Recently discovered Albian submarine diatremes sample tectonites from the Basque-Cantabrian Basin concealed basement. Garnet-sillimanite gneiss xenoliths provide petrologic, petrofabric, and radiometric evidence on a so-far unidentified early Permian (ca. 280–275 Ma) high-grade regional metamorphism. It is interpreted as a syntectonic event recorded at middle to lower crustal settings and related to the Variscan orogeny. This metamorphism is 25–30 m.y. younger than the Variscan high-grade regional metamorphism preserved in the axial zone of the Pyrenees and 15–30 m.y. younger than low-pressure plutonometamorphism documented in the core of the Ibero-Armorican arc. By contrast, it is 15–20 m.y. older than Variscan regional and contact metamorphism unraveled in basement outcrops around the Basque-Cantabrian Basin; it can be thoroughly discriminated from Proterozoic high-grade metamorphism preserved in lower crustal segments of the submerged north Iberian continental margin, and from low-pressure metamorphism related to mid-Cretaceous hyperextension predating the Alpine orogeny. The early Permian metamorphism reveals protracted and diachronic Variscan tectonomagmatic activity in the core of the Ibero-Armorican arc due to orocline buckling processes triggering lithosphere delamination, asthenosphere upwelling, and eventually heat and mass transfer through the continental crust.

The Pyrenees and their prolongation in the northern edge of the Iberian microplate (onshore and offshore) record a protracted geological history including various tectonometamorphic events. The most outstanding of them are those related to the Variscan (late Paleozoic) and Alpine (Paleogene) orogenies. These included various phases of regional deformation, orogenic and contact metamorphism, and magmatism (Choukroune, 1992; Barnolas et al., 1996; Ábalos et al., 2002; Capote et al., 2002). The Alpine orogeny was preceded by a Cretaceous lithospheric hyperextension process that also associated specific deformation and metamorphism (Lagabrielle and Bodinier, 2008; Jammes et al., 2009). Nevertheless, the geological history of the northern Iberian continental crust can be traced back to the Archean, in order to include Neoarchean and Paleoproterozoic magmatism (Guerrot et al., 1989), Mesoproterozoic exhumation of metamorphic rocks (unraveled in lower crustal exposures from the submerged continental margin; Gardien et al., 2000), and Neoproterozoic peripheral orogen tectonics (recorded in the onshore basement outcrops; e.g., Ábalos et al., 2002; Díaz-García, 2006; Casas et al., 2015).

Regional tectonic syntheses of the Variscan orogeny relate the basement of the Basque-Cantabrian Basin (BCB; Barnolas and Pujalte, 2004; Ábalos, 2016) to two major tectonic domains: the core of the Ibero-Armorican arc to the west (e.g., Weil et al., 2013; Murphy et al., 2016, and references therein) and the Pyrenean axial zone to the east (Barnolas et al., 1996; Carreras and Druguet, 2014). The outcrops of these domains are separated by the Mesozoic–Cenozoic sedimentary infill of the BCB, which geologically constitutes the western prolongation of the Alpine Pyrenees. The nature of the BCB concealed basement has remained essentially unknown to date.

In this article we report the occurrence of high-grade tectonites in the concealed basement of the BCB, sampled by submarine Albian diatremes. They bear fabrics (defined by a high-grade mineral assemblage) that resulted from orogenic metamorphism and contain neoformed metamorphic zircon grains and overgrowths. The relationships between petrofabrics and geochronology unveiled in this study allow the identification of a Permian deep-seated orogenic metamorphism that can be discriminated from the several other metamorphic events known in the area. They reveal a protracted Variscan tectonomagmatic activity in the core of the Ibero-Armorican arc; they also provide evidence of the currently missing (slightly older) deep crustal regional metamorphic counterparts of the well-known mid-shallow crustal (slightly younger) ductile deformations coeval with Variscan granite emplacement that predated postorogenic Permian magmatism.

Regional Geology

The BCB geological evolution relates to Mesozoic opening of the Bay of Biscay. A rift developed after Triassic and latest Jurassic–Early Cretaceous extensional stages and evolved to a hyperextended continental margin in the middle Albian (Jammes et al., 2009). As a result of this evolution, a sedimentary trough developed along the boundary separating the Iberian and European plates. It was inverted during the Paleogene (Ábalos, 2016, and references therein). Extreme stretching and crustal thinning permitted submarine lower crustal and upper mantle exhumation coeval with marine sedimentation (Lagabrielle and Bodinier, 2008; Clerc and Lagabrielle, 2014; Tugend et al., 2014; Masini et al., 2014), accompanied by alkaline magmatism in central parts of the trough between the Albian and the Santonian (Lamolda et al., 1983; Rossy, 1988; Ubide, 2013).

