Journal of Petrology

Journal of Petrology Advance Access originally published online on July 7, 2008
Journal of Petrology 2008 49(8):1427-1448; doi:10.1093/petrology/egn031

This Article Abstract FREE Full Text (PDF) Supplementary Data All Versions of this Article: 49/8/1427    most recent egn031v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (3) Request Permissions Google Scholar Articles by Bouvier, A.-S. Articles by Deloule, E. Search for Related Content GeoRef GeoRef Citation Social Bookmarking          What's this?

© The Author 2008. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org

Slab-Derived Fluids in the Magma Sources of St. Vincent (Lesser Antilles Arc): Volatile and Light Element Imprints

A.-S. Bouvier^1,*, N. Métrich² and E. Deloule¹

¹Centre De Recherche Pétrographiques Et Géochimiques, Nancy-Université, CNRS 54501 Vandoeuvre-Les-Nancy, France
²Laboratoire Pierre Süe, CNRS-CEA, Ce-Saclay, 91191 Gif-Sur-Yvette Cedex, France

RECEIVED DECEMBER 11, 2007; ACCEPTED MAY 6, 2008

	ABSTRACT

TOP ABSTRACT INTRODUCTION SAMPLE LOCATION AND DESCRIPTION ANALYTICAL METHODS MINERAL CHEMISTRY MELT INCLUSIONS DISCUSSION CONCLUSIONS: CONSTRAINTS ON THE... SUPPLEMENTARY DATA REFERENCES

 
It is generally accepted that the parental magmas of the Lesser Antilles arc were generated by partial melting of a mid-ocean ridge basalt (MORB)-type mantle source modified by slab-derived components. To determine the nature of these components, the H₂O, S, Cl, F, Li and B contents and ⁷Li, ¹¹B, ¹⁸O and ³⁴S compositions were systematically determined in olivine-hosted melt inclusions from St. Vincent (southern part of the arc). Both the geochemical and isotopic data define a broad compositional spectrum. On the whole, the melt inclusions have basaltic to CaO-rich (>15·0 wt %), SiO₂-poor (< 45·7 wt %) compositions. Most of the entrapped melts result from 10–20% batch partial melting of a MORB-type mantle source modified initially by dehydration fluids with low solute contents and a seawater-like chemical signature. As a result, the melt inclusions are enriched in B, Cl and H₂O compared with MORB and have ¹¹B up to + 15, ³⁴S of 2 and ¹⁸O down to + 3. In contrast, some others record initial magmatic heterogeneities that require input of fluids derived from (1) the dehydration of altered oceanic crust in agreement with the selective B enrichment (up to 53 ppm) in the melt and negative lithium isotopic compositions, and (2) the dehydration of sediments resulting in distinctive ¹¹B and ³⁴S (down to –20 and –8, respectively) and high Li contents in the melts. The CaO-rich melt inclusions cannot be distinguished from the others on the basis of their isotopic signatures. They possibly reflect magma interactions with CaO-rich, amphibole-bearing lithologies. Combination of our results with literature experimental data leads to the conclusion that St. Vincent basaltic melt inclusions—whose water content varies from 2·2 to 3·6 wt %—represent magmas derived from a rather limited portion of the mantle wedge, by partial melting at between 13 and 14·5 kbar and a restricted temperature range (1220–1190°C).

KEY WORDS: Lesser Antilles arc; subduction; melt inclusions; volatiles; stable isotopes

	INTRODUCTION

TOP ABSTRACT INTRODUCTION SAMPLE LOCATION AND DESCRIPTION ANALYTICAL METHODS MINERAL CHEMISTRY MELT INCLUSIONS DISCUSSION CONCLUSIONS: CONSTRAINTS ON THE... SUPPLEMENTARY DATA REFERENCES

 
A number of processes can generate melting in the mantle wedge above subducted slabs of oceanic lithosphere (e.g. Gaetani & Grove, 2007): (1) anhydrous decompression melting, essentially active in back-arc spreading centres; (2) hydrous partial melting within buoyant diapirs, a process that appears limited because of the balance between buoyancy forces and thermal equilibration time; (3) melting by reactive porous flow of slab-derived fluids, inferred to be the dominant process in subduction zones. Fluids released from both the recycled altered oceanic crust and subducted sediments are thought to play a significant role in magma genesis in the sub-arc mantle.

The intra-oceanic volcanic arc of the Lesser Antilles (Fig. 1) is the result of the relatively slow (2 cm/year), westward subduction of the Atlantic plate beneath the Caribbean plate (see Macdonald et al., 2000, for a review). From north to south along the arc, the lavas broadly vary in composition from low-K tholeiites (north) to calc-alkaline (central part) to silica-undersaturated basalts in the southern part where picrites and rare ankaramites occur (Brown et al., 1977; Macdonald et al., 2000). Compositional variations in the magmatic suites of the Lesser Antilles lavas result from the superimposition of different processes, including polybaric fractional crystallization, crustal contamination and magma mixing, combined with a variety of compositions and proportions of slab components added to the mantle source (Macdonald et al., 2000). High-magnesia basalts (HMB, mg-number >70, MgO >10 wt %), are regarded as the parental magmas of most of the Lesser Antilles lava series (e.g. Macdonald et al., 2000, and references therein; Pichavant et al., 2002) although Draper & Johnston (1992) proposed that high-alumina basalts and basaltic andesites were produced by melting of the subducted slab. The generation or final equilibration of primary magmas beneath the arc has been inferred to take place at pressures between 15 and 30 kbar, based on pseudo-ternary projection of inferred primary lava compositions in the Ol–Di–Pl–Qz anhydrous system (Macdonald et al., 2000). The extent to which the most primitive parental magmas have been affected by assimilation and fractional crystallization (AFC) process has also been widely debated, particularly at Grenada (Devine, 1995; Thirlwall et al., 1996; Macdonald et al., 2000). Thirlwall et al. (1996) proposed that even the picritic lavas have undergone minor crustal assimilation (2–5%), whereas Macdonald et al. (2000) suggested that only the ankaramitic ‘C-series’ at Grenada were affected by crustal contamination.

View larger version (22K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 1. Map of the Lesser Antilles arc modified after Poussineau (2005). St. Vincent is located in the southern part of the inner volcanic arc, and shown in the inset on the left-hand side of the figure, with the locations of the lapilli deposits (SVN4) and HMB lava samples analysed in this study (numbers refer to the name of the sample; e.g. 4 means SVN4).

