Description: The Earth's mantle is chemically and isotopically heterogeneous, and a component of recycled oceanic crust is generally suspected in the convecting mantle [Hofmann and White, 1982. Mantle plumes from ancient oceanic crust. Earth Planet. Sci. Lett. 57, 421–436]. Indeed, the HIMU component (high µ = 238U/204Pb), one of four isotopically distinct end-members in the Earth's mantle, is generally attributed to relatively old (≥ 1–2 Ga) recycled oceanic crust in the form of eclogite/pyroxenite, e.g. [Zindler and Hart, 1986. Chemical geodynamics. Ann. Rev. Earth Planet. Sci. 14, 493–571]. Although the presence of the recycled component is generally supported by element and isotopic data, little is known about its physical state at mantle depths. Here we show that the concentrations of Ni, Mn and Ca in olivine from the Canarian shield stage lavas, which can be used to assess the physical nature of the source material (peridotite versus olivine-free pyroxenite) [Sobolev et al., 2007. The amount of recycled crust in sources of mantle-derived melts. Science 316, 412–417], correlate strongly with bulk rock Sr, Nd and Pb isotopic ratios. The most important result following from our data is that the enriched, HIMU-type (having higher 206Pb/204Pb than generally found in the other mantle end-members) signature of the Canarian hotspot magmas was not caused by a pyroxenite/eclogite constituent of the plume but appears to have been primarily hosted by peridotite. This implies that the old (older than ~ 1 Ga) ocean crust, which has more evolved radiogenic isotope compositions, was stirred into/reacted with the mantle so that there is not significant eclogite left, whereas younger recycled oceanic crust with depleted MORB isotopic signature (〈 1 Ga) can be preserved as eclogite, which when melted can generate reaction pyroxenite.

Description: Late Cretaceous (66.2 ± 0.5 Ma amphibole and 66.7 ± 0.2 Ma phlogopite 40Ar/39Ar ages) nephelinitic volcanic rocks from Godzilla Seamount in the eastern North Atlantic (34°N latitude) have trace element and Sr–Nd–Pb–Hf-isotope compositions similar to the Enriched Mantle I (EM-I) endmember, except for their low 207Pb/204Pb relative to 206Pb/204Pb ratios (206Pb/204Pbin = 17.7, 207Pb/204Pbin = 15.34) plotting below the Northern Hemisphere Reference Line on the uranogenic Pb isotope diagram. O isotope data on amphibole separates are mantle-like (δ18O = 5.6–5.8‰). Age and location of the isolated Godzilla Seamount, however, preclude it from being derived from the Madeira or Canary hotspots, making a lower-mantle origin unlikely. Therefore we propose derivation from a shallow (lithospheric/asthenospheric) melting anomaly. As observed in mid-ocean-ridge and ocean-island basalts, there is a systematic decrease of 207Pb/204Pb ratios (and Δ7/4) in the individual EM-I endmember type localities towards northern latitudes with Godzilla lying on the extension of this trend. This trend is mirrored in ultra-potassic volcanic rocks such as lamproites and kimberlites, which reflect the composition of enriched subcontinental lithospheric mantle. Therefore, a global pattern in 207Pb/204Pb ratios and Δ7/4 is suggested. The geochemical composition of EM-I endmember type localities, including Godzilla lavas, and the enriched (DUPAL) anomaly in the southern hemisphere could reflect derivation from ancient, metasomatized subcontinental lithospheric mantle. We propose a two-stage model to explain the trace element and isotopic composition of the EM-I mantle endmember localities worldwide: 1) during the early history of the Earth, subcontinental lithosphere was metasomatized by melts from subducted slabs along convergent margins generating high μ (238U/204Pb) sources, and 2) as the Earth cooled, hydrous fluids replaced hydrous melts as the main slab component metasomatizing the subcontinental lithospheric mantle (generating EM-I sources with lower μ). In accordance with this model, the global variations in 207Pb/204Pb ratios and Δ7/4 could reflect geographic differences in μ and/or the age at which the transition from stages 1 to 2 took place in the Archaean lithosphere. The model would require a re-definition of the EM-I endmember to low 206Pb/204Pb, high 208Pb/204Pb (positive Δ8/4) but variable 207Pb/204Pb (positive and negative Δ7/4).

