Serpentinization, Carbonation, and Metasomatism of Ultramafic Sequences in the Northern Apennine Ophiolite (NW Italy)

Fluid‐rock interaction in ultramafic rocks considerably affects the chemical and isotopic composition of the oceanic lithosphere. We present a geochemical and petrological study of serpentinites and ophicalcites of the Northern Apennine ophiolite, Italy. This ophiolite sequence represents fragments of Jurassic oceanic lithosphere that have been denuded by low angle detachment faults, exposing peridotites on the ocean floor and triggering hydrothermal alteration. Seawater circulation is documented by (Jurassic) seawater‐like 87Sr/86Sr values and δ13C values of 1.1–3.0‰ in carbonate veins of the ophicalcites. Bulk rock ophicalcites have low 87Sr/86Sr values of 0.70489–0.70599, elevated SiO2 contents, and talc druses filling calcite veins that record Si‐metasomatism. In contrast, underlying serpentinites have 87Sr/86Sr values above Jurassic seawater values. Bulk rock δD and δ18O values of ophicalcites and serpentinites suggest interaction with an evolved seawater‐derived and/or magmatic fluid. These chemical signatures result from a complex history of serpentinization, carbonation, and metasomatism. Multiphase water‐rock interaction includes infiltration of basement‐derived fluids during initial mantle upwelling within an opening ocean basin, followed by localized high‐temperature fluid infiltration, extensive seawater circulation resulting in carbonation, and oxidation near the seawater‐exposed surface, and finally, fluid‐rock interaction with overlying mafic lithologies leading to Si‐metasomatism.

magnetite (Bach et al., 2004;Moody, 1976;O'Hanley, 1996). These reactions generally produce low-temperature, alkaline (pH ∼9-11), Ca-, and H 2 -rich fluids that induce carbonate formation upon interaction with seawater, such as those described from the Lost City hydrothermal field (LCHF) on the Atlantis Massif (AM) (Mid-Atlantic Ridge (MAR), 30°N; Kelley, Karson, et al., 2001;Ludwig et al., 2006). The LCHF represents one of the best examples of a modern serpentinite-hosted marine hydrothermal system and is characterized by up to 60 m high carbonate-brucite chimneys and towers that vent up to 95°C, high pH fluids and host diverse microbial communities (e.g., Brazelton et al., 2006;Kelley, Karson, Früh-Green, et al., 2005). Recent studies have shown that the evolution of peridotite-hosted hydrothermal systems is complex, commonly involving melt-rock interaction and a complex alteration history of serpentinization, talc metasomatism, carbonation, and low-temperature seafloor weathering and oxidation (e.g., Früh-Green et al., 2018;Paulick et al., 2006). In many cases, interaction of water with interspersed gabbroic intrusions causes wide-spread or localized high-temperature imprint and changes the chemical signatures of the basement rocks, as documented by distinct enrichments in Si, Al, Ca, and assemblages of Ca-rich amphibole, talc, and chlorite (Boschi et al., 2006a(Boschi et al., , 2006bRouméjon et al., 2018). In addition, high-temperature fluid-rock interaction associated with magmatism can produce extensive sulfide deposits (e.g., Garuti et al., 2008;Marques et al., 2007), whereas low-temperature serpentinization may facilitate formation of considerable amounts of microbially produced sulfide (e.g., Alt, Schwarzenbach, et al., 2013).
Here, we present a study of alteration processes and carbonate deposition in the Northern Apennine ophiolite in Italy with the aim to unravel its chemical evolution from serpentinite to ophicalcite formation. Ophiolites are fragments of oceanic lithosphere that have tectonically been emplaced on the continent (Coleman, 1977). Due to their exceptional exposure, they can provide unique insight into magmatic and hydrothermal processes that took place on the seafloor. The Northern Apennine ophiolite is considered a relict of oceanic lithosphere formed in the Piemont-Ligurian ocean. Previous research showed that this sequence records an oceanic alteration history similar to that described in a number of ultramafic core complexes along the MAR (e.g., Alt, Crispini, et al., 2018;Karson et al., 2006;Schwarzenbach, Früh-Green, Bernasconi, Alt, & Plas, 2013). The ophiolite only experienced a weak Alpine metamorphic overprint that did not exceed prehnite-pumpellyite facies conditions and did not modify the marine signatures (e.g., Cortesogno, Gianelli, & Piccardo, 1975;Garuti et al., 2008;Strating, 1991). Our study builds on previous work (Schwarzenbach et al., 2012(Schwarzenbach et al., , 2018, which was based on carbon and sulfur geochemistry and provided initial evidence for extensive seawater infiltration and variations in temperatures and fluid regime that also facilitated microbial activity in the serpentinites. Here, we present strontium, oxygen, and hydrogen isotope bulk rock data and in situ major and trace element compositions, which provide new, comprehensive insights into the hydrothermal and temporal evolution of this ophiolite sequence and the origin of the fluids that caused rock alteration. In particular, the good exposure of this sequence allows us to track its hydrothermal evolution during continuous uplift and exposure on the Jurassic ocean floor and the transition from serpentinite to ophicalcite formation. We compare this ancient system with the AM, which provides a modern analog for the N. Apennine ophiolite and allows us to evaluate the interplay of serpentinization, metasomatism, and carbonation, and their effects on mass transfer during the hydrothermal evolution of ultramafic rocks. This allows us to better understand the chemical evolution of the oceanic lithosphere, which is eventually subducted along convergent margins and effectively controls long-term geochemical cycles.