 
Lesser Antilles arc magmas have long been considered to originate from a normal mid-ocean ridge basalt (N-MORB) mantle source that has been affected by slab-derived fluid components produced by dehydration of altered oceanic crust and melting of subducted sediments (Hawkesworth et al., 1979; Thirlwall & Graham, 1984; White & Patchett, 1984; White et al., 1985; Davidson, 1985; Davidson & Harmon 1989; Thirlwall et al., 1994; Pearce & Peate, 1995; Smith et al., 1997; McDonald et al., 2000). The contribution of 0·5–3·5% sediments to the mantle source was first hypothesized by White et al. (1985) to explain the enrichment of lavas in radiogenic Pb (²⁰⁶Pb/²⁰⁷Pb = 20·16, ²⁰⁷Pb/²⁰⁴Pb = 15·85) specifically in the southern part of the arc. These proportions were re-evaluated to vary from 2% in the north to 15% in the south of the arc by Turner et al. (1996). Van Soest et al. (1998, 2002) also proposed that low ³He/⁴He ratios (3· 6–7· 6 R_A) in geothermal fluids and olivine crystals, and high ¹⁸O values (4·74–5·76) in olivines associated with radiogenic ⁸⁷Sr/⁸⁶Sr ratios (0·703970–0·705463; Davidson, 1985) in the lavas reflected the imprint of terrigenous sediments on the mantle wedge beneath the Martinique–Grenada section of the arc. The high ¹⁸⁷Os/¹⁸⁸Os ratios (0·134–0·164) of picritic lavas from Grenada were also interpreted as a sediment or slab-derived fluid signature (Woodland et al., 2002).

Direct evidence of multi-stage metasomatism events in the mantle wedge and peridotite–melt interactions is found in peridotite xenoliths exhumed by alkali basalts from Grenada, where harzburgites (residue of 22% fractional melting) are progressively transformed into lherzolites and ultimately into wehrlites (Parkinson et al., 2003). Further remobilization and interaction with ascending alkali basaltic melts are also evidenced in these xenoliths (Vannucci et al., 2007).

Recent studies of St. Vincent HMB have demonstrated that they were generated by around 15% partial melting of a MORB-type mantle source at 17 kbar (50–60 km), 1130°C, and an oxygen fugacity more oxidizing than the quartz–fayalite–magnetite solid buffer (FMQ + 1; Heath et al., 1998; Pichavant & Macdonald, 2003, 2007). This mantle source has been modified by the addition of slab-derived aqueous fluid components, which strongly influence the P–T conditions of mantle partial melting. Parental melts giving rise to the HMB of St. Vincent could have been generated at 1235°C, 11· 5 kbar when they contain 1· 5 wt % H₂O, or 1185°C, 16 kbar with 4·5 wt % H₂O (Pichavant et al., 2002; Pichavant & Macdonald, 2007).

Assessing the chemical and isotopic composition of the primary magmas and their volatile content is challenging because of their scarcity in the geological record. The aim of this study is to determine the initial major and volatile element and isotopic composition of the primary magmas and to identify the slab-derived components by analysing melt inclusions (M.I.) hosted in high-Fo (85–89 mol %) olivines that are thought to be in equilibrium with the most primitive basaltic magmas of the southern Lesser Antilles arc. We focused in this work on olivine-rich lapilli from St. Vincent (Fig. 1), located in the southern part of the arc, where relatively primitive basalts have been erupted as a consequence of fast ascent of magma without long-term storage in the crust (Heath et al., 1998; Macdonald et al., 2000).

The major element compositions of 200 M.I. were determined by electron microprobe. The H₂O, Li, B, Cl, F, S contents and ⁷Li, ¹¹B, ¹⁸O, ³⁴S of 50 selected M.I. were analysed by ion microprobe. The overall dataset allows us to discuss the petrogenetic processes that may have produced the St. Vincent magmas in the framework of the evolution of the Lesser Antilles arc.

	SAMPLE LOCATION AND DESCRIPTION

TOP ABSTRACT INTRODUCTION SAMPLE LOCATION AND DESCRIPTION ANALYTICAL METHODS MINERAL CHEMISTRY MELT INCLUSIONS DISCUSSION CONCLUSIONS: CONSTRAINTS ON THE... SUPPLEMENTARY DATA REFERENCES

 
Systematic sampling of scoria and lapilli deposits was undertaken on St. Vincent to collect suitable basaltic samples for melt inclusion studies and, in addition, some lava flows including HMB (Fig. 1). Pyroclastic deposits were collected: (1) on the west coast: fallout and lapilli deposits located south of Rose Bank, scoria from Jack Hill and Belleisle spatter cones; (2) from scoriae underlying the HMB lava flow at Ribishi point (SE); (3) from the pyroclastic sequence overlying the lava flows forming Jumby Point. Most of the samples, including the basaltic scoriae of the spatter cones (Jack Hill and Belleisle), contain crystallized or very small (< 20 µm) M.I., which were not appropriate for volatile studies, as previously reported at St. Vincent and more broadly in the Lesser Antilles arc (Devine, 1995). One unit of slightly stratified black lapilli deposits located on the west coast near Troumaka (SVN4b) was found to be particularly rich in olivine crystals that contain glassy inclusions. Unfortunately, the bulk-rock sample was not suitable for accurate chemical analysis because of its richness in olivine crystals and the alteration of its matrix. This unit could belong to the Yellow Tuff formation on the description given by Heath et al. (1998), who reported the presence of basaltic lapilli in fall layers of the Yellow Tuff formation in the Rose Bank area. More than 500 olivine crystals from the 0·5–1 mm grain size fraction of different crushed scoriae fragments from this unit (SVN4b) were hand-picked under a binocular microscope and embedded in epoxy. Their morphologies vary from euhedral (olivines with equant faces) to polyhedral, indicating different crystallization histories. About 200 olivine crystals were individually polished to expose their M.I. at the surface. Amongst the M.I. that were analysed for major elements, 50 were selected for secondary ionization mass spectrometry (SIMS) element and isotopic analyses.

