(Figures S2a-c) Phase diagrams of the upper crust, the lower crust and the upper mantle based on averaged compositions of the oceanic crust and mantle

The age of the subducting Nazca Plate off Chile increases northwards from 0 Ma at the Chile Triple Junction (46°S) to 37 Ma at the latitude of Valparaíso (32°S). Age-related variations in the thermal state of the subducting plate impact on (a) the water influx to the subduction zone, as well as on (b) the volumes of water that are released under the continental forearc or, alternatively, carried beyond the arc. Southern Central Chile is an ideal setting to study this effect, because other factors for the subduction zone water budget appear constant. We determine the water influx by calculating the crustal water uptake and by modeling the upper mantle serpentinization at the outer rise of the Chile Trench. The water release under forearc and arc is determined by coupling FEM thermal models of the subducting plate with stability fields of water-releasing mineral reactions for upper and lower crust and hydrated mantle. Results show that both the influx of water stored in, and the outflux of water released from upper crust, lower crust and mantle vary drastically over segment boundaries. In particular, the oldest and coldest segments carry roughly twice as much water into the subduction zone as the youngest and hottest segments, but their release flux to the forearc is only about one fourth of the latter. This high variability over a subduction zone of < 1500 km length shows that it is insufficient to consider subduction zones as uniform entities in global estimates of subduction zone fluxes. This article is protected by copyright. All rights reserved.

Noble metal abundances in ODP Leg 115 basalts (Table 1)

A comparison of 50 basalts recovered at Sites 706, 707, 713, and 715 along the Reunion hotspot trace during Ocean Drilling Program Leg 115 in the Indian Ocean shows that seafloor alteration had little effect on noble metal concentrations (Au, Pd, Pt, Rh, Ru, and Ir), determined by inductively coupled plasma-mass spectrometry (ICP-MS), which generally tend to decrease with magma evolution. Their compatible-element behavior may be related to the precipitation of Ir-Os-based alloys, chromite, sulfides, and/or olivine and clinopyroxene in some combination. The simplest explanation indicates silicate control of concentrations during differentiation. Basalts from the different sites show varying degrees of alkalinity. Noble metal abundances tend to increase with decreasing basalt alkalinity (i.e., with increasing percentages of mantle melting), indicating that the metals behave as compatible elements during mantle melting. The retention of low-melting-point Au, Pd, and Rh in mantle sulfides, which mostly dissolve before significant proportions of Ir-Os-based alloys melt, explains increasing Pd/Ir ratios with decreasing alkalinity (increasing melting percentages) in oceanic basalts. High noble metal concentrations in Indian Ocean basalts (weighted averages of Au, Pd, Rh, Pt, Ru, and Ir in Leg 115 basalts are 3.2, 8.1, 0.31, 7.3, 0.22, and 0.11 ppb, respectively), compared with basalts from some other ocean basins, may reflect fundamental primary variations in upper- mantle noble metal abundances

(Table 1) Nd and Sr isotopic analyses of ODP Leg 127 basalts

Sr and Nd isotopic compositions are reported for basaltic rocks collected during ODP Leg 127 from the Yamato Basin, a rifted backarc basin in the Japan Sea. The basalts are classified into two groups in terms of Nd isotopic composition: the upper sills at Site 797 are characterized by higher 143Nd/144Nd ratios (0.513083-0.513158, epsilon-Nd = 8.68-10.14) and the basalts from Site 794 and the lower sills at Site 797 have lower 143Nd/144Nd ratios (0.512684-0.512862, epsilon-Nd = 0.90-4.37). All of the basalts show higher Sr isotopic compositions than those of the mantle array, which is attributed to seawater alteration. The basalts with lower Nd isotopic values ranging in age from 20.6 to 17.3 Ma have tapped an enriched subcontinental upper mantle (SCUM) with the minor involvement of a depleted asthenospheric mantle (AM). Subsequent change in composition through the physical replacement of SCUM by AM yielded the basalts of the upper sills of higher Nd isotopic compositions. This event within the upper mantle was associated with the breakup of the overlying lithosphere during the rifting of the Japan Sea backarc basin.