Regional Geology
The ophiolite successions in the N. Apennine (Eastern Liguria, Italy) represent relicts of a former oceanic lithosphere of the Ligurian Tethys that separated the European and Apulian continental plates during the Late Jurassic-Cretaceous and locally developed a slow-to ultraslow-spreading ridge (Lagabrielle & Lemoine, 1997;Strating, 1991;Tribuzio, Thirlwall, & Vannucci, 2004). Along these spreading centers mantle exhumation was prevalent, whereas mature oceanic spreading was comparatively short-lived (Le Breton et al., 2021). Accordingly, these ophiolite sequences are characterized by a heterogeneous basement made up of gabbroic intrusions in mantle peridotites and are discontinuously overlain by ophicalcites, basaltic lavas, sedimentary breccias, and pelagic sediments (e.g., B. E. Treves and Harper, 1994). The peridotite basement varies in composition from depleted spinel harzburgites to fertile spinel lherzolites, and from dunites to plagioclase-enriched peridotites (Piccardo et al., 2014). MORB-type lavas form a thin and laterally discontinuous volcanic succession that was locally deposited directly on top of the mantle basement or the gabbros . Timing of gabbroic intrusions within the external and internal Ligurian units has been dated at 165-161 Ma based on U-Pb zircon ages and records the transition from an ocean-continent transition (OCT) zone to a slow-spreading ridge setting (Tribuzio, Garzetti, et al., 2016).
In the N. Apennine, the Internal Ligurian ophiolite units represent the oceanward fragments of oceanic lithosphere and include the Bracco and Val Graveglia Units (Figure 1). The Bracco Unit includes gabbroic intrusions in serpentinites that are discontinuously overlain by basaltic flows and pillow lavas with MORB affinity and Upper Jurassic to Cretaceous sediments (Cortesogno, 1981). The upper layer of the serpentinite basement becomes progressively more fractured, oxidized, and carbonated upwards, forming ophicalcites. The ophicalcite, also known as the Levanto Breccia, hosts an entangled polyphase complex of veins and fractures filled with calcite, talc, and a micritic to microsparitic matrix (Cortesogno, Galbiati, & Principi, 1980;B. E. Treves and Harper, 1994). Calcite precipitation occurred close to the seafloor surface where upwelling low-temperature hydrothermal fluids formed upon seawater-peridotite interaction and discharged through the fractured serpentinite basement (Schwarzenbach, Früh-Green, Bernasconi, Alt, & Plas, 2013). The ophicalcites have been interpreted as extensional fault rocks and products of marine hydrothermal activity related to detachment faulting, which display evidence for an evolution from ductile to brittle deformation during cooling and decompression of the oceanic lithosphere (B. Treves et al., 1995;B. E. Treves and Harper, 1994). The Levanto Breccia is overlain by an upper sedimentary sequence, the Framura Breccia, which has sedimentary features, sedimentary dykes and infills, and is characterized by reworking of the underlying ophicalcites.
The Val Graveglia Unit is situated northwest of the Bracco Massif and is characterized by serpentinized peridotites, gabbros, pillow basalts, pillow breccias, and an Upper Jurassic to Cretaceous sedimentary cover. Additionally, the Val Graveglia Unit hosts massive sulfide deposits that include stratiform deposits, which overlie serpentinites or are associated with pillow basalts overlying the serpentinite and stockwork veins in gabbro and basalt .

Sample Selection
Serpentinites, ophicalcites, and fault rocks from the Bracco Unit were systematically sampled in several active or abandoned quarries ( Figure 1) allowing for the recovery of fresh sample material unaffected by long-term weathering. Talc-tremolite fault schists were sampled from a dome-shaped structure in the Cava dei Marmi that represents a several meters-wide zone of strongly sheared fault rocks along the contact between the ophicalcite and the overlying volcano-sedimentary sequence (from here on referred to as Cava dei Marmi fault schists). We classify the ophicalcites into two groups according to their oxidation state and tectonic history: green and red ophicalcites. The green ophicalcites are defined here as completely serpentinized and variably calcite-veined massive peridotites. They have been fractured in situ and are characterized by the presence of cross-cutting generations of calcite veins. Locally, oxidation is observed along the calcite veins ( Figure 2a). The red ophicalcites typically overlie the green ophicalcites and are highly fractured, partly brecciated, and are interpreted as tectonic breccias (Figure 2b). They are characterized by an intense red color, crosscut by a myriad of calcite veins commonly containing talc in druses. The transition from serpentinites, to green and then to red ophicalcites is gradual and reflected by an increase in calcite veining and SCHWARZENBACH ET AL.  oxidation from bottom to top of the exposed sequence. A transect through all three lithologies was sampled to account for variations in veining and oxidation.
Serpentinites and tremolite-schists from the Val Graveglia Unit were sampled at one abandoned mining site close to the village of Libiola. In the study area, the serpentinites over-thrust the brecciated basalts and are therefore not associated with the underlying stockwork deposits. A prominent lens of tremolite-schist at the contact between the serpentinites and the brecciated basalts was sampled as well (Figures 2c and 2d) (from here on referred to as Libiola fault schists). In addition, gabbro samples from the Bracco Unit investigated by Molli (1995) were used in this study for strontium isotope analyses.

Petrographic Characterization
The mineralogy and petrology were investigated on thin sections with transmitted light microscopy and using a JEOL JSM-6390 LA scanning electron microscope with energy dispersive X-ray spectroscopy capabilities. Quantitative element concentrations were obtained by electron microprobe analysis carried out on a JEOL JXA-8200 at ETH Zurich. The beam was set to 15 kV accelerating potential, 1-10 µm beam size, and 5-20 nA beam current. Natural and synthetic standards were used for calibration. Additionally, X-ray diffraction (XRD) spectra were acquired on vein powders with a LynxEye X-ray diffractometer (Bruker AXS D8 Advance) at ETH Zurich to determine the carbonate mineralogy.

Major and Trace Element Analysis
Bulk rock powders of the serpentinites and ophicalcites were prepared by cutting away the outermost rind to remove contamination, crushing in a steel mortar, and grinding to a powder in an agate mill. Major and trace element analysis of the powders were obtained by Panalytical Axios wave-length dispersive X-ray fluorescence spectrometry (WD-XRF, 2.4 kV) at ETH Zurich after fusing the bulk rock powders to glass beads with Lithium-Tetraborate (ratio 1:5). Bulk rock trace element compositions were measured on glass beads with a laser-ablation micro-sampler coupled to an inductively coupled Elan 6000 plasma mass spectrometer (LA-ICP-MS). Boron and REE concentrations in ultramafic rocks were determined at the commercial Activation Laboratories Ltd. (Ancaster, Canada) by PGNAA Elemental Analyzer and by using its research quality Lithium Metaborate/Tetraborate Fusion ICP-MS method, respectively. Mineral REE compositions were measured on thin sections with a Resonetics Resolution 155 laser ablation system, coupled to a Thermo Element XR Sectorfield ICP-MS and using a diameter of 20-40 µm. Data reduction was performed using the Matlab-based code SILLS (Guillong et al., 2008).

Strontium Isotope Analyses
To prepare whole-rock powders of gabbros and ultramafic rocks for isotope analysis, rock samples were cut, separated from major calcite veins, crushed in a steel mortar, and ground in an agate mill. Strontium isotope analyses were performed on a Thermo Finnigan Neptune 53 MC-ICP-MS at ETH Zurich, after dissolution in a mixture of HF and HNO 3 according to ion exchange procedures developed by de Souza et al. (2010) and Deniel and Pin (2001). Because of the highly heterogeneous carbon content, the carbonate-rich green and red ophicalcites were decarbonized at least twice (to ensure complete carbonate removal) with 2M HCl prior to dissolution and the final residues were analyzed with a coulometer to determine if all carbonate had been dissolved (Vogel, 2016).
Calcite veins were micro-drilled to extract different vein generations and were dissolved with the following procedure: carbonate powders were dissolved in concentrated HCl, dried down, re-dissolved in concentrated HNO 3 , dried down and diluted 150 times with 2% HNO 3 prior to analysis. The long-term average value for 87 Sr/ 86 Sr of the NBS987 standard measured during the period of analysis is 0.710243 with a 2σ standard deviation of 21 μg/g (N = 43). The 87 Sr/ 86 Sr ratios were adjusted for mass bias by normalizing to a 87 Sr/ 86 Sr ratio of 0.1194 and are reported relative to the certified value of 87 Sr/ 86 Sr = 0.71024 for the NBS987 standard. In addition, the 87 Sr/ 86 Sr ratios have been corrected for their age by using the measured Rb/Sr to an average age of 160 Ma (Jurassic).

