A multi-isotope study (Fe, Ge, O) of hydrothermal alteration in the Limousin ophiolite (French Massif Central)

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.


Introduction
Ophiolite-derived rocks composed of ultrabasic (peridotites, serpentinites) and basic rocks (gabbros, basalts) are remnants of ancient oceanic lithosphere that were accreted in convergent settings by subduction/exhumation or obduction processes. Fluid-rock interactions in subducted plates and ophiolites may be sources of significant mass transfer between the oceanic lithosphere and the fluid and volatile phases. The oceanic lithosphere undergoes various stages of hydrothermal alteration (or hydrothermal metamorphism) from its formation along mid-oceanic ridge to its accretion in convergent settings (e.g. Beinlich et al., 2010;Cartwright and Barnicoat, 1999;Muehlenbachs, 1998): 1) High-temperature (high-T; T > 350°C up to late magmatic conditions) hydrothermally alteration close to spreading ridges and black smokers is responsible for Ca enrichment and Mg depletion in basic rocks, as well as for a decrease of the δ 18 O compared to mid-ocean ridge basalts (MORB; δ 18 O = +5.8 ± 0.3‰; Muehlenbachs, 1998, and references therein). Sulphides may precipitate on the ocean floor in black-smoker-type deposits, while gabbros are altered into epidosites under black-smokers; 2) Low-temperature (low-T; T from <100°C to 350°C) alteration in the upper part of the oceanic lithosphere is responsible for Mg enrichment, Ca depletion, and for a δ 18 O increase compared to MORB. The δ 18 O decreases with depth, i.e. with decreasing the amount of percolating fluids in the deeper parts of the ocean crust (gabbros); 3) Fluid-rock interactions during devolatilisation reactions related to subduction zone metamorphism if the oceanic lithosphere is subducted; 4) Fluid-rock interactions related to low-to medium-pressure metamorphism during obduction/exhumation of ophiolites related to orogenic processes; and 5) Migration of late-to post-orogenic fluids.
Studies on stable Fe and Ge isotopes showed that they can fractionate at both low-T and high-T magmatic and hydrothermal conditions and can be employed as tracers of magmatic, metamorphic and hydrothermal processes (for a review, see Sossi et al., 2016;El Korh et al., 2017a;Rouxel and Luais, 2017;Johnson et al., 2020).
During hydrothermal metamorphism of the oceanic lithosphere, significant changes in redox conditions are associated with dissolution and precipitation of Fe-bearing minerals on (sub-)seafloor. Because of the higher solubility of Fe 2+ compared to Fe 3+ in hydrous fluids, hydrothermal processes under reducing conditions produce fluids enriched in light-Fe isotopes, as well as high-T basalt-hosted vent fluids from midocean ridges with a δ 56 Fe lower of c. −0.2 to −0.7‰ than igneous rocks Rouxel et al., 2008). Fe is mainly transported as Fe-chloride complexes in fluids (Manning, 2004). Changes in the non-redox parameters [such as ligand composition (chloride, sulphide), speciation, Fe coordination] is also expected to trigger Fe isotope fractionation in Fe-bearing solutions (Hill et al., 2010).
Germanium speciation in fluids is relatively similar to that of aqueous Si (Wood and Samson, 2006). Because of the lower enthalpy and heat capacity of Si(OH) 4 (aq) species compared to Ge(OH) 4 (aq) species, the Ge/Si ratio of fluids in equilibrium with Ge-rich silicates increases with temperature (Pokrovski and Schott, 1998). Ge shares a strong affinity with Fe hydroxides and can co-precipitate with Fe during Fe 2+ oxidation or Fe 3+ hydrolysis (Pokrovsky et al., 2014). Thus, Ge behaviour varies according to the physico-chemical conditions of hydrothermal alteration of the basaltic crust: 1) Ge depletion during high-T alteration along hydrothermal vents if hydrothermal sulphides precipitate or; 2) Ge enrichment due to Ge adsorption by iron hydroxides (Escoube et al., 2015;Pokrovsky et al., 2014).