Rb, Sr, Sm and Nd concentrations and isotopic composition of basalts from DSDP Leg 74 (Table 1)

Basement intersected in DSDP holes 525A, 528 and 527 on the Walvis Ridge consists of submarine basalt flows and pillows with minor intercalated sediments. These holes are situated on the crest and mid and lower northwest flank of a NNW-SSE-trending ridge block which would have closely paralleled the paleo mid-ocean ridge (Rabinowitz and LaBrecque, 1979 doi:10.1029/JB084iB11p05973, Moore et al. (1983 doi:10.1130/0016-7606(1983)94<907:TWRTDS>2.0.CO;2). The basalts were erupted approximately 70 m.y. ago, an age equivalent to that of immediately adjacent oceanic crust in the Angola Basin and coraistent with formation at the paleo mid-ocean ridge (Moore et al., 1983). The basalt types vary from aphyric quartz tholeiites on the ridge crest to highly plagioclase phyric olivine tholeiites on the ridge flank. These show systematic differences in incompatible trace element and isotopic composition. Many element and isotope ratio pairs form systematic trends with the ridge crest basalts at one end and the highly phyric ridge flank basalts at the other. The low 143Nd/144Nd (0.51238), 206Pb/204Pb (17.54), 207Pb/204Pb (15.47), 208Pb/204Pb (38.14) and high 87Sr/86Sr (0.70512) ratios of the ridge crest basalts suggest derivation from an old Nd/Sm-, Rb/Sr- and Pb/U-enriched mantle source. This isotopic signature is similar to that of alkaline basalts on Tristan da Cunha but offset to significantly lower Nd and Pb isotopic ratios. The isotopic ratio trends may be extrapolated beyond the ridge flank basalts with higher 143Nd/144Nd (0.51270), 206Pb/204Pb (18.32), 207Pb/204Pb (15.52), 208Pb/204Pb (38.77) and lower 87Sr/86Sr (0.70417) ratios in the direction of increasingly Nd/Sm-, Rb/Sr- and Pb/U-depleted source compositions. These isotopic correlations are equally consistent with mixing of depleted and enriched end member melts or partial melting of an inhomogeneous, variably enriched mantle source. However, observed Zr-Ba-Nb-Y interelement relationships are inconsistent with any simple two-component model of magma mixing, as might result from the rise of a lower mantle plume through the upper mantle. Incompatible element and Pb isotopic systematics also preclude extensive involvement of depleted (N-type) MORB material or its mantle sources. In our preferred petrogenetic model the Walvis Ridge basalts were derived by partial melting of mantle similar to an enriched (E-type) MORB source which had become heterogeneous on a small scale due to the introduction of small-volume melts and metasomatic fluids.

Geochemistry of the oldest Atlantic oceanic crust

Controversy has surrounded the issue of whether mantle plume activity was responsible for Pangaean continental rifting and massive flood volcanism (resulting in the Central Atlantic Magmatic Province or CAMP, emplaced around 200 Ma) preceding the opening of the central Atlantic Ocean in the Early Mesozoic. Our new Sr-Nd-Pb isotopic and trace element data for the oldest basalts sampled from central Atlantic oceanic crust by deep-sea drilling show that oceanic crust generated from about 160 to 120 Ma displays clear isotopic and chemical signals of plume contamination (e.g., 87Sr/86Sr(i) = 0.7032-0.7036, epsilonNd(t) =+6.2 to +8.2, incompatible element patterns with positive Nb anomalies), but these signals are muted or absent in crust generated between 120 and 80 Ma, which resembles young Atlantic normal mid-ocean ridge basalt. The plume-affected pre-120 Ma Atlantic crustal basalts are isotopically similar to lavas from the Ontong Java Plateau, and may represent one isotopic end-member for CAMP basalts. The strongest plume signature is displayed near the center of CAMP magmatism but the hotspots presently located nearest this location in the mantle reference frame do not appear to be older than latest Cretaceous and are isotopically distinct from the oldest Atlantic crust. The evidence for widespread plume contamination of the nascent Atlantic upper mantle, combined with a lack of evidence for a long-lived volcanic chain associated with this plume, leads us to propose that the enriched signature of early Atlantic crust and possibly the eruption of the CAMP were caused by a relatively short-lived, but large volume plume feature that was not rooted at a mantle boundary layer. Such a phenomenon has been predicted by recent numerical models of mantle circulation.