Mid-Cretaceous Magmatic Rocks and Deep-Water Diatremes

Middle Albian–early Cenomanian turbidites and pelagic deposits constitute the so-called Black Flysch of the BCB (Souquet et al., 1985; Agirrezabala, 1996). It exhibits a complex lithostratigraphic and facies arrangement constrained by synsedimentary faults and diapirs (Poprawski et al., 2014) connected to marginal deltaic slope sedimentary systems (Robles et al., 1988). Submarine volcaniclastic rocks and pillow lavas dated as ca. 102 Ma (Castañares et al., 2001; López-Horgue et al., 2009) occur interbedded in the middle to upper part of the Black Flysch. Agirrezabala et al. (2014) and Agirrezabala (2015) described sills and laccoliths emplaced syndepositionally (∼500 m beneath the paleoseafloor) in the Black Flysch and in the underlying Monte Grande Formation.

Deep-water diatremes were discovered recently in the north-central BCB, in a cliff-dominated area along the current shore (Fig. 1). The diatremes are exposed with a semidiurnal periodicity during low tide periods. Their outcrops show that they crosscut the Monte Grande and Black Flysch units (Fig. 2). The disconformity-bounded Monte Grande Formation consists of sandstones, sandy mudstones rich in vegetal fragments, conglomerates, and fossil-rich limestones deposited in a retreating fan delta. The Black Flysch is composed of organic-rich mudstones, turbiditic sandstones, conglomerates, and siderite layers deposited in deep-water slope environments (Robles et al., 1988). The diatremes exhibit sections at different stratigraphic levels of the above units and are emplaced along a paleofault scarp sealed by the Black Flysch.

Diatreme outcrops present subcircular to elongate sections, inward-dipping steep walls, and smooth to very irregular sharp contacts with the moderately to steeply dipping host country rocks (Fig. 2). This permits us to envisage diatreme inverted conical shapes with subvertical axes. Intrusive contacts are irregular, smooth, or entangled by bedding-parallel diatreme offshoots. Country-rock fractures filled by the diatreme coexist with clastic dikes filling diatreme fissures. Diatremes are filled with hydrothermally altered, primary magmatic and sedimentary rocks, together with lithic fragments (sedimentary, volcanic, and scarce crystalline-rock xenoliths sampled for this study). The geological data gathered from field work currently in progress suggest that the outcrops represent lowermost and root zones of a diatreme volcanic system related to submarine eruption of an incipiently vesicular trachytic magma.

Igneous and Lithic Diatreme Infill

Diatreme infill consists of a mixture of volcaniclastic deposits and lithic clasts derived from a primary magma that have all been hydrothermally altered. Four diatreme-filling facies are distinguished: foliated breccia, vitric-lithic coarse lapilli tuff, vitric coarse lapilli tuff, and monomictic breccia. Clasts of solidified primary magma include fluidal and angular-shaped trachyte of fine lapilli to bomb size, accompanied by minor amounts of spheroidal pyroclasts (≤65 mm pelletal lapilli; Carracedo-Sánchez et al., 2015). Lithic clasts include fragments of the sedimentary host (≤1 m in size), recycled diatreme-filling fragments (≤0.25 m), and small (≤0.20 m) crystalline xenoliths.

Diatreme emplacement caused contact metamorphism and organic-matter carbon devolatilization in the host sediments. This resulted in the generation of methane- and CO2-rich fluids, the hydrothermal oxidation of the organic matter, and a CO2-derived dolomitization at temperature, T >400–500 °C (Agirrezabala et al., 2014; Agirrezabala, 2015) also facilitated the hydrothermal alteration of the xenoliths.

Crystalline Xenoliths

Crystalline xenolith lithologies include quartzite, high-grade pelitic gneiss, biotite micaschist, and foliated porphyric granodiorite. They were pervasively overprinted by microfracture networks (likely related to the country-rock explosive fragmentation) and metasomatized along with them.

An irregular gneiss xenolith of 6 × 20 cm (exposed dimensions) (Fig. 3A) shows phyllosilicate-rich and phyllosilicate-poor metamorphic differentiates parallel to a mineral foliation. The former domains preserve a high-grade metamorphic assemblage including garnet (Fig. 3B), sillimanite (Figs. 3C, 3D), quartz (Fig. 3E), rutile (Fig. 3F), ilmenite, graphite (Figs. 3G, 3H), apatite, monazite, and zircon. Accessory secondary minerals are leucoxene, chlorite, and widespread kaolinite (Figs. 3B, 3E, 3G). Its mineral diversity and microstructural features enable completion of a combined mineral chemistry, petrofabric, and geochronological study.

A granodiorite xenolith (smaller than the previous one) contains oriented plagioclase porphyrocrystals depicting a magmatic fabric without discernible tectonic overprint. Quartzite and biotite micaschist xenoliths exhibit tectonite fabrics. These xenoliths, however, lack zircons for dating or are inadequate for a combined mineral chemistry and petrofabric study due to their strong hydrothermal alteration.

The analytical techniques used for this study include determination of mineral compositions, mineral structural formulae, crystallographic preferred orientations, and radiometric dating using U, Th, and Pb isotopes.

