© C. Passos do Carmo et al., Published by EDP Sciences 2025

1 Introduction

Orogenic processes, and in particular partial melting of the deep crust, extraction of granitic magmas therefrom and their transfer to higher crustal levels, play a crucial role in the differentiation of the continental crust (Brown, 2001; Vanderhaeghe and Teyssier, 2001; Vanderhaeghe, 2009; Hawkesworth et al., 2010; Sawyer et al., 2011; Cawood et al., 2013; Moyen et al., 2017). In turn, these processes control the vertical distribution of the heat-producing elements K, U and Th and thus influence the thermal state of the crust over tens of million years (England and Thompson 1984; Sandiford et al., 2002).

The nature of the lower crust is ill–defined relative to the one of the upper crust (Rudnick, 1995; Hacker et al., 2011). Its architecture can nevertheless be deciphered from the structural, petrological, geochemical and geochronological analysis of exhumed metamorphic and plutonic rocks that are representative of crustal roots.

The Variscan Belt of Western Europe formed during the convergence between Gondwana and Laurussia in the late Paleozoic and resulted in the assembly of the Pangea supercontinent (Matte, 2001). In most areas of the Variscan Belt, tectonic accretion and continental collision were responsible for crustal thickening followed by thermal relaxation and/or lithospheric thinning, which triggered widespread partial melting of the orogenic root (Burg and Vanderhaeghe, 1993; Henk et al., 2000; Vanderhaeghe and Teyssier, 2001). This led to a voluminous magmatic event with contribution from both crustal and mantle sources (Finger et al., 1997; Janoušek et al., 2000; Laurent et al., 2017; Moyen et al., 2017; Jacob et al., 2021).

The French Massif Central (FMC) is part of the internal zone of the Variscan Belt (referred to as the Moldanubian Domain; (Ballèvre et al., 2014; Lardeaux et al., 2014) and records different stages of the orogenic evolution (from subduction to collision; (Matte 1986; Ledru et al., 1989; Lardeaux et al., 2001; Faure et al., 2009). The exposure in the FMC mainly consists of metasedimentary and meta-igneous-metasedimentary rocks affected by metamorphism from greenschist to upper amphibolite or even granulite facies (up to partial melting conditions), and abundant granitic plutons. Metamorphic rocks were mainly exhumed to the surface during the Carboniferous and thus correspond to the former upper and mid–Crust of the Variscan basement of Western Europe. Direct information on mid to lower Variscan crustal levels is accessible in parts of the Variscan Belt such as the Ivrea-Verbano Zone (Brodie et al., 1982; Zingg et al., 1990; Demarchi et al., 1998; Barboza et al., 1999; Handy et al., 1999) and the Pyrenees (Vielzeuf 1996; Olivier et al., 2008; Guille et al., 2019; Vielzeuf et al., 2021).

However, in the case of the FMC, information on the lower crust is accessible only through xenoliths sampled during Cenozoic volcansism, which are overall rare, unevenly distributed across the FMC (Leyreloup et al., 1977; Dostal et al., 1980; Downes and Leyreloup 1986; Leyreloup 1992; Costa and Rey 1995; Laurent et al., 2023), and do not allow to investigate the relationships between exposed lithological units and lower crustal rocks. As a result, the deep architecture of the continental crust in the FMC is still debated (see Vanderhaeghe et al., 2020 for a review).

A nappe stack comprised of an Upper Gneiss Unit (UGU), a Lower Gneiss Unit (LGU), and a Para-Autochthonous Unit (PAU) has been recognised in the Eastern part of the FMC (Burg and Matte, 1978; Ledru et al., 1989) and has been refined in the Western part with the addition of an intermediate allochthon (Girardeau et al., 1986; Berger et al., 2010). The reconstruction of this nappe stack has been based on the recognition of mylonitic contacts and of inverted metamorphic gradients (Burg et al., 1984; Burg et al., 1989). Several propositions have also been made regarding the nature of the crust beneath these nappes and the uncertainty to that respect is expressed in the “unknown Proterozoic basement” invoked by Faure et al., (2009). In the first crustal-scale model, (Matte 1986) proposed that the thrusts separating the nappes are rooted in the Moho, eluding the presence of migmatites exposed at the lowest structural level. In contrast, other authors proposed that migmatites coring crustal-scale domes such as the Velay dome, represent a partially molten crust that was exhumed during gravitational collapse of the orogenic crust (Gardien et al., 1990; Burg and Vanderhaeghe, 1993; Costa and Rey 1995; Ledru et al., 2001; Bouilhol et al., 2006; Gardien et al., 2011). The comparison of data from the FMC’s xenoliths with exposed sections of the Variscan crust in the Pyrenees and in the Alps served to propose that the lower crust beneath the migmatites is composed of a combination of felsic and mafic granulites that might correspond to a residue left after extraction of magmas from the partially molten orogenic root, associated with some mantle–Derived magmas and/or pre-Variscan mafic igneous rocks (Dostal et al., 1980; Vielzeuf et al., 1990; Costa and Rey, 1995; Vanderhaeghe, 2009; Laurent et al., 2023).

In the Pontgibaud region, the attribution of tectono-metamorphic units and the relationship between metamorphic and plutonic rocks are uncertain. This is exemplified by discrepancies, at the scale of the 1/50 000 scale geological map of Pontgibaud (Hottin et al., 1989) and the 1M geological map of France (Chantraine et al., 2003), in the attribution of the migmatitic gneisses exposed in the northern part of the studied area. This is of particular interest since the area has been identified as a potential target for deep geothermal energy production (Duwiquet et al., 2019; Duwiquet et al., 2021). Indeed, the potential for deep geothermal requires reconstructing the geothermal gradient as precisely as possible, which relies in part on the vertical distribution of heat-producing elements and thus on a model for the crustal-scale architecture. In order to fill this knowledge gap and improve the assessment of the geothermal potential of the Variscan crust of the FMC, the goal of this study is to characterise the Variscan geological record (structure, P–T–t–d -–Dhistory, magmatic transfers) of the Pontgibaud area using structural analysis, conventional petrography, phase equilibria modelling, bulk-rock geochemistry and U–Pb geochronology and isotopic analyses on accessory minerals.

2 Geological

2.1 Geology of the FMC

The FMC is part of the late Paleozoic Variscan orogenic belt that resulted of convergence between the supercontinents Gondwana and Laurussia, as well as the intervening crustal blocks Avalonia and Armorica, to form the Pangea supercontinent (Matte 2001). The FMC represents the inverted northern Gondwana margin (Lardeaux et al., 2014; Chelle-Michou et al., 2017; Couzinié et al., 2019). It was originally described as a stack of litho-tectonic nappes showing an inverted metamorphic gradient (Burg et al., 1984). Hereafter, we only describe the three main units, relevant for the investigated area. These units are from top to bottom and roughly from North to South (Burg and Matte, 1978; Ledru et al., 1989; Santallier et al., 1994; Faure et al., 2008; Lardeaux et al., 2014):

• The Upper Gneiss Unit (UGU) is composed of migmatitic gneiss and metasedimentary rocks that underwent Devonian eclogitic–Granulitic HP –HT metamorphism followed by early Carboniferous (360–350 Ma) HT retrogression into amphibolite facies (Gardien et al., 2011; Benmammar et al., 2020). This unit contains the so–Called Leptyno–Amphibolite Complex (LAC) that is interpreted to represent bimodal magmatic rocks with relicts of HP to UHP (eclogite–Facies) Silurian to Devonian metamorphism (Pin and Lancelot, 1982; Santallier et al., 1988; Gardien et al., 1990; Santallier et al., 1994; Lardeaux et al., 2001);

• The Lower Gneiss Unit (LGU) consists of metasedimentary and metaigneous rocks affected by lower Carboniferous amphibolite–Facies metamorphism characteristic of MP–MT conditions with a pressure peak at 350–340 Ma, followed by upper Carboniferous (330–300 Ma) HT metamorphism (Schulz et al., 1996; Mougeot et al., 1997; Schulz 2009; Schulz 2014; Do Couto et al., 2016; Chelle-Michou et al., 2017);

• The Para-autochthonous Unit (PAU) is made of metasedimentary and rare metaigneous rocks of usually greenschist- to lower amphibolite–Facies metamorphism that are overthrusted by the LGU and UGU units.

Both the UGU and LGU largely consist of Neoproterozoic to early Paleozoic (Cambrian–Ordovician) metasedimentary rocks and orthogneisses. Deposition of the precursor sedimentary sequences likely took place from the Ediacaran to the early Paleozoic (Melleton et al., 2010; Couzinié et al., 2019). The Lu–Hf signature of detrital zircons indicates that these sequences were fed by erosion of both the cratonic basement of northern Gondwana and juvenile Neoproterozoic terranes including the Arabian–Nubian shield and/or the Cadomian arc (Chelle-Michou et al., 2017; Couzinié et al., 2019; Avigad et al., 2022). Metasedimentary units are associated with former granites and minor mafic rocks of Ediacaran–Cambrian to Ordovician age, transformed into orthogneisses (Chelle-Michou et al., 2017; Couzinié et al., 2017; Lardeaux et al., 2014; Melleton et al., 2010) or alternate with the Leptyno-Amphibolite Complex. These meta-igneous rocks are described as products of two main igneous events at ca. 540 and ca. 480 Ma, respectively related to (i) inversion of an Ediacaran back-arc basin formed between the Cadomian arc and the northern Gondwana margin (Couzinié et al., 2017; 2019); and (ii) extension related to the opening of the Medio–Europeanpaleo-ocean and drift of the Armorica block from the Gondwana margin (Pin and Lancelot, 1982; Faure et al., 2009; Lardeaux et al., 2014; Chelle-Michou et al., 2017)). Markers of ocean closure by subduction are eclogites, blueschists and calc-alkaline plutons and rare volcanics (Berger et al., 2024). In the French Massif central, HP pressure metamorphism has been first documented to have occurred between Silurian and Early Devonian ((Pin and Lancelot, 1982); see also Lardeaux, 2014 and references therein) but recent dating in southern FMC yielded Upper Devonian to Lowermost Carboniferous ages for eclogitic metamorphism (Lotout et al., 2018; Lotout et al., 2020). This is in agreement with the magmatic record, in particular the various expressions of calc-alkaline active margin magmatism that also occurred during the Upper Devonian, especially in the Limousin area in the western FMC (Bernard–Griffiths and Cornichet, 1985; Shaw et al., 1993; Bertrand et al., 2001; Pin and Paquette, 2002).

