Geochemistry of Vein Calcites Hosted in the Troodos Pillow Lavas and Their Implications for the Timing and Physicochemical Environment of Fracturing, Fluid Circulation, and Vein Mineral Growth

Abstract Calcite veins hosted in pillow lavas of the Late Cretaceous Troodos suprasubduction zone ophiolite provide insights into the timing and physicochemical environment of postmagmatic fracturing and fluid circulation through oceanic crust. This study presents rare earth element and yttrium (REE+Y) concentrations, δ13C, δ18O, 87Sr/86Sr, and clumped isotopic (Δ47) compositions of vein calcites in order to investigate their fluid sources, formation temperatures, and precipitation ages. These geochemical data are combined with microtextural analyses. Intersections of 87Sr/86Sr ratios of vein calcites with the Sr isotope seawater curve suggest two distinct calcite veining phases. Major calcite veining within an interval of ~10 Myr after crust formation is characterized by microtextures that point to extensional fracturing related to crack and sealing, host rock brecciation, and advective fluid flow. These vein calcites show REE+Y characteristics, 87Sr/86Sr ratios, and clumped isotopic compositions indicative of precipitation from seawater at <50 °C. Extended fluid residence times intensified fluid‐rock interactions and lowered Y/Ho ratios of some blocky vein calcites, whereas crack and sealing resulted in pristine seawater signatures. Low 87Sr/86Sr ratios of localized high‐temperature blocky vein calcites point to the involvement of hydrothermal fluids. These calcites show Mn‐controlled oscillatory growth zonations that probably developed in a closed system out of equilibrium. Later calcite veining (<75 Ma) may have coincided with rotation and/or uplift of the Troodos ophiolite. Microtextures of these vein calcites indicate fluid diffusion and fracture‐independent crystallization pressure‐driven veining. Their variably modified seawater signatures resulted from diffusion‐related fluid interaction with hydrothermal sediments.

Based on geochemical compositions and stratigraphic position, the Pillow Lavas were originally subdivided into an ultrabasic-basic Upper (UPL) and a basic-acid Lower Pillow Lava (LPL) series (Gass, 1968;Moores & Vine, 1971) and related to spreading and subduction, respectively (Dilek et al., 1990;Thy & Esbensen, 1993). Later studies confirmed a twofold subdivision, but observed interbedded LPL and UPL and noted a lack of any systematic compositional change with stratigraphic depth (Woelki et al., 2019(Woelki et al., , 2018. In addition, they found evidence for fluid-induced melting in both series. Based on radiolarian biostratigraphy of localized sedimentary rocks intercalated with and overlying the pillow lavas (Blome & Irwin, 1985) as well as U-Pb geochronology of zircons from plagiogranites (Mukasa & Ludden, 1987), the interval of main pillow lava volcanism was dated to~92-90 Ma. The termination of major pillow lava volcanism is poorly constrained. Ar-Ar dating of a few localized boninites gave ages as young as 55.5 Ma possibly reflecting prolonged local magmatism. However, this seems unlikely long compared to the Izu-Bonin-Mariana SSZ forearc where magmatic activity lasted~2 Myr (Reagan et al., 2019).

Postmagmatic Evolution
Sedimentation of the Troodos pillow lavas initiated in the Turonian with the deposition of hydrothermal umbers (Bragina, 2008;Robertson, 1975). These are locally overlain by radiolarian mudstones, volcaniclastic sandstones, and claystones (Chen & Robertson, 2019;Robertson & Hudson, 1974). Area-wide deep-sea calcareous sediment deposition began in Maastrichtian (Robertson, 1977). In Maastrichtian-Campanian times the allochthonous Triassic Mamonia complex was accreted to the western Troodos oceanic crust (Bailey et (Quandt et al., 2019). Previous study areas focusing on pillow lava-hosted secondary carbonates (Coogan et al., 2019;Gillis et al., 2015;Gillis & Robinson, 1990;Weinzierl et al., 2018) are indicated by yellow dotted boxes. Geochemical data from these studies are compared to the data presented in this study (see discussion). Selected faults and inferred faults in the vicinity of sample locations are indicated and taken from the Geological Map of Cyprus (Constantinou, 1995).