Geochemistry and oxygen and iron isotope ratios of mafic rocks

Recycling of oceanic crust into the deep mantle via subduction is a widely accepted mechanism for creating compositional heterogeneity in the upper mantle and for explaining the distinct geochemistry of mantle plumes. The oxygen isotope ratios (d18O) of some ocean island basalts (OIB) span values both above and below that of unmetasomatised upper mantle (5.5 ± 0.4 per mil) and provide support for this hypothesis, as it is widely assumed that most variations in d18O are produced by near-surface low-temperature processes. Here we show a significant linear relationship between d18O and stable iron isotope ratios (d57Fe) in a suite of pristine eclogite xenoliths. The d18O values of both bulk samples and garnets range from values within error of normal mantle to significantly lighter values. The observed range and correlation between d18O and d57Fe is unlikely to be inherited from oceanic crust, as d57Fe values determined for samples of hydrothermally altered oceanic crust do not differ significantly from the mantle value and show no correlation with d18O. It is proposed that the correlated d57Fe and d18O variations in this particular eclogite suite are predominantly related to isotopic fractionation by disequilibrium partial melting although modification by melt percolation processes cannot be ruled out. Fractionation of Fe and O isotopes by removal of partial melt enriched in isotopically heavy Fe and O is supported by negative correlations between bulk sample d57Fe and Cr content and bulk sample and garnet d18O and Sc contents, as Cr and Sc are elements that become enriched in garnet- and pyroxene-bearing melt residues. Melt extraction could take place either during subduction, where the eclogites represent the residues of melted oceanic lithosphere, or could take place during long-term residence within the lithospheric mantle, in which case the protoliths of the eclogites could be of either crustal or mantle origin. This modification of both d57Fe and d18O by melting processes and specifically the production of low-d18O signatures in mafic rocks implies that some of the isotopically light d18O values observed in OIB and eclogite xenoliths may not necessarily reflect near-surface processes or components.

Helium and neon isotopes in oceanic crust of ODP Hole 118-735B

In an attempt to determine the helium and neon isotopic composition of the lower oceanic crust, we report new noble gas measurements on 11 million year old gabbros from Ocean Drilling Program site 735B in the Indian Ocean. The nine whole rock samples analyzed came from 20 to 500 m depth below the seafloor. Helium contents vary from 3.3*10**-10 to 2.5*10**-7 ccSTP/g by crushing and from 5.4*10**-8 to 2.4*10**-7 ccSTP/g by melting. 3He/4He ratios vary between 2.2 and 8.6 Ra by crushing and between 2.9 and 8.2 by melting. The highest R/Ra ratios are similar to the mean mid-ocean ridge basalt (MORB) ratio of 8+/-1. The lower values are attributed to radiogenic helium from in situ alüha-particle production during uranium and thorium decay. Neon isotopic ratios are similar to atmospheric ratios, reflecting a significant seawater circulation in the upper 500 m of exposed crust at this site. MORB-like neon, with elevated 20Ne/22Ne and 21Ne/22Ne ratios, was found in some high temperature steps of heating experiments, but with very small anomalies compared to air. These first results from the lower oceanic crust indicate that subducted lower oceanic crust has an atmospheric 20Ne/22Ne ratio. Most of this neon must be removed during the subduction process, if the ocean crust is to be recirculated in the upper mantle, otherwise this atmospheric neon will overwhelm the upper mantle neon budget. Similarly, the high (U+Th)/3He ratio of these crustal gabbros will generate very radiogenic 4He/3He ratios on a 100 Ma time scale, so lower oceanic crust cannot be recycled into either MORB or oceanic island basalt without some form of processing.