Recent studies have shown that the subducted oceanic crust can conserve the Fe and Ge isotopic composition of their hydrothermally altered basic protolith during high-pressure (HP) metamorphism (Beard and Johnson, 2004;Li et al., 2016;El Korh et al., 2017a, 2017bInglis et al., 2017), while Fe isotope fractionation may occur in HP blueschist-facies serpentinites (Debret et al., 2016). Deciphering the signatures of fluid-rock interactions in ophiolites can be relatively complex because these rocks often display various stages of fluid-rock interactions, fluid overprinting or polymetamorphism, especially in the case of subducted ophiolites. This study investigates nontraditional Fe and Ge isotope fractionation in ultrabasic (serpentinites) and basic rocks (amphibolite facies metagabbros) of the Limousin ophiolite (French Massif Central) that were not subducted during the Variscan orogeny. We aim to characterise the isotopic signatures of (sub-)seafloor hydrothermal alteration vs. magmatic processes and to understand processes controlling isotope fractionation during hydrothermal metamorphism in ancient non-subducted hydrothermally altered oceanic rocks.

Geological setting
The Limousin area is located in the northwestern French Massif Central. It is part of the European Variscan belt, which was formed by the Devonian to Carboniferous collision of Laurussia (formed by the assembly of Laurentia, Baltica and Avalonia continents) and Gondwana continents ( Fig. 1a) (e.g. Lardeaux et al., 2014;Matte, 2001). Two main collision stages have been characterised in the Western Variscan belt. The Variscan orogeny was initiated during the Middle Devonian by the closure of the oceanic domains (including the Rheic ocean), which were opened during the Cambro-Ordovician rifting between the continental domains, and by the dislocation of the northern margin of the Gondwana (e.g.; Kroner and Romer, 2013;von Raumer et al., 2015). The collision between Laurussia and Gondwana-derived terranes occurred during the Early Carboniferous and was followed by Late Carboniferous orogenic collapse (Kroner and Romer, 2013;Lardeaux et al., 2014;Franke et al., 2017).
The western European Variscan belt is composed of three major domains ( Fig. 1a): (1) the Rheno-Hercynian Zone (external domain), (2) the Saxothuringian Zone and, (3) the Moldanubian Zone (internal allochtonous domain) (e.g. Franke et al., 2017;Lardeaux et al., 2014). The allochthonous domain was formed by the superposition of nappes derived from peri-Gondwanan regions and includes a series of ophiolites and ophiolite-derived basic and ultrabasic rocks. (von Raumer et al., 2015, and references therein). In the Western Variscan belt, the emplacement of the ophiolite magmatic precursors along the Gondwana margin have been interpreted as the result of a late-Cambrian active margin (intra-continental back-arc basin rifting) that has followed the closure of the Proto-Rheic ocean during the Cambrian-Ordovician (von Raumer et al., 2015). Other studies consider the ophiolites as the remnants of a narrow ocean between Gondwana and Armorica named the "Galicia-South Brittany-Moldanubian" or "Medio-European" ocean (Matte, 2001;Faure et al., 2009;Lardeaux et al., 2014).
The FMC consists of a series of nappes, which were piled during the Devonian-Early Carboniferous, and display different units (Girardeau et al., 1986;Ledru et al., 1994;Faure et al., 2009). In the Limousin area, the Upper Allochthon (or Gartempe Unit) is formed by a series of low-grade Palaeozoic metasedimentary and metavolcanic associations (Fig. 1b). The Middle Allochthon (also known as the Upper Gneiss Unit), includes rocks from the "leptyno-amphibolite groups" (LAGs) (Santallier et al., 1988): paragneisses, leptynites and amphibolites of medium to high grade metamorphism, as well as migmatitic metagreywackes and relicts of eclogites and granulites. The Lower Allochthon (also known as the Lower Gneiss Unit) is mainly composed of metasedimentary rocks (paragneisses, micaschists, metashales and metagreywackes), as well as Late Proterozoic-Early Cambrian and Ordovician leucocratic orthogneisses. The Parautochthon basement is composed of metasediments and metagranites.
Remnants of subducted ophiolites occur as lenses of HP-UHP eclogites (zoisite-eclogites and kyanite-eclogites; Berger et al., 2010), which crop out at the basis of the Middle Allochthon. U-Pb age data indicate that eclogites have recorded a protolith age of 475-489 Ma (zircon), indicating that their protoliths were emplaced during the Cambro-Ordovician rifting (Berger et al., 2010). The UHP event (P~2.9 ± 0.5 GPa, T~660 ± 70°C) corresponds to a subduction at a depth of 100 km and is dated at 412 ± 5 Ma (Berger et al., 2010).