Sample Locations and Vein Microtextures
The vein samples used in this study were previously analyzed by Quandt et al. (2018) for microtextures and fluid inclusions. This chapter summarizes their findings and provides background information on the different vein types and their genetic implications. Vein and vesicle samples were taken from several pillow lava outcrops in the Troodos ophiolite including also locations that received little attention so far (Figure 1). In general, sampled veins extend over multiple lava pillows and crosscut pillow lava margins. Sampling of mineralized pillow lava interstices was avoided. Based on the occurrence of pillow lavas in the BG and the gradual transition from BG into LPL, it cannot be excluded that veins and vesicles in pillow lavas from the BG were sampled as well. The geological environments of the different sample locations and the respective vein microtextures are briefly described in the following. The different vein types are summarized in Figures 2a-2j Quandt et al., 2018, and references therein). (f) Inward growth of elongate-blocky calcites, growth competition due to inhibited nucleation of new crystals, and sealing along a median line are characteristic of syntaxial veining. Syntaxial vein calcites host decrepitated fluid inclusions (FIs) and show a high and homogenous cathodoluminescence (CL). (g) Growth-zoned blocky calcites host primary high-T fluid inclusions (~180-220°C) in Mn-poor zones. Decrepitated fluid inclusions are restricted to Mn-rich domains. Blocky calcite may also fill fluid-filled cavities where syntaxial (h) analcime or (i) quartz sealing was incomplete. (h) Blocky calcites associated with analcime show a highly repetitive and oscillatory growth zonation and host in some cases primary high-T fluid inclusions (~150-180°C), whereas (i) blocky calcites associated with quartz show a homogenous cathodoluminescence and decrepitated FIs. (j) In antitaxial veins, outward calcite fiber growth initiated at a median line and pushed the host rock aside without fracturing. Solid inclusion bands, cathodoluminescent bands, and decreasing cathodoluminescence in fiber growth direction are characteristic of antitaxial veins. They are also spatially associated with laminated micrite.
1. On the northeastern flank of the Troodos massif near the village of Margi, ultrabasic olivine-phyric pillow lavas from the uppermost UPL are exposed (Bailey et al., 1991). They are intercalated with and overlain by umbers. These hydrothermal Fe-and Mn-enriched sedimentary deposits (Robertson, 1975) are of Turonian-Santonian age (Bragina, 2008) and overlain by calcareous sedimentary rocks (Bailey et al., 1991). N-S striking normal faults formed a half-graben adjacent to the sample area in which hydrothermal and calcareous sediments accumulated (Bailey et al., 1991;Boyle & Robertson, 1984). The crosscutting and branching calcite vein networks analyzed in this study are ubiquitous structures and up to a few centimeters thick. Their proportion within the exposed rock volume may be as high as~10-20%. They have been previously described by Robertson (1975) who assumed that the calcite veins formed after the deposition of the umbers. Veins consist of calcite fibers with length-width ratios >10 oriented perpendicular to the wallrock. Fiber growth initiated at a median line from which the fibers grew outwardly incorporating solid inclusions bands and in cases tracking the vein opening trajectory. These features are typical for antitaxial veins that are related to fluid diffusion and crystallization pressure-driven veining independent of fracturing (e.g., Elburg et al., 2002;Means & Li, 2001;Meng et al., 2019;Wiltschko & Morse, 2001). This means that the rate of fiber growth exceeded the rate of fracture opening, which may have approximated zero (Hilgers et al., 2001). Under cathodoluminescence (CL) view, antitaxial vein calcites reveal luminescent bands perpendicular to the fibers and decreasing Mn concentrations in fiber growth direction. Some antitaxial veins are associated with different generations of crosscutting laminated micrite. 2. The Arakapas fault in the southern Troodos massif separates the stratigraphically intact Troodos ophiolite from the southern Limassol Forest where faulting disturbed the stratigraphic succession (Cann et al., 2001;MacLeod & Murton, 1993;Murton & Gass, 1986). There is broad consensus that the Arakapas fault represents a fossil oceanic transform fault along which sheeted dykes were sheared and mantle rocks exposed to the seafloor (e.g., Cooke et al., 2014;MacLeod & Murton, 1993;Moores & Vine, 1971;Simonian & Gass, 1978). The detailed geological evolution of the Arakapas fault and the Limassol Forest, however, is complex and controversial (e.g., Cann et al., 2001;MacLeod & Murton, 1993). In the Limassol Forest, the UPL and LPL outcrops are pervaded by millimeter to centimeter thick veins that may cement upper pillow lava breccias. These veins are composed of randomly distributed blocky calcite  Quandt et al. (2018). The stratigraphic unit of each sample is based on the Geological Map of Cyprus (Constantinou, 1995) and thin section petrography. Abbreviations: BG, Basal Group; LPL, Lower Pillow Lavas; UPL, Upper Pillow Lavas. crystals that enclose host rock fragments. Blocky veins may be associated with micrite clusters surrounding blocky calcites or laminated micrite. Straight single veins pervading the lower pillow lavas are less common. They consist of early-stage quartz followed by the precipitation of blocky calcite in cavities where quartz sealing was incomplete. Both blocky vein types indicate calcite precipitation into fluidfilled fractures or cavities under ongoing nucleation of new crystals (Bons et al., 2012). 3. In northwestern Cyprus the exposure of the Troodos ophiolite is discontinuous. Pillow lavas from the Akamas Peninsula are separated from the southern flank of the Troodos ophiolite by the Polis graben (Bailey et al., 2000;Borradaile & Lucas, 2003;Cameron, 1985;Swarbrick, 1993). To the south these units are bounded by serpentinite-filled sutures that give evidence of the amalgamation of the allochthonous Triassic Mamonia complex in Maastrichtian-Campanian times (Bailey et al., 2000;Lapierre et al., 2007;McPhee & van Hinsbergen, 2019). Veins were sampled from upper and lower pillow lava outcrops along the southern Troodos flank, 2-4 km east of the Evretou dam, and in the Akamas peninsula. In both areas, veins are predominantly millimeter thick straight single veins and occupy <5% of the exposed rock volume. Besides the typical blocky veins that also occur in other sample locations, microtextures indicate two related vein types, completely and incompletely sealed syntaxial veins, interpreted as mineralized extensional fractures or tension gashes. In completely sealed syntaxial veins, inward growth of elongate-blocky calcites sealed the fractures along a median line. Growth competition, that is, larger crystals that outgrow smaller ones, between neighboring calcites indicates that nucleation of new crystals was inhibited (Bons et al., 2012). These are the characteristics of pure syntaxial veining during which the crystal growth rates approximated the rate of fracture opening, that is, crack and sealing (Fisher & Brantley, 1992;Hilgers et al., 2001;Ramsay, 1980). Elongateblocky calcites host re-equilibrated/decrepitated fluid inclusions. In contrast, incomplete sealing by inward growth of elongate-blocky quartz or analcime resulted in fluid-filled space in which blocky calcite precipitated. Elongate-blocky analcime and quartz host primary high-T fluid inclusions (~180-240°C ). Some cavity-filling blocky calcites also host high-T fluid inclusions (~150-180°C) and show a complex and highly oscillatory growth zonation under CL view. Blocky calcites associated with quartz or analcime and blocky calcites unrelated to any early-stage noncarbonate mineral phase are summarized as blocky veins as both indicate precipitation into a fluid-filled space. 4. In the central-northern Troodos Stavros and Solea grabens large-scale extensional structures such as normal and low angle detachment faults were observed (e.g., Cooke et al., 2014;Hurst et al., 1994). Contemporaneously with or shortly after spreading these structures channelized hydrothermal fluid flow and were related to amagmatic extension and slow-spreading rates (Cooke et al., 2014;Hurst et al., 1994;Varga, 1991;Varga et al., 1999;Varga & Moores, 1985). Veins were sampled at the northernmost pillow lava exposure of the Troodos ophiolite along the northern extension of the Stavros graben. They occur as millimeter thick branching networks enclosing host rock fragments and are hosted in pillow lavas of the LPL or BG. Their proportion within the exposed rock volume is <5%. The sampled pillow lava outcrop is bounded by a fault whose sense of slip cannot be identified in detail due to strong alteration of adjacent rocks. This fault coincides with inferred NW-SE trending faults mapped in the Geological Map of Cyprus (Constantinou, 1995). In addition, an E-W trending fault interpreted as the northern transform fault was inferred from paleomagnetic data (Morris & Maffione, 2016). Veins are composed of small early-stage quartz crystals located along the vein margin and between blocky calcites. Blocky calcites show a Mncontrolled growth zonation under CL view. Mn-rich growth zones host decrepitated/re-equilibrated fluid inclusions, while Mn-poor zones host well-preserved primary fluid inclusions with seawater-like salinities from which high precipitation temperatures up to~220°C were inferred.

Carbon and Oxygen Isotopes
Sample powders for stable isotope analyses were produced using a computer-controlled micromill device equipped with a digital camera. Some vein samples consist of different calcite generations or show growth zonations. In some cases, it was possible to sample these different calcite generations/growth zones. Thus, the different growth zones of an individual blocky calcite (CY8_4) were sampled for stable isotopes. The different zones are indicated by capital letters behind the sample number (CY8_4 A to CY8_4 E). Similarly, blocky and micritic components of the same vein (e.g., CY10 A to CY10 C) and different stages of fiber growth (e.g., CY17_2 A to CY17_2 F) were sampled.