	ANALYTICAL METHODS

TOP ABSTRACT INTRODUCTION SAMPLE LOCATION AND DESCRIPTION ANALYTICAL METHODS MINERAL CHEMISTRY MELT INCLUSIONS DISCUSSION CONCLUSIONS: CONSTRAINTS ON THE... SUPPLEMENTARY DATA REFERENCES

 
Electron microprobe analysis (EMPA) of major elements
The major element compositions of olivine-hosted M.I., embayments that have preserved contact with the surrounding melt, host olivines, and spinels trapped in olivines were analysed using an SX50 CAMECA electron microprobe (Camparis-Jussieu, Paris). The analytical conditions were 5–10 µm beam size, 10 nA beam current, an average 15 s (up to 40 s for K₂O) counting time for each major element for M.I. analyses, and a 40 nA focused beam and 200 s counting time for olivine and spinel analyses. Sulphur, chlorine and phosphorus were analysed with a 30 nA beam current and 200 s counting time. EMPA and SIMS (see below) results for S and Cl were cross-checked on 11 inclusions.

The reproducibility of analyses was checked using Alv981-R23 and VG2 basaltic glasses and San Carlos olivine as internal standards (see Electronic Appendix 1, available for downloading at http://www.petrology.oxfordjournals.org). Uncertainties (given for 1) are less than 1% for SiO₂, Al₂O₃ and MgO, 1% for CaO, 2% for FeO, less than 3% for TiO₂ and Na₂O, 5% for K₂O, and 26% for MnO. For each M.I., three points were analysed on average. Olivine compositions were systematically determined near the M.I. (to assess any Fe loss), and at the cores and rims (to check for zoning).

SIMS analyses
The H₂O, Li, B, F, S and Cl contents of the M.I. and their Li, B, O and S isotopic compositions were analysed using the CAMECA IMS 1270 ion probe at CRPG (Nancy, France), with a 10 µm projected beam size for all analyses, using a primary beam accelerating voltage of primary beam of 13 kV for both O^– and Cs⁺ ions and a 10 kV secondary accelerating voltage. For all these measurements, olivines were mounted in gold-coated epoxy rings.

H₂O, Li, B, ⁷Li and ¹¹B analyses were performed with an O^– primary beam. The H₂O, Li and B contents were measured in a single analysis, with a primary beam intensity of 10–15 nA, a mass resolution of 1500 and an energy filtering of –60 V. Intensities of ¹H⁺, ⁷Li⁺, ¹¹B⁺ and ³⁰Si⁺ were measured by peak switching, in monocollection ion counting mode. H, Li and B were normalized to Si, and their ionization yields relative to Si were calibrated by measuring international and laboratory standards (data are listed in Table 1). Major element compositions of the standards are reported in Electronic Appendix 1, and measured and published ratios are reported in Electronic Appendices 2, 3 and 5. For ⁷Li and ¹¹B determinations, ⁶Li, ⁷Li, ¹⁰B and ¹¹B were measured with a 10–15 nA O^– primary beam, at a mass resolution of 1500 without energy filtering, over a period of 18 min (15 cycles). The instrumental isotopic fractionation of lithium and boron was determined by analysing standards with basaltic compositions (Table 1, Electronic Appendices 2, 4 and 6).

View this table: [in this window] [in a new window]   Table 1: Reference values for element and isotope analyses

 
The Cs⁺ primary beam was used for measuring F, Cl and S contents and ³⁴S, ¹⁸O, ¹⁹F^–, ³²S^–, ³⁵Cl^– and ³⁰Si^– were measured together in monocollection mode, during a 15 cycle analysis, at 3000 mass resolution and without energy filtering with a 10 nA primary beam. F, Cl and S were normalized to Si, and their ionization yields relative to Si were determined by measuring international or laboratory standards (Table 1, Electronic Appendices 1, 2 and 5). SIMS and EMPA measurements of sulphur are comparable, suggesting there is no matrix effect using the SIMS Cs ion source, as also proposed by Spilliaert et al. (2006). ³⁴S and ¹⁸O were measured in multicollection mode, with 3000 mass resolution, no energy filtering, a primary beam intensity of 10 nA, and counting time of 6 and 4 min each, respectively. For ³⁴S measurements, ³²S was measured with a Faraday cup (FC) and ³⁴S using an electron multiplier (EM), and the instrumental fractionation was determined on the glasses CY-82-29-3V and CY-82-31-2V (Table 1, Electronic Appendices 1, 4 and 6). For ¹⁸O, ¹⁶O and ¹⁸O were measured by FC, and calibrated against the MORB glasses CY-82-27-1V and CY-82-29-3V (Table 1, Electronic Appendices 1, 4 and 6), which have SiO₂ contents close to those of the melt inclusions (Electronic Appendix 1).

Because of the small size of the melt inclusions, several analyses had to be acquired successively on the same spot position. Therefore, we verified the reliability of multi-element analyses performed on the same spot, using either the O₂ or the Cs source, by duplicated or triplicated volatile and isotopic analyses on the same spot, or on different spots when the inclusion was large enough. Almost half of the inclusions were measured two or three times for H₂O, Li, B, F, S and Cl contents and ⁷Li, ¹¹B, ¹⁸O and ³⁴S, either in the same session, but not consecutively, or in different sessions, separated by several months. No systematic changes were observed; thus we conclude that there was no analytical bias caused by successive measurements on the same spot.

Uncertainties (1) defined using the various standards are as follows: 8% on water, 2% on Li, 4% on B, 7% on Cl, 3% on S, 6% on Cl, and ±1·2 on ⁷Li, ±2·1 on ¹¹B, ±0·16 on ⁸O and ±0·53 on ³⁴S (Electronic Appendices 2–4). The average analytical reproducibilities (mean standard deviation, 1) between sample replicate measurements were 15% on H₂O, 8 and 16% on Li and B, respectively, and 18, 12 and 14% on F, S and Cl, respectively. The sample reproducibilities for isotopic analyses are: ±1· 3 for ⁷Li, ±2·2 for ¹¹B, ±1· 25 for ¹⁸O and ±0·71 for ³⁴S.

Isotopic compositions are reported relative to reference standard values, as

where the standard is L-SVEC (⁷Li/⁶Li = 12·175). The same calculation is applied for the other isotopic compositions: ¹¹B is referenced to NBS 951 ¹¹B/¹⁰B ratio (4·044); ¹⁸O to the SMOW ¹⁸O/¹⁶O ratio (2005·2 x 10^–6) and ³⁴S to the Canyon Diablo ³⁴S/³²S ratio (4·43 x 10^–2).