Mineral Composition Microprobe Measurement

Standard 30-µm-thick polished rock sections were used for mineral microprobe analysis. Mineral composition measurements were performed at the Servicios Científico-Técnicos facility of the University of Oviedo with Cameca SX-50 and SX-100 automatic microprobes. The latter is equipped with five wavelength dispersive spectrometers, a dispersive energy spectrometer, and secondary electron, backscattered electron, and cathodoluminescence detectors. The operating parameters included a 10 s integration time, a 10 nA beam current, and a 15 kV accelerating voltage. Scanning electronic microscopy with energy-dispersive X-ray spectroscopy (EDX) was employed with a JEOL JSM-7000F field emission scanning electronic microscope (FE-SEM) at the University of the Basque Country (Electron Microscopy Facility-SGIker, Servicios Generales de Investigación; for further details, see following) in order to confirm the occurrence of some mineral phases such as graphite, quartz, pyrite, zircon, and monazite without a full compositional analysis.

Electron Backscattered Diffraction

Crystallographic preferred orientation measurements were performed at the University of the Basque Country (Electron Microscopy Facility-SGIker) with an automated electron backscattered diffraction (EBSD) system (Channel5, HKL) attached to a JEOL JSM-7000F FE-SEM. Standard 30-µm-thick polished rock sections were used for the EBSD microfabric study. They were ultrapolished with a colloidal silica suspension to remove surface damage and carbon coated to prevent charging; a copper tape was attached surrounding the measurement area to reduce charging effects. Samples were mounted in the apparatus on a stage tilted 70°, with the thin section long axis parallel to the SEM X-axis. The beam working distance was 20 mm (Prior et al., 1999) and the detector was placed at 185 mm. An acceleration voltage of 15 kV and a beam current of ∼4 nA were applied. Crystallographic orientations and orientation maps were obtained using the HKL CHANNEL 5.0 software package ( after automated EBSD analysis on predefined sampling grid steps of 20 µm, significantly smaller than the average grain size of the minerals. Crystallographic orientation solutions with mean angular deviation values >1.2° (between detected and simulated patterns) were rejected to assure EBSD measurement reliability. The data were filtered so that the orientation diagrams contain one orientation per grain. The grain detection technique considered a critical misorientation threshold of 10°, allowing boundary completion down to 5°. In order to avoid errors in the grain detection related to the eventual presence of nonindexed pixels and the large grain size of the analyzed minerals, only those grains sampled by more than two contiguous pixels were considered.

Optimal half-widths for calculation of the orientation distribution function (ODF) were determined with MTEX texture analysis software (; Hielscher and Schaeben, 2008; Bachmann et al., 2010; Mainprice et al., 2011). The strength of the fabrics was expressed through the J texture index (Bunge, 1982), which corresponds to the mean square value of the ODF. The calculations were done with the MTEX texture analysis software, too. Complementary fabric characterization was accomplished following eigenvalue analysis methods (Woodcock, 1977; Vollmer, 1990; Mainprice, 2016).

Laser Ablation–Inductively Coupled Plasma–Mass Spectrometry U-Pb Zircon Dating Method

U-Th-Pb isotopic measurement of zircon grains was conducted by laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS; Tiepolo, 2003). Table 1 summarizes the operating conditions of the LA-ICP-MS apparatus (Thermo Fisher Scientific iCAP Qc) at the Geochronology and Isotope Geochemistry facility of the SGIker (Faculty of Science and Technology, Universidad del País Vasco/Euskal Herriko Unibertsitatea, UPV/EHU, Bilbao, Spain). Equipment sensitivity was improved with a second vacuum plumb in the system interface. Signal smoothing was accomplished with the Tunheng and Hirata (2004) stabilizer.

The analyses involved the ablation of minerals in ∼30-µm-thick petrographic sections with a UP213 frequency-quintupled Nd:YAG (Nd-doped yttrium aluminum garnet) based laser ablation system (NewWave Research, Fremont, USA) coupled to a quadrupole based ICP-MS instrument with a dual pumping system to enhance the sensitivity. Nominal spot diameters of 30 µm associated with repetition rates of 10–5 Hz and laser fluence at the target of ∼5.2 J/cm2 were used for zircon dating. The ablated material was carried into He, and then mixed with Ar, before injection into the plasma source. The alignment of the instrument and mass calibration was performed before every analytical session using the NIST SRM 612 reference glass by inspecting the signal of 238U and by minimizing the ThO+/Th+ ratio (<<1%). The mean sensitivity on 238U at the instrumental conditions reported in Table 1, using a spot size of 100 mm, was ∼40,000 cps/ppm. The analytical method for isotope dating is basically similar to that reported in Tiepolo (2003) and Paquette and Tiepolo (2007). The signals of 204(Pb + Hg), 206Pb, 207Pb, 208Pb, 232Th, and 238U masses were acquired. The occurrence of common Pb in the sample can be monitored by the evolution of the 204(Pb + Hg) signal intensity, but no common Pb correction was applied owing to the large isobaric interference from Hg. The 235U signal is calculated from 238U on the basis of the ratio 238U/235U = 137.88. Single analyses consisted of 25 s of background integration with laser off followed by 45 s integration with the laser firing and a 30 s delay to wash out the previous sample (∼10 s for 6 orders of magnitude) and prepare the next analysis.

