An Integrated Study of the Serpentinite-Hosted Hydrothermal System in the Pollino Massif (Southern Apennines, Italy)

: A comprehensive study of the serpentinite and associated veins belonging to the Frido Unit in the Pollino Massif (southern Italy) is presented here with the aim to provide new constraints about the hydrothermal system hosted by the accretionary wedge of the southern Apennines. The studied serpentinites are from two di ﬀ erent sites: Fosso Arcangelo and Pietrapica. In both sites, the rocks show mylonitic-cataclastic structures and pseudomorphic and patch textures and are traversing by pervasive carbonate and quartz-carbonate veins. The mineralogical assemblage of serpentinites consists of serpentine group minerals (with a predominance of lizardite), amphiboles, pyroxene, chlorite, titanite, magnetite, and talc. In some samples, hydro-garnet was also detected and documented here for the ﬁrst time. As for cutting veins, di ﬀ erent mineralogical compositions were observed in the two sites: calcite characterizes the veins from Fosso Arcangelo, whereas quartz and dolomite are the principal minerals of the Pietrapica veins inﬁll, suggesting a di ﬀ erent composition of mineralizing ﬂuids. Stable isotopes of C and O also indicate such a di ﬀ erent chemistry. In detail, samples from the Pietrapica site are characterized by δ 13 C ﬂuctuations coupled with a δ 18 O shift documenting calcite formation in an open-system where mixing between deep and shallow ﬂuids occurred. Conversely, δ 13 C and δ 18 O of the Fosso Arcangelo veins show a decarbonation trend, suggesting their developing in a closed-system at deeper crustal conditions. Precipitation temperature calculated for both sites indicates a similar range (80 ◦ C to 120 ◦ C), thus suggesting carbonate precipitation within the same thermal system.


Introduction
Mantle peridotites are exposed on the seafloor at slow and ultraslow spreading mid-ocean ridges [1]. Seafloor spreading developed in areas characterized by tectonic extension and detachment faults, allowing uplift and exposure of mantle peridotite in oceanic core complexes [2][3][4][5][6][7]. Due to the interaction with seawater and deep hydrothermal fluids, seafloor peridotites are readily serpentinized over a wide range of conditions [8]. Serpentinites occurring in ophiolitic complexes are important for 0.6-0.8 GPa and temperatures of 350 • C, whereas the subsequent greenschist facies overprint took place at P = 0.4 GPa and T = 300-350 • C [37,59]. Cavalcante et al. [60] interpreted that also the nearby metasediments were affected by HP-LT conditions using illite crystallinity data and the b0 parameter of K-white mica in phyllite. Similar metamorphic conditions have also been documented by Invernizzi et al. [61] and Laurita and Rizzo [62] for the metabasites of the Frido Unit (200-300 • C and 0.6-0.8 GPa; 300-400 • C and 0.8-1.2 GPa, respectively).
to mobilization, fractionation, and redistribution of chemical elements during the emplacement of the Frido Unit serpentinites within the accretionary wedge.

Geological Background
The Pollino Massif is located in the southern Apennines at the Calabria-Lucania border zone ( Figure 1). It consists of tectonically juxtaposed thrust belts derived from the deformation of the African passive margin [38]. The deformation occurred between Oligocene and Pleistocene and involved the Ligurian ocean ophiolitic crust and its sedimentary cover [38,39]. In this area, the Liguride Complex [38], also defined as Liguride Units [40], is well exposed and is located in the highest position in the tectonics edifice of the southern Apennines. The Liguride Complex is derived from the NW subduction of the Tethyan ocean-continent transition zone and was divided into different tectonic units, where fragments of Jurassic oceanic crust [38], associated with slices of continental crust rocks, are preserved [38,39,41,42]. These terrains consist of a Mesozoic to Cenozoic flysch and a series of ophiolitic nappes, widely exposed along the whole Apennine Chain and in Calabria [38].  During blueschist facies conditions, crystallization of glaucophane occurred in metabasites [63], and of magnesio-riebeckite in schist associated with metabasites [64] and carpholite in some veins in metapelites and phyllites [37,65]. Afterward, in a brittle deformation regime, the most obvious structures are characterized by cataclastic bands well developed in the serpentinites [63].