	MINERAL CHEMISTRY

TOP ABSTRACT INTRODUCTION SAMPLE LOCATION AND DESCRIPTION ANALYTICAL METHODS MINERAL CHEMISTRY MELT INCLUSIONS DISCUSSION CONCLUSIONS: CONSTRAINTS ON THE... SUPPLEMENTARY DATA REFERENCES

 
Olivines show a large range of chemical variation from Fo₉₀ to Fo₇₂ (Fig. 2a). Of the more than 200 olivines analysed, a large proportion is rich in MgO and normally zoned, with Fo_89–85 cores and Fo_85–77 rims. Only rare Fe-rich olivines show reverse zoning. These observations indicate insignificant olivine re-equilibration and rapid transfer from their crystallization site to the surface. Their NiO and CaO contents vary from 0·07 to 0·34 wt % and from 0·09 to 0·28 wt %, respectively (Fig. 2b). As a whole, NiO decreases with the Fo mol % of olivine, whereas statistically 60% of the olivines have a CaO content between 0·15 and 0·20 wt % without a clear relationship between their Fo and CaO contents. Very few olivines have both a high CaO and Fo content (e.g. 0·26 wt % CaO in Fo_88·6; Electronic Appendix 7). Olivines with CaO <0·06 wt %, typical of mantle olivine (Parkinson et al., 2003), have not been found in our samples, but were observed in a few HMB lavas of St. Vincent (M. Pichavant, personal communication). In contrast, they can be common in primitive basalts from the Lesser Antilles arc; for example, in Grenada and Martinique (Parkinson et al., 2003, and unpublished data).

View larger version (30K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 2. (a) Compositional variation of St. Vincent olivines (Fo mol %). Most of the olivines have magnesian compositions, with Fo > 86. (b) Variation of CaO vs NiO (in wt %) in olivines. The grey field represents mantle olivine compositions (Parkinson et al., 2003). For comparison, data for olivines from ‘picritic’ lavas and lapilli deposits from Ilet à Ramier in Martinique are also plotted (authors’ unpublished data).

 
The olivines contain numerous Cr-spinel inclusions, of <10 to 20 µm in size, with a Cr number [Cr/(Cr + Al)] varying from 0·01 to 0·63 (Table 2; Figs 3 and 4a). They overlap the field reported for Cr-spinels in St. Vincent HMB (Heath et al., 1998; Pichavant & Macdonald, 2007). Amongst the analysed spinels (31), one-third are associated with melt inclusions and possibly favoured their trapping. These latter do not show significant difference from spinel isolated in the olivine except a few (four) grains that exhibit significant zoning with Cr number decreasing from 0·45 to 0·2 (Fig. 3) and Al₂O₃ and MgO contents increasing to 60 wt % and 20 wt %, respectively, in the late growth rim (Table 2; Fig. 4d). Al-rich spinels were previously reported in some basanitoids of Grenada (Arculus, 1974, 1978). Al-rich spinels and Al-rich rims surrounding Cr-rich core spinels have also been found in some arc basalts and interpreted as the crystallization products of anomalously Al-rich melts derived from the melting of amphibole-rich cumulates, probably in a sub-arc magma chamber (Della-Pasqua et al., 1995). However, we emphasize that the Al-rich spinel rims are found at the contact with M.I. and might simply testify to low-temperature growth at the contact with an Al-rich melt. Indeed, St. Vincent M.I. are all rich in Al₂O₃ with concentrations varying from 14·6 to 23·2 wt %, a range similar of that of the St. Vincent basalts (14·2–21· 5 wt %, Table 4 and unpublished data; Heath et al., 1998). Moreover, Al, Mg-rich spinels have been experimentally reproduced at temperatures of 1160–1180°C and pressures between 9·5 and 14·5 kbar in St. Vincent HMB (Pichavant et al., 2002; Pichavant & Macdonald, 2007), suggesting that these Al-rich spinel could crystallize from an Al-rich melt at upper mantle pressures.

View larger version (25K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 3. Variation of Cr-number [100 x Cr/(Cr + Al)] vs Mg-number [100 x Mg/(Mg + Fe2+)] in spinels hosted in olivine. Literature data are from [1] Pichavant & Macdonald (2007), [2] Pichavant et al., 2002 for experimental spinels, and [3] Heath et al. (1998) for spinels from St. Vincent post-caldera lavas.

 

View larger version (79K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 4. Photomicrographs of M.I. in St. Vincent olivines. (a) M.I. randomly distributed in the host crystal. Chromites are either associated with inclusions or isolated in olivine. (b) Isolated, rounded M.I. with a bubble and associated with a spinel. (c) Elongated inclusion with a small zoned spinel. (d) Detail of a spinel at the contact of a melt inclusion showing strong zoning on the micrometre scale with Cr number varying from 0·46 (core) to 0·03 (rim, sample 164, Table 2).

 

View this table: [in this window] [in a new window]   Table 2: Selected compositions of spinel–olivine pairs and of spinel associated with melt inclusions

 

	MELT INCLUSIONS

TOP ABSTRACT INTRODUCTION SAMPLE LOCATION AND DESCRIPTION ANALYTICAL METHODS MINERAL CHEMISTRY MELT INCLUSIONS DISCUSSION CONCLUSIONS: CONSTRAINTS ON THE... SUPPLEMENTARY DATA REFERENCES

 
Major element chemistry
The M.I. in the analysed sample (SVN4b) are randomly distributed in polyhedral olivine phenocrysts whose growth-rate led to trapping of a melt fraction inferred to be representative of the host magma (Faure & Schiano, 2005). They are free from cracks or other evidence of leakage. They are preserved as glass and range in size from 40 to 150 µm. As a whole they are ovoid or elongated, with a shrinkage bubble (Fig. 4b and c), and often associated with spinel. No sulphide globules were observed. Some embayments, representing open systems (Fig. 4a) and often containing microlites, were analysed to assess the composition of more evolved melts.

