Pre-Cadomian to late-Variscan odyssey of the eastern Massif Central, France: Formation of the West European crust in a nutshell

The East Massif Central (EMC), France, is part of the internal zone of the Variscan belt where late Carboniferous crustal melting and orogenic collapse have largely obliterated the pre- to early-Variscan geological record. Nevertheless, parts of this history can be reconstructed by using in-situ U-Th-Pb-Lu-Hf isotopic data of texturally well-defined zircon grains from different lithological units. All the main rock units commonly described in the EMC are present in the area of Tournon and include meta-sedimentary and meta-igneous rocks of the Upper Gneiss Unit (UGU) and of the Lower Gneiss Unit (LGU), as well as cross-cutting Variscan granitoid dikes and a heterogeneous granite coring the major Velay dome. Herein we demonstrate that the UGU and the LGU have markedly distinct zircon records. The results of this study are consistent with deposition of the protoliths of the paragneisses within a back-arc basin that was located adjacent to the Arabian-Nubian shield and/or the Saharan Metacraton during the late Ediacaran and collected detritus from the Gondwana continent. At ~ 545 Ma some of these sedimentary rocks were affected by a first melting event that formed the protoliths of the LGU orthogneisses, those of which subsequently remelted at ca. 308 Ma to form the Velay granite-migmatite dome. Protoliths of the UGU result mainly from a bimodal rift-related magmatism at ~ 480 Ma, corresponding to melting of the Ediacaran sediments and depleted mantle. Zircon rims from the UGU additionally provide evidence for a metamorphic/migmatitic overprint during the Lower Carboniferous (~ 350–340 Ma). Finally, several generations of granite dikes of which inherited zircons display characteristics of both the UGU and the LGU were protractedly emplaced from ~ 322 Ma to ~ 308 Ma, the youngest of which being coeval with the formation of the Velay dome. Our data further show that the vast majority of crustal material ultimately involved in the Variscan orogeny, which forms the present-day basement in the EMC, was derived from a sedimentary mixture of various components from the Gondwana continent deposited in Ediacaran times, with no evidence for the involvement of an older autochthonous crust.

These collisions resulted in the amalgamation of the supercontinent Pangea, and in the formation of a vast mountain belt, the amplitude of which is often compared to that of the modern-day Himalayas (Dewey and Burke, 1973;Kroner and Romer, 2013;Ménard and Molnar, 1988;Stampfli et al., 2013) although such a comparison has recently be questioned (Franke, 2014). The French Massif Central is one of the largest Variscan massifs and represents the exhumed internal zone of the orogen (e.g., Lardeaux et al., 2014; Fig. 1), thus offering insightful opportunities to study geologic processes that were at play in the Variscan orogenic core. In addition, pre-Variscan relicts (mostly gneisses and schists) of the eastern Massif Central (EMC; Fig. 1) are intimately associated with the Variscan granitoids and can further provide insight into the evolution of pre-Variscan terranes (i.e., timing of crustal formation and reworking, related geodynamic settings, and paleo-geographic positions) and into their role into the Variscan orogeny. Ultimately, the study of the EMC can provide a time transgressive window to study the protracted evolution of the local crust.
Zircon has proven to be a resistant mineral in a range of extreme geological processes from the surface to the deep Earth crust (e.g., Harley and Kelly, 2007). U-Pb dating of zircon

A C C E P T E D M A N U S C R I P T
is considered to be one of the most robust techniques for dating magmatic and metamorphic events (e.g., Corfu, 2013). In addition, combined U-Pb dating and Hf isotope analyses of zircon can provide key information regarding the source of the magma from which zircon crystallized as well as on the timing of crustal formation and reworking (Condie et al., 2011;Kemp et al., 2006). Moreover, combined U-Pb dating and Hf isotope tracing has proven efficient in retrieving reliable chrono-petrological information even from complexly zoned zircons grains formed and/or altered during multiple metamorphic, magmatic, and/or sedimentary processes (e.g., Gerdes and Zeh, 2009;Kemp et al., 2006;2009;Zeh et al., 2010aZeh et al., , 2010b. As a result, zircon is the perfect tool to investigate crustal evolution in the eastern Massif Central (EMC; Fig. 1) where very intense late-Variscan partial-melting and magmatism has significantly disrupted most of the pre-Variscan rock record.
Compared with other Variscan massifs such as Bohemia and Iberia, the EMC has been the subject of relatively few modern geochronological studies. The specific timing of pre-and syn-Variscan tectono-magmatic events, and their associated geodynamic-paleogeographic environments therefore remain poorly defined within the area. Herein, we present an integrated U-Pb dating (n = 771) and Hf isotopic (n = 339) study of magmatic, detrital and metamorphic zircon grains/domains from the Tournon area of the EMC (Figs. 1, 2). The Tournon area presents a complete cross-section through the classically described nappe pile and crosscutting magmatic rocks of the EMC (Lardeaux et al., 2014;Ledru et al., 1989). The area therefore represents a key locality for study of both the pre-and syn-Variscan history of the EMC. Our sampling strategy was designed to encompass the main lithological units within the Tournon area that are representative across the EMC. The new data are used to constrain the Precambrian paleogeographic position of the Massif Central terrane, to discuss the evolution of the north Gondwana margin throughout the Cadomian and Variscan orogenies and to provide a synoptic timeframe for Variscan magmatic and metamorphic

