019) or equilibrated and affected the whole basin as suggested by Saspiturry et al. (2020b) (Fig. 8).

Determining whether the measured T_max in the Pyrenean-Cantabrian domain points to an anomalously hot geothermal gradient during rifting and exhumation, or whether it can be considered as normal and comparable to that of young rifted margins is more difficult. There are two points that need to be recalled: 1) it is difficult to compare T_max data from the Pyrenees-Cantabrian domain with those from present-day rifted margins since few pre-rift sequences have been drilled in distal margins, and 2) T_max values are not allowing to determine the heat flow history of a margin, unless it can be assumed either that the margin was constantly thermally equilibrated, or that the thermal evolution was simple and controlled by depth-uniform lithospheric thinning. However, if T_max values are local or linked to a complex thermal evolution, including hydrothermal systems, magma and very high sedimentation rates, the values do not need to be linked to anomalous heat flow.

To evaluate whether the Pyrenean-Cantabrian domain had an anomalously hot geothermal gradient, the best places to compare with are the fossil remnants of the Alpine Tethys preserved in the Central Alps and that escaped an orogenic thermal overprint, like the Pyrenean counterparts. In the hyperextended and exhumed domains exposed in the Alps, one cannot find similar T_max values like those reported from the Pyrenees Incerpi et al. (2020b). There is, however, a fundamental difference between the distal margins in the Alps and the hyperextended basins in the Pyrenean-Cantabrian domain: while the former was sediment-starved, the latter show kilometre-thick syn-rift sequences. To explain values of 300 °C at a depth of 1 km in the absence of important sedimentary burial, the thermal gradient should be 300 °C/km, which is only possible if the heat flow is locally controlled by hydrothermal systems (see also Figure 6 in Lescoutre et al., 2019). Incerpi et al. (2020a, 2020b) analysed the diagenesis of pre- and syn-rift sediments from both the Central Alps and the Pyrenees using fluid inclusions, isotopes and U/Pb dating of pre-hyperextension carbonates from both the distal Tethys margins exposed in the Alps and the Mauléon basin. Their results show that the fluid history as well as the related temperatures were very similar between both sites, suggesting that both sites had similar high heat flow during hyperextension and mantle exhumation. Thus, the high T_max values from the Pyrenean-Cantabrian domain are certainly linked to a high heat flow during hyperextension, which is to be expected in (sediment-rich) hyperextended domains and predicted by numerical modelling (e.g., Lavier et al., 2019; Lescoutre et al., 2019; Duretz et al., 2019). Thus, while the heat flow may have been similar to that of other hyperextended and exhumed mantle domains, the high T_max values may have been anomalous in the Biscay-Pyrenean domain. If the basal heat flow would have been anomalous, one would have expected substantial production of tholeiitic magma, which is not observed. In contrast, the high T_max values may best be explained by the combination of high heat flow, high sedimentation rates in combination with hydrothermal activity during hyperextension and mantle exhumation.

To answer the question if we can use the T_max and a paleo-geothermal gradient from the distal part of the rift basin to assess the rheology of the crust during rifting, it is important to analyse the syn-kinematic minerals and textures linked to the fault rocks accommodating extension and exhumation. In the Mauléon basin, Corre et al. (2016) and Saspiturry et al. (2021) suggested that the hyperextended domain deformed in a dominantly ductile regime. Incerpi et al. (2020a) described calcite mylonites of syn-rift age suggesting that deformation may have been accommodated by ductile processes in carbonate-rich lithologies. Jammes et al. (2009) and Masini et al. (2014) described cataclasites and gouges along the Northern and Southern Mauléon Detachment faults in the Mauléon basin. These fault rocks indicate that final crustal and mantle exhumation occurred in the brittle field. This is also supported by the thermochronological data of Hart et al. (2016, 2017), which supports that hyperextension occurred at temperatures below 300 °C within the hyperextended crust, consistent with the occurrence of serpentinites and ophicalcites and linked isotope data (Clerc, 2012; DeFelipe et al., 2017). Thus, whether rifting and hyperextension occurred in the brittle or ductile regime remains at present debated. However, it is important to mention that hyperextension has been defined as the domain where the crust is in the brittle regime (Sutra et al., 2013). Moreover, numerical models show that exhumation of subcontinental mantle is difficult in the presence of ductile material since it prevents faults to penetrate into the mantle and exhume it (Manatschal et al., 2015). Moreover, ductile crustal rheologies and high thermal gradients tend to result in wide rift styles (Buck, 1991; Brune et al., 2017), which is not observed in the Biscay-former Pyrenean system. One may also note that the rift structures preserved offshore along the northern Iberia and the conjugate Armorican margins do not show any sign of either anomalous hot or ductile crust (Montadert et al., 1979; Thinon et al., 2003).

Fig. 8 Comparison of two published thermal models for the Mauléon basin: (a) the model from Saspiturry et al. (2020b) and (b) the model from Lescoutre et al. (2019). The thermal gradients at the top of the two models are from Saspiturry et al. (2020b) (see text for details and discussion).