Sampling and Analytical Method
A total of 26 representative samples of serpentinites dominated by carbonate-veins (SpFA) and serpentinites dominated by quartz-carbonate veins (SpPP) were collected at the Fosso Arcangelo site, near the San Severino Lucano village, and in the Pietrapica site, at the Calabria-Lucanian boundary ( Figure 1) ( Table 1).
The serpentinites outcropping at the Fosso Arcangelo site (SpFA) are brownish-grey-green and appear intensely reworked by strong brittle and ductile deformation, evidenced by several slip surfaces and by the presence of isoclinal and tight folds and some intrafolial folds often associated with crenulation cleavage. The fractures are commonly filled by carbonate minerals and display changes in both thickness and length (Figure 2a).
Serpentinites in this site crosscut by metadolerite dykes that were affected by metamorphism under relatively HP/LT (blueschist facies) conditions during the formation of the Apennine accretionary prism [45,[53][54][55]. The Pietrapica site is characterized by a complex exposure of dark-green cataclastic serpentinites (SpPP) with several slip surfaces and badland-like morphology. The serpentinite shows quartz-carbonate veins with talc-rich domains (Figure 2b). At the surrounding of the Pietrapica area, no rocks with any experience of HP/LT metamorphism occur. Petrographic characterization of all samples was carried out by optical microscopy on thin sections of rock samples oriented following their foliations and lineations. Serpentinites in this site crosscut by metadolerite dykes that were affected by metamorphism under relatively HP/LT (blueschist facies) conditions during the formation of the Apennine accretionary prism [45,[53][54][55]. The Pietrapica site is characterized by a complex exposure of dark-green cataclastic serpentinites (SpPP) with several slip surfaces and badland-like morphology. The serpentinite shows quartz-carbonate veins with talc-rich domains (Figure 2b). At the surrounding of the Pietrapica area, no rocks with any experience of HP/LT metamorphism occur.
Petrographic characterization of all samples was carried out by optical microscopy on thin sections of rock samples oriented following their foliations and lineations.
Mineralogy analyses were performed on randomly oriented powdered samples of both host-rocks and vein infill by using X-ray powder diffraction (XRPD) and μ-Raman techniques at the Department of Sciences, University of Basilicata (Potenza, Italy). XRPD analyses were accomplished by means of Siemens D5000 equipment with CuKα radiation, 40 kV and 32 mA, 2 s per step, and a step scan of 0.02° 2θ. Data were recorded between 5 and 70° 2θ for the bulk rock samples and from 15 to 70° 2θ for the vein infill. The mineral phase identification was carried out by means of the X'Pert HighScore Plus software (PANalytical 2001, Version 01 using the PDF-2 (2005) database. The μ-Raman analyses were carried out using a Horiba Jobin-Yvon LabRam HR800 spectrometer equipped with a HeNe laser source with a wavelength of 633 nm, a CCD detector operating at -70 °C, and an edge filter that excludes from detection shift below 150 cm −1 . A spectral resolution of 4 cm −1 was obtained by a holographic grating with 600 lines/mm. Correct calibration of the instrument was verified checking the position of the Si band at ±520.7 cm −1 . Output laser power was 20 mV, and measurements were performed using an optical microscope Olympus with objectives of 10×, 50×, and 100×. A laser beam spatial resolution of 1 μm was obtained with the 100× objective. Spectra results were from the average of 5 acquisitions of 10 s to optimize the signal/noise ratio. Two regions of the Raman spectra were investigated: 1200-150 cm −1 for structural bonding characterization and 3800-3500 cm −1 for the characterization of the hydroxyl groups. The minerals were identified based on the data reported in the online RUFF database.
Mineral chemistry was determined at the Centro Nacional de Microscopía Electrónica (CNME) of the Universidad Complutense (Madrid, Spain) by electron microprobe (EMP) analyses on the serpentinites and associated veins, using a JEOL Superprobe JXA-8900M equipped with four wavelength dispersive spectrometers. Silicate and oxide analyses were conducted at an accelerating voltage of 15 kV, an electron beam current of 20 nA, and a beam diameter of 5 μm. An accelerating voltage of 20 kV, an electron beam current of 10 nA, and a beam diameter of 5 μm were used for carbonate minerals. Each element was counted for 15 s. The following minerals were used as standards: sillimanite for the Al, albite for the Si and Na, Mineralogy analyses were performed on randomly oriented powdered samples of both host-rocks and vein infill by using X-ray powder diffraction (XRPD) and µ-Raman techniques at the Department of Sciences, University of Basilicata (Potenza, Italy). XRPD analyses were accomplished by means of Siemens D5000 equipment with CuKα radiation, 40 kV and 32 mA, 2 s per step, and a step scan of 0.02 • 2θ. Data were recorded between 5 and 70 • 2θ for the bulk rock samples and from 15 to 70 • 2θ for the vein infill. The mineral phase identification was carried out by means of the X'Pert HighScore Plus software (PANalytical 2001, Version 01 using the PDF-2 (2005) database. The µ-Raman analyses were carried out using a Horiba Jobin-Yvon LabRam HR800 spectrometer equipped with a HeNe laser source with a wavelength of 633 nm, a CCD detector operating at -70 • C, and an edge filter that excludes from detection shift below 150 cm −1 . A spectral resolution of 4 cm −1 was obtained by a holographic grating with 600 lines/mm. Correct calibration of the instrument was verified checking the position of the Si band at ±520.7 cm −1 . Output laser power was 20 mV, and measurements were performed using an optical microscope Olympus with objectives of 10×, 50×, and 100×. A laser beam spatial resolution of 1 µm was obtained with the 100× objective. Spectra results were from the average of 5 acquisitions of 10 s to optimize the signal/noise ratio. Two regions of the Raman spectra were investigated: 1200-150 cm −1 for structural bonding characterization and 3800-3500 cm −1 for the characterization of the hydroxyl groups. The minerals were identified based on the data reported in the online RUFF database.
Mineral chemistry was determined at the Centro Nacional de Microscopía Electrónica (CNME) of the Universidad Complutense (Madrid, Spain) by electron microprobe (EMP) analyses on the serpentinites and associated veins, using a JEOL Superprobe JXA-8900M equipped with four wavelength dispersive spectrometers. Silicate and oxide analyses were conducted at an accelerating voltage of 15 kV, an electron beam current of 20 nA, and a beam diameter of 5 µm. An accelerating voltage of 20 kV, an electron beam current of 10 nA, and a beam diameter of 5 µm were used for carbonate minerals. Each element was counted for 15 s. The following minerals were used as standards: sillimanite for the Al, albite for the Si and Na, almandine for the Mn and Fe, kaersutite for Mg, Ti and Ca, microcline for the K, fluorapatite for the P, Ca, F, Cl, and Ni, Cr pure metals. Corrections were made using the ZAF (Z: atomic number; A: absorption; F: fluorescence) method. The estimation uncertainties for major and minor elements were determined for each analysis, which have uncertainties from ±0.8% to ±5%. The structural formula of amphiboles was recalculated on the basis of 23 oxygens and classified by using the amphiboles nomenclature suggested by Leake et al. [66,67]. The andradite structural formula Minerals 2020, 10, 127 6 of 31 was recalculated on the basis of 24 oxygens. The structural formula of pyroxene was recalculated on the basis of 6 oxygens.
A total of 21 vein infills were selected for stable oxygen and carbon isotope analysis including 15 samples from the SpFA and 6 samples from the SpPP. In detail, about 0.1 mg of powder samples were put in a 12 mL screw cap Exetainer(R) vial and then flushed with pure helium to remove the air in the headspace. Subsequently, about 50 µL of 100% H 3 PO 4 was added to each sample for the conversion to carbon dioxide. The analyses were performed using a Thermo GB-II peripheral coupled with a Thermo Delta V Plus CF-IRMS at "Istituto Nazionale di Geofisica e Vulcanologia-Sezione di Palermo (Italy) laboratories" ( standards show that the accuracy for low-temperature measurements is better than ±0.1 • C. Critical point standards show that accuracy for high-temperature measurements is better ±0.1 • C. Homogenization temperatures (Th) have been interpreted as minimum entrapment temperatures. In this case, no pressure corrections were applied because a pressure determination would involve too many error-prone assumptions without an independently obtained value of pressure. The interpretation of Th as minimum entrapment temperatures is a typical procedure in working with Th data [69,70]. To interpret salinity, a NaCl-H 2 O model (using the equations from [71]) was used on the basis of the observed ice-melting temperatures of the last ice crystal Tm(Ice) from fluid inclusions.