We do not observe any relationship between major element composition and the size or shape of the inclusions. However, the M.I. have undergone variable extents of olivine post-entrapment crystallization. We calculate the K_D [(FeO/MgO)_ol/(FeO/MgO)_melt] for each melt inclusion, using an Fe₂O₃/FeO melt ratio of 0·4. This ratio was calculated using the spinel compositions (Maurel & Maurel, 1982) and the Kress & Carmichael (1991) equation for fO₂ at NNO + 1·4 (NNO = [log₁₀fO_2sample – log₁₀ f O_2(NNO)]) that was experimentally determined for St. Vincent HMB (Pichavant et al., 2002; Pichavant & Macdonald, 2003). The calculated K_D of the inclusions varies from 0·4 to 0·1. This range is much larger than the average value of 0·32 ± 0·03 obtained for St. Vincent HMB (Pichavant et al., 2002) and the theoretical value of 0·31 ± 0·01 calculated for the M.I. in this study using the K_D model of Toplis (2005). The lowest K_D values reflect post-entrapment crystallization on the inclusion walls. In addition, some inclusions, essentially those trapped in olivine Fo_>86, may have also suffered Fe loss, resulting in high K_D (>0·32) and lower FeO contents (7·1 wt % on average) than the HMB lava from Black Point (9·5 wt %, Table 3), inferred to be representative of the primary magmas (Pichavant et al., 2002). Because some of the M.I. that have not suffered Fe loss have FeO contents as high as those measured in the HMB, we have used this value for the M.I. that have experienced Fe loss. Post-trapping olivine crystallization (PEC%) was evaluated using the Petrolog software (Danyushevsky et al., 2000, 2002). This correction leads to an uncertainty of 5% on element abundances and affects the Fe/Mg ratio of the inclusions but not the ratios between other elements. The percentage of PEC ranges from 1· 7 to 11· 1% (Table 3). The data for M.I. discussed hereafter are corrected for PEC.

View this table: [in this window] [in a new window]   Table 3: Selected PEC-corrected compositions of St. Vincent melt inclusions analysed by SIMS 1270

 

View this table: [in this window] [in a new window]   Table 4: Compositions of bulk HMB lava samples from St. Vincent

 
Nearly 200 M.I. were analysed in olivines (Fo_89·9 to Fo_<83) for their major element contents using the electron microprobe. All of them display a wide compositional spectrum, with SiO₂ varying from 40·2 to 55·9 wt %, and K₂O from 0·25 to 0·85 wt %. In a plot of K₂O vs SiO₂ (Fig. 5a) two populations can be identified. In the first population, K₂O and SiO₂ are positively correlated. These M.I. are less evolved than those previously analysed in Yellow Tuff samples (Heath et al., 1998). More specifically, the M.I. trapped in Fo_85–89 closely resemble the whole-rock basaltic samples from St. Vincent (with SiO₂ = 47·7–46·2 wt % and K₂O = 0·34–0·30 wt %; Fig. 5a). They display similar Na₂O/K₂O (0·11) and range in CaO/Al₂O₃ (0·58–0·76) (Fig. 5b and c) and MgO contents (6·3 to 9·1 wt %) (Fig. 5d). The most evolved melt inclusions (SiO₂ > 50 wt %, K₂O >0·5 wt %, MgO <4 wt %; Fig. 5a and d) are those trapped in Fe-rich olivines (Fo <83) and embayments that remained in contact with the surrounding matrix. Increasing SiO₂ and K₂O contents are correlated with decreasing Na₂O contents (Fig. 5b), reflecting late-stage melt differentiation.

View larger version (30K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 5. Variations of K2O with SiO2 (a) and CaO/Al2O3 (c), and variations of Na2O (b) and MgO (d) with K2O wt % in olivine-hosted M.I. The M.I. compositions (filled and open squares represent ‘normal’ and CaO-rich M.I., respectively) are corrected for post-entrapment crystallization and normalized to 100%. The evolved M.I. are melts trapped in Fo<83 and glass embayments (asterisks). For comparison, the compositions of the St. Vincent HMB and basalt whole-rocks (W.R.) are also plotted (this study; Heath et al., 1998) and of the M.I. (not corrected for PEC) in pyroxenes and plagioclases of the pyroclastic fall Yellow Tuff deposits, dated between 3600 and 4500 years BP (crosses, [1], Heath et al., 1998).

 
A second population departs from this general evolution trend in having low SiO₂ (43·7–45·9 wt %) contents, weakly negatively correlated with K₂O (Fig. 5a). Their CaO content reaches 14·1–16·9 wt % with CaO/Al₂O₃ > 0·80 (Fig. 5c). These Ca-rich M.I., mostly trapped in olivine Fo_{87· 0–89·9}, display the same range of MgO contents as the ‘normal’ basaltic M.I. (Fig. 5d). They are nepheline-normative, as commonly reported for M.I. in arc settings (Schiano et al., 2000), unlike M.I. from MORB and ocean island basalt (OIB), which are richer in SiO₂, hypersthene-normative, and inferred to be trapped at higher pressures (Médard et al., 2004). No clear correlation was observed between the M.I. CaO content and that of their host. Combination of careful optical observations and the major element compositions of the M.I. does not reveal a relationship between the melt CaO content or its CaO/Al₂O₃ and the presence of spinel, the late growth of which could have affected their composition after trapping. As a whole, the EMPA totals (including volatile contents) are between 97 and 99 wt % with no clear relationship with the SiO₂ content. We thus consider these low SiO₂ contents to be reliable, representative of Ne-normative, Ca-rich melts. Totals (major elements + volatile contents) as low as 97 wt % could result from a combination of the sample preparation technique and the analytical uncertainties (see Analytical Methods section).

SIMS analysis
We have carefully selected 49 M.I. and entrapped glasses for SIMS analysis of H₂O, B, Li, Cl, F, S and ⁷Li, ¹¹B, ¹⁸O, ³⁴S on the basis of their major element compositions. Representative analyses are reported in Table 3 and the complete dataset of SIMS analyses is provided as supplementary data (Electronic Appendix 7).

Most M.I. (41) hosted in olivine Fo_85–89 were considered to be representative of basalts, having K₂O contents of 0·25–0·46 wt % (Fig. 5a), CaO/Al₂O₃ ratios between 0·65 and <0·80 (Fig. 5c) and MgO between 6·3 and 9·1 wt % (Fig. 5d). In addition, six M.I. with high CaO/Al₂O₃ and three evolved glass embayments (K₂O > 0·5 wt %) representative of late-stage magma evolution were analysed by SIMS for comparison.

