Geochemistry and Microtextures of Vein Calcites Pervading the Izu‐Bonin Forearc and Rear Arc Crust: New Insights From IODP Expeditions 352 and 351

Abstract International Ocean Discovery Program Expeditions 352 and 351 drilled into the Western Pacific Izu‐Bonin forearc and rear arc. The drill cores revealed that the forearc is composed of forearc basalts (FAB) and boninites and the rear arc consists of FAB‐like rocks. These rocks are pervaded by calcite veins. Blocky vein microtextures enclosing host rock fragments dominate in all locations and suggest hydrofracturing and advective fluid flow. Significant diffusion‐fed and crystallization pressure‐driven antitaxial veining is restricted to the rear arc. The lack of faults and presence of an Eocene sedimentary cover in the rear arc facilitated antitaxial veining. Rare earth element and isotopic (δ18O, δ13C, 87Sr/86Sr, and Δ47) tracers indicate varying parental fluid compositions ranging from pristine to variably modified seawater. The most pristine seawater signatures are recorded by FAB‐hosted low‐T (<30 °C) vein calcites. Their 87Sr/86Sr ratios intersect the 87Sr/86Sr seawater curve at ~35–33 and ~22 Ma. These intersections are interpreted as precipitation ages, which concur with Pacific slab rollback. Boninite‐hosted low‐T (<30 °C) vein calcites precipitated from seawater that was modified by fluid‐rock interactions. Mixing calculations yield a mixture of >95% seawater and <5% basaltic 87Sr/86Sr. In the rear arc, low‐T rock alteration lowered the circulating seawater in δ18O and 87Sr/86Sr. Thus, vein calcites precipitated from modified seawater with up to 20–30% basaltic 87Sr/86Sr at temperatures up to 74 ± 12 °C. These results show how the local geology and vein growth dynamics affect microtextures and geochemical compositions of vein precipitates.


Introduction
The knowledge of the stratigraphy, structure, and geochemical composition of the oceanic crust is based on ophiolites and drill cores from the International Ocean Discovery Program (IODP) and its predecessors. Since ophiolites mostly experienced a complex history of uplift and emplacement on continental crust, their structures and geochemical compositions may have been overprinted and/or altered. In situ drilling through oceanic crust can provide more pristine rocks and structures that have not been involved in uplift and emplacement processes. With the aim of investigating the tectonomagmatic processes of subduction initiation, IODP Expedition 352 drilled into the Izu-Bonin forearc crust in the Western Pacific ( Figure 1) and successfully cored 1.22 km of volcanic rocks (Reagan et al., 2015). The four drill sites revealed two well-preserved geochemically distinct volcanic rock sequences-forearc basalts (FAB) and boninites-each related to different stages after subduction initiation of the Pacific Plate beneath the Philippine Sea Plate in the early Eocene (Reagan et al., 2017;Reagan et al., 2019). Drill cores that were recovered during the preceding IODP Expedition 351 ( Figure 1) unveiled that the Izu-Bonin rear arc crust is composed of FABlike rocks but lacks boninites (Arculus, Ishizuka, Bogus, Gurnis, et al., 2015;Arculus, Ishizuka, Bogus, & the Expedition 351 Scientists, 2015;Hickey-Vargas et al., 2018;Ishizuka et al., 2018).
Here we present a multidisciplinary study on vein calcites pervading the volcanic units of the Izu-Bonin forearc (IODP Expedition 352) and rear arc (IODP Expedition 351), which links geochemical with microtextural data. This study (1) investigates the physicochemical conditions of fluid flow and vein calcite formation, (2) evaluates the extent of fluid-rock interaction, (3) constrains the precipitation ages of seawater-derived vein calcites, (4) examines if the different calcite vein types correlate with specific geochemical signatures, and (5) compares the postmagmatic evolution of the Izu-Bonin forearc and rear arc. For this purpose, different geochemical tracers and proxies are used, and the local geological characteristics of the forearc and rear arc are considered. REE+Y concentrations, as well as radiogenic strontium ( 87 Sr/ 86 Sr), stable oxygen (δ 18 O) and  (Reagan et al., 2015) and IODP Expedition 351 Site U1438 situated in the rear arc (Amami-Sankaku basin) (Arculus, Ishizuka, Bogus, & the Expedition 351 Scientists, 2015). (b) Schematic E-W cross section through the Izu-Bonin forearc indicating the lithological succession and drill sites (Reagan et al., 2015). carbon (δ 13 C), and clumped (Δ 47 ) isotopic compositions record the geochemical compositions of circulating fluids and temperature of vein mineral precipitation. The new clumped isotope method (Ghosh et al., 2006) has been rarely applied on secondary calcites from the oceanic crust (Coogan et al., 2019;Quandt, Micheuz, Kurz, Kluge, et al., 2019). In contrast to δ 18 O thermometric calculations, clumped isotopes yield precipitation temperatures without requiring knowledge of the parental δ 18 O fluid composition, which in turn may be calculated (Ghosh et al., 2006). Intersections of 87 Sr/ 86 Sr ratios of purely seawater-derived vein calcites with the well-established Sr isotope seawater curve (McArthur et al., 2001) give estimates on the time of calcite precipitation (e.g., Hart & Staudigel, 1978). These geochemical tracers are combined with Raman spectroscopy, and thin section and cathodoluminescence (CL) petrography, which provide insights into the vein mineralogy and modes of fracturing, fluid flow, and fracture sealing. This combination of geochemical and microtextural data improves the understanding of vein formation in the oceanic crust.

Geological Setting of the Izu-Bonin Forearc and Rear Arc
The prominent seafloor topography of the Izu-Bonin trench in the Western Pacific reflects the steep subduction of the Pacific Plate under the Philippine Sea Plate over a distance of 3,000 km (e.g., Bloomer et al., 1995;Faccenna et al., 2018;Holt et al., 2018;Stern & Bloomer, 1992). In order to better understand the initiation of this intraoceanic subduction zone, IODP Expedition 352 penetrated the volcano-stratigraphic succession of the Izu-Bonin forearc crust (Reagan et al., 2015). Four sites revealed FAB (Sites U1440 and U1441) and boninites (Sites U1439 and U1442), which are stratigraphically and geochemically distinguishable ( Figure 2) (Reagan et al., 2015;Reagan et al., 2017;Reagan et al., 2019). Each volcanic rock sequence formed as a response to the consecutive stages of subduction initiation and represents a modern analog to many suprasubduction zone ophiolites (Reagan et al., 2017;Reagan et al., 2019).

Vein Macrostructures and Host Rock Alteration
Macroscale vein structures and host rock alteration characteristics were described by Arculus , Ishizuka, Bogus, & the Expedition 351 Scientists (2015), Reagan et al. (2015), and Hickey-Vargas et al. (2018) and are summarized here. The drill cores revealed that the Izu-Bonin forearc and rear arc sites are pervaded by vertical to subvertical (mostly >45°) veins at different depths. Veins document mineralized (1) (tapered) tension fractures; (2) hybrid (tension and shear) fractures; (3) tension gashes linked to releasing bends and extensional step-overs, which are arranged along shear fractures; and (4) extensional horsetail splay faults emanating from single sets of shear fractures with top-down displacement (normal sense of shear). These structures occur independently of the drill site and depth ( Figure 2) and may be conjugated, constituting  (Arculus, Ishizuka, Bogus, Gurnis, et al., 2015;Arculus, Ishizuka, Bogus, & the Expedition 351 Scientists, 2015;Reagan et al., 2015;Reagan et al., 2017;Reagan et al., 2019) and occurrence of vein types with depth in meters below seafloor (mbsf). complex vein networks. The typical proportion of veins within a rock volume is <10% but may increase up tõ 75% where fault-related breccias are cemented.
The extent of host rock alteration varies and depends on drill site and lithology. In general, microcrystalline groundmasses, olivine, orthopyroxene, and plagioclase phenocrystals may be replaced by smectite group clay minerals, zeolites, and calcite. Glass may be fresh, devitrified, or altered to clay minerals, zeolites, and palagonite. Chlorite is an additional alteration product in FAB-like rocks at Site U1438 where alteration is localized and variable. At Site U1439 the degree of alteration of boninites increases with depth. Where host rocks are intensely brecciated, pervasive alteration resulted in significantly modified geochemical host rock compositions. FAB at Site U1440 are comparatively fresh. More intense alteration is restricted to the transition zone and dolerite dykes. At Site U1441 FAB alteration is variable and decreases with depth. In contrast to boninites, FAB and FAB-like rocks show argillaceous vein selvedges and vein-parallel alteration halos. Hence, alteration in FAB and FAB-like rocks tends to be more localized, whereas boninites were rather pervasively altered.

