There are three methods of investigating the oxygen-isotope composition of oceanic crust: (1) data from drilling-obtained samples; (2) study of obducted ophiolites; and (3) from crustal-derived xenoliths; however, each approach has limitations. Drilling capabilities are limited to the upper 1–3 km of the oceanic crust. Ophiolites are segments of oceanic crust and underlying mantle exposed at the Earth's surface, but their origin and emplacement are still debated. Nonetheless, the study of ophiolite sequences has enabled many of the fundamental observations regarding ancient-oceanic crust and the oxygen-isotope signature of the upper mantle. Analyses of ophiolite sequences (e.g., Samail) reveal δ18O variations ranging from isotopically enriched in relation to MORB (δ18O > MORB) to isotopically depleted (low δ18O ≤ MORB), interpreted to show the effects of seawater interaction at different temperatures. Mantle-derived xenoliths also display compelling geochemical evidence suggesting they are remnants of subducted ocean crust, and may in fact, be more representative samples. Some kimberlite-hosted diamondiferous eclogite xenoliths, with crustal protoliths, reveal a marked oxygen-isotope disparity between oceanic crust and the upper mantle; i.e., a bimodal distribution of both mantle-like (δ18O ≈ 5.6‰) and isotopically enriched oxygen (δ18O ≈ 6.8‰). The isotopically high δ18O values are often observed in eclogite xenoliths with ultramafic protoliths, which form beyond the low-temperature alteration zone. This finding is in contrast to traditional oxygen-isotope versus depth-curves of ophiolite sequences. Here, we present a compiled data set from Siberian eclogites and pyroxenites that illustrates this bimodality, using magnesium number (Mg#) as a depth proxy, and investigate the conditions that could potentially result in the observed “abnormal” oxygen-isotope variations.
 The current understanding of the origin and evolution of ancient oceanic crust is predominantly based upon geochemical and petrological studies of obducted ophiolite sequences, traditionally thought to represent oceanic crust, generated at spreading ridges [Gass, 1968; Dewey, 1976; Coleman, 1977]. The term “ophiolite” describes an assemblage of mafic and ultramafic rocks arranged in the following sequence: a basal ultramafic complex, consisting of harzburgite, lherzolite, and dunite in variable proportions, often with metamorphic overprinting; above this, an ultramafic-mafic gabbroic complex that grades from cumulate peridotites and pyroxenites to more typical gabbroic varieties; followed by a mafic sheeted-dyke complex and mafic volcanic complex, consisting of basaltic pillow lavas with various sodic felsic intrusive and extrusive structures (Figure 1) [Coleman, 1977]. The structure of the ophiolite sequence suggests that oceanic crust has formed in an extensional environment similar to modern fast-spreading mid-ocean ridges (MOR). Typically, this oceanic crust is recycled back into the mantle through the process of subduction; however, on some rare occasions, fragments of this MOR lithosphere are obducted onto continental margins [Dilek and Furnes, 2011].
 If we assume a similar formation history for all ophiolites, they can potentially yield important information about MOR processes, including the geochemical (e.g., oxygen isotope) evolution of such systems. However, many researchers have highlighted evidence for a complex tectonic environment during formation of oceanic crust suggesting that ophiolite structure, thickness, and geochemical characteristics may vary significantly as a function of spreading rate and tectonic setting [Pearce et al., 1984; MacLeod and Rothery, 1992; Searle and Cox, 1999; MacLeod and Yaouancq, 2000; Nicolas et al., 2000; Robertson, 2002; Harper, 2003; Metcalf and Shervais, 2008; Dilek and Furnes, 2011]. Two main groups of ophiolites have been defined: (1) subduction-related ophiolites, which include suprasubduction zone (SSZ-type) and volcanic-arc-type ophiolites and (2) subduction-unrelated ophiolites, which include continental-margin-type, MOR-type, and plume-type ophiolites, all of which generally show MORB-like characteristics [Dilek and Furnes, 2011].