The Limousin ophiolite belongs to the suite of oceanic rocks recognised in the Moldanubian Zone (von Raumer et al., 2015). It corresponds to a series of 1-5 km wide non-subducted ophiolite massifs, forming a 25 km long thrust sheet of basic and ultrabasic rocks in the  Berger et al., 2005, and El Korh et al., 2019 upper part of the Middle Allochthon (Dubuisson et al., 1989);Berger et al., 2005Berger et al., , 2006. The Limousin ophiolite has been interpreted as the plutonic sequence of a lherzolite-harzburgite ophiolite type, emplaced in a slow-spreading mid-ocean ridge (Berger et al., 2006). The ophiolite bodies are composed of a series of serpentinised ultrabasic rocks and basic amphibolite-facies rocks, such as diopside-bearing harzburgites, harzburgites, dunites, wehrlites, harzburgites, troctolites, (meta)gabbros and amphibolites (Berger et al., 2005). Further to magma emplacement, the rocks were submitted to a pervasive seafloor hydrothermal alteration under low-P conditions (~0.2 GPa), with temperature decreasing from high-T late-magmatic conditions to lower-T greenschist-zeolite metamorphic facies (Berger et al., 2005). The Limousin ophiolite-derived rocks have not been affected by a pervasive Variscan orogenic metamorphism, which probably Table 1 Description, provenance and mineral assemblage of the studied samples. Korh et al. (2019. ⁎ See Fig. 1. ⁎⁎⁎ The mineral abbreviations are from Kretz (1983); Fe sulph: Fe-sulphides (pyrite). took place before nappe emplacement and accretion of ophiolitederived rocks (Berger et al., 2005).
Both studied amphibolite types display compatible and incompatible trace elements (including REE and HFSE and transition metals) typical of MORB (Sun and McDonough, 1989), with (Ce/Yb) PM ratio of 0.60 to 0.66. Only the LILE, Li, Pb and Sr abundances are slightly higher than N-MORB, which can suggest mobile element enrichment during hydrothermal alteration. Actually, amphibolites have Li abundances and δ 7 Li typical of hydrothermally altered sheeted dykes and gabbros that have interacted with heavy-Li fluids (seawater or upwelling hydrothermal fluids) under high-T conditions (El Korh et al., 2019). Equilibration between amphibole and plagioclase in amphibolites has occurred between c. 570 and 750°C during high-T hydrothermal alteration (Berger et al., 2005).  Korh et al., 2019, that will be discussed below.

O isotopes
The oxygen isotope compositions of whole rocks were determined in the Stable Isotope Laboratory of the Institute of Institute of Earth Surface Dynamics (IDYST), University of Lausanne. Oxygen was extracted from 2 to 3 mg of rock powder using a 10 W New Wave CO 2 laser and F 2 gas. Isotopic composition of extracted oxygen was analysed with a Thermo-Finnigan MAT 253 gas source mass spectrometer. δ 18 O values are expressed in ‰ relative to VSMOW (Vienna Standard Mean Ocean Water) and are corrected to the LS_1 quartz standard (in-house standard of the University of Lausanne; 18.1‰). The LS_1 standard was analysed to monitor data accuracy and reproducibility (daily average value: +18.02 ± 0.322‰; n = 2) ( Table 3).