A C C E P T E D M A N U S C R I P T
events in the EMC. Finally, we discuss the genesis and reworking of the western European continental crust.
Dates between 432 and 408 Ma (ID-TIMS U-Pb on multigrain zircon fractions, LA-ICPMS U-Pb on zircon, and EMPA U-Th-Pb on monazite) on eclogite and granulite samples have been attributed to HP metamorphism (Berger et al., 2010;Do Couto et al., 2015;Ducrot et al., A C C E P T E D M A N U S C R I P T 1983; Paquette et al., 1995;Pin and Lancelot, 1982) related to early Variscan subduction of this(ese) oceanic domain(s) (Faure et al., 2009;Ledru et al., 1989;Matte, 1991).
Subsequently, exhumation of the UGU in the northern part of the Massif Central was driven by isothermal decompression accompanied by anatexis between 389 and 375 Ma (Boutin and Montigny, 1993;Costa and Maluski, 1988;Duthou et al., 1994). Available data suggest that exhumation of the UGU was diachronous from north to south: while in the northern part (Lyonnais area, Northeast of St-Etienne, Fig. 1), the UGU had cooled below 300°C by 350-340 Ma ( 40 Ar/ 39 Ar data on amphibole, biotite and muscovite; Costa et al., 1993), partial melting of the UGU during exhumation was ongoing at 345 ± 10 Ma in the southern Massif Central (Pin and Lancelot, 1982). This event likely corresponds to the stacking of the UGU nappe(s) atop the Lower Gneiss Unit (see below).
The area south of the Velay granite-migmatite dome consists of greenschist to loweramphibolite facies metasediments belonging to the "Para-autochtonous Unit" (Fig. 1).
Metamorphism was polyphased from ca. 340 to 310 Ma and related to thrusting, granite emplacement and the far field effect of the Velay granite-migmatite dome (Bouilhol et al.,

A C C E P T E D M A N U S C R I P T
2006; Faure et al., 2001). It has been suggested that these metasediments were deposited along the North Gondwana margin, presumably from the Ediacaran through the Lower Cambrian (Melleton et al., 2010).

The Tournon areageology and sampling
The Tournon area is located along the eastern border of the Velay dome along the Doux valley a few kilometers from the town of Tournon (Fig. 1). It forms part of a series of N-S

A C C E P T E D M A N U S C R I P T
aligned klippen of the UGU that rest atop the LGU, and are dissected by a series of NE-SWstriking dextral strike-slip faults (Gardien, 1993). On the eastern flank of the synformal UGU klippe (Fig. 2), the base of the UGU is characterized by a bimodal association dominated by amphibolite (sample TN13) with alternating layers of orthogneisses and minor sillimanitebearing gneisses (sample TN21), a lithological association typical of the LAC. These rocks record MP-MT metamorphism that reached partial melting conditions (as shown by concordant leucosomes), but locally preserves relicts of metabasite boudins and enclaves having been subjected to granulitic and eclogitic conditions (Gardien, 1993 tend to be less foliated than late dikes (samples TN11 and TN32).
On the western flank of the klippe, the UGU is separated from a homogeneous biotitebearing granite phase (sample TN01) of the Velay granite-migmatite complex by a prominent NNW-SSE striking, sub-vertical mylonitic zone (Fig. 2). The more typical, heterogeneous Velay granite (sample TN09) is exposed further West in the Doux valley, below migmatitic cordierite-bearing gneisses and augen gneisses (Fig. 2). This heterogeneous granite contains many fragments of country rocks (including an enclave of porphyritic biotite granite dated at 321.9 ± 1.3 Ma; Laurent et al., 2017) and is characterized by the presence of cockade-type cordierite (see Barbey et al., 1999).
The contact between the UGU and the LGU is exposed along the eastern flank of the klippe, where it is transposed into the synmigmatitic foliation. The LGU in this area is made up by mylonitic orthogneisses, migmatitic paragneisses (samples TN17 and TN46) and small

A C C E P T E D M A N U S C R I P T
lenses of amphibolite (Fig. 2). The degree of migmatization of the LGU broadly increases toward the east, reflecting an evolution towards lower structural (Fig. 2). Further east, the LGU is intruded by the 321.1 ± 1.1 Ma-old (U-Pb zircon) Tournon porphyritic biotite granite (Laurent et al., 2017) which belongs to the suite of the "peri-Velay" granites .
Detailed descriptions and coordinates of all samples investigated are provided within the electronic supplementary material (ESM1).