 The Samail ophiolite is one of the best exposures in the world, with crustal thicknesses similar to those of oceanic crust generated at modern fast-spreading ridges [Coleman, 1977; Gregory and Taylor, 1981; Dilek et al., 1998; Shervais, 2001; Dilek and Newcomb, 2003]. As a result, Samail has been extensively studied and is considered the archetypal ophiolite locality. Several authors [Shervais, 2001; Metcalf and Shervais, 2008; Dilek and Furnes, 2009, 2011] suggest that Samail is a SSZ-type ophiolite, resulting from oceanic-crust generation in a subduction rollback regime. As extension occurs, SSZ ophiolites form in nascent back-arc to fore-arc environments, as well as in oceanic and continental back-arc basins [Pearce et al., 1984; Stern and Bloomer, 1992; Shervais, 2001; Metcalf and Shervais, 2008; Dilek et al., 2007; Dilek and Polat, 2008; Dilek and Furnes, 2011]. Consequently, the extending lithosphere is fed by melts from the asthenosphere below [Stern and Bloomer, 1992; Shervais, 2001]. An important consequence of the SSZ model of formation for ophiolites is that they clearly have a complicated formation history, and their geochemical signatures are likely not representative of typical oceanic crust. Consequently, geochemical studies [e.g., Gregory and Taylor, 1981] of ophiolite suites may not provide the most representative information about the composition of typical oceanic crust.
 Alternatively, eclogitic xenoliths in kimberlites may be considered more representative of ancient oceanic crust; therefore, they provide more reliable records of geochemical systematics (e.g., oxygen isotopes) than obducted ophiolites, which have undergone a complicated tectonic history. Although this approach is somewhat less direct, it can potentially reveal useful information about ancient oceanic crust and the upper mantle. Eclogites result from prograde metamorphism of subducted altered oceanic crust and are thought to retain isotopic characteristics of the subducted oceanic crust. Periodically, xenoliths of eclogite are transported back to the surface through kimberlitic volcanism [Helmstaedt and Doig, 1975; Ongley et al., 1987; Shervais et al., 1988; Taylor and Neal, 1989; Neal et al., 1990; Jacob et al., 1994; Snyder et al., 1997; Jacob, 2004]. The recognition of a crustal protolith for eclogites began with studies of carbon-isotopes in diamonds within xenoliths [Sobolev, 1977; Deines, 1980; Javoy et al., 1986; Boyd et al., 1987; Deines et al., 1987; Sobolev et al., 1997, 2003; Cartigny, 2005]. Eclogitic diamonds possess a wide range of δ13C values (−44 to +5‰) that cannot be explained by a mantle origin. Similarly, the study of oxygen isotopes in eclogitic xenoliths [Garlick et al., 1971], also revealed δ18O values outside of the canonical mantle range (δ18O = 5.3 ± 0.6‰). Furthermore, major-element and trace-element constraints have been employed to classify eclogites into Groups A, B, and C, thus revealing information as to their source protolith [Shervais et al., 1988; Taylor and Neal, 1989; Neal et al., 1990; Taylor and Anand, 2004]. In their review paper, Taylor and Anand  highlighted the different chemical constraints that could be used to identify eclogite xenoliths as having crustal protoliths, such as the occurrence of nonmantle oxygen-isotope signatures, which were identified as an important characteristic, along with the presence of negative Eu-anomalies in clinopyroxene, which indicate accumulation of plagioclase. These findings were based upon numerous studies of kimberlite-hosted eclogite xenoliths from various cratonic settings, including South Africa and Siberia [Garlick et al., 1971; Shervais et al., 1988; Taylor and Neal, 1989; Caporuscio and Smyth, 1990; Neal et al., 1990; Jacob et al., 1994; Snyder et al., 1997; Valley et al., 1998; Schulze et al., 2000, 2001; Taylor et al., 2003c; Jacob, 2004; Spetsius et al., 2008; Jacob et al., 2009; Riches et al., 2010], most of which report δ18O values within the mantle range, but also display outlying δ18O values that plot both above and below 5.5‰ [Eiler et al., 2000; Eiler, 2001].
 In a similar way to crustal ophiolite sequences, eclogites that exhibit oxygen-isotope signatures outside of the mantle range have generally been attributed to presubduction, high-temperature and low-temperature alteration [Gregory and Taylor, 1981]. Using samples from a representative cross section of the Samail ophiolite, Gregory and Taylor  produced a generalized plot of oxygen isotope (δ18O) versus depth, as shown in Figure 1, which they interpret to be the result of fluid alteration at different temperatures causing oxygen-isotope enrichment (δ18O > MORB) at low-temperatures and oxygen-isotope depletion (δ18O < MORB) at high temperatures. Other ophiolite complexes across the globe have been observed to agree, in part, with the Gregory and Taylor  curve, but have also reported enriched oxygen-isotope values at depth [e.g., Schiffman et al., 1984; Lécuyer and Fourcade, 1991; Pearson et al., 1991; Harper, 2003]. These enriched signatures are again accredited to low-temperature alteration, but there is a lack of a definitive process for the introduction of low-temperature fluids to such depths within the ophiolite sections.