Geochemistry, Geophysics, Geosystems
Stable carbon and oxygen isotopic compositions of 38 sample powders in total were analyzed in two different laboratories.
(1) At the Institute of Earth Sciences, University of Graz, the calcite powders of 20 samples were reacted with oversaturated 100% ortho-phosphoric acid at 70°C in a Kiel II automated reaction system. Measurements were performed using a Delta Plus isotope-ratio mass spectrometer. Replicate analyses for standards (in-house and NBS 19) and samples had a reproducibility better than ± 0.05‰ VPDB (Vienna Pee Dee Belemnite) for δ 13 C and ± 0.15‰ VPDB for δ 18 O.
(2) At the JR-AquaConSol GmbH, Graz (Austria), the calcite powders of another 18 samples were analyzed using a GasBench II coupled to a Finnigan DELTA Plus XP mass spectrometer applying the phosphoric acid digestion method at 75°C and subsequent continuous-flow isotope ratio mass spectrometry (CF-IRMS). Reproducibility of replicate analyses for standards (in-house and NBS 19) and samples was better than 0.1‰ VPDB for both δ 13 C and δ 18 O values. Stable carbon and oxygen isotope analyses are archived at PANGAEA Data Archiving & Publication (Quandt et al., 2019).
The temperature-dependent fractionation of oxygen isotopes enables the calculation of calcite formation temperatures (e.g., Epstein et al., 1953;McCrea, 1950;Urey, 1947). These were calculated using the calcitewater fractionation curve of Friedman and O'Neil (1977), which is valid for temperatures between 0 and 500°C, and assuming equilibrium precipitation from a parental fluid with a δ 18 O value of −1‰ VSMOW (Vienna Standard Mean Ocean Water; Pucéat et al., 2003).

Clumped Isotopes
Sample powders for clumped isotope analyses were produced using a hand-held dentist drill. For each sample ≥100-mg powder was prepared for clumped isotope analyses. Pre-screening X-ray diffraction analyses on aliquots of the sample powders were conducted at the Institute of Applied Geosciences at Graz University of Technology prior to clumped isotope measurement in order to confirm the purity of calcite powders (for analytical details see Boch et al., 2019). For each replicate clumped isotope analysis 2-3 mg of the sample aliquots was inserted into the reaction vessel using a glass sample holder together with~1 ml 105% phosphoric acid at the bottom of the reaction vessel. After a 20-min evacuation period, acid digestion with ortho-phosphoric acid was started in the stirred reaction vessel at 90°C for 10 min. The reactant CO 2 was directly and continuously collected in a liquid N 2 -cooled glass trap and then cleaned after a procedure described by Dennis and Schrag (2010), including cryogenic distillation and separation of CO 2 from water using a dry-ice ethanol cooled glass trap. The remaining CO 2 was transferred to the mass spectrometer by passing through a Porapak Q trap at −35°C (see also Kluge et al., 2018).
Sample measurements were conducted using a Finnigan MAT 253 Plus at the Institute of Environmental Physics, Heidelberg University according to the procedures of Dennis et al. (2011) and Huntington et al. (2009). Measurements consisted of eight acquisitions, each including a peak center, background measurements, and an automatic bellow pressure adjustment aimed at a 6 V signal at mass 44, with ten cycles per acquisition, and a 26-s integration time per individual cycle. The sample gas was measured against an inhouse standard and transferred to the absolute reference frame (Dennis et al., 2011) using multiple carbonate standards (ETH1-4, Carrara Marble) and equilibrated gases.
The δ 18 O thermometer relies on temperature-dependent isotope fractionation and requires the knowledge of the δ 18 O value of the parental fluid from which a mineral (e.g., calcite) precipitated (Urey, 1947). Clumped isotopes, however, can provide temperature estimates independently of the parental fluid δ 18 O composition, which in turn may be calculated from the clumped isotope measurements (e.g., Ghosh et al., 2006). Based on the inorganic calcite calibrations of Daëron et al. (2019), Kele et al. (2015), Kelson et al. (2017), and Kluge et al. (2015), Δ 47 values were converted into clumped isotope formation temperatures (T Δ47 ), which were then used to calculate the parental fluid δ 18 O compositions, from which the calcites precipitated following the calibration of Coplen (2007 Geochemistry, Geophysics, Geosystems temperature samples require a temperature of ≥100°C over a period of 100 Myr in order to induce significant reordering of the clumped isotopes (Coogan et al., 2019;Henkes et al., 2014;Passey & Henkes, 2012). Since secondary fluid inclusion trails and planes, which are probably related to postcrystallization microfracturing, record temperatures <180°C (Quandt et al., 2018) and a long thermal event contrasts the geological history of the Troodos ophiolite (Mukasa & Ludden, 1987;Robertson, 1977), we feel confident that the blocking temperature has not been exceeded and the clumped isotope system remained isotopically closed.

Strontium Isotopes
Sample powders for strontium isotope analyses were produced using a hand-held dentist drill. In order to avoid dissolution of noncarbonate minerals or host rock constituents, approximately 30-mg powder per sample were dissolved in purified 0.4 M HCl. Sr was subsequently separated via ion exchange column chemistry using Sr-specific extraction chromatographic resin (Eichrom®, USA) and diluted double-distilled nitric acid. For element elution and Sr collection 5 ml of 3 M HNO 3 and 1 ml of 0.1 M HNO 3 were used, respectively (for details see Stammeier et al., 2018). Strontium isotope measurements were performed on a Nu Plasma II multicollector inductively coupled plasma mass spectrometer (MC-ICP-MS, Nu instruments, Wrexham, UK) equipped with a 0.1 ml/min MicroMist glass nebulizer at the NAWI Graz Central Lab for Water, Minerals and Rocks. Briefly, sample measurements were performed (1) in wet-plasma mode 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 86 Sr/ 88 Sr = 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 yielded 87 Sr/ 86 Sr ratios of 0.710253 ± 0.000067 (2 SD; n = 12). Total procedural blanks were below 1.2 ng Sr and thus negligible compared to analyte signals. All samples were corrected relative to the value of 0.710250 for NIST SRM 987. 87 Sr/ 86 Sr analyses are archived at PANGAEA Data Archiving & Publication (Quandt et al., 2019).
The 87 Sr/ 86 Sr ratios of samples were used to date calcite precipitation. This represents a suitable approach since radiogenic Sr isotopes show neither mineral-nor temperature-related fractionation (Kawahata et al., 2001). If a calcite precipitates from seawater, its 87 Sr/ 86 Sr ratio reflects the seawater composition at that moment assuming no further Sr isotope exchange after crystallization, that is, a closed system. Thus, the intersection of the 87 Sr/ 86 Sr ratio of a sample with the well-established Sr isotope seawater curve (McArthur et al., 2001) represents a likely precipitation age that is most precisely constrained where the curve is well-defined and has a steep slope (Hart & Staudigel, 1978). Where local minima and maxima cause sigmoidal curve shapes (e.g., 87 Sr/ 86 Sr ratios between 0.70783 and 0.70770), samples display multiple intersections and consequently multiple apparent ages. If fluid-rock interactions contributed mantle 87 Sr/ 86 Sr to the fluid from which a calcite precipitated, intersections represent maximum ages in most cases, mainly due to increasing 87 Sr/ 86 Sr seawater ratios since~90 Ma.
Considering the onefold crack and sealing characteristic of completely sealed syntaxial veins, both, fracturing and simultaneous precipitation may be dated (Ramsay, 1980;Roberts & Walker, 2016). If intersection ages approximate the timing of the earliest host pillow lava formation (≥90 Ma; Mukasa & Ludden, 1987;Osozawa et al., 2012), fracturing independent of the growth mechanism may be dated as well, because veining postdates pillow lava crystallization. Low Rb concentrations of calcite (93% of all laser ablation spot analyses <1 ppm Rb or even below detection limit, remaining spots mostly <4 ppm) do not require the calculation of initial 87 Sr/ 86 Sr ratios.