Description: Young (≤3Ma) lavas from volcanic centers along a 75-km N–S transect from northern Madeira Island to a submarine volcanic field 55km south of Madeira exhibit distinct spatial geochemical variations. From south to north along this transect, there is a general decrease in Fe(8) (total Fe as FeO normalized to MgO=8wt.%), K(8), Pb isotope ratios and Zr/Hf and an increase in Al(8) and Nd isotope ratios, reflecting an increasing contribution of depleted source components to the north. The south to north spatial variations mimic the ca. 6million year temporal geochemical evolution on Madeira with the isotopic composition of the volcanic rocks becoming progressively more depleted through time. The temporal and spatial variation in chemical composition of lavas is consistent with melt extraction from a heterogeneous blob of upwelling mantle consisting of enriched garnet pyroxenite/eclogite material (recycled hydrothermally altered oceanic crust) in a depleted ultramafic/peridotitic matrix (recycled lower oceanic crust and/or lithospheric mantle). The geochemical transect provides new insights into the spatial heterogeneity of the Madeira plume pulse and confirms models for the temporal evolution of ocean island volcanoes.

Description: Boron and Pb isotopic compositions together with B–U–Th–Pb concentrations were determined for Pacific and Indian mantle-type mid-ocean ridge basalts (MORB) obtained from shallow drill holes near the Australian Antarctic Discordance (AAD). Boron contents in the altered samples range from 29.7 to 69.6 ppm and are extremely enriched relative to fresh MORB glass with 0.4–0.6 ppm B. Similarly the δ11B values range from 5.5‰ to 15.9‰ in the altered basalts and require interaction with a δ11B enriched fluid similar to seawater ∼ 39.5‰ and/or boron isotope fractionation during the formation of secondary clays. Positive correlations between B concentrations and other chemical indices of alteration such as H2O CO2, K2O, P2O5, U and 87Sr/86Sr indicate that B is progressively enriched in the basalts as they become more altered. Interestingly, δ11B shows the largest isotopic shift to + 16‰ in the least altered basalts, followed by a continual decrease to + 5–6‰ in the most altered basalts. These observations may indicate a change from an early seawater dominated fluid towards a sediment-dominated fluid as a result of an increase in sediment cover with increasing age of the seafloor. The progression from heavy δ11B towards lighter values with increasing degrees of alteration may also reflect increased formation of clay minerals (e.g., saponite). A comparison of 238U/204Pb and 206Pb/204Pb in fresh glass and variably altered basalt from Site 1160B shows extreme variations that are caused by secondary U enrichment during low temperature alteration. Modeling of the U–Pb isotope system confirms that some alteration events occurred early in the 21.5 m.y. history of these rocks, even though a significant second pulse of alteration happened at ∼ 12 Ma after formation of the crust. The U–Pb systematics of co-genetic basaltic glass and variably low temperature altered basaltic whole rocks are thus a potential tool to place age constraints on the timing of alteration and fluid flow in the ocean crust.

Description: We report new data on the major and trace element composition of melt inclusions in spinel phenocrysts (Mg# = 0.7-0.8, Cr/(Cr + Al) = 0.32-0.52, TiO2 = 0.06-0.60 wt.%) from Cretaceous MORB-like basalt (La/Yb = 0.94, Th/Nb = 0.055, Th/La = 0.041) in the Kamchatsky Mys ophiolites (Eastern Kamchatka). The melt inclusions preserved primitive melts (Mg# up to 0.72), which are remarkably depleted in incompatible trace elements compared to common MORBs. Numerous ultra-depleted inclusions from the studied sample have extraordinarily low Na2O (0.20-0.67 wt.%), TiO2 (0.16-0.5 wt.%), K (1.5-25 ppm), La (0.015-0.040 ppm), Zr (0.9-2 ppm), B (0.01-0.03 ppm), Ti/Zr = 300-1074, La/Yb = 0.008-0.075 and represent the most depleted melts known until now. The ultra-depleted melts from the Kamchatkan ophiolites are only comparable to a single melt inclusion from MORB of 9 degrees N Mid-Atlantic Ridge [Sobolev and Shimizu, Nature 363 (1993) 151-154] yet have higher FeO, CaO, heavy rare-earth element (Dy, Er, Yb) contents and lower Na2O and SiO2. These melts, possibly the last melt fractions produced in an upwelling mantle column, could represent the highest degrees (up to similar to 20%) of near-fractional melting of mantle with T-p >= 1400 degrees C, which started melting at similar to 75 km depth and continued to shallow depths of similar to 20 km. The presence of melts ranging in composition from ultra-depleted to compositions similar to Mauna Loa Volcano, Hawaii, high potential mantle temperature and association with rocks akin the Cretaceous Hawaiian tholeiites suggest that the trace element depleted melts preserved in spinel phenocrysts could have originated from extensive melting of a depleted component intrinsic to the Hawaiian plume or ambient upper mantle entrained and heated up at the plume margins. (C) 2009 Elsevier B.V. All rights reserved.