Sample Preparation
Raman spectroscopy and CL petrography were performed on polished thin sections. Rock slices used for thin section preparation were subsequently used for laser ablation analyses. For stable, radiogenic Sr, and clumped isotope analyses, fine-grained powders of vein calcites were obtained using a computer-controlled (x-y-z position) micromill device (ESI, Portland) with 100 μm diameter (equipped with a digital camera) or a handheld dentist drill at the NAWI Graz Geocenter, Institute of Earth Sciences, University of Graz (Austria). Fine-grained powders necessary for all isotope analyses were produced by separate drilling into vein calcites hosted in rock slices. Except for a few reported samples, this means that every analysis is based on a different sample powder even if two analyses belong to the same sample. This enabled sampling of different calcite generations and vein branches from the same sample. It also facilitated the detection of potential isotopic heterogeneities. The low thickness of some veins and large amount of powder necessary for Sr isotopic analyses required separate drilling spots in order to obtain sufficient sample powder and to avoid contamination with host rock material.

Raman Spectroscopy
Confocal Raman spectra indicative of vein mineral phases were obtained with a HORIBA Jobin Yvon LabRam-HR 800 Raman microspectrometer at the NAWI Graz Geocenter, Institute of Earth Sciences, University of Graz (Austria). Crystals within polished thin sections were excited at room temperature with the 532 nm emission line of a 100 mW Nd-YAG laser through an OLYMPUS 100X objective (numerical aperture 0.9). The laser spot on the surface had a diameter of approximately 2 μm and a power of about 20 mW. Light was dispersed by a holographic grating with 1,800 grooves/mm. The slit width was set to 100 μm. The dispersed light was collected by a 1,024 × 256 nitrogen-cooled, open-electrode charge-coupled device detector. Spectra were recorded unpolarized. Band shifts were calibrated by regularly adjusting the zero position of the grating and controlled by measuring the Rayleigh line of the incident laser beam. The detection range of the Raman shift lies between 100 and 1,200 cm −1 using an acquisition time between 10 and 20 s.

Cathodoluminescence
CL of polished and carbon-coated thin sections was analyzed using a hot cathode CL microscope (Lumic HC5-LM) at the NAWI Graz Geocenter, Institute of Earth Sciences, University of Graz (Austria). A voltage of 13-14 kV accelerated electrons under a vacuum of <10 −5 mbar and produced true-color CL in the visible spectrum. CL photographs were acquired in real time using an attached digital camera.

Stable Oxygen and Carbon Isotopes
A set of 38 representative samples was chosen for stable oxygen and carbon isotope analyses. After the reaction of the sample powders with oversaturated 100% orthophosphoric acid at 70°C in a Kiel II automated reaction system, stable oxygen and carbon isotopes were measured using a Delta Plus isotope ratio mass spectrometer at the NAWI Graz Geocenter, Institute of Earth Sciences, University of Graz (Austria). The powders of further 22 samples were reacted with phosphoric acid at 75°C and subsequently measured by continuous-flow isotope ratio mass spectrometry using a GasBench II coupled to a Finnigan DELTAplus XP mass spectrometer at the JR-AquaConSol GmbH, Graz (Austria). The reproducibility of replicate analyses for standards  and samples of both measurements was better than ±0.15‰ Vienna Pee Dee Belemnite (VPDB) for δ 18 O and ±0.1‰ VPDB for δ 13 C.
Precipitation temperatures were calculated assuming equilibrium precipitation from a parental fluid with a δ 18 O value of −1‰ Vienna Standard Mean Ocean Water (VSMOW) (e.g., Epstein et al., 1953;McCrea, 1950;Urey, 1947) corresponding to ice-free Eocene seawater (Burgess et al., 2008;Zachos et al., 2001) and using the calcite-water fractionation curve of Friedman and O'Neil (1977), which is valid for temperatures between 0 and 500°C. Additional parental fluid δ 18 O values of 0 and +1‰ VSMOW were also applied in order to consider potential 18 O depletion or enrichment due to fluid-rock interactions or fluid mixing.

Clumped Isotopes
The clumped isotope compositions of vein calcites were determined at the Department of Earth Sciences, Geological Institute, ETH Zurich (Switzerland) using a Thermo Fisher Scientific 253Plus mass spectrometer coupled to a Kiel IV carbonate preparation device, following the method described in Schmid and Bernasconi (2010), Meckler et al. (2014), and Müller et al. (2017). The Kiel IV device included a PoraPakQ trap to eliminate potential organic contaminants. Prior to each sample run, the pressure-dependent backgrounds were determined on all beams to correct for nonlinearity effects in the mass spectrometer according to Bernasconi et al. (2013). During each run, 18 replicates of 130-150 μg of different samples and 6 replicates of each of the four carbonate standards, ETH-1, ETH-2, ETH-3, and ETH-4 (Bernasconi et al., 2018), were analyzed. Each analysis consisted of 18 cycles of sample and reference gas comparison, with 26 seconds integration. The calculations and corrections were done with the software Easotope (John & Bowen, 2016) using the revised "Brand parameters" for 17 O correction as suggested by Daëron et al. (2016). Individual sample replicate and standard measurements are archived at PANGAEA Data Archiving and Publication (Quandt, Micheuz, Kurz, Bernasconi, et al., 2019).
Temperatures were calculated using the Kele et al. (2015) calibration recalculated with the "Brand parameters" and the new accepted values for the ETH standards as reported in Bernasconi et al. (2018). We use this calibration because it has been produced with the same instrumentation and data reduction schemes as the data presented in this study and thus guarantees robust temperature estimates avoiding uncertainties related to poor interlaboratory comparability (Petersen et al., 2019).
In contrast to the δ 18 O geothermometer, the clumped isotope method yields temperature estimates without the necessity to know the parental fluid δ 18 O value (e.g., Ghosh et al., 2006). Conversely, this parental fluid δ 18 O value was calculated based on the clumped isotope temperature using the calibrations of Coplen (2007) and Kim and O'Neil (1997). Both calibrations result in a range of parental fluid δ 18 O values for every sample, which reflects the ongoing debate on the isotopic fractionation factor.
A theoretically long heating rate of ≥90-100°C over tens of Myr is necessary to reorder the 18 O-13 C bonds of calcite (Passey & Henkes, 2012;Stolper & Eiler, 2015). Because of this long duration compared to the relatively short phase of magmatic activity (~2 Myr) in the Izu-Bonin forearc (Ishizuka, Tani, et al., 2011;Reagan et al., 2013;Reagan et al., 2019), we suggest that the clumped isotope temperature and thus calculated parental fluid δ 18 O values have not been affected by reordering.