The SpFA
The SpFA consist of serpentinite breccias with locally protomylonitic fabric and crosscutting carbonate veins. The serpentinite breccias (Figure 3a) are characterized by angular and irregularly shaped serpentinite grains embedded in the carbonate matrix. In the protomylonitic portion, breccias show differentiated crenulation cleavage with antisymmetric microfolds. The main foliation is defined by cleavage domains (limbs of microfolds) and microlithons (fold hinge areas). The main foliation is well-developed and is marked by hydro-andradite. This schistosity refolded the previous foliation ( Figure 3b) defined by recrystallized serpentine, metamorphic clinopyroxene, and accessory minerals such as titanite and magnetite. The SpFA samples consist of a primary mineral assemblage made up of olivine, orthopyroxene and clinopyroxene (augite), and a serpentinite assemblage consisting of serpentine group minerals (mainly lizardite and a minor amount of chrysotile and antigorite), tremolite, diopside, and clinochlore. Accessory minerals are garnet, titanite, magnetite, and carbonate phases. As recently documented by Dichicco et al. [73], locally edenite amphibole may occur in chlorite-free samples.
Lizardite + magnetite mesh texture or hourglass structures [74][75][76] usually occur in the SpFA showing cores of relict olivine grains replaced by calcite that locally are cross-cut by carbonate veins (Figure 3c). Along cleavage planes, lamellae and fibers of lizardite and chrysotile are spread on bastite pseudomorphs after pyroxene.
Orthopyroxene is replaced by bastite pseudomorphs with exsolution lamellae of lizardite or fine-grained diopside aggregates. Primary magmatic clinopyroxene (augite) is preserved or replaced by metamorphic diopside. Spinel is replaced by Cr and Al-magnetite at the core and clinochlore at the rim.
Several types of carbonate veins have been recognized in the SpFA samples. As shown in Figure 3d,e, veins crosscut the bulk rock and are distinguished in sheeted (micrite-filled veins, carbonate veins with The SpFA samples consist of a primary mineral assemblage made up of olivine, orthopyroxene and clinopyroxene (augite), and a serpentinite assemblage consisting of serpentine group minerals (mainly lizardite and a minor amount of chrysotile and antigorite), tremolite, diopside, and clinochlore. Accessory minerals are garnet, titanite, magnetite, and carbonate phases. As recently documented by Dichicco et al. [73], locally edenite amphibole may occur in chlorite-free samples.
Lizardite + magnetite mesh texture or hourglass structures [74][75][76] usually occur in the SpFA showing cores of relict olivine grains replaced by calcite that locally are cross-cut by carbonate veins (Figure 3c). Along cleavage planes, lamellae and fibers of lizardite and chrysotile are spread on bastite pseudomorphs after pyroxene.
Orthopyroxene is replaced by bastite pseudomorphs with exsolution lamellae of lizardite or fine-grained diopside aggregates. Primary magmatic clinopyroxene (augite) is preserved or replaced Minerals 2020, 10, 127 8 of 31 by metamorphic diopside. Spinel is replaced by Cr and Al-magnetite at the core and clinochlore at the rim.
Several types of carbonate veins have been recognized in the SpFA samples. As shown in Figure 3d,e, veins crosscut the bulk rock and are distinguished in sheeted (micrite-filled veins, carbonate veins with serpentine, carbonate veins with amphibole, fibrous calcite veins) and carbonate types displaying different thickness. Veins occasionally form an anastomosing network with acicular, fibrous, and radial serpentine and amphibole crystals (Figure 3f).

