Abstract

The processes that drastically thinned and rapidly exhumed the Alborán Domain in southern Spain are poorly understood. Geological maps, cross-sections and a synthesis of previous work in the Alpujárride Complex provide a new structural framework for the Alborán Domain, with major revision of the previously defined lithotectonic units. Alborán continental lithosphere was buried and thickened in the Paleogene (D1) and subsequently thinned during late-orogenic extension in the early Miocene. Regional patterns of stretching lineations and late recumbent folds record important spatiotemporal changes in the kinematics of extension. East-directed shear (D2) is preserved in deep orogenic levels and thinned the crustal sequence during rapid exhumation and high-temperature metamorphism. North-directed shear (D3/D4) is preserved in upper orogenic levels and reworked earlier structures during the later stages of exhumation and cooling. This large-scale switch in the direction of finite stretching resulted from the interaction between components of orogen-parallel (D2) and orogen-perpendicular (D3/D4) extension, and associated vertical and lateral gradients in strain. We provide a novel kinematic model for gravity-driven extensional collapse of the Alborán Domain, compatible with both westward motion of the Alborán Domain relative to the Iberian margin and subsequent northward extensional flow and emplacement of Alborán lithosphere onto the Iberian margin.

Supplementary material: A list of sample names, coordinates and descriptions, plus a Google Earth .kmz file of sample locations is available at https://doi.org/10.6084/m9.figshare.5508643. A geological sketch map of the Sierra de las Estancias, indicating the trend of stretching lineations in the Alpujárride Complex and location of major geological boundaries is available at https://doi.org/10.6084/m9.figshare.5513398.

During the late stages of continental collision, thickened continental crust situated in the hinterland region of an orogenic system can be drastically thinned (i.e. late-orogenic extension), resulting in the exhumation of deep crust and mantle. Regional changes in the direction of late orogenic extension are documented in orogenic hinterlands, notably in the Betic–Rif arc (Orozco et al. 2017; Williams & Platt 2017; this study), Variscan Iberia (Díez Fernández & Martínez Catalán 2012; Arango et al. 2013), Internal Alps (Merle & Brun 1984; Ratschbacher et al. 1989; Ring & Merle 1992), Aegean region (Walcott & White 1998; Gessner et al. 2001; Jolivet et al. 2008), and Himalayan belt (Coleman 1996; Parsons et al. 2016). The magnitude and direction of extension depends on both plate boundary conditions and contrasts in gravitational potential energy within the orogenic system (Platt 2007). Therefore, kinematic reconstructions are crucial to our understanding of tectonic processes within mountain belts.

The Alborán Domain forms the westernmost part of the Alpine orogenic system and the extensional hinterland to the Betic–Rif arc. The tectonic history of the Alborán Domain is exceptionally complicated, which has in large part contributed to the multitude of contrasting geodynamic models proposed for the Cenozoic evolution of the region. Models involving subduction of oceanic lithosphere below the Alborán upper plate, propose that slab rollback and back-arc extension thinned the Alborán Domain (Lonergan & White 1997; Vergés & Fernàndez 2012). In contrast, models for removal of mantle lithosphere below the Alborán Domain provide an effective mechanism to rapidly exhume deep rocks during extension. Such models include delamination (Calvert & Sandvol 2000), convective removal of mantle lithosphere (Platt & Vissers 1989; Vissers 2012) or a combination of slab rollback and delamination (Bezada et al. 2013).

This study focuses on the kinematic evolution of the Alpujárride Complex, the largest tectonic complex of the Alborán Domain. A reassessment and synthesis of tectonic events in the Alpujárride Complex is timely because:

  1. Alternative mechanisms have been proposed for the exhumation of the Alborán Domain, including exhumation during accretion in the early stages of mountain building (Booth-Rea et al. 2005; Rossetti et al. 2005) and exhumation during extension in the late stages of mountain building (Williams & Platt 2017).

  2. Exhumed continental crustal sections record distinct metamorphic histories, deformation stages and kinematic patterns. Integrating geologic data across the entire Alpujárride Complex enables us to distinguish between the different geodynamic models.

  3. A regional synthesis of new and published structural data throughout the Betics is required, with a discussion of previous regional studies in the Alpujárride Complex (Tubía et al. 1992; Rossetti et al. 2005; Orozco et al. 2017).

Background

Tectonic history of the Betic–Rif arc

The Betic–Rif arc is a small and tightly arcuate mountain belt located in southern Spain and northern Morocco, and surrounds the Alborán Sea (Fig. 1). Like the rest of the circum-Mediterranean orogenic system, it formed as a result of the complex pattern of plate motions following the breakup of Pangaea, starting in Triassic time. Sinistrally oblique rifting between Africa and Eurasia created a set of minor oceanic basins and microcontinental fragments. This was followed by Africa–Eurasia convergence, which commenced in the late Cretaceous and led to the subduction and accretion of oceanic lithosphere and relatively buoyant continental lithosphere in the Paleogene (Rosenbaum et al. 2002). During the later stages of subduction, widespread extensional deformation exhumed deep crust and mantle in the Oligocene/early Miocene, and created the young ocean basins and arcuate orogenic belts in the western Mediterranean region (Rosenbaum & Lister 2004).

The Betic–Rif arc is divided into two zones: an internal extensional hinterland and an external thrust belt. The internal zone of the Betics contains metamorphosed units of the Alborán Domain and the Nevado–Filábride Complex. The Alborán Domain is flanked to the north and west by siliciclastic sediments of the flysch unit, and by basin and shelf sediments of the external thrust belt, derived from the former continental margins (Fig. 1). In the early Miocene, widespread extension and thinning of continental crust in the Alborán Domain coincided with thrusting of the external thrust belt onto the passive Iberian and African margins in a radial pattern around the Betic–Rif arc (Platt et al. 2003c).

The Alborán Domain

The bulk of the Alborán Domain is derived from an exotic continental fragment with no obvious affinity to either the African or Iberian continental margins. The history of Paleogene accretion and crustal thickening (i.e. D1) that formed the Alborán Domain is debated (Michard et al. 2002; Vergés & Fernàndez 2012; Platt et al. 2013; Casciello et al. 2015). However, it is generally accepted to have formed a high collisional ridge during NW dipping subduction of the African plate (including oceanic lithosphere and the so-called Alkapeca continental block) below the distal Iberian margin, with the palaeoposition of the trench in the Balearic and Ligurian seas (Lonergan & White 1997; Michard et al. 2002; Rosenbaum & Lister 2004; van Hinsbergen et al. 2014). This tectonic configuration is in part supported by the paucity of Alpine high-pressure (high-P) metamorphism in the lower Sebtide Complex of the Rif (Gueydan et al. 2015). Finally, in the Miocene, the Alborán Domain drifted westward with respect to Iberia and Africa, and Africa–Iberia convergence slowed down to a minimum (Rosenbaum et al. 2002; Platt et al. 2003c).

In southern Spain the now thinned Alborán thrust stack is divided into two tectonic complexes: non-metamorphic to weakly metamorphosed rocks of the Maláguide Complex, and low to high-grade rocks of the underlying Alpujárride Complex. The two complexes are separated by a major low-angle normal fault with ENE displacement, active in the early Miocene (Lonergan & Platt 1995). Similar tectonic histories are documented in the equivalent Moroccan units, the Ghomaride and Sebtide Complexes (see summary in Gueydan et al. 2015).

Tectonic events in the Maláguide Complex

The Maláguide Complex is the structurally highest part of the Alborán Domain, and many exposures lie along the tectonic boundary with the external thrust belt (Figs 1 and 2). It contains the most complete stratigraphy of the Alborán Domain, including Paleozoic, Mesozoic and early Tertiary sections (Lonergan 1993), and preserves evidence for both Variscan and Alpine tectonic events (Fernández-Fernández et al. 2007; Martín-Algarra et al. 2009). Towards the base of the Maláguide Complex, Paleozoic phyllites and quartzites locally contain late andalusite and biotite (Tubía et al. 1993; Cuevas et al. 2001), and have yielded early Miocene muscovite Ar–Ar and zircon fission-track cooling ages (Platt et al. 2003a; Esteban et al. 2013). This confirms that the late thermal overprint recorded in the Maláguide Complex is associated with early Miocene high temperature (high-T) metamorphism in the underlying Alpujárride Complex. It also suggests that the two complexes were exhumed and cooled together, during which extensional detachments thinned the previous Maláguide thrust stack (Booth-Rea et al. 2002a, 2004; Fernández-Fernández et al. 2007).