10 The granulite conundrum

Among the most heavily debated topics in the OROGEN project was, and still is, the lithological architecture and thickness of the crust and lithosphere prior to rifting. The reason is that it has a major impact on the interpretation of the subsequent rifting thermal and isostatic evolution and impacts the kinematics of detachment faults. A key issue in these debates, is the crustal position of the granulites prior to rifting. Since different interpretations have been proposed, all of which have major implications on the rift evolution, we refer to this problematic as the “granulite conundrum”, a topic that is in the heart of the discussion about inheritance (see also Cadenas et al., 2021).

Granulites occur throughout the Biscay-Pyrenean domain. They are often linked with exhumed mantle rocks and high T/low P pre-rift sediments. Granulites can be found either as outcrops flooring thick Apto-Cenomanian syn-rift sediments, or as clasts in breccias within these sediments. There is a consensus that the granulites formed during the Carboniferous at a late stage of the Variscan orogeny (Vielzeuf and Pin, 1989), although the exact conditions and geodynamic setting of their formation remains unclear. Indeed, the crustal depth at which the granulites were before the onset of rifting and how and when they were exhumed to the seafloor remain debated. Three different models have been proposed (Fig. 9):

• Clerc and Lagabrielle (2014) proposed a scenario (model 1 in Fig. 9) where (re)granulitisation and exhumation occurred during the Cretaceous rifting. This model was based on structural observations in the Agly massif and implies very high thermal gradients during rifting and a fast exhumation within a dominantly ductile crust.

• Saspiturry et al. (2019) suggested (model 2 in Fig. 9) based on observations from the Basque massifs that the granulites were exhumed to shallow levels already during the Permian via the formation of a core complex. This model was based on field observations and structural observations (see Fig. 4 (Sect. 6) and Figure 5s in Saspiturry et al., 2019). In this model, the granulites can be considered as belonging to the pre-rift upper crust and do not carry any information about rifting and the related geothermal gradient.

• Jammes et al. (2009), Masini et al. (2014) and Hart et al. (2017) proposed (model 3 in Fig. 9) that the granulites were in the upper middle crust (at about 10–15 km assuming an equilibrated thermal gradient of 30 °C.km⁻¹) before rifting and were exhumed to the seafloor during the Cretaceous rifting via extensional detachment faults. This was based on Ar/Ar ages on biotite and other thermochronological data (Masini et al., 2014; Hart et al., 2017).

Models 1 and 3 imply that granulites had to be exhumed in the footwall of a rift-related exhumation fault and that their top basement corresponds to an exhumation fault (e.g., Tb_f; see discussion in Sect. 6.1; Eq. (4)). The difference between models 1 and 3 is that in model 1 exhumation occurs in the ductile field, while in model 3 the granulites were already in the brittle field at the onset of rifting, as in model 2. The main difference between models 2 and 3 is that in model 2 the granulites need to be in the hanging wall of an older detachment fault thinning the crust since they were already exposed. In contrast, in model 3 they are in the footwall of a detachment fault. Although the three models are mutually exclusive for a given site, further research will be necessary to determine whether all granulites have the same origin and exhumation history at the scale of the entire Biscay-Pyrenean domain (see also Cadenas et al., 2021).

Some inconsistencies appear when comparing new thermochronological data with the three models. For instance, the results of Odlum and Stockli (2020) do not support the (re)granulitisation during Cretaceous rifting in the Agly massif. The assumption that the granulites were already exhumed near the surface in a core complex-type structure in Permian time is neither supported by clasts of granulites reworked in the Triassic sandstones, nor by the thermochronological data of Hart et al. (2016, 2017) and Masini et al. (2014).

It is also interesting to compare models 2 and 3 (see Fig. 9) and to explore their implications. Both are based on observations made at the western Mauléon Basin. Model 3 assumes that the granulites were exhumed during rifting and need therefore to be in the footwall of a major north dipping extensional detachment (e.g., Tb_f), which is compatible with the kinematic data determined on fault rocks from an exposed detachment fault in the Mauléon basin (Masini et al., 2014). In contrast, model 2 places the granulites in the pre-rift upper crust (e.g., Tb_{MH_initial}), i.e., in the hanging-wall of a more symmetric basin with the main detachment dipping towards the south. This explains why the two different interpretations of the pre-rift position of the granulites results in different rift models for the hyperextended Mauléon Basin.

Cadenas et al. (2021) compared the extensional structures and the tectono-stratigraphic architecture of related syn-rift sediments exposed in the Labourd massif (Basque Massifs) with those seismically imaged and dredged offshore at the Le Danois High (central North Iberian margin). The offshore-onshore comparison shows striking similarities and is in line with an exhumation of the granulites at the Le Danois High in a brittle regime, along the footwall of a seismically imaged extensional detachment system. The authors suggest that exhumation of granulites occurred during Aptian to Cenomanian time along a north-dipping extensional detachment systems covered by syn-rift breccias at both the Biscay and the Pyrenean rift systems. Finding a definitive answer to the granulite conundrum will provide further constraints to discriminate between the two rift models and will also provide further constraints on the thermal structure of the hyperextended basins in the Biscay-Pyrenean domain.

Fig. 9 Three competing models to explain the “granulite conundrum” in the Biscay-Pyrenean domain. See text for a more detailed description of the models and their implications.