Radiogenic Strontium Isotopes
Prior to Sr separation via ion exchange column chemistry, 15-30 mg powder per sample were dissolved in double-distilled 0.4 M HCl in order to avoid dissolution of noncarbonate vein or host rock constituents. Subsequently, Sr separation was performed via ion exchange column chemistry using Sr-specific extraction chromatographic resin (Eichrom®, USA) and diluted double-distilled 3 M HNO 3 for column conditioning. Five milliliter of 3 M HNO 3 and 1 ml of 0.1 M HNO 3 were used for element elution and Sr collection, respectively (for details, see Stammeier et al., 2018).
Strontium isotope measurements were carried out in wet-plasma mode using a Nu Plasma II multicollector inductively coupled plasma mass spectrometer (Nu instruments, Wrexham, UK) equipped with a 0.1 ml/min MicroMist® glass nebulizer and an ASX-110FR Flowing Rinse MicroAutosampler (Teledyne-Cetac Technologies) at the NAWI Graz Geocenter, Central Lab for Water, Minerals and Rocks, Graz University of 10.1029/2019GC008745 Geochemistry, Geophysics, Geosystems QUANDT ET AL. Technology (Austria). In summary, samples were measured (1) with sensitivities for 88 Sr being generally better than 20 V per 500 μg/ml, (2) as one block of 25 cycles with an integration time of 5 s, and (3) with background determined on half-masses prior to each block measured. Measured isotope ratios were internally normalized to an 86 Sr/ 88 Sr ratio of 0.1194. The analytical uncertainty of the 87 Sr/ 86 Sr measurement is ±0.00001, and repeated analysis of NIST NBS 987 (National Institute of Standards and Technology, Gaithersburg, USA) within the analytical session gave 87 Sr/ 86 Sr ratios of 0.710253 ± 0.000067 (2 SD; n = 12). Total procedural blanks below 1.2 ng Sr are negligible compared to analyte signals.
The decay of 87 Rb to 87 Sr affects the 87 Sr/ 86 Sr sample ratio and may require the calculation of the initial 87 Sr/ 86 Sr ratio at the time of sample formation. However, Rb is usually not incorporated into the calcite lattice (McArthur et al., 2012) and analyzed Rb concentrations are <1 ppm. In addition, the decay rate of Rb is low compared to the maximum calcite formation ages, which are constrained by the emplacement of host rocks in Eocene times. Therefore, calculation of initial 87 Sr/ 86 Sr ratios at the time of calcite formation is not required.

Rare Earth Elements and Yttrium
Prior to the analysis, rock slices containing calcite veins were polished and cleaned. REE+Y concentrations of vein calcites were determined using a 193 nm New Wave ArF excimer laser coupled to an Agilent 7500cx inductively coupled plasma mass spectrometer (LA-ICP-MS) at the NAWI Graz Geocenter, Central Lab for Water, Minerals and Rocks, Graz University of Technology (Austria). At a frequency of 7 Hz, the laser ablated between 6 and 20 spots per sample. Each laser spot was set to 75 μm diameter and placed along profiles through the vein samples. The acquisition time was set to a 30 s gas blank followed by a 60 s dwell time for each spot. The NIST standard reference material SRM 612 (National Institute of Standards and Technology, Gaithersburg, USA) was used to standardize the LA-ICP-MS analyses (Jochum et al., 2012, and references therein) before and after a block of 20 sample spot measurements. SRM 610 and 614 (National Institute of Standards and Technology, Gaithersburg, USA) and synthetic carbonate standard MACS-3 (U.S. Geological Survey) were analyzed as unknowns to monitor accuracy. All reference materials were reproduced for most elements within errors for the used elements. The relative uncertainty of REE+Y analyses of the NIST SRM 612 standard measurements was <5%. Time-resolved LA-ICP-MS data reduction was carried out using the GLITTER 4.0 software (Macquarie University, Sydney). The element concentrations are archived at PANGAEA Data Archiving and Publication (Quandt, Micheuz, Kurz, Bernasconi, et al., 2019).
In order to detect potential contamination of vein calcites during LA-ICP-MS analyses due to the ablation of clay minerals and/or zeolites, thresholds of Al and Zr concentrations were applied. Al concentrations <2000 ppm and Zr concentrations <5 ppm are interpreted as uncontaminated calcite analyses (Schier et al., 2018). Most analyzed samples have Zr and Al concentrations below the respective thresholds. Only 10 out of 135 laser ablation spot analyses reveal Al concentrations slightly above the Al threshold (Al <8,400 ppm). Most spot analyses with Al >2,000 ppm show similar REE+Y distribution patterns as spot analyses with Al <2000 ppm from the same sample. Therefore, these analyses are considered as representative and were not rejected.

Vein Mineralogy
Calcite is the dominant vein mineral throughout the drill cores from the Izu-Bonin forearc and rear arc. Other common mineralogical phases include (1) euhedral zeolites that typically occur as early-stage precipitates preceding calcite, (2) late-stage zeolite veins crosscutting calcite veins, (3) accessory palagonite and opaque minerals associated with calcite, (4) narrow argillaceous vein selvedges occasionally accompanied by alteration halos, and (5) host rock solid inclusion bands and trails. In the Izu-Bonin rear arc, zeolites are completely absent. The only noncarbonate phases in rear arc veins are argillaceous vein selvedges and host rock bands, trails, and fragments.

Vein Microtextures
Based on microtextures and crystal habits, calcite veins were subdivided into syntaxial, blocky, and antitaxial veins (Bons et al., 2012). Syntaxial veins texturally transition into blocky and antitaxial veins. The textural continuum between syntaxial and antitaxial veins constitutes another vein type termed transitional. A few veins solely consist of microcrystalline calcite (micrite), which also occurs as an additional component in blocky, antitaxial, and transitional calcite veins.

Syntaxial Veins
Syntaxial vein calcites (Figures 4a-4d) developed elongate-blocky crystal habits, whose length-width ratios are <10. Calcite growth initiated at both sides of typically few millimeters thick fractures and terminated at a median line. Larger crystals overgrew smaller ones, resulting in an increase in the crystal width and a decrease in the number of grains toward the median line, that is, growth competition (Bons et al., 2012). Median lines and elongate-blocky crystal habits, however, may be incompletely developed, implying a textural transition into blocky veins. This applies to approximately 50% of all syntaxial veins. CL of syntaxial vein calcites reveals nonluminescent cores and red-orange luminescent grain boundaries. The occurrence of unambiguous syntaxial veins is restricted to the lower halves of the forearc drill cores ( Figure 2). Such well-developed samples exhibit parallel vein margins, indicating the opening trajectory of the fracture.

Blocky Veins
Blocky calcite veins (Figures 4e-4i) are the dominant vein type and occur independently of depth, host rock, and drill core ( Figure 2). They form millimeters to centimeters thick straight veins or irregular networks of branching veins that are composed of randomly distributed equant (blocky) calcite crystals. Blocky calcites enclose rectangular host rock fragments, palagonite shards, and antitaxial vein fragments with preserved median lines, or cement breccias. Argillaceous selvedges that may be accompanied by dark alteration halos mark the vein-host rock transition. Zeolites either developed early-stage euhedral crystal habits preceding blocky calcite precipitation or formed anhedral crystals in late-stage veins crosscutting blocky calcite veins. CL of blocky calcites reveals diversely developed growth zonations. Simple growth zonations are composed of nonluminescent cores and thin red-orange luminescent grain boundaries. More complex oscillatory growth zonations consist of one to four thin red-orange luminescent zones in addition to the luminescent grain boundary. Early-stage and late-stage zeolites preceding blocky vein calcites and crosscutting them, respectively, display a blue CL.

