Subduction factory processes beneath the Guguan cross-chain, Mariana Arc: no role for sediments, are serpentinites important?

We need to understand chemical recycling at convergent margins and how chemical interactions between subducted slab and the overlying mantle wedge affect mantle evolution and magmagenesis. This requires distinguishing contributions from recycled individual subducted components as well as those contributed by the mantle. We do this by examining magmatic products generated at different depths above a subduction zone, in an intra-oceanic arc setting. The Guguan cross-chain in the intra-oceanic Mariana arc overlies subducted Jurassic Pacific plate lithosphere at depths of ~125--230 km and erupts mostly basalt. Basalts from rear-arc volcanoes are more primitive than those from the magmatic front, in spite of being derived by lower degrees of melting of less-depleted mantle. Rear-arc magmas also show higher temperatures and pressures of equilibration. Coexisting mineral compositions become more MORB- or OIB-like with increasing height above the subduction zone. Trace element and isotopic variations indicate that the subduction component in cross-chain lavas diminishes with increasing depth to the subduction zone, except for water contents. There is little support for the idea that melting beneath the Mariana Trough back-arc basin depleted the source region of arc magmas, but melting to form rear-arc volcanoes may have depleted the source of magmatic front volcanoes. Enrichments in rear-arc lavas were not caused by sediment melting; the data instead favor an OIB-type mantle that has been modestly affected by subduction zone fluids. Our most important conclusion is that sediment fluids or melts are not responsible for the K--h relationship and other cross-chain chemical and isotopic variations. We speculate that an increasing role for supercritical fluids released from serpentinites interacting with modestly enriched mantle might be responsible for cross-chain geochemical and isotopic variations. Electronic Supplementary Material Supplementary material is available for this article at and is accessible for authorized users.     

Petrogenesis of Volcanic Rocks from Saipan and Rota, Mariana Islands, and Implications for the Evolution of Nascent Island Arcs

An ⁴⁰Ar/³⁹Ar age of 45·1 Ma determined for lavas fromnorthern Saipan confirms that these high-silica rhyolites eruptedduring the ‘proto-arc’ stage of volcanism in theIzu–Bonin–Mariana system, which is characterizedelsewhere by eruption of boninitic lavas. Incompatible traceelement concentrations and Sr, Hf, Nd, and Pb isotope ratiosfor these rhyolites are transitional between those of c. 48Ma boninitic lavas and post-38 Ma ‘first-arc’ andesitesand dacites from Saipan and Rota that have typical subduction-relatedcompositions. These transitional compositions are modeled bycrystal fractionation of parental tholeiitic basalt combinedwith assimilation of young boninitic crust. A second stage ofRayleigh fractionation in the upper crust is required by SiO₂concentrations that exceed 77 wt % and near-zero compatibleelement concentrations. First-arc magma compositions are consistentwith fractionation of basalt and assimilation of crust similarin composition to the first-arc magmas themselves. The mantlesources of the proto-arc and first-arc lavas from Saipan andRota are similar to those of Philippine back-arc basin basaltsbased on Nd and Hf isotopic compositions. The Pb isotope compositionsof these lavas are between those of Pacific sea-floor basaltsand Jurassic and younger cherty and clay-rich sediments. Thiscontrasts with the boninitic proto-arc volcanic rocks from Guamand Deep Sea Drilling Project Sites 458 and 459 that have Pbisotope compositions similar to Pacific basin basalts and volcaniclasticsediments. The preferred explanation for the difference in thenature of proto-arc volcanism between Saipan and other fore-arclocations is that the crust ceased extending 3–4 Myr earlierbeneath Saipan. This was caused by a change from mantle upwelling,fore-arc extension, and shallow melting to an environment dominatedby more normal mantle wedge convection, stable crust, and deepermelting. KEY WORDS: rhyolite; andesite; Mariana arc; isotope ratios; trace elements     

Subduction and melting processes inferred from U-Series, Sr-Nd-Pb isotope, and trace element data, Bicol and Bataan arcs, Philippines