Rare Earth Elements and Yttrium Analyses
Rare earth elements and yttrium (REE+Y) concentrations of vein and vesicle calcites were measured by laser ablation on polished rock slices 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 Central Lab for Water, Minerals and Rocks. A 193 nm wavelength laser, pulsed at 7 Hz, ablated between 3 and 20 spots per sample. Spots were set to 75 μm in diameter and ablated along line profiles. After acquisition of a 30 s gas blank, a 60 s dwell time for each spot proceeded. Before and after every 20 spots, the NIST standard reference material (

Geochemistry, Geophysics, Geosystems
(National Institute of Standards and Technology, Gaithersburg, USA) and synthetic carbonate standard MACS-3 (United States Geological Survey) were analyzed as unknowns to monitor accuracy. All reference materials were reproduced for most elements within errors for the used elements. REE+Y analyses of the NIST SRM 612 standard measurements yielded a relative uncertainty of <5%. Time-resolved LA-ICP-MS data reduction was performed using GLITTER 4.0 software (Macquarie University, Sydney). REE+Y analyses and trace element ratios are archived at PANGAEA Data Archiving & Publication (Quandt et al., 2019).
In general, analytical interferences caused by barium oxide and shale normalization may induce apparent Eu anomalies (Tostevin et al., 2016). While Ba concentrations and Ba/Eu ratios of the calcites from this study are low (0.8 ppm Ba on average and Ba/Eu <100 for 102 out of 104 samples) and do not correlate with Eu anomalies, shale-normalized Eu anomalies are consistently~30% larger than chondrite-normalized Eu anomalies (e.g., McDonough & Sun, 1995). This implies that only Eu anomalies >1.3 reflect true positive anomalies.
Aluminosilicates such as clay minerals and zeolites are potential contaminators in the calcite veins during laser ablation. Therefore, only analyses with Zr <5 ppm and Al <2,000 ppm (Schier et al., 2018) are considered as clean calcites. Except for four red micrites and four blocky calcites with Al >2,000 ppm, all samples are below these cutoff values. Nevertheless, we did not reject these spot measurements since their REE+Y distribution patterns are indistinguishable from REE+Y distribution patterns of spot measurements with Al <2,000 ppm from the same sample. Hence, we are confident that clay minerals and zeolites did not affect the REE+Y measurements.  (Coogan et al., 2019;Gillis, 1987;Gillis et al., 2015;Gillis & Robinson, 1990). Corresponding δ 18 O formation temperatures (T δ18O ) range from 7 to 32°C (Figure 3).

Carbon and Oxygen Isotopes
Vein microtextures and stable isotope compositions do not correlate systematically except for antitaxial fibrous calcites from the Margi area that define a trend from the Upper Maastrichtian marine calcite field towards slightly lighter δ 13 C (up to −2.3‰ VPDB), not covered by previous studies (Figure 3). Isotope Maastrichtian marine calcite (Frank & Arthur, 1999) and secondary calcites from Troodos pillow lavas, sheeted dykes, and gabbros (Coogan et al., 2019;Gillis, 1987;Gillis et al., 2015;Gillis & Robinson, 1990;Staudigel & Gillis, 1990;Vibetti, 1993). Enveloped samples (grey dotted contour) correspond either to Mn-poor (red dashed contour) or Mn-rich growth zones of a single calcite crystal shown in detail in Figure 8a. Black framed samples host high-T fluid inclusions (Quandt et al., 2018). Temperature axis is based on the temperature relation of Friedman and O'Neil (1977)

Clumped Isotopes
Calcite formation temperatures (T Δ47 ) calculated using four different inorganic calibrations yield consistent results within a limited temperature range whose average approaches the T Δ47 after Kelson et al. (2017;Figure 4;Quandt et al., 2019). Average sample T Δ47 range from 20 ± 8 to 123 ± 18°C. T Δ47 >50°C are restricted to a single blocky vein calcite sample that also hosts high-T fluid inclusions (Quandt et al., 2018).
Parental fluid δ 18 O compositions for individual samples vary significantly depending on the calcite-water fractionation factor applied. In most cases, however, calculated fluid δ 18 O values are within the conservatively estimated range of seawater (−2 to +2‰ VSMOW; Muehlenbachs, 1998;Figure 4;Quandt et al., 2019). A good correspondence with seawater is particularly given by using the Δ 47 -T calibration and the oxygen isotope fractionation 18 α(CaCO 3 -H 2 O) of Daëron et al. (2019) for extremely slowly growing carbonates. Average parental fluid δ 18 O values of low-T blocky and syntaxial veins approach the Cretaceous seawater δ 18 O composition of −1‰ VSMOW (Pucéat et al., 2003). Antitaxial vein calcites and associated micrites trend to slightly higher δ 18 O values >−1‰ VSMOW, but are still within the seawater range. A high-T growth zoned blocky vein calcite shows enriched parental fluid δ 18 O compositions up to +4.3‰ VSMOW (Figure 4). Since this sample displays a distinct δ 18 O zonation (Figure 3), its Δ 47 results are examined in detail in the discussion.