The SpPP
The SpPP are characterized by serpentinite breccias with quartz-carbonate rich veins. The serpentinite assemblage is made up of lizardite and minor chrysotile and antigorite, carbonate minerals, amphibole minerals (actinolite, tremolite), clinochlore, Cr-spinel, quartz, and talc. Magnetite is the only accessory mineral. Similarly to the SpFA, in the SpPP, lizardite occurs in the relict mesh texture and in the matrix.
An irregular patchy texture with carbonates growing after serpentine (Figure 4a) characterizes the SpPP samples. The carbonates occur as microcrystals, together with talc and rare fibrous tremolite. They occur as elongated rombohedric crystals ( Figure 4b) only in the veins. Locally, the SpPP samples are characterized by talc-rich domains where quartz and carbonate minerals are also present. Quartz is in micrometer sub-grains, showing undulatory extinction and dynamic recrystallization [77] and intergrowths with carbonate crystals (Figure 4c). Talc, mostly associated with serpentine (lizardite and chrysotile) and chlorite, occurs as massive coarse-to medium-grained aggregates, fine fibers, and/or tabular crystals with perfect cleavage on the [001] plane ( Figure 4d). serpentine, carbonate veins with amphibole, fibrous calcite veins) and carbonate types displaying different thickness. Veins occasionally form an anastomosing network with acicular, fibrous, and radial serpentine and amphibole crystals (Figure 3f).

The SpPP
The SpPP are characterized by serpentinite breccias with quartz-carbonate rich veins. The serpentinite assemblage is made up of lizardite and minor chrysotile and antigorite, carbonate minerals, amphibole minerals (actinolite, tremolite), clinochlore, Cr-spinel, quartz, and talc. Magnetite is the only accessory mineral. Similarly to the SpFA, in the SpPP, lizardite occurs in the relict mesh texture and in the matrix.
An irregular patchy texture with carbonates growing after serpentine ( Figure 4a) characterizes the SpPP samples. The carbonates occur as microcrystals, together with talc and rare fibrous tremolite. They occur as elongated rombohedric crystals ( Figure 4b) only in the veins. Locally, the SpPP samples are characterized by talc-rich domains where quartz and carbonate minerals are also present. Quartz is in micrometer sub-grains, showing undulatory extinction and dynamic recrystallization [77] and intergrowths with carbonate crystals ( Figure 4c). Talc, mostly associated with serpentine (lizardite and chrysotile) and chlorite, occurs as massive coarse-to medium-grained aggregates, fine fibers, and/or tabular crystals with perfect cleavage on the [001]

Mineral Chemistry
To better characterize silicate and carbonate minerals, an EMPA analysis was performed on selected samples of serpentinites and associated veins, from Fosso Arcangelo (SpFA5 and SpFA39) and Pietrapica

Mineral Chemistry
To better characterize silicate and carbonate minerals, an EMPA analysis was performed on selected samples of serpentinites and associated veins, from Fosso Arcangelo (SpFA5 and SpFA39) and Pietrapica (SpPP34A) sites. Amphiboles were analyzed in the host rock and associated veins, pyroxene and garnet were analyzed in the host rock, carbonate minerals were analyzed in the veins only (Tables 2-5).

Sample Code
SpFA39 No

Mineralogy
According to petrographic observations, the XRD analysis revealed that serpentinite samples from both studied sites are made up of serpentine polimorphs, mainly lizardite and, in minor amounts, chrysotile and antigorite, amphibole-like minerals, mainly actinolite and tremolite, clinochlore, magnetite, and calcite. Diopside and hydro-andradite are also present in the SpFA only.
As for veins, a different mineralogical composition was detected for the two analyzed sample groups. The vein infill of the SpFA samples consists of prevalent calcite and traces of aragonite and rhodochrosite. In these samples, traces of silicate minerals, such as serpentine, actinolite, and tremolite, were also detected. The infill of veins traversing the SpPP is dominated by dolomite and Mg-calcite, with quartz as the sole silicate phase.
µ-Raman spectroscopy has been used as a complementary technique to the X-ray diffraction analysis to better identify the carbonate minerals (calcite, aragonite, and dolomite) of both serpentinites and crosscutting veins. The optical vibrations are internal vibrations of the CO 3 group (three Raman bands lying between 1500 and 700 cm −1 ) and external or lattice vibrations involving translation and librations of the CO 3 groups relative to the Ca or Mg atoms (500-100 cm −1 ) [80]. In our samples, calcite is characterized by a dominant Raman band at 1091 cm −1 , minor bands at 713, 280, and 155 cm −1 and a very weak band at 1439 cm −1 (Figure 6a). The Raman spectrum for aragonite signals are detected for a dominant Raman band at 1086 cm −1 , two strong bands at 212 and 150 cm −1 , and three weak bands at 703, 250, and 180 cm −1 (Figure 6b). In dolomite, the main peak in the Raman spectrum occurs at 1103 cm −1 , whereas the weak peaks are at 730, 305, and 180 cm −1 (Figure 6c).