Tectonic events in the Alpujárride Complex

Rocks of the Alpujárride Complex are intensively deformed, and preserve multiple phases of tight to isoclinal folds and crenulation cleavages. Regional deformation stages are recognized in the Alpujárride Complex: crustal thickening during high-P metamorphism (D1), exhumation during high-T metamorphism (D2), late recumbent folding (D3), and low-angle normal faulting (D4). The timing and significance of these tectonic events, in particular exhumation (D2) and major recumbent folding (D3), are controversial. This has led to three different tectonic models for the Alborán Domain (see Williams & Platt 2017, p. 6, fig. 4 for a summary):

  1. synorogenic exhumation during subduction in the Paleogene (Tubía et al. 1992; Simancas & Campos 1993; Jolivet et al. 2003; Booth-Rea et al. 2005; Rossetti et al. 2005; Vergés & Fernàndez 2012)

  2. intraorogenic exhumation during alternating stages of contractional and extensional deformation in the late Oligocene (Azañón et al. 1997, 1998; Balanyá et al. 1997; Azañón & Crespo-Blanc 2000)

  3. late-orogenic exhumation during extension in the early Miocene (Platt & Vissers 1989; Lonergan & White 1997; Platt et al. 2013; Orozco et al. 2017; Williams & Platt 2017)

Pre-Alpine events (pre-D1)

The Alpujárride Complex stratigraphic sequence is composed of Paleozoic graphitic schists, Permo-Triassic non-graphitic schists and late Triassic carbonate rocks. Evidence for pre-Alpine metamorphic events in the Alpujárride Complex are locally preserved in graphitic schists, which contain undated and relict chloritoid, garnet, staurolite, kyanite, andalusite and cordierite (Sánchez-Navas et al. 2012, 2016; Manjón-Cabeza Córdoba et al. 2014). Variscan zircons are documented in garnet granulites, gneiss bodies and intrusions towards the bottom of the crustal sequence (Zeck & Whitehouse 2002; Ruiz Cruz & Sanz de Galdeano 2014; Sánchez-Navas et al. 2014, 2017), as well as in crustal units below the Ronda peridotites (Esteban et al. 2011b; Acosta-Vigil et al. 2014). However, structures (e.g. stretching lineations) and kinematics associated with Variscan orogeny in the Alpujárride Complex are pervasively overprinted, and therefore poorly constrained. Variscan tectonic events are also preserved in the Maláguide Complex (Martín-Algarra et al. 2009), Nevado–Filábride Complex (Gómez-Pugnaire et al. 2012) and Sebtide Complex (Michard et al. 1997; Bouybaouene et al. 1998; Rossetti et al. 2010; Gueydan et al. 2015).

Burial and crustal thickening in the Paleogene (D1)

Subduction of continental crust up to 60 km depth in the Paleogene is evidenced by relict high-P assemblages (kyanite + Mg-chloritoid + carpholite) (Goffe et al. 1989, 1996; Azañón & Goffe 1997; Azañón et al. 1998) and rare eclogite relics (Tubía & Ibarguchi 1991; Tubía et al. 1997, 2009). The exact timing of high-P metamorphism is controversial, with white mica Ar–Ar crystallization ages ranging from 48 to 25 Ma (Monié et al. 1991, 1994; Platt et al. 2005). Early crustal thickening (D1) is evidenced by relict foliations (S1) included within porphyroblasts, early folds and cleavages in the Sierra Alhamilla which have yielded Eocene white mica Ar–Ar ages (Platt et al. 2005), and southward thrusting and imbrication in the Sierra Alhamilla (Fig. 3j) (Platt et al. 1983) and central Betics (Williams & Platt 2017). These features suggest that the Alborán Domain formed a thick accretionary wedge in the Paleogene.

Following subduction and burial, medium P/T metamorphism associated with the static crystallization of garnet, staurolite and kyanite (interkinematic between D1 and D2) suggests widespread thermal relaxation (Azañón et al. 1998; Platt et al. 2013; Manjón-Cabeza Córdoba et al. 2014; Williams & Platt 2017). Structures associated with crustal thickening were strongly overprinted by extensional deformation in the early Miocene (see section ‘Late-orogenic extension in the Alpujárride Complex’), and therefore the kinematics of D1 are unknown.

Rapid exhumation in the early Miocene (post-D1)

Several important features demonstrate that the Alpujárride Complex was rapidly exhumed during late-orogenic extension in the early Miocene. The Paleogene record of high-P and medium P/T metamorphism was pervasively overprinted by high-T and low-P metamorphism in the early Miocene, evidenced by the late crystallization of sillimanite, andalusite and cordierite, breakdown of muscovite, and partial melting in deep crustal levels (Garcia-Casco et al. 1993; García-Casco & Torres-Roldán 1996; Argles et al. 1999). Published pressure–temperature (P–T) paths for the Alpujárride Complex are characterized by isothermal decompression, whereby deep rocks were exhumed from the kyanite stability field into the sillimanite and andalusite stability fields (see Williams & Platt 2017, p. 3, fig. 3). This requires a combination of rapid exhumation and external heat input (Platt et al. 1998; Platt & Whitehouse 1999). High-T metamorphism (23–19 Ma zircon U–Pb crystallization ages) was immediately followed by cooling (22–19 Ma mica Ar–Ar cooling ages, and c. 18 Ma zircon/apatite fission-track ages), and demonstrates that exhumation in the early Miocene was rapid and simultaneous throughout the Alborán Domain (Platt et al. 2003b). Furthermore, well-preserved disequilibrium metamorphism in the Alpujárride Complex is diagnostic of rapid decompression followed by fast cooling during an extensional collapse event (García-Casco & Torres-Roldán 1996).

Late-orogenic extension in the Alpujárride Complex

The Alpujárride Complex is characterized by structures that formed during late-orogenic extension (Orozco et al. 2017; Williams & Platt 2017). These include vertically condensed metamorphic sequences (D2), subhorizontal foliations and stretching lineations (D2/D3), late recumbent folds (D3) and low-angle normal faults (D4), described below.

Vertical thinning of the crustal sequence (D2)

Large bodies of continental mantle lithosphere exhumed in the early Miocene attest to drastic vertical thinning of the Alborán Domain, and are exposed in the western Betics (Tubía et al. 1993, 2004; Platt et al. 2003a; Garrido et al. 2011) and Rif (Afiri et al. 2011; Frets et al. 2014; Gueydan et al. 2015). The most impressive attenuated crustal sequences in the Alpujárride Complex overlie the peridotite bodies and, from bottom to top, comprise garnet granulite, migmatitic gneiss, sillimanite–andalusite schist, chlorite–kyanite phyllite, and non-metamorphic units of the Maláguide Complex (Torres-Roldán 1981; Balanyá et al. 1993). The present-day structural thickness of this sequence is only c. 5 km, therefore extension during the early Miocene thinned the orogenic crust by at least c. 10 times its original thickness (c. 50 km) (Argles et al. 1999; Platt et al. 2003a). The most pervasive fabric in the Alpujárride Complex is S2–L2, and is often referred to as the ‘principal foliation’. S2–L2 formed during east-directed shear, and is associated with vertical condensation of the metamorphic sequence (Tubía et al. 1992; Balanyá et al. 1997; Azañón & Crespo-Blanc 2000) and extensional detachments (Argles et al. 1999; Platt et al. 2005).

Stretching lineations and directions of extension (D2/D3)

In principle, stretching lineations can be used to reconstruct the kinematics of subduction (Díez Fernández & Martínez Catalán 2012), back-arc extension (Jolivet et al. 2008, 2009) and extensional collapse (Díez Fernández et al. 2012). Large-scale changes in the trend of stretching lineations are located in crustal units below the peridotites (Esteban et al. 2008; Tubía et al. 2013), in crustal sections directly above the peridotites (Balanyá & García-Dueñas 1987; Balanyá et al. 1997) and in the central and NE Betics (Simancas & Campos 1993; Platzman & Platt 2004; Rossetti et al. 2005; Williams & Platt 2017). Orthogonal patterns of east-directed and later north-directed shear in the Alpujárride Complex have been attributed to crustal thickening in the Paleogene (Cuevas 1991; Tubía et al. 1992; Simancas & Campos 1993; Rossetti et al. 2005) and recently to extensional collapse in the early Miocene (Platt et al. 2013; Orozco et al. 2017; Williams & Platt 2017).

In a recent structural review of the Alpujárride Complex, Orozco et al. (2017) propose that the major change in shear direction in the central Betics, near Motril (Fig. 1), formed between two major extensional shear zones. Important questions remaining from their study include:

  1. were both shear zones active at the same time?

  2. what is the nature of the transition between these shear zones?

  3. what is the geometry of the shear zones, i.e. where are the lower and upper boundaries located?

  4. was exhumation accommodated by ductile or brittle deformation, i.e. what is the relationship between stretching lineations, major recumbent folds and the extensional detachments which bound them?

Late recumbent folds (D3)

Late recumbent folds (D3) formed during north-directed shear and refolded the condensed crustal sequence and high-T metamorphic isograds (Azañón et al. 1996, 1998; Williams & Platt 2017). Major folding (D3) has been associated with contraction during crustal thickening in the Paleogene (Tubía et al. 1992; Simancas & Campos 1993; Rossetti et al. 2005), a renewed stage of thrusting in the early Miocene (Azañón et al. 1997, 1998; Balanyá et al. 1997; Azañón & Crespo-Blanc 2000; Booth-Rea et al. 2005) or extension during the late stages of crustal thinning in the early Miocene (Orozco 1998; Platt 1998; Orozco et al. 2017; Williams & Platt 2017). Because the primary orientation of structures following the D2 event is poorly constrained, D3 fold geometry and kinematics on its own cannot be used to distinguish between these different tectonic settings. The relationship between D3 structures and the metamorphic history is therefore also an important criterion (Harris et al. 2002).