Rare Earth Elements and Yttrium
Post-Archean Australian Shale (PAAS)-normalized (McLennan, 1989) REE+Y distribution patterns of vein calcites are characterized by enrichments of the heavy relative to the light REE, and display Ce, Eu, and Y anomalies (Figures 6a-6c; Quandt et al., 2019). Micrites mostly mimic the REE+Y patterns of their spatially associated vein calcites. Vein calcites have two to five orders of magnitude higher total REE+Y contents than seawater and hydrothermal fluids (Bau & Dulski, 1999;Zhang & Nozaki, 1996) and similar contents as Troodos pillow lava glasses (Regelous et al., 2014). Most samples exhibit seawater-like distribution patterns.
The REE+Y patterns are dominated by negative Ce anomalies, but variations from negative to positive Ce anomalies occur along vein profiles (Figures 6a-6c Table 2.   (Bau & Dulski, 1999), and seawater (Zhang & Nozaki, 1996). High-T blocky vein calcites are composed of Mn-rich and Mn-poor growth zones whose REE+Y patterns are distinguished from each other in Figure 8b.

Geochemistry, Geophysics, Geosystems
Taken together, the entire sample suite shows a trend from pristine seawater-derived calcite precipitates to calcites that precipitated from modified seawater or a hydrothermally influenced fluid. This trend is representatively shown in Figure 7 where Y/Ho ratios and Eu/Eu* PAAS of the sample suite are compared with reference data. Seawater has Y/Ho ratios between 57 and 110 and uniform Eu/Eu* PAAS of~1 (Zhang & Nozaki, 1996). Hydrothermal fluids show lower Y/Ho ratios between 28 and 45 and higher Eu/Eu* PAAS between 8 and 57 (Bau & Dulski, 1999). The Troodos pillow lava glasses have uniform Y/Ho ratios of~27 and Eu/Eu* PAAS of~1.4 (Regelous et al., 2014). Some calcites show REE+Y characteristics that fall within the range of the pristine seawater end-member, whereas others trend toward the hydrothermal fluid endmember. This results in a curvilinear trend for the sample suite indicative of a range of fluid compositions. The absence of purely hydrothermal fluid signatures and the dominance of seawater-like characteristics may be the result of seawater-diluted hydrothermal fluids and insignificant fluid-rock interactions, respectively.  (Quandt et al., 2018). High-T blocky vein calcites are composed of Mn-rich and Mn-poor growth zones whose distinct Y/Ho and Eu/Eu* PAAS are shown in Figure 8b. Since PAAS-normalized Eu anomalies (Eu/Eu* PAAS ) are~30% larger than chondrite-normalized Eu anomalies, only Eu/Eu* PAAS >1.3 are interpreted as true Eu anomalies (green arrow). Very high Eu/Eu* PAAS may be caused by analytical interferences with BaO.

Geochemistry, Geophysics, Geosystems
The trend is based on the fractionation of the REE+Y under specific physicochemical conditions. The REE+Y have similar chemical properties, occur and behave as a coherent group, and may substitute for Ca 2+ in the calcite lattice (Zhong & Mucci, 1995). However, Y and Eu may fractionate from their neighboring rare earth elements by different processes, which are briefly explained. Y and Ho are considered to be geochemical twins, but the preferential sorption of Ho on particulate matter in marine environments causes temperature-independent fractionation from Y and results in a positive Y anomaly of seawater (Bau et al., 1997;Bau & Dulski, 1999;Möller, 2002;Nozaki et al., 1997). Eu occurs in +3 and +2 oxidation states. Eu 2+ fractionates from its neighboring trivalent REE under reducing conditions and dominates over Eu 3+ at temperatures ≥250°C, whereas pressure, pH, and the alteration of plagioclase do not significantly affect the Eu 2+ /Eu 3+ redox potential (Allen & Seyfried, 2005;Bau, 1991;Danielson et al., 1992;Sverjensky, 1984). A ≥250°C hot fluid preserves its positive Eu anomaly after cooling, while mixing with a fluid without positive Eu anomaly (e.g., seawater) reduces the positive Eu anomaly (Bau et al., 2010). This temperature relationship is represented by vein calcites that are characterized by high-T fluid inclusions as well as positive Eu/Eu* PAAS (CY5 and CY8).
The findings of this study are compatible with previous studies in which stable and clumped isotopic compositions also indicated low-T secondary calcite precipitation from seawater (Coogan et al., 2019;Gillis et al., 2015, and references therein;Gillis & Robinson, 1985Weinzierl et al., 2018). These studies mainly focused on the classical pillow lava outcrops on the northern flank of the ophiolite (e.g., Akaki and Margi area), whereas the vein calcites from this study were sampled south of the Arakapas fault, along the southern Troodos flank including the Akamas peninsula, and in the northernmost Troodos ophiolite.
High-T vein calcite precipitation and modified seawater signatures are subordinate features, which are particularly observed along the southern Troodos flank, Limassol Forest, and northernmost Troodos ophiolite. Similarly, Weinzierl et al. (2018) observed a few veins and vesicle calcites with 87 Sr/ 86 Sr ratios below the Sr isotope seawater curve and very low δ 18 O compositions, which were interpreted as a result of high formation temperatures. The high-T vein calcites sampled in this and in the study by Weinzierl et al. (2018) have in common that they represent subordinate features in the respective data sets. In both studies, pillow lavahosted high-T (>100°C) vein calcites mainly occur in the LPL and/or Basal Group. The Basal Group represents the lower pillow lava-sheeted dyke transition where an increase in temperature and a decrease in permeability was suggested (Gillis & Robinson, 1990;van Everdingen, 1995).
Sr isotopes of secondary calcites have been widely used to date their crystallization by matching the 87 Sr/ 86 Sr ratio of a calcite sample with the Sr isotope seawater curve (McArthur et al., 2001) and assuming precipitation from unmodified seawater (Hart & Staudigel, 1978). This assumption is valid for most samples as discussed above. Therefore, sample intersections with the Sr isotope seawater curve between~92 and~82 Ma are interpreted as precipitation ages. The hydrothermal signatures of high-T blocky (no intersections) and antitaxial vein calcites (intersections between~75 and~34 Ma) suggest the involvement of mantlederived 87 Sr/ 86 Sr ratios (Table 2). Accordingly, their ages will be discussed in detail in chapter 5.2.2 and 5.2.3, respectively.
Secondary mineralization as well as an area-wide thick sedimentary cover decrease the permeability of the oceanic crust (e.g., Fisher, 1998), which is typically highest in the uppermost pillow lavas and decreases downward into the sheeted dykes (e.g., Coogan & Gillis, 2018, and references therein). This depth-dependent distribution of permeability is reflected by the occurrence of the different veins and vein networks studied here. The UPL are characterized by pervasive centimeter thick vein networks, while veins in the LPL are less frequent, millimeter thick, and usually occur as single veins. Moreover, veins from the LPL and BG pillow lavas show a higher mineralogical diversity (calcite, SiO 2 phases, different zeolites), more complex fluid 10.1029/2019GC008369 Geochemistry, Geophysics, Geosystems origins, and higher formation temperatures than veins from the UPL (Quandt et al., 2018). Similarly, the pillow lava-hosted high-T (>100°C) veins from Weinzierl et al. (2018) belong to the Basal Group. Taken together, this indicates that the LPL/Basal Group represent the stratigraphic level where high-T hydrothermal fluid circulation transitions into the overlying low-T seawater circulation cell.
In summary, this investigation confirms previous studies and complements them by providing vein calcite data from previously not sampled pillow lava locations. The findings of this and previous studies together imply that large parts of the Troodos pillow lava section underwent a similar major phase of low-T calcite veining that lasted for~10 to~20 Myr. Since seawater-derived low-T vein calcites occur independently of the sample location, we suggest that the different geological environments did not significantly affect the formation temperature and geochemical signatures of the vein calcites. The only exceptions are found in the northernmost Troodos ophiolite where high-T vein calcites were observed and in the Margi area where antitaxial veins occur that are absent in all other sample locations. Since geochemical signatures correlate with the different vein types and vary over small scales such as growth zones, the geochemical compositions and their implications will be discussed in detail for the different vein types in the following. This contributes to an improved understanding of syntaxial, blocky, and antitaxial veining within the oceanic crust.