Stable Isotope Analyses
Hydrogen and oxygen isotope analyses of silicates were carried out on bulk rock powders of the serpentinites and green ophicalcites. Hydrogen isotope (δD) compositions were analyzed using the high-temperature (1,400°C) reduction method with He as a carrier gas and a TC-EA (Flash1112HT, Thermo Fisher) interfaced to a Delta Plus XP IRMS at IGG-CNR, Pisa (Italy). Oxygen isotope values were determined on whole-rock powders after decarbonization of all the samples with 2M HCl (see Section 3.3). Measurements were conducted at the Institute of Earth Sciences at the University of Lausanne (Switzerland) using a CO 2 laser fluorination system coupled to a Finnigan MAT-253 mass spectrometer. Results are given in standard δ-notation, expressed relative to the Vienna Standard Mean Ocean Water (V-SMOW) in permil (‰). Replicate hydrogen isotope analyses of internal standards give an average precision better than ±5‰. Oxygen analyses were duplicated or triplicated and corrected to the LS-1 in house standard, which yielded an average value of 18.1 ± 0.6‰. However, typical precision for the standard analyses is ±0.2‰. Higher errors during analysis were likely caused by oxidation during fluorination of the fine-grained powders, leading to higher background interferences.
Oxygen and carbon isotope analyses of micro-drilled calcite veins were analyzed on a GasBench II with a continuous-flow IRMS setup, after the method of Breitenbach and Bernasconi (2011). Isotope ratios are reported in permil (‰) using the conventional δ-notation and with respect to the V-SMOW for oxygen and Vienna Pee Dee Belemnite (V-PDB) for carbon. Standard deviation was 0.05‰ for both δ 13 C and δ 18 O.

Sample Description
Macroscopic and microscopic investigation of the serpentinites, and green and red ophicalcites indicate a similar protolith but highly variable alteration characteristics. The serpentinites from the basement of the Bracco Unit are completely serpentinized harzburgites and clinopyroxene-poor lherzolites. The serpentinites and ophicalcites show a typical mesh texture with no relicts of primary minerals except for Cr-rich spinel, which is a common magmatic mineral in peridotites and is strongly altered, as well as rare diopsidic and partly serpentinized clinopyroxene that is also most likely of primary magmatic origin. The groundmass of the ophicalcites commonly contains talc and chlorite as microscopic intergrowths in the serpentine mesh and amphibole either around bastites that replace orthopyroxene or forming veins ( Figure 2). Oxidation is observed in both green and red ophicalcites. In the green ophicalcites, oxidation of the serpentinite mesh texture initiates along calcite veins locally producing a reddish halo along the veins (Figure 2a). In contrast, the red ophicalcites show pervasive oxidation of the groundmass as defined by the high abundance of hematite (as determined by polarization microscopy) and/or other Fe-oxides and -hydroxides finely dispersed in the groundmass, which gives the intense red color (Figure 2b). Similarly, calcite veining is significantly more abundant in the red ophicalcites, where calcite veins mostly follow mm to cm wide fractures. In the red ophicalcites, numerous generations of calcite veins can be distinguished based on fabrics and crosscutting relationships ( Figure 3): (1) Early calcite veins, in long planar fractures and locally including fragments of serpentine fibers; (2) Thinner ribbon calcite veins as bands of subparallel calcite veins that commonly show extensional structures; (3) Vein networks including jigsaw puzzle breccias, consisting of a disarray of calcite veinlets with complex geometry and a clear hydrofracture texture; (4) Breccias with pink/gray calcite-micrite matrix and serpentinite clasts; and (5) Late-stage calcite veins filling thick and branching irregular fractures. These latter veins cut across older features and reactivate older surfaces. Furthermore, late-stage calcite veins are associated with talc druses in the center of the veins, as previously reported in detail by B. Treves et al. (1995).
The Cava dei Marmi fault schists are strongly sheared, highly heterogeneous, and characterized by the abundance of talc, tremolite, and calcite. Serpentinite occurs as rare clasts and chlorite in pockets. Such assemblages are typical for metasomatic rocks observed in detachment fault zones on the seafloor and suggests an ocean floor setting (Boschi et al., 2006a(Boschi et al., , 2006bKarson et al., 2006). However, tectonic reworking during the Alpine orogenesis cannot be excluded completely.
Mg-rich gabbros, sampled in the Bracco Pass and the Bonassola area, intrude the serpentinites in the Bracco Massif and commonly show medium-to coarse-grained porphyroblastic textures, consisting mainly of clinopyroxene and plagioclase. High temperature shear zones are preserved within sections of undeformed gabbros. In the Bonassola area the effect of hydrothermal circulation in the gabbros is also apparent. Widespread fractures filled with Mg-hornblende (± plagioclase), which replaces the clinopyroxene, occur as parallel swarms crosscutting the high-T shear zones. For a detailed description of the gabbros, we refer to Molli (1995) and Tribuzio, Renna, et al. (2014).
Macro-and microscopically, the Val Graveglia serpentinites are similar to the Bracco serpentinites with a dominance of serpentine and magnetite and no primary mineral phases preserved other than spinel. Sulfide minerals, amphibole, and talc are rare in these serpentinites. Only the Libiola fault schists are characterized by an abundance of elongated tremolite crystals and are randomly crosscut by tremolite veins, whereas serpentine is rare and only present as angular clasts. Chlorite is visible in pockets, which are locally crosscut by tremolite veins.