Data were corrected for U-Pb and Th-Pb fractionation occurring during laser sampling and for instrumental mass discrimination (mass bias) by standard bracketing with repeated measurements of the GJ-1 zircon standard (Jackson et al., 2004). At the beginning and at the end of every run, repeated analyses of the 91500 zircon standard (Wiedenbeck et al., 1995) and the Mud Tank standard (Black and Gulson, 1978), treated as unknowns, controlled independently the reproducibility and accuracy of the corrections. The data reduction was carried out with Iolite 3.32 (Paton et al., 2011) and VizualAge (Petrus and Kamber, 2012) using the GJ-1 zircon standard (Jackson et al., 2004). For each analysis, the time-resolved signals of single isotopes and isotope ratios were monitored and carefully inspected to verify the presence of perturbations related to inclusions, fractures, mixing of different age domains, or common Pb. Calculated ratios were exported and concordia ages and diagrams were generated using the Isoplot/Ex v. 2.49 package (Ludwig, 2001). Concentrations in U-Th-Pb were calibrated relative to the certified contents of the GJ-1 zircon standard (Jackson et al., 2004). Percentage concordance was calculated as [(206Pb/238U age)/(207Pb/206Pb age)]×100 (Meinhold et al., 2011).

Xenolith Mineralogy and Phase Relationships

The primary constituents of the gneiss xenolith high-grade assemblage are garnet, sillimanite, quartz, rutile/ilmenite, and graphite (Fig. 3). Volumetrically minor mineral constituents are pyrite, zircon, and monazite. Widespread hydrothermal alteration affected the xenoliths and is manifested by variable retrogression of most primary minerals by kaolinite. The structural formulae of these minerals were calculated by charge balance criteria (e.g., Spear, 1993; Brandelik, 2009) following various procedures for different phases. The weight percentage of each oxide analyzed and the calculated structural formulae are presented in Table DR1 (garnet; 32 analyses), Table DR2 (sillimanite; 14 analyses), Table DR3 (rutile; 5 analyses), Table DR4 (ilmenite; 11 analyses), Table DR5 (variably altered rutile and ilmenite; 13 analyses), Table DR6 (plagioclase; 14 analyses), and Table DR7 (kaolinite; 69 analyses) in the GSA Data Repository1.

In the gneiss xenolith, homogeneous, unzoned garnet is intensely corroded (Fig. 3B). It is rich in almandine (∼70%) and pyrope components (∼25%, see Table DR1) and poor in grossular (<5%). Sillimanite (Figs. 3C, 3D; Table DR2) is nearly pure Al silicate. Rutile and ilmenite (Fig. 3F; Tables DR3 and DR4) are stoichiometrically nearly pure. Both are variably altered to opaque phases with 96%–11% TiO2, and Al2O3 in the range 28%–1% (Table DR5). Graphite (Figs. 3B–3E, 3G, 3H) is chemically pure and fully ordered C (as determined by EDX).

The granodiorite xenolith mineralogy is dominated by quartz and plagioclase (essentially andesine); the anorthite molar fraction of the latter varies between 0.43 and 0.51 (Table DR6), although pure albite was also detected.

The most widespread alteration product of both the gneiss and the granitoid samples is kaolinite (Table DR7). In the gneiss this hydrated Al-silicate usually bears a small amount of K and very low contents of Mg and Fe.

The equilibrium garnet-sillimanite-rutile-quartz-graphite high-grade mineral assemblage does not expedite the use of specific thermobarometers based upon mineral compositions (e.g., Spear, 1993). Only semiquantitative determinations are possible after the evidence of sillimanite stability, the occurrence of garnet, and the absence of evidence of partial melting. The phase relationships in the KFMASH (K2O-FeO-MgO-Al2O3-SiO2-H2O) pelite system (Spear, 1993, p. 373) show that the stability conditions of the gneiss primary assemblage range between ∼570 and 670 °C at 0.4–0.6 GPa. As regards the subsequent retrogression, the small K, Mg, and Fe contents determined for kaolinite (Table DR7) are suggestive of a retrogressive origin after muscovite. Muscovite replacement and sillimanite partial retrogression by kaolinite (along fractures and reworked subgrain boundaries) might have involved an intermediate stage of pyrophyllite formation at temperatures of 400–425 °C (for a wide pressure range; Spear, 1993). Transformation of pyrophyllite to kaolinite is favored by water availability in the system under temperature and pressure conditions of 325–350 °C and 0.1–0.2 GPa, respectively (Thompson, 1970).

Xenolith Petrofabric

Quartz ribbon grain shape fabrics and graphite and/or phyllosilicate crystals appear arranged in two orientations, one parallel to the internal foliations of sillimanite porphyrocrystals and a second one systematically oblique to it (at ∼30°; Fig. 4). This microstructure of the gneissic xenolith is reminiscent of an S-C fabric, well known as shear sense indicators in rocks ductilely deformed under regimes incorporating rotational components. Sillimanite occurs as elongated porphyroblastic aggregates as large as 3 × 8 mm defining the rock foliation. It associates internal and external foliations, the geometrical relationships of which point to a syntectonic mineral growth (Fig. 4). Sillimanite aggregates consist of fully recrystallized, variably misoriented small grains. They are traversed by fractures normal to the foliation due to stretching parallel to it.