Antitaxial Veins
Especially in FAB-like rocks from the Izu-Bonin rear arc antitaxial veins are prominent structures and approximate the abundance of blocky veins (Figure 2), whereas in the forearc antitaxial veins are rare (boninite) or absent (FAB). Antitaxial veins (Figure 4j-4n) are composed of fibrous calcites with length-width ratios ≫10. Calcite fibers typically grew from a median line outward, incorporating host rock fragments as solid inclusion bands and trails (Bons et al., 2012). Antitaxial veins either (1) have single median lines in the vein center (symmetrical) or closer to the vein margin (asymmetrical), (2) have multiple median lines often accompanied by micritic infill (multiply (a)symmetrical), or (3) lack median lines (unitaxial). In   Bons et al. (2012). The inward growth of syntaxial elongate-blocky calcites sealed mode I fractures along a median line (d). Blocky calcites grew within fluid-filled fractures (i). Blocky calcites enclose host rock (e, f) and antitaxial vein fragments (g) and developed Mn-controlled growth zonations in some cases (h). Preserved median lines and vein margins of antitaxial vein fragments enable the reassembly of corresponding fragments (g). Diffusion-fed and crystallization pressure-driven outward growth of antitaxial calcite fibers started at a median line and displaced the host rock (k, n) without fracturing (e.g., Elburg et al., 2002;Meng et al., 2019;Wiltschko & Morse, 2001). Fiber growth faces episodically incorporated solid inclusion bands composed of host rock material. Individual fibers join corresponding markers along the vein margins and indicate the opening trajectory (k, l) (e.g., Durney & Ramsay, 1973;Ramsay & Huber, 1983). The textural continuum between syntaxial and blocky veins is displayed in (c) where the median line is slightly discontinuous (bottom left). In (j) and (m) calcites developed weak growth competition and length-width ratios~10 indicative of the transitioning from syntaxial to antitaxial veining (transitional veins). general, median lines and vein margins are parallel to each other. Antitaxial veins are associated with branching vein systems and occur at different depths. Fragments of antitaxial veins may be incorporated into blocky veins ( Figure 4g). Calcite fibers display a high red-orange CL, which slightly attenuates along fiber long axis.

Transitional Veins
Transitional vein calcites have length-width ratios~10, which mark the textural continuum between syntaxial elongate-blocky (<10) and antitaxial fibrous (>10) (Bons & Jessell, 1997) calcite shapes (Figures 4j and  4m). Similar to antitaxial veins, transitional veins display median lines, which are often spatially associated with micrite, and luminescent inclusion bands. In contrast, slight growth competition of calcite crystals as well as CL characteristics showing nonluminescent cores and red-orange luminescent grain boundaries conform to the syntaxial vein type. Transitional veins pervade FAB and boninites but lack in FAB-like rocks from the rear arc where the characteristics of antitaxial veining are well developed (Figure 2).

Micrite Veins
Veins may be completely (micrite veins) or partly (blocky, antitaxial, and transitional veins) filled with micrite ( Figure 5). Micrite veins are often multiply laminated and contain host rock fragments and palagonite shards. Individual layers are distinguishable by their colors comprising shades of white, green, and brown. Nonlaminated micrite occurs in spatial association with blocky veins in which micrite shows a patchy occurrence as well as antitaxial and transitional veins where micrite concentrates along the median lines. They typically display a more whitish color and contain less noncarbonate phases (e.g., zeolites and clay minerals) than do laminated micrites. The occurrence of micrite is restricted to branching vein networks, particularly in the upper levels of the Izu-Bonin forearc, whereas the rear arc lacks any micritic components ( Figure 2). CL of micrite depends on its mineralogical purity. Pure micrite exhibits highly redorange luminescent crystals. With increasing amount of noncarbonate phases within micrite accumulations or layers, CL may progressively diminish up to nonluminescence.

Stable Oxygen and Carbon Isotopes
Vein calcites have δ 18 O values that range from −15.3‰ to +3.5‰ VPDB and δ 13 C values that vary between −1.1‰ and +2.4‰ VPDB ( Figure 6 and Table 1). These oxygen isotopic compositions correspond to δ 18 O formation temperatures (T δ18O ) between 0 and 140°C using the calcite-water fractionation curve of Friedman and O'Neil (1977) and assuming equilibrium precipitation from seawater with a range of δ 18 O compositions from −1‰ to +1‰ VSMOW (Table 1). Most samples, however, have δ 18 O values >−0.5‰ VPDB equivalent to formation temperatures <30°C and δ 13 C values >1.0‰ VPDB, resulting in a point cluster, which overlaps with the field defined by the compilation of Eocene deep-sea benthic foraminifera (Zachos et al., 2001). Boninite-hosted vein calcites plot within this point cluster and display a smaller δ 18 O variation (−1.0‰ to +1.6‰ VPDB) than do FAB-hosted vein calcites (−0.4‰ to +3.5‰ VPDB). Vein calcites from the Izu-Bonin rear arc define a trend toward significantly negative δ 18 O values as low as −15.3‰ VPDB, which is not found in vein calcites from the other drill sites.
A general trend toward more negative δ 18 O values with depth below seafloor is observed in vein calcites from the forearc (Figure 7). Oxygen isotopes of vein calcites from Ocean Drilling Program (ODP) Site 786 in the northern Izu-Bonin forearc (Alt et al., 1998) and ODP Site 801 in the Western Pacific Pigafetta basin (Alt & Teagle, 2003) extend this trend to greater depths. Below 450−550 m below seafloor, oxygen isotopic variation increases significantly including also vein calcites from the rear arc.
Independently of the host rock, δ 18 O and δ 13 C compositions define fields related to the vein type ( Figure 6). The lowest δ 18 O compositions (<−3.2‰ VPDB) predominantly refer to blocky vein calcites. Syntaxial vein calcites plot uniformly within the point cluster defined by samples with δ 13 C >1.0‰ VPDB and δ 18 O > −0.5‰ VPDB. They have rather high δ 13 C values ≥1.9‰ VPDB and well-constrained δ 18 O compositions between 0.0‰ and −0.4‰ VPDB, resulting in a narrow field for this vein type. The range of δ 13 C compositions of antitaxial vein calcites includes the lowest δ 13 C values (<+1.8‰ VPDB) of all samples and plots below the field of syntaxial veins. Transitional veins marking the textural continuum between syntaxial and antitaxial veins span a field whose δ 13 C values overlap with the upper range of antitaxial veins and the lower range of syntaxial veins. Stable isotope compositions of separately drilled powders of the same Figure 6. Stable carbon and oxygen isotopic compositions of vein calcites subdivided into drill site (colors) and vein type (symbols). Temperature axis was calculated assuming a seawater composition of −1‰ VSMOW and using the calcitewater fractionation curve of Friedman and O'Neil (1977). Stable isotope compositions of 14 vein samples were taken from the clumped isotope analyses (see also Table 1). The global Eocene deep-sea stable isotope record of carbonates refers to Zachos et al. (2001).

10.1029/2019GC008745
Geochemistry, Geophysics, Geosystems  Geochemistry, Geophysics, Geosystems sample show a good consistency in the carbon isotope composition, but some samples, especially from the rear arc, reveal strong inhomogeneities in δ 18 O (Table 1).

Clumped Isotopes
The analyzed vein calcites reveal consistently low clumped isotope temperatures (T Δ47 ) ranging from 9 ± 5 to 74 ± 12°C (Figures 8 and Table 2). The mean T Δ47 of 33°C separates these temperatures into two groups. The higher T Δ47 (>33°C) belong to veins from the Izu-Bonin rear arc, whereas the lower T Δ47 (<33°C) refer to veins hosted in FAB and boninites from the Izu-Bonin forearc.
Independently of vein type and host rock, most parental fluid δ 18 O compositions ( Figure 8 and Table 2) fall within the conservatively estimated seawater range (−2‰ to +2‰ VSMOW) (Hoefs, 2015) or are slightly enriched relative to seawater (up to +2.6‰ VSMOW). Two pairs of values of blocky veins from the Izu-Bonin rear arc, however, show significantly lower calculated compositions than seawater (−3.3‰ to −9.7‰ VSMOW), whereas another blocky vein from the same location falls within the seawater field (−1.8‰ to +0.2‰ VSMOW). Stable and clumped isotopic analyses of separately drilled powders of the same sample yielded similar δ 18 O and δ 13 C values for most vein calcites, but deviations from each other by up to 3.5‰ VPDB especially exist for vein calcites from the rear arc (Tables 1 and 2).

