Abridged English version. – The Variscan Pyrenean belt (fig. 11) has been for long famous for its Late Carboniferous LP-HT metamorphism, characterised by the prograde succession, in medium grade metapelites, of biotite, cordierite, andalusite and sillimanite, together with staurolite and garnet [Guitard et al., 1996]. However, the discovery of two kyanite generations lead Azambre and Guitard  to propose a polymetamorphic evolution, with an early (MI) and a late (MIII) kyanite-bearing Barrovian stage, preceding and following the main LP-HT stage (MII).
The Variscan orogeny in the Pyrenees occurred from Namurian to Early Stephanian (c. 325-300 Ma), following the deposition of thick Ediacarian-Ordovician silico-clastites, Silurian to Early Caboniferous carbonates, and pre-orogenic Mid-Carboniferous flyschs.
Two main tectonic events are recorded, each one subdivided into regionally correlated sub-events (phases) (table I1), allowing a detailed correlation between tectonics, metamorphism and plutonism. The Namurian to Westphalian D1 event (c. 325-310 Ma) resulted in a S-vergent fold and thrust belt (with 100–150 km of N-S shortening) and the development of the main, sub-horizontal, Sr schistosity (D1c phase), coeval with MI. The Westphalian-Early Stephanian D2 event (310-300 Ma) was more complex. First, a syn-convergence extensional phase (N-vergent backfolds and E-W extension) resulted in the E-directed escape of the upper crust (D2a phase). Then, a renewal of the N-S shortening was marked by large upright anticlines (domes) and narrower synclines, with up to 10 km amplitudes (e.g., the Canigou anticline-Villefranche syncline pair) (D2b phase). Both D2a and D2b were coeval with MII and the emplacement of early granitoid sills and laccoliths (e.g., the Ansignan hypersthene-granite in the Agly Massif). Later on, D2 evolved into a transcurrent regime, with belt-parallel dextral transpression (D2c and D2c phases). D2c was coeval with the main stage of granite emplacement under low-grade conditions, allowing the expression of a conspicuous Mγ contact metamorphism (e.g., Mont-Louis pluton). D2d ended the D2 event, with the development of retrograde dextral-reverse mylonites. The late MIII metamorphic event encompassed D2c and D2d (and possibly D2b).
The early MI Barrovian metamorphic event
The MI Barrovian metamorphic event resulted from the crustal thickening associated with the development of the D1 intra-cratonic wedge. It was of low-grade, with a chlorite-muscovite Sr schistosity, in the part of the belt that was subsequently overprinted by the syn-MII transformation of chlorite into biotite. The only remnants of MI medium-grade conditions are found as early kyanite in the deepest domains of the Castillon, St-Barthélémy, Agly and Aston massifs, being there obliterated under high-grade MII conditions, and in the core of the Canigou anticline (Velmanya, point v in fig. 22), where a relict kyanite-staurolite-anorthite paragenesis is known, shielded by MII cordierite. The reconstructed P-T conditions at the thermal peak of MI are 5 kbar (19 km) and 575oC (fig. 22), implying the existence of a (now eroded) major D1 nappe (≥ 7 km thick).
The main MII LP-HT metamorphic event
Structural domes and medium– or high-grade MII zones are broadly coincident, high-grade conditions being only encountered in the core of the Albères massif, the southern Aston Dome and the North-Pyrenean massifs (grading there up to the LP granulite facies) (fig. 11).
Subdivisions of the MII event
The prograde MII metamorphism is essentially syn-D2a, with clear syn-kinematic growth of the medium-grade minerals, and the main regional tectono-metamorphic D2a/MII structure is evidently deformed and strongly folded by the D2b phase: the D2b domes are basically post-metamorphic. However, a detailed examination of the blastesis-deformation relationships shows that staurolite is pre- to-synkinematic for D2a, whereas andalusite is strictly synkinematic (and consequently is often observed shielding the staurolite), cordierite being syn-to post-kinematic and syn-D2b in some instances. This allows a subdivision of the MII event into three stages:
– MIIs, pre-to-syn-D2a, characterised by the staurolite-andalusite (And1 without cordierite) association, with development of a staurolite zone grading downwards into an andalusite (St → And1) zone.
– MIIa, syn-to post-D2a (but always developed prior to D2b), characterised by the cordierite (Cord1)-andalusite (And2) association (without staurolite), with development of a thin cordierite zone grading downwards into an andalusite (Cord1 → And2) zone.