11 Reactivation of hyperextended segmented basins: a new emerging concept

One of the main results in studying magma-poor rifted margins and hyperextended rift basins is the recognition, characterization and mapping of rift domains referred to as proximal, necking, hyperextended, exhumed and oceanic domains (Sutra et al., 2013; Tugend et al., 2014). Each of these domains is characterized by its crustal nature, characteristic structures, total accommodation space and rheological profile (see Fig. 10; Lescoutre and Manatschal, 2020). Tugend et al. (2014) and Chenin et al. (2017) suggested a direct link between rift and orogenic domains, an assumption that is further supported by the results of the OROGEN project.

The systematic study of the central North Iberian margin, the Basque-Cantabrian Basin and the western Mauléon Basin documented in Cadenas et al. (2020), Lescoutre and Manatschal (2020), Miró et al. (2021), and Gómez-Romeu et al. (2019) (Fig. 6) has enabled to describe the role of rift inheritance during reactivation of these segmented hyperextended rift systems. Convergence at the North Iberian margin initiated in the exhumed mantle domain using a decoupling level in the serpentinized mantle. In the more reactivated domains (Basque-Cantabrian and Mauléon Basins), it is difficult to document the initial stage of convergence, in particular at segment centres. Rocks derived from the exhumed mantle and hyperextended domains, including exhumed mantle, granulites and pre-rift sediments with a HT/LP overprint, are only locally preserved in syn-rift sedimentary breccias. The occurrence of these breccias is a fingerprint that enables to argue for the presence of these domains in the former rift section. Following the inception of reactivation, shortening can take three different avenues: (1) shortening continues but the decoupling level changes (e.g., central section in the Basque-Cantabrian basin; Fig. 6); (2) shortening creates new structures preserving former rift structures (e.g., eastern section across the Mauléon-Basque massifs; Fig. 6) or (3) shortening jumps to a new location (e.g., Cantabrian section; Fig. 6). While the last example is complex and needs further studies, the first two examples may be representative for the reactivation of segment centres and boundaries, respectively. In the next two sections we propose a conceptual model to explain the sequential reactivation of segmented rift domains that is based on the studies of Tugend et al. (2014), Lescoutre and Manatschal (2020) and Miró et al. (2021).

Fig. 10 Conceptual model illustrating the in-sequence reactivation of rift-inherited decoupling levels at a segment centre and explaining the transition from subduction to soft collision, to hard collision, and to final late collision (for details and discussion see text). Modified from Lescoutre and Manatschal (2020) and Miró et al. (2021).

11.1 Rift-inherited decoupling levels and their sequential reactivation: a conceptual model

In an idealized schematic section though a magma-poor rifted margin, each rift domain is defined by a crust with a different shape, thickness, composition, and rheology (Fig. 10). The oceanic domain is made of idealized Penrose magmatic crust coupled to the underlying mantle. It terminates continent ward at the breakup point (BP), which marks the transition to the domain of exhumed mantle and hyperextended crust. The exhumed mantle-hyperextended crust domain is characterized by hydration processes that become dominant within the uppermost 3 km below seafloor and result in massive serpentinization. The effect of serpentinization results in an efficient decoupling level that is easy to reactivate. Depth and rheological characteristics of serpentinized mantle are described in Gillard et al. (2019). The exhumed mantle domain is characterized by a rugose top basement with basement highs and local magmatic additions, and the hyperextended domain by the occurrence of a wedge of hydrated continental crust (Nirrengarten et al., 2016). The continent-ward limit of the hyperextended crust corresponds to the oceanward limit of the necking zone. It is defined to occur where the crust is ca. 10 km thick and is named the Coupling Point (CP) (Sutra et al., 2013) (Fig. 10). The CP plays a key role during reactivation, since it juxtaposes crusts with different properties: On the oceanward side the crust is thin, hydrated, and brittle and it overlies serpentinized mantle. On its continent-ward side crust is thicker, includes residual ductile levels in quartz-rich layers, and is coupled to the underlying, non-serpentinized mantle. The CP marks also the limit across which potential decoupling levels are changing. While on its oceanward side serpentinized mantle or inherited low-angle extensional detachment faults can be reactivated, these decoupling levels do not exist on its continent-ward side. There, decoupling can occur either in ductile crustal levels or within the salt or clay-rich levels on top of the crust. Last but not least, the CP also juxtaposes thin crust that can be subducted against thicker, more buoyant crust.

