The coupled Iberian oroclines of the western European Variscan orogen accommodated >1100 km of post-Variscan orogen-parallel shortening at translation rates in excess of 5 cm⋅yr–1. Palinspastic restoration of the Iberian coupled oroclines reveals a north-south–trending 2300-km-long Variscan ribbon bound by ophiolite-bearing allochthons. The requirements for orocline formation, including continuing subduction and the consumption of vast tracks of oceanic lithosphere, cannot be reconciled within traditional models that view the Variscan as a record of the closure of a single (Rheic) ocean resulting in terminal Gondwana-Laurussia collision and a stable Pangea supercontinent. Paleomagnetic data from the Gondwana-derived Variscan autochthon indicate its mid-Paleozoic separation (as the Armorican microplate) and the growth of a second mid-Paleozoic (Paleotethys) ocean. The Gondwanan stratigraphic and faunal character of the Variscan autochthon can be reconciled by a model within which counterclockwise separation results in an Armorican ribbon continent that remains close or connected to Gondwana to the south and extends north toward Laurasia between an older (Rheic) ocean to the west and a newly opened (Paleotethys) ocean to the east. Geologic and paleomagnetic data further indicate that, despite translating steadily northward, Gondwana remained separated from Armorica throughout the various stages of Variscan orogenesis. We explain Pangean amalgamation as being coincident with buckling of a linear Armorican ribbon continent caught between Laurussia to the north and the northward-migrating Gondwana to the south. In this model, Variscan orogenesis is explained in terms of individual accretionary events generated through continuous, west-dipping subduction along the margins of the Armorican ribbon continent.

The episodic amalgamation and subsequent breakup of supercontinents plays a significant role in Earth system evolution (e.g., Murphy and Nance, 2013; Nance and Murphy, 2013; Bradley, 2011; Eyles, 2008; Worsley et al., 1986). Recognition of the Proterozoic supercontinents Nuna (Colombia) (Hoffman, 1997; Rogers and Santosh, 2002) and Rodinia (e.g., McMenamin and McMenamin, 1990; Hoffman, 1991) has been essential for constraining the episodicity of supercontinent formation. However, our efforts to decipher the mechanisms and processes involved in the supercontinent cycle begin with the record of Earth’s most recent supercontinent, Pangea (Wilson, 1966; Bullard et al., 1965; Wegener, 1912).

The Early Jurassic configuration of Pangea at the onset of its 180 Ma breakup is constrained by paleomagnetic poles, morphological fit, and lithological correlations along the Atlantic continental margins, and the age and magnetic striping of the oceanic crust that floors the Atlantic Basin (Stampfli and Borel, 2002; Torsvik et al., 2001; Bullard et al., 1965). Pangean amalgamation culminated in the mid-Paleozoic with the collision of the southern megacontinent Gondwana with the previously amalgamated Laurussia (Laurentia plus Baltica) to the north (Nance et al., 2010; Ziegler, 1988; Van der Voo, 1983). Although the Variscan-Alleghanian-Ouachita system of western Europe and eastern North America is commonly interpreted as providing an orogenic record of terminal Gondwana-Laurussia collision, we posit that this interpretation is inconsistent with available geological and paleomagnetic data. Here we (1) summarize the data that constrain the relative timing of Pangean amalgamation and Variscan orogenesis, and (2) present an alternative model wherein the Variscan orogen is interpreted as a product of accretionary orogenesis affecting an independent Armorican (Van der Voo, 1979) ribbon continent. Geological and paleomagnetic data together dictate that construction of the Variscan orogen predates Gondwana-Laurussia collision, necessitating the more complex story proposed herein.