 In the present study, we systematically reevaluate eclogite and pyroxenite xenolith oxygen-isotope data from well-characterized Siberian kimberlite pipes, in order to better constrain the variation in oxygen-isotope values of oceanic crust. The primary aim of this work is to explain the greater variability in oxygen-isotope ratios with depth within oceanic-crust sequences, highlighting that greater δ18O variation occurs in eclogite and pyroxenite xenoliths than is observed in ophiolites, and that these variations are characteristic of fluid alteration of mafic to ultramafic rocks under low-temperature conditions immediately prior to slab subduction.
2. Data Source
 Eclogitic xenoliths erupted in kimberlites within the Siberian craton are considered to be crustally derived [Snyder et al., 1997, 1999; Sobolev et al., 2003; Taylor et al., 2003a; Jacob, 2004; Riches et al., 2010], meeting all geochemical constraints put forth by Taylor and Anand . As a result, Siberian eclogites provide excellent materials for the study of mantle and crustal oxygen. Here, we present compiled oxygen-isotope data (n = 20) from the Komsomolskaya kimberlite pipe, Siberia [Carmody et al., 2013] and data (n = 105) from previous oxygen-isotope studies of other Yakutian kimberlite-hosted eclogites [Taylor et al., 2003b; Spetsius et al., 2008; Riches et al., 2010]. All Komsomolskaya and Nyurbinskaya oxygen-isotope data are displayed in Figure 2, with histograms illustrating the distribution of garnet δ18O values in relation to the mantle range.
 Notably both kimberlite pipes in this study are diamond-bearing and located within the Daldyn-Alakit region of the Yakutian kimberlite province, with Komsomolskaya lying within the Alakit-Markhinsky field and Nyurbinskaya within the Nakynsky field (Figure 3). Nyurbinskaya is geologically well known for producing one of the largest arrays of diamondiferous xenoliths in the world; and for this reason, it has been extensively studied for oxygen isotopes, showing large δ18O variations (+4.43 to +9.65‰), and the highest δ18O values reported for eclogites [Spetsius et al., 2006, 2008]. Komsomolskaya is also diamondiferous, producing high-gem-quality stones but at lower grades.
3. Siberian Eclogite Results—Oxygen Isotope Versus Estimated Depth of Oceanic Crust
 It is widely accepted that MOR basalts undergo extensive fractional crystallization at low-to-moderate pressures, and evolve from primitive ultramafic melts to more evolved mafic melts [Sparks et al., 1980]. Early plate tectonic models for the formation of oceanic crust are based on the narrow compositional range of erupted basalts, evidence for magma mixing in MORB, and the ophiolite analogue for layered gabbros. These models argue for the existence of large magma chambers periodically refilled by the influx of new magma [e.g., Stakes et al., 1984]. Later geophysical studies of fast-spreading ridges negated that view by denying the existence of large magma bodies in the crust, and allowing only a small magma cupola above a larger partially molten mush zone [Reid et al., 1977; McClain et al., 1985].
 Recent models for ocean crust formation based on these geochemical and geophysical observations include the “gabbro glacier” model and its variants, with lower crust formation by the ductile flow of solid material both downward and outward from a single shallow-axial magma chamber [Quick and Denlinger, 1993; Boudier et al., 1996; Boudier and Nicolas, 2011]. Alternatively, others have proposed “sheeted sill” models, which invoke the “gabbro glacier” model only in the upper, foliated gabbros; the layered gabbros formed by multiple injections of magma form a series of stacked sills [Bedard, 1993; Kelemen et al., 1997; Korenaga and Kelemen, 1997].
 Studies of the plutonic sections of ophiolites have shown that the bulk of the section comprises layered gabbros, which exhibit only minor variations in composition due to fractionation (represented by Mg#s of ∼78–85; Mg# = Mg/Mg+Fe). This is overlain by a thinner section of foliated gabbros that are more evolved (Mg#s 74–82) than the layered gabbros, and finally by isotropic or varitextured gabbros with even lower Mg#s of 40–75 [Pallister and Hopson, 1981; MacLeod and Yaouancq, 2000; Nicolas et al., 2000; Nicolas and Boudier, 2003]. The uppermost volcanic and sheeted dike sections of the ophiolite represent magma compositions with Mg#s < 60.
 These differences in Mg# correspond to an approximate depth model for oceanic crust that comprises three broad depth zones that can be used to estimate stratigraphic depth within the oceanic crust. These three depth zones are: (1) a shallow zone of volcanics and sheeted dikes with Mg#s < 60; (2) an intermediate zone of noncumulate gabbros with Mg#s ∼60–75; and (3) a deep zone of layered gabbros with Mg#s ∼75–85. By assuming bulk-rock Mg# as an approximate depth proxy within these three broad zones, the eclogite samples of this study can then be placed into a stratigraphic cross section for comparison with the Samail ophiolite [Gregory and Taylor, 1981].