Fabric orientation distributions of the high-grade gneiss minerals are presented in Figure 5 using lower hemisphere, equal area stereographic diagrams. The projection plane of the stereoplots is close to the structural XZ planes (X—lineation direction; XY—foliation plane). The macroscopic foliation is represented there as the equatorial diameter (E–W), while the lineation is horizontal within the same plane.

Sillimanite crystallographic axes exhibit a strong preferred orientation (crystal preferred orientation, CPO; Fig. 5A), with the poles to (010) normal to the foliation and the [001] axes forming a girdle with a clear maximum parallel to the lineation. The CPOs of rutile, graphite, and apatite (Figs. 5B, 5C, 5D, respectively) exhibit an internal consistency and reproduce similar geometrical relationships between specific mineral axes/planes and the macroscopic foliation/lineation. This suggests that the high-grade assemblage was syntectonic.

Graphite crystals 100 × 500 µm and much smaller flakes are subparallel to the gneiss foliation. The larger crystals are usually deformed internally by complex angular and kink folds, giving rise to extremely thin (<5 µm) graphite sheet displacements and separations, the voids being filled by kaolinite (Figs. 3G, 3H). These features point to noncoaxial deformation regimes, including flattening components normal to the foliation and stretching parallel to it.

Lozenge-shaped quartz ribbons are made of polycrystalline quartz (Figs. 3D, 3E, and 4). Thick ribbons contain quartz with subgrain boundaries arranged parallel and normal to the foliation, forming a mosaic microstructure typical of high-T plastic deformation. Thin ribbons contain dynamically recrystallized grains and slightly misoriented subgrains with internal microstructures denoting intracrystalline creep. Quartz CPOs (Fig. 5E) contain c-axis submaxima oblique but close to the foliation pole and the lineation, connected by an ill-defined girdle. This petrofabric is ambiguous; on one hand, it might reflect the activation of basal <a> and prism [c] intracrystalline slip (characteristic of high-T quartz deformation, concomitantly with ribbon-quartz microstructure). On the other hand, it might reflect basal <a> slip under noncoaxial, low-T deformation dominated by flattening components. A poor definition of quartz <a> CPOs precludes further judgment. Quartz grain boundaries and subboundaries are usually the sites of kaolinite precipitation as microveins (some antitaxial), pointing to quartz crystalline discontinuity brittle reactivation during explosive volcanism (Fig. 3E).

Xenolith Geochronology

Zircon and monazite dating was attempted with the LA-ICP-MS. While zircon dating was successful, monazite analyses revealed large amounts of common Pb that in most cases impeded determination of reliable crystallization ages for this mineral.

The analyzed zircon grains vary in sectional shape (Fig. 6) from rounded (likely recycled) to subidiomorphic (likely igneous or overgrown around inherited grains); some have rounded cores (Figs. 6A, 6B). Internal zircon features reveal irregular or diffuse zoning and either homogeneous or rythmic overgrowths suggestive of magmatic growth zoning structures (e.g., Figs. 6B, 6F). Compositionally, zircons exhibit large ranges of Pb, Th, and U contents (Table 2). Low Th/U ratios (<0.5, although usually <0.05) characteristic of metamorphic zircons (Heaman et al., 1990; Hoskin and Ireland, 2000; Rubatto, 2002) are frequent (85% of the zircons), whereas Th/U ratios >0.5, typical of a magmatic origin, are less common.

The U-Pb spot measurements performed after zircon grains permitted us to select 20 analytically valid analyses (out of 24) from which 11 radiometric ages resulted concordant within the 95%–105% range (gray shaded in Table 2; Fig. 7). The nearly concordant U-Pb radiometric ages can be arranged in three groups (gray-shaded bands in Fig. 8): 497 ± 12 to 443 ± 13 Ma (late Cambrian–Ordovician), 346 ± 8 to 323 ± 11 Ma (early Carboniferous), and 282 ± 6 to 275 ± 5 Ma (early Permian). A supplementary inherited component corresponds to discordant Paleoproterozoic (Statherian) and subconcordant Mesoproterozoic (Tonian–Stenian) ages.

Early Carboniferous ages were found both in zircon cores and in cyclic overgrowths around inherited Ordovician rounded cores. The early Permian zircon group is represented by homogeneous cores in likely neoformed, small, and subidiomorphic zircon grains (276 ± 5 Ma and 280 ± 5; Figs. 6A, 6C, 6D), and by irregular and/or diffusely zoned overgrowths around Ordovician (282 ± 6 Ma; Figs. 6E, 6F) or older cores (Fig. 6B), or around early Carboniferous overgrowths (295 ± 11 Ma, 282 ± 6 Ma). In all cases zircon Th/U ratios are very low (usually <0.05), independent of the large ranges of Pb, Th, and U contents measured. This feature is characteristic of metamorphic zircons and contrasts with the Th/U ratios >0.5 of most older zircons (Table 2), suggestive of a magmatic origin.