Traditional tectonic models view Variscan orogenesis as a consequence of Pangea-forming Gondwana-Laurussia collision involving the closure of a single intervening ocean termed the Rheic. The Variscan autochthon is of Gondwanan affinity and consists of an early Paleozoic passive margin sequence constructed unconformably on top of a limited late Precambrian sequence deformed through late Ediacaran (Cadomian) orogenesis. Allochthonous terranes of the Variscan orogen include continental terranes of peri-Gondwanan affinity and oceanic terranes containing ophiolites and associated accretionary complexes (Fig. 1). The Rheic Ocean is defined as the ocean that opened during the early Paleozoic drift of peri-Gondwanan terranes (e.g., Avalonia-Meguma) from the northern margin of West Gondwana. With the opening of the Rheic, passive margin sedimentation was initiated along the north-facing Variscan margin of Gondwana. The boundary between Variscan autochthon and peri-Gondwanan allochthon is thus interpreted to mark the location of the Rheic Ocean suture. The suture trace is commonly characterized by ophiolite or ophiolitic mélange considered to represent either direct lithospheric remnants of the Rheic Ocean or suprasubduction zone ophiolites generated during its closure (Murphy et al., 2006; Nance and Linnemann, 2008; Nance et al., 2010, 2012). Although a notable number of Variscan tectonic models have proposed Gondwana-Laurussia collision involving one or more intervening microcontinents and associated minor oceanic domains (e.g., Lardeaux, 2014; von Raumer et al., 2009; von Raumer, 2013; Stampfli et al., 2002; Franke et al., 2000; Tait et al., 2000; Matte, 1986, 1991, 2001), such models remain highly variable and widely debated.

The Variscan autochthon is preserved most completely within the Iberian Massif of Spain and Portugal, across which the orogen is deformed into the S-shaped coupled Cantabrian and Central Iberian oroclines (Shaw et al., 2012, 2014; Martínez Catalán, 2011; Aerden, 2004; Dvorak, 1983; Du Toit, 1937). The autochthon is divisible into a series of tectonostratigraphic domains distinguished primarily on the basis of pre-Variscan stratigraphy and Variscan deformational style. From the core of the more northern Cantabrian orocline outward, these domains include a thin-skinned foreland fold and thrust belt, and external and internal hinterland domains that become progressively more deformed and metamorphosed toward the shallowly emplaced ophiolite-bearing Galicia–Trás-os-Montes allochthon at the core of the Central Iberian orocline in northwestern Iberia (Fig. 1) (e.g., Martínez-Catalán et al., 2007). A fourth autochthonous domain within southern Iberia, the Ossa Morena, is characterized by a unique pre-Variscan paleogeographic history and is separated from the more northern massif by the crustal-scale Badajoz-Córdoba sinistral shear zone (BCSZ) (e.g., Quesada and Dallmeyer, 1994). Sedimentological and paleontological constraints indicate that during the early Paleozoic, the foreland and hinterland autochthonous domains to the north of the BCSZ formed a 2300-km-long linear portion of the north-facing, northern passive margin of West Gondwana. The margin extended east-west (in present-day coordinates) along the northern limits of northeast Africa and Arabia (Shaw et al., 2012, 2014, and references therein). A distinct absence of late Mesoproterozoic–early Neoproterozoic (1.1–0.9 Ga) detrital zircons within Ediacaran–lower Ordovician strata of the Ossa Morena autochthonous domain indicate that this region (along with similarly characterized autochthonous domains within the Bohemian and Armorican massifs) occupied a different position along the early Paleozoic Gondwana passive margin, perhaps adjacent to the West African craton (e.g., Fernández-Suárez et al., 2002; Linnemann et al., 2004). Cooling ages spanning 370–330 Ma along the BCSZ (40Ar/39Ar on amphibole and mica; Quesada and Dallmeyer, 1994) indicate that the Ossa Morena domain had reached its current position relative to the more northern autochthon by the middle stages of the Variscan orogeny.