Mineralogy
According to petrographic observations, the XRD analysis revealed that serpentinite samples from both studied sites are made up of serpentine polimorphs, mainly lizardite and, in minor amounts, chrysotile and antigorite, amphibole-like minerals, mainly actinolite and tremolite, clinochlore, magnetite, and calcite. Diopside and hydro-andradite are also present in the SpFA only.
As for veins, a different mineralogical composition was detected for the two analyzed sample groups. The vein infill of the SpFA samples consists of prevalent calcite and traces of aragonite and rhodochrosite. In these samples, traces of silicate minerals, such as serpentine, actinolite, and tremolite, were also detected. The infill of veins traversing the SpPP is dominated by dolomite and Mg-calcite, with quartz as the sole silicate phase.
μ-Raman spectroscopy has been used as a complementary technique to the X-ray diffraction analysis to better identify the carbonate minerals (calcite, aragonite, and dolomite) of both serpentinites and crosscutting veins. The optical vibrations are internal vibrations of the CO3 group (three Raman bands lying between 1500 and 700 cm −1 ) and external or lattice vibrations involving translation and librations of the CO3 groups relative to the Ca or Mg atoms (500-100 cm −1 ) [80]. In our samples, calcite is characterized by a dominant Raman band at 1091 cm −1 , minor bands at 713, 280, and 155 cm -1 and a very weak band at 1439 cm −1 (Figure 6a). The Raman spectrum for aragonite signals are detected for a dominant Raman band at 1086 cm −1 , two strong bands at 212 and 150 cm −1 , and three weak bands at 703, 250, and 180 cm −1 (Figure 6b). In dolomite, the main peak in the Raman spectrum occurs at 1103 cm −1 , whereas the weak peaks are at 730, 305, and 180 cm −1 (Figure 6c).

Carbon and Oxygen Stable Isotope Analyses
Results of isotope analyses of carbonate phases in the veins of SpFA and SpPP are presented in Tables 6  and 7.
The C and O isotope ratios of carbonates from the veins of the SpFA have two distinct ranges. The δ 13 C values range from −0.81‰ to +2.16‰ and from −2.  [82], and Kim and O'Neil [83]. Instead, for vein samples of SpPP, dolomite-water fractionation curves of Schmidt et al. [84] and Horita [85] have been considered. In both areas, the oxygen isotope composition of water (δ 18 O) has been assumed to be 0‰ similar to modern seawater, as used by Agrinier et al. [86]. Depending on the fractionation factors available from the literature, for the vein samples of SpFA we have obtained an equilibrium temperature ranging from 83 to 117°C (Δ 18 Ocalcite from [81]), from 81 to 121 °C (Δ 18 Ocalcite from [82]), and from 76 to 107 °C (Δ 18 Ocalcite from [83]).

Carbon and Oxygen Stable Isotope Analyses
Results of isotope analyses of carbonate phases in the veins of SpFA and SpPP are presented in Tables 6 and 7.
Based on the XRPD results, calcite and dolomite have been considered as dominant carbonate phases in the veins of the SpFA and SpPP, respectively. Accordingly, equilibrium temperatures for SpFA veins were computed from δ 18 O data and considering the calcite-water fractionation curves of O'Neil et al. [81], Friedman and O'Neil [82], and Kim and O'Neil [83]. Instead, for vein samples of SpPP, dolomite-water fractionation curves of Schmidt et al. [84] and Horita [85] have been considered. In both areas, the oxygen isotope composition of water (δ 18 O) has been assumed to be 0% similar to modern seawater, as used by Agrinier et al. [86]. Depending on the fractionation factors available from the literature, for the vein samples of SpFA we have obtained an equilibrium temperature ranging from 83 to 117 • C (∆ 18 O calcite from [81]), from 81 to 121 • C (∆ 18 O calcite from [82]), and from 76 to 107 • C (∆ 18 O calcite from [83]).    [81]. c Temperature calculated using the equation proposed by [81]; [82]. d Temperature calculated using the equation proposed by [83]. e δ 13 C CO2 fluid calculated using the equation proposed by [87] considering b . f δ 13 C CO2 fluid calculated using the equation proposed by [87] considering c . g δ 13 C CO2 fluid calculated using the equation proposed by [87] considering d . Table 7. C and O isotope data and calculated fluid isotopic composition for the dolomite in serpentinites from the Pietrapica site.  Similarly, for the vein samples of SpPP, the equilibrium temperature was inferred in the range between 105 and 110 • C (∆ 18 O dolomite from [84]) and between 83 and 87 • C (∆ 18 O dolomite from [85]).
The pristine δ 13 C CO2 (gas) values of the fluid from which veins were formed were computed from the δ 13 C CaCO3 values and the calculated carbonate deposition temperature for each carbonate sample. We have assumed the achievement of the isotope equilibrium between fluid CO 2 and a carbonate mineralogical phase (calcite in the SpFA veins and dolomite in SpPP veins) and applied the following equation: where δ 13 C carb.min is the isotope composition of the considered carbonate mineralogical phase, and ∆ carb.min-CO2 is the equilibrium fractionation factor for carbon between the carbonate mineral and CO 2 and calcite [87] and between CO 2 and dolomite [85] (Tables 6 and 7). Based on the equilibrium temperature estimated using the fractionation factors computed by [81], [82], and [83], the average δ 13 C CO2 values of SpFA veins is in the range from −4.36% to −5.02% . Slightly more negative average δ 13 C CO2 values (from −6.86% to −8.32% ) were obtained for SpPP veins if we consider the equilibrium temperature estimated following [84] and [85], respectively.