Serpentine, Talc and Chlorite
The texture and chemical composition of serpentine strongly depends on the primary mineral phases it replaces. Serpentine after olivine is the main phase forming the groundmass and commonly develops a mesh or ribbon texture. The meshes typically are formed by fibrous serpentine at the rims and very fine-grained, homogeneous, and locally isotropic serpentine in the mesh cores suggesting a dominance of lizardite and chrysotile. Orthopyroxene has been completely replaced by aggregates of fibrous serpentine minerals that retain the prismatic shape of the original orthopyroxene grains and form bastites. In some of the ophicalcites, serpentine veins cut the mesh texture and were reactivated by calcite veining or vice versa ( Figure 2g). No evidence of serpentine recrystallization was observed that would indicate a regional metamorphic SCHWARZENBACH ET AL.  overprint, which is consistent with maximum metamorphic conditions around prehnite-pumpellyite facies as constrained by Garuti et al. (2008).
The chemical variation of serpentine (serpentine mesh texture, bastite, and veins), chlorite, and talc from serpentinites and ophicalcites is presented in Figure 4. In the serpentinites, only serpentine with compositions very close to lizardite was analyzed and no talc or chlorite was detected. In all samples (i.e., serpentinites, green and red ophicalcites), serpentine after olivine shows a large variation in Mg# (Mg/(Mg + Fe)) of 0.76-0.89, whereas bastites have a narrower range (0.81-0.93). Mesh cores have lower Mg# than mesh rims, indicating an enrichment in FeO tot from rim to core of the serpentine meshes with up to 9.8 wt.% FeOtot in mesh cores (with no magnetite present). Bastites have highly variable compositions within individual grains and show higher contents of Cr 2 O 3 and Al 2 O 3 , and low contents of NiO compared to mesh serpentine, reflecting the primary orthopyroxene composition (see Table S2). Serpentine veins are generally poorer in FeO tot compared to groundmass serpentine.
Mineral assemblages with talc and chlorite were only detected and analyzed in the ophicalcites. In addition, mixing trends between serpentine and chlorite, and between serpentine and talc suggest fine-grained mineral intergrowths on a submicrometer scale ( Figure 4). Polarization microscopy indicates that talc occurs in pockets in the core of the serpentine mesh texture, as fine-grained rims around bastites, and intergrown within serpentine in the bastites (Figures 2e and 2f). It is also present as fine intergrowths with amphibole and chlorite forming veins. In some red ophicalcites, coarse-grained talc-druses (up to 3 cm wide) occur in late-stage calcite veins (Figures 2b and 3). In the Cava dei Marmi fault schists that border the ophicalcites, talc occurs as fine-grained aggregates in pockets and, rarely, within the cores of serpentine meshes. Only analyses of talc veins (within calcite veins) from the red ophicalcite yielded pure talc analyses, whereas all other occurrences are fine serpentine-talc intergrowths. Chlorite forms either fan-shaped aggregates in the serpentine groundmass, where it occurs with talc and amphibole as vein-like features, or replaces pyroxene SCHWARZENBACH ET AL.  in bastites (Figures 2g and 2h). In most cases, chlorite forms fine-grained intergrowths with serpentine ( Figure 4). In addition, chlorite is associated with altered Cr-rich spinel in the serpentinites and ophicalcites.

Amphibole
Amphiboles are present mostly in the ophicalcites and in the Cava dei Marmi and Libiola fault schists. They are colorless, calcic in composition, and form needles and blades at the rim of bastites or partly replacing bastites. They are also found as veins and in pockets, or in the serpentine mesh texture as very fine intergrowths with serpentine, talc, and chlorite (Figures 2e and 2h). In the fault schists, amphiboles are present as veins or in pockets together with chlorite. Following the classifications of Leake et al. (1997), the amphiboles are mostly magnesio-hornblende with lesser tremolites ( Figure 5) and show no clear distinction between occurrences in veins and the fault schists. However, the chemical compositions of the amphiboles that form rims around bastites and those that replace bastites are more variable and also include edenite compositions.

Calcite Veins
The ophicalcites have a complex network of multiple generations of veins, ranging in size from a mm-scale microscopic network that appears to hydrofracture the mesh texture, to macroscopic veins that are several centimeters thick and several meters in length. The veins are composed mainly of calcite as previously also determined by Schwarzenbach, Früh-Green, Bernasconi, Alt, and Plas (2013). No aragonite, Mg-calcite or dolomite were detected by XRD or microscopic observations, although B. E. Treves and Harper (1994) described calcite fibers and rosettes that may represent pseudomorphs of former aragonite crystals. Locally, calcite is present in the core of the serpentine meshes.

Minor Phases and Oxides
All magnetite is most likely of secondary origin associated with serpentinization and carbonation. Very small magnetite grains and oxide needles are distributed homogeneously in the serpentinite mesh texture or form thick oxide bands at the rim of serpentine veins as observed by polarization microscopy. Magnetite also occurs locally as euhedral crystals in calcite and serpentine veins. Hematite, Fe-oxides and/or Fe-hydroxides give the noticeable red color in the red ophicalcites and occur along calcite veins in the green ophicalcites. In some cases, ghost-structures of hematite in calcite veins can be found, reflecting older serpentine mesh structures. Almost no magnetite could be observed in the bastites.
Cr-rich spinel is observed in all thin sections and shows incipient alteration to Fe-chromite ± chlorite along the edges. The composition of Cr-spinel varies strongly between the different samples: Cr 2 O 3 -contents range from 25.8 to 38.1 wt.%, MgO-contents are between 7.8 and 17.5 wt.% and Al 2 O 3 -contents are between 18.9 and 39.0 wt.%. Sulfide minerals are abundant in calcite veins and occur locally within the serpentine mesh. Sulfides include pentlandite, pyrrhotite, pyrite, millerite, siegenite and rare chalcopyrite, and are described in detail by Schwarzenbach, Früh-Green, Bernasconi, Alt, Shanks III, et al. (2012).

Bulk Rock Major and Trace Element Geochemistry
Bulk rock major element compositions of serpentinites and ophicalcites are plotted (on an anhydrous basis) in Figure 6 together with serpentinites and talc-amphibole schists from the AM (Boschi et al., 2006a) and fresh peridotites from the Internal Ligurian units (Mt. Fucisa; Rampone et al., 1996). The chemical composition of the serpentinites overlaps with the serpentinites from the AM and with the Internal Ligurian peridotites; however, our samples on average show high Fe 2 O 3 and higher Al 2 O 3 contents compared to the AM serpentinites. The gradual transition from serpentinite to ophicalcite from the Bracco Unit is highlighted by an increase in CaO (up to 6.4 wt.%), due to carbonate veining, and by an increase in SiO 2 and a decrease in MgO, due to the formation of amphibole/talc. A similar trend is found for the AM talc schists, especially with regard to SiO 2 and MgO contents.   (Rampone et al., 1996), the Atlantis Massif serpentinites (AM serpentinites) and talc-amphibole schists (AM talc schists) (Boschi et al., 2006a). Total iron calculated as ferrous iron. The ophicalcites are generally enriched in SiO 2 , Al 2 O 3 , and CaO, and depleted in MgO compared to basement serpentinites reflecting formation of talc, amphibole, and calcite, the latter in veins.
In situ mineral analyses of the mesh serpentine and bastite minerals yielded different REE patterns than the serpentinite bulk rock analyses (Figure 7c). They are characterized by highly variable and depleted LREE compositions and a nearly flat pattern for the HREE. Both groundmass serpentine and bastites display slight positive Eu anomalies, but variable (positive and negative) Ce anomalies. Some of the bastites have slightly higher REE compositions than the serpentine groundmass and reflect the precursor pyroxene: LREE depletion, and a flat HREE pattern. One analysis of clinopyroxene yielded elevated REE compositions compared to the bastites and serpentines.  (Table 1).
In addition to the serpentinites and ophicalcites, gabbros from the Bracco unit were analyzed for their Sr isotope composition. Overall, the studied samples have a large range in 87 Sr/ 86 Sr ratios (Figure 8)

Carbon and Oxygen Isotope Systematics of Calcite Veins
The carbonate veins have δ 13 C values of 1.1-3.0‰ (V-PDB) (Figure 9a). The δ 18 O values range from 15.4 to 24.6‰ (V-SMOW) and record variations in precipitation temperatures (Figures 3 and 9b, Table S4). Calcite precipitation temperatures were calculated using the calcite-water fractionation equation of Friedman and O'Neil (1977) assuming an initial seawater composition of δ 18 O = −1‰ for an ice-free Earth. This yields an overall trend of decreasing temperature from 108°C in the early and ribbon calcite veins to 41°C in the late stage calcite veins. These temperatures agree with results of Schwarzenbach, Früh-Green, Bernasconi, Alt, & Plas (2013) that showed a decrease with age from 151°C to 49°C. This trend is associated with early calcite veins with the lowest δ 13 C values (≈1‰) and 87 Sr/ 86 Sr ratios close to seawater composition, and late calcite veins with δ 13 C values of 2-3‰ and lower 87 Sr/ 86 Sr ratios (Figure 9a).