Description: New 238U–230Th–226Ra and 231Pa–235U disequilibria data measured by TIMS are presented for ridge-centered MORB glasses dredged during the R/V Sonne 158 cruise at the Galápagos or Cocos-Nazca Spreading Center (GSC) between 86.0°W and 92.3°W. The application of U-series isotopes to the GSC region, situated a few hundred kilometres to the north of the Galápagos hotspot, allows assessment of fundamental questions related to the dynamics of plume–ridge interaction. These include (1) the relationship between long-lived source variations, U-series disequilibria and extent of differentiation, (2) partial melting during solid upwelling, and (3) the nature and rates of plume–ridge mass transfer. The along axis U-series disequilibria variation show gradational patterns that locally are correlated with geochemical and isotopic parameters such as La/Sm, Tb/Yb, 206Pb/204Pb and 143Nd/144Nd as well as major element compositions. The correlation of (230Th)/(238U) with radiogenic isotopes and Tb/Yb provides constraints on the plume source influence on the melting process, reflecting an increase in the amount of melting at depth in the presence of garnet or aluminous clinopyroxene. Moreover, the correlation between U-series signatures, radiogenic isotopes, incompatible element abundance and MgO content indicates a causative relationship between the melting of plume source materials and how these lavas differentiate at shallow depths. We speculate that this involves loss of alkalis from ascending melts to shallow peridotite and crustal gabbro, resulting in increased olivine fractionation from the magmas. The U-series data place stringent constraints on the timing of plume–ridge mass transfer and thus distinguish whether mass transfer occurs by movement of melts or solid mantle. Within the likely conditions proposed by the model of (Braun and Sohn [EPSL 213 (2003): 417–430] and with knowledge of (231Pa)/(235U) and (230Th)/(238U) observed in Galápagos Islands lavas [A. Saal, personal communication], we show that all 226Ra excess will be lost and the initial 231Pa and 230Th excesses will be largely decayed. Therefore, we conclude that the plume influence on the GSC lavas results from a solid mantle flow process instead of through migration of plume-derived melts to the ridge.

Description: The shift of lava geochemistry between volcanic front to rear-arc volcanoes in active subduction zones is a widespread phenomenon. It is somehow linked to an increase of the slab surface depth of the subducting oceanic lithosphere and increasing thickness of the mantle wedge and new constraints for its causes may improve our understanding of magma generation and element recycling in subduction zones in general. As a case study, this paper focuses on the geochemical composition of lavas from two adjacent volcanic centres from the volcanic front (VF) to rear-arc (RA) transition of the Southern Kamchatkan subduction zone, with the aim to examine whether the shift in lava geochemistry is associated with processes in the mantle wedge or in the subducted oceanic lithosphere or both. The trace element and O–Sr–Nd–Hf–Pb (double-spike)-isotopic composition of the mafic Mutnovsky (VF) and Gorely (RA) lavas in conjunction with geochemical modelling provides constraints for the degree of partial melting in the mantle wedge and the nature of their slab components. Degrees of partial melting are inferred to be significantly higher beneath Mutnovsky (∼18%) than Gorely (∼10%). The Mutnovsky (VF) slab component is dominated by hydrous fluids, derived from subducted sediments and altered oceanic crust, eventually containing minor but variable amounts of sediment melts. The composition of the Gorely slab component strongly points to a hydrous silicate melt, most likely mainly stemming from subducted sediments, although additional fluid-contribution from the underlying altered oceanic crust (AOC) is likely. Moreover, the Hf–Nd-isotope data combined with geochemical modelling suggest progressive break-down of accessory zircon in the melting metasediments. Therefore, the drastic VF to RA shift in basalt chemistry mainly arises from the transition of the nature of the slab component (from hydrous fluid to melt) in conjunction with decreasing degrees of partial melting within ∼15 km across-arc. Finally, systematic variations of key inter-element with high-precision Pb-isotope ratios provide geochemical evidence for a pollution of the Mutnovsky mantle source with Gorely melt components but not vice versa, most likely resulting from trench-ward mantle wedge corner flow. We also present a geodynamic model integrating the location of the Mutnovsky and Gorely volcanic centres and their lava geochemistry with the recently proposed thermal structure of the southern Kamchatkan arc and constraints about phase equilibria in subducted sediments and AOC. Herein, the slab surface hosting the subducted sediments suffers a transition from dehydration to melting above a continuously dehydrating layer of AOC. Wider implications of this study are that an onset of (flush-) sediment melting may ultimately be the main trigger for the VF to RA transition of lava geochemistry in subduction zones.