In the case of the FMC the production of large amounts of granitic magma is attested by the widespread exposition of migmatites and granitic bodies showing zircon crystallization ages from approximately 360 Ma to 290 Ma (see compilation in Vanderhaeghe et al., 2020, and references therein). The magmatic activity in this interval is characterised by melting of dominantly crustal sources (represented by pre --Variscan, both metasedimentary and metaigneous rocks), leading to emplacement of dominantly peraluminous, S-type granites (see Moyen et al., 2017) and references therein). In the eastern part of the FMC (i.e., east of the Sillon Houiller strike-slip fault; Fig. 1), crustal melting resulted in the development of the large (ca. 100 km in diameter) Velay granite-migmatite dome (Burg and Vanderhaeghe, 1993; Barbey et al., 1999, 2015; Ledru et al., 2001) and is accompanied by significant contribution of lithospheric mantle sources, as shown by the coeval emplacement of peraluminous granites and “vaugnerites” (Mg–K rich lamprophyres) and associated granitoids from ca. 350 to 300 Ma (von Raumer et al., 2014; Couzinié et al., 2016; Laurent et al., 2017; Moyen et al., 2017; Werle et al., 2023).

Fig. 1 (a) General context of the French Massif Central. (b) Sketch map of the Variscan basement of the Pontgibaud region showing the spatial distribution of LGU, UGU and PAU units, and localities where these nappes were defined (Sioule Series, Artense Dome and Haut Allier). Modified after Chantraine et al., (2003). (c) Geological map of the Pontgibaud region. The dashed black line represents the UGU/LGU contact proposed by Chantraine et al. (2003). Note that paragneiss are either attributed to LGU (north-west of Gelles granite) or UGU (south-east of Gelles granite). Modified after Hottin et al., (1989).

2.2 Pontgibaud geology

The Pontgibaud region is located in the Eastern French Massif Central (in relation to the Sillon Houiller fault zone), between the Sioule, Artense and Haut-Allier regions, which are type localities where the UGU/LGU relationships have been previously established (Fig.1B–C). The Pontgibaud area is bounded by the Sillon Houiller sinistral fault zone to the West and the Aigueperse-St-Sauves fault zone to the East. The geological map of the studied area was established based on the work of (Fernandez, 1969; Tempier, 1969; Tempier, 1974; Fernandez and Tempier, 1977; Négroni, 1981), and the geological map of Hottin et al (1989). The main metamorphic lithological units are, from S to N (i) aluminous micaschists comprised of alternating quartz–feldspar rich metyagrauwacke and aluminosilicate-rich metapelite; (ii) biotite + sillimanite ± muscovite paragneisses, locally containing garnet; (iii) minor and poorly exposed, quartzo–feldspathic orthogneisses; (iv) metatexitic paragneiss, represented by biotite ± cordierite ± garnet ± sillimanite; and (iv) cordierite diatexite.

The structural record of the studied area has been documented (Fernandez, 1969; Tempier, 1974; Fernandez and Tempier, 1977; Négroni 1981). These data are synthesised in the 1/50 000 scale geological map of Pontgibaud (Hottin et al., 1989). Following these authors and our own field observations, micaschists and paragneiss display a shallow–dipping foliation, associated with rootles isoclinal folds that are correlated with kilometre scale isoclinal folds with NE–SW striking axes plunging dominantly to the SW. This foliation is affected by three generations of open upright folds with, in chronological order, E–W, NW–SE, and N-S striking axial planes, the latter being preferentially expressed to the NW of the Gelles pluton. In the migmatites, the foliation is steeply dipping to vertical and its direction displays a chaotic pattern on the map scale.

Metamorphic rocks are intruded by biotite ± cordierite- and muscovite-bearing Variscan granitic plutons, namely the Claveix fine–Grained granite, forming an intrusive body feeding a ca. 6 km-long N–S trending porphyritic dyke; and the Gelles porphyritic granite, forming a ca. 12 x 6 km pluton associated with volumetrically minor, satellite granite porphyry dykes. Based on field relationships, the Gelles granite is interpreted as intrusive in the Claveix granite (Fernandez and Tempier, 1977; Hottin et al., 1989). The detailed analysis of the Gelles pluton served to develop the structural analysis of granitoids based on the preferred orientation of feldspar megacrysts (Fernandez, 1969; Fernandez and Tempier, 1977). The magmatic foliation, depicted by the preferred orientation of feldspar megacrysts and by microgranular enclaves, shows a roughly concentric distribution, parallel to the contacts of the oval-shaped pluton. The northern and southern margins of the plutons are steeply dipping. The western contact of the pluton with the metamorphic hosts is gently dipping toward the West, whereas the eastern contact is steeply dipping toward the East. The lineation marked by the preferred orientation of the long axis of feldspar megacrysts is shallowly dipping towards the West in the western part of the pluton and is steeply dipping towards the East in the eastern part.

In early works (Fernandez, 1969; Tempier, 1974; Fernandez and Tempier, 1977; Négroni, 1981) the micaschists and paragneiss of the southern part of the studied area were represented structurally on top of the sillimanite- and cordierite-bearing migmatites, and this transition was interpreted as a metamorphic gradient. This proposition has been subsequently challenged by tentative correlations of the different litho-tectonic units to the nappe structure of the French Massif Central, but the attribution of these units to either UGU or LGU is still unclear. In the 1/50 000e map, the attribution of the metamorphic rocks to the LGU or UGU has essentially been made based on the presence of amphibolite lenses in the migmatitic gneisses; and correlations with rocks exposed in the Artense dome further SW (Hottin et al., 1989). Accordingly, the micaschists are attributed to the LGU, the migmatites to the UGU and the paragneisses to a third metamorphic unit referred to as “Undifferentiated Gneiss Unit” (Hottin et al., 1989). In contrast, on the geological map of France at the 1:1 000 000/ scale (Chantraine et al., 2003), the paragneisses are attributed to either the LGU to the west of the Gelles Granite, or to the UGU to the southeast (Chantraine et al., 2003) (see dashed black line in Fig.1C).

3 Materials and Methods

Following field studies and preliminary petrographic investigation on a set of 30 samples, eight of them were selected for further investigation, as representatives of the main identified lithological units in the Pontgibaud area: one micaschist (PG21-04), two paragneisses (PG20-01 and PG21-28), two metatexitic paragneisses (PG20-02 and PG21-20, sampled at the same locality), one diatexite (PG20-4), one sample of porphyritic-texture Claveix granite along the N-S trending dyke (PG20-05), and one sample of Gelles granite (PG20-03). Sample locations are reported in Figure 1.

Sample preparation was performed at the Géosciences Environnement Toulouse - Observatoire Midi-Pyrénées (GET-OMP) laboratory, France. Rock samples were sawed to prepare 30 µm-thick petrographic thin sections and, for some of them,ca. 50 µm-thick sections for i-N-S itu monazite geochronology by LA-ICP-MS. The samples were reduced and subsequently crushed using a steel jaw crusher and a disc mill (all samples except PG21-28 paragneiss), a fraction of each sample was further crushed to a sub-µm fraction in an agate swing mill for whole rock major and trace elements analysis. Major elements compositions are shown in the pseudosection equilibrium diagrams (Figs. 12–13) ; complete data and ALS analytical procedure are reported in supplementary material and in Table 1.

3.1 Pseudosection calculation

The sub-µm rock powders were sent to ALS Global Geochemistry for whole-rock analysis, using conventional ICP-OES (for major elements) and ICP-MS (for trace elements) analysis following alkali fusion and wet acid digestion of the samples. The results were used as representative of bulk sample compositions for thermodynamic modelling.

Pseudosection calculations combined with thin section observation and major–Element analyses of main mineral phases were performed in order to obtain the metamorphic P–T conditions recorded by the samples. Major element analysis on feldspar, muscovite, biotite and garnet were performed using an CAMECA SXFive electron microprobe at the Centre de Microcaractérisation Raimond Castaing in Toulouse, France, in order compare the mineral compositions with calculated mineral isopleths. Analytical conditions were an acceleration voltage of 15 kV, a beam current of 10 nA for the micas and 20 nA for garnet and feldspar, and, a spot size of 2 µm.

Phase diagrams and isopleths were calculated using Perple_X software (Connolly 2009) with thermodynamic databases and mineral compositions from Holland & Powell, (2011), using the 9–Components system SiO₂–Al₂O₃––FeO–MnO–MgO–Na₂O––CaO–K₂O–TiO₂. The water content was set at saturation for modelling. Ferric iron is not included in the model due to the lack of major phases containing Fe+³⁺. Solution models used for modelling are: muscovite (Coggon and Holland, 2002; Auzanneau et al., 2009), biotite (Tajčmanová et al., 2009), feldspar (Newton et al., 1980; Holland and Powell, 2003), garnet (White et al., 2000), chlorite (Holland et al., 1998), sanidine (Waldbaum and Thompson, 1968), ilmenite (White et al., 2000), cordierite (ideal), staurolite (Holland et al., 1998), orthopyroxene (Holland and Powell, 1996), and melt (White et al., 2001). The whole-rock compositions and representative mineral compositions are reported in the supplementary material and in Tables 1 and 2.

3.2 U-(Th-)–Pb geochronology

3.2.1 Zircon

Zircon separation was performed from the <500 µm sample fractions using standard heavy mineral concentration techniques (Wilfley shaking table, manual panning, Frantz magnetic separator and heavy liquids when necessary). The concentrated zircons were then hand-picked using a binocular microscope, set in 1-inch epoxy resin mounts and polished to a sub–Equatorial section. Attention was given to selecting zircons of different shapes and sizes in order to obtain representative zircons for each sample, especially for the populations of detrital grains in metasedimentary rocks.

The internal structures of the zircons were characterised using the Tescan rainbow cathodoluminescence (CL) detector attached to the Tescan Vega4 Scanning Electron Microscope (SEM) at GET. Conditions were an acceleration voltage of 10 kV and beam current of 3 nA.

Simultaneous analyses of U–Pb isotopes in zircons were conducted by laser ablation– inductively coupled plasma– mass spectrometry (LA-ICP-MS) at the Service ICP-MS of Observatoire Midi-Pyrénées (OMP-UAR831). The analyses were carried using a NWRfemto (Elemental Scientific Instruments) solid-state femtosecond laser ablation system set to UV mode (257 nm wavelength) coupled with an Element XR (ThermoScientific) sector–Field ICP-MS. Measurements were performed with a laser spot diameter of 20 µm, a repetition rate of 6 Hz and an energy density of ca. 2.5 J/cm². Ablation was performed in the built-in, dual-volume, fast-washout ablation cell (<1 cm³ effective volume) flushed with He carrier gas (ca. 0.6 L/min) to which was admixed Ar make-up gas (ca. 0.88 L/min) downstream of the ablation cell before introduction in the plasma. A typical run consisted of 30 seconds of background signal acquisition, followed by 30 seconds of ablation and 10 seconds of washout. The list of acquired masses, corresponding dwell times, details about the instrument optimization and data acquisition procedures are provided in the Supplementary Table X.

