The Origin of Carbonate Veins Within the Sedimentary Cover and Igneous Rocks of the Cocos Ridge: Results From IODP Hole U1414A

Abstract Carbonate veins in the igneous basement and in the lithified sedimentary cover of the Cocos Ridge at International Ocean Discovery Program (IODP) Hole 344‐U1414A reveal the hydrologic system and fluid‐rock interactions. IODP Hole 344‐U1414A was drilled on the northern flank of the Cocos Ridge and is situated 1 km seaward from the Middle America Trench offshore Costa Rica. Isotopic and elemental compositions were analyzed to constrain the fluid source of the carbonate veins and to reveal the thermal history of Hole 344‐U1414A. The formation temperatures (oxygen isotope thermometer) of the carbonate veins in the lithified sedimentary rocks range from 70 to 92 °C and in the basalt from 32 to 82 °C. 87Sr/86Sr ratios of the veins in the altered Cocos Ridge basalt range between 0.707307 and 0.708729. The higher ratios are similar to seawater strontium ratios in the Neogene. 87Sr/86Sr ratios lower 0.7084 indicate exchange of Sr with the igneous host rock. The calcite veins hosted by the sedimentary rocks are showing more primitive 87Sr/86Sr ratios <0.706396. The isotopic compositions indicate seawater, modified into a hydrothermal fluid by subsequent heating, as the main fluid source. Low‐temperature alteration and the presence of a high‐temperature fluid resulted in different carbonate precipitates forming up to several cm thick veins. The geochemical data combined with age data of the sedimentary rocks suggest intraplate seamount volcanism in the area between the Galapagos hot spot and the Cocos Island as an additional heating source, after the formation of the Cocos Ridge at the Galapagos hot spot.


Introduction
At the erosive plate boundary offshore Costa Rica the Cocos Plate is being subducted under the Caribbean Plate and the highstanding aseismic Cocos Ridge (CCR) has lifted the seismogenic zone into the accessible reach of scientific oceanic drilling. Drilling on the subducting and overriding plate was achieved as part of the Costa Rica Seismogenesis Project (CRISP) during International Ocean Discovery Program (IODP) Expedition 344. CRISP has the main objective to clarify processes that control nucleation and seismic rupture of large earthquakes at erosional subduction zones (Harris et al., 2013a). A particular target is how the hydrogeochemical processes within the subduction zone affect seismicity and chemical alteration (Solomon et al., 2009). Circulating and trapped fluids within the sediments and ocean crust of the incoming plate can be incorporated into the subduction system, enter the subduction zone and may be released into fracture systems within the overriding plate. The understanding of the impact of these fluids on the material in the subduction zone is also essential for a better understanding of the seismogenesis in subduction zones. Therefore, vein-hosting sedimentary and igneous rock material from IODP Hole U1414A, located on the incoming Cocos Plate 1 km seaward from the Middle America Trench offshore the western margin of Costa Rica (Figure 1a), was investigated. The results from IODP Hole U1414A describe the hydrologic system and identify fluid pathways and fluid-rock interactions within the upper crustal sections of the oceanic Cocos Plate (Harris et al., 2013b).
The sedimentary and igneous rocks of IODP Hole U1414A are transected by various hydrothermal veins that are mainly filled by carbonates. These hydrothermal veins host inclusions of low to seawater-like salinity fluids (salinity from 0 to 3.4 mass% NaCl, except vesicles with ca. 5 mass% NaCl) trapped over a wide temperature range from 40 to approximately 400°C. These fluid inclusions reveal multiple fluid entrapment stages suggesting that fluid flow occurred over several episodes (Brandstätter et al., 2016).

RESEARCH ARTICLE
Hydrothermal fluid flow in the uppermost oceanic crust is generally characterized by penetration of cold seawater into the porous, permeable igneous rocks. This seawater heats up with depth and evolves into hydrothermal fluids that become buoyant and ascend toward the seafloor as hot vent fluids (Alt, 1995). This circulation can result in strong fluid-rock exchange causing major chemical changes to the host basalts and influence the chemistry of the seawater (e.g., Alt & Teagle, 1999;Coggon et al., 2004Coggon et al., , 2010Staudigel et al., 1981Staudigel et al., , 1996. Evidence of such fluid circulations in the upper oceanic crust is provided by hydrothermal veins and by mineralogical changes within the host basalts. Veins can give important information about the source and composition of the migrating fluids and fluid circulation and offer the opportunity to reveal the thermal history, past fluid circulation and fluid-rock interactions (e.g., Blyth et al., 2000;Burkhard & Kerrich, 1988;Dietrich et al., 1983;Drake & Tullborg, 2009;Rye & Bradbury, 1988). Vein minerals in both the igneous basement rocks and the overlying lithified sedimentary rocks sampled from IODP Hole U1414A allow the reconstruction of element ratios in the fluids from which they are precipitated (Coggon et al., 2004(Coggon et al., , 2010. In this study, we present analyses of the elemental and isotopic composition of calcite and aragonite veins (CCV) and of the surrounding host rocks to unravel the thermal and fluid history at IODP Hole 344-U1414A and to clarify the temperature variations derived from fluid inclusions. Oxygen and strontium isotopic compositions of CCV provide useful constraints on fluid sources and reflect the extent of the chemical exchange between seawater and the host rocks of the upper oceanic crust (Bach et al., 2011;Coggon et al., 2004;Muehlenbachs & Clayton, 1976). We used 87 Sr/ 86 Sr ratios, oxygen, carbon isotopic, and elemental compositions, particularly the rare earth elements and Yttrium (REY), to define the fluid source and pathways of paleofluids through the sedimentary cover and the igneous basement, and to reveal fluid-rock interaction processes.