Detailed Insights by Combining Geochemistry and Microtextures 5.2.1. Low-T Blocky and Syntaxial Veining
Low-T blocky veins associated with rock brecciation and advective fluid flow are the dominant vein type and occur in the UPL and LPL. This implies that hydrofractures and cooling fractures create significant permeability in the oceanic crust at different depths. In contrast, extensional fault-related syntaxial veins are less common. The geochemical seawater and seawater-like signatures of low-T blocky and syntaxial vein calcites suggest that their 87 Sr/ 86 Sr ratios intersecting the Sr isotope seawater curve between~92 and~82 Ma represent crystallization ages ( Figure 5). Double intersections of 87 Sr/ 86 Sr ratios of the oldest blocky vein calcites with the Sr isotope seawater curve approximate earliest pillow lava formation and imply fracturing betweeñ 92 and~90 Ma shortly after pillow lava solidification ( Figure 5). Syntaxial vein calcites are more suitable to date fracturing. Their onefold crack and sealing nature implies fracture opening in an extensional regime and contemporaneous mineral growth that terminates in complete fracture sealing along a median line (Bons et al., 2012;Fisher & Brantley, 1992;Quandt et al., 2018;Ramsay, 1980;Roberts & Walker, 2016). This process may reduce the fluid residence time within the fracture and thus prevents extensive fluid-rock interaction as indicated by seawater-like 87 Sr/ 86 Sr, δ 13 C, δ 18 O, and Δ 47 isotope signatures, Y/Ho ratios ≥40, and negative Ce/Ce* PAAS . Low T Δ47 and T δ18O <40°C argue against significant seawater heating. The unmodified seawater signature and the crack and sealing processes infer that intersections with the Sr isotope seawater curve at~88 and~82 Ma represent reliable fracturing and vein precipitation ages ( Figure 5). Syntaxial calcite veins lack the typical characteristics of host rock brecciation and their formation may be related to the extensive regime of the Late Cretaceous Troodos SSZ (e.g., Pearce & Robinson, 2010;Varga et al., 1999). Pure syntaxial calcite veins are restricted to the UPL, whereas initial syntaxial veining in the LPL was probably interrupted by fracture opening rates that exceeded mineral growth rates and resulted in fluid-filled cavities. Incompletely sealed syntaxial analcime and quartz veins hosting late-stage blocky calcites give some indication of this process.
Simultaneous fracture opening and calcite sealing may imply fast growth rates. As a result, isotopic disequilibrium due to preferential, kinetically controlled incorporation of light isotopes may occur (Watkins et al., 2014). It also explains the increased entrapment of numerous particularly large and primary two-phase fluid inclusions (Quandt et al., 2018) that are particularly prone to postentrapment modifications (Goldstein, 2001) and usually not observed at precipitation temperatures <50°C (Pagel et al., 2018).
In contrast, blocky vein calcites tend to have lower Y/Ho ratios and slightly more enriched parental fluid δ 18 O compositions than syntaxial veins. This might be the result of longer fluid residence times during which slight fluid-rock interaction occurred. Due to the low precipitation temperature, however, the isotopic seawater composition was not perceptibly altered.
Micrites may be spatially associated with blocky veins. They may be laminated or form clusters around blocky calcites. In both cases their isotopic signatures are similar to the blocky calcites they are associated with indicating low-T precipitation from seawater. Therefore, laminated micrites may represent marine 10.1029/2019GC008369 Geochemistry, Geophysics, Geosystems sedimentary infill, that is, neptunian dykes (e.g., Lehner, 1991, and references therein). Where micrites lack lamination, their formation may be related to abiogenic in-situ precipitation or calcite grain size reduction during repeated fracturing and host rock brecciation.
Intersection ages ( Figure 5) interpreted as the time of vein calcite precipitation and relative ages are consistent for a~23 Ma old low-T seawater-derived late-stage calcite from an incompletely sealed syntaxial analcime vein that crosscuts an~85 Ma old low-T blocky calcite vein (Figure 9d). This Early Miocene age falls into the temporal framework of minor Troodos uplift to hemipelagic or shallow water depth ( Figure 5; Avigad et al., 2016;Kinnaird et al., 2011;Morag et al., 2016;Robertson et al., 1991;Robertson et al., 2012). In a previous study, the formation of palygorskite and gypsum veins were also related to ophiolite emplacement (Gillis & Robinson, 1990). The elevated precipitation T Δ47 of this sample (36 ± 9°C), however, is relatively high for shallow Mediterranean seawater.

High-T Blocky Veining
Blocky vein calcites that host high-T fluid inclusions (up to~220°C) are also characterized by Mn-controlled growth zonations (Quandt et al., 2018). In vein calcites (CY8) from the northernmost Troodos ophiolite, these growth zonations are suitable for stable isotope and REE+Y analyses using micromilling and the laser ablation technique, respectively. These analyses show that Mn-poor and Mn-rich zones have distinguishable stable isotope and REE+Y characteristics that are summarized in Figure 8.
Mn-poor zones host well-preserved high-T primary fluid inclusions with seawater-like salinities (Quandt et al., 2018). Fluid inclusion and δ 18 O-based thermometers yield coinciding peak temperatures (219°C and 215°C, respectively) for Mn-poor zones. In contrast, Mn-rich zones lack primary fluid inclusions, but host re-equilibrated and decrepitated fluid inclusions (Quandt et al., 2018). Mn-rich zones are also characterized by higher δ 18 O values equivalent to precipitation temperatures <70°C. The high precipitation temperatures of Mn-poor zones correspond well to positive Eu/Eu* PAAS (>1.7), which are typical for ≥250°C hot hydrothermal fluids (Bau & Dulski, 1999;Sverjensky, 1984). Positive Eu/Eu* PAAS are preserved after cooling and decrease after mixing with seawater (Bau et al., 2010). Mn-rich zones show lower Eu/Eu* PAAS (1.1-1.5) than Mn-poor zones (>1.7). In accordance with seawater-like salinities of fluid inclusions from Mn-poor zones, Y/Ho ratios (40-56) tend to be higher in Mn-poor zones and consistently are above the minimum Y/Ho ratio for seawater (36) (Tostevin et al., 2016). Mn-rich zones have lower Y/Ho ratios (33-48) indicating stronger fluid-rock interaction. Ce/Ce* PAAS overlap but tend to be slightly higher in the Mn-rich zones.
While well-preserved primary fluid inclusions are located in Mn-poor growth zones (Quandt et al., 2018) and hence fluid inclusion temperatures refer only to these zones, the relatively large amount of sample material  (Quandt et al., 2018) from the same zones.