 In Figure 2, we note that two distinct oxygen-isotopic regimes can be distinguished in the Komsomolskaya eclogite data set; the first with an average δ18O value that corresponds to the range of values from garnet and zircons in equilibrium with mantle magmas (δ18O +5.3 ± 0.6‰) [Valley, 2003; Valley et al., 2003], and a second with 18O-enriched values (δ18O >7.0‰). Eclogitic xenoliths from Nyurbinskaya also show distinct isotopic regimes, i.e., mantle-like (δ18O +5.3 ± 0.6‰) and enriched δ18O ≈ 6.0‰. Figure 4 shows the same oxygen-isotope variations using Mg# as a proxy for depth in the section.
 The calculated whole-rock (WR) Mg# for the sections of Samail ophiolite are based on model compositions from the literature [Pallister and Gregory, 1983]. The oxygen-isotope value for clinopyroxene was calculated using the garnet oxygen-isotope value, and published fractionation factors (Δgrt−cpx) between garnet and clinopyroxene, assuming equilibrium [Ongley et al., 1987]. However, previous work has shown a minimal fractionation between garnet and clinopyroxene; thus, the garnet value is argued to approximate the whole rock [Valley et al., 1998; Schulze et al., 2001; Valley, 2003; Valley et al., 2003]. Studies have shown fractionation between clinopyroxene and garnet does occur, resulting in Δgrt−cpx from 0.15 to 0.76 [Boyd et al., 1987; Ongley et al., 1987; Deines and Haggerty, 2000], however, the higher value is likely to be from unequilibrated samples, as this magnitude of fractionation cannot be achieved at mantle temperatures. The clinopyroxene δ18O values were calculated using a more widely accepted fractionation factor of Δ ≈ 0.3 [Kohn and Valley, 1998; Valley, 2003] and are simply used to provide a comparison, not an absolute value. Whole-rock reconstructions were completed for both oxygen isotope and Mg# in order to directly compare these sample suites to the previous data of Gregory and Taylor . Previous studies on eclogite mineralogy indicate that the ratio of garnet to clinopyroxene can vary significantly, with garnet comprising between 30 and 90% of the rock and pyroxene between 8 and 70% [Sobolev et al., 1994]. We adopted an intermediate average assuming a modal mineralogy of 40% garnet and 60% clinopyroxene in order to reconstruct both Mg# and δ18O. The 1σ uncertainty on a given reconstructed Mg# is approximately ±0.002.
 According to the most commonly used isotopic cross section of Gregory and Taylor , measured δ18O values of the Oman ophiolite are primarily a function of hydrothermal alteration at different temperatures. Samples that retain near-MORB-like isotopic signatures represent the most pristine oceanic crust, having been subjected to little or no hydrothermal alteration. Some lateral variation in δ18O versus depth within the Samail ophiolite has also been attributed to the presence of late-stage plagio-granite intrusions [Stakes and Taylor, 1992]. Of the Komsomolskaya samples, one pyroxenitic sample displays a mantle-like oxygen-isotope signature; due to its high Mg#, we conclude that it is ultramafic in origin and is likely derived from the layered gabbro and peridotite portions of the oceanic crust. The remaining δ18O-enriched samples are interpreted to represent oceanic crust that has undergone hydrothermal alteration at low temperatures (i.e., <350°C), within the upper 1.5 km of the ophiolite sequence [Gregory and Taylor, 1981]. Whereas this may be true for eclogite samples thought to originate from a shallower stratigraphic level, the remaining δ18O-enriched samples are derived from depths well beyond the low-temperature alteration zone, and thus, contradict the classic Gregory and Taylor  interpretation. In Figure 4, both eclogites and pyroxenites from the noncumulate gabbroic portion of oceanic crust display δ18O-enriched signatures, however, when compared to Figure 1, samples at corresponding stratigraphic levels show δ18O-depleted signatures relative to MORB (i.e., δ18O < 5.6‰). Indeed, crustally derived Siberian eclogites, representing the ultramafic portion (lower one third) of oceanic crust, do not correspond to the oxygen-isotope model of Gregory and Taylor . The eclogites exhibit δ18O-enriched signatures, but are estimated to originate from lithologies significantly deeper than that of the low-temperature alteration zone.