Geological Setting and Sample Description
The investigated CCV were sampled in drill cores recovered from IODP Hole U1414A, located 1 km seaward of the Middle America Trench offshore Costa Rica, where the Cocos Plate, that previously formed at the Cocos-  Nazca spreading center (also known as Galapagos spreading center), subducts beneath the Caribbean Plate ( Figure 1a). This region is a seismically active area with a history of Mw > 7 (Mw = moment magnitude scale) earthquakes (Harris et al., 2013a). The complex history of the Cocos Plate, formed at the Cocos-Nazca spreading center (also known as Galapagos spreading center), is dominated by the interaction of the Galapagos hot spot and the Cocos-Nazca spreading center resulting in the formation of the highstanding NE-SW trending aseismic CCR (Barckhausen et al., 2001;Brandstätter et al., 2016;Harpp et al., 2005;Hoernle et al., 2002;Meschede & Barckhausen, 2000). The NE end of the tholeiitic CCR was dated at 13.0-14.5 Ma (Werner et al., 1999) and tephra layers within soft sediment layers from Site U1381, about 11 km southeast from Hole U1414A, record Plinian eruptions from 16.5 to 8 Ma (Schindlbeck et al., 2015). A younger bathymetric features in this area are the~2 Ma alkalic Cocos Island volcanoes and an adjacent group of seamounts ( Figure 1a) that form the Cocos Island Province (Harpp et al., 2005;Schindlbeck, Kutterolf, Freundt, Andrews, et al., 2016;Werner et al., 2003).  . The presence of a Late Miocene to Pliocene sediment hiatus (1.5 to 8 Ma) at Site U1381 is in strong contrast to the continuous sedimentation of the closely located Site U1414 and could be the result of temporally and regionally interplays of morphology, erosion, and biogenic productivity .
The igneous basement broadly comprises aphyric to highly phyric massive and thin basaltic flows. The abundance of phenocrysts, such as plagioclase, clinopyroxene, and olivine, varies. The basalts are slightly to strongly hydrothermally altered, characterized by brecciation, partial replacement of groundmass mineralogy, partial to complete replacement of phenocrysts, particularly olivine and clinopyroxene, and veins, some with alteration halos. There is a lack of oxidative secondary minerals, such us iron-oxyhydroxide and celadonite (Harris et al., 2013b). Vesicle abundance varies throughout the different igneous basement units, and vesicles are partially to completely infilled by secondary minerals, particularly saponite and smectite, and fibrous calcite and quartz (Harris et al., 2013b).
Hydrothermal veins are distributed within the lithified sedimentary rocks of Unit III and the igneous basement. In the lithified sedimentary rocks, the veins are mainly filled by blocky, coarse calcite, crosscutting the rarely appearing fibrous radial quartz veins. Noticeable are wall rock fragments embedded within the vein filling.
The vein filling in the basalt consists of clay minerals (smectite and saponite), calcium carbonates, quartz, and pyrite. All four minerals occur in centimeter-thick veins, in some cases with several generations of calcium carbonate. Pyrite is present in veins and in both host rocks. Veins in the basalt are more or less surrounded by a brown selvage of predominantly clay minerals resulting from basalt alteration. The veins in the basalt occur within different growth morphologies: syntaxial, antitaxial, and composite veins (Brandstätter et al., 2016). Syntaxial veins are mainly composed of blocky or coarse aragonite, antitaxial veins solely of fibrous calcite. Composite veins show spherulitic blocky quartz in the center, sometimes overgrown by small calcite grains. Quartz is surrounded by fibrous and/or blocky calcite and a clay selvage (Brandstätter et al., 2017).
The CCV hosted by the sedimentary units and the CCR basalt are described as CCV-S and CCV-B, respectively, in the following chapters.