87
Sr/ 86 Sr ratios vary from 0.70574 to 0.70836, but most samples cluster between 0.70770 and 0.70784 (Table 3). This cluster intersects the Sr isotope seawater curve (McArthur et al., 2001) for the Eocene to present day between~52 and~33 Ma ( Figure 9). The sigmoidal shape and shallow slope of the Sr isotope seawater curve until~35 Ma leads to multiple intersections. One higher 87 Sr/ 86 Sr ratio (0.70836) refers to a single FAB-hosted blocky vein calcite that intersects the seawater curve at~22 Ma. Lower 87 Sr/ 86 Sr ratios (<0.70770) lack any intersection with the seawater curve and refer to all veins from the Izu-Bonin rear arc as well as a few blocky and antitaxial veins hosted in boninites from the Izu-Bonin forearc. Antitaxial and transitional vein calcites mostly have 87 Sr/ 86 Sr ratios below the seawater curve, whereas blocky vein calcites cover the whole 87 Sr/ 86 Sr range of the sample suite.   (Figure 10b). The sample suite defines a trend from Eocene seawater toward altered host rock compositions. Vein calcites from the Izu-Bonin rear arc show the most host rocklike compositions. -normalized (McLennan, 1989) REE+Y concentrations of vein calcites exhibit heavy REE-enriched distribution patterns with negative to slightly positive Ce, positive Eu, and positive Y anomalies ( Figure 11). Vein calcites have higher concentrations in REE+Y than do seawater (Zhang & Nozaki, 1996) and hydrothermal fluids (Bau & Dulski, 1999). Here the term hydrothermal fluid is defined as seawater that entered the oceanic crust and interacted at depth with rocks under temperatures up to >400°C. This involves the enrichment of dissolved metals and gases in the fluid phase (Koschinsky, 2014 (Alt et al., 1998) and ODP Site 801 in the Western Pacific Pigafetta basin (Alt & Teagle, 2003). Temperature axis was calculated as in Figure 6.
Eu/Eu* PAAS and Y/Ho ratios of each sample are plotted in Figure 12. The subdivision into vein type and host rock reveals specific compositional fields for the different vein types as well as distinct fields for veins with the same host rock. The fields for antitaxial veins and veins from the Izu-Bonin rear arc are indistinguishable. Both are characterized by a significant variation of Eu/Eu* PAAS and a small range of Y/Ho ratios. In contrast, the fields for veins from the forearc and fields for syntaxial and blocky veins show a higher variation in Y/Ho, whereas the range of Eu/Eu* PAAS is restricted. Transitional veins marking the textural continuum between antitaxial and syntaxial veins display the low Y/Ho ratios of antitaxial veins and the restricted Eu/Eu* PAAS of syntaxial veins. All fields overlap in the center of the plot (Y/Ho = 36 ± 8, Eu/Eu* PAAS = 1 ± 0.5) where the point density is highest.

Discussion
5.1. Mode of Fracturing, Fluid Flow, and Fracture Sealing/Mineral Growth 5.1.1. Syntaxial Veins Most veins studied here are the result of mineral growth within fluid-filled fractures. The mode of fracturing may be deduced from the crystal shape and vein microtextures such as median lines and solid inclusion bands (Bons et al., 2012). The syntaxial veins are interpreted as mineralized mode I fractures possibly associated with extensional faulting. Their characteristic growth competition, the presence of a continuous median line, and the lack of solid inclusion bands suggest onefold crack and sealing (Bons et al., 2012;Fisher & Brantley, 1992;Ramsay, 1980). This implies that the rate of mineral growth approximated the rate of fracture opening (Fisher & Brantley, 1992;Hilgers et al., 2001). Extensional veining at Sites U1439-U1441 fits well into the concept of an extensional forearc regime where normal faults, grabens, and half-grabens (Christeson et al., 2016;Robertson et al., 2018) related to slab rollback and trench retreat are observed Kurz et al., 2019). In accordance with this interpretation, the Izu-Bonin rear arc Site U1438 lacks syntaxial veins as well as large-scale extensional structures (
Since syntaxial and blocky veins show a textural continuum and distinguishing features such as growth competition and median lines may be incomplete, rates of fracture opening to fracture sealing must have varied continually. Therefore, the typical characteristics of syntaxial veins are less evolved where mineral growth did not keep pace with fracture opening, resulting in mineral growth within fluid-filled space (Bons et al., 2012;Fisher & Brantley, 1992;Hilgers et al., 2001). Pure growth within fluid-filled space is unambiguously expressed by blocky veins in which ongoing nucleation and mineral growth without preferred orientation (Bons et al., 2012) resulted in irregularly arranged blocky calcites. Euhedral, early-stage zeolites on the vein margins give further evidence of mineral growth within fluid-filled fractures.
During their growth, calcites in veins from the Izu-Bonin forearc developed Mn-activated (Fairchild, 1983) growth zonations consisting of one to four luminescent zones including the grain boundary ( Figure 4h). Growth zonations are commonly interpreted as a result of changing fluid compositions associated with an open system, varying redox, temperature or pressure conditions, alternating growth rates, or closed system-related geochemical self-organization out of equilibrium (Dromgoole & Walter, 1990;Meyer, 1984;Ortoleva et al., 1987;Reeder et al., 1990;Shore & Fowler, 1996;Wang & Merino, 1992).
Changing fluid compositions may originate from mixing of seawater with hydrothermal fluid and/or different rates of fluid-rock interaction. The fine and repetitive character of the more complex oscillatory growth zonations consisting of more than two zones (e.g., BON-14 and FAB-16/19; Figure 4h), however, argues against changing fluid compositions and points to geochemical selforganization. During this process calcites autonomously develop growth zonations without extrinsic control (Ortoleva et al., 1987). Therefore, geochemical self-organization occurs in a closed system out of equilibrium (Reeder et al., 1990). This implies that the stable and clumped isotopic compositions of calcites showing complex growth zonations may not reflect the composition of the fluid from which they precipitated.

10.1029/2019GC008745
Geochemistry, Geophysics, Geosystems luminescent grain boundary may reflect Mn adsorption on the calcite surface (Franklin & Morse, 1983;McBride, 1979) at the end of crystal growth when the Mn/Ca ratio of fluid was increased due to preferred Ca incorporation into the core.

Antitaxial Veins
The origin of antitaxial veining and its mode of fluid supply are still under debate (Bons et al., 2012). In previous studies on natural antitaxial fibrous veins, crack and sealing (Ramsay, 1980) and microbe-induced calcite fiber growth (Bons et al., 2009;Bons & Montenari, 2005) were proposed as potential formation processes. However, the majority of previous studies agrees on fracture-independent veining solely due to the crystallization pressure of fibrous mineral growth involving considerable diffusive fluid supply (e.g., Elburg et al., 2002;Meng et al., 2019;Wiltschko & Morse, 2001). This hypothesis is corroborated by laboratory experiments (e.g., Bons & Jessell, 1997;Means & Li, 2001;Taber, 1916) and numerical simulations (e.g., Hilgers et al., 2001), in which antitaxial fibrous microtextures were successfully generated due to the crystallization pressure of fibers.
These natural and synthetic antitaxial veins from the literature show microtextural characteristics that are strongly reminiscent of the antitaxial veins studied here. We therefore suggest that fluid diffusion and the crystallization pressure of calcite fibers induced antitaxial veining in the Izu-Bonin rear arc and subordinately in the forearc. This is in accordance with the conspicuous high number of antitaxial veins in the rear arc where faults representing efficient advective fluid channels were not observed and a sedimentary cover reducing the permeability of the oceanic crust was present during the Eocene (Arculus, Ishizuka, Bogus,  Fisher, 1998;Pearson & Lister, 1973). This geological setting limited the advective fluid transport in the rear arc and provided conducive conditions for antitaxial veining.

Transitional Veins
The textural continuum between syntaxial and antitaxial veins reflects varying rates of mineral growth to fracture opening (Bons et al., 2012;Fisher & Brantley, 1992;Hilgers et al., 2001). The highest rates are recorded where fracture opening approximated zero and antitaxial calcite fibers display extreme lengthwidth ratios (Bons et al., 2012;Hilgers et al., 2001). Transitional veins have length-width ratios~10 and thus imply reduced rates of mineral growth to fracture opening. Transitional veins occur predominantly in the Izu-Bonin forearc Sites U1439-U1441, whereas the rear arc Site U1438 exposes a high number of purely antitaxial veins with length-width ratios >100. We therefore suggest that the extensional regime in the Izu-Bonin forearc (Christeson et al., 2016;Kurz et al., 2019) inhibited widespread antitaxial veining. Transitional veins may accordingly reflect antitaxial veining that was preceded, accompanied, or succeeded by extensional fracturing, whose path of propagation was predetermined by zones of weakness such as preexisting (antitaxial) veins.