The results of the OROGEN project support the idea that subduction initiates in the exhumed mantle domain (see also Dielforder et al., 2019). Data shows that the exhumed mantle and hyperextended domains can be subducted during initial soft collision, which explains their absence (except for small blocks) in the reactivated segment centres. Since the necking domain is made of >10 km thick crust, too buoyant to be subducted, the major change in collision occurs when the CP starts to be involved in collision. This event marks the transition from soft to hard collision. In both the Basque-Cantabrian and Mauléon basins, hard collision is controlled by the necking zone that acts first as a buttress. Subsequent shortening either reactivates ductile levels within the residual continental crust, resulting in a thick-skinned reactivation, or reactivates continuous low friction levels in the sedimentary section (e.g., salt), resulting in a thin-skinned reactivation. Thick-skinned reactivation results in the stacking of crustal nappes that induces crustal thickening, creation of orogenic topography, erosion and exhumation. In contrast, thin-skinned deformation results in a nappe stack that consists of supra-salt units only (Jourdon et al., 2020). Combinations of thick- and thin-skinned reactivation can also be observed. Thus, the presence or absence of salt is key in controlling the reactivation during hard collision (Miró et al., 2021). In the presence of salt, a complete decoupling between the supra-salt and sub-salt sequence occurs, resulting in allochthonous convergence (e.g., section across the Basque-Cantabrian basin). During final collision, shortening is accommodated along thrust systems that are ramping up towards the foreland, often reactivating former exhumation faults. The occurrence of rift basins in the stretching domain can result in local complexities during final convergence, but on a first-order, final collision can be described with the Coulomb Wedge theory, indicating that rift inheritance plays at this stage a subordinate role. Collision stops and/or reorganizes either when the forces available to accommodate shortening are inferior to the bulk yield strength of the deforming lithosphere, or when the differential gravitational potential energy between the orogen and the adjacent foreland becomes too high. Both situations occur when the proximal domain starts to be involved into the collisional process. The sequential reactivation of rift domains, as discussed in this section and shown in Figure 10, can be found in segment centres (e.g., section through the central Basque-Cantabrian; Fig. 6).

11.2 Reactivation of overstepping en-échelon segment boundaries

A more complex reactivation of rift domains is observed at the Pyrenean-Cantabrian junction. This domain has been re-interpreted by Lescoutre et al. (2021) as an en-échelon array formed by the overstepping terminations of the Basque-Cantabrian and Mauléon Basins. As shown by Lescoutre and Manatschal (2020), decoupling levels used in segment centres (see previous section) are forced to produce shortcuts at segment boundaries in order to link up with the main decoupling levels in the adjacent segment center (Fig. 11). Such shortcut structures at segment boundaries result in a change of the exhumed and hyperextended rocks from a footwall position to a hanging wall position in the main thrusts accommodating shortening. This explains the preservation of inherited rift structures, including detachment surfaces, exhumed crustal and mantle rocks in the nappe stack. Thus, in contrast to segment centres where rift-related detachment faults are reactivated and most of the exhumed mantle and hyperextended crust are subducted, rift structures and exhumed material are preserved during reactivation of segment boundaries. This is well documented in the Labourd massif, where rift-related detachment faults and the related crustal section are preserved (Masini et al., 2014; Lescoutre et al., 2019).

It is interesting to note that the preservation of the rift structures coincides with the occurrence of a high density/velocity mantle body imaged by potential field and tomographic methods (Jammes et al., 2010; Martin et al., 2020; Wang et al., 2016). A likely interpretation is that this body corresponds to mantle rocks exhumed at or near the seafloor during hyperextension (see also results of the Maupasacq experiment; Lehujeur et al., 2021). The preservation of such a high velocity body within a shallow crustal level in the orogen linked to a segment boundary is compatible with the model shown in Lescoutre and Manatschal (2020) (see also Fig. 11). High density bodies imaged by positive gravity anomalies occur at several other places along the Biscay-Pyrenean system (e.g., Parentis, Basque-Cantabrian Basin, and Saint Gaudens; Martin et al., 2020) but further studies will be necessary to determine whether their presence is linked to the reactivation of rift segment boundaries.

Other observations that may help to define rift-inherited segment boundaries have been discussed in Lescoutre et al. (2021) and are shown in Figure 11. A striking observation at the Pyrenean-Cantabrian junction is the along-strike change of structures in the internal part of the orogenic section. While at segment centres the orogen is defined by the pull-up of the sediment infill from the former hyperextended basin (for details see previous section), at segment boundaries it is dominated by basement forming a keystone. This keystone contains preserved rift structures, and upper crustal- and locally also exhumed lower crustal rocks, both of which are bounded by symmetric retro- and pro-thrust faults. The along-strike transition between rift segment center and boundary is marked by thrusts that reactivate former extensional detachment faults. These thrusts carry sediments in their hanging wall (e.g., Lakoura thrust and Licq fault) and place younger rocks on older rocks. In contrast, thrusts in the keystone such as the Roncesvalles fault (see Figure 1 in Lescoutre et al., 2021) show the classical old over young relationship. Therefore, mapping and correlating thrust structures from one rift segment towards the adjacent segment is challenging. It may explain why the Pyrenean-Cantabrian junction remains one of the most debated parts in the Pyrenean system. Similar situations as that described from the Pyrenean-Cantabrian junction, may occur elsewhere in the Biscay-Pyrenean system as discussed in Cadenas et al. (2021).

Fig. 11 Reactivation of overlapping en-echelon rift segments based on the example of the Pyrenean-Cantabrian junction showing the importance of shortcutting structures and their role in preserving rift structures (modified from Lescoutre and Manatschal, 2020). (a) Schematic 3D evolution of the Pyrenean-Cantabrian junction highlighting the structural evolution during the subduction (Late Santonian-Paleocene) and collisional (Eocene-Miocene) phases. (b) Rift domains map of the Pyrenean-Cantabrian junction. (c) Schematic kinematic restoration of the rift domains map (panel b) at Late Cenomanian (end of hyperextension), Late Santonian to Paleocene (early convergence), and Eocene to present-day (late and post convergence) stages. The locations of the cross-sections shown in panel d are represented. (d) Cross sections across the Pyrenean-Cantabrian junction (see panel c for location) showing the evolution for the Late Cenomanian (end of hyperextension), the Late Santonian to Paleocene stage (early convergence), and the Eocene to present-day (late- and post-convergence) stages. Abbreviations used in the Figures: BCB: Basque-Cantabrian Basin; MB: Mauléon Basin; SPF: South Pyrenean Fault; NdM: Nappe des Marbres; Lak: Lakoura thrust; Gav: Gavarnie thrust; Ronces: Roncesvalles thrust; NPFT: North Pyrenean Frontal Thrust; NPF: Northern Pyrenean Fault.