Methods-Analytical Techniques
The selection of samples for the isotopic and elemental composition analyses were based on microscopic thin section observations, fluid inclusion data (Brandstätter et al., 2016), and aim to represent the entire sequence of the recovered drill core. Sedimentary and basaltic host rock samples were drilled manually with a hand-held carbonite-tipped drill, and CCV were drilled with a computer-controlled micromill-sampler (ESI New Wave, drilling head 100 μm). Visibly altered portions and heterogeneous areas of the veins were avoided. Whole rock samples of the host rocks, the extracted carbonate phase of the powder of the bulk sedimentary host rock, and the CCV from both rock types were analyzed for strontium isotopic and elemental composition in the Isotope Geochemistry

Strontium Isotopic and Elemental Composition
Ten CCV-S from sedimentary Unit III and seven CCV-B (five calcite and two aragonite) from the basalt, the corresponding host rocks as well as the extracted carbonate phase of the siliceous and calcareous cemented sedimentary host rock, were analyzed for elemental composition and strontium isotopes in the Isotope Geochemistry Laboratories of the National Oceanography Centre Southampton, University of Southampton. Carbonate samples were dissolved in 5% acetic acid, the carbonate phases were leached from the sedimentary host rocks with 5% acetic acid, and the whole rocks were digested using HF/HNO 3 . Trace and selected major elements were measured by inductively coupled plasma-mass spectrometry using a Thermo Scientific X-Series II. Samples and standards were spiked with internal standard elements and corrected for interferences and blank and then calibrated using a suite of international rock standards (JB-3, JB-1a, JGb-1, BHVO-2, and BIR-1, reference data from GeoReM database; Jochum et al. 2005). Long-term accuracy relative to reference values is 3-5%. Sr was isolated with 80 μl Sr-Spec columns and eluted with 3 M HNO 3 . The dried down Sr fraction was loaded onto a single Ta filament with a Ta activator solution and analyzed on a Thermo Scientific Triton Thermal Ionization Mass Spectrometer, using a static procedure with amplifier rotation (on a 88 Sr beam of 2 V) for 300 ratios. Fractionation was corrected using an exponential correction normalized to 86 Sr/ 88 Sr = 0.1194. NIST 987 was run as a reference standard and the long-term average (150 analyses) on this instrument is 0.710245 ± 0.000025 (2SD).
Additionally five samples, two fibrous calcite vesicles, and three veins with fibrous calcite and blocky quartz were dissolved by dilute (2%) HNO 3 to avoid dissolution of clay minerals. Sr was separated using Sr-specific extraction chromatographic resin (Eichrom) columns. About 5 ml of 3 M HNO 3 were used for element elution and 1 ml 0.1 M HNO 3 for Sr collection. Strontium isotopic ratios were determined on a Nu Instruments Plasma II multicollector-inductively coupled plasma-mass spectrometry at the NAWI Graz Central Lab for Water, Minerals and Rocks, Graz University of Technology. NIST SRM 987 yielded 87 Sr/ 86 Sr ratios of 0.710253 ± 0.000067 (2σ; n = 12) and all samples were corrected relative to the value of 0.710250 for NIST SRM 987.

Results
The carbon sources can be constrained by the carbon isotopic composition of the calcite, reflecting the δ 13 C of dissolved CO 2 (Ohmoto & Rye, 1979). The oxygen isotopic composition of the vein calcite relates to the δ 18 O of the water from which it precipitated and the temperature-dependent oxygen isotope fractionation between calcium carbonate and the fluid (Kim & O'Neil, 1997;Urey, 1947;Zheng, 1999). Fields of typical marine limestones, burial cements, and hydrothermal calcites and a mixing line of seawater carbon and mantle-derived carbon are plotted in Figure 2c (Nelson & Smith, 1996;Rollinson, 1993). The 87 Sr/ 86 Sr ratio of the calcium carbonate veins records the Sr-isotopic composition of the fluids from which they precipitate, that in turn reflects the composition of contemporaneous seawater, pore fluid compositions, and fluid-host rock Sr exchange (e.g., Coggon et al., 2004Coggon et al., , 2010. The chemical evolution of the fluids from which the CCV were precipitated is reflected by changes in the trace element concentrations (Coggon et al., 2004(Coggon et al., , 2010.