We present U-series, Sr-Nd-Pb isotope, and trace element data from the two principal volcanic chains on Luzon Island, developed over oppositely dipping subduction zones, to explore melting and mass transfer processes beneath arcs. The Bataan (western) and Bicol (eastern) arcs are currently subducting terrigenous and pelagic sediments, respectively, which have different trace element and isotopic compositions. The range of (²³⁰Th/²³⁸U) disequilibria for both arcs is 0.85-1.15; only lavas from Mt. Mayon (Bicol arc) have ²³⁰Th activity excesses. Bataan lavas have higher ⁸⁷Sr/⁸⁶Sr and lower ¹⁴³Nd/¹⁴⁴Nd than Bicol lavas (⁸⁷Sr/⁸⁶Sr = 0.7042-0.7046, ¹⁴³Nd/¹⁴⁴Nd = 0.51281-0.51290 vs. ⁸⁷Sr/⁸⁶Sr = 0.70371-0.70391, ¹⁴³Nd/¹⁴⁴Nd = 0.51295-0.51301) and both arcs show steep linear arrays towards sediment values on ²⁰⁷Pb/²⁰⁴Pb vs. ²⁰⁶Pb/²⁰⁴Pb diagrams. Analysis of incompatible element and isotopic data allows identification of a sediment component that, at least in part, was transferred as a partial melt to the mantle wedge peridotite. Between 1% and 5% sediment melt addition can explain the isotopic and trace element variability in the rocks from both arcs despite the differences in sediment supply. We therefore propose that sediment transfer to the mantle wedge is likely mechanically or thermally limited. It follows that most sediments are either accreted, reside in the sub-arc lithosphere, or are recycled into the convecting mantle. However, whole-sale sediment recycling into the upper mantle is unlikely in light of the global mid-ocean ridge basalt data. Fluid involvement is more difficult to characterize, but overall the Bicol arc appears to have more fluid influence than the Bataan arc. Rock suites from each arc can be related by a dynamic melting process that allows for ²³⁰Th ingrowth, either by dynamic or continuous flux melting, provided the initial (²³⁰Th/²³²Th) of the source is ∼0.6-0.7. The implication of either model is that inclined arrays on the U-Th equiline diagram may not have chronologic significance. Modeling also suggests that U-series disequilibria are influenced by the tectonic convergence rate, which dictates mantle matrix flow. Thus with slower matrix flow there is a greater degree of ²³⁰Th ingrowth. While other factors such as prior mantle depletion and addition of a subducted component may explain some aspects of U-series data, an overall global correlation between tectonic convergence rate and the extent of U-Th disequilibria may originate from melting processes.     

Evidence for adiabatic decompression melting in the Southern Mariana Arc from high-Mg lavas and melt inclusions

Unusually magnesian (Mg# ∼76) basalts have been sampled from a small submarine volcano situated on the Mariana arc magmatic front. Total alkalis range from 1.7 to 1.94%, Al₂O₃ from 9.09 to 10.3% and CaO from 13.9 to 14.09%. These lavas can be classified based on mineralogy as picrite and ankaramite. Olivine-hosted melt inclusions (MIs) have median MgO contents of 17.17–17.86 wt%, 0.35–0.5% TiO₂, 42–50% SiO₂ and 1.66–3.43% total alkalis, which suggest that the parental magmas were primitive mantle melts. Trace element concentrations for both MIs and lavas are arc-like, although more depleted than most arc lavas. Chlorine (182–334 ppm) and H₂O contents (0.11–0.64 wt%) in the MIs are consistent with the estimated median oxygen fugacities (log ΔFMQ of + 1.53–1.66) which lie at the low end of the range estimates for arc basalts and picrites (ΔFMQ = + 1 to + 3). Isotopic compositions of Sr, Nd, Hf and Pb are similar to those of other Mariana arc lavas and indicate derivation from an Indian Ocean mantle domain. The averaged magmatic temperature estimate from several geothermometers was 1,367°C at 1–1.5 GPa. We propose that high-Mg magmagenesis in this region results from the adiabatic decompression melting of relatively anhydrous but metasomatized mantle wedge. This melting is attributed to enhanced upwelling related to unusual tectonics on the over-riding plate related to a tear or other discontinuity on the subducted slab.Electronic Supplementary Material Supplementary material is available for this article at and is accessible for authorized users.     

Multi-stage mixing in subduction zones: Application to Merapi volcano (Java island, Sunda arc)