12 The lessons learned from the OROGEN project

12.1 What’s new and what remains debated?

The OROGEN project provided access to new field and seismic reflection data, better and higher resolution tomographic imaging, and new thermochronological and thermal data that enabled to propose, test and calibrate new interpretations for the tectonic evolution of the peri-Iberian system. Here we focus on the role of inheritance within the incomplete Wilson cycle recorded by the geology of the Biscay-Pyrenean system. Although at this stage the results do not allow us to unravel in detail how inheritance controlled the evolution of this system, some new ideas and concepts emerged from the OROGEN project. This includes a new definition of inheritance that includes “interface shape inheritance” and its application to the Biscay-Pyrenean system (Sect. 7). Other new results include the kinematic restoration of sections back to the pre-rift stage conserving crustal area and pre-rift marker horizons length (Sect. 6). Moreover, the in-sequence reactivation of rift inherited decoupling levels, and the antagonistic behaviour of rift segment centres and boundaries during reactivation enable to propose a refinement of previously published concepts explaining the reactivation of segmented rifted margins (Sect. 11). On a larger scale, the new interpretations of the kinematic evolution of the Europe-Iberia plate boundary across the Biscay-Pyrenean system suggest a multistage rift system. This conclusion has far-reaching implications not only for the local and regional interpretations, but also for the understanding of the Alpine and Mediterranean domains that are kinematically linked with the Atlantic system through the Iberian plate.

Although the concepts of hyperextension and mantle exhumation have been well established at present-day magma-poor rifted margins and the fossil Alpine Tethys margins, these classical models have been challenged within the OROGEN project. At present, mutually exclusive models exist to explain the hyperextended systems of the Pyrenean-Cantabrian domain (Sect. 6.2). The competing models differ in the polarity of faults, the amount of asymmetry and the style of deformation, implying different thermal and rheological evolutions that have first-order implications on the role of inheritance (Sect. 9.2). While one group of models suggests high geothermal gradients during rifting and a thermally equilibrated stage at the end of rifting, the other group of models suggests high geothermal gradients at the end of rifting (and as a result of rifting), as well as a thermal state that is asymmetric and not equilibrated at the end of rifting. At this stage it is too early to reject one of the models, especially because the existence of competing models generates debates that are necessary for prolific research. A way to test and sort between the two groups of models is to solve the granulite conundrum, i.e., to understand the position and exhumation history of the granulites before and/or during rifting (Sect. 10).

12.2 Assessing the role of inheritance: what are the problems?

The Biscay-Pyrenean system is one of the few sites worldwide that allows to address the importance of inheritance at different stages within an incomplete Wilson Cycle. Advantages of this system include excellent datasets, the long history of investigation on the stacking of three main tectonic events, namely a Variscan orogenic system that has been overprinted by a multistage Mesozoic rifting, which was in turn reactivated in its easternmost parts, leading to the development of the Pyrenees. That only few Variscan structures have been used or reactivated by Mesozoic rift structures in the Pyrenean domain suggests that the former orogenic inheritance did not systematically control later rifting, contrary to common belief. While prominent first-order structures such as the Pamplona or Leiza faults have been regarded as inherited and/or reactivating Variscan and/or Permian structures in the past, new studies question the existence of the Pamplona fault and reinterpret the Leiza fault as a second-order structure (Lescoutre et al., 2021). Thus, assessing the role of inheritance is complicated and cannot only rely on comparing strike directions or on geological intuition.

Another learning from the OROGEN project is that it is not only important to determine the initial conditions, i.e., the genetic code of a tectonic system, but also to understand how it interacts and evolves during subsequent events. Understanding why some structures have not been reactivated is as important as understanding why other structures have been used during subsequent events. It is important to realize that structural, compositional, and thermal inheritance does not always control subsequent events, as demonstrated by the few reactivated Variscan structures in the Biscay-Pyrenean domain. Thus, defining the role of inheritance is not only about defining and mapping inheritance, but requires also determining what type of inheritance controls what, how, when, where and at what scale.

While some of the rift- or orogenic system complexities can be elegantly explained by integrating inheritance, the danger is that inheritance is used as a “deus ex machina” to explain whatever cannot be understood with classical models. Thus, the use of inheritance must be based on a foundation that is coherent with what we know of a geological system. The aim is not to find solutions by replacing unknown by inheritance. Conversely, we need to be even more aware of potential inconsistencies, incoherencies and shortcuts that may be based on the misapplication of some concepts.