The resulting intensities were processed offline with the Igor Pro Iolite v2.5 software (Hellstrom et al., 2008), using the VizualAge data reduction scheme (Petrus and Kamber, 2012) for U–Pb dating. The GJ1 zircon reference material (using isotope ratios from (Horstwood et al., 2016) was used as primary standard for correction of instrumental drift, mass bias and laser-induced U/Pb fractionation (following (Paton et al., 2010)) through conventional standard-sample bracketing (Hellstrom et al., 2008). Secondary zircon standards Plešovice (Sláma et al., 2008), Mud Tank (Horstwood et al., 2016) and AUSZ7-1 (Kennedy et al., 2014) were analyzed as unknowns to confirm the accuracy of the results. The obtained U–Pb ages for all secondary standards (see Supplementary Table 1) are accurate within the total bulk reproducibility of the method, which is 2 per cent relative (2σ) (Horstwood et al., 2016). The data from unknowns and reference materials are provided in Supplementary Table X. Results were plotted in Concordia diagrams and Concordia or Discordia ages were calculated whenever possible, using the IsoplotR online package (Vermeesch, 2018). For age calculations, data treatment was performed as proposed by (Spencer et al., 2016) through outlier rejection as to obtain a suitable MSWD based on the constraints from (Wendt and Carl, 1991). Systematic uncertainties were propagated in the uncertainties of pooled ages (i.e. calculated from the average of several individual data points) as recommended by Horstwood et al., (2016).

3.2.2 Monazite

In situ U–Pb dating of monazite was performed directly in thin sections of paragneiss PG20-01, metatexitic paragneiss PG20-02; and diatexite PG20-04. Monazites in thin sections were imaged using the same SEM instrument as mentioned above in backscattered electron (BSE) mode and with X-ray maps.

For U–Pb isotopic analyses, the same LA-ICP-MS system as mentioned for zircon U–Pb dating was used, with the same analytical conditions except for a 7 µm spot size and timing parameters of 20 seconds background signal acquisition, 20 seconds of ablation and 10 seconds of washout delay. Data processing was performed using the same software and procedures as for zircon U–Pb dating, using the Moacyr monazite reference material (using isotope ratios from Gasquet et al., (2010) as primary standard and Manangoutry (Paquette and Tiepolo, 2007) as secondary standard. Details about the analytical conditions and results on the secondary reference material are provided in the Supplementary Table Y.

3.3 Zircon Lu-Hf isotopes

Zircon Lu–Hf isotopic analysis were performed by laser ablation– multi–Collection– inductively coupled plasma– mass spectrometry (LA-MC-ICP-MS) at OMP. A high-resolution MC-ICP-MS Neptune Plus (Thermo Scientific) was coupled to the same femtosecond laser ablation system as used for U–Pb geochronology. Analyses were performed on some selected zircons, in the same spot locations as previously dated with the U–Pb technique or within the same growth zone, and attention was given to select zircons with >90 per cent of concordance and representative of all groups of ages. The ablation, aerosol transport and timing parameters were the same as for zircon U–Pb measurements, except for a 50 µm spot diameter and repetition rate of 7 Hz. More details about the analytical parameters are reported in Supplementary Table X.

The data were processed offline with the Igor Pro-based Iolite v2.5 software, using an in-house data reduction scheme. Instrumental mass bias for Yb and Hf isotopes were corrected to the natural abundance ratios¹⁷³ Yb/¹⁷¹Yb and¹⁷⁹ Hf/¹⁷⁷Hf of Chu et al., (2002). Mass bias for Lu isotopes was assumed identical to that for Yb isotopes. The isobaric interferences of¹⁷⁶ Yb and¹⁷⁶ Lu on¹⁷⁶ Hf were corrected using the natural abundance ratios of¹⁷⁶ Yb/¹⁷³Yb = 0.79502 and¹⁷⁶ Lu/¹⁷⁵Lu = 0.02656. The Yb- and Lu-poor zircon reference material Mud Tank (Woodhead and Hergt, 2005) was used as primary reference material to correct for instrumental drift and any offset on the interference–Corrected¹⁷⁶ Hf/¹⁷⁷Hf ratios (generally not exceeding 1 to 2 εHf-units). Accuracy and external reproducibility of the method were controlled by repeated analyses of reference zircon standards 91500 (Wiedenbeck et al., 1995), GJ-1 (Morel et al., 2008), Maniitsoq (Marsh et al., 2019), Plešovice (Slamá et al., 2008) and Temora (Woodhead and Hergt, 2005). The¹⁷⁶ Hf/¹⁷⁷Hf ratios obtained on all reference materials are within uncertainties identical to the recommended values (see Supplementary Table X). Calculations of initial¹⁷⁶ Hf/¹⁷⁷Hf ratios and εHf were performed using a¹⁷⁶ Lu decay constant of 1.865 ×10^–11 (Scherer et al., 2001) and¹⁷⁶ Hf/¹⁷⁷Hf and¹⁷⁶ Lu/¹⁷⁷Hf values for the chondritic uniform reservoir (CHUR) recommended by (Bouvier et al., 2008).

4 Results

4.1 Field Observations

Our field observations confirm that the metamorphic units in the Pontgibaud region, as described in Tempier (1969; 1974) and Hottin et al., (1989), are dominated by metasedimentary rocks that correspond to alternation of quartz–Feldspar rich and aluminosilicate rich micaschists that progressively grade into paragneiss migmatites towards the North (Fig. 2). In the following, we present outcrop-scale features of these rocks and describe their regional-scale structural relationships. In the map of figure 2, the structural measurements correspond to the ones reported by Tempier, (1974) and Hottin, (1989) for the metamorphic rocks and by Négroni (1981) and Fernandez, (1969) for the Gelles pluton, complemented with our own data. Note that in previously published maps, the authors did not distinguish the different generations of structures, and the structural symbols correspond to dominant structure identified at the outcrop scale. The representation on the cross sections of the different generations of structures relies on our outcrop scale observations and on the correlations with previously published structural measurements. A detailed description of the mineral assemblage and texture of the studied samples is provided in the next section.

The micaschists and paragneisses, which enclose rare, km-scale lenses of fine–Grained orthogneiss with a granitic composition, are located on a topographic high, to the South of the Gelles pluton. Outcrops are generally variably weathered road cuts and small quarries. Metasediments are composed essentially of quartz, plagioclase, and biotite, differing in the presence of andalusite and higher content of muscovite in the micaschist, whereas the paragneiss contains sillimanite with rare garnet. At the outcrop scale, the alternation of coarse–Grained, quartz-rich layers and fine–Grained, mica-rich layers is interpreted to represent the former sedimentary bedding (Fig. 3B). This bedding is transposed into a major sub-horizontal S/_0/n foliation, also marked by the preferred orientation of micas, which are in an axial planar position of rootless isoclinal folds identified in the field (Figs. 2, 3B) and described by Tempier (1974). The isoclinal folds display a dominant NE–SW trending axis and associated with the development of a discrete axial planar schistosity S+_n+1 (Fig. 3A–B). This main foliation is trending roughly parallel to lithological contacts that delineate an arch-shape from a preferred NE–SW to E–W orientation to the East of the Gelles pluton and scattered around a - N-S direction in the paragneiss domain to the West of the pluton (Fig. 1). The main foliation is also transposed into NW-SE-trending subvertical upright folds in the western part of the studied area, close to the Sillon Houiller and into an E–W trending subvertical schistosity in an axial planar position of upright folds, more markedly toward the transition with the migmatites.

The micaschists and paragneisses to the south of the Gelles pluton, grade into migmatites that form the bulk of the exposure in the northern part of the studied area (Fig. 1). The transition from micaschists to migmatites occurs progressively over a couple kilometers and is marked, from South to North by the succession of metatexites and diatexites with a gradual increase of leucosome proportions. Migmatite outcrops are generally relatively small, discontinuous and variably weathered road cuts, such that the relationships with other lithologies are difficult to establish. The metatexitic paragneiss, which only corresponds to a kilometre-thick zone between paragneiss and diatexite, is characterised by the alternation of centimetre to decimetre thick leucosome and mesosome layers, which delineate a synmigmatitic foliation. Leucosome veins concordant to the foliation are in textural continuity with veins discordant to the foliation (Fig. 3C–D), pointing to syntectonic melt segregation and magma migration at the grain to outcrop scales. Diatexites are dominated by a granite with a heterogeneous texture and containing abundant cordierite as well as numerous enclaves of metatexitic paragneiss (Fig. 3–E).

Along the southern margin of the migmatites, the synmigmatitic foliation of the metatexitic paragneiss displays roughly the same strike as the S/_0/n foliation of the micaschists and paragneisses but is in general steeper (Fig. 2). In contrast, the orientation of the synmigmatitic foliation of the diatexites, defined by dismembered schlieren and restitic enclaves in the granitic matrix, is more heterogeneous and does not display a preferred orientation. The synmigmatitic foliation in both, metatexite and diatexite, is generally concordant to lithological contacts, possibly delineating in kilometre scale subdomes within the migmatites (Fig. 3C).

The metamorphic rocks are intruded by the Claveix and Gelles granitic plutons, which crosscut the roughly E–W trending contact between the micaschist-paragneiss series and the migmatites (Fig. 1 and 2). The main body of the Claveix granite (Hottin et al., 1989) consists of homogeneous, equigranular, two-mica- and cordierite-bearing granite, which is poorly exposed. It grades to the north into a broad, N-S directed dyke associated with porphyry-like granite bodies with sub-volcanic textures (Figs. 2 and 3F). Locally, the dyke clearly crosscuts the foliation of migmatites and shows a chilled margin with quartz and feldspar phenocrysts enclosed in a very fine–Grained groundmass (Fig. 3F). The Gelles pluton forms a low-topography flat zone in which isolated exposures are generally strongly weathered, whereas fresh outcrops are mainly restricted to the pluton’s margins, especially to the east along the escarpment of the Sioule valley. Granite forming the core of the pluton is a coarse–Grained, porphyritic biotite–Cordierite- and minor muscovite-bearing granite containing K–Feldspar phenocrysts up to 5 cm long (Fig. 3 G-H). Tourmaline may be present as a disseminated mineral (Hottin, 1989) or, together with quartz, as veinlets cross–Cutting the granite. The Gelles pluton is dominated by a shallow–Dipping magmatic fabric underlined by the preferred orientation of K–Feldspars phenocrysts. Along the western contact of the Gelles pluton, this magmatic fabric is shallowly dipping toward the West and displays an E–W trending preferred orientation of the long axis of K–Feldspar phenocrysts. These phenocrysts locally display bookshelf docking and are affected by ductile deformation marked by a C/S fabric consistent with a top to the East sense of shear (Fig. 2 and 3G). Along the eastern contact, the magmatic fabric is dipping to the East and is crosscut by mylonitic to cataclastic shear zones steeply dipping to the East that are running parallel to the N-S trending faults indicated on the map (Fig. 2). The shear direction is indicated by the stretching direction of quartz and feldspar in mylonite and by mechanical striaes in cataclasite. Deflection of the ductile fabric into these shear zones is consistent with a top to the East sense of shear.

Fig. 2 Interpretative cross-sections (locations of the profiles are shown in Fig. 1c) showing relationships between micaschist, paragneiss, migmatites and plutonic rocks, and structural observations.