The textural relationships and Th/U isotopic ratios described here for the 282 ± 6 to 275 ± 5 Ma (early Permian) zircons indicate that they either overgrew preexisting grains or were neoformed (Fig. 6) during a high-T thermal event. In the former case, zircon cores (usually Ordovician) and early Carboniferous rims (both overgrown by the irregular and/or diffusely zoned early Permian zircon coats) have characteristics usually associated with an igneous origin. Radiometric evidence of metamorphic zircon growth before the early Permian is rare (a single, 97% concordant, 295 ± 11 Ma age; Table 2) and overlaps the early Permian event identified herein. This hampers consideration of multiple thermal events that would permit a disconnection between an older high-grade metamorphism and a younger zircon growth. Consequently, we interpret that the formation of the garnet-sillimanite gneiss and metamorphic zircon growth were eventually related to a single thermal overprint.

The data and observations presented here suggest that early Permian high-grade regional metamorphic rocks occur in the concealed basement of the BCB and that ascending magma carried them to the seafloor during diatreme formation in Albian time. Because early Permian metamorphic neoformed zircons and zircon overgrowths occur in a gneissic rock with a syntectonic microstructure and a clearly oriented high-grade assemblage mineral petrofabric, metamorphism should be related to the Variscan orogeny. As discussed in the following, this Permian deep crustal event is not consistent with known regional Variscan tectonometamorphic events in the area surrounding the BCB that are 15–30 m.y. older.

Currently, the basement outcrops nearest the diatreme site are >120 km to the west (pre-Mesozoic outcrops of the Iberian Massif) and >80 km to the east (pre-Mesozoic outcrops of the Pyrenees); the former correspond to the core of the Ibero-Armorican arc (the Pisuerga-Carrión domain of the Cantabrian Zone foreland), whereas the latter relate to more internal zones of the Variscan orogen in the northern arm of the arc (currently forming the westernmost outcrops of the Pyrenean Axial Zone in the so-called Basque massifs; e.g. Ábalos et al., 2002, and references therein). In none of such onshore areas, however, have deep-seated tectonometamorphic events been described, other than those related to contact aureoles of latest Carboniferous and Permian intrusions (Pesquera, 1985; Gallastegui et al., 1990).

The closing stages of the Variscan orogeny in the Ibero-Armorican arc and the Pyrenees include Carboniferous flysch basin fill, late Carboniferous deformation, regional metamorphism, and emplacement of late Carboniferous–Permian granites in a context of regional transpression followed by orogenic collapse (Laumonier et al., 2010; Martínez-Catalán et al., 2014; Murphy et al., 2016). In the Pyrenees the ages of synorogenic flysch deposits (Martínez et al., 2016) range between ca. 340 Ma (Visean, Middle Mississippian) and ca. 315 Ma (Bashkirian, Early Pennsylvanian), and compare to similar successions in the core of the Ibero-Armorican arc. The 315 Ma age approaches the time of flysch penetrative deformation, which is postdated by magmatic intrusions (granitoid plutons, granite dikes, and dolerite dikes; Denèle et al., 2014; Pereira et al., 2014).

The closure of the Ibero-Armorican arc between 310 and 305 Ma produced separate magmatic pulses, including mantle and lower crustal melting at 305–300 Ma (latest Carboniferous), 300–292 Ma (earliest Permian Asselian Age), and 292–286 Ma (early Permian Sakmarian Age; cf. Gutiérrez-Alonso et al., 2011b), interpreted to have been caused by lithospheric delamination triggered by the Ibero-Armorican arc development (Gutiérrez-Alonso et al., 2011a). According to Gutiérrez-Alonso et al. (2011a), after 285 Ma the thermal regime precluded significant mantle or crustal melting and shallow crustal emplacement of voluminous magmatic rocks. Fernández-Suárez et al. (2006) reported geochronological evidence from lower crustal xenoliths of the Spanish Central System (∼300 km to the south of the BCB) of a slightly older extensive recrystallization and/or migmatization over a 30 m.y. period, with peaks at 298 and 284 Ma; they related such lower crustal tectonothermal events to two pulses of post-tectonic granitoids, ca. 300–295 Ma and 288–280 Ma.