4. Discussion—Oxygen-Isotope Variability in Oceanic Crust; Subducted Versus Obducted
 As a result of the pioneering study of Gregory and Taylor , the Samail oxygen profile has become the “standard” for oceanic crustal profiles. However, oxygen-isotope variations observed in Siberian eclogites suggest that oxygen isotope versus depth-profiles for ophiolite sequences may be more complex and that Samail-like profiles may not necessarily be applicable to all samples of oceanic crust. In order to accurately interpret oxygen-isotope signatures revealed from natural samples, it is important to understand the various different physical processes that control oceanic-crust production and modification. Obducted ophiolite sequences currently form the basis of our understanding of oceanic-crust structure and geochemistry. However, significant geochemical differences (e.g., oxygen-isotope variations) are evident between obducted-ophiolite sequences and eclogites. The following discussion, therefore, focuses upon the physical effects that can control oxygen-isotope ratios in ophiolites, with a particular emphasis on the observed disparity between ultramafic and mafic portions of subducted and obducted sequences.
4.1. Obducted SSZ-Type Ophiolites
 Previous oxygen-isotope data from the Samail ophiolite has revealed that the upper sections (pillow lavas, sheeted dykes, and upper-gabbroic cumulates) are 18O-enriched, with a range in δ18O values between +12.7‰ (at the top of the pillow basalts) and slightly depleted δ18O values of +4.9‰ within the gabbro cumulates [Gregory and Taylor, 1981]. The gabbroic cumulates and the ultramafic-mafic complex toward the middle of the sequence exhibit δ18O values varying between typical mantle whole-rock values (δ18O ≈ +5.5 ± 0.2‰) [Mattey et al., 1994; Valley, 2003] and δ18O-depleted values, between +3.7 and +4.9‰. Interestingly, toward the base of the ophiolite sequence, ultramafic harzburgites, lherzolites, and dunites are δ18O-depleted (δ18O = +4.0 to +6.0‰), falling slightly below the δ18O mantle values. However, further studies of the Samail ophiolite and associated plagio-granites and highly chloritized units [Stakes and Taylor, 1992] have shown δ18O-enriched signatures (δ18O >6.0‰) down to the Mohorovičić (MOHO) discontinuity.
 Similar Samail-like oxygen isotope versus depth-profiles are also observed in oceanic crust sampled during the Oceanic Drilling Program (ODP) and Deep Sea Drilling Program (DSDP), which have extracted cores from up to 3 km within the oceanic-crust sequence. The early drill-cores probed the upper 1 km of the crustal sequence, penetrating into the sheeted-dike complex and recorded enriched δ18O signatures for the extrusive pillow basalts (δ18O = +6.1 to +8.5‰), trending toward more MORB-like and δ18O-depleted values into the sheeted dikes (δ18O = +5.4 to −3.6‰) [Lawrence et al., 1975; Alt et al., 1986; Alt and Bach, 2006; Spivak et al., 2008]. Subsequent drilling expeditions extended drill cores to depths beyond the sheeted dikes and into the layered gabbros (∼3 km below the surface). At such depths, gabbros are δ18O-depleted, relative to unaltered basalts [Lecuyer and Reynard, 1996; Gao et al., 2006, 2012]. The ocean drilling programs have greatly increased the available ophiolite data set, providing access to large sections of the upper oceanic crust; however, current drilling capabilities are limited and available drill-cores do not penetrate into the ultramafic section of the ophiolite sequence, and thus, drill-core data still only provide us with a relatively limited understanding of oxygen isotopes in ophiolite sequences.
 Oxygen-isotope variations as a function of depth within the Samail ophiolite sequence demonstrate the influence of hydrothermal alteration at both low and high temperatures and suggest the presence of two circulatory systems: (1) circulation of low-temperature seawater in the upper oceanic crust and (2) circulation of high-temperature hydrothermal fluids, potentially of magmatic origin, in the lower oceanic crust [Gregory and Taylor, 1981; Schiffman et al., 1984; Alt et al., 1986; Lécuyer and Fourcade, 1991; Gao et al., 2006]. Hydrothermal alteration at low temperature (i.e., <350°C) has been experimentally shown to result in higher δ18O values in the rocks, whereas high temperature, (i.e., >350°C), alteration causes lower δ18O values [Urey, 1947; O'Neil et al., 1969]. The temperature dependence of oxygen-isotope fractionation is related to atomic mass; its effect on lattice vibrations, and consequently variations in equilibrium exchange constants, occur with temperature [Urey, 1947; Chacko et al., 2001]. In general, igneous rocks that are isotopically distinct from MORB must have interacted with material (e.g., seawater or sediments) near the Earth's surface [Taylor and Sheppard, 1986]. Using this principal, it is therefore, possible to trace the extent of low-temperature or high-temperature alteration with depth within oceanic-crust sequences and provide explanations for anomalous horizons.