Fig. 3 Outcrop features of (A) Ms–Bt micaschist and macroscopic sample (inset) affected by crenulation folding with N–S trending axes; the observation plane of the macroscopic sample is marked in white. (B–D) Bt–Sil paragneiss affected by an isoclinal fold (B), and displaying concordant to discordant leucosomes (C, D). (E) Migmatite outcrop displaying foliated metatexite crosscut by Crd-bearing diatexite. (F) Migmatite and Claveix granite discordant contact, inset showing a macroscopic sample of the granite. (G–H) Gelles granite exhibiting (G) C–S fabric and (H) oriented K-feldspar phenocrysts in a vertical plane (pink pencil as scale). Structural measurements are reported in dip direction/dip.

4.2 Petrography

4.2.1 Micaschists and paragneiss

Sample PG21-04 is representative of the micaschists located to the SW of the study area, composed of mostly biotite, muscovite, quartz and plagioclase with an alternation of mica-rich and quartz -rich layers (Fig. 4A). Andalusite is the main index metamorphic mineral and is localised in the mica-rich layers. It is present as sub-idioblastic crystals oriented parallel to the main foliation and containing inclusions of muscovite and biotite (Fig. 4B). Quartz-rich layers present polygonal grain boundaries attesting for grain boundary migration during solid-state recrystallization. Accessory minerals comprise rutile, ilmenite, tourmaline, apatite, zircon and monazite. Mica-rich layers contain centimeter-sized round and fractured porphyroclasts porphyroblasts (possibly cordierite, see below), completely transformed in sericite. These porphyroclasts contain inclusion of white mica and ilmenite that delineate a schistosity continuous to the one of the matrix but they are also wrapped into the foliation, which points to their syntectonic crystallization. Similar pseudomorphosed porphyroclasts have been previously interpreted as cordierite described in this area (Hottin et al., 1989). Biotite contains inclusions of zircon, monazite and rutile. Ilmenite is present as small crystals in the matrix and as sub-idioblastic to xenoblastic crystals enclosed in the porphyroblasts. Some ilmenite inclusions reach a length up to 200 µm and contain quartz inclusions.

Paragneisses PG20-01, PG21-28 and metatexitic paragneiss PG20-02 present a paragenesis composed of plagioclase + biotite ± K–Feldspar ± sillimanite ± garnet together with accessory zircon, monazite and apatite. The main foliation in all samples is marked by the preferred orientation of biotite. A network of concordant to discordant leucosome veins points to syntectonic melt segregation. In sample PG20-01 sillimanite occurs as fibrolite aggregates mostly concordant to the main foliation, usually embedded in a fine–Grained quartz–muscovite matrix (Fig. 4D) suggesting the retrograde metamorphic reaction:

K–Feldspar + sillimanite + biotite → muscovite + quartz

Sillimanite aggregates are either concordant with, folded in, or overprinting the foliation, which suggests that deformation occurred in the stability field of sillimanite. The K–Feldspar and plagioclase crystals are respectively rounded and elongated, and contain numerous quartz, biotite and more rarely muscovite inclusions. K–Feldspar and plagioclase form aggregates that are wrapped into mica-rich layers marking the main foliation forming a symmetric sigma-type shape pointing to syn-kinematic recrystallization. Chlorite is locally replacing biotite, and, along with rutile, is more common in the leucosome. As documented in the micaschists, the sample contains porphyroclasts, presumably cordierite pseudomorphs, totally altered to white mica (Hottin et al., 1989). Leucosome veins contain tabular and skeletal muscovite replacing sericitised idioblastic minerals (presumably cordierite), and rarely relics of sillimanite. Quartz of leucosome veins displays serrated to lobate grain boundaries typical of dynamic recrystallisation by bulging and grain boundary migration (GBM).

Sample PG21-28, at northwest of the studied area, is a fine–Grained garnet-biotite-plagioclase-quartz paragneiss with a lepidoblastic texture defined by the orientation of biotite. Xenoblastic to sub-idioblastic crystals of garnet occur typically in biotite-rich domains, frequently as atoll crystals (Fig. 4E). Garnet crystals, when not as atoll texture, comprise an inclusion-rich core and an inclusion-poor rim. The inclusions consist mostly of quartz and apatite, rarely biotite, muscovite and rutile. Smaller square-shape inclusions, analysed by EDS, typically present C and O peaks pointing either to organic matter or to decrepitated fluid inclusions. Rare fibrolite and mica flakes are interlayered with biotite.

Fig. 4 Photomicrographs of (A–B) micaschist PG21-04: (A) in cross-polarised light (XPL), showing a pseudomorph, presumably after cordierite; (B) andalusite occurrence parallel to the main foliation. (C) Bt–Sil paragneiss PG20-01 in plane-polarised light (PPL), showing replacement of sillimanite by muscovite. (D) Grt–Bt paragneiss in PPL, showing garnets with inclusion-rich cores oriented along foliation. (E) Bt–Sil metatexitic paragneiss PG20-02, exhibiting a pseudomorphed crystal (probably Crd) replaced by sillimanite + muscovite. (F) Crd–bearing diatexite sample PG20-04, showing poikiloblastic texture of K-feldspar enclosing plagioclase crystals. Note white arrows indicating interconnected interstitial quartz, plagioclase with limpid rims, and resorbed plagioclase. (G) Claveix granite showing porphyritic texture with globular, corroded quartz. (H) Gelles granite in XPL, showing coarse-grained matrix texture.

4.2.2 Metatexitic paragneiss

Metatexitic paragneiss sample PG20-02 resembles PG20-01 in mineralogy. They are characterised by a synmigmatitic foliation delineated by the alternation of mesosome and concordant leucosome veins. The mesosoma has a granoblastic texture and is made of tabular crystals of plagioclase, quartz, biotite and sillimanite aggregates. Sillimanite is also present as inclusions in the core of plagioclase crystals. Early muscovite also occurs in this sample as inclusion in biotite or plagioclase. The texture of the metatexitic paragneiss PG20-02 is characterised by rounded porphyroclasts and interstitial quartz films interpreted as a former melt phase wetting the grains. Concordant leucosome veins are composed mainly of quartz associated to plagioclase and K–Feldspar, they are folded and boudinaged. Quartz grains are characterised by polygonal boundaries that attest for high-temperature recrystallization. Pseudomorphosed porphyroclasts are replaced by sericite and fine–Grained acicular crystals of sillimanite (Fig. 4C). These porphyroclasts are wrapped into the main foliation, are surrounded by pressure shadows and contain helicitic inclusions indicating a syntectonic growth (Fig. 4C).

4.2.3 Diatexites

Sample PG20-04 is a diatexite composed of quartz, plagioclase, K–Feldspar, biotite, muscovite, cordierite, and chlorite together with accessory zircon, monazite, apatite and rutile. Its texture is heterogeneous and the granitic matrix contains enclaves of metatexites. At the microscopic scale the matrix is either made of equigranular euhedral plagioclase and K–Feldspar crystals with interstitial quartz, or of poikilitic K–Feldspar containing plagioclase inclusions (Fig. 4F). Polycrystalline domains with quartz present chessboard microstructure and ubiquitous serrated grain boundaries are indicative of solid-state deformation at high temperature. Enclaves of metatexites consist of biotite, sillimanite, plagioclase and quartz and display a metamorphic foliation. A pervasive alteration is marked by transformation of feldspars, cordierite, and sillimanite in micas.

4.3 Plutonic rocks

4.3.1 Claveix pluton

The main body of the Claveix pluton is described as an equigranular, fine-grained cordierite and biotite-bearing granite containing late muscovite. The sample PG20-05, investigated for U–Pb geochronology, was sampled NW of the Gelles pluton, in a body of porphyry-like granite associated with the Claveix granite (Fig. 1) near the contact with migmatites. It is a fine–Grained granite with feldspar crystals up to 2 mm (10 per cent) embedded in a fine–Grained feldspar + quartz darker matrix, defining a sub-volcanic texture. It is composed of plagioclase, K–Feldspar, quartz and minor chloritised biotite (Fig. 4G). Accessory minerals are zircon and apatite. Feldspars are euhedral, whereas quartz is sub–Euhedral to anhedral/skeletal.

4.3.2 Gelles pluton

The predominant facies of the Gelles pluton is a granite exposed in the centre of the studied area straddling the contact between migmatites and metasedimentary rocks. At the mesoscale, it is a leucocratic granite characterised by a porphyritic-textured, defined by K–Feldspar phenocrysts (20 per cent) in a matrix that consists of plagioclase (30 per cent), K–Feldspar (20 per cent), quartz (30 per cent) and biotite (Fig. 4H) with, more rarely, cordierite, muscovite and tourmaline. Plagioclase often display inclusion-rich cores. Accessory minerals are zircon and apatite. Tabular K–Feldspar phenocrysts are up to 3 cm size, often presenting oriented fabric defining a dominantly subhorizontal magmatic fabric.

4.3.3 Chemistry of the Claveix and Gelles granitic plutons

Whole-rock bulk compositions obtained in this study for Claveix and Gelles granites samples along with data available in Hottin et al., (1989) (n = 5 for Claveix granite and n = 7 for Gelles granite), plot in the field of peraluminous granites, ranging in composition from A/CNK (molar Al₂O₃/[CaO+Na₂O+K₂O]) 1.11–1.3 and A/NK (molar Al₂O₃/[Na₂O+K₂O]) 1.4-1.6. Both granites display similar concentrations of SiO₂(ranging from 69–74 wt. per cent), K₂O (4 wt. per cent), Al₂O₃ (15 wt. per cent) and CaO (1 wt. per cent). They slightly differ in terms of FeO_T + MgO contents (3.21 wt. per cent for Claveix and 3.05 wt. per cent for Gelles, see Table 1). In diagrams such as A/CNK vs. A/NK and Na₂O+K₂O vs. SiO₂ the samples from this study and the ones compiled in Hottin et al., (1989) yield values within the Variscan S-type granites (Couzinié et al., 2017; Moyen et al., 2017).

4.4 Mineral Chemistry

The compositional characteristics of the main minerals are presented in Fig. 5 and summarised below.

Fig. 5 Mineral chemistry results from electron microprobe analyses in micaschist (PG21-04), paragneiss (PG20-01, PG20-02, PG21-28) and diatexite (PG20-04) samples. (A) Feldspar ternary diagram. (B) Fe + Mg versus Si contents in muscovite. (C–D) Biotite Si and Ti a.p.f.u. compositions versus Mg# (100 × Mg/[Mg + Fe2+]). (E) Garnet composition shown as molar proportions of end-member compositions (Alm = almandine, Grs = grossular, Prp = pyrope, Sps = spessartine) across a crystal profile.

4.4.1 Plagioclase (Fig. 5A)

All metamorphic samples present plagioclase composition below 20 mol. per cent anorthite content (molar ratio Ca/[Ca+Na]). The composition of plagioclase in metatexitic paragneiss sample PG20-02 (8 mol. per cent An) is intermediate between the ones of plagioclase in paragneiss PG20-01 (12 mol. per cent An) and in diatexite PG20-04 (2 mol. per cent An).