Alteration History Based on Microstructures
The mineralogical and geochemical characteristics of the studied serpentinites indicate a multiphase alteration process with infiltration of fluids along passive tension fractures and locally focused along fault zones. The chemical zoning of the mesh-texture serpentine after olivine (i.e., Fe-poor rim vs. Fe-rich core) SCHWARZENBACH ET AL. reflects the multistage process of serpentinization. A first stage of alteration likely resulted in the formation of Fe-rich brucite (as an intermediate phase and not preserved) and Fe-poor serpentine, which was followed by increasing H 2 O-and Si-activities that resulted in progressive dissolution and replacement of olivine and Fe-rich brucite by Fe-poor serpentine and magnetite forming the magnetite-bearing veins (e.g., Beard et al., 2009;Boschi, Dini, Baneschi, et al., 2017;Schwarzenbach, Caddick, et al., 2016). A later stage of alteration is indicated by relict olivine in the core of the meshes replaced by Fe-rich serpentine, which suggests that Si-activities increased to the point where brucite and magnetite were instable (e.g., Früh-Green et al., 2004;Grozeva et al., 2017;Klein et al., 2014;Mayhew & Ellison, 2020). This interpretation is supported by local occurrences of talc in the mesh cores as well as fine-grained serpentine-talc intergrowths in the mesh texture ( Figure 4). In addition to serpentinization and extensive carbonation, variations in bulk rock chemistry associated with talc, chlorite, and amphibole with highly variable compositions ( Figure 6) indicate variations in fluid/rock ratios and/or fluid compositions and temperatures during progressive rock alteration (Malvoisin, 2015). Secondary amphibole and/or talc rims around bastites have been described in many oceanic serpentinites and their origins are commonly linked to early stages of mantle hydration associated with the interaction of peridotites with hydrothermal fluids during mantle upwelling (e.g., Agrinier & Girardeau, 1988;Früh-Green et al., 1996). However, in the N. Apennine serpentinites, the amphibole veins and patches cutting the serpentine mesh texture, and amphibole of hornblende to edenite composition replace bastites, which document a later stage of localized fluid infiltration likely at amphibolite-facies conditions (e.g., Schmidt, 1992 Alt, Shanks III, et al., 2012, analyzed for this study. b Samples collected by Molli (1995) and analyzed for this study.
Trace and rare Earth element patterns as well as stable and radiogenic isotope signatures provide ideal tracers for fluid sources and fluid-rock interaction processes. Importantly, major and trace element concentrations in peridotites are also modified by melting and refertilization by melt-rock interaction (Niu, 2004;Paulick et al., 2006). In the following, we first discuss the REE patterns to account for possible melt-rock interaction that could have taken place before, during or after hydrothermal alteration and then combine these observations with isotope signatures to constrain the hydrothermal alteration history and the impact of different fluid sources during serpentinization, carbonation and Si-metasomatism.

Distinguishing Melt-Rock and Fluid-Rock Interaction Processes
Melt impregnation and melting processes are typically reflected by the addition of SiO 2 , REE, and HFSE (e.g., Nb, Ta, Zr, Hf). Melt-rock interaction causes addition of LREE and HFSE to the rock in about equal proportions, whereas interaction with aqueous solutions leads to a more distinct enrichment in LREE than in HREE and HFSE (Paulick et al., 2006). In the Bracco Unit, the chondrite-normalized REE patterns of the ophicalcites are similar to the basement serpentinites ( Figure 7b). HREE (Gd-Lu) patterns are rather flat in both serpentinites and ophicalcites, whereas the LREE (La-Eu) contents are more variable and enriched compared to the unaltered peridotites of the Internal Ligurian units (Rampone et al., 1996). Interestingly, the in situ analyses of the minerals show no LREE enrichment in the serpentinites nor the ophicalcites (Figure 7c).
LREE enrichment in serpentinites compared to unaltered peridotites has been broadly documented in midocean ridge serpentinites and has been associated with the mobility and incorporation of LREE in serpentinites during hydrothermal processes (Niu, 2004;Paulick et al., 2006). Records of hydrothermal processes are preserved in both the serpentine minerals and some ophicalcite bulk rocks that display negative Ce anomalies in the REE patterns ( Figure 7b) and have been linked to the presence of seawater-derived fluids or the oxidation of Fe 2+ during serpentinization (Niu, 2004). However, some researchers (Bodinier & Godard, 2003;Boschi et al., 2006a) suggest that the LREE enrichment observed in ophiolitic and oceanic harzburgites may result from the infiltration of small volumes of evolved, LREE-enriched melts that precipitate minute amounts of barely detectable LREE-rich mineral phases. Because the in situ REE analyses of the serpentine minerals in the N. Apennine sequences show strong LREE depletions, we infer that a smallscale, "cryptic" melt impregnation may have occurred before serpentinization began, and any LREE-enriched phase that contributes to the LREE-enriched bulk rock pattern was not measured by the in situ analyses of the major mineral phases.
SCHWARZENBACH ET AL.