Analytical Methods
Zircon separation was carried out at the University of Geneva. Rock samples were crushed and <400 µm sieved fractions were processed using a Wilfley shaking table, a Frantz magnetic separator and heavy liquids (methylene iodide) to extract zircons. Zircons grains were handpicked under a binocular microscope, mounted in epoxy, cut in half, and polished to expose their interior. In order to obtain the largest variety of grain types and minimize sampling bias, zircons were selected randomly without filtering for inclusion-free, fracturefree, core-free or transparent grains. All grains were subsequently imaged with cathodoluminescence (CL) on a JEOL CamScan scanning electron microscope (using a 10 kV accelerating voltage) at the University of Geneva (Fig. 3).
U-Th-Pb isotopic data were obtained by laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS) at the Laboratoire Magmas et Volcans of Clermont-Ferrand, France, using an Agilent 7500cs ICP-MS equipped with a dual pumping system to enhance sensitivity , and interfaced to a Resonetics Resolution M-50 193 nm ArF Excimer laser system. Data were corrected offline for U-Pb fractionation and instrumental mass discrimination. These corrections and the determination of the U-Th concentrations were carried out by normalizing to the GJ-1 zircon standard (Jackson et al.,

A C C E P T E D M A N U S C R I P T
2004). The 91500 zircon standard (Wiedenbeck et al., 1995), was used as secondary standard to control the reproducibility and accuracy of the method.
Lu-Hf isotope measurements were performed by laser ablation, multi-collector inductively coupled plasma mass spectrometry (LA-MC-ICPMS) at the Goethe University, Frankfurt am Main (Germany) and the University of Geneva (Switzerland). All spots for Lu-Hf isotope analyses were placed "on top" of previous U-Pb dating spots, or alternatively within the same internal zircon domains as identified in CL images, exclusively where concordant U-Pb dates were obtained. Accuracy and external reproducibility of the method were controlled by repeated analyses of 176 Hf/ 177 Hf ratios in reference zircon standards GJ-1 (Morel et al., 2008), Plešovice (Sláma et al., 2008), Temora (Woodhead and Hergt, 2005), Mud Tank (Woodhead and Hergt, 2005) and 91500 (Blichert-Toft, 2008) Details concerning analytical methods, instrument setup, data reduction protocols and results on zircon reference materials for both U-Pb and Lu-Hf used in both laboratories can be found in the electronic supplementary material (ESM 2).

Results
In this study, only zircon dates with a concordance level of 98-102% concordance (defined as  Fig. 7) and a huge scatter in εHf t from +8 to -27 (Fig. 8).

Upper gneiss unit
In the UGU gneisses (TN07, TN12, TN21), the prominent peak of Lower Ordovician dates at ca. 480 Ma was most frequently obtained from zircon rims overgrown on older cores (~70%), and less often from entire grains (~30%; Figs. 3, 4). Furthermore, in contrast to the very scattered Hf isotopic signature of zircons older than 550 Ma (εHf t of -28 to +10), this Ordovician peak shows a relatively homogeneous Hf isotopic signature (εHf t of -1 to -7, excluding few outliers; Fig. 8). We interpret those observations as reflecting an important ca.
480 Ma thermal event that resulted in partial melting of the protolith, zircon resorption, and subsequent crystallization of isotopically homogeneous overgrowths. The overgrowths generally have lower Th/U ratios (<0.2) than the older Neoproterozoic cores (>0.2; Fig. 7), which may indicate that zircon recrystallized in presence of a competing phase for Th, such as monazite. down to 0.002) compared with the cores (0.05-1; Fig. 7). This observation is commonly considered characteristic of metamorphic zircon where recrystallization purges Th and Pb from the crystal lattice in the metamorphic domains, resulting in low Th/U rim and resetting of the U-Pb system (e.g., Hoskin and Schaltegger, 2003). In the banded gneiss (TN12), the