4.4.2 Muscovite (Fig. 5B)

The composition of muscovite in all samples displays between 3.02 and 3.12 cations of Si per formula unit (a.p.f.u., calculated in the basis of 11 Oxygen atoms) and Fe+Mg in a range between 0.08 and 0.14 cations a.p.f.u. The Fe+Mg content of sericite in the porphyroclasts in sample PG20-02 is higher (0.15– 0.18 cations a.p.f.u.) and may reflect the influence of the composition of the pseudomorphosed porphyroclast (presumably cordierite). Muscovite inclusions in plagioclase from paragneiss (PG20-01) present higher values of Si up to 3.2 cations (a.p.f.u.).

4.4.3 Biotite (Fig. 5C–D)

Biotite compositions from micaschist PG21-04 and paragneiss PG21-28 present the highest values of XMg (molar ratio Mg/[Mg+Fe+²⁺]) of 0.48-0.50 and the lowest values of Ti (between 0.10 and 0.12 cation a.p.f.u., calculated in the basis of 11 Oxygen atoms) of all analysed samples. Biotites from diatexite PG20-04 show lower values of XMg (ca. 0.3) and higher values of Ti (between 0.2 and 0.25 cation a.p.f.u.) than those of the other samples. Si content in biotite does not seems to vary and is approximatively 2.5 cations a.p.f.u. for all samples.

4.4.4 Garnet (Fig. 5E)

Garnet crystals from the investigated samples are typically almandine (>57 mol. per cent) in composition. Electronic microprobe profiles of selected crystals show a significant Mn zonation with X_sSp values ranging from 20 mol. per cent in the core to 30 mol. per cent in the rim and less important zonations of XAlm (from 57 to 61 mol. per cent respectively) and X_Pyp (from 10 to 13 mol. per cent respectively). Grossular content never exceeds 3 mol. per cent, rendering GASP estimations not reliable as mentioned by Holdaway (2001).

4.5 U-Pb–Pb dating on zircon

4.5.1 Zircons textures

Cathodoluminescence images of representative zircon grains of each sample are presented in Figure 6 along with spot location of U–Pb and Lu–Hf analysis. Zircons in micaschist (PG21-04), paragneiss (PG20-01), metatexitic paragneiss (PG20-02) and diatexite (PG20-04) samples are typically rounded and vary in size from 50 to 200 µm, with a few grains up to 200 µm in the diatexite. The zircon grains typically show a -Core–rim zonation internal texture, characterised by truncation of the core zoning pattern by the rim. The cores are generally oscillatory or sector zoned and CL-brighter than the rims. These general features do not apply to all grains and in particular, the core/rim volume ratio is very variable from a grain to another.

Zircons from the Claveix-related porphyritic granite (PG20-05) and of the Gelles granite -(PG20-03) are euhedral, usually prismatic, especially in PG20-03. Zircons from PG20-05 present oscillatory zoning in most crystals. Some of them show a distinct euhedral light interface between a dark rounded core and an oscillatory zoned mantle, as represented by zircon V-23 in Figure 6. Zircons in sample PG20-03 show more complex -Core–rim zonation relationships than in PG20-05, although with commonly oscillatory-zoned, CL-brighter cores and CL–Darker rims showing faint or no zoning.

Fig. 6 Representative cathodoluminescence images of zircons from (A) PG20-01 paragneiss; (B) PG21-04 micaschist; (C) PG20-02 paragneiss; (D) PG20-03 Gelles granite; (E) PG20-04 diatexite; and (F) PG20-05 Claveix granite.

4.5.2 U-Pb–Pb dates

Micaschist sample PG21-04 exhibits U–Pb date distribution as the two paragneiss samples, with a prominent peak of Ediacaran dates (550–650 Ma) tailing down to ca. 800 Ma. However, older dates are much less frequent than observed in other samples, with only 3 concordant analyses yielding dates >800 Ma. The youngest concordant zircon population in this sample provides a Concordia date of 572.5 ± 7.2 Ma (MSWD = 1.5, n = 21, Fig. 7 A–B).

Some analyses in paragneiss sample PG20-01, mostly from zircon rims, do not provide concordant U–Pb dates, due to variable amounts of common Pb and/or Pb loss (Fig. 7C). Concordant analyses (n = 91), show a large variety of U–Pb dates, ranging from ca. 3325 to 545 Ma with a distinguished important population at ca. 600 Ma (Fig. 7C, inset). In addition, the date distribution is characterised by a few analyses with Archean dates, a Paleoproterozoic population between 2300 Ma to 1800 Ma, and a gap between 1400 Ma to 1000 Ma (Fig. 14C). The youngest population of this sample provides a Concordia date of 571.8 ± 8.2 Ma (MSWD = 0.75; n = 13, Fig. 7D).

Metatexitic paragneiss PG20-02 sample shows a very similar date distribution as that from PG20-01 presented above, except for the higher proportion of concordant Archean and Paleoproterozoic dates, notably a prominent peak at ca. 2000 Ma (Fig. 7E). The youngest concordant zircon population in this sample provides a Concordia date of 588.9 ± 7.1 Ma (MSWD = 1.6, n = 24, Fig 7E).

Diatexite sample PG20-04 presents a very large range of zircon U–Pb dates, with a statistical distribution very close to that of PG20-01 (Fig. 7F), i.e., an important population with concordant Neoproterozoic (909-543 Ma) dates, a lack of Mesoproterozoic (<1700 Ma and >1000) grains and minor populations of Paleoproterozoic (2303-1700 Ma) dates. Many analyses, especially from zircon rims, are strongly discordant presumably due to the presence of common Pb and/or Pb loss.

In the Claveix porphyritic granite sample PG20-05, there are as few concordant grains as in sample PG20-03 (Fig. 8A) but they all cluster together, defining a Concordia date of 338.0 ± 7.3 Ma (MSWD = 1.4; n = 6) (Fig. 8B). This date is identical within uncertainty to a lower intercept date of 333.7 ± 6.8 Ma (MSWD = 1.9; n = 28) calculated by the Discordia method including most of the variably discordant analyses spreading along a mixing line between radiogenic and common Pb (Fig. 8A).

Zircons from the Gelles granite sample PG20-03 present a large proportion of discordant dates, but concordant analyses show a dominant population at ca. 320 Ma, together with minor Neoproterozoic and Paleoproterozoic dates (Fig. 8C). Six out of 40 analysis from the Gelles granite are concordant and yield a Concordia date of 323.3 ± 9.7 Ma (MSWD = 2.2; n = 6) (Fig. 8D). Many of the discordant analyses define a linear trend connecting these concordant points to a common Pb endmember (Fig. 8C) and thus possibly corresponding to a mixing line between radiogenic and common Pb. The lower intercept, Discordia date obtained from most analyses along this trend is 323.1 ± 5.4 Ma (MSWD = 1.6; n = 22), which is statistically undistinguishable from the Concordia date.

Fig. 7 Concordia diagrams (Tera–Wasserburg, 238U/206Pb vs. 207Pb/206Pb) and kernel density estimate (KDE) plots showing the distribution of U–Pb isotopic analyses for metasedimentary rocks: (A–B) PG21-04 micaschist; (C–D) PG20-01 paragneiss; (E) PG20-02 paragneiss; and (F) PG20-04 diatexite. In (A), (C), (E) and (F), grey circles indicate rim analyses; in (B) and (D), empty circles are excluded from the calculation of Concordia dates. Only analyses with less than 5 per cent discordance (206Pb/238U ÷ 206Pb/207Pb age × 100) are included in the KDE plots.

Fig. 8 Concordia Tera–Wasserburg diagrams and KDE plots showing the distribution of U–Pb isotopic analyses for plutonic rocks: (A-B) Claveix-related porphyritic granite; (C-D) Gelles granite. Empty circles are excluded from the calculation of Concordia and Discordia dates; grey circles indicate rim analyses.

4.6 Lu-Hf–Hf analysis on zircon

Lu–Hf isotopic analysis of concordant zircon from all metamorphic and magmatic samples show a wide range of εHf(t) values from +13.6 to -25.5 (typical uncertainties are ± 1.5 εHf(t) units) (Fig. 9A–B). Zircons from micaschist PG21-04, paragneiss PG20-01, metatexitic paragneiss PG20-02 and diatexite PG20-04, as well as the few >500 Ma zircons from Claveix porphyritic granite PG20-05 and Gelles granite PG20-03, overlap in both age and εHf(t) values. The dominant population of zircon with Ediacaran U–Pb dates found in all samples notably spreads over the entire range of εHf(t) values, whereas the Paleoproterozoic and Archean populations show a more limited variability (ca. +5 to -10 with only few outliers at more positive or negative values).

For the two plutonic rocks, the initial εHf(t) values were calculated for each granite based on the dates of the main zircon populations (see section about U–Pb dating). The Claveix porphyritic granite shows on average less radiogenic zircon Hf isotopic composition than the Gelles granite, with average εHf(t) values of -5.2 ± 0.6 (2 S.D., n = 7) and -2.2 ± 1.2 (2 S.D., n = 8), respectively.

Fig. 9 Zircon Lu–Hf compositions for samples from this study and from the eastern French Massif Central (FMC) literature, expressed as εHf(t) versus U–Pb age, with (a) zircon ages from 200 Ma to 3500 Ma; and (b) zircon ages from 200 Ma to 800 Ma. Grey and pink areas represent inherited cores and magmatic zircons from S-type granites (Moyen et al., 2017), respectively. Zircon data for the Velay unit, LGU and UGU are compiled from (Chelle-Michou et al., 2017; Couzinié et al., 2019).

4.7 Monazite dating

4.7.1 Monazite textures

Monazites from the investigated samples range in size from 10 to 50 µm. In the paragneiss PG20-01 and metatexitic paragneiss PG20-02, monazite occurs mostly as inclusion in biotite, along its cleavage planes or as elongated crystals or clusters of crystals along the grain boundaries of minerals defining the main foliation (Fig. 10 A–B). Monazite is also present in sillimanite aggregates. In the diatexite sample PG20-04, monazite occurs as inclusion in feldspar or quartz of the granitic matrix, or within the metatexites enclaves associated with biotite and sillimanite,i.e., the mesosome (Fig. 10C). Chemical maps reveal that monazite grains from all samples do not usually exhibit internal zonation, although some rare grains show a core enriched in Th related to rims.

Fig. 10 Representative backscattered electron images of metamorphic monazites from paragneiss PG20-01 and PG20-02, and diatexite PG20-04, with corresponding lower-magnification and higher-brightness images showing the textural context of the grains. White circles indicate the location of analysis spots and corresponding individual 206Pb/238U dates.