Isotopic Composition 4.1.1. Carbon and Oxygen Isotopes
Carbon isotopes of the CCV-S and CCV-B vary between À2.7 and À0.6‰ and À2.9 and 0.7‰, respectively, with a slight increase of δ 13 C with depth in the basalt (Table 1 and Figure 2a). The aragonite veins in the basalt have relatively low δ 13 C values from À2.9 to À1.5‰. The higher values are typical for modern seawater (around 0‰), whereby the lower values indicate low-temperature interaction of seawater or pore fluids with the basalt (e.g., Alt & Teagle, 2003). The δ 18 O values of the CCV-S cluster in a narrow range (17 to 19.7‰ SMOW), in contrast to the CCV-B that show a much wider range of values from 18.5 to 28.2‰ SMOW, indicating various formation temperatures of CCV-B (calcite and aragonite; Figure 2b). The stable isotopic values of the corresponding sedimentary host rock lie between 0.6 and 1.9‰ for carbon isotopes and between 21.0  Teagle et al., 1996). The dashed line shows mixing between mantle-derived carbon and seawater carbon (Stakes & O'Neil, 1982) and in black the field of the isotopic composition of the mantle (Pineau & Javoy, 1983). Fields of isotopic compositions of carbonates derived from Nelson and Smith (1996) and Rollinson (1993).

10.1029/2018GC007729
Geochemistry, Geophysics, Geosystems and 28.0‰ SMOW for the oxygen isotopes (Table 1). In Figure 2c carbon and oxygen isotope fields for burial cements, marine limestones, and hydrothermal calcites (Nelson & Smith, 1996;Rollinson, 1993) are represented. CCV in oceanic crust from Site 801 and Hole 896A, the calcareous sedimentary host rock, and the majority of the CCV-B plot in the field of marine limestones, whereas the CCV-S are mainly in the field of hydrothermal calcites.

Strontium Isotopes
Strontium isotopes derived from shipboard measured pore fluids within the upper poorly indurated part of Unit III show a decrease of 87 Sr/ 86 Sr ratios with depth, from 0.70884 at 308 m below seafloor to 0.70861 at the deepest measureable sample (Figure 3 and Table 2). Intense carbonate recrystallization and cementation inhibited the liberation of pore fluids below 337 m below seafloor (Ross et al., 2015).
The analyzed vein material of sedimentary Unit III obtained 87 Sr/ 86 Sr ratios from 0.70548 to 0.70639 and the corresponding sedimentary host rocks from 0.707481 to 0.708517 (the extracted carbonate phase of the host rock from 0.708403 to 0.708561).
The CCR basalt yield typical 87 Sr/ 86 Sr ratios for altered basalt (Dasch et al., 1973); however, the basaltic host rock of the two aragonite samples shows lower ratios, in direction to fresh basalt. The CCV-B yield 87 Sr/ 86 Sr ratios from 0.70729 to 0.70873 (Table 2). The measured 87 Sr/ 86 Sr ratios of the majority of the CCV-B and the carbonate phase of Unit III are close to the range of the seawater composition during Miocene times ( Figure 3). Sr isotopic composition of seawater increased from approximately 0.7088 in the Middle Miocene to approximately 0.70916 in recent times (McArthur et al., 2001). 87 Sr/ 86 Sr ratios of CCV-B with quartz as additional mineral phase obtained lower 87 Sr/ 86 Sr ratios than syntaxial, blocky aragonite veins, and the fibrous calcite veins.   Geochemistry, Geophysics, Geosystems the fluid from which the carbonates were precipitated, with due regard to the temperature of vein formation and the knowledge about the temperature dependence of element partitioning between mineral and fluid (Coggon et al., 2010).
REE + Y (REY) data of all carbonate rocks is presented in Post-Archean Australian Shale (PAAS)-normalized diagrams (Pourmand et al., 2012;Taylor & McLennan, 1985). Shale-normalized (SN) elemental concentrations were used to calculate elemental anomalies. Different REY anomalies were used to constrain the fluid sources and the paleo-redox conditions, normalized to PAAS (Pourmand et al., 2012). REY data of the CCR basalt are normalized to REE chondrite (Sun & McDonough, 1989). REEs occur in trivalent state and behave as a coherent group with a variation in chemical properties (e.g., Zhong & Mucci, 1995). Ce and Eu have the ability to adopt different oxidations states in aqueous solutions and can be used to constrain precipitation environments of carbonates (Bolhar et al., 2004;Bolhar & Van Kranendonk, 2007). Trivalent Ce can be oxidized to less soluble tetravalent Ce in oxygenated water; this fractionation leads to a negative Ce anomaly (Ce SN / Ce SN * = Ce SN /(Pr SN^2 )/Nd SN ; Lawrence et al., 2006;Tostevin et al., 2016) in modern well-oxygenated seawater. Under extremely reducing conditions and higher temperatures, trivalent Eu can be reduced to soluble divalent Eu (eg., Bau, 1991;Danielson et al., 1992;Wang et al., 2014). Positive Eu anomalies (Eu SN / Eu SN * = 2*Eu SN /(Sm SN + Gd SN ); Lawrence et al., 2006;Tostevin et al., 2016) are typical for hydrothermal vent fluids, can be the result of plagioclase recrystallization reactions during magmatic rock alteration, and are mainly regarded as an indicator for hydrothermal input (e.g., Bolhar & Van Kranendonk, 2007;Wang et al., 2014;Wheat et al., 2002;Zhang et al., 2015). Positive La anomaly (La SN /La SN * = La SN /(Pr SN *(Pr SN /Nd SN )^2); Lawrence et al., 2006) and positive the Gd anomaly (Gd SN /Gd SN * = Gd SN /((2*Tb SN )-Dy SN ); Bolhar et al., 2004) are characteristic for modern seawater REY systematics. The Yttrium anomaly (=Y/Ho ratio) serves for the identification of the sedimentary origin and different water types/fluid sources (e.g., Bau & Dulski, 1999;Bolhar et al., 2004;Bolhar & Van Kranendonk, 2007;Wang et al., 2014;Zhang et al., 2015). The twin elements Y and Ho are geochemically similar but can differ in their complexation behavior and behave incoherently in marine environment due to lower stabilities of surface complexes of Y (e.g., Bau et al., 1997;Bolhar & Van Kranendonk, 2007;Pack et al., 2007). For modern oxygenated seawater the typical Y/Ho ratio is >45 and for marine carbonates 44-74 (Bau et al., 1996;Bolhar & Van Kranendonk, 2007). With increase of water depth and under anoxic conditions the molar ratio decreases, due to preferential sorption of Ho relative to Y on Fe and Mn oxyhydroxides (Bau & Dulski, 1999 Geochemistry, Geophysics, Geosystems mol; Table 3). The Mg concentration in the calcite veins within the basalt shows a broader distribution compared to CCV-S (Figure 4).

Fluid Source of the CCV
The CCV-S and CCV-B can be clearly distinguished by their oxygen and strontium isotopic composition, except three transitional samples (from 53R-1-W 126/131 to 61R-1-W 45/49), which are composite CCV with quartz yielding similar oxygen isotope values like the CCV-S and less radiogenic 87 Sr/ 86 Sr ratios than the other CCV-B ( Figure 6). A further CCV with quartz (sample 57R-1-W 38/43) indicates lower temperatures than the other transitional samples. This, however, could be attributed to the sample powder for the oxygen isotopic analysis, which was taken from the outer vein area, where fibrous calcite is intergrown with clay  Geochemistry, Geophysics, Geosystems minerals. Formation temperatures (oxygen isotope thermometer) of a similar vein sample (53R-1-W 126/131), where the powder was taken from the vein center, are significantly higher. There is a positive correlation between the strontium ratios and the oxygen isotopic composition of the CCV in both rock types, that is, an increased temperature with decreased 87 Sr/ 86 Sr ratios ( Figure 6). Less radiogenic strontium ratios results from seawater-basalt and seawater-sediments interaction and suggests higher formation temperatures for the CCV (Alt & Teagle, 2003).
Sr/Ca and Mg/Ca ratios of the CCV-B trend to higher values with higher formation temperature and broader distribution, compared to CCV-S, which could be attributed to multiple calcite precipitation phases within the basalt. The results of this study confirm the observation of Coggon et al. (2004) that calcites precipitated from a hotter fluid, tend to have rather basaltic, less radiogenic 87 Sr/ 86 Sr ratios, and lower Mg/Ca and Sr/Ca concentrations (Figures 4 and 6). The CCV-S yield only limited ranges of Sr/Ca and Mg/Ca ratios for the whole temperature range, compared to the carbonate phase of the host sediments that have higher and more broadly distributed Sr/Ca and Mg/Ca values. The leached carbonate phase from the bulk sedimentary host rock is more heterogeneous, due to primary calcareous components and younger calcareous cements, and consists of more radiogenic strontium ratios than the bulk sedimentary host rock. This difference is presumably due to the silicate-derived 87 Sr, the existence of cryptic microveins and recrystallization in the bulk rock can be excluded as 87 Sr source, due to the dissolution from the bulk sedimentary host rock.
The fluid source of the less radiogenic CCV-S with higher formation temperatures could be a deeper sourced hot fluid that altered the basalt and penetrated the lithified sedimentary rocks of Unit III. The lower temperatures determined for the CCV-B suggest that the first vein formation in the basalt was dominated by invaded seawater interacting with the cooling porous basalt.
The carbon isotopic values of the CCV yield a range from À3 to +0.7‰. The higher values are typical for marine carbonates (mean δ 13 C value around 0‰; Nelson & Smith, 1996) and the lower values indicate a fluid source more enriched in carbon, for example, from biogenic carbon and/or from a carbonic fluid derived from the altered CCR basalt. Magmatic carbon has a δ 13 C value around À5‰, mantle-derived rocks have δ 13 C values in a range of À8 to À4‰ PDB (Pineau & Javoy, 1983), and vent fluids show values from À8 to À2‰ (Craig et al., 1980;Shilobreeva et al., 2011). Biogenic material, however, is rare; few fossils of radiolarians were found in Unit III (Sandoval et al., 2017). Compared with the results of Stakes and O'Neil (1982), the carbon isotopic values in this study plot within the field of calcite veins hosted by hydrothermally altered midocean-ridge basalt.