Fluid Inclusions Hosted by Quartz in Sppp Veins
Most of the fluid inclusions in the quartz are arranged along lines of crystal growth, and thus, they are considered as primary and/or pseudosecondary fluids according to the criteria defined by [88] ( Figure 7a). Some fluid inclusions occur along secondary trails and necking down is occasionally seen. In Table 8, the primary or secondary origin of each fluid inclusion is indicated. When possible, fluid inclusion assemblages [89] have been analyzed. Two major types of fluid inclusions were recognized at room temperature: predominantly all-liquid fluid inclusions (single-phase inclusions: L H2O ) and in less amount liquid-vapor inclusions (biphasic inclusions: L H2O + V H2O ) (Figure 7b). All-liquid fluid inclusions usually nucleated a small gas bubble with little heating (around 50 • C) (Figure 7c). This means that they are in a metastable state out of their stability field. Only the smallest ones (usually <5 µm) remain all-liquid after heating. These ones cannot be used for microthermometric studies because a bubble is required for temperature determinations.

Fluid Inclusions Hosted by Quartz in Sppp Veins
Most of the fluid inclusions in the quartz are arranged along lines of crystal growth, and thus, they are considered as primary and/or pseudosecondary fluids according to the criteria defined by [88] (Figure 7a). Some fluid inclusions occur along secondary trails and necking down is occasionally seen. In Table 8, the primary or secondary origin of each fluid inclusion is indicated. When possible, fluid inclusion assemblages [89] have been analyzed. Two major types of fluid inclusions were recognized at room temperature: predominantly all-liquid fluid inclusions (single-phase inclusions: LH2O) and in less amount liquid-vapor inclusions (biphasic inclusions: LH2O + VH2O) (Figure 7b). Allliquid fluid inclusions usually nucleated a small gas bubble with little heating (around 50 °C) ( Figure  7c). This means that they are in a metastable state out of their stability field. Only the smallest ones (usually <5 μm) remain all-liquid after heating. These ones cannot be used for microthermometric studies because a bubble is required for temperature determinations.    Collectively, the two types of fluid inclusions have rounded and sub-rounded shapes and exhibit a relatively wide range of liquid/vapor volume ratios (some with ratios around 95:5 and others with ratios between 50:50 and 90:10), indicating a heterogeneous entrapment.
We did not find any evidence for the presence of CO 2 -and CH 4 -phases (Tm = −56.6 and −147.0 • C for CO 2 -and CH 4 , respectively). In fact, during the cooling phase, we went down until liquid nitrogen temperatures and no melting process different than the melting of ice was observed. This means that the gas is likely water vapor (Figure 7d). However, the presence of another gas different from water vapor cannot be discarded. If present, this gas would have very low density (H 2 or He), so that would be undetectable by microthermometry. The temperatures of final ice-melting Tm(Ice) values range from −0.3 to −1.9 • C. Many inclusions present positive ice-melting temperatures, which means that they are under high pressure out of their stability field. Temperatures of the first melting (eutectic temperature, Te) were observed around −30 • C, which is the metastable temperature of the H 2 O + NaCl system. Based on the Tm(Ice) and taking into account the H 2 O + NaCl system, the biphasic inclusions are found to be of low salinity, between 0.53 and 3.23 NaCl mass % equivalent (using equation by [71]). All the biphasic inclusions homogenize to liquid, with the final homogenization temperature (Th) present in two different ranges of temperature: the first one from 93 to 140 • C and the second one from 185 to 335 • C.