4.7.2 Monazite Dates

Monazite U–Pb data from all samples show a simple distribution characterised by a dominant, if not exclusive population of concordant Carboniferous dates (Fig. 11). For all samples, it is possible to calculate Concordia dates using a majority of the data points, respectively 347.2 ± 3.8 Ma (MSWD = 1.3; n = 20) and 342.2 ± 4.6 Ma (MSWD = 1.4; n = 15) for two different thin sections of paragneiss sample PG20-01; 347.1 ± 4.2 Ma (MSWD = 0.38; n = 48) for metatexitic paragneiss sample PG20-02; and 357.6 ± 1.5 Ma (MSWD = 1.5; n = 30) for the diatexite PG20-04. There is no clear correlation between U–Pb age and textural location. Within a given sample, monazite grains from different textural positions, or different chemical domains within single grains (e.g. Th-rich core vs. Th-poor rim), most commonly show overlapping U–Pb dates. Conversely, some monazites showing significantly older dates (.ei. >360 Ma), or on the contrary, younger dates as in the case of the diatexite in which a few concordant analyses (n = 4) decreased to ca. 300 Ma, do not seem to be associated to a peculiar textural position as they are present both in leucosome and as inclusions in biotite.

On the other hand, in a few cases, distinct U–Pb dates were obtained from different spots within single crystals of monazite, pointing to a zonation that is not identified from BSE images or X-ray maps.

Fig. 11 Concordia Tera–Wasserburg diagrams and KDE plots for monazites from paragneiss (a–b) PG20-01, (c) PG20-02, and diatexite (d) PG20-04. Empty circles represent analyses excluded from the calculation of Concordia dates.

4.8 Discussion

4.8.1 A metamorphic gradient from micaschists-paragneiss to diatexites

Our field investigations confirm the structural and metamorphic trends of the studied area near Pontgibaud, as originally proposed (Tempier, 1969; Tempier, 1974; Fernandez and Tempier, 1977). Namely, micaschsits and paragneiss exposed to the south of the studied display a shallow–Dipping dominant foliation that is transposed in a steep–Dipping foliation toward the Sillon Houiller to the West and into a step–Dipping synmigmatitic foliation of the migmatitic paragneiss. This transposition is consistent with a metamorphic gradient delineated by metamorphic isograds mapped by Tempier (1967). We further document this metamorphic gradient with-P–Testimates obtained from pseudosection calculations.

The main mineral assemblage of sample micaschist PG21-04 is plagioclase + muscovite + biotite + quartz with rare andalusite and altered porphyroblast interpreted as pseudomorphs after cordierite (Fig. 4A and B). This assemblage is stable in a restricted-P–Tfield at 2.3 ± 0.3 kbar and 570 ± 20 °C, within the stability field of rutile that occur in the matrix of the sample. Ilmenite may have formed at higher temperature conditions at ca. 650 °C. With decreasing temperature from the Bt Fsp Crd San Qz Ilm field, K–Feldspar and cordierite modes decrease while muscovite, plagioclase and biotite modes increase in proportion reaching relatively equal amounts of 10-15 per cent in the lower temperature stability field containing andalusite and rutile, consistent with thin section observations.

Paragneiss sample PG20-01 contains biotite + plagioclase + K–Feldspar + quartz + sillimanite and cordierite replaced by sericite. Concordant leucosome is observed in macroscopic specimen. Muscovite as inclusion in biotite and feldspar suggests that it corresponds to a relic from a lower grade paragenesis. The mineral assemblages correspond to a restricted-P–Tfield around 3 ± 1 kbar and 650 ± 20 °C. In this field 10-20 per cent liquid is produced, and biotite composition is consistent with those observed in thin section (X_Fe ca. 55, Ti = 0.13-0.19 a.p.f.u). Sillimanite replacement by muscovite suggests a path with decreasing temperature, where muscovite increases abruptly from this field while sillimanite decreases, and the anorthite content of plagioclase composition decreases down to 15 per cent X_An. This trajectory probably extended to the biotite + plagioclase+ cordierite + muscovite + quartz field at ca. 2.6 kbar and ca. 530 °C (Fig.12C–D).

The dominant mineral paragenesis of paragneiss sample PG21-28, sillimanite + garnet + biotite + plagioclase with rare rutile inclusions in garnet and biotite-sillimanite intergrowth, is stable in a restricted-P–T field between 660 °C ± 10 °C and 5.3 ± 1.0 kbar (Fig.13A-B). Compositional zoning in garnet indicates a decreasing temperature towards the Bt+Fsp+Grt+Ilm+Ms+Qz field at lower temperature, consistent with minor late muscovite and rare ilmenite observed in the matrix. From this field, sillimanite, garnet, biotite and plagioclase modes increase with decreasing muscovite. The Mn enrichment observed in garnets’ compositional profiles (Fig. 5E), along with isopleths calculated for this sample, collectively suggest reequilibration at lower temperature compared to the core. The -Core–rim zonation zonation is consistent with a retrogressive path implying a temperature decrease of about 70°C. Leucosome is not recognised neither at microscopic nor macroscopic scale but thermodynamic modelling indicates an increase in melt up to 10 vol. per cent with increasing temperature within the inferred stability field. This suggests that some melt might have escaped this rock.

Metatexitic paragneiss sample PG20-02 contains biotite + plagioclase + quartz + sillimanite as well as porphyroblasts interpreted as cordierite pseudmorphs replaced by sillimanite-rich assemblages. Minute garnet is locally observed but could not be analysed due to its small size (< 200 µm) and the presence of numerous graphitic inclusions. Isolated leucosomes (max 10 vol. per cent) were also observed at the macroscopic scale. The mineral assemblages correspond to a restricted-P–Tfield around 5 ± 1 kbar and 650 ± 50 °C. The composition of biotite (X_Fe = 55–58 and Ti = 0.15–0.19 a.p.f.u.) indicates maximum temperatures around 675°C just above 5 kbar (Fig.13C-D). Note that large biotite crystals included into cordierite pseudomorphs have systematically low X_Fe values (ca. 55) compatible with growth of cordierite at pressure between 5 and 6 kbar. Garnet is predicted at these conditions but in very low proportions (1 vol per cent), in agreement with petrographic observations. Retrogression of large cordierite porphyrolasts into sillimanite-rich assemblages suggests decreasing temperature between 675 and 650°C where the proportion of cordierite decreases and the one of sillimanite increases. The complete retrogression of cordierite porphyroclasts is consistent with a retrograde path extending toward the cordierite-out line located near 600°C between 5 and 4 kbar. The presence of plagioclase grains with sillimanite inclusions is also consistent with such a retrograde path, along which plagioclase content increases steadily (from 20 vol. per cent around 675 °C to 40 vol. per cent at 650 °C near 5 kbar) while sillimanite mode only increases moderately. Accordingly, plagioclase An content is compatible with temperatures below 650 °C. Collectively, the textural record, mineral modes and compositions thus indicate a retrograde path starting around 675 °C and 5.5 kbar down to 600 °C and 4.5 kbar.

These data are consistent with a continuous, high-temperature//low-pressure metamorphic gradient from micaschists-paragneiss to migmatites with a pressure that ranges from ca. 2 to 5 kbar and temperature increasing from ca. 570 to 660 °C from South to North. We propose that this metamorphic gradient is representative of the thermal state of the Variscan orogenic crust after the climax of crustal thickening and following significant thermal relaxation.

Fig. 12 Results of thermodynamic modelling shown as P–T pseudosections (a, c) and isopleth and isomode contours (b, d) for micaschist PG21-04 (a–b) and paragneiss PG20-01 (c–d). Bulk compositions (in oxide wt. per cent) used for calculations are indicated. Coloured lines represent the appearance of index minerals. Arrows indicate P–T paths from peak to retrograde conditions.

Fig. 13 Results of thermodynamic modelling shown as P–T pseudosections (A, C) and isopleth and isomode contours (B, D): (A–B) for paragneiss PG21-28, and (C–D) for paragneiss PG20-02. Bulk compositions (in oxide wt. per cent) used for calculations are indicated. Coloured lines represent the appearance of index minerals. Arrows indicate P–T paths from peak to retrograde conditions.

4.8.2 Interpretation of zircon/monazite data

The youngest population of zircons in micaschist PG21-04, paragneiss PG20-01, and metatexitic paragneiss PG20-02 provides Concordia dates of 572.5 ± 7.2 Ma (MSWD = 1.5, n = 21, Fig. 6F), 571.8 ± 8.2 Ma (MSWD = 0.75; n = 13, Fig. 7B) and 588.9 ± 7.1 Ma (MSWD = 1.6, n = 24, Fig. 6C), respectively. Considering the metasedimentary nature of these samples and the likely detrital nature of the zircon population with strongly variable U–Pb dates, these dates are interpreted as reflecting the maximum deposition age of the sedimentary protoliths. Zircons from diatexite sample PG20-04 do not provide any reliable age that would correspond to Variscan metamorphism. The range of U–Pb dates is therefore interpreted as reflecting mainly inherited zircon from - meta–sedimentary protoliths, similar to the micaschist and paragneiss.

The Concordia dates of 338.0 ± 7.3 Ma and 323.3 ± 9.7 Ma obtained on euhedral, oscillatory-zoned magmatic zircon of the Claveix porphyritic granite (PG20-05) and the Gelles granite (PG20-03), are interpreted to represent the ages of crystallization of these plutons. In previous work, the Claveix granite was already identified to be older than the Gelles granite (Fernandez, 1969; Négroni, 1981; Hottin et al., 1989) and was argued to be synchronous with the emplacement of Upper Visean volcanic tuff at ca. 335–330 Ma (Hottin et al., 1989). This magmatic event at 340–330 Ma has been documented throughout the northeastern FMC (Binon and Pin, 1989; Laurent et al., 2017). In turn, the crystallisation age of the Gelles pluton is consistent with the previously published Rb–Sr mineral isochron age of 319 ± 12 Ma (Fernandez, 1969).

Monazite grains in paragneiss samples, which show a dominant population of concordant dates, are unzoned, embedded in the minerals of the main paragenesis of the rock (biotite, sillimanite, and quartzo–feldspathic leucosome) and are and aligned with the main foliation of the rock. Accordingly, monazite likely crystallised in the-P–Tconditions inferred for the main paragenesis of the samples and monazite U–Pb Concordia dates ranging from 357.6 ± 1.5 Ma to 342.2 ± 4.6 Ma are interpreted as dating the corresponding metamorphic event. The small number of concordant ages older than ca. 360 Ma were all obtained on crystals included in high-temperature minerals (biotite, sillimanite) or assemblages (leucosome), thus possibly corresponding to the age of the onset or prograde stage of the HT event.

In diatexite PG20-04, monazite dates as old as 600 Ma are interpreted as inherited, which suggests that the temperature of partial melting was not high enough to insure complete dissolution-resetting of pre–Existing monazite and zircon grains (Rubatto et al., 2001; Kelsey et al., 2008; Yakymchuk and Brown, 2014; Guergouz et al., 2018). This is consistent with the lack of metamorphic rims in zircons from micaschist, paragneiss, metatexite and diatexite samples. This feature was recognised in other migmatites of the FMC corresponding to low-temperature melting (Couzinié et al., 2021) and, predicted from thermodynamic modelling, which shows that at <700 °C, a maximum 10 per cent zircon and 40 per cent monazite from the rock can be dissolved in the melt phase (Kelsey et al., 2008).