Late-Triassic carbonates and marbles are intensely deformed and folded in the western Betics (Sanz de Galdeano & Andreo 1994, 1995; Orozco & Alonso-Chaves 2012), central Betics (Sanz de Galdeano 1986, 1988, 1990; Simancas & Campos 1993; Sanz de Galdeano & López-Garrido 1999, 2003) (Fig. 3a) and eastern Betics (Sanz de Galdeano 1985; Campos & Simancas 1989; Cuevas et al. 1990; Orozco et al. 2004; Sanz de Galdeano 2009; Sanz de Galdeano & López Garrido 2014a, b), and carbonate units were the focus of the Orozco et al. (2017) study. However, determining the age and significance of deformation fabrics in these carbonate units is hindered by a lack of metamorphic index minerals. In contrast, Paleozoic and Permo-Triassic schists preserve the relationship between high-T decompressional metamorphism and D3 deformation, and confirm that major D3 folds formed during crustal extension in the early Miocene (Williams & Platt 2017; this study).

Lithotectonic units and low-angle normal faults (D4)

The Alpujárride Complex was originally mapped as a stack of thrust sheets (Aldaya 1969; Aldaya et al. 1979). Once it was demonstrated that the Alpujárride Complex had been dramatically extended (e.g. Platt & Vissers 1989; Vissers et al. 1995), these thrust sheets were reinterpreted as lithotectonic units bound by low-angle normal faults (from top to bottom, the Adra, Salobreña, Herradura, Escalate and Lújar–Gador units) (Fig. 2). The correlation of these lithotectonic units across the Alpujárride Complex was based on high-P mineral assemblages (Azañón et al. 1994; Balanyá et al. 1998; Azañón & Crespo-Blanc 2000), and these units are described as the lower, intermediate and upper tectonic units by Tubía et al. (1992, 1997) and Orozco et al. (2017). However, these tectonic units show marked vertical and lateral variations in metamorphic grade associated with the early Miocene high-T event, which complicates definitions based on high-P assemblages. Also, many of the mapped tectonic boundaries between tectonic units are inherited from original definitions based on stratigraphic or tectonic omissions, repetitions and inversions, and do not relate to the metamorphic evolution or to the history of late-orogenic extension (Platt 1998). In the central Betics, Williams & Platt (2017) reinterpreted several of these tectonic boundaries as stratigraphic contacts, early thrust faults (D1), metamorphic isograds (D2), low-angle normal faults (D4) and late normal faults (post-D4). The lithotectonic unit concept clearly requires reassessment throughout the Alpujárride Complex, which provides the motivation for this study.

Low-angle normal faults (D4), many of which bound the extensional units described above, cut the exhumed crustal sequence (Fig. 2). These formed during the waning stages of north–south-directed crustal extension and cooling that formed late orogenic basins in the early Miocene (Fig. 3i) (Crespo-Blanc et al. 1994). These low-angle faults are in turn cut by steeper SW-directed normal faults associated with the middle to late Miocene exhumation of the Nevado–Filábride Complex (post-D4) (Martínez-Martínez & Azañón 1997) (Fig. 3k). Regionally, late Miocene upright contractional folds (post-D4) deform earlier structures, and are associated with broad dome structures, for example the Sierra Alhamilla and Sierra Nevada antiforms (Crespo-Blanc et al. 1994).

The Nevado–Filábride Complex: an early Miocene subduction event

The Nevado–Filábride Complex is a remnant of the Iberian margin, subducted in the early Miocene below the extending and hot Alborán Domain (Platt et al. 2006; Kirchner et al. 2016), then exhumed in the middle and late Miocene within a subduction channel (Behr & Platt 2012, 2013). The final stages of exhumation were accommodated by west- to SW-directed motion of the Alpujárride Complex relative to the Nevado–Filábride Complex on a major ductile to brittle shear zone, which brought the Nevado–Filábride Complex in contact with the overlying Alpujárride Complex (Jabaloy et al. 1993). This phase of deformation was accompanied by the formation of a series of middle to late Miocene extensional basins, many of which are bounded by SW-directed normal faults (post-D4) that cut Alpujárride rocks (Morales et al. 1990; Galindo-Zaldívar et al. 1999; Meijninger & Vissers 2006; Platt et al. 2013; Do Couto et al. 2014).

Results

In the Results section we use field observations (Fig. 3) and microstructural/petrographic analysis of 413 oriented rock samples (Fig. 4 & Supplementary material) to describe important tectonic and metamorphic events in the western, central, SE and NE Betics. We provide geological maps (Fig. 5) and cross-sections (Fig. 6) which illustrate how metamorphic zones, stretching lineations, km-scale fold structures and low-angle normal faults formed during late-orogenic extension which intensely thinned and modified the previous thrust stack. We also provide a synthesis of kinematic data throughout the Alpujárride Complex (Figs 7 and 8), including data collected from 28 previous studies (Table 1 and 2). Lastly, we discuss our new kinematic interpretations in the context of previous studies and provide a revised tectonic model for the Alborán Domain (Figs 911).

Western Betics

East-directed shear and the early stages of exhumation (D2)

The Alpujárride Complex in the western Betics stretches from Estepona to Málaga, and contains peridotite bodies overlain by drastically attenuated crustal sequences (Fig. 5a and b). During high-T metamorphism, east-directed extension thinned the entire crustal section (Argles et al. 1999; Frasca et al. 2016) and formed the main S2–L2 fabric (Fig. 4a and b). S2–L2 comprises a pervasive foliation with an ESE- to ENE-trending stretching lineation (Fig. 8a–d). In deep crustal levels, S2–L2 formed during the growth of sillimanite, and completely overprinted relict fabrics and assemblages associated with crustal thickening (D1). This suggests that D2 formed during the early stages of exhumation, during decompression into the sillimanite stability field. In shallower crustal levels, the L2 stretching lineation is defined by the alignment of minerals such as kyanite, amphibole and epidote (Fig. 4c).

D2 structures, such as sheath folds, isoclinal folds, boudinaged foliations and low-angle shear bands formed during high strain east-directed shear (Orozco & Alonso-Chaves 2012; Frasca et al. 2016). The large-scale geometry of D2 folds has not been reconstructed because D2 deformation is intense and high strain, and D2 folds strongly transpose layering. Large-scale recumbent D2 folds should produce local repetitions of the stratigraphy, but due to the aforementioned reasons, evidence for this in the western Betics is lacking.

North-directed shear, late recumbent folds and the late stages of exhumation (D3)

Throughout the Alpujárride crustal sequence, D2 structures were deformed by north-directed shear (D3). D2 folds are refolded by recumbent D3 folds and, because D2 and D3 fold axes are generally subparallel, the most common fold interference patterns are refolded isoclines (Fig. 3c). Andalusite systematically overprints S2–L2 and is synkinematic with recumbent D3 folds (Fig. 4e). This suggests that D3 folds formed during the later stages of exhumation and crustal thinning, as the rocks rapidly decompressed into the andalusite stability field. This sequence of early exhumation (D2), associated with the growth of sillimanite, and late exhumation (D3), associated with the growth of andalusite, is also observed in the central Betics (Williams & Platt 2017).

Kinematics of the mantle–crust shear zones

During late-orogenic extension in the early Miocene, normal sense mantle–crust shear zones exhumed bodies of mantle peridotite and juxtaposed them below deep crustal rocks (Balanyá et al. 1993; Frasca et al. 2016). Relict metamorphic zones, foliation trajectories and stretching lineations within the peridotite bodies are discordantly overprinted by peridotite mylonites of the mantle–crust shear zone (Johanesen et al. 2014). In the Sierra Bermeja, top-to-the-SW shear in the peridotite mylonites (Van der Wal & Vissers 1996; Vissers 2012; Hidas et al. 2013; Johanesen & Platt 2015; Gueydan & Frasca 2017) is opposite to top-to-the-NE shear in the overlying garnet granulites and migmatitic gneisses of the crustal section (Balanyá et al. 1997). In the Carratraca massif, Frasca et al. (2016) also document opposite senses of shear in the mantle-crust shear zone (top-to-the-west) and middle/upper crust (top-to-the-east). This may suggest that the mantle–crust boundary is either an extraction fault (Froitzheim et al. 2006) formed by zippering of an originally conjugate system of shear zones (Passchier & Platt 2016; Platt & Passchier 2016), or it accommodated opposite senses of shear either side of a continental rift (Gueydan & Frasca 2017).

The kinematics of the mantle–crust shear zones is different, and perhaps earlier, from shear in the overlying crustal sequence. For example, in the Sierra Bermeja, NE/SW-directed conjugate shear in the mantle–crust shear zone is oblique to north-trending L2 stretching lineations and D3 recumbent fold axes in the overlying Jubrique crustal section (Fig. 5a). In the Carratraca massif and Sierra de la Robla, shear is top-to-the-SSE in the marginal peridotites and garnet granulites, transitional between top-to-the-SSE and -ENE in the migmatite gneisses, and top-to-the-ENE (D2) in the overlying schists and phyllites (Argles et al. 1999 and our observations) (Figs 5b and 8k). East-directed shear (D2) is also documented in the crustal sequence of the Santi-Petri dome, c. 10 km SE of the Carratraca massif (Cuevas et al. 2001; Booth-Rea et al. 2003) (Figs 5b and 8l). In the Sierra Alpujata, top-to-the-ESE shear is documented in both the mantle–crust shear zone and overlying crustal sequence, so no change in kinematics is observed (Tubía et al. 1993) (Fig. 5a and 8j).