Fe isotopes
The whole rock Fe isotope compositions were measured in liquid mode by multi-collector inductively coupled plasma mass spectrometry (MC-ICP-MS) at the CRPG-Nancy using a NeptunePlus spectrometer (ThermoFisher Scientific, Germany and USA), following the procedure  described in Liu et al. (2014) and El Korh et al. (2017a). Fe was separated according to the following chemical procedure: c.10 mg of powdered samples were dissolved in three steps using: 1) a 2:1 mixture of HF (28 N) and HNO 3 (15 N) on a hot plate at 90°C; 2) HNO 3 (15 N) at 60°C; and 3) HCl (7 N) at 60°C. After sample centrifugation, Fe was isolated through AG-MP-1 resin-exchange chromatography columns. Fractions were eluted with HCl (2 N) and dried down. Fe recovery is 88-99%. The dry fractions were then dissolved in HNO 3 (7 N) and dried down, before being re-dissolved in HNO 3 (0.05 N) for MC-ICP-MS analyses. The MC-ICP-MS instrument was equipped with a standard nebuliser and cyclonic chamber, as well as Ni skimmer and sampler cones in standard geometrical configuration. In order to solve argide polyatomic interferences on Fe masses ( 40 Ar 14 N on 54 Fe; 40 Ar 16 O on 56 Fe; 40 Ar 16 O 1 H on 57 Fe), analyses were performed in static mode at a high resolution (M/ΔM ≈ 8000) on the left side (low mass) of the Fe peaks. The cup configuration consisted of: Low 3 ( 53 Cr), Low 2 ( 54 Fe), Axial ( 56 Fe) and High 1 ( 57 Fe). 53 Cr was used for the correction of the isobaric interference between 54 Cr and 54 Fe intensities. Gas flow rates, torch parameters and ion lenses were optimised by measurement of the IRMM-014 ultrapure Fe standard. Prior to optimising the peak shapes and centring, the correction coefficients between the Faraday cups were determined by gain calibration. Analyses were carried out on 2-5 ppm Fe solutions, diluted in HNO 3 (0.05 N). Analytical sequences consisted of 50 cycles (10 min in total), with 8 s integration time, after 13 min washout in HNO 3 (0.5 N) and 8 min washout in HNO 3 (0.05 N). Each sample analysis was sequentially bracketed by a measurement of the IRMM-014 standard. Fe isotope compositions are reported in ‰ standard delta values (δ 56 Fe and δ 57 Fe), relative to the IRMM-014 standard.
Data bulk external reproducibility and accuracy were estimated by replicate analyses of the PCC-1 peridotite (Austin Creek, California, USA) geostandard (USGS) which was submitted to the same chemical and analytical procedure as the samples. The mean δ 56 Fe and δ 57 Fe values are within uncertainty with the data reported in the literature (Table 4).

Ge isotopes
The whole rock Ge isotope ratios were measured using the Neptune Plus spectrometer MC-ICP-MS at the CRPG-Nancy (Table 5). Ge dissolution and separation followed the chemical procedure given in Luais (2012). The powdered samples (c. 300 mg) were dissolved in a 3:1 mixture of HF (28 N) and HNO 3 (15 N) at 60-65°C, because of the strong Ge volatility. Several steps of leaching with HF and centrifugation allowed Ge recovery in the supernatant fraction (Luais 2012). Ge was isolated using two types of resin-exchange chromatography columns: 1) a AG 1-X8 anion exchange resin (chloride form, 200-400 mesh, 2 ml) with HF (1 N), where the Ge fraction was collected with HNO 3 (0.2 N); 2) an AG 50 W-X8 (hydrogen form, 200-400 mesh, 2 ml cationicexchange resin, where Ge was eluted with HNO 3 (0.5 N) (Ge recovery: 100%) (Luais 2012). Ge isotope analyses were carried out on 10 ppb diluted Ge standards and samples in HNO 3 (0.01 N). Standards and samples were doped with the NBS SRM 994 Ga reference international isotopic standard ( 69 Ga/ 71 Ga = 1.50676; Machlan et al., 1986), in a 10:1 ratio to monitor mass bias accuracy and instrumental drift during the analytical session.
The MC-ICP-MS spectrometer was equipped with a hydride generator introduction system (HGIS) to increase the sensitivity (Florin et al., 2020). Samples are mixed with a high-reducing solution of NaBH 4 -NaOH in excess, which allows conversion of volatile aqueous species (GeOH 4 ) to gaseous (GeH 4 ) hydride species (Dedina and Tsalev 1995), with a yield of 100%. Isobaric interferences that do not form hydrides (such as argides, Zn, NiO, and FeO) are thus neutralised. Analyses were performed in static mode at a low resolution (M/ΔM = 400), with Ni skimmer and sampler cones placed in standard geometrical configuration. The cup configuration consisted of: Low 3 ( 68 Zn), Low 2 ( 69 Ga), Low 1 ( 70 Ge), Axial ( 71 Ga), High 1 ( 72 Ge), High 2 ( 73 Ge), High 3 ( 74 Ge). 68 Zn was used to correct the isobaric interference between 70 Ge and 70 Zn mathematically. Gas flow rates, torch parameters and ion lenses were optimised by measurement of the NIST SRM 3120a ultra-pure Ge standard (Luais 2012). The correction coefficients between the Faraday cups were calculated by a gain calibration before optimising peak shapes.