CCV in basalts of ODP drill sites located on the Cocos Plate (Hole 896A) and Pacific Plate (Site 801) show generally positive δ 13 C values with a mean value of 0.6 and 2.1‰, respectively (Alt & Teagle, 2003;Teagle et al., 1996). The interaction of seawater with the basalt and the sedimentary host rocks is perceptible in the δ 13 C values and the 87 Sr/ 86 Sr ratios. Compared to the strontium isotope seawater curve (McArthur et al., 2001), the pore fluids from sample 38R-2, 66-102 cm, indicate an age around 17 Ma, whereas the Ar/Ar dating of tephras from sample 38-R, 76-78 cm yield an age of 12 Ma . This suggests that the isotopic composition of the pore fluids was modified by fluid-rock interaction. Correlation of the CCV-B with the strontium isotope stratigraphy (McArthur et al., 2001) make sense for five samples, assuming that the maximum age of the CCR basalt of Hole U1414A is 20 Ma. One vesicle sample, two fibrous CCV-B, and two aragonite samples show a decrease of age with decreasing depth. The CCV-B with quartz as additional mineral phase are showing the opposite trend (Figure 3). Modern seawater contains a large range of Y/Ho from 33 to 40 in near shore or restricted settings to 40-70 in open marine settings (Bau & Dulski, 1999;Bolhar & Van Kranendonk, 2007;Nozaki & Zhang, 1995;Zhang et al., 1994). Four CCV-B samples (two calcite and two aragonite) plot together with the carbonate phase samples of the sedimentary host rock in the field of near shore or restricted setting for seawater. This confirms the assumption that the first vein formation in the basalt and the lithification and cementation of Unit III sedimentary rocks are related. This happened probably near the ridge axis of the Cocos-Nazca spreading center at shallow water conditions or in an area with restricted connection to an open ocean system (Figure 8a). Volcanological, geochemical, morphological, and geophysical data indicate that parts of the CCR formed under subaerial or shallow-marine conditions (Hoernle et al., 2000;Werner et al., 1999). Compared to oxic and anoxic seawater (Bau et al., 1997), the elemental anomalies of the CCV-B plot rather close to anoxic seawater and the anomalies of the CCV-S closer to oxic seawater, except the Ce anomaly. The Y/Ho values of the CCV-S plot in the field of seawater in open marine settings. This would imply that the formation of these veins happened off ridge axis somewhere on the way from the Cocos-Nazca spreading center to the Middle America Trench (Figure 8b). However, strontium isotope ratios of the CCV-S preclude pure seawater as fluid source and it should be taken into account that various parameters, such as pH, redox state, temperature of the system, and sorption along the pathway of the fluid control the REY exchange between the fluid and mineral (Debruyne et al., 2016).
The CCV-B and CCV-S shale normalized REE distribution patterns show HREE enrichment relative to LREE and MREE and positive La anomaly, which is typical for marine carbonates and indicating a derivation of the REE from seawater (Bolhar et al., 2004;Tostevin et al., 2016;Webb & Kamber, 2000;Zhang & Nozaki, 1996, 1998. The enrichment of HREE in the seawater is attributed to their different complexation behavior; where LREE and MREE are more preferentially adsorbed and HREE are preferentially remain in solution (e.g., Bolhar et al., 2004). Modern seawater is depleted in Ce, resulting in a negative Ce anomaly (e.g., Bolhar et al., 2004;Nozaki, 2001). Except for one sample (40R-1-W 6/9), a clear negative Ce anomaly is absent. The weak positive Ce anomaly values for CCV-B and the lack of Fe-oxyhydroxides and celadonite imply that the veins precipitated more likely under reduced conditions. Possible Ce mobilization during basalt alteration could result in Ce enrichment in the veins (Patino et al., 2003). The Europium anomaly is generally regarded as an indicator for hydrothermal input (e.g., Bolhar & Van Kranendonk, 2007;Wang et al., 2014;Wheat et al., 2002). Both CCV have slightly positive to positive values and indicate an influence of a hydrothermal fluid; this applies particularly to the CCV-S and to the CCV-B in highly altered basalt. The REY and isotopic values suggest that seawater and modified hot seawater are the main fluid source for both CCV. Fluid circulation occurred through both host rocks, resulting in a basaltic hydrothermal input in the CCV-S and in the CCV-B. The elemental and isotopic compositions of the CCV-B confirm the results of previous analyses that carbonate precipitation occurred in multiple stages (Brandstätter et al., 2017).