Mineral Assemblage
As reported by previous studies, the mineral assemblages and texture of serpentinites of the Frido Unit show evidence of ocean floor metamorphism [54,55]. In addition to minerals typical of worldwide serpentinites, including serpentine minerals, amphiboles, pyroxene, chlorite, titanite, and magnetite, the Frido Unit serpentinites are characterized by talc and hydro-garnet. In particular, the presence of garnet in the serpentinitic rocks from the studied area has been documented here for the first time.
Hydro-andradite, containing variable amounts of TiO 2 (0.75 to 3.60 wt %, Table 3), occurs in several mineral assemblages of serpentinites, among which the "serpentine + diopside + magnetite" association is one of the most common. Hydro-garnet is stable in these rocks over a wide range of oxygen fugacities and Ca activities, and its stability is controlled by the following reaction [90]: The titanian hydro-andradite may form in both magmatic and hydrothermal systems. According to [91], in fact, the presence of such a mineral has been documented in rocks associated with silica undersaturated magmatic systems (carbonatites, kimberlites, alkaline intrusions) as well as with hydrothermally alterated oceanic lithosphere (for example the Sanbagawa metamorphic complex, the mid-Atlantic ridge, the Nagaland ophiolite belt), testifying intermediate to low (150 to 300 • C) temperature fluids with low SiO 2 activity.
The hydrothermal activity is also thought of as responsible for the talc occurrence in the studied serpentinites. Frost et al. [90] stated that hydrothermal fluids in equilibrium with basic rocks may have high enough silica activity to alter serpentine to talc following the reactions below [16]: Based on the mineralogical composition of studied samples, as also confirmed by the petrographic observations (Figure 4d), the above reactions mirror the rock-fluid interaction processes that have involved samples from the Pietrapica site only, in which talc-rich domains and dolomite have been detected. However, in such samples along with talc and carbonate minerals, quartz is present as well. According to Moore and Rymer [92], a large amount of dissolved silica may be supplied to serpentinite rocks during the hydrothermal alteration of serpentine. Therefore, the quartz in the studied samples could represent the result of direct precipitation from a silica-rich fluid derived from the breakdown of serpentinite-forming silicates (serpentine).
The mode of occurrence of quartz in the SpPP veins suggests a further consideration of the chemistry of the mineralizing fluid in the Pietrapica area. The petrographical study of veins, in fact, has shown that intergrowth structures characterize quartz and dolomite in those samples, and this is consistent with the hypothesis of a contemporaneous formation of silicate and carbonate phases from the same source fluid. As a consequence, the chemical features of fluid inclusions in the quartz can be assumed as representative of the chemistry of the whole mineralizing fluid.
As for veins from the Fosso Arcangelo site, no hypothesis can be made on the basis of their mineralogical composition only. These veins are dominated by calcite as the principal carbonate phase. Calcite is a common mineral because it may form in a great variety of geological settings. It represents an important rock-forming mineral in sedimentary rocks, can be an essential component of metamorphic and igneous rocks, and is common in hydrothermal environments [93]. In particular, in geothermal systems, the calcite formation is chiefly favored by boiling, dilution, and condensation processes that control its occurrence, distribution, and stable isotope composition [94]. Further, during serpentinization, the mineralogical and geochemical processes transforming the oceanic lithosphere usually produce Ca-rich fluids that can migrate in the hydrothermal system and promote carbonates precipitation (mainly calcite) as serpentinite matrix and/or infill of veins and veinlets [35,95,96]. In the SpFA veins, calcite locally is in association with serpentine and amphibole crystals. Habitus of crystals and the lack of a preferential orientation of these silicates allow us to suppose that they were englobed into the hydrothermal fluid during its migration through the serpentinite host rocks. The lobate contacts between serpentine and/or amphibole crystals and calcite support this hypothesis.