Many studies have argued for the contribution of at least three components, namely the mantle wedge, the subducted oceanic crust, and its sediment cover, to describe the geochemistry of island arc volcanics. However, isotope correlations reflecting a simple binary mixing can be observed at the scale of a single arc island or volcano. Here we investigate the possibility that these trends reflect pseudo-binary mixing relationships in a three-component system. We present a simplified, two-stage model for the systematic isotope modelling of a cogenetic suite of arc lavas. This includes metasomatism of portions of the mantle wedge by hydrous phases released from the down-going oceanic crust, and sediments, followed by progressive mixing and melting. A consequence of this model is that it leads to a two end-member mixing process from the mantle wedge, oceanic crust, and sediment components. To solve the model we reduce it to a step-by-step procedure combined with a Monte-Carlo simulation. The procedure consists of: (i) producing a large number of random values on each variable of the model; (ii) using the computed values to calculate the isotopic compositions of lavas; and (iii) comparing the obtained isotopic compositions with measured data. Applied to a new set of Sr, Nd, and Pb isotope data for volcanics (basalts, basaltic andesites, trachybasalts, and basaltic trachyandesites) from Merapi volcano (Java island, Sunda arc), the model successfully reproduces the binary mixing relationships previously documented for the medium-K and high-K lava series from this volcano, thus giving further support to the hypothesis that this distinction is inherited from the primary magmas and primarily reflects a property of the mantle source. The results allow identification of a set of numerical values for bulk partition coefficients (solid/hydrous fluid, and solid/H₂O-rich melt) and variables (e.g., the mass ratio between the metasomatizing phase and the mantle wedge), which can be used for quantitative arc-lava petrogenetic calculations. They also require a direct relationship between dehydration of the slab and melting of the metasomatized mantle wedge. Finally, our evaluation shows that for isotope modelling of the Merapi lavas, the two-stage procedure is controlled more by the considered source components (mantle wedge, oceanic crust, sediments, and their derivatives) than by the various processes involved (dehydration, melting, and mixing).     

Mafic Late Miocene–Quaternary volcanic rocks in the Kamchatka back arc region: implications for subduction geometry and slab history at the Pacific–Aleutian junction

New ⁴⁰Ar/³⁹Ar and published ¹⁴C ages constrain voluminous mafic volcanism of the Kamchatka back-arc to Miocene (3–6 Ma) and Late Pleistocene to Holocene (<1 Ma) times. Trace elements and isotopic compositions show that older rocks derived from a depleted mantle through subduction fluid-flux melting (>20%). Younger rocks form in a back arc by lower melting degrees involving enriched mantle components. The arc front and Central Kamchatka Depression are also underlain by plateau lavas and shield volcanoes of Late Pleistocene age. The focus of these voluminous eruptions thus migrated in time and may be the result of a high fluid flux in a setting where the Emperor seamount subducts and the slab steepens during rollback during terrain accretions. The northern termination of Holocene volcanism locates the edge of the subducting Pacific plate below Kamchatka, a “slab-edge-effect” is not observed in the back arc region.     

The origin of high-Mg magmas in Mt Shasta and Medicine Lake volcanoes,Cascade Arc (California): higher and lower than mantle oxygen isotope signatures attributed to current and past subduction

We report the oxygen isotope composition of olivine and orthopyroxene phenocrysts in lavas from the main magma types at Mt Shasta and Medicine Lake Volcanoes: primitive high-alumina olivine tholeiite (HAOT), basaltic andesites (BA), primitive magnesian andesites (PMA), and dacites. The most primitive HAOT (MgO > 9 wt%) from Mt. Shasta has olivine δ¹⁸O (δ¹⁸O_Ol) values of 5.9–6.1‰, which are about 1‰ higher than those observed in olivine from normal mantle-derived magmas. In contrast, HAOT lavas from Medicine Lake have δ¹⁸O_Ol values ranging from 4.7 to 5.5‰, which are similar to or lower than values for olivine in equilibrium with mantle-derived magmas. Other magma types from both volcanoes show intermediate δ¹⁸O_Ol values. The oxygen isotope composition of the most magnesian lavas cannot be explained by crustal contamination and the trace element composition of olivine phenocrysts precludes a pyroxenitic mantle source. Therefore, the high and variable δ¹⁸O_Ol signature of the most magnesian samples studied (HAOT and BA) comes from the peridotitic mantle wedge itself. As HAOT magma is generated by anhydrous adiabatic partial melting of the shallow mantle, its 1.4‰ range in δ¹⁸O_Ol reflects a heterogeneous composition of the shallow mantle source that has been influenced by subduction fluids and/or melts sometime in the past. Magmas generated in the mantle wedge by flux melting due to modern subduction fluids, as exemplified by BA and probably PMA, display more homogeneous composition with only 0.5‰ variation. The high-δ¹⁸O values observed in magnesian lavas, and principally in the HAOT, are difficult to explain by a single-stage flux-melting process in the mantle wedge above the modern subduction zone and require a mantle source enriched in ¹⁸O. It is here explained by flow of older, pre-enriched portions of the mantle through the slab window beneath the South Cascades.     