Fluid-Rock Interaction
The elemental composition of CCV can be strongly influenced by the chemical development of the fluid, by fluid-rock interaction, as well as by alteration and leaching of elements from the host rock (e.g., Coggon et al., 2004). Certain major and trace elements, such as Mn, Fe, Mg, Na, and Sc, indicate fluid rock interaction in   Alt & Teagle, 2003) and from Hole 896A in young, 6.8 Ma crust (Teagle et al., 1996). terms of leaching from the basaltic host and incorporation into the CCV-B. This geochemical exchange is confirmed by chondritic Y/Ho values (26-28) for CCV-B in highly altered basalt (Figure 7a). Elevated Mg, Mn, and Fe values for the calcite veins in the basalt, compared to the aragonite veins and CCV-S, suggest element mobilization due to basalt alteration (e.g., partially altered to complete dissolved olivine; Bischoff & Dickson, 1975). At ODP Site 801 Alt and Teagle (2003) found highly altered basalts associated to deposits of low-temperature hydrothermal fluids, indicating losses of Fe, Mn, Mg, Ca, Na, and Sr.
Elevated and scattered values for the lithophile elements Zr, Hf, and Th of CCV-S samples indicate the imprint of the siliceous host rock (e.g., Bolhar & Van Kranendonk, 2007). Particularly sample 40R-1-W 6/9 shows strongly enriched values of Th, Hf, and Zr and anomalies differ compared to the other CCV-S, which can be the result of contamination of the inhomogeneous sedimentary host rock during drilling of the >1-cm-thin calcite vein (e.g., Bolhar & Van Kranendonk, 2007). The U concentration of the sedimentary rocks can depend on the depositional and environmental conditions. Rapid sediment deposition limits U concentration due to slow diffusion rates below the sediment-water interface (Wignall, 1994). Furthermore, under oxic conditions U is soluble; hence, under reducing conditions U may be more enriched (Algeo & Maynard, 2004). The U content is also strongly influenced by fluid-rock interactions (Bach et al., 2011). A depletion in U is attributed to hydrothermal interaction of the fluid with the igneous host rock (Bach et al., 2011;Michard et al., 1983), which is also seen in our samples with lower U concentrations in the CCV-B. Particularly the strontium isotopic composition of CCV indicates intense exchange of Sr between the fluid and the host rocks.

Thermal and Tectonic Evolution
First fibrous calcite and clay mineral vein formation in the CCR basalt occurred shortly after the basalt formation close to the Cocos-Nazca spreading center. Veins were formed, due to temperature differences between the basalt and the seawater, resulting in thermal contraction and seawater invasion near or off-axis ( Figure 8a; Brandstätter et al., 2017). The similar trend of elemental and isotopic composition of some CCV-B and the carbonate phase of the sedimentary host rock suggest that first CCV-B precipitation was followed by sediment cementation and lithification from the same fluid source. The invaded seawater was heated up to 60°C by the slowly cooling basaltic lavas. The aragonite veins show the highest 87 Sr/ 86 Sr ratios (~0.7087) and low formation temperatures. Due to their localized occurrence in the upper part of the igneous basement, the time of formation is not clear but could be around 17 Ma according to the strontium seawater curve.
Previous fluid inclusion studies showed a similar fluid chemistry for both CCV with a mean salinity of 2.4 mass % NaCl, and with 5.0 mass % NaCl for vesicles. The fluids yield temperatures up to 400°C in CCV and quartz veins and nearly undeformed, intragranular fluid inclusions yield entrapment temperatures from approximately 70 to 200°C in the CCV in both host rocks (Brandstätter et al., 2016). These temperatures are much higher compared to the calculated carbonate formation temperatures (from δ 18 O values) of the CCV-S (60°C to 90°C) and CCV-B (32°C to 82°C). Bach et al. (2011) described similar differences in temperatures of CCV in peridotite fault rocks. The isotopic and elemental characteristics imply that the CCV-S and the transitional CCV-B were precipitated from possibly 350°C hot hydrothermal fluids, but the calculated formation temperatures, derived from δ 18 O values, were significantly lower. This is attributed to conductive cooling during slow ascent from deeper sources (Bach et al., 2011). In this study, the formation temperatures, derived from oxygen stable isotope compositions, present the precipitation temperatures of the carbonates, whereas the high temperatures of fluid entrapment in quartz precipitates may result from convective and/or advective heating of the fluid. All CCV, independent of the host rock, and the carbonate phase of the sedimentary host rock indicate in varying degrees hydrothermal influence, shown by a slight positive to positive Eu anomaly, which represents the imprint of the altered basalt and the exchange with the fluid. The stable and strontium isotopic compositions of the CCV suggest also the presence of a hot, CO 2 rich fluid. This implies that after lithification of the sediments high