10.1029/2019GC008369
Geochemistry, Geophysics, Geosystems that is necessary for Sr and clumped isotope measurements represents a mixture of Mn-poor and Mn-rich growth zones. Therefore, the measured clumped isotope (123 ± 18°C) and fluid inclusion temperatures (~180-220°C; Quandt et al., 2018) are distinct from each other and require some further discussion.
The clumped isotope signal might be composed of two end-members that refer to the high-T fluid inclusionhosting Mn-poor zone and the Mn-rich zone whose formation temperature is unknown. However, the formation temperature of the Mn-rich zone has to be lower than 100°C in order to counterbalance the high-T (~180-220°C) Mn-poor zone and to result in the intermediate clumped isotope temperature (123 ± 18°C).
Since the growth zonation is oscillatory (i.e., multiple Mn-poor and -rich zones alternate), hot and cold fluid conditions should alternate a few times until fracture sealing is completed. We suggest that this represents an unlikely scenario. Instead, we suppose that the growth zonation is related to geochemical selforganization. This process describes how a closed system autonomously develops a pattern such as a growth zonation without an external input or control (e.g., Ortoleva et al., 1987;Reeder et al., 1990;Wang & Merino, 1992). Such processes are out of equilibrium and typically accompanied by complex oscillatory growth zonations, which are controlled by the growth-rate inhibiting Mn 2+ incorporation into the calcite (Reeder et al., 1990;Wang & Merino, 1992). Another high-T fluid inclusion-hosting blocky calcite shows a similar but much finer and more repetitive oscillatory growth zonation that is accompanied by (intra)sectoral zones. This sample is even less compatible with an environment of changing fluid compositions, because numerous small injections of Mn-rich fluids would be required. Geochemical self-organization is associated with closed-system disequilibrium precipitation. As a consequence, oxygen isotopic compositions and possibly REE+Y concentrations must be treated with caution.
The 87 Sr/ 86 Sr ratios of the respective sample are slightly higher than the estimated 87 Sr/ 86 Sr compositional range for hydrothermal fluids (0.70470-0.70590) from the Troodos ophiolite (Bickle & Teagle, 1992). REE +Y distribution patterns show seawater (positive Y anomalies) as well as hydrothermal characteristics (positive Eu anomalies). Based on these observations, we suggest that the high-T blocky vein calcite precipitated from seawater that was modified by fluid-rock interaction and/or mixing with a hydrothermal fluid. This is supported by clumped isotopes-based enriched parental fluid δ 18 O values relative to Cretaceous seawater. A potential hydrothermal fluid might have risen from depth along the faults that were observed in the sample area. Although the low 87 Sr/ 86 Sr sample ratios do not intersect the Sr isotope seawater curve, we suggest crystallization during the major volcanic phase between~92 and~90 Ma (Mukasa & Ludden, 1987; Figure 5) in order to provide sufficient heat from ongoing pillow lava emplacement or the magma chamber to achieve the high formation temperatures (~180-220°C).

Late-Stage Antitaxial Veining
Antitaxial vein calcites (CY17_2, CY17_6) are distinguished from all other vein calcite types by their slightly negative δ 13 C values (Figure 3), elevated 87 Sr/ 86 Sr ratios ( Figure 5), high Eu/Eu* PAAS accompanied by low Y/Ho ratios (Figure 7), and localized occurrence restricted to the Margi area (Table 1 and Figure 1). These δ 13 C and REE+Y characteristics suggest precipitation from slightly modified seawater. Slightly elevated parental fluid δ 18 O compositions relative to Cretaceous seawater support this or are the result of mainly increasing δ 18 O compositions after the Cretaceous (Zachos et al., 2001). Seawater modification may involve Sr exchange between seawater and host rock. Therefore, 87 Sr/ 86 Sr intersections with the Sr isotope seawater curve represent maximum precipitation ages and antitaxial vein calcites may have formed at any time between~75 Ma and today postdating Turonian-Santonian umber deposition (Bragina, 2008;Robertson, 1975). Micrites associated with antitaxial veins show similar geochemical signatures as antitaxial vein calcites, but indicate older maximum ages (~85 Ma). Their lamination points to sediment infilling into fractures.
The close spatial association of antitaxial calcite veins including micrites with hydrothermal umbers in the Margi area (Robertson, 1975) and the positive Eu/Eu* PAAS in both materials suggest some relation. The positive Eu/Eu* PAAS observed in the sample material might be related to precipitation from a hydrothermal fluid, but the late-stage formation of antitaxial calcite veins (≤75 Ma) after major volcanism (~92-90 Ma) argues against any hydrothermal fluid source. We therefore hypothesize that the hydrothermal umbers inherited its geochemical signature to the antitaxial vein calcites and micrite. This model requires fracturing of the pillow lavas and overlying sediments in order to inject unconsolidated calcareous sediment into fractures within the pillow lavas (i.e., neptunian dyke) where it is now preserved as micrite. Several N-S striking 10.1029/2019GC008369 Geochemistry, Geophysics, Geosystems normal faults run through the Margi area but are older than the maximum ages of micrite and antitaxial vein calcites (Boyle & Robertson, 1984). Instead, Troodos microplate rotation from Campanian to Early Eocene (Clube et al., 1985;Morris et al., 1990) and/or subduction reactivation and minor uplift in Late Oligocene to Early/Middle Miocene (Kinnaird et al., 2011;Main et al., 2016;Morag et al., 2016;Robertson, 1977Robertson, , 1998Robertson et al., 2012) might have triggered fracturing of the pillow lavas and overlying sediments. The proposed temporal framework for the formation of antitaxial calcite veins and associated micrites overlaps with both phases. Subsequent to fracturing, seawater pervaded the fractures and transported  (Alt, 1995;Fisher, 1998). A diffusive interface separates this seawater-like circulation cell from the hotter and more saline brine cell (Bischoff & Rosenbauer, 1989;Kelley et al., 1992;Nehlig, 1993). Extensional faults (Hurst et al., 1994;Varga et al., 1999) and fractures channelized seawater downflow and hydrothermal fluid upflow. (b) The majority of these fluid channels was sealed within~10 Myr by seawater-derived low-T blocky and syntaxial vein calcites. The permeability of the volcanic section decreases with depth (e.g., Coogan & Gillis, 2018, and references therein). This is reflected by higher vein density, greater vein thicknesses, and more complex vein networks in the UPL. In the LPL/BG veins occur as single straight veins or less complex vein networks. Low-T precipitation dominates throughout the whole pillow lava section, but high-T conditions and hydrothermal fluids may occur locally. High-T blocky vein calcites show a growth zonation associated with geochemical self-organization and precipitated from a seawater-diluted hydrothermal fluid. (c) Later veining is largely restricted to antitaxial calcite veins, which formed independently of fracturing and advective fluid flow. Their fibrous calcite morphology indicates diffusive fluid transport and crystallization pressure-driven vein growth. Variably modified seawater signatures are the result of fluid interaction with hydrothermal umbers during diffusive fluid transport. This implies that their formation ages may be younger than their intersections with the Sr isotope seawater curve suggest. Thus, antitaxial veining may temporally overlap with rotation or uplift. (d) Localized late-stage calcites may fill cavities in incompletely sealed analcime (Anl) veins that crosscut older blocky veins. The relative age of this crosscutting relationship is in accordance with intersections with the Sr isotope seawater curve.