Emplacement of the peridotite bodies

Immediately following early Miocene exhumation, the peridotite bodies were emplaced onto continental crust while they were still hot enough to partially melt the underlying crust (Tubía et al. 1997, 2013; Esteban et al. 2008, 2011a; Frasca et al. 2017). Emplacement was accommodated by narrow and high-temperature shear zones at the base of the peridotites, associated with top-to-the-north shear in the Guadaiza and Istán tectonic windows (Esteban et al. 2008; Acosta-Vigil et al. 2014) (Fig. 8p), and top-to-the-ENE shear in the Ojén tectonic window (Tubía et al. 1997, 2013) (Fig. 8o). The orientation of dyke intrusions in the peridotites suggests that the Carratraca massif was also emplaced during top-to-the-ENE shear (Cuevas et al. 2006). Whether emplacement occurred within a contractional or extensional tectonic regime is controversial (Platt et al. 2003a, 2013; Hidas et al. 2013; Tubía et al. 2013).

Significance of the Blanca unit

Surrounding the base of the Sierra Alpujata peridotites is a sequence of cordierite, sillimanite and K-feldspar-bearing gneisses associated with high-T emplacement of the peridotite (Tubía et al. 2013), juxtaposed next to high-grade schists, amphibolites, and strongly deformed marbles, ascribed to the Blanca unit and exposed in the Sierra Blanca and Sierra de Mijas (Orozco & Alonso-Chaves 2012) (Fig. 5a). The structural relationship of the Blanca unit with respect to the peridotites is complicated (Fig. 6a), but it is generally accepted that the Blanca unit is separated from the base of the peridotites by a low-T shear zone (Navarro-Vilá et al. 2007; Tubía et al. 2009; Bartoli et al. 2016), and that parts of the Blanca unit are steeply dipping and locally overturned in the inverted limb of a D3 fold (Sanz de Galdeano & Andreo 1994, 1995; Tubía et al. 2009; Esteban et al. 2011b). Whether the Blanca unit structurally underlies (Tubía et al. 1997; Balanyá et al. 1998; Navarro-Vilá et al. 2007; Mazzoli et al. 2013; Acosta-Vigil et al. 2014; Frasca et al. 2016, 2017) or overlies (Sanz de Galdeano & López Garrido 2016) the subperidotite complex and the peridotites is debated (Fig. 6a).

ESE-trending stretching lineations, sheath folds, isoclinal folds and boudinaged layering (D2) in the Blanca unit (Orozco & Alonso-Chaves 2012) are consistent with ESE-directed shear and S2–L2 fabrics documented in the crustal section overlying the Sierra Alpujata peridotites (Fig. 8j). D2 fold axial planes and S2 fabrics were deformed by tight folds (D3) at lower strain, and due to reorientation locally dip steeply to the south and north (Fig. 6a). The curved geometry of D2 and D3 fold axial traces is therefore not a topographic feature (Sanz de Galdeano & Andreo 1994, 1995). The Blanca unit forms an overall arched geometry; foliations, tectonic boundaries and D3 folds are NNE-trending in the Sierra Blanca, swinging to ENE-trending in the Sierra de Mijas (Fig. 5a).

The structure of the Blanca unit marbles has been interpreted as a result of interference between early east–west- and later north–south-trending folds (Sanz de Galdeano & Andreo 1994, 1995) and, more recently, as a km-scale D3 sheath fold (Orozco & Alonso-Chaves 2012; Orozco et al. 2017). Several features suggest to us that the Blanca unit marbles comprise a km-scale D2 sheath fold, later reworked by D3 folds:

  1. the abundance of D2 folds with isoclinal and sheath geometry in outcrop, associated with boudinaged layering and a strong penetrative S2–L2 fabric (Orozco & Alonso-Chaves 2012; Tubía et al. 2013 and our observations);

  2. the large variance in the orientation of D2 fold axes (Orozco & Alonso-Chaves 2012), and curved geometry of major D2 fold axial traces (Fig. 5a) suggest a major D2 sheath geometry;

  3. the opposite vergence of km-scale D2 isoclinal folds and eye-shaped D2 fold structures documented in the Sierra de Mijas and Sierra Blanca (Sanz de Galdeano & Andreo 1994, 1995);

  4. the north–south-trending neck between the Sierra Blanca and Sierra de Mijas marble bodies suggests km-scale boudinage structures associated with east–west extension (i.e. D2) (Sánchez-Gómez et al. 2002; Orozco et al. 2017);

  5. D2 folds are reworked by lower strain D3 folds and form fold interference patterns. S2–L2 fabrics are not strongly overprinted by D3, and there is no penetrative S3–L3 fabric.

On the southeastern margin of the Sierra Blanca, near Ojén, stretching lineations and D3 folds in the Blanca unit and subperidotite complex are NNE-trending (Fig. 8o). We propose that this is a local effect; S2 foliations containing originally ENE-trending L2 lineations (e.g. Sierra de Mijas) have been reoriented and steepened towards sub-vertical within the inverted limb of a major NNE-trending D3 fold.

Central Betics

The Alpujárride Complex in the central Betics is situated between Málaga and Motril, and contains a sequence of sillimanite + K-feldspar gneisses through chlorite phyllites with a present-day structural thickness of ≥4 km. This suggests vertical thinning of the crustal sequence up to 3 times its original thickness (c. 12 km) (Williams & Platt 2017). Correlation of lithotectonic units across the central and eastern Betics is poor, in particular between Torrox and Motril (Sanz de Galdeano & López-Garrido 2003; Orozco et al. 2017) (Fig. 2). We provide a new map and cross-sections reinterpreting the metamorphic sequence, kinematics of deformation (D2/D3), geometry of major D3 folds and significance of tectonic contacts, e.g. low-angle normal faults (D4) (Figs 5c and 6).

Superposed directions of shear (D2/D3)

The central Betics is characterized by a large-scale swing in stretching lineation and fold axis orientation (Simancas & Campos 1993; Rossetti et al. 2005; Orozco et al. 2017; Williams & Platt 2017). The swing is progressive over a c. 60 km wide region, with ESE-trending lineations near Vélez-Malaga and Torrox (Fig. 8a), ENE-trending lineations west of Almuñécar (Fig. 8c,d), NE/NNE-trending lineations east of Almuñécar (Fig. 8e), and north-trending lineations in the Sierra Lújar region (Fig. 8f). We associate this swing with a spatial and temporal kinematic transition during extensional exhumation: east-directed shear (D2) dominant in the western and central Betics (Fig. 5a–c) is overprinted by north-directed shear (D3) dominant in the eastern Betics (Fig. 5d).

During the early stages of exhumation (D2), east-directed shear thinned the crustal sequence and exhumed deep rocks to middle/upper crustal levels. In the western and central Betics, including the Benamocarra unit (Cuevas et al. 2001), approximately east-trending stretching lineations (L2) formed during the growth of sillimanite in high-grade rocks. During the later stages of exhumation (D3), north-directed shear produced recumbent D3 folds that refold S2–L2 fabrics and high-T metamorphic isograds (Fig. 6). This also produced a new generation of NNE-trending stretching lineations (L3) which formed during the growth of andalusite in high-grade rocks (Fig. 8e). Therefore, the transition from east-directed shear (D2) to north-directed (D3) shear occurred during rapid exhumation from the sillimanite into the andalusite stability field (Williams & Platt 2017).

Composite zone of shear (D2/D3)

Between Almuñécar and Motril both L2 and L3 stretching lineation generations are preserved (Fig. 8e). This is a zone of composite D2 and D3 shear, and represents the interface between ductile D2 and D3 extension. Furthermore, north-directed shear (D3) reoriented L2 lineations and D2 fold axes towards a NE-trend (Williams & Platt 2017). The eastern Betics is separated from this zone by a late normal fault (post-D4), which we refer to as the Motril fault, and contains strong S3–L3 fabrics which formed during cooling, from ductile to semi-brittle conditions. In these low-grade schists, the absence of index minerals hinders determination of the relative age of deformation fabrics, making their age and relevance controversial (discussed in the section ‘Deformation events in the SE Betics’).

Gradients in D3 strain

In medium- and low-grade schists between Almuñécar and Adra, approximately ENE-trending D3 fold axes are oblique to the approximately NNE-trending L3 stretching direction (Figs 5c, d and 8e, g, h). In contrast, low-grade schists and carbonates on the south and east border of the Sierra de Lújar contain approximately north-trending D3 fold hinges subparallel to the L3 stretching direction (Fig 8f). These folds have curved hinge line geometries and have been rotated parallel to the L3 stretching direction (Cuevas 1989b, 1991; Cuevas et al. 1990; Rossetti et al. 2005). This suggests that the structurally deepest L3 lineations formed during relatively low strain top-to-the-north shear (D3), and a gradient in top-to-the-north (D3) strain, increasing upward through the crustal section, reoriented D3 fold hinge lines towards the north-directed L3 extension direction. The Sierra Lújar region therefore represents the upper part and high strain zone of D3 deformation. Elsewhere in the eastern Betics, north-trending L3 mineral lineations are perpendicular to east/west-trending D3 folds in the Sierra Gador (Orozco et al. 2017), Sierra Alhamilla (Fig. 8q) and Sierra de Baza (Fig. 8r), and oblique to NE/SW-trending D3 folds in the Sierra de las Estancias (Fig. 8s), suggesting these are relatively low-strain D3 zones.