勘察加岛弧岩浆岩研究进展

西太平洋分布了全球大部分的洋内俯冲带,也是全球沟-弧-盆体系最发育的地区。勘察加(Kamchatka)半岛位于俄罗斯远东地区,地处太平洋西北部(51°~60°N、155°~164°E),是全球环太平洋岛弧的重要组成部分。前人对勘察加岛弧岩石地幔源区性质、熔融过程、岩浆结晶分异及熔/流体交代过程进行了详细的研究,并获得了丰硕的成果。最新的研究进展表明:(1)勘察加岛弧前缘火山和中部火山的源区主要为亏损地幔,而弧后区域则存在较为富集的地幔贡献;(2)勘察加岛弧不同区域的地幔源区流体性质具有一定的差异,导致从前缘火山至中部火山,地幔熔融程度逐渐降低;(3)勘察加岛弧不同区域岩石地球化学成分存在差异,而且,沿穿弧剖面某些元素或同位素(如δ11 B)表现出系统变化的特征,反应了俯冲板片流体通量和流体性质的差异;(4)勘察加半岛部分多期次火山(如Klyuchevskoy火山)地球化学成分复杂,可能反应了源区熔融条件的不同和岩浆结晶分异过程;(5)勘察加岛弧北部与阿留申岛弧近直角相交,导致异常的构造背景,促使该区域形成了具有埃达克质特征的岛弧岩浆。     

Origin of high field strength element enrichment in volcanic arcs: Geochemical evidence from the Sulu Arc, southern Philippines

Lavas from the Sulu Arc, southern Philippines, exhibit an enrichment in high field strength elements (HFSE) that represents a departure from the typical volcanic arc geochemical signature. It has been postulated that this relative enrichment arises from metasomatism of mantle wedge peridotites by melts derived from the subducting oceanic lithosphere, through formation of amphibole which subsequently breaks down and enriches the mantle source of parental arc magmas in HFSE. Divergent chemical and isotopic characteristics between Sulu Arc HFSE-enriched lavas and the Sulu Sea crust being subducted—the presumed source of slab-derived melts—render it unlikely, however, that HFSE enrichment arises from the influence of such melts. New geochemical data suggest that the varying degrees of HFSE enrichment in Sulu Arc lavas are instead the result of variable amounts of mixing between enriched and depleted mantle end-components—the sources of South China Sea intraplate lavas and Sulu seafloor basalts, respectively—within a compositionally heterogeneous mantle wedge.     

New Age and Geochemical Data from the Southern Colville and Kermadec Ridges,SW Pacific: Insights into the recent geological history and petrogenesis of the Proto-Kermadec (Vitiaz) Arc

The intra-oceanic Kermadec arc system extends ~1300 km between New Zealand and Fiji and comprises at least 30 arc front volcanoes, the Havre Trough back-arc and the remnant Colville and Kermadec Ridges. To date, most research has focussed on the Kermadec arc front volcanoes leaving the Colville and Kermadec Ridges virtually unexplored. Here, we present seven ⁴⁰Ar/³⁹Ar ages together with a comprehensive major and trace element and Sr-, Nd-, and Pb-isotope dataset from the Colville and Kermadec Ridges to better understand the evolution, petrogenesis and splitting of the former proto-Kermadec (Vitiaz) Arc to form these two remnant arc ridges. Our ⁴⁰Ar/³⁹Ar ages range from ~7.5–2.6 Ma, which suggests that arc volcanism at the Colville Ridge occurred continuously and longer than previously thought. Recovered Colville and Kermadec Ridge lavas range from mafic picro-basalts (MgO = ~8 wt%) to dacites. The lavas have arc-type normalised incompatible element patterns and Sr and Pb isotopic compositions intermediate between Pacific MORB and subducted lithosphere (including sediments, altered oceanic crust and serpentinised uppermost mantle). Geochemically diverse lavas, including ocean island basalt-like and potassic lavas with high Ce/Yb, Th/Zr, intermediate ²⁰⁶Pb/²⁰⁴Pb and low ¹⁴³Nd/¹⁴⁴Nd ratios were recovered from the Oligocene South Fiji Basin (and Eocene Three Kings Ridge) located west of the Colville Ridge. If largely trench-perpendicular mantle flow was operating during the Miocene, this geochemical heterogeneity was likely preserved in the Colville and Kermadec sub arc mantle. Between 4.41 ± 0.35 and 3.40 ± 0.24 Ma some Kermadec Ridge lavas record a shift from Colville Ridge- to Kermadec arc front-like, suggesting the proto-Kermadec (Vitiaz-) arc split post 4.41 ± 0.35 Ma. The Colville and Kermadec Ridge data therefore place new constraints on the regional tectonic evolution and highlight the complex interplay between pre-existing mantle heterogeneities and material fluxes from the subducting Pacific Plate. The new data allow us to present a holistic (yet simplified) picture of the tectonic evolution of the late Vitiaz Arc and northern Zealandia since the Miocene and how this tectonism influences volcanic activity along the Kermadec arc at the present.     