A C C E P T E D M A N U S C R I P T
176 Hf/ 177 Hf t of these metamorphic zircon rims connect with their magmatic ca. 480 Ma cores along a trend characterized by 176 Lu/ 177 Hf ~ 0.015 (Fig. 11), consistent with an average crustal composition and thus, with the expected clastic sedimentary protolith of sample TN12 (see above). Collectively, these results suggest that these zircon rims result from incipient partial melting of the gneiss, under near-solidus temperature conditions where the Th/U zircon/melt distribution coefficient is dramatically reduced owing to temperature-dependent lattice-strain partitioning constrains (Blundy and Wood, 1994). In contrast, 176 Hf/ 177 Hf t ratios of the metamorphic rims in the amphibolite (TN13) plot at significantly lower values than what would be expected from pure metamorphic zircon alteration ( Fig. 11), i.e. U-Pb age reequilibration without changing the Hf isotopic composition (Amelin et al., 2000;Gerdes and Zeh, 2009;Lenting et al., 2010). This necessarily requires that some amount of nonradiogenic Hf was added to the system during the metamorphic event, and incorporated in the newly crystallized zircon rims. A fluid or melt phase derived from the surrounding UGU gneisses would represent a plausible source of such non-radiogenic Hf because those gneisses, as indicated by sample TN12, show a much lower 176 Hf/ 177 Hf t than the amphibolite at the time of this metamorphic overprint (Fig. 11).
In summary, zircon U-Pb and Lu-Hf data support the interpretation that the youngest zircon rims in samples TN12 and TN13 formed by incipient dissolution/recrystallization of the original (detrital or magmatic) zircon, with subordinate Hf mobility possibly promoted by a coeval melt or fluid phase during an Upper Devonian to early Carboniferous (Variscan) migmatitic-metamorphic episode.

Lower gneiss unit
In the two migmatitic paragneisses (TN17 and TN46), zircons with concordant dates ranging from the Neoproterozoic (ca. 530 Ma) to the Mesoarchean, display scattered ɛ Hf t (-20 to  . 7). This suggests that they have a different origin and may derive from a single igneous unit emplaced between 550.2 ± 3.1 Ma (TN01) and 544.3 ± 3.1 Ma (TN09), possibly represented by the migmatitic augen orthogneisses (not dated) located NW of the study area ( Fig. 2). This assumption is supported by the fact that most orthogneisses of the LGU in the Velay area belong to a former, widespread S-type granitic batholith emplaced in the late Ediacaran at ca. 545 Ma (Mintrone, 2015;R'Kha Chaham et al., 1990). The broad and generally low Th/U values (0.02-1) of the ca. 550 Ma zircons from the Velay granite ( Fig. 7) suggest that they co-crystallized with a Th-rich accessory mineral such as monazite. Such monazite is documented in peraluminous orthogneisses from the southern Velay dome (Be Mezeme et al., 2006) and as inherited grains (with an age of ca. 540 Ma) within the microgranites associated with the Velay granite (Didier et al., 2013).

A C C E P T E D M A N U S C R I P T
In samples TN01 and TN09, few zircons cores overgrown by ca. 550 Ma rims have a large spread of both Neoproterozoic U-Pb dates and ɛ Hf t (-20 to +8), and more restricted Th/U ratios (2.5 to 0.3), all of which overlap with those from the detrital zircons of the paragneisses (Figs. 7, 8). Therefore, we interpret these older zircons either as (i) inherited from the source of the 550 Ma granites which may be comparable to the LGU paragneisses (maybe slightly older), (ii) assimilated from the LGU paragneisses upon emplacement of the 550 Ma granites or during the formation of the Velay granite, or (iii) a combination thereof.
However, it is noteworthy that at least some of the ca. 550 Ma zircons with Th/U ratio over 0.2 and ɛ Hf t between -0.5 and -2.5 (i.e., belonging to the main inherited population in the Velay granite) might be of detrital origin (assimilated or inherited) and cannot individually be discriminated from those magmatic zircons having similar Th/U and ɛ Hf t .
In the Velay granites, a few entire grains with oscillatory zoning or thin rims overgrown on older cores record Carboniferous dates (309 ± 2.6 Ma for TN01 and 307.5 ± 2.0 Ma for TN09; Figs. 3, 5), interpreted as the age of crystallization of the Velay granite at Tournon. These ages are consistent with other geochronological data obtained from the Velay granites-migmatites, all yielding ages in the range 300-310 Ma (Couzinié et al., 2014;Didier et al., 2013;Laurent et al., 2017;Mougeot et al., 1997). In the LGU migmatitic paragneiss TN17, the age of the migmatization is well recorded by zircon rims on older cores or entire grains (307.4 ± 2.3 Ma) and overlap within uncertainty with the age of the Velay granite as dated on TN01 and TN09 (Fig. 5). This event is, in contrast, poorly recorded in sample TN46 (only 2 concordant analyses around 310 Ma), which is consistent with its apparent lower degree of partial melting.
In the LGU, the few concordant analyses with dates ranging from ca. 310 to 540 Ma most probably represent lead loss of the ≥540 Ma-old zircons (Fig. 5) during Variscan events (which would still result in concordant analysis within the analytical precision). This

A C C E P T E D M A N U S C R I P T
interpretation is supported by the fact that these zircons were exclusively found in the two Velay granite samples (TN01 and TN09), where partial resetting of the U-Pb system in inherited zircons is likely to have taken place owing to dissolution/recrystallization, whereas zircons dates from the two LGU paragneisses samples (TN17 and TN46) clearly show a complete gap between ca. 550 and 320 Ma (Fig. 5).