Major D3 folds in the central Betics

Kilometre-scale recumbent folds (D3) have a north-vergent geometry, defined by north-closing hinges cored by high-grade schists, and south-closing hinges cored by low-grade schists and carbonates (Fig. 6). Overturned D3 fold limbs are associated with south-vergent outcrop-scale folds and locally invert the metamorphic sequence and reorient S2–L2 fabrics. The clearest example of a major overturned fold limb is located to the south of the Sierra de Tejeda and Sierra de Almijara, and locally places steeply dipping sillimanite schists above low-grade schists and carbonates (Alonso-Chaves & Orozco 2012; Orozco et al. 2017) (Figs 6b,c). Major D3 folds between Vélez-Málaga and Nerja can now be correlated with those in the Almuñécar region (Williams & Platt 2017) (Fig. 6d).

Low-angle normal faults (D4) and late normal faults (post-D4)

Recumbent D3 folds are cut and dismembered by low-angle normal faults with north-displacement (D4). In the Sierra de Tejeda and Sierra de los Guájares, low-angle normal faults place klippen of sillimanite schists (Adra and Herradura units) above chlorite schists and carbonates (Herradura unit) (Fig. 6b-d; Williams & Platt 2017). In the Sierra de Tejeda, SW-trending D3 fold axes (Fig. 8c) suggest the low-angle normal fault (D4) cut similarly oriented structures positioned in the footwall and to the SE, near Almuñécar. Similarly, in the Sierra de los Guájares, SE-plunging D3 fold axes (Fig. 8b) are likely related with folds positioned to the SW, near Torrox.

Later normal faults with SW-displacement (post-D4) further thinned the crustal sequence (Alonso-Chaves & Orozco 1998, 2007; Orozco & Alonso-Chaves 2002). The clearest example is the major Maro-Nerja normal fault (post-D4) which placed sillimanite schists directly on top of low-grade units and is associated with a large fault gouge zone (Figs 5c and 6b–d). We reinterpret several low-angle normal faults (D4) which bound lithotectonic units as SW-directed late normal faults (post-D4). For example, the klippe of upper extensional unit (Adra unit) in the Sierra de Chaparral (Rossetti et al. 2005; Orozco et al. 2017) (Fig. 2) is, in fact, bound by a steep normal fault with SW-displacement (Fig. 3k). This fault uplifted high-grade schists in the footwall, and juxtaposed them next to low-grade schist and carbonates in the hanging wall. Other steep faults (post-D4) have recently been identified in the Sierra de los Guájares, and near to Almuñécar and Motril (Williams & Platt 2017).

Motril fault (post-D4)

The Motril fault is an important and perplexing structure (see Fig. 5c). It shows the following characteristics:

  1. it is a major north–south-trending lineament that separates D2 and D3 domains;

  2. in its current form it is a discrete brittle fault and there is no evidence for a wide shear zone associated with displacement on the fault;

  3. the change in stretching lineation trend is dramatic across the fault, suggesting it is a late structure;

  4. it cuts north-directed low-angle normal faults (D4) and is cut by SW-directed normal faults (post-D4).

We have reinterpreted the Motril fault as a steep east-dipping normal fault that juxtaposes low-grade rocks of the eastern Betics above high-grade rocks of the central Betics (Williams & Platt 2017). In contrast to previous interpretations (e.g. Orozco et al. 2017), this means that the western and central Betics mainly expose deep Alpujárride levels, and the eastern Betics expose predominantly upper Alpujárride levels. Below we provide two potential interpretations of how the Motril fault formed.
  1. Perhaps the Motril fault was originally a top-to-the-east detachment fault which bounded ductile D2 deformation in the footwall to the west, and D3 deformation in the hanging wall to the east. The fault may have acted as a decoupled interface between D2 and D3 deformation, and became a localized brittle detachment at a late stage in its evolution.

  2. The location of the Motril fault coincides with the western termination of the Nevado–Filábride Complex (Figs 1 and 2) which suggests a link to the exhumation of the Nevado–Filábride Complex. Perhaps it was a conjugate fault to the Nevado–Filábride/Alpujárride detachment. If so, it must have transferred displacement on to the detachment, as it does not cut it. The detachment may have functioned as an extraction fault, allowing extraction of the deeper Alpujárride units westwards, so the higher Alpujárride units now lie directly on the Nevado–Filábride Complex.

Around the margins of the Sierra Nevada, a perplexing aspect concerns the widespread outcropping of low-grade Alpujárride and Maláguide units in an apparently low structural position directly above the Nevado–Filábride Complex (Platt et al. 2013). This has been interpreted as evidence for a late thrusting event (D3) in the Alpujárride Complex, which placed high-grade units towards the top of the nappe stack and low-grade units towards the bottom (Balanyá et al. 1997; Azañón & Crespo-Blanc 2000; Booth-Rea et al. 2005). We suggest instead that the Motril fault likely acted as an extraction fault that placed low-grade units towards the east. During its exhumation in the middle to late Miocene the Nevado–Filábride Complex cut upwards through the entire eastern Alpujárride section (Figs 2 and 6e) and now crops out below low-grade Alpujárride units (García-Dueñas et al. 1992; Crespo-Blanc 1995). During SW/NE extension, major normal faults (post-D4) cut through the Alpujárride Complex and soled on the Alpujárride/Nevado–Filábride detachment (Crespo-Blanc 1995; Martínez-Martínez & Azañón 1997, 2002; Augier et al. 2005). This is supported by scattered outcrops of Maláguide very near to the detachment, close to the towns of Berja and Ugíjar (Fig. 2).

SE Betics

South of the Nevado–Filábride window, between Motril and Carboneras, the Alpujárride Complex contains monotonous low- and medium-grade rocks without sillimanite that preserve Paleogene high-P metamorphism (Goffe et al. 1989; Azañón & Goffe 1997; Booth-Rea et al. 2002b, 2005) and medium-P/T metamorphism (Fig. 5d). These rocks largely escaped the early Miocene thermal overprint, and Paleogene mica Ar–Ar ages in low-grade phyllites suggest some exhumation before the early Miocene event (Platt et al. 2005). These rocks were finally exhumed in the early Miocene during pervasive north-directed shear and folding (D3) and, like deep crustal levels of the western and central Betics, contain fabrics (S2 and S3) associated with late-orogenic extension (Orozco 1998; Platt et al. 2003b; Orozco et al. 2004, 2017).

Deformation events in the SE Betics

Due to the low metamorphic grade, the timing of deformation events in the SE Betics is particularly controversial (Rossetti et al. 2005; Williams & Platt 2017). Below we provide a summary of deformation events based on our microstructural and petrological observations (Fig. 4).

The earliest foliation (S1) formed during the growth of Mg-rich chloritoid (Fig. 4g) and is preserved within relict garnet, staurolite and kyanite porphyroblasts (Fig. 4f and h) (Williams & Platt 2017). In deep crustal levels S2 wraps these relict porphyroblasts and is folded by D3 (Fig. 4f). D3 crenulations and hinges are overprinted by andalusite (Fig. 4f), confirming that major north-vergent D3 folds in the SE Betics formed during ductile exhumation, and before cooling and low-angle normal faulting (D4). In upper crustal levels S2 was intensely reworked and overprinted by north-directed shear (D3) (Fig. 3e–h) during low-grade and non-metamorphic conditions (Fig. 4g, h). This produced the main structural fabric in the SE Betics, S3–L3.

North-directed shear in upper crustal levels (D3)

The SE Betics is defined by north-vergent folds (D3), pervasive axial planar fabrics (S3) and north-trending stretching lineations (L3). Sigmoidal quartz lenses (Fig. 3e,h), asymmetrical strain shadows around relict garnets (Fig. 4h), rotational chloritoids and N-S trending mineral lineations (Fig. 4g) are consistent with top-to-the-north shear during D3 extension. D3 progressed during the late stages of exhumation in a transition from ductile to brittle conditions. Finally S3 fabrics and D3 folds are cut by north-directed shear bands which formed during strain hardening (late-D3), and low-angle normal faults which formed during cooling (D4) (Fig. 3e).

In the region between Motril and Adra (Fig. 5d), S3–L3 fabrics are strongly developed in high strain zones and major D3 fold hinge regions (e.g. Cuevas & Tubía 1990; Cuevas 1991; Rossetti et al. 2005). Both S3 and L3 dip and plunge to the south (Fig. 8f–h), locally very steeply (Fig. 3g, h). Due to the competence contrast, S2 foliations, D2 fold hinges and early boudinaged layering are typically preserved in strong quartzites and carbonates. For example, in the high strain D3 zone near Sierra Lújar and Castell de Ferro, carbonate units preserve east–west-trending folds (D2) superposed by north–south-trending folds (D3) (Sanz de Galdeano 2009; Sanz de Galdeano & López Garrido 2014a). S3–L3 is strongly developed in weaker and anisotropic schists (Fig. 3e–h).

Major D3 folds in the SE Betics

In the SE Betics several repetitions of stratigraphy and metamorphic grade were originally interpreted as thrusts or low-angle normal fault contacts bounding the lithotectonic units (Aldaya 1969; Tubía et al. 1992; Simancas & Campos 1993). Many of these contacts have since been reinterpreted as major fold hinges and inverted limbs belonging to km-scale D3 folds (Azañón et al. 1997; Orozco 1998; Orozco et al. 2004, 2017) (Fig. 5d). These folds deformed metamorphic isograds and formed during extensional collapse (Azañón et al. 1996; Orozco et al. 2017; Williams & Platt 2017) (Fig. 6e–g). We therefore correlate these structures to major D3 folds in the western and central Betics (Fig. 6a–d). Their position in the hanging wall block of the Motril fault suggests they represent the uppermost part of the D3 fold sequence.