– MIIb, post-D2a and syn-D2b, characterised by a large cordierite (Cord2) zone developed at the expense of an-dalusite (And → Cord2), only found in the core of the D2b anticlines (e.g., the Garonne dome).
Thus, although MII is basically pre-D2b, and the MIIs and MIIa medium-grade isogrades are folded, it appears that metamorphism was still active in the cores of the ascending D2b domes (MIIb). Moreover, in the core of some domes, prograde sillimanite is also syn-kinematic of the D2b phase, and the sillimanite-muscovite isograde may obliquely overprint the MIIa isogrades, as in the Canigou dome. This is related to the syn-D2b emplacement of granite sheets (e.g., the Canigou granite) and may be interpreted as an aureola of “regional-contact” metamorphism, noted MIIγ, that was evidently coeval with MIIb, and enhanced its effects.
P-T-t path of the MII event
The P-T-t path of the MII event may be described using the petrogenetic grids of Pattison et al.  and Pattison and Vogl  (fig. 33). From MIIs to MIIb, it records a prograde anti-clockwise path, following a post-MI clockwise exhumation path, with ≥ 7 km eroded (fig. 2B2). The MIIs pressure was close to 3 kbar (10–11 km) in the St zone and decreased to 2.5 kbar (9 km) at the MIIa stage (And2 isograde), for an estimated temperature of 540oC (based on the triple point of Holdaway , the thermobarometer of Pattison et al.  and independent fluid inclusion data by Kister et al. ). A further pressure decrease, down to 2 kbar (7 km), and a temperature increase (up to 600oC) is registered in the MIIb cordierite zone in the core of active D2b domes. Except for the cores of the domes, MIIa remained the peak temperature event, and during MIIb pressure remained constant (or was re-increasing in the syncline cores) and temperature was constant or decreasing. At the end of the MII event (MIIb-MIIγ), extreme conditions of c. 4 kbar and 700–730oC are recorded in the deepest parts of the belt, where anatexis, succeeding to a sillimanite-K-feldspar zone, is observed, as in the Albères Massif and some North-Pyrenean Massifs.
The MII metamorphism as a syn-tectonic plutono-metamorphic event
Based on the observation of the deep crust outcropping in the North Pyrenean massifs, Vielzeuf [in Guitard et al., 1996] concluded that emplacement of mafic melts in the Carboniferous lower crust was responsible for the MII metamorphism. At the beginning of the process, a regional thermal anomaly is superimposed to the middle crust (MIIs-MIIa), directly reflecting the emplacement of mafic sills in the underlying lower crust (fig. 4A4). Heat is transferred conductively and, most likely, advected by the aqueous-carbonic fluids issued from the devolatilising lower crust (fluid inclusion data). Heat advection by melts characterised the end of the MII event, with development of more or less local thermal anomalies: still “regional” (MIIbγ) as in the Garonne dome, or directly liked to sheet-like granite intrusions (MIIγ) as at the bottom of the Mont-Louis pluton (fig. 4B4) or at the contact of the Canigou granite (fig. 4C4).
The late MIII Barrovian metamorphic event
The MIII event is mainly characterised in the eastern massifs (Albères, Cap de Creus), where a retrogressive kyanite (so-called “hysterogenic” kyanite) is overprinting high-grade assemblages. Although poorly expressed, MIII minerals in these massifs define two zones, with an external chloritoid zone and an internal kyanite-staurolite zone. A MIII chloritoid zone (sillimanite → chloritoid) is also observed in the core of the Canigou dome. Under the kyanite-staurolite equilibrium hypothesis, the peak MIII P-T conditions in the eastern massifs are estimated at 5 kbar and 575oC, that would imply a pressure increase of 1 to 1.5 kbar (4–6 km deepening) starting from the end of MII, associated with a severe temperature decrease of 150oC. Such an overpressure cannot be due to the D2d dextral-inverse mylonites. However, a fluid inclusion study [Kister et al., 2003] demonstrated that the rocks of the Villefranche syncline did register a pressure increase at the D2b stage, i.e., experienced effective downwards displacement during the syncline formation, and it may be estimated that, in the core of the syncline, a depth increase of 7–8 km could have been attained. Now, in the Cap de Creus massif, the highest MIII grade is observed in the core of the D2b Birba syncline, analogous to the Villefranche syncline. Thus, D2b deepening in the syncline cores may have contributed to the pressure increase. An additional increase may have been provided by sedimentary accumulation in an overlying (and now eroded) syn-orogenic basin (fig. 55). While such a process may explain the development of MIII associations in the D2b synclines, it remains to explain its appearance in the anticlines (Albères, Canigou). However, in the same fluid inclusion study referred to just above [Kister et al., 2003], it is demonstrated that, post-dating D2c and the late pluton emplacement, the studied area suffered a severe isobaric temperature drop, allowing the appearance of chloritoid in the Canigou core (fig. 55). A similar explanation may hold for the Albères massif, if it is accepted there that late kyanite and staurolite were not in equilibrium: starting from the peak MII conditions (c. 4 kbar and 650o–700oC), a strong isobaric cooling would have allowed the successive appearance of staurolite and kyanite.