http://www.sciencedirect.com/science/article/pii/S1674987114000875

The origin and petrogenesis of the Cameroon Volcanic Line(CVL),composed of volcanoes that form on both the ocean floor and the continental crust,are difficult to understand because of the diversity,heterogeneity,and nature of available data.Major and trace elements,and Sr-Nd-Pb isotope data of volcanic rocks of the CVL spanning four decades have been compiled to reinterpret their origin and petrogenesis.Volcanic rocks range from nephelinite,basanite and alkali basalts to phonolite,trachyte and rhyolite with the presence of a compositional gap between Si O258e64 wt.%.Similarities in geochemical characteristics,modeled results for two component mixing,and the existence of mantle xenoliths in most mafic rocks argue against significant crustal contamination.Major and trace element evidences indicate that the melting of mantle rocks to generate the CVL magma occurred dominantly in the garnet lherzolite stability field.Melting models suggest small degree(3%)partial melting of mantle bearing(6e10%)garnet for Mt.Etinde,the Ngaoundere Plateau and the Biu Plateau,and5%of garnet for the oceanic sector of the CVL,Mt.Cameroon,Mt.Bambouto,Mt.Manengouba and the Oku Volcanic Group.The Sr-Nd-Pb isotope systematics suggest that mixing in various proportions of Depleted MORB Mantle(DMM)with enriched mantle 1 and 2(EM1 and EM2)could account for the complex isotopic characteristics of the CVL lavas.Low Mg number(Mg#100 Mg O/(Mg O t Fe O))and Ni,Cr and Co contents of the CVL mafic lavas reveal their crystallization from fractionated melts.The absence of systematic variation in Nb/Ta and Zr/Hf ratios,and Sr-Nd isotope compositions between the mafic and felsic lavas indicates progressive evolution of magmas by fractional crystallization.Trace element ratios and their plots corroborate mantle heterogeneity and reveal distinct geochemical signatures for individual the CVL volcanoes.     

Drastic shift in lava geochemistry in the volcanic-front to rear-arc region of the Southern Kamchatkan subduction zone: Evidence for the transition from slab surface dehydration to sediment melting

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.     

Young basalts of the central Washington Cascades, flux melting of the mantle, and trace element signatures of primary arc magmas

Basaltic lavas from the Three Sisters and Dalles Lakes were erupted from two isolated vents in the central Washington Cascades at 370–400 ka and 2.2 Ma, respectively, and have distinct trace element compositions that exemplify an important and poorly understood feature of arc basalts. The Three Sisters lavas are calc-alkaline basalts (CAB) with trace element compositions typical of most arc magmas: high ratios of large-ion-lithophile to high-field-strength elements (LILE/HFSE), and strong negative Nb and Ta anomalies. In contrast, the Dalles Lakes lavas have relatively low LILE/HFSE and no Nb or Ta anomalies, similar to ocean-island basalts (OIB). Nearly all Washington Cascade basalts with high to moderate incompatible element concentrations show this CAB or OIB-like compositional distinction, and there is pronounced divergence between the two magma types with a large compositional gap between them. We show that this trace element distinction can be easily explained by a simple model of flux-melting of the mantle wedge by a fluid-rich subduction component (SC), in which the degree of melting (F) of the peridotite source is correlated with the amount of SC added to it. Distinctive CAB and OIB-like trace element compositions are best explained by a flux-melting model in which dF/dSC decreases with increasing F, consistent with isenthalpic (heat-balanced) melting. In the context of this model, CAB trace element signatures simply reflect large degrees of melting of strongly SC-fluxed peridotite along relatively low dF/dSC melting trends, consistent with derivation from relatively cold mantle. Under other conditions (i.e., small degrees of melting or large degrees of melting of weakly SC-fluxed peridotite [high dF/dSC]), either OIB- or MORB (mid-ocean ridge basalt)-like compositions are produced. Trace element and isotopic compositions of Washington Cascade basalts are easily modeled by a correlation between SC and F across a range of mantle temperatures. This implies that the dominant cause of arc magmatism in this region is flux melting of the mantle wedge. Received: 2 March 1999 / Accepted: 18 August 1999     

Magmatic evolution of the South Shetland Islands,Antarctica, and implications for continental crust formation