One notable example of a major D3 fold is located along the SE border of the Sierra Lújar. Here, garnet-bearing schists (Adra and Salobreña units) form the core of a north-closing fold hinge, and locally overlie low-grade phyllites and carbonates (Escalate and Lújar-Gádor units) in the inverted limb of the south-closing Lújar syncline (Fig. 6e). The Lújar syncline was interpreted as a sheath fold by Orozco et al. (2004, 2017), which formed during high strain north-directed shear (D3). Another example is north of Adra, along the Río Grande river, where chloritoid-bearing graphitic schists (Salobreña unit) in a north-closing fold hinge locally overlie low-grade non-graphitic phyllites and carbonates (Escalate and Lújar–Gádor units) in the inverted limb of the south-closing Turón synform (Azañón et al. 1997; Orozco et al. 2004, 2017) (Fig. 6g).

Large recumbent D3 folds are also documented east of Adra, in the Sierra Alhamedilla (Balanyá et al. 1987; Campos & Simancas 1989; Orozco 1998), the Sierra de Gádor (Orozco et al. 2004, 2017) and the Sierra Alhamilla (Platt et al. 1983). In the Sierra Alhamilla, north/south-aligned elongate caliche nodules form a stretching lineation on fold axial planes in a high strain overturned limb of the major east/west-trending D3 fold (Fig. 8q). However, phyllites at this location yielded an Ar–Ar age of 48 Ma (Platt et al. 2005), suggesting that these lineations and folds may be earlier (i.e. D1) than D3 structures in garnet + staurolite + plagioclase schists further down the section.

Relationship between stretching lineations, recumbent folds and low-angle normal faults

In the central and eastern Betics, km-scale north-vergent folds (D3) with regionally north dipping enveloping surfaces locally inverted the metamorphic sequence and were later cut by low-angle normal faults (D4) with large northward displacement (Fig. 6). This thinned the crustal sequence and placed low-grade rocks above high-grade ones (e.g. the Contraviesa fault, Fig. 6f).

In the SE Betics, low-angle normal faults (D4) with NNW-displacement were active in the early and middle Miocene (Crespo-Blanc et al. 1994; Martínez-Martínez & Azañón 1997, 2002). S3 foliations containing NNE-trending L3 stretching lineations are cut by faults, shear bands and drag folds associated with top-to-the-NNW shear (D4) (Fig. 8h). Near Albuñol, tectonic windows of low-grade phyllites and carbonates (Escalate and Lújar–Gador units) are separated by a discrete low-angle normal fault from overlying garnet and biotite schists (Adra and Salobreña units) (Fig. 6f). NNW-directed shear band lineations and low-angle normal fault slickenlines (D4) are recorded within the tectonic windows (Azañón et al. 1995; Sanz de Galdeano & López Garrido 2014b), and the hanging wall blocks (Fig. 5d). The Turón synform (D3) is cut by a major D4 low-angle normal fault called the Turón detachment (Figs 3i and 6g). This is described as the upper boundary of the major north-directed extensional shear zone (D3) (Orozco et al. 2017), and the basal detachment fault of the D4 extensional system (Crespo-Blanc et al. 1994; Martínez-Martínez & Azañón 2002; Orozco et al. 2004).

NE Betics

We define the NE Betics as the region between Granada and Cartagena north of the Sierra Nevada (Fig. 1). In this region the Alpujárride Complex consists predominantly of fine-grained schists and phyllites (e.g. Crespo-Blanc 1995; Booth-Rea et al. 2002a, 2004) which occasionally preserve high-P assemblages (Booth-Rea et al. 2002b) and garnet, staurolite, kyanite and late andalusite (Platt et al. 1983; Platzman & Platt 2004; Booth-Rea et al. 2005). During large-magnitude ductile extension in the early Miocene, extensional detachments (D2) placed low-grade rocks directly above high-grade rocks and are documented in the Sierra de las Estancias, Río Grande section and Sierra Alhamilla (Platt et al. 1983, 2005; Orozco 1998; Platzman & Platt 2004). These detachments and associated S2 foliations were refolded by major north-vergent folds (D3) during continued crustal extension (Vissers et al. 1995; Platt 1998).

Shear fabrics and mineral lineations in the NE Betics suggest a top-to-the-north stretching direction. In the Sierra de las Estancias, deep units below the Estancias detachment yielded mica Ar–Ar cooling ages of c. 20 Ma (Platt et al. 2005). In these units stretching lineations trend NE and are synkinematic with andalusite (Platzman & Platt 2004) (Figs 4i, 8s). These lineations are therefore likely to be associated with D3 deformation. In upper units above the Estancias detachment and close to the Alpujárride/Maláguide detachment, stretching lineations trend ENE. Mica Ar–Ar ages of c. 30 Ma from these upper units probably represent cooling (Platt et al. 2005) and suggest that these lineations may be older and associated with D2 extension.

In the eastern part of the Sierra de las Estancias and Sierra de Tercia fold axes are north-trending and parallel to north-trending stretching lineations (Booth-Rea et al. 2004) (Fig. 8t). However, due to low-grade units which lack index minerals, the age of these structures is poorly constrained. Listric low angle normal faulting (D4 and post-D4) north of the Sierra Nevada has largely dismembered the Alpujárride crustal section (Crespo-Blanc 1995; Booth-Rea et al. 2002a; Augier et al. 2005).

A new tectonic model for the Alborán Domain

The Alborán Domain: a thinned orogenic sequence

We propose a new structural and kinematic framework for the Alborán Domain during early Miocene extension (Figs 911). The western Betics and Rif contain mantle peridotites overlain by high-grade crustal units (Platt et al. 2003a; Johanesen et al. 2014; Gueydan et al. 2015), which we interpret as the deepest levels of the Alpujárride Complex. Deep crustal levels are characterized by condensed metamorphic isograds and pervasive ductile D2 deformation. In contrast, the eastern Betics is formed of low-grade units and brittle D2 extensional detachments which we interpret as the upper levels of the Alpujárride Complex. This suggests increasing peak-temperature conditions downwards and westwards throughout the Alpujárride Complex in the early Miocene. The thermal structure of the Alborán Domain therefore represents a continental lithospheric sequence, stretched and thinned during large-magnitude ductile extension (e.g. Platt et al. 2003a; Negro et al. 2006; Frasca et al. 2016).

Our interpretations require a complete structural rearrangement of the Alpujárride crustal sequence. Previous studies are largely based on the lithotectonic unit concept developed by Aldaya (1969), which interpret the Alpujárride Complex as a thickened nappe stack. These interpretations place the lowest-grade lithotectonic units at the bottom of the now thinned thrust stack (e.g. Lújar-Gador unit, Fig. 2), and vice versa (Tubía et al. 1992, 1997; Azañón et al. 1994; Balanyá et al. 1998; Rossetti et al. 2005; Orozco et al. 2017). The geometry of the two major extensional shear zones described by Orozco et al. (2017) is based on the outdated lithotectonic unit concept. We therefore propose a new temporal and spatial relationship between the two major zones of east-directed and north-directed shear, described below.

Stretching lineations in the Alpujárride Complex

Stretching lineations record the direction of finite extension during ductile exhumation of deep crustal rocks (Jolivet et al. 2009; Díez Fernández & Martínez Catalán 2012; Díez Fernández et al. 2012). Changes in regional patterns of stretching lineation orientation may be related to one or more of the following effects (Williams & Platt 2017, and references therein).

  1. heterogeneity in the magnitude and direction of shear combined with strain partitioning during formation of the stretching lineation;

  2. more than one generation of stretching lineation, associated with separate deformation events;

  3. reorientation of a linear fabric due to later deformation.

The regional orientation of stretching lineations in the Alpujárride Complex is remarkably consistent across the Betic mountain belt (Fig. 7). However, there are several major changes in stretching lineation trend and shear direction. The most impressive change in stretching lineation trend is in the central Betics (Figs 5c and 7). According to Orozco et al. (2017) two wide and orthogonal extensional shear zones accommodated the major change in stretching direction during a single D3 subduction-related event. In this interpretation stretching lineations and major recumbent folds in the western and central Betics formed by D3 shear within a large east-directed ductile shear zone (Alonso-Chaves & Orozco 2012; Orozco & Alonso-Chaves 2012; Orozco et al. 2017). However, east–west-trending stretching lineations and major folds in this area formed during high strain extension and the growth of sillimanite (e.g. the Sierra Blanca sheath fold), and we interpret them as D2 structures. These were later deformed by major D3 folds during the growth of andalusite (Fig. 4d and e), producing km-scale fold interference structures (D2/D3) (Fig. 6a and c). In our interpretation D3 folds in the western and central Betics formed during the later stages of decompression and a switch to north-directed shear, and can be correlated with major north-vergent D3 folds in the eastern Betics Fig. 6e–g. Structures associated with east-directed shear (D2) are systematically overprinted by north-directed shear (D3) throughout the entire Alpujárride Complex (Tubía et al. 1992; Rossetti et al. 2005; Williams & Platt. 2017). Therefore, the orthogonal stretching directions described by Orozco et al. (2017) are not confined to two regionally separated and contemporaneous D3 shear zones. We suggest that rocks were progressively exhumed through the deforming orogen, and moved from the D2 deformational field into the D3 deformational field (Fig. 11). These processes in the vertical direction are not accounted for by the two-shear zone model of Orozco et al. (2017).