Analyses were carried out on 2-4 ppm Fe solutions, diluted in HNO 3 (0.05 N). Analytical sequences consisted of 60 integration cycles (9 min in total), after 200 s washout in HNO 3 (0.6 N) and 8 + 2 min washout in HNO 3 (0.01 N). Each sample measurement was sequentially bracketed by a measurement of the NIST SRM 3120a standard. Ge isotope compositions are expressed in ‰ standard delta values (δ 72 Ge, δ 73 Ge and δ 74 Ge), relative to the NIST SRM 3120a standard. The long-term stability of the ICP mass spectrometer, the external reproducibility and accuracy of the measurements were estimated by replicate analyses of the JMC and Aldrich Ge solution standards. The average δ 72 Ge, δ 73 Ge and δ 74 Ge values (Table 5) are consistent with their respective reference values (Escoube et al., 2012;Luais 2012).

O isotopes
O isotope compositions of the studied serpentinites, amphibolites and UHP eclogite samples are given in Table 3 (Fig. 3).

Fe isotopes
The Fe isotopic compositions of the studied serpentinites, amphibolites and UHP eclogite are given in Table 4. The reproducibility on δ 56 Fe for all samples is generally lower than ±0.043‰ at 2σ SE. All data plot within uncertainty on the slope of the equilibrium and kinetic theoretical mass fractionation lines in a δ 57 Fe vs. δ 56 Fe diagram (Fig. 4a). The δ 56 Fe values of serpentinites vary between +0.14 and +0.18‰ (Fig. 4a). Amphibolites have δ 56 Fe values varying from +0.03 to +0.16‰, whithin the same range as serpentinites. The two groups of amphibolite facies rocks can be distinguished, even if their δ 56 Fe ranges In panels d and f, the correlation coefficient R 2 is given for the whole series of amphibolites, as well as for the three samples that correlate well with the δ 56 Fe (excluding sample CLUZ1a). Error bars are at 2σ SE external reproducibility.
In amphibolites, the δ 56 Fe values do not correlate with SiO 2 , Ni, Cr, and Ge contents, nor with Al 2 O 3 /SiO 2 , Fe 3+ /ΣFe and (Ce/Yb) PM values (Figs. 5, 6 and 7). The δ 56 Fe shows a good negative correlation with the Fe 2 O 3 tot content only in 3 samples (Fig. 5c), while the Fe 3+ /ΣFe only varies slightly with the Fe 2 O 3 tot content (Fig. 6b). The δ 56 Fe also increases with the increase of XMg and Y/Ti (Figs. 5 and 7), and with the decrease of Na 2 O, Ba and Rb (Figs. 5f and 7e-f). While there is no clear correlation between δ 56 Fe and CaO values, the highest δ 56 Fe is measured in sample that shows the highest CaO abundance (CLUZ4; Fig. 5e).

Ge isotopes
The Ge isotopic compositions of the studied samples are presented in Table 5. All samples have a reproducibility on δ 74 Ge lower than ± 0.097‰ at 2σ SE. All data plot within uncertainty on the slope of the equilibrium and kinetic theoretical mass fractionation lines in a δ 72 Ge vs. δ 74 Ge diagram (Fig. 4b). Serpentinites have δ 74 Ge values ranging from +0.48 and +0.93‰ that are similar to heavier than the values measured in ultrabasic rocks and MORB (Fig. 8). Amphibolites display a narrow range of δ 74 Ge values varying from +0.72 to +0.78‰, similar within uncertainty to the heaviest values measured in basalts (Fig. 8).
In serpentinites, the δ 74 Ge displays a poor negative correlation with SiO 2 , as well as a good positive correlation with the (Ce/Yb) PM ratio (3 samples out of 5). The δ 74 Ge does not show any correlation with CaO, Fe 2 O 3 tot , Ni, Cr, Ge, concentrations, nor with Al 2 O 3 /SiO 2 , XMg and Fe 3+ / ΣFe ratio (Figs. 8 and 9). Because of the narrow range of δ 74 Ge values in amphibolites, no correlation can observed with major and trace element concentrations (Figs. 8 and 9).