Lavas from the South Shetland Islands volcanic arc (northern Antarctic Peninsula) have been investigated in order to determine the age, petrogenesis and compositional evolution of a long-lived volcanic arc constructed on 32-km-thick crust, a thickness comparable with average continental crust. New ⁴⁰Ar–³⁹Ar ages for the volcanism range between 135 and 47 Ma and, together with published younger ages, confirm a broad geographical trend of decreasing ages for the volcanism from southwest to northeast. The migration pattern breaks down in Palaeogene time, with Eocene magmatism present on both Livingston and King George islands, which may be due to a change in both subduction direction and velocity after c. 60 Ma. The lavas range from tholeiitic to calc-alkaline, but there is no systematic change with age or geographic location. The compositions of lavas from the north-eastern islands indicate magma generation in a depleted mantle wedge with relatively low Sr and high Nd isotopic compositions and low U/Nb, Th/Nd and Ba/Nb ratios that was metasomatized by hydrous fluids from subducted basaltic oceanic crust. Lavas from the south-western islands show an additional sedimentary influence most likely due to fluid release from subducted sediments into the mantle wedge. Although magmatic activity in the South Shetland arc extended over c. 100 m.y., there is no evolution towards more enriched or evolved magmas with time. Few South Shetland arc lavas are sufficiently enriched with incompatible elements to provide a potential protolith for the generation of average continental crust. We conclude that even long-established subduction zones with magmatic systems founded on relatively thick crust do not necessarily form continental crustal building blocks. They probably represent only the juvenile stages of continental crust formation, and additional re-working, for example during subsequent arc-continental margin collision, is required before they can evolve into average continental crust.     

Isotopic and trace element constraints on the petrogenesis of lavas from the Mount Adams volcanic field,Washington

Strontium, Nd, Pb, Hf, Os, and O isotope compositions for 30 Quaternary lava flows from the Mount Adams stratovolcano and its basaltic periphery in the Cascade arc, southern Washington, USA indicate a major component from intraplate mantle sources, a relatively small subduction component, and interaction with young mafic crust at depth. Major- and trace-element patterns for Mount Adams lavas are distinct from the rear-arc Simcoe volcanic field and other nearby volcanic centers in the Cascade arc such as Mount St. Helens. Radiogenic isotope (Sr, Nd, Pb, and Hf) compositions do not correlate with geochemical indicators of slab-fluids such as (Sr/P) _n and Ba/Nb. Mass-balance modeling calculations, coupled with trace-element and isotopic data, indicate that although the mantle source for the calc-alkaline Adams basalts has been modified with a fluid derived from subducted sediment, the extent of modification is significantly less than what is documented in the southern Cascades. The isotopic and trace-element compositions of most Mount Adams lavas require the presence of enriched and depleted mantle sources, and based on volume-weighted chemical and isotopic compositions for Mount Adams lavas through time, an intraplate mantle source contributed the major magmatic mass of the system. Generation of basaltic andesites to dacites at Mount Adams occurred by assimilation and fractional crystallization in the lower crust, but wholesale crustal melting did not occur. Most lavas have Tb/Yb ratios that are significantly higher than those of MORB, which is consistent with partial melting of the mantle in the presence of residual garnet. δ ¹⁸O values for olivine phenocrysts in Mount Adams lavas are within the range of typical upper mantle peridotites, precluding involvement of upper crustal sedimentary material or accreted terrane during magma ascent. The restricted Nd and Hf isotope compositions of Mount Adams lavas indicate that these isotope systems are insensitive to crustal interaction in this juvenile arc, in stark contrast to Os isotopes, which are highly sensitive to interaction with young, mafic material in the lower crust.     

Magnesian andesite and dacite lavas from Mt. Shasta,northern California: products of fractional crystallization of H<Subscript>2</Subscript>O-rich mantle melts