Description: The Canary plume is one of the few plumes that can be traced to the core–mantle boundary, making it an excellent location for studying mantle melting dynamics. We performed a comprehensive study of the subaerial shield stage lavas erupted on Gran Canaria, Tenerife, La Gomera and La Palma because these rocks are believed to have been formed by the greatest degree of partial melting of the source and thus can provide direct insights into the origin of plume material and deep crustal recycling. The most primitive picritic to alkali basaltic and basanitic shield stage lavas are moderately enriched in light rare earth elements ([La/Sm]n = 2.1–4.6), strongly enriched in Nb and Ta ([Nb/La]n = 0.8–1.5, [Ta/La]n = 1.3–1.8) and depleted in K and Pb ([K/La]n = 0.2–0.7; [Pb/La]n = 0.2–0.3), resembling HIMU-type magmas. Fractional crystallization and phenocryst accumulation are processes affecting the compositions of parental magmas. The Sr–Nd–Pb isotope data (87Sr/86Sr = 0.702966–0.703312, 143Nd/144Nd = 0.512884–0.512929, 206Pb/204Pb = 19.49–20.27, 207Pb/204Pb = 15.59–15.66, 208Pb/204Pb = 39.21–39.81) and the relationship between Sr isotopes in whole rocks and O isotopes in olivine phenocrysts (δ18O = 4.3–5.8 ± 0.3‰) argue against extensive magma contamination at crustal depths. Varying, but relatively low, degrees of partial melting (1–10%) of a peridotite source containing 2–20% garnet can explain the Zr/Y range between 6.7 and 10.3 but require unrealistically high amount of garnet (up to 30%) to account for higher (up to 11.8) Zr/Y ratios in the lavas from Gran Canaria. This observation, as well as systematically lower CaO/Al2O3 ratios, higher SiO2 and NiO contents at a given MgO in the Gran Canaria lavas, as compared with those from other Canary Islands, imply melting of a plume probably containing garnet-bearing recycled component in a form of eclogite. The presence of amphibole and/or phlogopite is essential to produce variations in Ba/Nb (3.6–7.7), Ba/Th (51–108), K/La (65–295) and Nb/U (42–85) ratios in the shield stage magmas. We propose a scenario in which the magmas are derived through partial melting of a plume probably containing a recycled component with “ghost” amphibole and/or phlogopite signature; fluid-mobile elements compatible in amphibole and phlogopite could be preferentially lost to the subarc mantle during subduction-related breakdown of these phases. The relationships between He (3He/4He = 5.7–9.3 RA), Nd, Sr and Pb isotope ratios support the presence of a recycled component in the source of the Canary shield stage magmas. The observed relatively low 40Ar/36Ar isotopic ratios between ∼ 310 and ∼ 450 indicate contamination of plume-derived magmas by atmospheric air.