10.1029/2019GC008369
Geochemistry, Geophysics, Geosystems unconsolidated calcareous sediment into the fractures where it interacted with the hydrothermal umbers. As a consequence, seawater acquired the hydrothermal signature of the umbers. This signature is now preserved in the micrites. Further fluid diffusion through the pillow lavas carried the hydrothermal signature to the site where calcite fibers precipitated. This fluid diffusion is indicated by antitaxial calcite fiber veins, which are generally interpreted as diffusion-fed structures that formed independently of fracturing solely due to the crystallization pressure of the calcite fibers (e.g., Elburg et al., 2002;Means & Li, 2001;Meng et al., 2019;Wiltschko & Morse, 2001).
Ce/Ce* PAAS , δ 13 C, and δ 18 O values vary along the calcite fibers and infer changing physicochemical conditions during their growth, e.g., variable fluid chemistry and/or growth rate. Occasionally, positive Ce/Ce* PAAS correlate with elevated concentrations of redox-sensitive Mn and imply a reduction to Mn 2+ that substitutes for Ca 2+ in the calcite lattice. Continuous Mn-controlled CL bands (Quandt et al., 2018) perpendicular to the fiber growth direction provide evidence for these short redox-reducing episodes that particularly occur during early-stage fiber growth near the median line. Clumped isotope temperatures (27 ± 8 and 20 ± 8°C) fall within the low-T range.
In summary, blocky and syntaxial veining occurred within~10 Myr after crust formation and contemporaneously with initial Late Cretaceous sediment deposition on the seafloor. Both processes reduced the permeability of the oceanic crust (e.g.,  and thus hampered further fluid circulation and secondary mineralization within the pillow lavas. Blocky and syntaxial veining and initial sediment deposition on the seafloor are postdated by antitaxial veining that is independent of fracturing and advective fluid circulation. This might imply that antitaxial veins form when the oceanic crust is largely sealed. Therefore, antitaxial veins may represent an efficient late-stage veining mechanism.

Conclusions
1. The Troodos SSZ oceanic crust (Figure 9a) was exposed to pronounced fracturing and subordinate extensional faulting. These fractures and faults increased the permeability of the oceanic crust, channelized seawater downflow, and provoked secondary mineralization. This secondary mineralization was dominated by low-T (<50°C) blocky and syntaxial vein calcites, which precipitated from pristine to slightly modified seawater under predominantly oxidizing redox conditions (Figure 9b). Together with Late Cretaceous sediment deposition on the seafloor (e.g., Chen & Robertson, 2019;Robertson, 1975Robertson, , 1977, these veins reduced the permeability of the oceanic crust within 10-20 Myr (~92-72 Ma) as indicated by intersections of 87 Sr/ 86 Sr ratios of samples with the Sr isotope seawater curve. These results are in accordance with and complement previously published studies that focused on the northern Troodos flank (Coogan et al., 2019;Gillis et al., 2015, and references therein;Gillis & Robinson, 1985Weinzierl et al., 2018). 2. Geochemical signatures of localized high-temperature blocky vein calcites (up to~220°C based on fluid inclusions) point to high-temperature fluid-rock interaction or the involvement of hydrothermal fluids. They are characterized by Mn-controlled oscillatory growth zonations, which probably developed in a closed system out of equilibrium ( Figure 9b). Hence, stable and clumped isotopic compositions must be treated cautiously. High-temperature blocky vein calcites probably formed during volcanic activity between~92 and~90 Ma. 3. Subsequent secondary calcite mineralization is largely restricted to low-T (<30°C) antitaxial calcite veining ( Figure 9c) and very localized cavity fillings (Figure 9d). The slightly modified seawater signatures of antitaxial vein calcites are the result of diffusion and fluid interaction with hydrothermal umbers, which overlie the pillow lavas. Antitaxial calcite veining initiated at~75 Ma or later and lasted until~67 Ma at least. Spatially associated micrite probably represents remobilized and reworked sediment. 4. The investigated sample suite defines a curvilinear trend from pristine seawater precipitates such as syntaxial vein calcites to antitaxial and high-temperature blocky vein calcites characterized by fluid-rock interaction and/or the involvement of a hydrothermal fluid. 5. Calcite vein types are distinguished from each other by their unique geochemical compositions that are related to distinct crystal growth mechanisms. Low-T blocky calcites precipitated into fractures or cavities filled with seawater. As a consequence of extended fluid residence times, fluid-rock interaction slightly modified the seawater composition. Rapid crack and sealing in contrast prevented extensive 10.1029/2019GC008369 Geochemistry, Geophysics, Geosystems fluid-rock interaction and resulted in distinct seawater signatures of syntaxial vein calcites. Antitaxial veining is mainly driven by diffusion and the crystallization pressure of calcite fibers. This decoupling from fracturing/faulting and advective fluid flow probably represents an effective veining mechanism within widely sealed oceanic crust in general. 6. This study demonstrates that a combination of microtextural and multiproxy geochemical analyses (e.g., REE+Y, stable, radiogenic and clumped isotopes) is essential in order to understand the timing and variable physicochemical environment of vein mineral growth. Without the discovery of potential growth zonations or crack and sealing textures, geochemical details would remain undetected and interpretations could be oversimplified.