Fluid Sources During Serpentinization and Carbonation
Seawater is the primary fluid source in peridotite-hosted hydrothermal systems, as is reflected by seawater-like 87 Sr/ 86 Sr values found in serpentinites and their carbonate veins, such as from the Atlantic Ocean (e.g., Delacour et al., 2008;Vils et al., 2009). In the N. Apennine ophiolite, geochemical evidence for seawater-infiltration is documented in the carbon isotope compositions (see also Schwarzenbach, Früh-Green, Bernasconi, Alt, & Plas, 2013) and the 87 Sr/ 86 Sr values of some micro-drilled carbonate veins (Figure 9a). The early calcite veins have 87 Sr/ 86 Sr ratios of 0.70661-0.70718 and δ 13 C values around 1‰, both reflecting average mid-to late Jurassic seawater compositions ( 87 Sr/ 86 Sr ≈ 0.7068-0.7070, δ 13 C ≈ 1‰, respectively) (Veizer et al., 1999;Wierzbowski et al., 2017). The effects of seawater infiltration within the serpentinites and ophicalcites are further documented by the enrichment of FMEs with distinct positive anomalies in Cs, U, and Sr ( Figure 7a). These enrichments are typical for seafloor-exposed serpentinites (Deschamps et al., 2012;Peters et al., 2017) and indicate infiltration of seawater-derived fluids during the main stages of serpentinization. The N. Apennine ophicalcites show stronger enrichments in Cs, Sr, and Rb (and variable enrichments in U) (see Figure S1) suggesting overall higher fluid-rock ratios during formation of the ophicalcites compared to the serpentinites.
Oxygen and hydrogen isotope compositions provide further constraints on fluid sources, water-rock ratios, and temperatures during water-rock interaction (e.g., Früh-Green et al., 1996;Sakai et al., 1990). Previous studies have suggested that fluid-rock interaction in the N. Apennine serpentinites took place at ∼150°C-240°C based on bulk rock δ 18 O values of 6.7-9.9‰ and assuming interaction with Jurassic seawater (Schwarzenbach, Früh-Green, Bernasconi, Alt, & Plas, 2013). The δ 18 O WR values of the samples studied here are within a slightly larger range (5.1-14.8‰; Table 1), but overall suggest a similar temperature range of 180°C -280 °C depending on which fractionation factor is used (see Figure S2). Comparison to available δD and δ 18 O data from mid-ocean ridge serpentinites shows considerable overlap (Figure 10a). The range in δ 18 O data mostly reflects variable serpentinization temperatures and variable water-rock ratios, with δ 18 O values >5.5‰ (i.e., mantle values) generally suggesting temperature <250 °C and δ 18 O values <5.5‰ indicating higher temperatures (e.g., Alt, Shanks, et al., 2007;Barnes, Paulick, et al., 2009;Boschi, Bonatti, et al., 2013;Früh-Green et al., 1996;Sakai et al., 1990;Wenner & Taylor, 1973). In addition, the range in δD values of −86 to −59‰ determined in our study provide further evidence that most serpentinites studied to date have strongly negative δD values (Figure 10a). The origin of these negative δD values is still controversial. A common interpretation is that they require interaction with either evolved seawater-derived and/or magmatic waters, which are characterized by negative δD values (Früh-Green et al., 1996;Shanks et al., 1995;Wenner & Taylor, 1973). Alternatively, Früh-Green et al. (1996) proposed that the negative δD values are correlated to the formation of considerable amounts of H 2 due to oxidation of ferrous iron in olivine ± pyroxene. Finally, negative δD values in ophiolites potentially also derive from interaction with meteoric waters, which could have interacted with the serpentinites subsequent to their emplacement on the continent. This has previously been inferred based on highly variable δD values that correlated with mineral textures (Kyser et al., 1999). In this case, the δD values are lowered to more negative values commonly reaching δD values of −100‰ and lower (Barnes, Selverstone, & Sharp, 2006;Kyser et al., 1999). In the N. Apennine samples, the δD values have a relatively narrow range, which overlaps considerably with SCHWARZENBACH ET AL.  the δD and δ 18 O values seen in abyssal serpentinites and serpentinites in OCT settings (Agrinier & Girardeau, 1988;Barnes, Paulick, et al., 2009;Boschi, Bonatti, et al., 2013;Früh-Green et al., 1996) (Figure 10a). In addition, we find no evidence for past or current interaction by meteoric waters (as e.g., seen in the Oman ophiolite (Kelemen et al., 2020)), nor do we find mineralogical evidence for serpentine recrystallization or surface weathering that would indicate either interaction with meteoric fluids or the effect of low-grade metamorphic overprint. Sample recovery from quarries also allowed for fresh sample material to be collected. Thus, we argue that the measured δD values are most likely the result of seafloor processes and that interaction with an evolved seawater-derived fluid possibly with a magmatic component resulted in the negative δD values and affected both bulk serpentinites and ophicalcites in about equal proportions (Früh-Green et al., 1996;Shanks et al., 1995).

Silica Metasomatism
The abundance of talc (macroscopically seen in Figures 2b and 3) and amphibole in the ophicalcites as well as serpentine-talc intergrowths in the serpentinite mesh textures (Figure 4) document an increase in SiO 2 bulk content from the basement serpentinites to the ophicalcites, and suggests localized Si-metasomatism. Si-metasomatism in the ophicalcites is also expressed by lower MgO/SiO 2 ratios ( Figure 6) compared to the basement serpentinites, as is seen in a plot of MgO/SiO 2 vs. Al 2 O 3 /SiO 2 (Figure 11), where the terrestrial array is defined by the successive magmatic depletion of a primitive mantle during partial melting (Hart & Zindler, 1986;Jagoutz et al., 1979). An off-set to lower MgO/SiO 2 values in most abyssal serpentinites has previously been related to partial MgO loss during seafloor weathering through brucite dissolution and incongruent olivine dissolution (Niu, 2004;Paulick et al., 2006;Snow & Dick, 1995). However, the macroscopic abundance of talc in the Bracco ophicalcites leads us to infer that the lower MgO/SiO 2 ratios are more likely the result of Si addition rather than extensive Mg loss (Malvoisin, 2015).
Si-metasomatism and/or talc formation have previously been documented in peridotites from the MAR, including ODP Leg 209 near the 15°20′N fracture zone (Paulick et al., 2006) and the AM (Boschi et al., 2006a(Boschi et al., , 2006b. In both cases, alteration is associated with the input of fluids derived from peridotite or gabbro alteration at depth, which resulted either in pervasive talc alteration in the peridotites (Paulick et al., 2006) or by localized influx of oxidizing, Si-Al-Ca-rich fluids produced talc, amphibole, and chlorite schists along fault zones (e.g., at the AM; Boschi et al., 2006a;Boschi et al., 2006b). Accordingly, we suggest that in the rocks studied here, Si-metasomatism could have been caused by two possible mechanisms: 1) seawater mixing with fluids generated from the serpentinization of peridotites, specifically pyroxenes, at increasing water-rock ratios or 2) interaction with fluids with a component derived from a mafic source. Both processes have been thermodynamically modeled by Malvoisin (2015) and resulted in formation of talc coexisting with either serpentine or carbonate. Considering the amount of talc precipitated as druses in the ophicalcites and the shift to higher SiO 2 bulk rock contents in most ophicalcites, we tend toward an externally derived fluid to account for the late stage Si-metasomatism.
SCHWARZENBACH ET AL. Isotope ranges for unaltered mantle (5.8 ± 0.3‰) and abyssal, ocean-continent transition zones (OCT), subduction zone and arc settings (SZ), and ophiolitic serpentinites (including subducted ophiolites) from Agrinier and Girardeau (1988), Barnes, Eldam, et al. (2013), Barnes, Paulick, et al. (2009), Boschi, Bonatti, et al. (2013), Früh-Green et al. (1996, Sakai et al. (1990), Taylor (1973, 1974 Sr/ 86 Sr compositions than Jurassic seawater, suggesting the influence of fluids with more radiogenic Sr. Jurassic seawater and mantle as in Figure 8. Higher enrichments in Sr, Rb, and Cs in the ophicalcites provide further evidence for a mafic source of the fluid, such as gabbroic intrusions or basaltic lavas (von Damm, 1990) (Figure 7a). Interestingly, Ba, which is considered to be highly mobile in high-temperature fluids and can be enriched in fluids derived from young gabbroic intrusions, is not markedly enriched. In addition, there is a lack of metal enrichments (e.g., Cu or Zn), which is also typical for high-temperature fluid-rock interaction (e.g., Niu, 2004;von Damm, 1990). This would suggest that the fluids that produced the Si-metasomatism in the ophicalcites were dominantly lower temperature fluids and hence, would not have affected Ba and metal concentrations. Additional evidence for mafic-derived fluids is provided by the Sr isotope signatures. The final formation of talc filling druses within pre-existing carbonate veins in the ophicalcites ( Figure 3) correlates with an increase in the 87 Sr/ 86 Sr ratios from seawater-like compositions in the early calcite veins toward mantle or MORB-like 87 Sr/ 86 Sr ratios in the later stage carbonate veins (Figure 9c). In addition, we calculated simple isotope mixing trajectories between serpentinites and mafic-derived end-member fluids (Figure 10b). For this we used different Sr concentrations for the serpentine (4-15 μg/g), the highest serpentinite compositions ( 87 Sr/ 86 Sr = 0.7085) and the average of the analyzed gabbroic rocks ( 87 Sr/ 86 Sr = 0.7037), which is assumed to approximate the composition of the mafic-derived fluids. All the ophicalcites and Cava dei Marmi fault schists indicate variable proportions of mixing between the two end-members. Similarly, most late calcite veins have 87 Sr/ 86 Sr ratios between those of Jurassic seawater and the gabbros (Figures 9 and 10b). We infer that the trend to lower 87 Sr/ 86 Sr ratios reflect increasing influx of mafic-derived fluids with ongoing rock fracturing associated with more pronounced Si-influx in the ophicalcites than the underlying serpentinites. Associated δ 13 C-values of 2-3‰ overlap with values from carbonate vent samples, breccias, and carbonate infillings from the LCHF and are attributed to the input of 13 C-enriched carbon associated with ongoing microbial activity (e.g., methanogenesis) (Früh-Green et al., 2003). An abundance of methanogenesis agrees with the decreasing temperatures preserved in the δ 18 O carb values (Figure 9b), which increasingly favors microbial activity. Alternatively, they may be the result of closed system fluid evolution of the dissolved inorganic carbon during water-rock interaction with the mafic lithologies.
The Cava dei Marmi fault schist that displays a high δ 18 O serp value yielded low 87 Sr/ 86 Sr ratios (Figure 10b), which are in the same range as the ophicalcites, suggesting that these fault schists and the ophicalcites were altered by the same or compositionally similar fluids. In contrast, the Libiola fault schist is characterized by elevated 87 Sr/ 86 Sr ratios (up to 0.71847), and based on the oxygen composition, has the highest formation temperatures (lowest δ 18 O value). Such compositions may imply that a sedimentary influx took effect during hydrothermal alteration on the ocean floor or that the schists may have been reactivated during Alpine orogenesis.