temperatures, derived from a shallow magma chamber or intrusives, affected the CCR basalt and the sedimentary cover ( Figure 8b). This could be the trigger of hydrofracturing in the lithified sedimentary rocks and subsequent vein formation, as well as precipitation of quartz in some CCV-B (Brandstätter et al., 2016(Brandstätter et al., , 2017. Brandstätter et al. (2016) discussed three potential heat sources for an additional heating event, (a) the Cocos-Nazca spreading center, due to the ridge jump at 14.5 Ma; (b) a combination of Cocos-Nazca spreading center and Galapagos hot spot activity; or (c) seamount intraplate volcanism in the area of Cocos Island in the Plio-Pleistocene. Schindlbeck, Kutterolf, Freundt, Andrews, et al. (2016) cite a number of proposed models (Castillo et al., 1988;Harpp et al., 2005;Herbrich et al., 2015;O'Connor et al., 2007;Werner et al., 2003) to explain such young volcanism along the CCR, and the models coincide that Cocos Island Seamount Province volcanism has its origin from plume-related mantle movements. The special location of the Cocos Island at the eastern edge of an extinct small spreading center and the delimitation from the thickened lithosphere of the CCR by a transform fault could be the reason for the significant larger melt catchment area of Cocos Island compared to adjacent seamounts farther away from the transform fault (Schindlbeck, Kutterolf, Freundt, Andrews, et al., 2016). Herbrich et al. (2015) propose that contrary to the CCR the Cocos Plate Seamount province has its origin on EPR generated crust/lithosphere. The formation of the seamounts is the result of northward transport of plume material from the Northern Galapagos area and incorporation of depleted asthenosphere material. By reaching the thinner EPR lithosphere, melting occurs due to decompression or is the result of upwelling at the edge of a viscous relict root of accumulated plume material at the base of the lithosphere (Herbrich et al., 2015). The models of Schindlbeck, Kutterolf, Freundt, Andrews, et al. (2016) and Herbrich et al. (2015) endorse that there is a melt catchment area between the Galapagos hot spot and Cocos Island with lateral distribution towards the lithosphere of the CCR.
The elemental data presented in this study, particularly the high Y/Ho ratios for the CCV-S, and the tephra age in Unit III, suggest rather intraplate seamount volcanism northward of the Galapagos hot spot as possible heat source, than the Cocos-Nazca spreading center. Due to fast plate tectonic movement of 96 mm/a (>11 Ma) to 72 mm/a (<10 Ma), Hole 344-U1414A was between the Cocos Island and the spreading center 12 Ma ago (Schindlbeck et al., 2015) and the subsequent younger magmatic processes of the Cocos Island Seamount Province affected Hole 344-U1414A. While approaching the Middle America Trench cooling was  (Brandstätter et al., 2016), and deformation conditions (Brandstätter et al., 2017).

Conclusion
The differences between the CCV-S and the CCV-B are most pronounced in their REY patterns, especially Y/Ho ratios, Mn, Fe, Sc, Na, Zr, Hf, Th, and U content, in their Mg/Ca ratios, oxygen isotopic composition and in their 87 Sr/ 86 Sr ratios. These differences are mainly attributed to different fluid sources and to the corresponding host rock and the interaction with the fluid. The isotopic and elemental composition of aragonite and fibrous calcite veins in the CCR basalt indicate that invading seawater from restricted marine setting was the fluid source and/or precipitation occurred under restricted conditions. The chemical composition of the CCV-B in highly altered basalt corresponds rather to the geochemistry of the basaltic host rock and to a hydrothermal fluid, than to seawater. The high Y/Ho ratios and positive La and Gd anomalies suggest that CCV-S were precipitated from seawater of more likely open marine setting or from a fluid chemical changed by fluid-host rock interaction. The positive Eu anomaly indicates that all CCV were hydrothermally influenced and the strontium and oxygen isotopic composition of the CCV-S and of the composite CCV-B samples imply that seawater was modified into a hot, hydrothermal fluid. Fibrous calcite veins in the basalt were precipitated from invaded seawater near the ridge axis. These veins in the CCR basalt, as well as lithified sedimentary rocks, were penetrated by a hot and chemically modified fluid, resulting in vein formation in the sedimentary rocks and quartz and calcite precipitation in the CCV-B. The lack of high-temperature alteration in the basalt could be attributed to a high fluid flow rate, evidenced by hydrofracturing in the sedimentary rocks of Unit III. The geochemical results combined with the age data of Unit III and the models for formation of the Cocos Plate Seamounts after Herbrich et al. (2015) and Schindlbeck, Kutterolf, Freundt, Andrews, et al. (2016) constrain intraplate seamount volcanism in the area between the Galapagos hot spot and the Cocos Island, probably in the Plio-Pleistocene, as the source for subsequent heating after CCV formation.