Granite dikes
All four granite dikes samples show a group of Carboniferous zircon rims and entire oscillatory-zoned grains that have been used to calculate U-Pb dates ranging from 323.3 ± 3.5 Ma (TN14) to 307.9 ± 3.3 Ma (TN32; Figs. 3, 6). These dates are fully consistent with crosscutting relationships and furthermore overlap with ages recently obtained on Variscan granitoids in the eastern FMC (340-300 Ma; Laurent et al., 2017). Therefore, these dates are considered to represent the emplacement ages of the dikes.
All samples further show a range of older zircon, mainly >470 Ma with scattered ɛ Hf t from -18 to +12 and predominantly obtained from zircon cores (Fig. 8), which would correspond to either inherited grains from the source, or zircon sampled from wall rock during magma ascent and emplacement. Few grains dated at ca. 480 Ma display positive ɛ Hf t comparable to those from the crosscut amphibolite (sample TN13) while others with ɛ Hf t between 0 and -5 are akin to those from the UGU gneisses (Fig. 8) LGU (Figs. 7, 8). These elements suggest that inherited and/or xenocrystic zircons from the granite dikes are of various magmatic and detrital origins and have been sourced from both the UGU and the LGU. This pattern of ages and isotopic compositions of inherited zircons is

A C C E P T E D M A N U S C R I P T
also notably similar to that observed from other Variscan granitoids in the area (Couzinié et al., 2014;Laurent et al., 2017;Moyen et al., 2016).
This characteristically North Gondwanan signature is obvious in the repartition of zircon ages obtained from the Tournon area (Fig. 9). The Stenian-Tonian ages (ca. 1050-950 Ma) can be attributed to the assembly of the Rodinia supercontinent during the Grenvillian orogenies (Bradley, 2011;Rino et al., 2008), while ages of 700 to 550 Ma correspond to the assembly of Gondwana during several (Pan-African) collisional orogenies, and by the formation of an Andean-type orogenic belt along its northern margin during the Ediacaran,

A C C E P T E D M A N U S C R I P T
the Cadomian-Avalonian belt (Bradley, 2011;Linnemann et al., 2008;Nance and Murphy, 1994;Rino et al., 2008;Veevers, 2004). Extensional rifting of the northern Gondwana margin during the early Paleozoic is recorded by the ca. 480 Ma age peak. This event resulted in the opening of the Rheic Ocean and probably other smaller oceanic domains (e.g., Galicia-Massif Central Ocean(s)), and the drift of the Avalonian continental ribbon away from the main Gondwana supercontinent (e.g., Gerdes and Zeh, 2006;Linnemann et al., 2008;Nance et al., 2010;Ballèvre et al., 2014). Finally, the closure of these oceans from the Devonian to the Carboniferous led to the formation of the supercontinent Pangea during the Variscan orogeny (e.g., Kroner and Romer, 2013;Matte, 2001, Nance et al., 2010, as is partly recorded by the ca. 360-300 Ma date cluster in the Tournon area (Fig. 9).
Although Non-metric multi-dimensional scaling analysis ( Fig. 12b; Vermeesch, 2013) show that Israeli and Moroccan zircons are the closest neighbors to those from the EMC. Additional visual comparison of the age spectra (Fig. 12a) suggests that the zircon population from Tournon is much more akin to the one of the Israeli sediments because:  sediments from Morocco lack the Meso-Neoarchean zircon age peak seen in Tournon  (Fig. 10), similar to that of exposed magmatic rocks from the Eburnean/Birimian belts of the West African craton (Block et al., 2016) but not observed in the Tournon (meta)sediments, in which the Paleoproterozoic zircons have negative εHf t (Fig. 10). In contrast, Paleoproterozoic zircons from the Israeli sediments have negative εHf t similar to those from Tournon (Morag et al., 2011a), which clearly suggests that pre-Cadomian detritus of the EMC were not derived

A C C E P T E D M A N U S C R I P T
from the West African craton, but rather from the same massifs that sourced the Israeli sediments. Most Neoproterozoic (800-650 Ma) detrital zircons from the EMC show superchondritic εHf t values overlapping with those from Neoproterozoic clastic and magmatic rocks from the Arabian-Nubian Shield ( Fig. 10; Morag et al., 2011b;Avigad et al., 2015). However, the Arabian-Nubian Shield cannot be considered as the only source of detritus for the EMC Ediacarian sediments, because 67% of the 580-650 Ma detrital zircons from the latter have clearly less radiogenic Hf isotopic compositions (ɛ Hf t from 0 to -28; Fig. 10). This indicates that the Tournon detritus were also fed by exposed granitoid rocks