Mt. Shasta andesite and dacite lavas contain high MgO (3.5–5 wt.%), very low FeO*/MgO (1–1.5) and 60–66 wt.% SiO₂. The range of major and trace element compositions of the Shasta lavas can be explained through fractional crystallization (~50–60 wt.%) with subsequent magma mixing of a parent magma that had the major element composition of an H₂O-rich primitive magnesian andesite (PMA). Isotopic and trace element characteristics of the Mt. Shasta stratocone lavas are highly variable and span the same range of compositions that is found in the parental basaltic andesite and PMA lavas. This variability is inherited from compositional variations in the input contributed from melting of mantle wedge peridotite that was fluxed by a slab-derived, fluid-rich component. Evidence preserved in phenocryst assemblages indicates mixing of magmas that experienced variable amounts of fractional crystallization over a range of crustal depths from ~25 to ~4 km beneath Mt. Shasta. Major and trace element evidence is also consistent with magma mixing. Pre-eruptive crystallization extended from shallow crustal levels under degassed conditions (~4 wt.% H₂O) to lower crustal depths with magmatic H₂O contents of ~10–15 wt.%. Oxygen fugacity varied over 2 log units from one above to one below the Nickel-Nickel Oxide buffer. The input of buoyant H₂O-rich magmas containing 10–15 wt.% H₂O may have triggered magma mixing and facilitated eruption. Alternatively, vesiculation of oversaturated H₂O-rich melts could also play an important role in mixing and eruption.     

Contribution of slab-fluid in arc magmas beneath the Japan arcs

Identifying the amount and composition of slab-derived fluid and its spatial variation is key to quantifying fluid processes in subduction zones. Based on the isotopic systematics of arc lavas, we found regional variations over the Japan arcs in terms of the amount and composition of slab-derived fluid added to the melting source region. The average amounts of slab-derived fluid differ among the arcs: 2.6 wt.% for Central Japan, 0.69 wt.% for Ryukyu, 0.17 wt.% for NE Japan, and 0.12 wt.% for both Kurile and Izu–Bonin. These differences may be attributed to the arc setting (oceanic or continental) and the geometry of the slabs. Contribution of sediment involved in the slab-derived fluid is dominant in NE Japan compared to the Izu–Bonin and Central Japan arcs. This could be attributed to mechanical features such as fractures near the subducting plate surface, in addition to the arc setting and the slab geometry. Therefore, the amount and composition of slab-derived fluid are thought to be controlled not only by the thermal conditions, but also by the tectonic and mechanical features around the subduction zone. On top of the variability of slab-derived fluid, the mantle wedge shows the regional variation in terms of proportion of the Pacific-type and Indian-type MORB-source mantle components, which also contributes to the compositional variations of arc magmas.     

On the origin of andesite in the northern Mariana Island Arc: Implications from Agrigan

Magmatic evolution on the active volcano of Agrigan in the northern Mariana Island Arc is interpreted as resulting in the production of calc-alkaline andesites by the fractional crystallization of high-alumina basalt. Basaltic products predominate, but the ratio of andesites to basalts increases with time up to an event of voluminous andesitic pyroclastic ejection accompanied by caldera-collapse; post-collapse lavas are entirely basaltic. Moderate iron-enrichment is demonstrated for the volcanic suite, with indications of a progressive, pre-caldera decrease in iron-enrichment; post-caldera lavas display a return to moderate Fe-enrichment. Overall, the lavas are enriched in the LIL elements (K, Rb, Ba, Sr) and depleted in Ti, Mg, Cr, and Ni. From the oldest to the youngest pre-caldera volcanic sequence, the LIL elements increase 3-6X while Ca and Mg decrease by 50% or more. Approximately constant K/ Rb (430±60) and ⁸⁷Sr/⁸⁶Sr (0.7032–0.7034) indicate consanguinity of the basalts and the andesites. Cumulate plutonic xenoliths, common in the lavas, are composed of mineral phases also encountered as phenocrysts. The following order of crystallization is indicated: olivine; anorthite-bytownite; clinopyroxene; orthopyroxene and titanomagnetite. Co-existing xenolithic olivines (Fo_74–83) and plagioclase (An_88–96) are typical of calc-alkaline island-arc assemblages and contrast with assemblages in the tholeiites from the Mariana Trough to the west. The relatively fayalitic composition and low abundances of Ni in olivines and Cr in clinopyroxenes indicate equilibrium with an already-fractionated liquid. These data, along with structural evidence, high Ca in the olivines, and comparison of the observed assemblages with experimental studies, suggests that these xenoliths formed as crystal cumulates at the floor of a shallow ( 7 km) crustal magma chamber.Major element modeling studies using the separation of observed xenocrystic and phenocrystic phases from assumed parental liquids reproduce the observed temporal and geochemical variations in the lavas. Trace element modeling parallels this evolution with the exception of Cr and Ni in the andesites. An extensive (16.3 km³) gabbroic body is required by this modeling to be present beneath Agrigan to produce the inferred volumes of the various lithologies preserved in the volcano's evolution. The sum of stratigraphic, geochemical, and isotopic evidence on Agrigan supports the derivation of calc-alkaline andesite by the removal of about 75% solids from a high-alumina basalt accompanied by a process of K and Rb enrichment, such as volatile-transfer. Considerations of ⁸⁷Sr/⁸⁶Sr, ¹⁴³Nd/¹⁴⁴Nd, and ³He/⁴He isotopic data indicate that the source region of these parental liquids lies in the mantle, not subducted crust. In the northern Marianas, the model of a shallow-crustal origin for andesite is preferred over one requiring andesite generation in the deeper mantle and/or subducted slab.     