A second distinct ophiolitic assemblage within southern Iberia is considered to mark the location of the Rheic Ocean suture. The Beja Acebuches ophiolite and its spatially associated Pulo de Lobo accretionary mélange together characterize the Pulo de Lobo (Rheic) suture. The suture defines the boundary between the Ossa Morena autochthonous domain to the north and a peri-Gondwanan terrane referred to as the South Portuguese zone to the south. The ophiolite-bearing Galicia–Trás-os-Montes allochthon of northwestern Iberia is interpreted as a klippe thrust northward from the Pulo de Lobo suture to its south (e.g., Martínez Catalán et al., 2007). However, the age of emplacement and emplacement geometry of the two ophiolitic complexes differ significantly. Ophiolites within the Galicia–Trás-os-Montes allochthon yield protolith ages of ca. 395 Ma (Pin et al., 2006, 2002; Díaz García et al., 1999) and were emplaced above the northwestern Iberian autochthon ca. 377 Ma (40Ar/39Ar cooling ages; Dallmeyer et al., 1997). In contrast, metamorphic cooling ages within the Beja Acebuches ophiolite (Fonseca and Ribiero, 1993) suggest that ocean closure along the Pulo de Lobo suture occurred ca. 340 Ma. Oceanic closure was achieved through northward-directed subduction as indicated by southward structural vergence across the Pulo de Lobo suture. Voluminous Middle Devonian to early Visean (ca. 390–340 Ma) calc-alkaline magmatism characterizes the Ossa Morena autochthon, consistent with its location being within the upper plate to the Pulo de Lobo subduction zone.

The western European Variscan orogen (Fig. 1) is characterized by a series of oroclinal bends. Of these, the S-shaped Cantabrian–Central Iberian pair of the Iberian massif is the best constrained. The more northern and convex toward the west Cantabrian orocline preserves the Variscan foreland at its core, while allochthonous complexes (the Galicia–Trás-os-Montes zone) occupy the core of the more southern and convex toward the east Central Iberian orocline (Du Toit, 1937; Aerden, 2004; Martínez Catalán, 2011; Shaw et al., 2012). Recognition of the S-shaped Cantabrian–Central Iberian oroclines is incompatible with the traditional interpretation of a direct link from southwestern Iberia to the northwestern Armorican massif and southern British Isles around the so-called Ibero-Armorican arc (e.g., Martínez Catalán et al., 2007; Robardet, 2002; Matte, 1991; Robardet and Gutiérrez-Marco, 1990). A new interpretation of Variscan geometry beyond the Iberian massif is thus required, but cannot be constrained by current geological data sets.

Structural, stratigraphic, paleomagnetic, and sedimentological studies collectively demonstrate that the Iberian coupled oroclines formed by vertical-axis buckling of an originally linear 2300-km-long segment of the Variscan orogen (Fig. 2) (Gutiérrez-Alonso et al., 2004; 2011a, 2011b; Pastor-Galán et al., 2011; Shaw et al., 2012, 2014, 2015; Weil et al., 2010). The spatiotemporal distribution and geochemical attributes of widespread post-Variscan (ca. 310–285 Ma) magmatism (Fig. 3) have been used to argue in favor of a plate-scale model for orocline formation within which buckling involved both the crust and lithospheric mantle (Gutiérrez-Alonso et al., 2004, 2011a, 2011b). Orocline formation occurred over a 20 m.y. time interval bracketed by the late Carboniferous (ca. 310 Ma) cessation of east-west Variscan shortening (in present-day coordinates) and the early Permian (ca. 290 Ma) unconformable overlap of terrigenous sedimentary successions that yield expected paleomagnetic poles (e.g., Weil et al., 2010). The coupled Cantabrian and Central Iberian oroclines are therefore post-Variscan structures.

Palinspastic restoration of the Iberian coupled oroclines (Fig. 2) shows that the Galicia–Trás-os-Montes allochthon and Pulo de Lobo suture were located on opposite sides of the initially linear 2300-km-long Iberian segment of the orogen (Shaw et al., 2015; Shaw and Johnston, 2016). The Galicia–Trás-os-Montes ophiolitic nappe structurally overlies the western hinterland of the preoroclinal, linear north-south–striking orogen, whereas the Pulo de Lobo ophiolite is along the eastern, foreland side of the restored orogen. The two ophiolite-bearing complexes therefore represent two different oceanic tracts, and indicate that the Iberian segment of the Variscan orogen formed a ribbon continent with oceanic lithosphere extending along both sides. Ophiolite emplacement ages young from the Galicia–Trás-os-Montes allochthon (ca. 377 Ma) to the Pulo de Lobo suture (ca. 350 Ma), in the same direction as structural vergence.