12.3 Implementation of “inheritance” in conceptual, analogue, and numerical models

Numerical and analogue models commonly try to explore the physical processes of systems by defining boundary conditions and investigating the parameter space. Each model relies on initial conditions that are fundamental but difficult to define in geological systems. In most models, the initial conditions are simple layer-cakes with either one or several inherited features (e.g., weak seeds). Thus, models also include “inheritance” and their initial conditions are actually not so different from those we assume for the Biscay-Pyrenean system at the onset or rifting. Thus, so-called “accordion” models presented by Jammes et al. (2014) may be a good way to model incomplete Wilson Cycles, since they develop more appropriate initial conditions for the orogenic reactivation, including thinned crust and lithosphere prior to convergence (i.e., “interface shape inheritance”). They may also account for the main decoupling horizons such as hydrated levels, remnant ductile levels or salt (e.g., Jourdon et al., 2019, 2020). This may explain why 2D numerical models are able to reproduce first-order structures observed at rifted margins and orogenic systems (e.g., keystone in extensional systems and orogenic prism in collisional orogens), as well as the in-sequence reactivation observed at segment centers of hyperextended basins. If one compares models using slightly different initial conditions but the same boundary conditions and physical parameters, the major differences arise at the initial stage while the final results are comparable. Thus, it appears that at a mature stage a tectonic system depends little on the initial conditions, but more on the applied boundary conditions and physical parameters. This suggests that, after an initial stage of self-organization, a tectonic system evolves into a “steady-state system”. Thus, when modelling tectonic systems, a choice may have to be made: (1) either a model tries to include acceptable initial conditions in order to test the initial stages of a system, or (2) models can only be interpreted for structures that are established at the moment when the initial conditions are not controlling anymore the model results.

Another important point when designing models is to define the controlling initial perturbance. Is it thermal, structural or compositional or a combination of the three? A viable way to test the role of inheritance may be to compare models with different initial heterogeneities (seeds) and to compare their initial evolution and to test at what stage they converge and start to resemble. This would allow to define the transition from a situation where the experiment is controlled by the initial conditions/inheritance to a stage when it is controlled by the boundary conditions and applied physical parameters. From our observations, such a stage could correspond when the Coupling Point (CP) get involved in convergence, i.e., the transition from soft- to hard collision. More generally, a key question is at what stages of a Wilson Cycle what type of inheritance is the most important. The observations made in the OROGEN project suggest that the main rift inheritance controlling the reactivation of a margin relates to the distribution of decoupling levels. Reactivation initiates outboard in the hydrated mantle and migrates inboard to ductile mid crustal levels and/or weak low frictional layers in the sedimentary cover. However, in 3D, rift architecture and in particular rift segmentation appear to be a first-order control in explaining along-strike non-cylindrical structures and complexities. This suggests that integrating the third dimension in numerical models is a prerequisite to understand the tectonic evolution of convergent or divergent boundaries.

12.4 Emerging concepts and applicability to global systems

It is likely that the orogenic evolution of systems that went through a full Wilson Cycle, i.e., the closure of a mature and wide oceanic system, is controlled by long-lasting subduction-related magmatic arc systems and slab dynamics and cannot be compared with the Biscay-Pyrenean system. Thus, what can we learn from orogens that result from incomplete Wilson Cycles, i.e., the closure of narrow/immature hyperextended domain, and how can these learnings be applied to the mature orogenic systems? One of the major learnings is that in the Biscay-Pyrenean system we can mainly observe the initial stages of subduction and collision. Since all orogens, even the mature ones, had to initiate, these are precisely these early stages, often neither defined nor observed or understood that are important and where our learnings can be applied. For orogens where collision occurs after a mature, long-lasting subduction, the hanging wall may be completely obliterated by arc magmatism, while such a system does not exist in a Pyrenean-type system. However, the incoming lower-plate margin may be reactivated under similar conditions like those observed at the Biscay-Pyrenean system. Thus, some of the learnings from the OROGEN project, such as the in-sequence reactivation of rift-inherited decoupling levels or the role of segment centers versus boundaries may be applicable to more mature systems. This is well shown by Tavani et al. (2021) who successfully applied the in-sequence concept initially proposed by Tugend et al. (2014) and Lescoutre and Manatschal (2020) to the Apennine, Oman, Zagros and Taiwan. Other examples where the learnings of the OROGEN project may be applied are the Alps in Western Europe and the Oriental Cordillera in Colombia.

12.5 Recognizing remnants of former rifts and rifted margins in orogens

Our study shows that a prerequisite to investigate the role of rift inheritance and predict its effects is the mapping of rift domains and its comparison with the along strike variability and/or non-cylindricity of internal parts of orogens. Two cases may be found: at former segment centres a complete reactivation of the former margin is likely, and most of the former distal margin is subducted. In such systems it is important to evaluate the decoupling levels and to see if their sequential reactivation can be observed. Key questions that need to be addressed include: Are there hints for exhumed mantle and/or lower crust material reworked in syn-rift breccias? Are there decoupling levels, or remnants of characteristic fault rocks such as gouges or cataclasites related to such structures? Is the accretionary prism formed by syn-rift sediments that contain tectono-sedimentary breccias sampling remnants of the former distal margin? Moreover, can a change in the nature of the decoupling level be observed at the transition from soft- of hard collision and is shortening focused in the ductile crust and/or in the low-friction sediment layers?