In the Pyrenees, quick cooling followed late Carboniferous (Ghzelian) emplacement of granitoids (e.g., 301.7 ± 7.3 Ma zircon crystallization in the Maladeta igneous complex, 300 ± 2 Ma in the Lys-Caillaouas complex, or 299 ± 4 crystallization age of the Boí granodiorite; Metcalf et al., 2007; Pereira et al., 2014; Esteban et al., 2015). Sillimanite-bearing gneisses overprinted by granite intrusions record slightly older metamorphic events there (305 ± 6 Ma; Metcalf et al., 2007). Esteban et al. (2015) identified a coeval 307 ± 3 Ma metamorphic event and related it to gneiss dome formation. The ages reported enable linking congruent Variscan deformation phases, high-T metamorphism, and pluton emplacement (Laumonier et al., 2010; Denèle et al., 2014), and demonstrate that a 5–7 m.y. lag exists between orogenic metamorphism and granitoid emplacement. Rhyodacitic and ignimbrite flows and granite dikes were emplaced in the early Permian (Cisuralian), during a time span dated between ca. 295 and 275 Ma (Pereira et al., 2014; Martínez et al., 2016). Pereira et al. (2014) determined within this time span crystallization ages for the Vielha granodiorite as young as 276 ± 3 Ma (zircon crystallization) and 280 ± 3 Ma (zircon concordia). Unconformable Permian molasse and younger successions intercalate magmatic rocks that record the transition to postorogenic rifting. Lamprophyre and dolerite dikes in the basement of Permian sedimentary basins intruded in the middle Permian (267 ± 11 Ma; cf. Galé, 2005) and mid-late Permian transition (259.2 ± 3.2 Ma cf. Rodríguez-Méndez et al., 2014), seemingly later than the youngest period of Permian hypabyssal magmatism (Guadalupian; cf. Galé, 2005). This is represented by subalkaline-alkaline trachyandesites, dolerites, and lamprophyre sills and/or dikes emplaced under late and post-orogenic transpressive, transtensive, or extensional conditions.

Permian crystalline rocks in the western Pyrenees (Cinco Villas massif) include hypabyssal magmatic rocks similar to those described here (Lago et al., 2004) and the Aya granite. The latter lacks evidence of mineral plastic deformation and was emplaced syntectonically in a dextral transcurrent regime (Olivier et al., 1999) at 267.1 ± 1.1 Ma (Denèle et al., 2012). Granite emplacement occurred after the calc-alkaline Cisuralian magmatic cycle and before the alkaline Guadalupian one. The pluton’s contact aureole exposes andalusite and cordierite-bearing rocks and rare pegmatite veins containing fibrolitic sillimanite (Pesquera, 1985). Pesquera and Velasco (1988) and Pesquera and Pons (1990) interpreted the very low- to low-grade regional metamorphism of Carboniferous country rocks as syntectonic, and inferred a syntectonic emplacement of the pluton. Permian emplacement of the Aya granite postdates granite emplacement in the Pyrenean Axial Zone (Denèle et al., 2014; Pereira et al., 2014) by at least 10 m.y., and usually by as much as 30–35 m.y.

None of the crystalline and metamorphic rocks discussed to this point compare to the diatreme xenoliths in metamorphic grade (e.g., in the Cinco Villas massif) or in metamorphism age (e.g., in the Pyrenean Axial Zone). A similar conclusion can be reached when considering late Carboniferous low-grade metamorphic rocks and Permian intrusions from the BCB basement exposed in the Pisuerga-Carrión domain of the Cantabrian Zone (Gallastegui et al., 1990). The western Pyrenean massifs also bear the fingerprints of mid-Cretaceous tectonization and/or metamorphism (Clerc et al., 2015); this is also at odds with the basement sampled by the diatremes.

Concealed basement rocks were drilled in the offshore BCB (∼20 km to the north and northeast of the area studied herein) in the 1980s and 1990s during hydrocarbon exploration of the Gaviota and Albatros gas fields, respectively. The lithologies recovered are very low-grade Carboniferous flysch and Permian organic-rich post-tectonic successions (Cortés et al., 2005). These are reminiscent of the Cinco Villas massif and differ in metamorphic grade from the diatreme xenoliths sampled.

Other high-grade metamorphic rocks were drilled and picked by submersible surveys of exhumed middle and lower crustal segments of the Iberian continental margin, currently exposed in the deep seafloor (Capdevila et al., 1974; Guerrot et al., 1989). These crystalline rocks bear Neoarchean and Paleoproterozoic igneous zircons and were already exhumed in the Mesoproterozoic (Gardien et al., 2000). Similar domains are currently under kilometer-thick Mesozoic–Quaternary continental margin cover; they might also exist onshore, concealed under the BCB. However, they can be discarded as equivalents of the basement sampled by the diatremes because they lack a Variscan metamorphic imprint.

Independent evidence on the nature of the pre-Mesozoic basement surrounding the BCB was provided by study of detrital zircon and metamorphic clasts from the diatreme-hosting Black Flysch. Puelles et al. (2014) showed that Ordovician orthogneisses and country rocks of Gondwanan affinity were exposed during the Albian in the Black Flysch source areas around the BCB. The Cinco Villas and Ursuya metamorphic massifs of the west Pyrenean realm were rejected as source areas because of the lack of evidence of Carboniferous and Permian magmatism (Denèle et al., 2012) and of mid-Cretaceous tectonization or metamorphism in their inherited zircon record (Pesquera and Velasco, 1988; Clerc et al., 2015). The mid-Cretaceous exposed source areas were ascribed to tectonometamorphic equivalents of the Armorican Massif and the French Massif Central, currently submerged in the Landes platform (Puelles et al., 2014). In any case, those areas are also at odds with the BCB concealed basement sampled by the diatremes because of the absence of Permian metamorphic rocks.