Orocline formation followed a 90° shift in the orientation of principal compressive stress away from the orogen-normal stress responsible for Variscan orogenesis. Post-Variscan orogen-parallel stress resulted in >1100 km of orogen-parallel shortening (by oroclinal buckling) accommodated at translation rates in excess of 5 cm yr–1 (Shaw et al., 2015; Shaw and Johnston, 2016). The magnitude of orogen-parallel shortening requires the removal of a surface area of >1 × 106 km2 of lithosphere (Fig. 4), suggesting that the orogen was bound on at least one side by oceanic lithosphere subducted during orocline formation (Johnston et al., 2013). The rapid rate of translation during orocline formation is consistent with deformation having been driven by the subduction of oceanic lithosphere with which the orogen was coupled. The 90° post-Variscan change in regional compression, >1100 km of orogen-parallel shortening, and continuing subduction yielding rapid translation and the consumption of >1 × 106 km2 of oceanic lithosphere cannot be reconciled within traditional models that view the Variscan as a record of terminal Gondwana-Laurussia collision yielding a stable Pangea.

The Gondwanan affinity of the Variscan autochthon is established through sedimentological and paleontological studies (Noblet and Lefort, 1990; Fernández-Suárez et al., 2002; Robardet, 2003; Linnemann et al., 2004; Shaw et al., 2014). However, paleomagnetic data from mid-Paleozoic rocks in the autochthonous domains of the Iberian, Armorican, and Bohemian Massifs (see compilation in Shaw and Johnston, 2016) place these regions at significantly lower paleolatitudes than the north African Gondwanan margin to which they are linked (Fig. 5). These paleomagnetic data imply the separation and independent northward drift of a microplate termed Armorica (Van der Voo, 1979, 1982), consisting of the full extent of the Variscan autochthon (see discussion in Shaw and Johnston, 2016). Separation of Armorica from Gondwana requires that a second ocean, termed the proto-Tethys or Paleotethys (e.g., Stampfli et al., 2002; Perroud et al., 1984), grew in the wake of the northward-migrating Armorica, reaching a maximum width of ∼3500 km at 380 Ma. The timing of Armorican separation depends on the validity of an outlying paleomagnetic pole, the 409 Ma Aïr pole of Hargraves et al. (1987). Acceptance of the Aïr pole (Torsvik et al., 2012) implies separation of Armorica and Gondwana between 410 and 400 Ma (Fig. 5). However, exclusion of the anomalous Aïr pole yields a smoother Gondwana apparent polar wander path, and implies separation of Armorica in the early to middle Silurian (ca. 430 Ma; Fig. 5).

The existence of an Armorican microplate and its associated Paleotethys Ocean is seemingly at odds with faunal and stratigraphic data interpreted to show that the Variscan autochthon remained sedimentologically and paleontologically linked to north African Gondwana throughout the Paleozoic (Robardet, 2003; Linnemann et al., 2004). However, the presence of distinct ophiolite-bearing allochthons (i.e., distinct oceanic sutures) on either side of the initially linear Iberian segment of the Variscan orogen suggests that the autochthon separated. Shaw and Johnston (2016) explained the Gondwanan faunal ribbon and stratigraphic character of Armorica by modeling it as a ribbon continent that remained close or connected to Gondwana to the south and extended north, separating the Paleotethys Ocean to its east from the Rheic Ocean to its west (Fig. 6A). A similar model was proposed by Stampfli et al. (2002). The Shaw and Johnston (2016) model is consistent with the paleomagnetically constrained north-south trend for the palinspastically restored Iberian coupled oroclines, and assumes counterclockwise rotation of Armorica during separation resulting in the Pyrenean segment of the restoration at its northern end (Fig. 6B). The paleomagnetically constrained maximum width of the Paleotethys exceeds the 2300 km length of the palinspastically restored Iberian oroclines; however, the nature and number of oroclinal bends across the full extent of the western European Variscan orogen have yet to be determined (Warr, 2012).