Nd and Sr isotopic composition of Cretaceous MORB

Chemical and isotopic (Nd and Sr) compositions have been determined for 12 Cretaceous basaltic samples (108 Ma old) from Holes 417D and 418A of Legs 51,52 and 53. We have found that: (1) The chemical compositions are typical of MORB. They do not vary systematically with the stratigraphic positions of the analyzed samples; thus, the chemical evolution is independent of the eruption sequence that occurred at this Cretaceous ridge. (2) REE patterns for all rocks are characterized by a strong LREE depletion with (La/Sm)N = 0.38-0.50; no significant Eu anomalies are found; HREE are nearly flat or slightly depleted towards Yb-Lu and have 12-18 * chondritic abundances. Combining the results of previous studies, it suggests that no significant temporal and spatial variation in magma chemistry (especially for LIL elements) has occurred in the 'normal' ridge segments over the last 150 Ma. (3) lsotopically, 143Nd/144Nd ratios vary from 0.513026 to 0.513154, corresponding to epsilon-Nd(0) = +7.5 to +10, and they fall in the typical range of MORB. However, these rocks have unexpectedly high 87Sr/86Sr ratios (0.70355-0.70470) which are attributed to the result of seawater-rock interaction. (4) The Nd model ages (Tin), ranging from 1.53 to 2.47 (average 2.06) AE, suggest that the upper mantle source(s) underwent a large scale chemical differentiation leading to LREE and other LIL element depletion about 2 AE ago, assuming a simple two-stage model. More realistically, the variation in Tm(Nd) or epsilon-Nd could be derived from mixing of heterogeneous mantle sources that were a consequence of continuous mantle differentiation and continental formation. (5) Because of the low mg values (0.52-0.63), the analyzed basaltic rocks do not represent primary liquids of mantle melting. The variation in La/Sm ratios and TiO2 are not compatible with a model in which all rocks are genetically related by a simple fractional crystallization. Rather, it is proposed that the basaltic rocks might have been derived from some heterogeneous upper mantle source with or without later magmatic mixing, and followed by some shallow-level fraetionations.

Chemical composition of basalts from ODP holes at the Kerguelen Plateau

Petrographic and geochemical study of basalts in the Kerguelen Plateau basement revealed changes in composition and character of volcanism during development of this tectonovolcanic structure. The Kerguelen Plateau is one of the largest intraplate rises in the World Ocean. It started to form about 120 Ma ago. Age of basalts and overlying sediments shows that plateau formation was in the northwest direction. Basalts of the Kerguelen Plateau basement are products of tholeiitic melts in terms of geochemistry, but differ from mid-ocean ridge basalt (MORB). They are enriched in incompatible trace elements and rare earth elements (REE) relative to MORB, and degree of enrichment varies in basalts from different segments of the plateau. Composition of basalts does not directly depend on their age. Specific features of plateau magmatism are commonly explained in terms of a long-living deep magma plume, which variously interacted with a depleted upper mantle source at different stages of plateau formation. However, taking into account block morphology and deep structure of the plateau, one can suggest that plateau volcanism was initiated by a large fault. As the volcanism prograded to the northwest, depth of fault penetration into the mantle changed. Composition of basalts in the plateau basement was also governed by formation depth of primary melts.

Geochemistry of igneous rock samples of ODP Site 125-786

A morphologically complex igneous basement was penetrated at Leg 125 Site 786 beneath approximately 100 m of Eocene-Pleistocene sediments at 31°52.45 'N, 141°13.59'E in a 3082-m water depth. The site is located on the forearc basement high (FBH) of the Izu-Bonin (Ogasawara) Arc. In the broadest terms, the sequence in Hole 786B consists of a basal sheeted dike complex, heavily mineralized in places, with overlying pillow lavas giving way to a complex and repeated sequence of interlayered volcanic breccias and lava flows with some thin sedimentary intervals. The sequence has been further cut by dikes or sills, particularly of high-Ca and intermediate-Ca boninite, and is locally strongly sheared by faulting. The whole basement has been covered with middle Eocene-early Pleistocene sediments. A monomict breccia forms the shallowest portion of Hole 786B and a polymict breccia having Mn-oxide-rich clast coatings and matrix forms the deepest part of Hole 786A (-100-160 mbsf). The basement is tectonized in some places, and a mineralized stockwork is present in the deepest part of Hole 786B. A wide variety of rock types form this basement, ranging from mafic to silicic in character and including high-, intermediate-, and low-Ca boninites, intermediate- and low-Ca bronzite andesites, andesite, dacite, and rhyolite groups. Intragroup and intergroup relationships are complicated in detail, and several different upper mantle source(s) probably were involved. A significant role for orthopyroxene-clinopyroxene-plagioclase fractionation is indicated in the mafic-intermediate groups, and the most probable complementary cumulates should be noritic gabbros. Many overall similarities but some subtle differences are noted between the igneous basement at Site 786 and the subaerial outcrops of the FBH to the south in the type boninite locality of Chichijima. Both suites were derived by hydrous melting of a relatively shallow, refractory (harzburgitic) upper mantle source. These Bonin forearc basement rocks are similar in many respects to those of Eocene-Oligocene age now forming the forearc of the Marianas at Leg 60 Site 458 and on Guam. In sharp distinction, the geochemistry of the Eocene-Pleistocene ash sequences overlying the Bonin FBH must have been derived from a very different upper mantle source, implying considerable across-strike differences in sub-arc mantle composition.