Micrite Veins
Micrite-filled fractures with similar petrographic characteristics as the micrite veins studied in this investigation have been discovered in numerous drill cores recovered during IODP Expeditions and on land (e.g., Christie et al., 2001;Clerc et al., 2014;Quandt et al., 2018;Schroeder et al., 2015). Their origin and formation, however, is often not discussed in detail and remains vague. In studies focusing on the alteration of Oligocene, and Miocene refer to McArthur et al. (2001). The range of host rock reference values comprises boninites, FAB, and FAB-like rocks, which were recovered during IODP Expedition 351, submersible dives, and seafloor dredging (Ishizuka et al., 2006;Ishizuka, Taylor, et al., 2011;Reagan et al., 2010;Yogodzinski et al., 2018). Yogodzinski et al. (2018) define 87 Sr/ 86 Sr ratios <0.70340 as representative for preserved magmatic compositions. Higher 87 Sr/ 86 Sr ratios are probably the result of host rock-seawater interaction . Boninites tend to have higher 87 Sr/ 86 Sr ratios than FAB (Reagan et al., 2010). Note that the x-axis is compressed in order to show the full range of 87 Sr/ 86 Sr ratios. O isotopic compositions of vein calcites. The sample suite defines a trend from a seawater end-member to an altered host rock end-member. Host rock reference values refer to Eiler et al. (2000), Reagan et al. (2010), and Staudigel et al. (1995). Seawater reference values refer to McArthur et al. (2001) and Zachos et al. (2001). the oceanic crust, fluid-mediated injection or infill of calcareous sediments into fractures or cavities, that is, neptunian dykes (e.g., Lehner, 1991, and references therein), is often suggested (Christie et al., 2001;Clerc et al., 2014;Schroeder et al., 2015).
Though the Izu-Bonin forearc is characterized by a long sedimentary hiatus until 35 Ma, bypassing sediment (Robertson et al., 2018) was probably captured by fractures connected to the seafloor. Positive δ 13 C values preclude any significant contribution from organic and mantle-derived carbon. Instead, stable isotopic  Bau and Dulski (1999), and Zhang and Nozaki (1996), respectively. REE+Y data on boninites from the Izu-Bonin trench are limited. They show similar PAAS-normalized patterns as FAB but have lower heavy rare earth element concentrations (e.g., Reagan et al., 2010).

10.1029/2019GC008745
Geochemistry, Geophysics, Geosystems compositions are compatible with the precipitation from seawater and mobilized sediment. Therefore, the micrite-filled veins may represent neptunian dykes. This explains the observed lamination and the increased occurrence in the upper levels of the drill cores, whereas the deeper rear arc drill cores lack any micritic component ( Figure 2). Deeper micrites lacking lamination as well as noncarbonate phases and being spatially associated with blocky and fibrous calcites, by contrast, may have been precipitated from seawater similarly to blocky calcites. Furthermore, REE+Y distribution patterns exhibit almost consistently the typical negative Ce and the positive Y anomalies of seawater (Zhang & Nozaki, 1996), whereas pronounced positive Eu anomalies indicative of hydrothermal fluid involvement (Bau & Dulski, 1999) are absent. Positive Eu anomalies are caused by the fractionation of Eu 2+ from the trivalent REE+Y and give rough information on the fluid temperature evolution prior to mineral precipitation (Bau et al., 2010). In contrast to Eu +3 and the neighboring trivalent REEs, Eu 2+ is stable at temperatures ≥250°C (Sverjensky, 1984). The resulting fractionation is only weakly dependent on pressure, pH, and the alteration of plagioclase (Allen & Seyfried, 2005;Bau, 1991;Danielson et al., 1992;Sverjensky, 1984). Cooling of a ≥250°C hot hydrothermal fluid does not alter the positive Eu anomaly, and mixing with cool seawater only diminishes the Eu anomaly (Bau et al., 2010). Figure 12. Y/Ho and PAAS-normalized Eu anomalies (Eu/Eu* PAAS ) subdivided into drill site (colors) and vein type (symbols). Due to the inheritance of Eu anomalies from shale normalization relative to chondrite normalization, only Eu/ Eu* PAAS >1.3 are considered as true Eu anomalies (green arrow). Host rock and seawater reference values refer to Shervais et al. (2019) and Zhang and Nozaki (1996), respectively. REE+Y data on boninites from the Izu-Bonin trench are limited and incomplete. Available Y/Ho ratios and Eu/Eu* PAAS tend to be slightly higher in boninites (e.g., Reagan et al., 2010). Hydrothermal fluids typically have Y/Ho ratios from 28 to 45 and Eu/Eu* PAAS >7 (Bau & Dulski, 1999).

Geochemistry, Geophysics, Geosystems
Consistently negative Ce/Ce* PAAS and the corresponding absence of pronounced positive Eu/Eu* PAAS point to well-oxygenated seawater that was not modified by fluid-rock interactions as indicated by Y/Ho ratios mostly >40. These characteristics are based on the redox-sensitive fractionation of Ce +4 from the trivalent REE and the preferred sorption of Ho relative to Y on particulate matter in marine environments, respectively (Bau et al., 1997;Bau & Dulski, 1999;Möller, 2002;Nozaki et al., 1997).
Stable oxygen and clumped isotope geothermometry uniformly reveal temperatures <30°C. This lowtemperature calcite formation agrees with the locally associated occurrence of the low-temperature zeolite mineral phillipsite, which is stable below 100°C and represents an alteration product of basaltic glass (Chipera & Apps, 2001). The decrease of δ 18 O values equivalent to increasing formation temperatures with increasing depth below seafloor (Figure 7) may be the result of a geothermal gradient. In a few cases, repeated δ 18 O analyses of individual blocky calcites reveal varying δ 18 O values (Table 1). Together with the occurrence of complex growth zonations (e.g., , this probably reflects variable physicochemical conditions during calcite growth (e.g., growth rate and temperature), changing fluid compositions due to fluid mixing or global changes in seawater chemistry over time, or geochemical self-organization. Most blocky calcites (e.g., FAB-7/9/12/17/18), however, show consistent δ 18 O values and simple growth zonations indicative of equilibrium precipitation from seawater.
Based on these geochemical interpretations and the predominance of blocky veins associated with brecciated host rocks, the Izu-Bonin forearc at Sites U1440 and U1441 must have experienced extensive hydrofracturing accompanied by minor extensional fracturing indicated by rare syntaxial veins. Hydrofractures propagated upward to the seafloor where the advective fluid circulation system was recharged (Alt, 1995). Argillaceous vein selvedges formed prior to blocky calcites as the result of fluid-rock interaction. They formed in equilibrium with the prevailing physicochemical conditions and reduced the host rock area in fractures exposed to circulating fluids. Therefore, the vein selvedges possibly attenuated further fluidrock interaction.