A C C E P T E D M A N U S C R I P T
The age spectra and Hf isotopic data of zircons in the samples from the LGU and UGU show characteristic differences, which point to a different timing of pre-Variscan formation (Figs. The late Ediacaran magmatism strikingly dominates the inherited zircon record of the Velay granites TN01 and TN09 (Fig. 5) and is also most likely represented by the migmatitic augen orthogneiss present NW of the study area. Similar orthogneisses, locally migmatitic and exposed all around the Velay dome have been dated around the Ediacaran-Cambrian boundary (ca. 530-550 Ma) which confirms that they represent a widespread, self-consistent unit (Be Mezeme et al., 2006;Caen-Vachette, 1979;Mintrone, 2015;Montel et al., 1992;R'Kha Charam et al., 1993;Weisbrod et al., 1980). Elsewhere in the western and southern sides of the Massif Central (Limousin, Montagne Noire), various orthogneisses yielded similar Ediacaran to Cambrian ages (Alexandre, 2007;Alexandrov et al., 2001;Duthou et al., 1984 andreferences therein, Lafon, 1984;Lévêque, 1985;Melleton et al., 2010).

A C C E P T E D M A N U S C R I P T
The geodynamic significance of this important late Ediacaran to early Cambrian granitoid formation remains poorly constrained in the EMC. It is generally accepted that during the late Neoproterozoic exotic terranes were accreted to the North Gondwana margin to form the Cadomian domain (see Garfunkel, 2015 and references therein). In the European Variscides, most of the Cadomian outcrops actually record only the latest Ediacaran period and are made of a thick siliciclastic marine sedimentary pile derived from both the Gondwana hinterland and from local magmatic sources (e.g., a Cadomian magmatic arc; Garfunkel, 2015). Only few segments of the Cadomian domain keep record of a late Ediacaran magmatic arc and orogenic event (Garfunkel, 2015). In turn, the Ediacaran record is dominated by wide basinal areas that were intruded by voluminous granitoids and usually record limited subcontemporaneous deformation (Abbo et al., 2015;Garfunkel, 2015 and references therein). In many places, of which domains now adjacent to the EMC such as the Maures Massif and the eastern Pyrenees, the lack of a significant Cadomian-aged deformation, and the presence of bimodal magmatism with mantle (suprasubduction tholeiitic to calc-alkaline metabasite) and crustal (peraluminous metavolcanics/metagranitoids) affinities dated at ca. 560-540 Ma suggest that a large part of the Cadomian domain was undergoing crustal stretching (e.g., Bellot et al., 2010;Castiñeiras et al., 2008;Garfunkel, 2015;Innocent et al., 2003).
Collectively these data are consistent with the emplacement of a widespread ca. 550 Ma-old granitic pluton or batholith within a large, late Ediacaran sedimentary sequence and is now mostly involved into Velay dome. It possibly originated from crustal melting triggered by the incipient inversion of an Ediacaran back-arc basin (Mintrone, 2015). Nevertheless, the relation of these late-Ediacaran granite intrusions in the EMC to a possible subduction zone along the northern border of the same crustal block remains unclear and structural evidence for basin inversion has not been described thus far. In any case, it must be noted that these late Ediacaran meta-sedimentary and meta-igneous units do represent the oldest protoliths

A C C E P T E D M A N U S C R I P T
identified within the EMC to date. Indeed, all zircons older than ca. 550 Ma in the Tournon samples arguably correspond to detrital zircons within paragneisses of the UGU and LGU (see §6).
The time interval between ca. 540-490 Ma is marked by an apparent gap in the zircon record of the Tournon area (Fig. 9), unlike in the adjacent massifs (e.g., the Alpine massifs  ). We propose that during the Cambrian, a stable continental platform was established in at least part of the future EMC and that continued tectonics and related magmatic activity (e.g., von Raumer and Stampfli, 2008) was rather focused in peripheral areas.
After this magmatic lull, zircons from the UGU testify that enhanced thermal regime and bimodal magmatic activity resumed around 480 Ma (Figs. 4, 8), which is the age of mafic magma emplacement (protolith of the amphibolite TN13) and of partial melting in the Ediacaran (meta)sediments (recorded in the UGU gneisses). This bimodal magmatism is best explained in a context of a thermal event caused by crustal extension, thinning and eventually rifting.
Throughout Europe, such a Lower Ordovician bimodal magmatism is very well represented. A wealth of evidence shows that during this period, the Avalonian terranes drifted from northwest Gondwana, which caused the diachronic west to east opening of the Rheic ocean (e.g., Linnemann et al., 2008;Nance et al., 2010;von Raumer and Stampfli, 2008 von Raumer et al., 2013;Villaseca et al., 2015). Despite some disagreement about the source and geodynamic implication of such magmatism, it is generally agreed that at this period the northern Gondwana margin was under widespread extensional tectonics that locally led to the formation of slow-spreading oceanic crust, distinct from that of the Rheic Ocean (Berger et al., 2006;Bouchardon et al., 1989;Briand et al., 1991;Díez Fernández et al., 2012;Matte, 2001;Montes et al., 2010;Pin and Marini, 1993;von Raumer and Stampfli, 2008). In the Massif Central, this "oceanic" domain is often referred to as the Galicia-Brittany