Evidence for Fluid Input From Continental Basement Rocks
Interestingly, most of the serpentinites have higher 87 Sr/ 86 Sr values than Jurassic seawater (Figure 8), which varied between 0.7068 and 0.7072 in the Middle to Late Jurassic (Wierzbowski et al., 2017), whereas 87 Sr/ 86 Sr values above 0.7078 were only reached at ∼74 Ma (Jones et al., 1994). 87 Sr/ 86 Sr values above seawater compositions have also been documented in some serpentinites from the MAR (e.g., Mével, 2003;Snow et al., 1993) suggesting a 87 Sr reservoir during hydrothermal circulation. Snow et al. (1993) argue that high 87 Sr/ 86 Sr ratios could be generated by infiltration of detrital sediments penetrated through micro-cracks into serpentinites. This process is not directly related to serpentinization but can considerably alter the 87 Sr/ 86 Sr compositions of abyssal serpentinites (Mével, 2003). Thus, one possibility would be that interaction of fluids with the sedimentary Framura Breccia overlying the ophicalcites could have had an effect on the serpentinites. However, the ophicalcites, which are underlying the sedimentary Framura Breccia and overlying the SCHWARZENBACH ET AL.  (Boschi et al., 2006) AM serpentinites (Boschi, 2006) Global data set (Niu et al., 2004) Serpentinites Ophicalcites

Cava dei Marmi Fault schists
Internal Ligurides (Rampone et al., 2006) Ter res tria l Arr ay Talc alteration Brucite serpentinites, do not show this more radiogenic Sr enrichment. Alternatively, continental basement rocks have high 87 Sr/ 86 Sr ratios (>0.720 to 0.750; Allègre, 2008) similar to sedimentary rocks and, thus, fluids that interacted with continental basement could have modified the 87 Sr/ 86 Sr ratios of the serpentinites. Recent models suggest that continental basement was likely present during initial stages of rifting and opening of the Piemont-Ligurian ocean, where subcontinental mantle was sporadically exposed to seawater in a narrow ocean basin (Le Breton et al., 2021). Similarly, felsic continental crust associated with subcontinental mantle in the N. Apennines have been interpreted as evidence for the early stages of continental breakup and opening of the oceanic basement within an OCT (Marroni et al., 1998) and have been compared to observations from the Iberia-Newfoundland margins where tilted continental basement blocks and oceanic crust periodically occur over a distance of 150 km within the OCT (McCarthy et al., 2018;Whitmarsh et al., 2001). Such a setting would allow infiltration of fluids that had interacted with continental basement rocks during initial stages of peridotite hydration, shifting the 87 Sr/ 86 Sr ratios of the serpentinites to higher values. This interpretation, however, assumes that the early continental Sr isotope signatures are preserved throughout the hydrothermal evolution of the system. Alternatively, we would have to envisage exchange with continental rocks and preferential fluid circulation solely in the serpentinites during the Alpine orogeny, for which we have no further evidence.