We propose the large-scale change in stretching direction resulted from a combination of the effects listed above. East-directed shear (L2), prominent in the western and central Betics, is overprinted by north-directed shear (L3), prominent in the eastern Betics (Williams & Platt 2017; this study). The composite zone of D2 and D3 shear in the central Betics exposes an important spatial transition in the Alborán Domain; the swing in stretching lineation trend represents the interplay between decreasing D2 strain and increasing D3 strain across the deforming orogen, which progressively reoriented the finite extension direction. Because exhumation was rapid (<3 myr) (Platt et al. 2003b), D2 and D3 were active close together in time, and in some areas may be represented by a single deformation event. Therefore, D2 and D3 likely represent both spatial and temporal differences in kinematics within the deforming orogen (Fig. 11).

Gradients in finite strain have been described in the subperidotite units (Esteban et al. 2008), in the peridotites towards the mantle–crust shear zone (Précigout et al. 2013), and in the Adra unit of the eastern Betics (Cuevas 1991). We suggest that large-scale gradients in finite extensional strain were important during the exhumation of the Alpujárride Complex. Deep orogenic levels of the western Betics were exhumed during high-T conditions and high strain east-directed extension (D2) which thinned the lithospheric sequence by a factor of at least ten times (Platt et al. 2003a; Frasca et al. 2016). In the central Betics the crustal sequence was thinned by a factor of about three times (Williams & Platt 2017). Therefore, the magnitude of east-directed shear (D2) increased downwards and westwards through the orogen. Throughout the Alpujárride Complex, D2 structures are progressively deformed, reoriented and overprinted by north-directed shear and north-vergent recumbent folds (D3) at lower pressure conditions. This is associated with an increase in D3 finite strain upward and eastward through the crustal sequence. East-trending L2 stretching lineations and D2/D3 fold hinges in the central Betics were progressively rotated and reoriented towards the north-directed stretching direction (L3) in the eastern Betics (Fig. 8a–e).

Significance of major recumbent folds (D3)

Kilometre-scale D3 folds formed during horizontal extension and vertical shortening: S2 fabrics were dipping at a moderate angle with respect to the north-directed shear zone, before being folded within the contractional field of the instantaneous strain ellipsoid (Platt 1982; Froitzheim 1992; Harris et al. 2002; Orozco et al. 2004; Arango et al. 2013; Díez Fernández et al. 2017; Williams & Platt 2017) (Fig. 10c). In the western and central Betics, D3 fold axes are subparallel to D2 fold axes and L2 stretching lineations (Figs 3c,d, 5a–c and 8a–d,i,m). This is perhaps due to (a) rotation of the D3 fold axes during continued and broadly contemporaneous east-directed shear (D2), (b) relatively low D3 finite strain in deep orogenic levels of the western Betics and (c) rheological factors, for example, strong L2 anisotropy controlled the orientation of later D3 folds. In the SE Betics, D3 fold axes trend obliquely to L3 stretching lineations by c. 30–60° (Fig. 8g and h), suggesting partial rotation of the D3 fold hinge line toward the north-stretching direction (Cuevas 1991). The region near the Sierra Lújar is a high strain D3 zone, with full rotation of D3 fold axes parallel to the L3 north-stretching direction (Fig. 8f). This gradient in D3 strain is consistent with a change in geometry of km-scale D3 folds; recumbent folds and elongated domes in the Turón region transition to sheath folds in the Sierra Lújar (Orozco et al. 2017, p. 56, fig. 4).

Significance of low-angle normal faults (D4)

In the SE Betics, stretching lineations and late recumbent folds are interpreted to have formed by north-directed drag along extensional detachments (D4), for example the Turón detachment (Orozco 1998; Orozco et al. 2004, 2017; Rossetti et al. 2005) (Fig. 2 and 3i). However, these structures formed during ductile deformation (D3) (Figs 3e–h and 4f–h), and are systematically cut by semi-brittle shear bands and discrete low-angle normal faults (D4) with scattered NNW/SSE-trending striae (e.g. Cuevas et al. 1986; Tubía et al. 1992; Williams & Platt 2017) (Figs 3e and 6f, g). These relationships suggest that recumbent folds (D3) and stretching lineations (L3) were formed and largely exhumed before being cut by low-angle normal faults (D4) during the waning stages of north-directed extension, exhumation and cooling.

Low-angle normal faults (D4) record the transition from ductile to brittle conditions, and are more prominent in the upper crustal levels of the eastern Betics than in the deeper orogenic levels of the western and central Betics (Crespo-Blanc et al. 1994; Crespo-Blanc 1995; Martínez-Martínez & Azañón 1997). We therefore suggest that D2 and D3 represent large-magnitude ductile extension and exhumation of the Alpujárride Complex, and D4 low-angle normal faults represent the final increment of extension at the top of the deforming orogen, during rapid cooling. Late-orogenic extension in the Alborán Domain records the interplay between processes acting at the bottom (D2) and towards the top (D3/D4) of the deforming orogen (Fig. 11).

Dextral shear due to westward motion of the Alborán Domain (post-D4)

In other parts of the Alpujárride Complex, changes in regional patterns of consistent stretching lineation trend are clearly associated with reorientation due to later deformation (i.e. post-D4). During westward motion of the Alborán Domain relative to Iberia in the early Miocene, distributed dextral shear in the Alborán Domain may have reoriented L2 and L3 lineations. Dextral shear may have conditioned the vector of L2 extension parallel to the trend of the orogen (east/west trending) and rotated the vector of L3 extension oblique to the trend of the orogen (NNE/SSW trending) (Fig. 7). These orientations are consistent with a decreasing component of dextral shear, as westward motion of the Alborán Domain slowed down, either due to limited plate motions as the orogen evolved or increased mechanical coupling between the Alborán upper plate and Iberian lower plate in the eastern Betics (Marín-Lechado et al. 2017a).

Major strike-slip corridors and strike-slip faults also accommodated westward collision of the Alborán Domain onto the Iberian margin (Platt et al. 2003c; Frasca et al. 2015; Gueydan & Frasca 2017). Regional patterns of north-trending L2 stretching lineations are recorded in the Sierra Bermeja region, and are orthogonal to east-trending L2 stretching lineations in the Sierra Alpujata region (Fig. 5a). The change in stretching direction occurs across the major Albornoque strike-slip fault, which bounds the northern margin of the Sierra Blanca. Dextral motion along the fault cut major D3 folds (Sanz de Galdeano & Andreo 1995; Tubía et al. 2013; Frasca et al. 2015) and contributed to the arched geometries (i.e. curved anticline traces) of the Sierra Blanca and Guadaiza window. The orthogonal stretching lineation pattern resulted from large clockwise rigid-body rotations in the Miocene (Platzman 1992; Platzman et al. 2000; Michard et al. 2002; Berndt et al. 2015; Gueydan & Frasca 2017), rather than strain partitioning within a transpressional setting (Tubía et al. 2013). Similarly, the Maro-Nerja normal fault was active in the Tortonian with a major dextral component of slip (Alonso-Chaves & Orozco 1998, 2007) (Fig. 5c) and accommodated large clockwise rigid-block rotation (Crespo-Blanc et al. 2016). The fault cuts through the external thrust belt in the NW (Martínez-Martínez & Azañón 1997; Crespo-Blanc et al. 2016), and the offshore Alborán Basin in the SE (Comas et al. 1999; Martínez-García et al. 2013), and separates predominantly ESE-trending L2 stretching lineations near Torrox (Fig. 8a) from ENE-trending L2 stretching lineations in the Sierra de Tejeda and Almuñécar region (Fig. 8b–d). We suggest that clockwise rigid-body rotation produced the discrete change in L2 stretching lineation orientation. Finally, late Tortonian to Pliocene uplift and open folding (post-D4) reoriented structures, for example, the broad change in ENE- through ESE-trending lineations across the Santi Petri dome (Cuevas et al. 2001; Booth-Rea et al. 2003) (Fig. 5b).

Stretching lineations in the other tectonic units

Subperidotite and deep crustal units

The subperidotite units represent the deepest crustal levels of the Alborán Domain, and many aspects of their origin and formation are enigmatic (Bartoli et al. 2016). Granulites and migmatites beneath the peridotites preserve both Variscan and Alpine tectonic fabrics (Acosta-Vigil et al. 2014). We note similar stretching directions and senses of shear recorded in both subperidotite units and respective crustal sequences overlying the Sierra Bermeja, Sierra Alpujata and Carratraca peridotite bodies (i.e. D2) (Fig. 5a and b). This suggests to us that the peridotites were emplaced onto the crust along an east-directed low-angle normal sense shear zone during thinning of the entire Alpujárride sequence (Platt et al. 2003a, 2013), rather than during west-directed thrusting associated with westward collision of the Alborán Domain onto the Iberian basement (Sánchez-Gómez et al. 2002; Navarro-Vilá et al. 2007; Tubía et al. 2013; Frasca et al. 2015), or inversion of thinned back-arc lithosphere during slab rollback (Hidas et al. 2013; Précigout et al. 2013; Frasca et al. 2017; Gueydan & Frasca 2017).