 Fluid circulation and the mechanisms introducing fluids into ophiolite sequences are key points to address when discussing oceanic-crust alteration. To interpret the oxygen-isotope signatures exhibited by oceanic crust (ophiolites, eclogites, and oceanic-drill cores), it is important to consider the extent of fluid circulation, the temperature at which fluids interact with the rocks, and time scales over which these events occur, relative to formation of the crustal sequence. Hydrothermal alteration is a widely accepted mechanism for modifying the geochemistry and oxygen-isotope composition of the oceanic crust [Muehlenbachs and Clayton, 1972; Wolery and Sleep, 1976; Gregory and Taylor, 1981; Alt et al., 1986; Muehlenbachs, 1998; Alt and Teagle, 2003; Gao et al., 2006; Wilson et al., 2006; Gao et al., 2012]. However, the timing and extent of alteration processes is not well constrained. Alt et al.  argued that oceanic crust that formed at mid-oceans ridges is subject to two-fluid circulatory systems; an upper low temperature, seawater system, which affects the pillow-basalt portion of oceanic crust and a deeper, initially high temperature, hydrothermal circulatory system, which transports δ18O-enriched fluids upward through the oceanic crust. Notably, the origin and isotopic composition of the deeper hydrothermal fluid is not well constrained in Alt et al. . We propose that the deeper fluids are likely MORB-equilibrated fluids, based upon previous studies of fluids from volcanic regions; these report that near the volcanic edifices, fluids are equilibrated with the mantle, with regards to stable isotopes and noble gases, e.g., helium and carbon [Taylor, 1990; Poreda et al., 1992; Hilton et al., 1998; Barry et al., 2013]. The presence of two-fluid circulation systems, overprinting each other over time, is used by Alt et al.  to explain the “enriched-depleted-enriched” versus depth sequence observed within DSDP hole 504B; however, the authors provide no indication of the timing of the overprinting events.
 Obducted ophiolite sequences in Northern California have also been shown to contain δ18O-enriched horizons (+5.8 to +7.6‰) within the amphibolite facies and unmetamorphosed rocks of the plutonic member of the sequence [Schiffman et al., 1984]. The oxygen-isotope signature of this portion of the ophiolite is comparable to the δ18O values reported for samples from Komsomolskaya. Schiffman et al.  invoked a model of both discharging (upward flowing) and recharging (downward flowing) fluids as a mechanism to generate the observed variation in δ18O with depth. In this model, the discharging fluid is argued to be of hydrothermal/metamorphic origin (magmatic water) that could result in 18O-enrichment at depths, due to the interaction of water and rock at small water-rock ratios (W/R) of approximately 0.5 [Bowers and Taylor, 1985]. The recharging fluids, however, represent the downward flow of seawater, which are isotopically distinct from the discharging waters and depleted relative to Standard Mean Ocean Water (SMOW) following isotopic exchange in the upper crust. When the W/R is low (i.e., <0.5), this process is argued to result in δ18O-enrichment at higher thermal gradients or greater depths. The relative contribution of recharge and discharge fluid flow could thus be used to explain the heterogeneous oxygen isotopic signatures observed in ophiolites and in particular the δ18O-enriched ultramafic portions. The dynamics of the upper seawater cycle is relatively well constrained, with respect to seawater infiltration and interaction with ophiolites and oceanic drill cores; however, the deeper hydrothermal cycle is not nearly as well characterized. Thus, the exact nature of interaction between the two circulations cells is not well understood and has not been effectively modeled.
4.2. Recycled Crustal Eclogites of Siberia
 The eclogitic xenoliths of this study were transported to the surface via the eruption of the Komsomolskaya and Nyurbinskaya kimberlites. The geochemical characteristics of these eclogites indicate that they were originally derived from the lower, mafic/ultramafic portions of subducted oceanic crust and that these geochemical signatures are preserved during eclogite formation. The ultramafic portions of the subducted oceanic crust display a δ18O-enriched signature, despite being estimated to occur at depths below the low-temperature alteration zone. Ultimately the enriched oxygen-isotope signatures at depth are most likely linked to deep seawater circulation, but the conditions and mechanisms of this exchange are unclear.