Geochemistry of basalts fropm DSDP Leg 65 basalts

Samples of basalt collected on Leg 65 near 22°N on the East Pacific Rise all display the depleted light rare-earth pattern of "normal" oceanic crust. Consequently the La/Ta ratio is close to 18, as opposed to the value of 9 associated with the flat or enriched patterns found along parts of the Mid-Atlantic Ridge and the Emperor Seamount chain. The Leg 65 samples are chemically similar to those from the CYAMEX area at 21 °N and to the Leg 54 samples from 9°N, suggesting homogeneity of the upper mantle under the northern part of the East Pacific Rise over a minimum distance of about 1500 km. The geochemistry of the rocks and their field relationships with respect to depth and distance from the axis of the Rise show no pattern of distribution linked to the degree of fractional crystallization and thus cast doubt on any possible model involving large, long-lived magma chambers at the axis of the Rise.

Chemical and isotopic compositions of sediments from DSDP Hole 21-204

The highly depleted intra-oceanic Tonga-Kermadec island arc forms an endmember of arc systems and a unique location in which to isolate the effects of the slab flux. High precision TIMS uranium, thorium, strontium, neodymium, and lead isotopes, along with complete major and trace element data, have been obtained on an extensive sample set comprising fifty-eight lavas along the arc as well as nineteen samples of the subducting sediments at DSDP site 204 just to the east of the Tonga-Kermadec trench. Ca/Ti and Al/Ti ratios extend from values appropriate to an N-MORB source in the southern Kermadecs to very high ratios in Tonga interpreted to reflect increasing degrees of depletion of the mantle wedge due to backarc basalt extraction. The isotope data emphasize the need for four components in the petrogenesis of the lavas: (1) the mantle wedge; (2) a component with elevated 207Pb/204Pb towards which the Kermadec and southern Tongan lavas extend; (3) a component characterised by high 206Pb/204Pb, Ta/Nd, and low 143Nd/144Nd observed only in the northernmost Tongan islands of Tafahi and Niuatoputapu; (4) a fluid component characterised by strong enrichments of Rb, Ba, U, K, Ph, and Sr, relative to Th, Zr, and the REE and producing large 238U excesses ((230Th/238U) = 0.8-0.5) in the more depleted lavas. The mantle wedge (Component 1) is isotopically similar to the source of the Lau BABB. Component 2 is average pelagic sediment on the downgoing Pacific plate as observed at DSDP sites 595/596 and in the upper sections of the sediment pile at DSDP site 204. Mass balance calculations indicate that less than 0.5% is recycled into the arc lavas; essentially all the subducted sediment is returned to the upper mantle (~0.03 km**3/yr). Exceptionally low concentrations of Ta and Nb relative to Th and the LREE requires that this sediment component is added as a partial melt which was in equilibrium with residual rutile or ilmenite. Component 3 is identified as volcaniclastics from the Louisville Ridge which comprise the lower 44 m of the sediment section intersected at DSDP site 204. These volcaniclastics are spatially restricted to the vicinity of the Louisville Ridge and provide a unique sediment tracer which can be used to show that it takes 4 Myr from the time of subduction to its first appearance in the arc lava signature. Component 4, the fluid contribution to the lava source is inferred to contribute ~1 ppm Rb, 10 ppm Ba, 0.02 ppm U, 600 ppm K, 0.2 ppm Ph, and 30 ppm Sr. It has 87Sr/86Sr = 0.7035 and 206Pb/204Pb = 18.5 and thus it is inferred to have been derived from dehydration of the subducting altered oceanic crust. U-Th isotope disequilibria reflect the time since fluid release from the subducting slab and a reference line through the lowest (230Th/232Th) lavas constrains this to be 30000-50000 yr. The U-Th and Th-Ra isotope systematics are decoupled, and it is suggested that Th-Ra isotope disequilibria record the time since partial melting and thus indicate rapid channelled magma ascent. Olivine gabbro xenoliths from Raoul are interpreted as cumulates to their host lavas with which they form zero age U-Th isochrons indicating that minimal time was spent in magma chambers. The subduction signature is not observed in lavas from the backarc island of Niuafo'ou. These were derived from partial melting of fertile peridotite at 130-160 km depth with melt rates around 0.0002 kg/m**3/yr.