Temperature of Precipitation, Fluid Composition, and Sources
Stable isotope (carbon and oxygen) geochemistry provides relevant constraints about metasomatic processes involving carbonates [97][98][99][100][101]. As previously stated, based on δ 18 O data and assuming the water/mineral isotope equilibrium, equilibrium temperature fluctuates in a narrow range for both the vein samples (T = 80-120 • C for SpFA and T = 80-110 • C for SpPP). Nonetheless, the veins belonging to different outcrops show different prevalent carbonate minerals (calcite in SpFA and dolomite for SpPP) and occurred under similar thermal regimes. The lack of a positive correlation between δ 13 C CaCO3 and δ 18 O CaCO3 (Figure 8) seems to indicate that the depositional temperature controls, exclusively, the isotope signature of the carbonate veins. Therefore, the computed δ 13 C CO2-gas values are representative of the original CO 2 supplied during vein formation.
The average fluid δ 13 C CO2 inferred from the fluid-carbonates isotope equilibrium range from −4.36% to −5.32% in the veins of the SpFA and from −6.86% to −8.32% in the veins of the SpPP.
The range of the C-isotope composition of both vein groups is slightly more negative than that of seawater carbonates (δ 13 C CO2 around 0% ), whereas it lies fully within the range of δ 13 C CO2 values typically associated to mantle carbon (−8% < δ 13 C CO2 < −4% ; [102][103][104]. Degassing of CO 2 -rich fluids during their rising from the underlying lithospheric mantle might lead to the precipitation of calcite [105]. Thus, the inferred δ 13 C CO2 values are consistent with a magmatic CO 2 component in the hydrothermal fluids. However, the available data do not allow us to rule out that carbon in these veins might derive from other sources than the mantle. In fact, the mixing between fluids having different isotope carbon isotope composition (e.g., seawater and carbon derived from organic rich sediments or from the oxidation of methane having δ 13 C CO2 < −15% ) would reproduce δ 13 C CO2 values in the same range to those inferred for the SpFA and SpPP veins.
Alternatively, decarbonation of marine sediments having a typical isotope signature (e.g., δ 13 C close to 0% , [106] would produce a lowering of δ 13 C, with or without change in δ 18 O, and an O shift can be observed only when decarbonation is driven by infiltration by externally derived H 2 O-rich fluids [34]. range for both the vein samples (T = 80-120 °C for SpFA and T = 80-110 °C for SpPP). Nonetheless, the veins belonging to different outcrops show different prevalent carbonate minerals (calcite in SpFA and dolomite for SpPP) and occurred under similar thermal regimes. The lack of a positive correlation between δ 13 CCaCO3 and δ 18 OCaCO3 (Figure 8) seems to indicate that the depositional temperature controls, exclusively, the isotope signature of the carbonate veins. Therefore, the computed δ 13 CCO2-gas values are representative of the original CO2 supplied during vein formation. Figure 8. δ 13 CV-PDB vs. δ 18 OV-SMOW plot of calcite/dolomite in the serpentinites from the Fosso Arcangelo site (SpFA = green diamonds) and Pietrapica quarry (SpPP = red squares). Geochemical trends of carbonates affected by dissolutions [99], carbonate reduction [101], and carbonation and decarbonation [35] are also shown.
The average fluid δ 13 CCO2 inferred from the fluid-carbonates isotope equilibrium range from −4.36‰ to −5.32‰ in the veins of the SpFA and from −6.86‰ to −8.32‰ in the veins of the SpPP.
The range of the C-isotope composition of both vein groups is slightly more negative than that of seawater carbonates (δ 13 CCO2 around 0‰), whereas it lies fully within the range of δ 13 CCO2 values typically associated to mantle carbon (−8‰< δ 13 CCO2< −4‰; [102][103][104]. Degassing of CO2-rich fluids during their rising from the underlying lithospheric mantle might lead to the precipitation of calcite [105]. Thus, the inferred δ 13 CCO2 values are consistent with a magmatic CO2 component in the hydrothermal fluids. However, the available data do not allow us to rule out that carbon in these veins might derive from other sources than the mantle. In fact, the mixing between fluids having different isotope carbon isotope composition (e.g., seawater and carbon derived from organic rich sediments or from the Figure 8. δ 13 C V-PDB vs. δ 18 O V-SMOW plot of calcite/dolomite in the serpentinites from the Fosso Arcangelo site (SpFA = green diamonds) and Pietrapica quarry (SpPP = red squares). Geochemical trends of carbonates affected by dissolutions [99], carbonate reduction [101], and carbonation and decarbonation [35] are also shown.
As shown in Figure 8, samples from the two studied sites fall in different fields of the δ 18 O vs. δ 13 C diagram. In detail, the SpFA veins overlap the pathway typical of the decarbonation process as suggested by [35] and references therein. Such a process likely developed at depth in the crust in a "closed system" wherein no external fluid supply can occur.
On the contrary, for the SpPP site, vein samples are characterized by an oxygen isotope shift toward more positive δ 18 O values. Moreover, significant differences in the mineralogical assemblage were also found in SpFA and SpPP veins, being calcite prevalent in the former veins and dolomite and quartz in the latter ones. Intergrowth structures between dolomite and quartz crystals suggest the hypothesis of a contemporaneous formation of silicate and carbonate phases in the SpPP veins.
Data of fluid inclusions in the quartz (in SpPP veins) show abundant aqueous (L H2O -V H2O ) and low salinity features (between 0.53 and 3.23 NaCl mass % equivalent). Further, the decrease of salinity associated with the decrease of homogenization temperature (335-185 • C and 140-93 • C) may be related to a large infiltration of shallow, diluted, and fresh waters that also led to a progressive cooling of the hydrothermal system. Therefore, it seems that SpPP veins are consistent with a crystallization in an open-system at shallower crustal conditions. Accordingly, all these features, even not well-constrained would suggest the hypothesis that SpFA and SpPP veins were deposited under different boundary conditions (e.g., temperature), and/or from parental fluids having different chemical composition.
Further detailed investigations on the fluid inclusions composition as well as on the oxygen isotopes of the silicate are requested to clarify the formation process of these veins and to fully understand the role and the compositions of the parental fluid(s).

Conclusions
Our study demonstrates that, in serpentinites of the Frido Unit (southern Apennines), different types of veins occur recording fluid production and migration in the accretionary wedge of the southern Apennines. We envisage at least two formation episodes for veins that crosscut serpentines from the studied sites (Fosso Arcangelo and Pietrapica) within the same thermal system (Figure 9a). We have identified a first vein group formed by decarbonation of serpentinites in a closed hydrothermal system that generated Ca-and CO 2 -rich fluids from which carbonates of the Fosso Arcangelo veins derive. Even if not well-constrained, such fluids could migrate toward shallower depths, modifying their composition by interaction with serpentinite host rocks. Then, fluid migration ended in an open hydrothermal system where mixing between deep Si and Mg enriched fluids and externally derived H 2 O-rich fluids occurred, promoting the quartz-carbonate vein formation in the Pietrapica serpentinites. Subsequently, serpentinite slices and associated veins were involved by tectonic activity related to the formation of the Liguride Complex and finally exhumed (Figure 9b).