Sillimanite + K-feldspar gneiss bodies located north of the Sierra Tejeda (Alonso-Chaves et al. 1985) (Fig. 8m), near Torrox (Cuevas et al. 1989) (Fig. 8n) and Melicena (Cuevas 1989b; Azañón et al. 1997) (Fig. 8g), contain stretching lineations with different orientations to the overlying units. They may therefore preserve Variscan deformation at the bottom of the orogenic crust, associated with Variscan metamorphism (e.g. Sánchez-Navas et al. 2017).

Maláguide Complex

Because the Maláguide Complex is predominantly non-metamorphic and the upper parts contain predominantly mixed Variscan and Alpine zircon/apatite fission-track ages, determining the age and significance of ductile fabrics, folds and stretching lineations is difficult. However, where stretching lineations are documented in the Maláguide Complex and Alpujárride/Maláguide bounding detachment zone (e.g. Balanyá & García-Dueñas 1987; Tubía et al. 1993; Lonergan & Platt 1995; Cuevas et al. 2001), they are subparallel to L2 stretching lineations in the underlying Alpujárride Complex (Figs 5a, b and 7). This suggests that the Maláguide Complex was thinned during D2 extension in the early Miocene and juxtaposed above high-grade Alpujárride units along the Alpujárride/Maláguide detachment (Fig. 11).

Nevado–Filábride Complex

The Nevado–Filábride formed during SE-directed subduction of the Iberian plate below the extending Alborán Domain in the early Miocene (Behr & Platt 2012) (Fig. 9d and 10c). It is possible that the early stages of subduction may have contributed to north-directed shear (D3/D4) in the Alpujárride Complex. North-directed shear (D3/D4) in the Alpujárride Complex of the eastern Betics is orthogonal and opposite to approximately WSW-directed shear associated with middle and late Miocene exhumation of the Nevado–Filábride Complex (Fig. 7). This suggests a major kinematic reconfiguration following early Miocene extensional collapse of the Alborán Domain, perhaps associated with underplating of the Nevado–Filábride Complex and changes in gravitational potential energy within the Betic–Rif system.

Geodynamic implications

Understanding how the Alborán Domain was exhumed is crucial for geodynamic models of the Betic–Rif orogenic system. Following crustal thickening (D1), continental lithosphere of the Alborán Domain was rapidly exhumed in the early Miocene during late-orogenic extension (D2/D3) and an elevated thermal gradient. Two endmember geodynamic models have been proposed to explain the exhumation of the Alborán Domain:

  1. back-arc extension – driven by subduction and slab rollback;

  2. extensional collapse – driven by high gravitational potential energy and detachment of a mantle lithosphere root by either convective removal or delamination.

The primary difference between these two models is the relative importance of external plate forces acting on the Alborán Domain (i.e. back-arc extension) v. internal body forces acting within the Alborán Domain (i.e. extensional collapse). The two models also make different predictions about the timing and amount of westward motion of the Alborán Domain with respect to Iberian continent (Platt et al. 2003c; Balanyá et al. 2007; Frasca et al. 2015), the location and type of magmatism (Turner et al. 1999; Duggen et al. 2004; Booth-Rea et al. 2007; Doblas et al. 2007; Rossetti et al. 2013; Varas-Reus et al. 2017) and the amount of oceanic lithosphere subducted (Platt & Houseman 2003; Spakman & Wortel 2004; Capitanio & Goes 2006; van Hinsbergen et al. 2014). These are controversial topics beyond the scope of this discussion.

The kinematic pattern of late-orogenic extension in the Alpujárride Complex enables us to distinguish between these models. The back-arc extension model predicts a downgoing slab beneath the overriding Alborán Domain in the early Miocene, which retreated rapidly and stretched the Alborán lithosphere (Lonergan & White 1997; Spakman & Wortel 2004; Jolivet et al. 2008; Orozco et al. 2017) (Fig. 10b). In this model back-arc extension would exhume the Ronda peridotite in a continental rift setting (Hidas et al. 2013; Précigout et al. 2013; Frasca et al. 2016; Gueydan & Frasca 2017). The change in stretching direction in the central Betics would therefore reflect a change in the direction of subduction within the orogenic system. Orozco et al. (2017) propose westward rollback of an east-directed subduction zone, followed by southwards rollback of a north-directed subduction zone to explain the kinematic pattern. However, the polarity of subduction beneath the Alborán Domain in the early Miocene is poorly constrained; models predict either SW- (Jolivet et al. 2008; Garrido et al. 2011; Varas-Reus et al. 2017) or NW- (Spakman & Wortel 2004; Vergés & Fernàndez 2012; Bezada et al. 2013) directed slab rollback. Also, the amount of north or south rollback would have been limited by the narrow width of the Alborán Basin (c. 200 km) and a large proportion of buoyant continental lithosphere within the Iberian and African margins (Capitanio & Goes 2006). Body forces were therefore necessary to drive extensional collapse of the Alborán Domain onto the Iberian continental margin, and thrusting within the external units.

The extensional collapse model predicts that following crustal thickening (D1) (Fig. 10a), removal of mantle lithosphere in the early Miocene led to high gravitational potential energy within the extensional hinterland (Fig. 10b) (Vissers 2012; Platt et al. 2013). In principle this would have extended the Alborán Domain in all directions, with extension faster normal to the length of the orogen (i.e. north–south) than parallel to the length of the orogen (i.e. east–west) due to larger contrasts in gravitational potential energy. However, the Alborán Domain was wedged between the Iberian margin to the north, and African margin to the south, preventing north–south extension at depth. This drove the Alborán Domain westwards towards relatively thin oceanic or continental crust (Platt 2007) and promoted east–west extension (D2) in deep orogenic levels. At the same time, north–south extension in the upper levels was not constrained in the same way, because the Alborán Domain could ride up and over the two continental margins. This led to coeval north-directed extension (D3) at shallow orogenic levels. Rocks in the Alborán Domain were exhumed by extension and moved up through the orogenic system, giving the overprinting of D2 by D3 (Fig. 11).

The component of D2 stretching increased downwards towards the base of the Alborán Domain, and westwards towards the foreland (Fig. 11). Mantle–crust extensional shear zones exhumed the Ronda peridotite during D2 and removal of mantle lithosphere (Johanesen et al. 2014). The relative importance of east-directed extension and pure shear extension during the D2 event is poorly constrained. Some sort of rollback was required to accommodate the westward motion of the Alborán Domain. We therefore propose a combination of westward tectonic flow associated with extensional collapse, plus a minor component of rollback towards the west to generate the east-directed extension in the Alborán Domain (D2) (Fig. 10b).

Following D2 the Alborán Domain was emplaced onto the Iberian margin and further thinned during north-directed extension (D3 and D4) (Fig. 10c and d). The change in extension direction in the central Betics resulted from progressive changes in the relative importance of body forces within the deforming orogen. The magnitude of gravitational potential energy contrasts parallel to the strike of the orogen was reduced during D2 extension. Therefore gravitational potential energy perpendicular to the strike of the orogen progressively became the primary body force, and this drove north-directed extensional flow of the orogen (D3). Furthermore, east-directed extensional flow during D2 led to a relative progressive increase in gravitational potential energy towards the east, which contributed to the upward and eastward gradient in D3 strain throughout the orogen.

The extensional collapse model predicts the systematic overprinting of D2 by D3 throughout the Alpujárride Complex, the progressive spatial and temporal change in extension direction from L2 to L3, and the rapid transition (<3 myr) from orogen-parallel to orogen-perpendicular extension. It is also consistent with subduction of the Iberian plate beneath the Alborán Domain after, and not during, the early Miocene extensional event (c. 21 Ma); the Nevado–Filábride Complex was subducted c. 18 Ma and escaped the high-T metamorphism that affected the Alborán Domain (Behr & Platt 2012; Kirchner et al. 2016).

Conclusions

We propose a new tectonic model for extensional collapse of the Alborán Domain in the early Miocene: high gravitational potential energy in the orogenic hinterland drove extensional collapse parallel to the strike of the orogen, and the westward motion of the Alborán Domain. This resulted in east-directed extension (L2) which pervasively thinned the Alborán crust and exhumed deep orogenic levels. Following D2, contrasts in gravitational potential energy increased perpendicular to the strike of the orogen. This resulted in north-directed extension (L3), which progressively overprinted D2 and produced major fold interference structures (D2/D3). D3 was associated with an upward gradient in strain and thinned upper orogenic levels. The central Betics is a spatial and temporal transition zone between orogen-parallel (D2) and orogen-perpendicular (D3) extension, and preserves composite approximately NE-directed shear. Low-angle normal faults (D4) dismembered the exhumed orogenic section during the later stages of north-directed extension and cooling. The kinematic pattern of late-orogenic extension in the Alborán Domain was finally modified by clockwise rigid-block rotations and strike-slip faults associated with the dextral motion of the Alborán Domain relative to Iberia, and normal faulting associated with exhumation of the Nevado–Filábride Complex (post-D4). Our tectonic reconstruction of the deforming Alborán Domain is compatible with geodynamic models that invoke gravity-driven extensional collapse rather than slab-driven back-arc extension.

Acknowledgements

The authors thank R. Díez Fernández and L. Jolivet for their helpful reviews and comments.

Scientific editing by Karel Schulmann