In contrast, at potential segment boundaries one would expect the preservation of rift inherited basement structures, including lower crustal and mantle rocks remaining in the nappe stack. In such examples one can expect the preservation of exhumation faults and related diagnostic fingerprints, which are exhumed basement overlain by tectono-sedimentary breccias reworking the underlying footwall (see Epin et al., 2017). The along-strike changes may be abrupt and difficult to map and understand. Thrusts may appear to stop abruptly and/or show a change from old over young polarity to young over old along one and the same thrust plane. Structures where sediments are transported along thrusts over basement, including tectono-sedimentary breccias reworking the underlying basement at their base are reminiscent of reactivated detachment faults. Examples have been described by Epin et al. (2017) from the Alps. The local geological complexity due to reactivation of former rift structures, in particular of extensional detachment systems, is difficult to understand, since it is local and only occurring in the more internal parts of orogenic systems. In contrast, at external parts structures appear to be more cylindric. In the Pyrenean example, such local complexities have either been ignored or interpreted as Variscan, post-Variscan or Alpine structures. Only very recently, such structures have been related to the former Mesozoic rifting. Most geologists working in orogens still have problems in defining rift-related inheritance and in particular former, often reactivated extensional detachment faults. In the Alps, Beltrando et al. (2014) and Epin et al. (2017) defined diagnostic “fingerprints” to identify such structures and discussed how they are reactivated. Many present-day debates in the Pyrenees are linked to the non-recognition or erroneous interpretation of inherited structures. Although it is important to be careful and not to use and abuse inheritance, ignoring it does not make interpretations better. Missing inherited structures and interpreting them as Alpine structures results in erroneous interpretations. In most cases, recognizing rift-inherited structures makes the apparently complex orogenic structures simpler (e.g., Beltrando et al., 2014). A key is to understand that hyperextended systems cannot be approximated by simple layer cakes, as was often done in the past.

13 Concluding remarks

Even if the main aim of this paper was to define the role of inheritance in the Biscay-Pyrenean system, the results are of broader importance and the ultimate aim is to transfer the learnings established in the OROGEN project to other, less explored hyperextended rift systems, magma-poor rifted margins and Alpine-type collisional mountain belts. The main lessons learnt from the Biscay-Pyrenean system that may be applied elsewhere include:

• the sequential reactivation of rift-inherited decoupling levels;

• the importance of the Coupling Point as a main, mappable boundary that may become the S-point sensu Willett et al. (1993) of a collisional system at the transition from soft- to hard collision;

• distinguishing between competing decoupling systems (salt/clay versus middle crust) that are at the origin of thin- versus thick-skinned reactivation;

• complexities at segment boundaries creating non-cylindrical structures in internal parts of orogens;

• along-strike variations from inherited reactivated structures to newly formed structures with changing hanging-wall and footwall.

It appears that the thermal state of a margin is less important than previously thought for the localization of subduction but may control the flexural behaviour of the lithosphere during collision (Mouthereau et al., 2013). In contrast, it seems that serpentinized mantle is a key in controlling onset of subduction. This may be explained by the low friction at its top contrasting with the strength of the underlying non-serpentinized mantle, and its high density comparing to the surrounding crustal rocks.

Last but not least, trying to integrate the role of rift inheritance in the understanding of orogens may also imply to change the approach used in the field and in interpreting datasets. While detailed structural analysis assuming layer cake geology may work for post-rift sequences/at external parts of the orogen, such methods are more difficult to use in distal margins/internal parts of orogens, where the inherited geological template is more complex.

Acknowledgments

This study was funded by the Orogen project, a tripartite joint academic-industry research program between the CNRS, BRGM, and Total R&D Frontier Exploration program. We thank Tiago Alves and an anonymous reviewer for their very constructive comments that helped to improve the manuscript as well as the Associate Guest Editor Stefano Tavani and the Guest Editor Olivier Lacombe for handling the manuscript.

References

All Figures

Fig. 1 (a) Main stages (rifting, seafloor spreading, subduction, and late orogenic collapse) within a Wilson Cycle are illustrated by the shapes of the three main interfaces, which are the Earth surface, the Moho and the base of the lithosphere. (b) Idealized lithosphere made of a thermally equilibrated layer cake versus a “real” post-orogenic lithosphere showing inherited structural and compositional complexity (modified after Manatschal et al., 2015). (c) Schematic evolution of the thermal state and gravitational potential energy depicted by the base of the lithosphere (1300 °C isotherm) and the surface topography within a Wilson Cycle. p.-o. ext.: post-orogenic extension.

In the text

Fig. 2 (a) Map of the Mesozoic rift basins between France and Spain along the Iberian-European plate boundary. Modified after Lescoutre et al. (2021) and Masini et al. (2014). (b) Rift domain map of the Pyrenean-Cantabrian area. STS: Santander Transfer System, W: Western section, C: Central section; E: Eastern section. Modified from Tugend et al. (2014) and Cadenas et al. (2020). CC: continental crust; OC: oceanic crust; OCT: ocean-continent transition.