O, Fe and Ge isotopic composition of serpentinites
Oxygen isotope fractionation in the oceanic lithosphere is strongly sensitive to hydrothermal alteration processes which can be distinguished by their distinct signatures (Cartwright and Barnicoat 1999;Muehlenbachs 1998). The oxygen isotopic composition of the Limousin serpentinites is typical of peridotites (+5.5 ± 0.2‰; Mattey et al., 1994) and altered ultrabasic rocks (0 to +6‰; Magaritz and Taylor, 1974) (Fig. 3). As the two ranges overlap, our data do not allow to determine whether O isotopes have fractionated during high-T hydrothermal alteration based on the δ 18 O alone. As serpentinites are strongly oxidised rocks (Fe 3+ /ΣFe from 0.58 to 0.71), the comparison of redox-sensitive isotopes, (such as Fe and Ge isotopes; Polyakov and Mineev 2000;Pokrovsky et al., 2014) with oxygen isotopes are expected to decipher magmatic and hydrothermal signatures in serpentinites.
The δ 74 Ge in serpentinites displays a good negative correlation with the δ 18 O (R 2 = 0.70) and a good positive correlation with the δ 56 Fe (R 2 = 0.70) (Figs. 13a and b), suggesting both Ge and Fe fractionation during hydrothermal alteration at T between 350 and 500°C. However, δ 56 Fe values in serpentinites show a larger deviation from the values typically measured in ultrabasic rocks than δ 74 Ge and δ 18 O values. Hence, oxidising conditions have enhanced fractionation of Fe isotopes to a larger extent than Ge isotopes. A. El Korh, M.-C. Boiron and D. Cividini Lithos 378-379 (2020) 105876 By contrast, the lowest δ 56 Fe and δ 74 Ge values and the highest δ 18 O value were measured in the sample from Le Cluzeau (CLUZ6), whose major element composition differs from the other serpentinites samples (higher CaO and Fe 2 O 3 tot contents and lower MgO abundance). Contrary to the serpentinites from La Flotte and Saint-Laurent that derive from the alteration of abyssal peridotites, sample CLUZ6 probably derives from a troctolite. Thus, its isotopic signature also reflect a difference in the initial protolith composition in addition to hydrothermal alteration.
6.2. O, Fe, and Ge isotope fractionation during high-temperature hydrothermal alteration in amphibolites Amphibolites have δ 18 O values (+6.2 to +6.6‰) that are slighlty higher than MORB, in agreement with the δ 18 O values measured in high-T hydrothermally altered basic rocks (Cartwright and Barnicoat 1999) (Fig. 3). Fe 2 O 3 tot compositions of amphibolites (5.3-8.6%) are typical of oceanic gabbros (2.47-11.1%; Kaczmarek et al., 2008). Similary, amphibolites Fe 3+ /ΣFe ratios (0.11-0.14) typical of MORB (Fe 3+ / ΣFe = 0.07-0.16; Christie et al. 1986;Cottrell and Kelley 2011). In particular, sample CLUZ4, which displays symplectites made of anorthitic plagioclase + hornblende, have conserved a MORB-like δ 56 Fe of +0.158 ± 0.040‰. Symplectites and enrichment in Al and Mg were interpreted as the result of an interaction of preexisting amphibole or clinopyroxene with hot seawater-derived fluids, at temperature conditions of the amphibolite-to-greenschist facies transition rather than an effect of the Variscan orogenic metamorphism (see discussion in Berger et al., 2005). Thus, Fe appears to be relatively immobile during high-T hydrothermal alteration. δ 56 Fe values of amphibolites (+0.03 to +0.16‰) are within the range of MORB, despite a lower δ 56 Fe value measured in one of the metagabbro samples (CLUZ5) (Fig. 10). The δ 56 Fe increase with the Fe 2 O 3 tot decrease and XMg and Y/TiO 2 increase argues that Fe isotopic composition reflects the protolith composition, despite an absence of correlation between the δ 56 Fe and SiO 2 and Al 2 O 3 /SiO 2 contents and between the δ 56 Fe and the fluid-immobile Cr and (Ce/Yb) PM ratio (Figs. 5 and 7). Assuming that Ba concentrations in amphibolites were not significantly modified by hydrothermal alteration, the negative correlation between δ 56 Fe and Ba values (Fig. 7f) may be caused by variations in the protolith composition as well, as Ba is compatible to mildly incompatible in plagioclase. By contrast, the δ 56 Fe decrease with the increase of Na 2 O, and Rb contents and with  However, the correlation between δ 18 O and δ 57 Fe may also reflect intensive mantle metasomatism (Gréau et al., 2011). Contrary to the highly oxidised serpentinites, high-T hydrothermal alteration in amphibolites did not trigger any Fe oxidation. This may be due to the precipitation of reduced Fe 2+ -rich alteration products (sulphides) (Rouxel et al., 2003) or to the low permeability of gabbroic rocks, which prevents hydrothermal fluids to migrate (Cartwright and Barnicoat 1999). Moreover, the various degrees of Fe oxidation states and Fe isotope fractionation between amphibolites and serpentinites can be explained by the different temperature of hydrothermal  alteration, which was lower in serpentinites (350-500°C) than in amphibolites (570-750°C) (Berger et al., 2005).