A geochemical study of magmatism associated with the initial stages of back-arc spreading

Bransfield Strait is a narrow basin separating the South Shetland Islands from the Antarctic Peninsula and is attributed to recent back-arc extension behind the South Shetland volcanic arc. The volcanic islands of Deception and Bridgeman are situated close to the axis of spreading, whereas Penguin Island lies slightly to the north of this axis. The mineralogy, petrology and geochemistry of the lavas of the three volcanoes have been studied in order to provide information on the nature of magmatism associated with the initial stages of back-arc spreading.Deception Island lavas range from olivine basalt to dacite, and all are highly sodic, with high Na/K, K/Rb, Ba/Rb and Zr/Nb ratios and with Ce_N/Yb_N = 2. Incompatible elements increase systematically between basalt and rhyodacite, while Sr decreases, suggesting that fractional crystallisation is the dominant process relating lava compositions. The rhyodacites have high concentrations of Zr, Y and the REE and negative Eu anomalies and are compositionally similar to oceanic plagiogranite. Bridgeman Island lavas are mostly basaltic andesites, but the levels of many incompatible elements, including REE, are significantly lower than those of Deception lavas, although Ce_N/Yb_N ratios and ⁸⁷Sr/⁸⁶Sr ratios (0.7035) are the same. Penguin Island lavas are magnesian, mildly alkaline olivine basalts with a small range of composition that can be accommodated by fractional crystallisation of olivine, clinopyroxene and/or chromite. Penguin lavas have higher ⁸⁷Sr/⁸⁶Sr (0.7039) and Ce_N/ Yb_N (4) ratios than Deception and Bridgeman lavas. The Rb/Sr ratios of Deception and Penguin basalts (ca. 0.01) are much too low to account for their present ⁸⁷Sr/⁸⁶Sr ratios.Modelling suggests that the source regions of the lavas of the three volcanoes share many geochemical features, but there are also some significant differences, which probably reflects the complex nature of the mantle under an active island arc combined with complex melting relationships attending the initial stages of back-arc spreading. Favoured models suggest that Bridgeman lavas represent 10–20% melting and the more primitive Deception lavas 5–10% melting of spinel-peridotite, whereas Penguin lavas represent less then 5% melting of a garnet-peridotite source. The mantle source for Bridgeman lavas seems to have undergone short-term enrichment in K, Rb and Ba, possibly resulting from dewatering of the subducted slab. Hydrous melting conditions may also account for the more siliceous, high-alumina nature and low trace element contents of Bridgeman lavas.     

Geochemical and isotopic variability in lavas from the eastern Trans-Mexican Volcanic Belt: Slab detachment in a subduction zone with varying dip

Strong compositional variations are observed in the late-Miocene to Quaternary volcanic rocks of the eastern Trans-Mexican Volcanic Belt. Geochemical and isotopic analyses of samples well constrained in age indicate an abrupt change in magma composition in the late-Miocene (∼ 7.5 Ma), when calc-alkaline, subduction-related magmatism was replaced by mafic, alkaline, OIB-like volcanism. Afterwards, volcanism migrated toward the trench and the erupted lavas showed increasing contributions of subduction components reflected in higher Th/Nb, La/Sm(n), Ba/Nb, and Ba/Th ratios. Lavas from volcanic fields located closer to the trench show clearer, although strongly variable, arc signatures as well as evidence of subducted sediment contributions. Farther from the trench, only lavas emplaced in late-Pliocene time appear to be slightly modified by subduction components, whereas the youngest Quaternary lavas can be regarded as intraplate lavas modified by crustal assimilation.The sudden change in magma composition in the late-Miocene is related to detachment of the subducting slab, which allowed the infiltration of enriched asthenospheric mantle into the mantle wedge. After detachment, the subducting plate started to increase its dip because of the loss of slab pull. This caused (1) the migration of the arc toward the trench, (2) convection of enriched asthenosphere into the mantle wedge, and (3) an increasing contribution of slab components to the melts, in a process that resulted in a highly heterogeneous source mantle. The variable contribution of subduction-related components to the magmas is controlled by the heterogeneous character of the source, the depth of the subducting plate, and the previous magmatic history of the areas.