On the other hand, the presence of a few (n = 4) concordant monazite analyses showing U–Pb dates younger than 357.6 ± 1.5 Ma in this sample could be interpreted as recording protracted monazite crystallization throughout metamorphism and subsequent melting. Alternatively, given the diatexite nature of PG20-04, monazites with a range of ages could be incorporated through mechanical mixture of fragments of rocks and melts from different protoliths, coming from a deeper crustal level and possibly recording different metamorphic ages. However, none of these two hypotheses is consistent with the fact that diatexite PG20-04 is cross–Cut with sharp contacts by Claveix porphyritic granite (sample PG20-05) that crystallised at ca. 338 Ma (see above), indicating that the migmatites had already cooled down and were significantly exhumed at this time. Therefore, this range of monazite ages likely rather reflects late disturbance of the U–Pb system, presumably due to the intrusion and/or late magmatic hydrothermal activity related to the Claveix and Gelles granite.

4.8.3 Links between metamorphic and magmatic units in the Pontgibaud area

As presented in the previous sections, the compositions, assemblages, metamorphic evolution and ages of micaschists-paragneiss, migmatites and granites in the Pontgibaud area support that these rocks are genetically linked. The gradual field transition from micaschists to migmatites and the corresponding, continuous Low-Pressure/High-Temperature metamorphic gradient further suggest that these rocks are part of the same lithotectonic unit showing different exposure levels (shallower in the South, deeper in the North). From this perspective, and the steep–Dipping synmigmatitic foliation of the migmatites would correspond to the transposition of the shallow–Dipping foliation of the metasedimentary rocks. U–Pb dating of metamorphic monazite points to broadly synchronous amphibolite–Facies metamorphism in all samples (Fig. 14A–B) at ca. 355-345 Ma. Lastly, the paragneiss and diatexite samples display a similar population of former detrital zircon ages, which is consistent with a petrogenetic link between these rocks (Fig. 14C–D), i.e., a prominent peak at 630 Ma, a lack of Stenian ages (1000-1200 Ma) and the existence of a Paleoproterozoic population of mainly Orosirian age (1800 to 2050 Ma). Former detrital zircon from the paragneiss samples and diatexite fully overlap in the Hf isotopic vs. U–Pb age space (Fig. 9), supporting that these rocks likely present a genetic link, i.e. represent a single litho-tectonic unit affected by a metamorphism that increased from South to North.

The Claveix and Gelles granites of the Pontgibaud area are peraluminous. They contain biotite, cordierite, muscovite and tourmaline, and thus correspond to Cordierite- and Muscovite Peraluminous Granites (CPG and MPG) in the classification of Barbarin (1999). Magmas of CPG/MPG compositions are common in the eastern FMC and were proposed to result from melting of the country rocks, dominantly of meta–sedimentary composition (Laurent et al., 2017; Moyen et al., 2017; Villaros et al., 2018). Inherited zircons from the plutonic rocks present the same U–Pb age distributions (Fig. 14) and Lu–Hf isotopic characteristics (Fig. 9) as those of the paragneiss and diatexite rocks, which support the hypothesis that granites of the Pontgibaud area are potentially sourced from magma extraction derived from partial melting of - meta–sedimentary rocks similar to the exposed ones, but at a deeper level of exposure (see next section for detailed discussion about this point).

Our results and interpretations have implications for the attribution of metamorphic rocks of the Pontgibaud area to the litho-tectonic units defined at the scale of the FMC. As introduced earlier (see Fig. 1 and Geological Setting), the northern vs. southern parts of the study area were respectively attributed to the UGU and LGU, with disputed location of the boundary zone depending on the authors and maps (Hottin et al., 1989; Chantraine et al., 2003). In any case, the superposition of the UGU over the LGU should be marked by a mylonitic contact and an inverted metamorphic gradient, which were not identified in the field. In contrast, as discussed here, all lines of evidence are rather consistent with all metamorphic rocks belonging to the same litho-tectonic unit affected by synchronous LP-HT metamorphism. The metamorphic conditions and LP-HT gradient are typical of the LGU in the region (Schulz, 2009) and in the E–FMC in general (Vanderhaeghe et al., 2020), in contrast with the HP-LT gradient recorded by UGU rocks. The age of the LP-HT event of ca. 357-345 Ma is consistent with metamorphic ages from the LGU elsewhere in the E–FMC, notably from the nearby Sioule series (Schulz 2009; Do Couto et al., 2016; Monnier et al., 2021) where UGU monazites are generally younger (340–320 Ma; Schulz, 2009) and in LGU diatexites in a similar structural position in the Livradois area (Gardien et al., 2011). Moreover, the range of maximum deposition ages of sedimentary protoliths in the Pontgibaud area (571.8 ± 8.2 Ma to 588.9 ± 7.1 Ma) well overlaps that of -meta–sedimentary rocks of the LGU (556.8 ± 5.1 Ma to 592.4 ± 5.5 Ma) as determined by Couzinié et al., (2019) in the Velay area. Finally, although zircon populations of metamorphic rocks in the Pontgibaud region show similar ages and Lu–Hf isotopic composition as those (Chelle-Michou et al., 2017; Couzinié et al., 2019) from both LGU and UGU samples (Fig. 9, 14), they strikingly lack the 480 Ma age peak that is typical for the UGU gneisses (Chelle-Michou et al., 2017). Based on these features, we attribute the metasedimentary rocks and migmatites exposed in the Pontgibaud area to the Lower Gneiss Unit.

Fig. 14 KDE plots of U–Pb dates (206Pb/238U for dates younger than 1200 Ma; 206Pb/207Pb for dates older than 1200 Ma), comparing zircon and monazite data from this study with published data from the eastern French Massif Central (FMC). (A) Monazite 206Pb/238U dates’ distribution for paragneiss and diatexite (this study). (B) Monazite and zircon 206Pb/238U dates for LGU paragneiss (this study) and zircon dates for LGU and UGU (from Chelle-Michou et al., 2017; Couzinié et al., 2019). (C–D) Distribution of zircon dates of metasedimentary rocks from this study and from LGU, UGU and orthogneiss in the eastern FMC (compiled from Chelle-Michou et al., 2017; Couzinié et al., 2017, 2019; Do Couto et al., 2016).

4.8.4 Implications for the crustal structure and geotherm

As discussed in the previous section, the Variscan metamorphic rocks of the Pontgibaud area likely belong to the LGU, in contrast with what had been proposed in previous works (Hottin et al., 1989; Chantraine et al., 2003). On a larger scale, the LGU rocks that outcrop with a main subhorizontal foliation in the Pontgibaud region may continue in depth northward and southward to the Sioule series and the Artense dome respectively, where the LGU is overlain by the UGU unit in km-scale folded structures (Fig. 15). Therefore, correlation of the tectono-metamorphic pattern described in this study with the data presented by Do Couto et al., (2016) and Brousse et al., (1990) allows the construction of a crustal model, where the plutonic units would be sourced in LGU migmatite as result of melt segregation, and the latter would represent an important proportion of the crust in depth, derived from partial melting of rocks similar to the metamorphic ones exposed at the surface (Fig. 15). Arguments supporting this interpretation are developed in the following.

Formation and extraction of large volumes of Crd-Bt granitic to granodioritic magmas such as those represented by the Claveix and Gelles plutons requires melting conditions of >800 °C, as indicated by phase equilibria modelling (Villaros et al., 2018) on other E–FMC granitoids. Moreover, experimental data of partial melting of E–FMC meta-sediments indicate that magmas compositionally similar to the Claveix and Gelles granites are typically reproduced by melting at ca. 850–875 °C and 5–10 kbar (Montel and Vielzeuf, 1997). Such a temperature was not reached by the exposed metamorphic rocks. Although some of the exposed migmatites are diatexites, they still contain abundant biotite pointing to temperatures <800 °C, such that they either may not all be “n situi” migmatites and represent local accumulation of melts transferred from lower crustal levels; and/or were formed by water–fluxed melting induced by the emplacement and crystallization of magmas sourced in the lower crust (as proposed in the Velay dome area; Barbey et al., 2015; Villaros et al., 2018, Couzinié et al., 2021). Either way, the presence of voluminous granite and diatexite at mid-/upper crustal levels requires transfer of granitic magma from lower crustal levels. This process seems to have widely occurred throughout the FMC domain in the Variscan orogen during the Carboniferous, as large volumes of crust–derived granites and diatexites were produced in this region during the Carboniferous (Laurent et al., 2017; Vanderhaeghe et al., 2020).

The process of deep partial melting and extraction of granitic melts typically leaves behind a residual, granulitic lower crust (Clemens 1990; Vielzeuf et al., 1990; White and Powell, 2002; Cawood et al., 2013; Clemens et al., 2020). In the E–FMC, this refractory lower crust is sampled by crustal xenoliths brought up by Cenozoic volcanoes (Leyreloup et al., 1977; Leyreloup, 1992) whose petrology and zircon U–Pb systematics are fully consistent with an origin through extensive melting and melt extraction to form the mid-/upper crustal granites and migmatites (Fig. 15; Vanderhaeghe, 2009; Villaros et al., 2018; Laurent et al., 2023). The metamorphic conditions recorded by the lower crustal xenoliths range from ca. 5 kbar and 700 °C to ca. 10 kbar and 950 °C (see Leyreloup, 1992; Laurent et al., 2023 and references therein),i.e. define a LP-HT metamorphic gradient that perfectly aligns with that recorded by the LGU rocks in the Pontgibaud area. Moreover, these lower crustal rocks underwent granitic melt loss from ca. 340 to 310 Ma (Laurent et al., 2023), in agreement with the crystallization ages of the Claveix and Gelles granites (ca. 340–320 Ma) obtained here. Altogether, granitic melts that fed the upper-/mid–Crustal plutons and diatexites in the Pontgibaud area were likely generated from a lower crustal source region that underwent a similar metamorphic evolution as that recorded by the granulite xenoliths sampled elsewhere in the E–FMC. Therefore, melt–depleted granulites complementary to the diatexites and granites exposed at the surface likely form the bulk of the lower crust beneath the Pontgibaud area.

The lower crustal granulite xenoliths have chemical and petrographic characteristics consistent with the loss of 25 to 30 vol. per cent melt (Laurent et al., 2023), in agreement with the minimum melt fractions required for efficient melt extraction (20-25 vol. per cent melt; Vigneresse et al., 1996). This requires that the granulitic lower crust represents 2 to 3 times the volume of granitic magma at the present level of exposure. Considering a crustal thickness of ca. 28 km in the Pontgibaud area (Zeyen et al., 1997) and a closed-system redistribution of material during Carboniferous melting and granitic magma transport, this entails that the crust consists of a vertical section comprising a 7–14 km-thick upper layer of granite and diatexite (± unmolten country rocks close to the surface), underlain by 14–21 km of refractory migmatite and granulite (Fig. 15). This must be considered as an extreme estimate considering that the upper crust is not entirely formed of liquid extracted from the lower crust.

Fig. 15 Model for the Variscan crustal structure in the Pontgibaud region. LGU paragneiss and migmatites are exposed from Pontgibaud to the St Gervais Massif northwards. The transition from paragneiss to metatexite and diatexite is gradual. At depth, migmatites are products of partial melting of LGU and are abundant down to the granulitic lower crust. In this model, granites, migmatites and paragneiss are genetically linked.