Amphibolites show a small range of δ 74 Ge values (+0.72 to +0.77‰), that are heavier than most basalts and gabbros (+0.37 to +0.74‰; for a review, see Rouxel and Luais 2017) and than highpressure metabasites (El Korh et al., 2017b) (Fig. 11). As observed for serpentinites, hydrothermal alteration of basic rocks may have been responsible for Ge isotope fractionation towards heavier values than their protolith. However, no δ 74 Ge vs. δ 56 Fe correlation is observed, as δ 56 Fe values in amphibolites do not differ significantly from the δ 56 Fe of basic rocks, while δ 74 Ge values show evidence of a definite deviation from the values typically measured in basic rocks (Fig. 13b). Hence, Ge isotope fractionation has prevailed over Fe isotope fractionation during hydrothermal alteration of basic rocks. Reducing conditions have enhanced Ge isotope fractionation towards compositions heavier than MORB. A similar behaviour has also been observed during fluid-rock interactions in the retrograde greenschist facies metabasites of the Ile de Groix: in the most retrogressed samples, rehydration reactions in a reducing context triggered Ge isotope fractionation towards heavy compositions compared to the more oxidised greenschists and high-grade facies rocks (El Korh et al., 2017b).

Concluding remarks
Three isotopic systems (O, Fe and Ge) were employed to investigate processes controlling isotope fractionation in ancient non-subducted hydrothermally altered oceanic rocks from the Limousin ophiolite. The two main lithologies (serpentinites and amphibolites) display variable O, Fe and Ge compositions. The ultrabasic and basic magmatic precursors of the Limousin ophiolite-derived rocks were significantly metasomatised during the pervasive hydrothermal alteration on seafloor that has followed magma emplacement. While Fe and Ge isotopic signatures of pre-Variscan hydrothermal processes were preserved, there is no evidence of subsequent isotope fractionation during fluid-rock interactions related to the Variscan collision in the FMC, even during nappe stacking, involving ophiolite-derived rocks, under high-temperature conditions leading regional-scale partial melting.
In particular, Fe and Ge isotopes shows an opposite behaviour with contrasting redox conditions. In the highly oxidised abyssal serpentinites, Fe isotopes (δ 56 Fe from +0.14 to +0.18‰) may fractionate significantly towards heavier values than their magmatic ultrabasic protolith. By contrast, the δ 74 Ge (+0.48 to +0.93‰) only increases during intensive hydrothermal alteration of ultrabasic rocks. Amphibolite facies metagabbros, can conserve MORB-like Fe isotopic compositions (+0.12 to +0.16‰) during hydrothermal alteration, but the δ 56 Fe can decrease (+0.03‰) with the increase of the δ 18 O during intensive hydrothermal alteration. Ge isotopes can fractionate towards heavier values than basic protoliths in metagabbros (+0.72 to +0.78‰) that have conserved a MORB-like Fe reduced state during hydrothermal alteration. Hence, coupled O-Fe-Ge isotopes are efficient tracers of magmatic vs. hydrothermal processes.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.