Description: New and published data on the composition of melt inclusions in olivine (Fo73–91) from volcanoes of the Kamchatka and northern Kurile Arc are used 1) to evaluate the combined systematics of volatiles (H2O, S, Cl, F) and incompatible trace elements in their parental magmas and mantle sources, 2) to constrain thermal conditions of mantle melting, and 3) to estimate the composition of slab-derived components. We demonstrate that typical Kamchatkan arc-type magmas originate through 5–14% melting of sources similar or slightly more depleted in HFSE (with up to ∼ 1 wt.% previous melt extraction) compared to MORB-source mantle, but strongly enriched in H2O, B, Be, Li, Cl, F, LILE, LREE, Th and U. Mean H2O in parental melts (1.8–2.6 wt.%) decreases with increasing depth to the subducting slab and correlates negatively with both ‘fluid-immobile’ (e.g. Ti, Na, LREE) and most ‘fluid-mobile’ (e.g. LILE, S, Cl, F) incompatible elements, implying that solubility in hydrous fluids or amount of water does not directly control the abundance of ‘fluid-mobile’ incompatible elements. Strong correlation is observed between H2O/Ce and B/Zr (or B/LREE) ratios. Both, calculated H2O in mantle sources (0.1–0.4%) and degrees of melting (5–14%) decrease with increasing depth to the slab indicating that the ultimate source of water in the sub-arc mantle is the subducting oceanic plate and that water flux (together with mantle temperature) governs the extent of mantle melting beneath Kamchatka. A parameterized hydrous melting model [Katz et al. 2003, G3, 4(9), 1073] is utilized to estimate that mantle melting beneath Kamchatka occurs at or below the dry peridotite solidus (1245–1330 °C at 1.5–2.0 GPa). Relatively high mantle temperatures (yet lower than beneath back-arc basins and ocean ridges) suggest substantial corner flow driven mantle upwelling beneath Kamchatka in agreement with numerical models implying non-isoviscous mantle wedge rheology. Data from Kamchatka, Mexico and Central America indicate that 〈 5% melting would take place beneath continental arcs without water flux from the subducting slab. A broad negative correlation appears to exist between crustal thickness and the temperature of magma generation beneath volcanic arcs with larger amounts of decompression melting occurring beneath thinner arc crust (lithosphere). In agreement with the high mantle temperatures, we observe a systematic change in the composition of slab components with increasing slab depth from solute-poor hydrous fluid beneath the volcanic front to solute-rich hydrous melt or supercritical liquid at deeper depths beneath the rear arc. The solute-rich slab component dominates the budget of LILE, LREE, Th and U in the magmas and originates through wet-melting of subducted sediments and/or altered oceanic crust at ≥ 120 km depth. Melting of the upper parts of subducting plates under water flux from deeper lithosphere (e.g. serpentinites), combined with high temperatures in the mantle wedge, may be a more common process beneath volcanic arcs than has been previously recognized.

Description: We measured oxygen isotope compositions of 34 adakites, high-Mg andesites, and lavas suspected to contain abundant slab and sediment melts from the Western and Central Aleutians, the Andes, Panama, Fiji, Kamchatka, Setouchi (Japan), and the Cascades. This suite covers much of the diversity of arc lavas previously hypothesized to contain abundant ‘slab’ melts. Measured and calculated values of δ18O for olivine phenocrysts in these samples vary between 4.88‰ and 6.78‰, corresponding to calculated melt values of 6.36‰ to 8.17‰. Values of δ18O for these samples are correlated with other geochemical parameters having petrogenetic significance, including Sr/Y, La/Yb, 87Sr/86Sr, and 143Nd/144Nd. Archetypical adakites from Adak Island (Central Aleutian) and Cook Island (Andean Austral zone), previously interpreted to be nearly pure melts of basaltic and gabbroic rocks in subducting slabs, have values of δ18O slightly higher than those of normal mid-ocean-ridge basalts, and in oxygen isotope equilibrium with typical mantle peridotite (i.e., their subtle 18O enrichment reflects their Si-rich compositions and low liquidus temperatures, not 18O-rich sources). Other primitive adakites from Panama and Fiji show only subtle sub-per mil enrichments in the source. This finding appears to rule out the hypothesis that end-member adakites are unmodified partial melts of basaltic rocks and/or sediments in the top (upper 1–2 km) of the subducted slab, which typically have δ18O values of ca. 9–20‰, and also appears to rule out them being partial melts of hydrothermally altered gabbros from the slab interior, which typically have δ18O values of ca. 2–5‰. One explanation of this result is that adakites are mixtures of partial melts from several different parts of the slab, so that higher- and lower-δ18O components average out to have no net difference from average mantle. Alternatively, adakites might be initially generated with more extreme δ18O values, but undergo isotopic exchange with the mantle wedge before eruption. Finally, adakites might not be slab melts at all, and instead come from differentation and/or partial melting processes near the base of the arc crust in the over-riding plate. High-Mg andesites and Setouchi lavas are commonly higher in δ18O than equilibrium with the mantle, consistent with their containing variable amounts of partial melts of subducted sediments (as we conclude for Setouchi lavas), slab-derived aqueous fluid (as we conclude for the Cascades) and/or crustal contaminants from the over-riding plate (as we conclude for Kamchatka).