Hydrothermal Evolution of the Northern Apennine Ophiolite
The data presented above suggest a complex alteration history and indicate that the serpentinites were dominated by different fluids than the overlying ophicalcites, but all record multiple phases of water-rock interaction, some of which result in pervasive alteration whereas other events were more localized (e.g., forming distinct veins). Using our new data with previously published interpretations, we reconstruct the temporal and chemical evolution of the studied rocks and infer the alteration history of the N. Apennine ophiolite sequence.
The formation of tremolite to hornblende rims around pyroxene suggests that the peridotites of the N. Apennine were initially hydrated at relatively high temperatures likely during initial mantle upwelling. These stages of mantle hydration as well as continuous mantle exhumation were affected by fluids that had reacted with either the continental basement or with sediments prior to peridotite infiltration ( Figure 12-stage 1). Since continental basement rocks and sedimentary sequences have considerably more radiogenic Sr compared to peridotites, even low amounts of fluid that had interacted with basement rocks or sediments would have substantially shifted the 87 Sr/ 86 Sr values of the serpentinized peridotites to higher values. The serpentinites from the Bracco and the Val Graveglia Units both show the same trends further supporting wide-spread and relatively pervasive fluid infiltration with a more radiogenic Sr signature. Seawater influx was overall relatively restricted as suggested by the lesser enrichment of FMEs such as Cs, Rb, Sr, U in the serpentinites compared to the ophicalcites. Based on the flat REE patterns of the peridotites cryptic melt impregnation likely also affected these rocks.
Serpentinization was also associated with the local influx of higher temperature fluids. This is documented by the formation of edenite and Mg-hornblende, indicating amphibolite-facies conditions, that form veins together with chlorite or replace bastites. In addition, high-temperature fluid influx is recorded by sulfur isotope signatures. Using in situ δ 34 S values in sulfide minerals and bulk rock multiple sulfur isotope signatures, Schwarzenbach, Gill, and Johnston (2018) document the impact of thermochemical sulfate reduction, which took place at ∼350°C-400°C and was associated with input of H 2 S leached from mafic rocks. This stage of alteration likely took place during continuous uplift of the peridotites during the main stage of serpentinization as some of the amphibole-chlorite-veins cut the serpentine mesh texture. Gabbroic intrusions most likely provided the heat for high-T fluid circulation (Figure 12-stage 2). At a similar time, the Cava dei Marmi fault schists likely formed within fault zones during peridotite exposure, similar to the processes described at the AM (e.g., Boschi et al., 2006a;Früh-Green et al., 2018;Rouméjon et al., 2018).
By continuous uplift of the peridotites, these rocks were exhumed to the seafloor where serpentinization fluids mixed with seawater in cracks and fractures producing carbonate veins in the serpentinites and ophicalcites ( Figure 12-stage 3). These veins document temperatures decreasing from 150°C to 41°C with increasing seawater influx (this study and Schwarzenbach, Früh-Green, Bernasconi, Alt, & Plas, 2013). As the system cooled, microbial activity took place as documented by sulfur isotope compositions in sulfide minerals (Schwarzenbach, Gill, & Johnston, 2018) and the δ 13 C values of the carbonates that point to methanogenesis. At this stage the highly oxidized ophicalcites would have been directly exposed to seawater. Extensive and pervasive interaction with seawater in the ophicalcites is indicated by greater enrichments in FMEs (Cs, Rb, Sr, and locally U) compared to the serpentinites. However, some of this enrichment may also be linked to later Si-metasomatism.
Interestingly, most of the Sr values of the calcite veins in the ophicalcites are significantly below Jurassic seawater and only a few of the veins reflect unmodified seawater based on their 87 Sr/ 86 Sr ratios. As discussed above, these low 87 Sr/ 86 Sr ratios can be correlated with the Si-metasomatism recorded in the ophicalcites and are likely produced by Si input from mafic rocks. However, the serpentinites underlying the ophicalcites have significantly higher 87 Sr/ 86 Sr ratios than the ophicalcites suggesting that the Si-metasomatism was most pronounced in the top most sections and decreased downwards, consistent with the presence of talc filling late stage carbonate veins. Underlying gabbro intrusions are therefore unlikely the fluid source. The most likely scenario is that the fluids leading to Si-metasomatism were derived from interaction with basalts, which are locally found overlying the ophicalcites (B. E. Treves and Harper, 1994). Consequently, we infer that seawater-derived fluids infiltrated the basalts after their emplacement, picked up the 87 Sr/ 86 Sr signal of the basalts and then infiltrated the ophicalcites to produce talc in druses within preexisting carbonate veins. As the fluids would have come from the top, restricted fluid infiltration associated with Si-metasomatism would not have affected the underlying serpentinites. In addition, it should be noted that the serpentinite protolith of the ophicalcites would initially have had an 87 Sr/ 86 Sr ratio above Jurassic seawater and would have later entirely shifted (during their conversion from serpentinites to ophicalcites) toward basalt-like  Manatschal & Müntener, 2009). Stage 1: initial stages of serpentinization by fluids with radiogenic Sr e.g., from continental basement rocks or sediments. Stage 2: continuous uplift and serpentinization with local influx of high temperature fluids (derived from gabbroic intrusions) producing rare talc, chlorite, and amphibole with hornblende to edenite compositions as veins or replacing pyroxene. Stage 3: final stages of alteration and exposure to seawater documented by carbonate veins with seawater-like Sr isotope compositions. Subsequent extrusion of basaltic lavas on top of the ophicalcites and continuous water-rock interaction led to Si metasomatism (with SiO 2 provided by the lavas) and talc formation predominantly in the ophicalcites. See text for discussion.

Conclusions
This study summarizes the extensive alteration history of ophicalcites and serpentinites in the N. Apennine ophiolite. The mineralogical and geochemical signatures of the serpentinites and ophicalcites record extensive metasomatism by fluids with variable chemical compositions and a cryptic melt impregnation pre-dating hydrothermal alteration. In particular, the chemical variations attest to fluids of variable origin, i.e., seawater-derived fluids that have interacted with different lithologies and imprint a chemical signature on the ultramafic basement as it is undergoing hydration. The serpentinites thereby record 1) hydration processes during early mantle upwelling within an opening ocean basin with input of a continental basement-derived fluid-as preserved in the high 87 Sr/ 86 Sr ratios; and 2) localized high-temperature fluid influx likely associated with continuous mantle exposure and local gabbroic intrusion as documented by amphiboles with hornblende to edenite composition and sulfide isotope compositions as previously reported. In contrast, the ophicalcites mostly record the late stages of seawater-rock interaction with extensive carbonation and interaction with basalt-derived fluids at low-temperatures as the mantle sequence was emplaced at the ocean floor, resulting in a gain of SiO 2 , CaO, and FMEs. Although all studied rocks record multiple phases of water-rock interaction, their chemical signatures are dominated by different and distinct events. Accordingly, the transition from serpentinite to ophicalcite as preserved in the N. Apennine ophiolite records the history of uplift and exposure on the seafloor. In addition, this alteration sequence documents that serpentinites can undergo a very complex hydrothermal evolution from serpentinization, carbonation to Si-metasomatism. The complex hydrothermal evolution of these systems is essential to understand the chemical and petrophysical evolution of the oceanic lithosphere and processes of mass transfer. Extensive carbonation and Si-metasomatism may eventually affect subduction zone processes, i.e., change the expected rheology, for example, due to higher Si contents, and mineral stabilities within the subducting slab and the subduction zone channel, and as a consequence may affect element cycling within subduction zones.

Data Availability Statement
All data sets for this research are included in this paper and its supplementary information files Tables S1-S5. In addition, all supplementary data is available online on PANGAEA at https://doi.org/10.1594/ PANGAEA.921751.

Acknowledgments
We thank Giancarlo Molli for help during fieldwork and for offering the gabbro samples for this study. We also thank Peter Ulmer and Derek Vance for their contribution to the interpretation of the data. We are grateful to Corey Archer, Giuditta Fellin, Ilaria Baneschi, Benita Putlitz, Ruth Hindshaw, and Gregory de Souza for their valuable help during the analytical procedures. This work was supported by Swiss SNF Grants 200021-134947 and 200020-146886 to Früh-Green. This contribution benefited greatly from the constructive comments by the associate editor, and extensive reviews by Benjamin Malvoisin and an anonymous reviewer, whom we gratefully thank. Open Access enabled and organized by Projekt DEAL.