Paleomagnetic data indicate that at the onset of Variscan orogenesis within the Iberian Massif, as recorded by the ca. 377 Ma obduction of the Galicia–Trás-os-Montes allochthon, the Paleotethys Ocean was at its widest, with Gondwana far removed from the deforming Variscan autochthon (Fig. 5). Despite translating steadily northward, Gondwana remained separated from Armorica throughout the various stages of Variscan orogenesis. Including the Late Devonian Galicia–Trás-os-Montes obduction, the main orogenic stages are Late Devonian to early Carboniferous hinterland crustal thickening along east-verging thrusts and folds; mid-Carboniferous (ca. 330 Ma) closure of the Pulo de Lobo suture; extensional collapse of the overthickened hinterland; and late Carboniferous (ca. 320–310) formation of the east-verging foreland fold and thrust belt (Martínez Catalán et al., 2007, and references therein). Northward migration of Gondwana proceeded at a steady rate, averaging 3 cm⋅yr–1 throughout Variscan orogenesis (from 380 to 310 Ma). However, upon cessation of east-west Variscan shortening at 310 Ma, Gondwana accelerated. From 310 to 290 Ma, coincident with the formation of the Iberian oroclines, Gondwana migrated northward at >6.5 cm⋅yr–1. These observations suggest that it is the Iberian oroclines, not the Variscan orogen, that formed during Pangean amalgamation and provide a record of its associated deformation.

If Variscan orogenesis is not the product of terminal Gondwana-Laurussia collision, then what caused it? Here we present an alternative model demonstrating that the nature and timing of the main stages of Variscan orogenesis may be explained in terms of individual accretionary events generated through continuous, west-dipping subduction of and along the margins of the Armorican ribbon continent, as proposed in Shaw and Johnston (2016).

The simplest model for producing the Armorican ribbon continent is counterclockwise rotation of the ribbon about a pole of rotation near its westernmost end (Fig. 6B). In such a model, the west end of the ribbon continent may remain fixed to Gondwana, while its easternmost end rotates the farthest from Gondwana, and forms the northern (Pyrenean) portion of the Iberian Variscides. Our focus here is only on the Iberian Variscides; the relative locations of the Armorican and Bohemian Massifs along the ribbon continent are unknown. Counterclockwise rotation of Armorica suggests that it was situated between an older ocean to the west and a newly opened ocean to the east. The presence of an older preexisting ocean to the west of the Armorican ribbon continent is consistent with shallow-marine offshore paleocurrent data from lower Ordovician strata of the Iberian autochthon (Shaw et al., 2012). This older ocean to the west is traditionally referred to as the Rheic Ocean; in our model the older ophiolite of the Galicia–Trás-os-Montes allochthon was in part derived from this ocean. The younger Pulo de Lobo suture on the eastern side of the orogen ribbon continent, which traditionally has been interpreted as having been derived from the Rheic Ocean, may represent in part the younger ocean (i.e., Paleotethys) that opened during separation of the ribbon continent from Gondwana. Neither the Galicia–Trás-os-Montes nor Pulo de Lobo sutures can be representative of oceanic closure during terminal Gondwana-Laurussia collision because both predate it.

Assuming separation of the Armorican ribbon continent from Gondwana at 430 Ma (Fig. 5), Variscan orogenesis began ∼50 m.y. later with the ca. 377 Ma obduction of the Galicia–Trás-os-Montes allochthon along its western margin. Interpretation of the Galicia–Trás-os-Montes allochthon as an oceanic arc fueled by west-dipping subduction of oceanic lithosphere continuous with the Armorican ribbon continent (Fig. 7A) is consistent with the presence of calc-alkaline orthogneisses and metagabbros with the composition of island arc tholeiites within the upper levels of the nappe stack (Andonaegui et al., 2002; Pin et al., 1992). Arc-continent (Galicia–Trás-os-Montes ocean arc–Armorican ribbon continent) collision (Fig. 7B) is interpreted as the driving mechanism for Late Devonian to early Carboniferous hinterland crustal thickening along east-verging thrusts and folds. Whole-rock and muscovite 40Ar/39Ar age analyses reveal diachronous 359–336 Ma development of axial planar cleavage from the allochthon eastward and a youngest age of thrusting at the eastern limits of the external hinterland domain at 321 Ma (Dallmeyer et al., 1997). Eastward-younging granitic magmatism initiating ca. 350 Ma within the hinterland domains of northern Iberia (Fernández-Suárez et al., 2000) suggests the continuation of west-dipping subduction along the eastern margin of the Armorican ribbon continent following collision and accretion of the Galician arc.