 Fractures within the upper 2–3 km of oceanic crust have been detected during seismic surveys [White et al., 1984; Yang et al., 1996; Dunn and Toomey, 2001]. Ophiolite sequences from Northern California exhibit δ18O-enriched rocks, which are attributed to either; deep-channelized penetration (up to 2 km) of low-temperature seawater at slow-spreading centers [Lécuyer and Fourcade, 1991]; and/or competition between up-welling and down-welling fluids in the hydrothermal system [Schiffman et al., 1984]. The deep-channelized alteration, evident in the Silurian Trinity ophiolite, was hypothesized to relate to deeper and wider faulting patterns, typical of slow-spreading centers. Slow-spreading centers also result in thinner ophiolite sequences [Reid and Jackson, 1981; Dick et al., 2003], permitting seawater percolation to greater depths within the ophiolite sequence, thereby facilitating isotopic exchange between seawater and basal rocks, resulting in δ18O-enriched rocks.
 Although the portions of ophiolites studied from California are part of the mafic-ultramafic complex, they are still relatively shallow (∼3–5 km) in comparison to the proposed ultramafic origin of the δ18O-enriched pyroxenites (>5 km) from Komsomolskaya and Nyurbinskaya, Siberia, used in this study. It is, therefore, unlikely that deep channels in slow-spreading centers, caused only by cooling, could extend to depths of the base cumulate gabbro where the enriched Siberian samples are proposed to have formed. Alternatively, we propose that the δ18O-encriched signatures observed in subduction-derived eclogites and pyroxenites, attributed to apparent low-temperature isotopic exchange at depths (>5 km) within oceanic crust, occurs due to near trench hydration. The process of near trench hydration occurs as the ophiolite moves away from the spreading center and subduction commences. Tectonic-related normal faults are either introduced or reactivated along oceanic spreading fabric in the upper, brittle portion of the subducting slab, resulting in widespread normal faulting. Such subduction-related bending faults have been widely observed and documented, including along large portions of the Pacific Rim [Masson, 1991]. Plate bending faults have been identified at subduction zones through the use of multibeam bathymetry [Kobayashi et al., 1998; Ranero et al., 2005], seismic reflection profiling [Ranero et al., 2003; Henrys et al., 2006], and earthquake focal mechanisms [Jiao et al., 2000; Christova and Scholz, 2003; Obana et al., 2012] and are argued to be active across the entire incoming plate, creating a tectonic fabric that cuts through the crust, often penetrating deep into the mantle [Ranero et al., 2003; Rüpke et al., 2010]. This portion of the oceanic plate is therefore suggested to be pervasively hydrated, with the occurrence of deep-faults promoting the hydration of the cold crust and upper mantle [Rüpke et al., 2004; Faccenda et al., 2009; Rüpke et al., 2010]. Evidence for this can also be found from seismic refraction studies of subduction trenches, which indicate low seismic velocities in the crust and upper mantle due to the presence of fluids [Contreras-Reyes et al., 2007; Ivandic et al., 2008].
 Furthermore, heat-flow studies indicate increased hydrothermal circulation, with subduction trenches exhibiting significantly lower heat-flow values in comparison to the global mean [Grevemeyer et al., 2005]. Therefore, we propose that infiltration of seawater could potentially induce low-temperature oxygen-isotope exchange in the lower portions of the ophiolite, and thus, explain the isotopic variations observed in ultramafic eclogites. Notably, it is the process of subduction and associated near-trench hydration that results in the observed oxygen-isotope disparity between obducted-ophiolite sequences and subducted-oceanic crust, as sampled by eclogites and pyroxenites. The crucial difference between the two representations of oceanic crust is therefore the extent of faulting and deep hydration that has occurred. We note that cooling-related fractures have been shown to only penetrate to shallow depths (<3 km), and therefore, if this sequence cools and is obducted onto continental margin, the record of low-temperature alteration is confined to shallow lithologies. However, if the ophiolite sequence is subducted, deep fractures introduced or reactivated during the bending of the plate can potentially allow the infiltration of low-temperature fluids to much greater depths, modifying the ultramafic portions to higher δ18O.
5. A Three Stage Model for Oxygen-Isotope Evolution in Ophiolite Sequences
 After reviewing the processes that can affect the oxygen-isotope ratios of oceanic crust at depth, we present an updated and comprehensive model of fluid alteration to provide an explanation for the observed isotopic trends exhibited by eclogites; this puts emphasis on the presence of δ18O-enriched samples that occur below the low-temperature alteration zone. The proposed model (illustrated in Figure 5) divides the alteration processes into three stages; (1) on-axis alteration; (2) alteration occurring just off-axis (Proximal); (3) alteration completed far off-axis, prior to subduction (Distal); and can be viewed simply as a competition between two types of circulating fluids
5.1. Stage One—“On-Axis” Alteration
 During the eruption of oceanic crust, at the axial ridge of the spreading center, infiltration of seawater into the ophiolite sequence occurs. Within the first few 100 m of the surface [Wolery and Sleep, 1976], the infiltrating seawater, penetrating through cooling cracks, is at low-temperature (<350°C) resulting in pervasive δ18O-enrichment of the surrounding rocks relative to MORB (δ18O > 5.5 ± 0.2‰). However, at depth, the rocks and fluids are in isotopic equilibrium, both exhibiting a MORB-like oxygen-isotope ratio. At this stage, an oxygen isotope versus depth-profile can be represented by that of Figure 5a.