In the text

Fig. 3 (a) Major orogenic domains and structures in the North Atlantic realm (modified from Chenin et al., 2015 and Chenin 2016). (b) Map showing (i) the rift domains and corresponding rifting duration (i.e., the time lag between necking and breakup), as well as the distribution of breakup-related magmatic additions for the North Atlantic rift system; and (ii) orogenic inherited structures and magmatic additions onshore Western Europe and eastern N-America (from Chenin 2016). (c) Map showing the correlation of parts of the Variscan orogen across the Biscay-Pyrenean system (from Martínez Catalán et al., 2007). (d) Distribution of mantle types in Europe at 110 Ma (Late Aptian) (from Picazo et al., 2016). (e) Interpreted seismic section through the Variscan orogenic area (from Bois et al., 1989; see map in panel b for location). VT = Variscan Thrust; VF = Variscan Front; the brown shading highlights the layered and highly reflective lower crust that is interpreted as either magmatic underplate or granulite-facies metamorphosed rocks (from Chenin et al., 2019).

In the text

Fig. 4 Restoration of the southern North Atlantic, with North America fixed on present-day coordinates after Nirrengarten et al. (2018) with modifications for the Biscay Pyrenean domain based on Frasca et al. (2021). (a) Santonian/Campanian (83 Ma), (b) Late Aptian/Albian (112 Ma), (c) Aptian (125 Ma), (d) Late Jurassic/Early Cretaceous (145 Ma), (e) end of Triassic (200 Ma). RECC: Reconstructed Edge of the Continental Crust. d.: domain; MOR: mid-oceanic ridge; EOB: East Orphan Basin; WOB: West Orphan Basin; PB: Porcupine Basin.

In the text

	Fig. 5 Kinematic restoration of a crustal section back to the pre-rift stage with area conservation as well as pre-rift marker horizon length conservation (see text for further explanations). Tb: top basement; Tbm: magmatic top basement (neww real estate); Tbf: fault top basement (new real estate); TbMH: stratigraphic marker horizon top basement.
In the text

Fig. 6 Sections through the Cantabrian (Western transect) and the Basque Cantabrian Basin (Central transect) (after Miró et al., 2021) and through the Mauléon-Basque massifs (Eastern section) (after Lescoutre et al., 2021). Sections show the present-day situation and restorations at middle Cenomanian (96 Ma) and at Barremian (124 Ma) times. These sections are restored respecting area conservation as well as pre-rift marker horizon length conservation. See Fig. 2b for section location. B.: basin; H.: high; BCB: Basque-Cantabrian Basin; SPBT: South Pyrenean Basal Thrust; NPFT: North Pyrenean Frontal Thrust.

In the text

Fig. 7 (a) Schematic view of the two possible fates for Triassic (pre-hyperextension) salt in a multistage rift system: the top-left panel shows no interference/reactivation between salt bearing basins (blue) and subsequent hyperextended systems (green); the top-right panel shows full interference/overprint of salt bearing basins during later hyperextension (modified from Miró et al., 2021). (b and c) Role of pre-rift salt during rifting and Alpine reactivation in the Asturias (b) and Basque-Cantabrian (c) transects (from Miró et al., 2021); Note that these sections are zooms on the sections displayed in Figure 6.

In the text

	Fig. 8 Comparison of two published thermal models for the Mauléon basin: (a) the model from Saspiturry et al. (2020b) and (b) the model from Lescoutre et al. (2019). The thermal gradients at the top of the two models are from Saspiturry et al. (2020b) (see text for details and discussion).
In the text

	Fig. 9 Three competing models to explain the “granulite conundrum” in the Biscay-Pyrenean domain. See text for a more detailed description of the models and their implications.
In the text

	Fig. 10 Conceptual model illustrating the in-sequence reactivation of rift-inherited decoupling levels at a segment centre and explaining the transition from subduction to soft collision, to hard collision, and to final late collision (for details and discussion see text). Modified from Lescoutre and Manatschal (2020) and Miró et al. (2021).
In the text

Fig. 11 Reactivation of overlapping en-echelon rift segments based on the example of the Pyrenean-Cantabrian junction showing the importance of shortcutting structures and their role in preserving rift structures (modified from Lescoutre and Manatschal, 2020). (a) Schematic 3D evolution of the Pyrenean-Cantabrian junction highlighting the structural evolution during the subduction (Late Santonian-Paleocene) and collisional (Eocene-Miocene) phases. (b) Rift domains map of the Pyrenean-Cantabrian junction. (c) Schematic kinematic restoration of the rift domains map (panel b) at Late Cenomanian (end of hyperextension), Late Santonian to Paleocene (early convergence), and Eocene to present-day (late and post convergence) stages. The locations of the cross-sections shown in panel d are represented. (d) Cross sections across the Pyrenean-Cantabrian junction (see panel c for location) showing the evolution for the Late Cenomanian (end of hyperextension), the Late Santonian to Paleocene stage (early convergence), and the Eocene to present-day (late- and post-convergence) stages. Abbreviations used in the Figures: BCB: Basque-Cantabrian Basin; MB: Mauléon Basin; SPF: South Pyrenean Fault; NdM: Nappe des Marbres; Lak: Lakoura thrust; Gav: Gavarnie thrust; Ronces: Roncesvalles thrust; NPFT: North Pyrenean Frontal Thrust; NPF: Northern Pyrenean Fault.

In the text