The Ossa Morena autochthonous domain translated northward along the sinistral the Badajoz-Córdoba strike-slip fault to reach its present position along the eastern margin of the southern Iberian segment of the Armorican ribbon continent by 350 Ma (Fig. 7C). The 390 Ma onset of calc-alkaline magmatism within the Ossa Morena zone (Nance et al., 2010) indicates that westward-directed subduction beneath it was ongoing since the Middle Devonian, and may suggest that collision of the Galician arc was diachronous south to north along the length of the ribbon. 40Ar/39Ar cooling ages of ca. 370–360 Ma on amphibole and 340–330 Ma on muscovite from ductile domains within the central Badajoz-Córdoba shear zone have been interpreted to date postmetamorphic transpressional uplift (Quesada and Dallmeyer, 1994). However, 340–330 Ma muscovite cooling ages are coeval with the ca. 340 Ma cessation of calc-alkaline volcanism within the Ossa Morena domain (Nance et al., 2010) and may reflect the termination of subduction due to closure of the ocean that was to the east of the ribbon continent. Ocean closure led to the accretion of the South Portuguese peri-Gondwanan allochthon along the Pulo de Lobo suture on the eastern margin of the southern Iberian segment of the Armorican ribbon continent (Fig. 7D).

Extensional collapse of the overthickened external and internal hinterland domains had peaked by 320 Ma, yet is commonly cited as the driving mechanism for late Carboniferous (ca. 320–310) formation of the east-verging Cantabrian foreland fold and thrust belt (e.g., Martínez Catalán et al., 2007). Instead, we suggest that imbrication of the foreland fold and thrust belt may be related to a ca. 320 Ma shift in the relative rates of convergence and roll-back with ongoing west-dipping subduction beneath the eastern margin of Armorica (Fig. 7E). As subduction stepped outboard following ca. 330 Ma collision of the peri-Gondwanan allochthon in southern Iberia, the entire length of the Iberian segment of the Armorican continental ribbon was characterized by active arc volcanism at the 310 Ma onset of oroclinal buckling (Figs. 7F, 7G).

The age distribution of what are commonly regarded as post-tectonic Variscan magmatic rocks (see compilation of Gutiérrez-Alonso et al., 2011b) (Fig. 2) is consistent with eastward migration of arc magmatism in northern Iberia as the core of the newly forming convex toward the west Cantabrian orocline pulled away from the plate boundary, and westward migration of arc magmatism in southern Iberia as convergence was accelerated at the hinge of the convex toward the east Central Iberian orocline. Permian–Carboniferous arc magmatism across Iberia ceased once oroclinal buckling, and hence Pangea amalgamation, was complete.

The post-Variscan formation of the coupled Iberian oroclines by vertical-axis buckling of an initially linear 2300-km-long segment of the Variscan orogen required >1100 km of orogen-parallel shortening and the consumption of vast tracks of oceanic lithosphere, neither of which could have been accommodated at the continental core of Pangea. The traditional interpretation of the western European Variscan orogen as the record of terminal Pangea-forming Gondwana-Laurussia collision must therefore be reconsidered. Paleomagnetic data indicate that (1) the Variscan autochthon (Armorica) separated from its north Gondwanan point of origin ∼50 m.y. prior to the onset of Variscan orogenesis and (2) Gondwana-Armorica collision postdates Variscan orogenesis but is coincident with formation of the Iberian coupled oroclines. The Variscan is an accretionary orogen attributable to continuous westward subduction leading to collision of the Armorican ribbon continent with an oceanic arc represented by the Galicia–Trás-os-Montes allochthon, and the subsequent accretion of the South Portuguese peri-Gondwanan allochthon. Late Carboniferous geometry of the Armorican ribbon continent is given by palinspastic restoration of Variscan oroclines, for which only the Cantabrian–Central Iberian pair are adequately constrained. The true record of Pangea amalgamation is recorded by oroclinal buckling of the Variscan orogen–Armorican ribbon continent in response to its entrainment between Laurussia to the north, and a northward-migrating Gondwana to the south.

Funding for the production of this manuscript was provided by a Natural Sciences and Engineering Research Council of Canada (NSERC) (RGPIN-2014-06533) Discovery Grant awarded to Stephen T. Johnston.