U1439 (Boninite, Forearc)
Boninite-hosted vein calcites from Site U1439 show similar isotopic signatures as FAB-hosted vein calcites, implying low-temperature precipitation under similar conditions. Stable oxygen and clumped isotopic compositions indicate precipitation temperatures <30°C and show a correlation with depth following a geothermal gradient. Further similarities include stable carbon and oxygen isotopic compositions overlapping with the global Eocene deep-sea carbonate field (Zachos et al., 2001), parental fluid δ 18 O compositions within the seawater range, and intersections of sample 87 Sr/ 86 Sr ratios with the Sr isotope seawater curve (McArthur et al., 2001).
However, some vein calcites (BON-10/11/19/20) have 87 Sr/ 86 Sr ratios slightly below the 87 Sr/ 86 Sr seawater curve, pointing to the involvement of mantle-derived Sr. Trace element characteristics show further differences between vein calcites from the different forearc sites. Boninite-hosted vein calcites have less pronounced negative or even positive Ce/Ce* PAAS , larger Eu/Eu* PAAS , and lower Y/Ho ratios than have FAB-hosted vein calcites, indicating a reducing environment and more intense fluid-rock interaction at Site U1439. These modified seawater signatures may be related to the lack of selvedges possibly representing a protection from fluid-rock interaction, higher extent of fluid-rock interaction as indicated by pervasively altered boninite compared to localized alteration patterns of FAB, and/or higher proportion of antitaxial and transitional veins, whose diffusion-fed growth facilitated host rock leaching. As a consequence of fluid-rock interaction, the alkalinity in the fluid was increased, leading to calcite supersaturation (Sample et al., 2017); the Y/Ho ratios of the fluid were reduced toward the host rock value (~26) ; and the fluid δ 18 O was slightly lowered relative to seawater (Gregory & Taylor, 1981;Muehlenbachs & Clayton, 1976). The latter is indicated by clumped isotopes, which yield slightly lower parental fluid δ 18 O compositions than FAB-hosted vein calcites. An increase in fluid alkalinity is probably accompanied by mobilization of Sr from the host rock resulting in an 87 Sr/ 86 Sr decrease of ambient seawater.
Both Mn-controlled blocky calcite growth zonation (e.g.,  and varying Ce/Ce* PAAS probably indicate redox-sensitive changes during mineral growth. Moreover, varying conditions during calcite growth are suggested by variable stable isotopic compositions of individual vein samples (BON-4/22/26) also observed among FAB-hosted vein calcites (Table 1). The textural continuum between syntaxial and antitaxial veins (i.e., transitional veining) is supported by stable isotopic and trace element characteristics in Figures 6 and 12  Similarly to the FAB Sites U1440 and U1441, hydrofracturing, advective fluid flow, and low-temperature calcite precipitation modified the boninitic crust at Site U1439. Elemental and isotopic tracers, however, indicate that the circulating seawater underwent more intense modification due to fluid-rock interactions than that at Sites U1440 and U1441. 5.2.3. U1438 (FAB-Like, Rear Arc) In contrast to the vein calcites from the Izu-Bonin forearc, the vein calcites from the rear arc show a considerable range of δ 18 O, δ 13 C, Δ 47 , and 87 Sr/ 86 Sr isotopic compositions, pointing to the formation under variable physicochemical conditions. On the one hand, this range includes seawater-like 87 Sr/ 86 Sr ratios and parental fluid δ 18 O values within the range of seawater (ASB-4/9). Together with corresponding seawater-like average Y/Ho ratios >36, Eu/Eu* PAAS ≤1.3, and Ce/Ce* PAAS <1 (ASB-4/11), precipitation from slightly modified seawater similar to the vein calcites from the forearc is suggested. On the other hand, the range of isotopic compositions also involves the lowest δ 18 O, parental fluid δ 18 O, δ 13 C, and 87 Sr/ 86 Sr values, as well as the highest formation temperatures and Eu/Eu* PAAS of all samples (ASB-1/3/5/10/12).
Very low parental δ 18 O fluid values relative to seawater (−3.3‰ to −9.7‰ VSMOW, ASB-1/5) are likely the result of low-T fluid-rock interaction (Gregory & Taylor, 1981;Muehlenbachs & Clayton, 1976). Low-T seawater-basalt interaction results in 18 O enrichment of the basalt and 18 O depletion in the fluid (Gregory & Taylor, 1981;Muehlenbachs & Clayton, 1976), from which the calcites precipitated. The corresponding samples have 87 Sr/ 86 Sr ratios below Eocene seawater values. Therefore, clumped and Sr isotopic compositions consistently indicate that vein calcites from the rear arc precipitated from seawater that was modified by fluid-rock interaction. As a result of this 18 O depletion in the parental fluid, stable oxygen isotope thermometry-based precipitation temperatures are overestimated. We therefore suggest that precipitation temperatures of vein calcites from the rear arc did not exceed the T Δ47 of 74 ± 12°C.
Compared to veining in the forearc, vein calcites from the rear arc show higher formation temperatures, indicate more altered seawater compositions, and are more often classified as antitaxial veins. These features can be related to the local geological setting of the rear arc. Veins from Izu-Bonin rear arc Site U1438 were recovered from significantly greater depth between 1,460 and 1,600 m below seafloor compared to the <500 m depth of the veins from the forearc (Figure 2). These differences may explain higher precipitation temperatures due to increasing temperature with increasing depth. Furthermore, the lack of major fault structures in the rear arc and the presence of a low-permeability sedimentary cover during the Eocene (Arculus, Ishizuka, Bogus, Gurnis, et al., 2015;Arculus, Ishizuka, Bogus, & the Expedition 351 Scientists, 2015) resulted in limited advective fluid circulation Fisher, 1998;Pearson & Lister, 1973) and provoked fluid diffusion instead. This facilitated the growth of diffusion-fed and fault-independent crystallization pressure-driven antitaxial veins (Elburg et al., 2002;Means & Li, 2001;Meng et al., 2019;Taber, 1916;Wiltschko & Morse, 2001). Diffusion in turn intensified fluid-rock interaction and resulted in varying degrees of seawater modification.

Chronology of Veining
Calcite that forms in the marine environment under equilibrium conditions records the 87 Sr/ 86 Sr ratio of the fluid from which it precipitates without fractionation (Kawahata et al., 2001). Therefore, intersections of Sr isotopic compositions of vein calcites with the Sr isotope seawater curve (McArthur et al., 2001) enable the relative dating of calcite precipitation (Hart & Staudigel, 1978) and fracturing in case of syntaxial crack and sealing veins due to simultaneous fracturing and calcite precipitation (Ramsay, 1980;Roberts & Walker, 2016). The potential involvement of mantle-derived strontium with 87 Sr/ 86 Sr ratios~0.70300 (Albarède et al., 1981) lowers the 87 Sr/ 86 Sr ratio of the sample. Due to increasing seawater 87 Sr/ 86 Sr ratios since~40 Ma (McArthur et al., 2001), a contribution of mantle strontium would increase the apparent age of a sample, and thus, estimated ages represent maximum ages. For instance, a sample with an 87 Sr/ 86 Sr ratio of 0.70780 intersects the Sr isotope seawater curve at~34 Ma. If this sample resulted from the mixture of seawater and basaltic 87 Sr/ 86 Sr (~0.70300), then a higher seawater 87 Sr/ 86 Sr ratio (>0.70780) equivalent to a younger intersection age would be required in order to result in the mixed sample 87 Sr/ 86 Sr ratio.
Except for vein calcites from the rear arc (ASB-1/4/5/8/9/14) and a few boninite-hosted vein calcites from the forearc (BON-10/11/19/20), vein calcites independently of the type intersect the Sr isotope seawater curve between~52 and~22 Ma. Due to the sigmoidal shape of the Sr isotope seawater curve, particularly boninite-hosted vein calcites show multiple intersections. Vein calcites from the rear arc and boninitehosted vein calcites are also characterized by precipitation from variably modified seawater due to fluid-rock interaction. This probably involved the exchange of Sr between basalt and seawater. We therefore conclude that vein calcites from the rear arc and boninite-hosted vein calcites cannot be dated reliably using this method. Thus, these vein calcites may have formed between respective host rock emplacement (≤52 Ma) and today.
In contrast, FAB-hosted vein calcites show the most pristine seawater signatures and, in most cases, single intersections with the Sr isotope seawater curve (FAB-4/10/18/22). Their stable and clumped isotopic compositions indicate precipitation from seawater, and REE+Y distribution patterns show the typical features of seawater (e.g., negative Ce and positive Y anomalies). FAB-hosted vein calcites also define the upper parts of the δ 18 O-87 Sr/ 86 Sr-depth correlations in Figures 7 and 10, which indicate the most seawater-like compositions. CL petrography shows that the corresponding samples (FAB-7/18/22) are characterized by a nonluminescent core and a thin luminescent grain boundary. Based on this observation, equilibrium precipitation from fluids with constant composition is inferred. We therefore suggest that vein calcite precipitation within FAB may be dated by using the Sr isotope seawater curve. Thus, most vein calcites hosted in FAB and showing single intersections with the Sr isotope seawater curve point to calcite precipitation between~35 and 33 Ma.
A single calcite vein (FAB-18) hosted in a FAB talus (Figure 2) intersects the Sr isotope seawater curve at~22 Ma (Figure 9). The FAB talus probably formed by erosion of extensional fault scarps in Miocene to Pliocene times (Robertson et al., 2018). The vein is intact and therefore appears to have formed after talus formation. This is in agreement with its intersection age, which represents a maximum precipitation age.