Sulphur, chlorine and fluorine
The chlorine content of the M.I. varies from 650 to 1855 ppm, a range that encompasses that of subduction-related basalts (500–2000 ppm; Wallace, 2005). In a plot of Cl vs K₂O (Fig. 6a), the majority of the primitive basaltic M.I. have high Cl contents and show selective Cl enrichment, possibly tracking the influence of contaminants with different Cl/K ratios, as discussed below. Only few of these M.I. plot on a trend defined by the evolved M.I. and the embayments, which is extended by the trend related to late-stage magma evolution observed in the most evolved M.I. from St. Vincent (Heath et al., 1998).

View larger version (26K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 6. Variations of chlorine (a), sulphur (b), fluorine (c), and H2O (d) vs K2O wt % in St. Vincent M.I. In each diagram the average MORB composition is shown (Chaussidon et al., 1991; Straub & Layne, 2003; Wallace, 2005). For F and H2O, evolved M.I. refer only to glass embayments selected for SIMS analysis to be representative of evolved composition (see text).

 
Sulphur abundances vary between 98 and 1653 ppm. The basaltic M.I. exhibit a large range of variation in their S contents from 610 to 1653 ppm (average 831 ppm), with no clear correlation between S and K₂O (Fig. 6b). The decrease of S with increasing K₂O in evolved M.I. and embayments tracks late-stage degassing.

Fluorine contents vary between 79 and 420 ppm, again with a large range of concentrations in the basaltic M.I. On average, F contents (230 ppm) are similar to those of MORB, although some F enrichment is accompanied by a slight decrease in K₂O (Fig. 6c). Fluorine and K₂O contents (F from 304 to 424 ppm) are generally positively correlated in the evolved M.I.

The CaO-rich inclusions (CaO/Al₂O₃ > 0·8) do not differ from the typical basaltic inclusions, having similar ranges in S (468–1033 ppm), F (79–245 ppm) and Cl (1087–1217 ppm) contents (Table 3, Fig. 6).

Water
Water contents range from 0·8 to 5·2 wt % (Fig. 6d). The inclusions hosted in Fo_{86·0–89·9} olivine show a broad range of water contents between 1·3 and 3·8 wt %. However, inclusions with the lowest water contents have a large vapour bubble and could have partly lost their initial water. Interestingly, the average water content varies from 3·2 wt % ± 0·7 (1) to 2· 4 wt % ± 0·6 (1) in M.I. trapped in Fo_>88 and Fo_86–88, respectively. The M.I. with the highest CaO/Al₂O₃ (>0·80) contain between 1· 5 and 2· 0 wt % of H₂O and cannot be distinguished from the other M.I. (Fig. 6d). No clear relationship between H₂O and K₂O or any other element has been observed.

As discussed above, inclusions in olivine Fo_>86 could have undergone some Fe loss. Consequently, we cannot exclude the possibility that some water has also been lost, as H⁺ diffuses rapidly in olivine (Mackwell & Kohlstedt, 1990). However, we have verified that there is no correlation between the post-entrapment evolution of the inclusions, water concentration and the extent of Fe loss.

We also note that M.I. with relatively high S and Cl (up to 1536 and 1855 ppm, respectively) do not have high water contents. In fact, Cl, S and H₂O concentrations do not co-vary. Similar conclusions can be drawn from literature data for other arcs where Cl concentrations of 1000 ppm are found in M.I. with 0·3–6·0 wt % H₂O (Wallace, 2005), although these inclusions represent magmas produced by variable degrees of differentiation. Thus, we suggest here that the variation in water contents in the primitive basaltic M.I. of St. Vincent is not related to degassing, but rather to variable H₂O enrichment of the original magmas.

The evolved M.I. and embayments (K₂O >0·5%) have the lowest water contents (down to 0·8 wt %) for higher K₂O, in agreement with late-stage water degassing, as also shown for sulphur.

Boron and lithium
Boron contents vary widely from 8 to 53 ppm (with one value down to 3 ppm). In Fig. 7a, a B–K trend (B/K = 2·6 x 10^–3) is defined by M.I. (6·3 wt % < MgO < 9·5 wt %) for which B contents vary from 8 to 18 ppm. This variability cannot be explained by crystal fractionation, as these inclusions exhibit similar degrees of differentiation, but instead must track heterogeneities in the magmas themselves. Above this baseline, a number of inclusions have relatively constant K₂O (0·3–0·4 wt %), but variable B contents (10–37 ppm B; one value at 53 ppm) suggesting selective B enrichment. Actually, the most primitive basaltic M.I. (MgO > 8·5 wt %) in olivine Fo_>88 have between 8 and 37 ppm of B. No clear relationship was observed between Cl and B variations. The evolved glass embayments have B contents ranging from 13 to 21 ppm.

View larger version (22K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 7. Variation of B (a) and Li (b) vs K2O wt % (normalized to 100%) in St. Vincent M.I. Embayments refer here only to glass embayments selected for SIMS analysis to be representative of evolved compositions.

 
Lithium concentrations range from 2·6 to 9·2 ppm in basaltic M.I. (average 4·5 ppm). Lithium tends to increase with K content (Fig. 7b). Whether or not two distinct trends exist is not clear. Glass embayments have relatively high Li contents (5·9–11· 4 ppm), probably caused by crystal fractionation as a result of the moderate incompatibility of Li.

The contrasting behaviour of B and Li is illustrated in Fig. 8a: a large number of M.I. show a narrow range in B concentrations (3–18 ppm) and a slightly larger range of Li contents (between 3 and 9 ppm) compared with the B-enriched M.I. In contrast, some M.I. display selective boron enrichment (up to 53 ppm) for little variation in Li concentration (3–5 ppm). The evolved embayments display Li enrichment with nearly constant B contents.

View larger version (25K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 8. Variation of B (a) and 7Li (b) vs Li. In each diagram, values for the different reservoirs for boron and lithium are shown: MORB (Bouman et al., 2004; Elliott et al., 2004), altered oceanic crust (AOC; boron data are from Spivack & Edmond, 1987; Chaussidon & Libourel, 1993; Chaussidon & Jambon, 1994; Li data from Bouman et al., 2004) and Lesser Antilles sediments [Li contents of ODP Leg 144-207 from Carpentier, 2007; 7Li of ODP site 27&543 from Bouman et al., 2004; B data from CRPG-SARM analysis (this study) and Leeman & Sisson, 1996]. Embayments refer here only to glass embayments selected for SIMS analysis to be representative of evolved compositions. Other symbols as in Fig. 5.