A C C E P T E D M A N U S C R I P T
Ocean (Matte, 2001) or Massif Central-Moldanubian Ocean (Tait et al., 1997). However, the continuity of the benthic faunas and paleomagnetic data preclude the existence of a large ocean between the Massif Central and Gondwana during the Ordovician and the Silurian (Cocks and Torsvik, 2002;Fortey and Cocks, 2003;Paris and Robardet, 1990;Shaw and Johnston, 2016;Tait et al., 2000). Accordingly, an Ocean-Continent transition (Lardeaux et al., 2014) or an hyperextended north Gondwana margin would represent more realistic settings for the Ordovician period.

Timeframe of Variscan events in the EMC
The new U-Pb and Lu-Hf isotopic data from zircons of the Tournon area record several Carboniferous metamorphic and magmatic events and thus place new constraints on the Variscan evolution of the EMC.
The data indicate that zircon grains/overgrowths in the banded gneiss (TN12) and the amphibolite (TN13) from the UGU (Figs. 7, 8) were formed during a migmatitic-metamorphic event at around 350-340 Ma (see §6.1 and Fig. 11). The lack of HP assemblage within the migmatitic portions of sample TN12 and TN13 shows that these ages unlikely reflect the timing of HT-HP metamorphism of the UGU (up to ca. 800°C, 15 kbar; Gardien, 1993).

A C C E P T E D M A N U S C R I P T
Instead we argue that these correspond to incipient anatexis and crystallization of the nappe on the retrograde path to amphibolite facies conditions.
Our early Carboniferous metamorphic ages for the LAC of Tournon are similar to a zircon ID-TIMS U-Pb age 345 ± 10 Ma in the Marvejols area (south Massif Central) where a bimodal series resembling that of Tournon records retrograde (after HP) amphibolite facies metamorphism (Pin and Lancelot, 1982). Our new data further suggest that in the southern part of the Massif Central the retrogression of the UGU to amphibolite facies conditions (Gardien, 1993) in association with incipient partial melting occurred significantly later (ca. 350-340 Ma) that in the northern part (389-375 Ma; Boutin and Montigny, 1993;Costa and Maluski, 1988;Duthou et al., 1994). This implies that rocks forming the UGU in the southern Massif Central were buried and exhumed later than in the northern Massif Central and/or that after syn-orogenic exhumation to mid-crustal depth, they remained partially molten for a longer time period.
The base of the UGU was later affected by intense magmatic activity with several generations of granite dikes emplaced from 323.3 ± 3.5 Ma (TN14) to 307.9 ± 3.3 Ma (TN32; Fig. 6). This timing of dike intrusion overlaps with the emplacement of numerous peri-Velay granite plutons, including the nearby Tournon granite (Fig. 2)

A C C E P T E D M A N U S C R I P T
the UGU nappe above the LGU must have been achieved by ca. 325 Ma, that is, prior to the emplacement of the oldest dikes.
The youngest dike (TN32; 307.9 ± 3.3 Ma) is concomitant with the emplacement of the Velay granites, dated at 309.6 ± 2.9 Ma (TN01) and 307.5 ± 2.0 Ma (TN09) in Tournon and in the 310-300 Ma time span elsewhere in the EMC (Couzinié et al., 2014;Didier et al., 2013;Laurent et al., 2017;Mougeot et al., 1997). Overall, data from the dikes argue for the presence of a 15.4 ± 4.8 Ma-long lasting thermal anomaly in the Tournon area, accompanied with protracted crustal partial melting and granite dike intrusion. The exhumation of the Velay dome marks the end of this high crustal thermal regime and suggests that the existence of a long-lived molten zone in the middle crust possibly controlled dome extrusion and gravitational collapse Vanderhaeghe, 2009;Vanderhaeghe and Teyssier, 2001). The nearly continuous intrusion of high-K mantle-derived magmas between 335 and 300 Ma in the EMC, provides a clue that protracted crustal melting was concomitant with asthenospheric mantle upwelling caused by late-orogenic lithosphere delamination (Laurent et al., 2017).

Constraints on crustal growth and recycling
The U-Pb vs εHf t data from the Tournon area ( Fig. 10) suggest that the EMC crust result from the reworking of several juvenile crustal components throughout the geological history.
Below we describe the age and origin of these crustal components and their respective reworking associated with each zircon date group from the Tournon area.