5 Conclusion

Field observations, metamorphic conditions and geochronological data presented in this study suggests that: 1) the Pontgibaud area exposes a continuous section along a mid–Carboniferous MP to LP/HT metamorphic gradient from micaschists to migmatites, belonging to the Lower Gneiss Unit of the Variscan orogenic crust; 2) this event is followed by the generation and emplacement of crust–derived granitoids represented by the Claveix and Gelles plutons.

Most metamorphic rocks consist of dominantly -meta–sedimentary lithologies with late Edicaran protoliths (maximum deposition ages of ca. 590–560 Ma) similar to those documented elsewhere in the FMC for the LGU. These sediments have recorded-P–Tconditions ranging from 2.3–5.3 kbar and 570° to 660°C, dated from 357.6 ± 1.5 to 342.2 ± 4.6 Ma by LA-ICP-MS U–Pb analyses of metamorphic monazite. The contact between the metasedimentary rocks and migmatites is crosscut by peraluminous, cordierite-biotite-bearing Claveix and Gelles granitic plutons dated at 338.0 ± 7.3 Ma and 323.3 ± 9.7 Ma respectively by U–Pb on zircon. Extensive melting and extraction of a large volume of granitic magma from these lithologies requires higher temperature conditions (800°C) than those recorded by the exposed rocks. Accordingly, we propose that the Pontgibaud area is underlain by metamorphic rocks with a similar protolith as the one exposed at the surface, but affected by higher grade metamorphism and magma extraction, which is consistent with the inferred residual nature of granulites sampled by Cenozoic volcanoes. The lower crust is interpreted as a high–grade, residual equivalent of the regionally exposed migmatitic -meta–sedimentary rocks (Leyreloup, 1992; Laurent et al., 2023).

Accordingly, we propose that this segment of the Variscan belt has recorded high temperature metamorphism, deformation, partial melting and melt segregation in an interval between ca. 350 Ma to 320 Ma, consistent with propositions where a time gap of 20-30 Ma after crustal thickening and/or increase of the heat flux through the Moho is required to trigger the pervasive melting of the crustal column up to the middle crust (Thompson and Connolly 1995; Vanderhaeghe and Teyssier 2001; Cavalcante et al., 2018). In this scenario, partial melting of the continental crust would be continuous in this 20-30 Ma interval, whereas segregation; accumulation, transport and emplacement of plutons such as Claveix and Gelles represent episodic events dated by U–Pb on zircon at 338 ± 7 Ma et à 323 ± 7 Ma, respectively.

The vertical distribution of lithologies between the upper to lower crust, the latter being ill–defined, may influence the geotherm and therefore the potential for deep geothermal energy production in a given region. Our study has clarified the vertical structure of the Variscan crust in the Pontgibaud area, which will allow to discuss the distribution of heat-producing elements as well as thermal conductivity within the crust column and thus more precisely model the current geotherm (Bellanger et al., 2019).

Acknowledgments

This study has received funding from the ANR (grant GERESFAULT number ANR-19-CE05-0043) and the grant “CNRS IRN FALCoL” to OV and from the CNRS-INSU TelluS-SYSTER program (grant “QuantHyCCO” to OL). The thorough review of four anonymous reviewers helped improve this manuscript.

Supplementary Material

Table X. LA-(MC-)ICP-MS U-Pb, trace element and Lu-Hf isotopic analyses on zircon - Metadata.

Table Y. LA-ICP-MS U-Pb isotopic analyses on monazite - Metadata.

References

All Tables

All Figures

Fig. 1 (a) General context of the French Massif Central. (b) Sketch map of the Variscan basement of the Pontgibaud region showing the spatial distribution of LGU, UGU and PAU units, and localities where these nappes were defined (Sioule Series, Artense Dome and Haut Allier). Modified after Chantraine et al., (2003). (c) Geological map of the Pontgibaud region. The dashed black line represents the UGU/LGU contact proposed by Chantraine et al. (2003). Note that paragneiss are either attributed to LGU (north-west of Gelles granite) or UGU (south-east of Gelles granite). Modified after Hottin et al., (1989).

In the text

	Fig. 2 Interpretative cross-sections (locations of the profiles are shown in Fig. 1c) showing relationships between micaschist, paragneiss, migmatites and plutonic rocks, and structural observations.
In the text

Fig. 3 Outcrop features of (A) Ms–Bt micaschist and macroscopic sample (inset) affected by crenulation folding with N–S trending axes; the observation plane of the macroscopic sample is marked in white. (B–D) Bt–Sil paragneiss affected by an isoclinal fold (B), and displaying concordant to discordant leucosomes (C, D). (E) Migmatite outcrop displaying foliated metatexite crosscut by Crd-bearing diatexite. (F) Migmatite and Claveix granite discordant contact, inset showing a macroscopic sample of the granite. (G–H) Gelles granite exhibiting (G) C–S fabric and (H) oriented K-feldspar phenocrysts in a vertical plane (pink pencil as scale). Structural measurements are reported in dip direction/dip.

In the text

Fig. 4 Photomicrographs of (A–B) micaschist PG21-04: (A) in cross-polarised light (XPL), showing a pseudomorph, presumably after cordierite; (B) andalusite occurrence parallel to the main foliation. (C) Bt–Sil paragneiss PG20-01 in plane-polarised light (PPL), showing replacement of sillimanite by muscovite. (D) Grt–Bt paragneiss in PPL, showing garnets with inclusion-rich cores oriented along foliation. (E) Bt–Sil metatexitic paragneiss PG20-02, exhibiting a pseudomorphed crystal (probably Crd) replaced by sillimanite + muscovite. (F) Crd–bearing diatexite sample PG20-04, showing poikiloblastic texture of K-feldspar enclosing plagioclase crystals. Note white arrows indicating interconnected interstitial quartz, plagioclase with limpid rims, and resorbed plagioclase. (G) Claveix granite showing porphyritic texture with globular, corroded quartz. (H) Gelles granite in XPL, showing coarse-grained matrix texture.

In the text

Fig. 5 Mineral chemistry results from electron microprobe analyses in micaschist (PG21-04), paragneiss (PG20-01, PG20-02, PG21-28) and diatexite (PG20-04) samples. (A) Feldspar ternary diagram. (B) Fe + Mg versus Si contents in muscovite. (C–D) Biotite Si and Ti a.p.f.u. compositions versus Mg# (100 × Mg/[Mg + Fe2+]). (E) Garnet composition shown as molar proportions of end-member compositions (Alm = almandine, Grs = grossular, Prp = pyrope, Sps = spessartine) across a crystal profile.

In the text

	Fig. 6 Representative cathodoluminescence images of zircons from (A) PG20-01 paragneiss; (B) PG21-04 micaschist; (C) PG20-02 paragneiss; (D) PG20-03 Gelles granite; (E) PG20-04 diatexite; and (F) PG20-05 Claveix granite.
In the text

Fig. 7 Concordia diagrams (Tera–Wasserburg, 238U/206Pb vs. 207Pb/206Pb) and kernel density estimate (KDE) plots showing the distribution of U–Pb isotopic analyses for metasedimentary rocks: (A–B) PG21-04 micaschist; (C–D) PG20-01 paragneiss; (E) PG20-02 paragneiss; and (F) PG20-04 diatexite. In (A), (C), (E) and (F), grey circles indicate rim analyses; in (B) and (D), empty circles are excluded from the calculation of Concordia dates. Only analyses with less than 5 per cent discordance (206Pb/238U ÷ 206Pb/207Pb age × 100) are included in the KDE plots.

In the text

	Fig. 8 Concordia Tera–Wasserburg diagrams and KDE plots showing the distribution of U–Pb isotopic analyses for plutonic rocks: (A-B) Claveix-related porphyritic granite; (C-D) Gelles granite. Empty circles are excluded from the calculation of Concordia and Discordia dates; grey circles indicate rim analyses.
In the text

Fig. 9 Zircon Lu–Hf compositions for samples from this study and from the eastern French Massif Central (FMC) literature, expressed as εHf(t) versus U–Pb age, with (a) zircon ages from 200 Ma to 3500 Ma; and (b) zircon ages from 200 Ma to 800 Ma. Grey and pink areas represent inherited cores and magmatic zircons from S-type granites (Moyen et al., 2017), respectively. Zircon data for the Velay unit, LGU and UGU are compiled from (Chelle-Michou et al., 2017; Couzinié et al., 2019).

In the text

	Fig. 10 Representative backscattered electron images of metamorphic monazites from paragneiss PG20-01 and PG20-02, and diatexite PG20-04, with corresponding lower-magnification and higher-brightness images showing the textural context of the grains. White circles indicate the location of analysis spots and corresponding individual 206Pb/238U dates.
In the text

	Fig. 11 Concordia Tera–Wasserburg diagrams and KDE plots for monazites from paragneiss (a–b) PG20-01, (c) PG20-02, and diatexite (d) PG20-04. Empty circles represent analyses excluded from the calculation of Concordia dates.
In the text

	Fig. 12 Results of thermodynamic modelling shown as P–T pseudosections (a, c) and isopleth and isomode contours (b, d) for micaschist PG21-04 (a–b) and paragneiss PG20-01 (c–d). Bulk compositions (in oxide wt. per cent) used for calculations are indicated. Coloured lines represent the appearance of index minerals. Arrows indicate P–T paths from peak to retrograde conditions.
In the text

	Fig. 13 Results of thermodynamic modelling shown as P–T pseudosections (A, C) and isopleth and isomode contours (B, D): (A–B) for paragneiss PG21-28, and (C–D) for paragneiss PG20-02. Bulk compositions (in oxide wt. per cent) used for calculations are indicated. Coloured lines represent the appearance of index minerals. Arrows indicate P–T paths from peak to retrograde conditions.
In the text

Fig. 14 KDE plots of U–Pb dates (206Pb/238U for dates younger than 1200 Ma; 206Pb/207Pb for dates older than 1200 Ma), comparing zircon and monazite data from this study with published data from the eastern French Massif Central (FMC). (A) Monazite 206Pb/238U dates’ distribution for paragneiss and diatexite (this study). (B) Monazite and zircon 206Pb/238U dates for LGU paragneiss (this study) and zircon dates for LGU and UGU (from Chelle-Michou et al., 2017; Couzinié et al., 2019). (C–D) Distribution of zircon dates of metasedimentary rocks from this study and from LGU, UGU and orthogneiss in the eastern FMC (compiled from Chelle-Michou et al., 2017; Couzinié et al., 2017, 2019; Do Couto et al., 2016).

In the text

Fig. 15 Model for the Variscan crustal structure in the Pontgibaud region. LGU paragneiss and migmatites are exposed from Pontgibaud to the St Gervais Massif northwards. The transition from paragneiss to metatexite and diatexite is gradual. At depth, migmatites are products of partial melting of LGU and are abundant down to the granulitic lower crust. In this model, granites, migmatites and paragneiss are genetically linked.

In the text