The source area of the diatreme xenoliths is linked to the geometrical center of the Ibero-Armorican arc, coinciding with the western end of its Pyrenean straight arm. This domain is acknowledged as an enigmatic segment of the Varsican orogen where mid-crustal high-T rock flow has been interpreted as related to coeval lithospheric buckling of the Cantabrian orocline transitional to shear deformation localized along the current Pyrenees (Gutiérrez-Alonso et al., 2011a, 2011b; Denèle et al., 2014). Gutiérrez-Alonso et al. (2011a, 2011b) and Denèle et al. (2014) advocated that asthenosphere upwelling after lithosphere delamination induced Variscan high-T–low-P metamorphism and magmatism, separated by a 5–7 m.y. lag. This hiatus might also be unraveled in the diatreme area if a similar relationship is assumed between the 282 ± 6 to 275 ± 5 Ma (early Permian) xenolith metamorphism and the emplacement of the nearest granitoid intrusion currently known (the 267.1 ± 1.1 Ma Aya granite; Denèle et al., 2012). The age relationships reported unravel the time scales of heat and mass transfer between lower and mid-crustal structural domains, that is, between the high-grade metamorphic realms and the realms hosting pluton intrusions above, separated by ∼10 km. Horizontal propagation of a heat front, and of the concomitant tectonothermal effects at a comparable rate (within an order of magnitude) might suffice to explain that, in the area here considered, Variscan orogenic metamorphism is 25–30 m.y. younger than in middle and lower crustal domain outcrops of the Pyrenean Axial Zone and the northwest Iberian Massif. This diachronism appears to be supported by granitoid emplacement ages and deserves further research. In any case, we foresee that the tectonic models proposed to explain the sequence of deformational, metamorphic, and plutonic events in the exposed Variscan areas (e.g., those proposed by Laumonier et al., 2010; Denèle et al., 2014, for the Pyrenean Axial Zone, and by Gutiérrez-Alonso et al., 2011a, for the northwest Iberian Massif) can account for the occurrence of early Permian high-grade metamorphic rocks in the concealed basement of the central area of the Basque-Cantabrian Basin.

The combined petrologic, petrofabric, and radiometric study of a diatreme gneissic xenolith reveals an early Permian orogenic metamorphism in the concealed basement of the BCB. It is a high-grade regional metamorphism that is 25–30 m.y. younger than the high-grade regional metamorphism of the Variscan Pyrenees, and 15–30 m.y. younger than the low-P plutonometamorphism of the core of the Ibero-Armorican arc. In addition, it is ∼15–20 m.y. older than the low-grade syntectonic metamorphism and thermal aureoles preserved by the country rocks of Permian granitoid intrusions from the BCB basement outcrops. The early Permian orogenic metamorphism departs from the Proterozoic high-grade metamorphism preserved in middle and lower crustal segments of the submerged north Iberian continental margin and from that recorded by the Variscan basement exposed in the Bay of Biscay in mid Cretaceous time.

We interpret this metamorphism as a deep crustal testimony of the Variscan orogeny, in particular as a record of rock fabric development related to pervasive ductile deformation, and of the mineral recrystallization and metamorphic reactions that took place under high-grade thermobaric conditions and moderate to low pressures. Vestiges comparable to these do not exist in surface outcrops of pre-Mesozoic rocks within the BCB or reasonably near it. The tectonic models proposed to explain the sequence of deformational, high-T–low-P metamorphism and magmatism in the exposed Variscan areas of the Pyrenean Axial Zone and the northwest Iberian Massif, involving asthenosphere upwelling after lithosphere delamination induced by buckling of the Cantabrian orocline, can account for the occurrence of early Permian high-grade metamorphic rocks in the concealed basement of the central area of the BCB.

Exhumation of the aforementioned Variscan deep crustal domains under a sedimentary cover would have been facilitated by low-angle normal faults (possibly extraction faults; Froitzheim et al., 2006) operative during Cretaceous hyperextension. Albian volcanism carried fragments of such a basement as xenoliths and emplaced them in submarine diatremes. All these processes predated Alpine deformation of the BCB. We interpret that the deep crustal source realm of the metamorphic xenoliths has not been displaced significantly by Alpine deformations (neither tangentially nor along strike) with respect to its Variscan counterparts at shallower crustal levels (currently cropping out). This assumption is based upon the fact that the Mesozoic cover was detached from the Variscan basement of the BCB during the Paleogene. In addition, Alpine inversion tectonics involving the BCB basement were accommodated notably by high-angle reverse faults.

Technical assistance of M.A. Fernández (Servicios Científico-Técnicos, Universidad de Oviedo), S. Fernández-Armas, and S. García de Madinabeitia (Servicios Generales de Investigación, Universidad del País Vasco/Euskal Herriko Unibertsitatea, UPV/EHU), and European Regional Development Fund and European Science Foundation funding are acknowledged. Financial support was provided by the Spanish Ministry of Economy, Industry and Competitiveness and the Fondo Europeo de Desarrollo Regional (MINECO/FEDER CGL2015-63530-P), and by the UPV/EHU (GIU15/05 and IT631-13 and IT930-16). The insightful reviews of three anonymous referees and of Science Editor Arlo Weil helped to significantly improve the original manuscript.

1GSA Data Repository Item 2017113, containing mineral composition microprobe measurement results (Tables DR1–DR7), is available at, or on request from