 
Isotopic compositions
The ⁷Li isotopic compositions vary considerably from values close to MORB [+7 for M.I., and +5 for MORB according to Bouman et al. (2004)] to strongly negative values (–9·6) for a restricted range of Li contents (Fig. 8b). Strong negative ⁷Li is inconsistent with diffusion of lithium from the inclusion towards its host. In fact, the more rapid diffusion of ⁶Li compared with that of ⁷Li (Lundstrom et al., 2005; Jeffcoate et al., 2007) would have resulted in heavier ⁷Li values in the inclusions compared with the initial MORB-like magma (+5, Bouman et al., 2004). Moreover, Li concentration and isotope profiles were measured in some olivines to test for possible Li diffusive loss from the inclusions (Electronic Appendix 8). These profiles, made in two olivines hosting M.I. (SVN142 and SVN108), have ⁷Li = –8·5 and –8·9, respectively. They show neither significant Li enrichment nor Li isotopic variation in olivine near M.I. We thus assume that ⁷Li values measured in the melt inclusions are representative of that of the trapped melt. Major fractionation observed between core and rim of olivine is probably due to Li diffusion during cooling (Beck et al., 2006). Glass embayments have variable Li isotopic composition (–8·2 to +2·3).

¹¹B values range from –25·6 to +11· 8 (average 0·9 for primitive M.I.). Inclusions in which boron contents (3·4–18 ppm) are positively correlated with K and Li (Figs 7a and 8a) show a large range of ¹¹B values from –16 to +12. In contrast, inclusions selectively enriched in B have heavier isotopic compositions and a more limited range of ¹¹B values from +3 to +11, the significance of which is discussed below.

Oxygen isotopic compositions of primitive basaltic M.I. vary from +3·2 to +10·1 (average +6·1). Glass embayments show a positive correlation between ¹⁸O and TiO₂. The latter can be related to crystallization of olivine ± pyroxene according to Eiler et al. (2000) and is consistent with fractional crystallization identified on the basis of major element compositions.

The sulphur isotopic compositions of the M.I. range from –9·0 to +7· 0 (on average, ³⁴S = +1· 1 for primitive basaltic M.I., Table 3). Glass embayments have more negative ³⁴S compositions, but the two measurements reported are not necessarily representative.

CaO-rich M.I. do not display distinctive isotopic compositions.

	DISCUSSION

TOP ABSTRACT INTRODUCTION SAMPLE LOCATION AND DESCRIPTION ANALYTICAL METHODS MINERAL CHEMISTRY MELT INCLUSIONS DISCUSSION CONCLUSIONS: CONSTRAINTS ON THE... SUPPLEMENTARY DATA REFERENCES

 
Most of the olivines have preserved their original zoning, suggesting direct transfer to the surface in the host magma. The observed chemical and isotopic variations of the M.I. are larger than analytical error and only partially explained by shallow-level differentiation processes. Instead, consistent with the relatively primitive nature of both the inclusions and their host olivine crystals, and with the chemical and isotopic heterogeneities previously observed for St. Vincent bulk lavas (Heath et al., 1998), the M.I. are considered to represent heterogeneities in the primary magmas. Their chemical and isotopic compositions suggest the involvement of melts from different sources, each with their own specific signatures.

Modified N-MORB source
The melt inclusions analysed in this study generally have major element compositions close to those of the St. Vincent basalts (see Results section), thought to be derived from modified MORB-source mantle (Heath et al., 1998; Pichavant et al., 2002; Pichavant & Macdonald, 2003, 2007).

Most of the analysed M.I. have F [230 ppm ± 80 (1)] and Li [4 ppm ± 1 (1)] contents and Li isotopic compositions (⁷Li from 2 to 6) in the range of MORB values (F = 200–300 ppm, Straub & Layne, 2003; Li 5 ppm and ⁷Li 4 ± 2, Chan et al., 1992; Elliott et al., 2004). Our data therefore support the idea that the source of the basaltic magmas of St. Vincent is predominantly similar to MORB-source mantle as previously proposed (Heath et al., 1998; Pichavant et al., 2002; Pichavant & Macdonald, 2003). In agreement with previous data for whole-rocks from the Lesser Antilles arc (Macdonald et al., 2000), our M.I. data show that there is a contribution from slab fluid components as described below. The first indication is the higher K₂O contents of the M.I. (0·25–0·50 wt %) compared with typical MORB. Indeed, Lassiter et al. (2002) suggested that, with a maximum of 50 wt % of the K₂O extracted from the subducted crust during dehydration, a corresponding input of 0·20–0·25 wt % K₂O to the mantle source is possible. Moreover, dehydration and/or melting of Lesser Antilles arc subducted sediments with high K₂O [from 0·98 to 1· 6 wt %, Ocean Drilling Program (ODP) Leg 144, and Barbados; Carpentier, 2007] will produce fluids or melts rich in potassium. Other elements, such as Cl and water, should also provide clues for slab contributions. The relatively high chlorine content and Cl/H₂O ratio of the M.I. are generally enhanced by differential fractionation of Cl and H₂O during fluid expulsion from the slab and/or migration through the mantle wedge prior to melting as proposed by Kent & Elliott (2002). The variation of the Cl/H₂O ratio (from 0·01 to 0·17) is consistent with contributions of between 2 and 15% equivalent NaCl following Wallace (2005). The nature of these slab contributions is discussed below on the basis of the chemical and isotopic SIMS analyses of the M.I.

Seawater-like signature
St. Vincent basaltic M.I. are all enriched in H₂O, Cl and B compared with N-MORB by a factor from 5 to >150 (Fig. 9a–c), even the most primitive M.I. (MgO >8·5 wt %). The latter contain 3·2 wt % H₂O, 820–1402 ppm Cl and 8–37 ppm B, implying the interaction of aqueous fluids with the mantle wedge.

View larger version (20K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 9. Histograms of H2O content in wt % (a) and chlorine (b) and boron (c) contents in ppm where M.I. representative of St. Vincent basalts (0·65 < CaO/Al2O3 0·80) are distinguished from CaO-rich inclusions (CaO/Al2O3 > 0·8). MORB data for H2O from Baker et al. (1994) and Sobolev & Chaussidon (1996); for Cl from Lassiter et al. (2002), and references therein; for B from Straub & Layne (2002).