5.2. Stage Two—“Proximal” Off-Axis Alteration
 The second stage of alteration begins as the newly formed oceanic crust moves away from the spreading center, toward the oceanic plateau. The base of the oceanic crust is still influenced by the MORB-equilibrated fluids due to the proximity of the crust to the volcanic center; thus, the ultramafic rocks continue to exhibit a MORB-like oxygen-isotope ratio. However, further cooling of the upper crust increases the depth of fractures delivering seawater into the sheeted-dike portion of the sequence. As seawater permeates, it increases in temperature to >350°C, at which point, isotope exchange results in the depletion of the rocks to δ18O values lower than MORB. Therefore, the oxygen-isotope ratio versus depth profile is characterized by an upper-enriched zone, an intermediate-depleted zone and a MORB-like base (Figure 5b).
5.3. Stage Three—“Distal” Off-Axis Alteration
 The final stage is a low-temperature alteration (i.e., overprinting) throughout the entire ophiolite sequence, including the basal ultramafics, as the crust moves away from the spreading center and is no longer affected by upwelling, MORB-equilibrated, hydrothermal fluids, related to the magma chamber. As the crust approaches the subduction zone, the descent of the preceding crust causes flexure of the plate, which results in deep normal faulting to occur in the brittle upper portion of the slab, through which low-temperature seawater can interact with the ultramafic portions of the ophiolite sequence. The continued isotopic exchange with depth results in an overall enrichment of the oxygen isotope versus depth profile, shifting the oxygen-isotope ratios of all portions of the crust to higher δ18O values, but not completely erasing previous alteration patterns (Figure 5c).
 The study of ophiolite sequences (e.g., Samail) and eclogitic xenoliths from kimberlite pipes with oceanic crustal protolith prior to subduction provides information regarding the petrology and geochemistry of ancient oceanic crusts. Here, we have demonstrated that oxygen-isotopes characteristic of the various stratigraphic regimes in a reconstructed ophiolite can vary considerably. Consequently, the utilization of obducted ophiolite sequences, as a standard for typical oceanic crust, may cause significant differences in the interpretation and assumptions we make about upper mantle geochemistry and diamond petrogenesis. Fresh oceanic basalts have been used extensively to accurately define the average mantle δ18O value. Deviations away from this value have been attributed to low-temperature or high-temperature hydrothermal alteration by fluid circulation, when viewed in ophiolite sequences and within some mantle-eclogitic xenoliths. However, oxygen-isotope measurements of eclogitic garnets have generated data sets with at least a bimodal distribution of oxygen-isotope signatures, with one average centered about the mantle-mean (+5.6‰) and the other centered on a δ18O-enriched signature (δ18O ≈ 7.0‰). By evaluating the physical and chemical conditions that oceanic crust is subjected to during formation and subduction, we have highlighted ways in which enriched-oxygen-isotope signatures can be generated at depths far below the low-temperature alteration zone. Based upon a logical tectonic sequence, we have presented a model for the evolution of isotopic signatures within oceanic crust. The enriched oxygen-isotope signatures at depth are most likely linked to fluid circulation and low-temperature isotopic exchange, as a result of deep fractures caused by tectonic stresses during plate-flexures prior to subduction. The Samail ophiolite, therefore, is a good representation of oceanic material that has not undergone subduction; but, it cannot be used to explain all mantle-eclogitic material, since it has not undergone additional deep-fracture processing and near-trench hydration immediately prior to subduction. Application of this sequence to all oceanic-crust material, i.e., other ophiolites or eclogites, must therefore, be done with caution.
 Most of the samples from which data were derived in this paper originated from samples in the collections of Slava Spetsius, which were published previously. For his generosity, we are greatly indebted. John Valley provided a fine review of an earlier version of this paper and made several useful suggestions. Craig Grimes and an anonymous reviewer are also thanked for their in depth reviews and insightful suggestions. In addition, John Pernet-Fisher and Geoffrey Howarth (UT) also provided helpful reviews, with many useful comments. The major portion of the study presented in this paper was funded by NSF grant EAR-1144337 (LAT) and the Planetary Geosciences Institute at the University of Tennessee. We also acknowledge NSF, Division of Earth Sciences (EAR), for providing funding (EAR-1144559) to PHB.