Discussion and conclusion
The youngest pre-orogenic flyschs are dated (in the Axial Zone) from the Namurian-Westphalian boundary (315±5 Ma), thus setting a minimal age for D1-MI. On the other hand, in the northern Pyrenean Agly massif, the Ansignan hypersthene-granite, which is coeval with MII, is dated at around 315-305 Ma, and the associated norites, likely testifying for the mafic magmatism at the origin of the heat flux responsible for MII, are themselves dated at c. 315 Ma. Finally, the large syn-D2c (post-MII) granite plutons are all dated at 307±3 Ma (i.e., close to the Westphalian-Stephanian boundary). Taken together (with the possibility of a slight diachronism between the North Pyrenean massifs and the Axial Zone, and, within the Axial Zone, between east and west), these data indicate that the MI-MII transition and the whole D2a–c/MII development took place in a very restricted time interval (c. 10 Ma), in Westphalian to Stephanian times.
Crustal rheology and orogenic development
At the end of the Namurian crustal subduction (D1-MI), the Pyrenean crust, that had been thickened with at least a doubling of the upper crust thickness, had begun to experience uplift and erosion. This exhumation process rapidly changed from retrograde to prograde (MIIs-MIIa) during the D2a (MII) syn-convergence extensional phase.
The D2a sub-event was marked by the development of three interrelated processes: (i) isotherm upwelling, regional stratiform MII metamorphism and partial melting in the middle crust, as a result from the intrusion, in the lower crust, of mafic magmas of mantellic derivation; (ii) thinning of the thickened crust; (iii) first arrival of granite plutons in the middle crust. It is thought, according to Vielzeuf [inGuitard et al., 1996], that these processes were initiated by a lithospheric delamination process.
At the end of D2a, the crustal rheology had been modified, with a partially melted middle crust that received granitic melts issued from the melting of the lower crust. This highly ductile middle crust was sandwiched between a thick (≥ 10 km) rigid upper crust and a less ductile granulitised hot lower crust (800o–900oC), thus allowing the progressive decoupling of the upper and lower crust from D2a to D2c. The buckling of the upper crust, with formation of the large upright D2b folds, became therefore possible, forcing the injection of deep anatectic melts in the anticline cores (a probable explanation of the MIIbγ thermal culmination), and creating, in the deepened syncline cores, the strong pressure increase that favoured MIII inception.
However, the MII isogrades are frozen in their folded position, indicating that cooling of the belt had indeed begun since at least the end of the D2b phase. The cooling was sufficiently rapid to be expressed in the Axial Zone by a sub-isobaric temperature decrease, at the origin of the MIII Barrovian and retrograde event, coeval with the late D2c and D2d phases. In the North Pyrenean Massifs, where the D2d phase was extensive, the retrograde MIII event could not be expressed, due to both decompression and thermal effects of the extension.
A summary of this complex evolution is given in figure 66. Finally, the interrelated D2 and MII events appear as the record, in the middle-upper crust, of a very short, but very intense heating event that strongly modified the rheologic behaviour of the crust inherited from the D1 crustal subduction and allowed a transitory decoupling of the upper and lower crust. The isobaric MIII event records an exceptionally rapid return to the “normal” thermal and rheologic structures of the crust.
The rapidly changing tectonic and thermal conditions that characterise the Variscan Pyrenees during the D2 event may be understood if the position of the Pyrenees within the southern branch of the West European Variscan belt is considered (fig. 77).