A C C E P T E D M A N U S C R I P T
crust has been extracted from a depleted mantle source at 3.4-3.1 Ga. These model ages are comparable to those obtained during previous studies from basement rocks of the West African and Congo cratons (Eglinger et al., 2016;Batumike et al., 2009;Potrel et al., 1996;1998;Tchameni et al., 2001; Fig. 10). Variable εHf t values (from 0 to -10) of some Paleoproterozoic zircons (2.1-1.9 Ga) may be explained by the interactions between Archean crust (3.0-2.6 Ga) and juvenile magmas during the "Birimian" or "Eburnean" orogenies. This in Central Germany (Zeh et al., 2001;Gerdes and Zeh, 2006;Linnemann et al. 2008Linnemann et al. , 2014 and from the Iberian Massif, Spain and Portugal Teixeira et al., 2011).
Most zircons from Tournon that crystallized in the 800-650 Ma time span show positive εHf t up to +11 (81%), thus pointing to a period of significant juvenile crust formation ( Fig. 10). Similar age-Hf isotope signature are consistently reported from basement rocks of the Arabian-Nubian Shield ( Fig. 10; Morag et al., 2011b), and from some Gondwana-derived sediments in Europe Linnemann et al., 2008;Orejana et al., 2015). This

A C C E P T E D M A N U S C R I P T
observation together with the Ediacaran paleogeographic position of the EMC (see §7.1), suggests that a crustal component similar in age and Hf isotopic composition to the rocks of the Arabian-Nubian Shield may have contributed to build the EMC crust.
Alternatively, this large spread of εHf t may be explained as a sedimentary mixture (deposited during the late Neoproterozoic along the northern Gondwana margin) of which detritus were derived from different crustal regions of the Cadomian/Pan-African orogens, in which magmatism was characterized by various proportions of respectively Archean/Paleoproterozoic, and juvenile Neoproterozoic crustal sources (Morag et al., 2011a).
This option is preferred for the EMC as there is so far no compelling evidence for autochthonous crust older than ca. 550 Ma in the EMC (Mintrone, 2015; this study), further supporting the distal origin and the detrital nature of these zircons. Nevertheless, this does not rule out the possible co-existence of Paleoproterozoic crust and Neoproterozoic subductionrelated magmatism, similar to that reported is some areas of the Cadomian orogen (e.g., in northern Brittany: Ballèvre et al., 2001;Tauride block: Abbo et al., 2015).
Many zircons of the ca. 550 Ma-old population show a very limited spread in εHf t from 0 to -5, especially those from the Velay granites in the LGU, which we interpret as inherited from a Cadomian (meta)granitic batholith (see §6.2 and Mintrone, 2015; R'Kha

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T Chaham et al., 1990). The slightly sub-chondritic Hf isotope signature of zircons from these magmas is rather consistent with a source represented by a relatively homogeneous mixture of juvenile Neoproterozoic and Archean to Paleoproterozoic crustal material. Such a source is possibly represented by the UGU and LGU paragneisses which, as discussed earlier, were fed by mixed detritus from these two crustal components (Figs. 8, 10).
Apart from a minor juvenile input recorded by the emplacement of the amphibolite Nd isotope data from the FMC granitoids Pin and Duthou, 1990;Turpin et al., 1990). Such model ages clearly have no geological significance as zircon in all Gondwana derived sediments throughout Europe provide no evidence for magmatism during this period (Balintoni et al., 2014;Díez Fernández et al., 2012;Drost et al., 2011;Gebauer et al., 1989;Gerdes and Zeh, 2006;Linnemann et al., 2014;Morag et al., 2011a;Pastor-Galán et al., 2013;Pereira et al., 2012a;2012b;Shaw et al., 2014;Sirevaag et al., 2016;Villaseca et al., 2016;Williams et al., 2012;Zeh and Gerdes, 2010). Instead, this Hf isotope signature would correspond again to a mixed source consisting of Neoproterozoic Neoproterozoic (0.6-0.8 Ga); and (iii) that the crust of the EMC, and possibly large portions of the Western European crust as well, derive from a mixture of these three components as a result of tectono-magmatic and sedimentary processes that occurred during the Cadomian/Pan-African orogenies.

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T Ga (e.g., Kemp et al., 2006). DM field is bounded by the models of Griffin et al. (2002)   Many 380-340 Ma-old zircon rims and grains from TN13 show significantly lower 176 Hf/ 177 Hf

A C C E P T E D M A N U S C R I P T
ratios than what is predicted from zircons recrystallization (black dashed arrow), indicating that their formation requires the input of externally-derived, non-radiogenic Hf, most likely derived from the surrounding gneisses (represented by TN12). (B) non-metric multi-dimensional scaling plot of the north Gondwana and EMC dataset (Vermeesch, 2013). Data have been compiled by Meinhold et al. (2013)