Quantification of the Extent of Fluid-Rock Interaction
The geochemical compositions of vein calcites indicate that they precipitated from variable fluid compositions ranging from pristine seawater (FAB-hosted vein calcites) to weakly and significantly modified seawater compositions due to fluid-rock interaction (rear arc and boninite-hosted vein calcites). This fluidrock interaction will be quantified in the following using simple binary mixing calculations. These calculations are based on a pristine seawater end-member with a fixed 87 Sr/ 86 Sr ratio for a given time and a basaltic end-member with a fixed 87 Sr/ 86 Sr ratio. This means that a precipitation age equivalent to the fixed seawater 87 Sr/ 86 Sr ratio is assumed and the necessary fraction of host rock 87 Sr/ 86 Sr is calculated in order to yield the sample 87 Sr/ 86 Sr ratio.
Since veining may have occurred before, contemporaneously with, or after pervasive host rock alteration, two approaches are presented. The first approach assumes pristine seawater reaction with a fresh basaltic host rock, whereas the second approach supposes pristine seawater reaction with an altered basaltic host rock. The latter is supported by some calcite veins that crosscut altered phenocrystals and zeolite veins, which represent alteration products.
For both approaches a range of pristine seawater 87 Sr/ 86 Sr ratios corresponding to~34 Ma (0.7078),~22 Ma (0.70836), and~5 Ma (0.70903) (McArthur et al., 2001) were used. The two older ages correspond to the assumed precipitation ages of FAB-hosted seawater-derived vein calcites. For fresh and altered basaltic end-members, 87 Sr/ 86 Sr ratios of 0.7032 and 0.7050 were used, respectively (Reagan et al., 2010;Yogodzinski et al., 2018). Between the Eocene and today, the Sr concentration of seawater has mainly decreased from~14 to~7.8 ppm (Coogan, 2009). Based on Coogan (2009), Sr concentrations of 12, 10, and 8 ppm, representative for seawater at 34, 22, and 5 Ma, respectively, were used for the calculations. FAB and boninites have average Sr concentrations of~100 ppm (Reagan et al., 2010). During seawater-induced alteration, Sr is mobilized but may be incorporated into alteration phases balancing the whole rock Sr concentration (Hart, 1971;Humphris & Thompson, 1978;Menzies & Seyfried, 1979). Therefore, 100 ppm Sr was used for both approaches. Mixed 87 Sr/ 86 Sr ratios were calculated for 1%, 5%, 10%, 20%, 30%, and 40% fractions of basaltic 87 Sr/ 86 Sr. Figure 13 shows the results of the mixing calculations. Due to its high Sr concentration, the basaltic endmember controls the 87 Sr/ 86 Sr ratio of the mixed fluid. The different time-dependent seawater 87 Sr/ 86 Sr ratios have an impact on the seawater-dominated mixtures but do not significantly affect host rockdominated mixtures. The lowest 87 Sr/ 86 Sr ratio of the sample suite recorded by a vein calcite from the rear arc (ASB-5) may have resulted from mixing of~90% seawater with~10% fresh host rock. If the host rock was altered before veining, 20% to 30% host rock-derived 87 Sr/ 86 Sr was necessary. The remaining rear arc vein calcites (ASB-1/4/8/9/14) have higher 87 Sr/ 86 Sr ratios, which require host rock fractions <10%. The basaltic 87 Sr/ 86 Sr fractions for boninite-hosted vein calcites depend on the time of precipitation but are generally <5%. In summary, most rear arc and boninite-hosted vein calcites precipitated from seawaterdominated fluids with <10% basaltic 87 Sr/ 86 Sr contribution.

Conclusion
Geochemical and microtextural observations indicate that calcite veining in the Izu-Bonin forearc and rear arc is dominated by different processes under diverse physicochemical conditions. These differences can be related to the local geological settings. In the forearc, blocky calcite veins predominate over syntaxial and   (Coogan, 2009;McArthur et al., 2001). These seawater compositions reacted with fresh (0.70320, 100 ppm) and altered basalt (0.70500, 100 ppm) (Reagan et al., 2010;Yogodzinski et al., 2018 Geochemistry, Geophysics, Geosystems fluid flow through the forearc crust. The rear arc, in contrast, shows approximately equal proportions of blocky and antitaxial calcite veins. The lack of major faults and the early presence of a sedimentary cover (Arculus, Ishizuka, Bogus, Gurnis, et al., 2015;Arculus, Ishizuka, Bogus, & the Expedition 351 Scientists, 2015) potentially hampered advective fluid circulation and provided conducive conditions for diffusionfed and crystallization pressure-driven antitaxial veining. Therefore, antitaxial veins may constitute an efficient veining mechanism within oceanic crust where faults and other advective fluid paths are absent or have previously been sealed by secondary mineralization and/or sedimentation.
As a consequence of these dissimilar geological settings and different vein types in the forearc and rear arc, vein calcites reveal varying geochemical signatures indicative of distinct geological and physicochemical processes. The sample suite defines a mixing curve from pristine seawater to variably modified seawater. The most pristine geochemical seawater signatures are recorded by FAB-hosted vein calcites whose oxygen and clumped isotopic compositions reveal precipitation temperatures <30°C. Sr/ 86 Sr and parental fluid δ 18 O compositions than FAB-hosted vein calcites indicating weak, lowtemperature fluid-rock interaction. This is in agreement with reduced Y anomalies and increased Eu anomalies relative to seawater. Simple binary mixing calculations yield <5% basaltic 87 Sr/ 86 Sr contribution presuming vein calcite precipitation at~34,~22, and~5 Ma.
A few blocky vein calcites from the forearc developed Mn-controlled oscillatory growth zonations or show varying Ce anomalies along line profiles through the veins. These observations indicate varying physicochemical conditions during calcite growth (e.g., varying growth rates, temperatures, redox conditions, and/or fluid compositions) or geochemical self-organization. The latter implies crystal growth within a closed system out of equilibrium (e.g., Ortoleva et al., 1987;Reeder et al., 1990). The geochemical interpretation of such samples has to be treated with care as they might be the result of disequilibrium precipitation.
Vein calcites from the rear arc show variable isotopic and trace element signatures due to varying degrees of fluid-rock interaction. Vein calcites with 87 Sr/ 86 Sr ratios slightly below the seawater curve, parental fluid δ 18 O compositions within the range of seawater, and seawater-like REE+Y patterns are also observed among boninite-hosted vein calcites in the forearc and indicate precipitation from slightly modified seawater due to weak fluid-rock interaction. More modified 87 Sr/ 86 Sr ratios considerably below the seawater curve, very low parental fluid δ 18 O compositions, and elevated precipitation temperatures up to 74 ± 12°C are distinct characteristics of vein calcites from the rear arc. These geochemical characteristics are probably related to the greater depth from which the veins were recovered and low-temperature fluid-rock interaction during which up to 20-30% basaltic 87 Sr/ 86 Sr was contributed to the fluid mixture from which the calcites precipitated.
This study shows the benefits and additional insights provided by the integration of vein microtextures into a geochemical study and demonstrates how the interplay of the local geological setting, mode of fluid flow and fracturing, and the physicochemical environment determines the geochemical composition of vein calcites and their microtextures.