 
A large proportion of the M.I. have B concentrations and ¹¹B (+10) higher than N-MORB (0·4 ppm and –5; Straub & Layne, 2002). This is illustrated in Fig. 10, which is an interpretative plot of the isotopic compositions and concentrations of boron in the M.I. This diagram, modified from Rose et al. (2001), identifies the different slab-derived components able to change the composition of a MORB-type mantle source. The overall influence of seawater-like fluids (B 4·4 ppm; ¹¹B +40 for seawater; Straub & Layne, 2002) is demonstrated by the high ¹¹B and B enrichment of the M.I., implying that the source rock for these melts is enriched in both B and ¹¹B. We calculate a mixing curve between a depleted MORB mantle (DMM) with B from 0·05 to 0·3 ppm and ¹¹B from –6·5 to –1·5 (Chaussidon & Libourel, 1993; Chaussidon & Jambon, 1994) and a seawater-like fluid. A large proportion of the B and ¹¹B compositions of the basaltic M.I. could be explained by the addition of 18–40% of the B in the mantle by seawater-like fluids and 10–20% partial melting of this modified mantle source. For these calculations we assumed batch melting, ¹¹B unchanged during melting (Rose et al., 2001) and a melt/rock partition coefficient of 0·015 (You et al., 1996; Rose et al., 2001).

View larger version (28K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 10. Interpretative diagram modified after Rose et al. (2001) showing the evolution of B vs 11B in St. Vincent M.I. The MORB-type mantle source (Chaussidon & Libourel, 1993; Chaussidon & Jambon, 1994) is mixed with seawater-like fluids (4·4 ppm B, 11B = +40; Straub & Layne, 2002). The percentage of melting of the modified mantle is calculated assuming batch melting and using a melt/rock partition coefficient of 0·015 (You et al., 1996). The content and isotopic composition of boron in altered oceanic crust (AOC) and sediments after variable extents of dehydration and in their extracted fluids are calculated using a Rayleigh distillation model assuming DBfluid/solid = 50 (Rose et al., 2001) and isotopic fractionation between fluid and solid f–s = +5. Theoretical trajectories for dehydrated AOC and sediments are represented by the dashed lines 1 and 2, respectively, and those for the fluids produced by variable dehydration (expressed in per cent) by the continuous lines 3 and 4, respectively. The initial boron contents (11–65 ppm) of the Lesser Antilles terrigenous sediments (field SED) of the ODP Leg 144 silica-rich units were analysed at SARM CRPG. Their Li isotopic compositions of –10 are from Smith et al. (1997) and are closely comparable with those used by Rose et al. (2001). The initial AOC composition (AOC) is from Spivack & Edmond (1987), Chaussidon & Libourel (1993) and Chaussidon & Jambon (1994). For comparison, seawater (Straub & Layne, 2002) and carbonate compositions (B element analyses from SARM-CRPG; this study) and B isotopic compositions from Deyhle et al. (2001) are also shown. Mixing curves between MORB-source mantle and seawater indicates that a maximum of 40% of the initial B derives from seawater. Mixing curves between the modified mantle melt and fluids resulting from dehydration of AOC and sediments show that the M.I. record a contribution of up to 40% of fluids issued from 4·5% AOC dehydration and up to 45% of fluids resulting from 6% dehydration of terrigenous sediments. It should be noted that all St. Vincent M.I. are enriched in boron with respect to MORB (see text for details). Embayments refer here only to glass embayments selected for SIMS analysis to be representative of evolved compositions.

 
The input of seawater-like fluids into the mantle source is qualitatively in agreement with the Cl enrichment and the heaviest ³⁴S of most of the M.I. (2·0 in M.I., seawater: +20, Taylor & Sheppard, 1986) compared with typical N-MORB compositions (0, de Hoog et al., 2001; Fig. 11b). Because of the high fluid/rock partition coefficients of B and Cl (e.g. D_B^fluid/solid = 50; You et al., 1996), this fluid should be rich in these two elements. Consequently, the amount of fluid required for the observed enrichments should not be unusually large: an input of 18–40% of B in the source (0·7–1· 6 ppm B added) corresponds to 0·3–0·7% seawater-derived fluid, using D_B^fluid/solid = 50. Moreover, after Lassiter et al. (2002), assimilation of 0·3% of a 50% NaCl brine could produce 1000 ppm excess Cl. The limited shift toward low ¹⁸O values in the M.I. also suggests addition of a relatively small amount of this fluid. This idea is reinforced by the lithium isotope data: the M.I. do not show specific enrichment in ⁷Li compared with MORB (Fig. 8), although seawater has much higher ⁷Li (32, Chan et al., 1992). The negligible effect of the fluid on the ⁷Li of the magma is accounted for by the low Li content of seawater (0·18 ppm).

View larger version (26K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 11. Evolution of boron (a) and 34S (b) vs 18O in St. Vincent M.I. The main reservoirs for boron, 34S and 18O are indicated in the two plots: notably sediments (terrigenous and carbonates, B contents from CRPG-SARM analysis, this study; 18O from Eiler et al., 2000; 34S from de Hoog et al., 2001); the theoretical composition of AOC-derived fluids (Leeman & Sisson, 1996; Alt & Shanks, 2006); MORB and arc basalts (18O from Eiler et al., 2000; Straub & Layne, 2002; 34S from de Hoog et al., 2001). Negative 34S could be explained by a component derived from subducted black shales (see text). Embayments refer here only to glass embayments selected for SIMS analysis to be representative of evolved compositions.

 
We thus argue that dehydration aqueous fluids with a low solute content and a seawater-like signature, probably released from the slab at temperatures as low as 650°C (Hermann et al., 2006), are responsible for a first-stage contamination of the MORB mantle source of the St. Vincent basalts. This initial contamination is recorded by all the M.I. that were analysed, independently of the more specific signatures discussed below. This interpretation agrees with the high ¹¹B in fluids released at the onset of the slab-dehydration process relative to MORB values (Marschall et al., 2007). Following this line of reasoning, serpentinites fluxed by fluids could be dragged down to the sub-arc region and release B-rich, high-¹¹B metasomatizing fluids at greater depth when they exceed their stability limit.

Altered oceanic crust derived fluids
The Cl enrichments illustrated in Fig. 4a strongly suggest a contribution of other fluids in addition to the aqueous fluid with a seawater-like isotopic signature. We investigate the nature of these fluids by combining chemical data with specific isotopic signatures of the M.I. In particular, the B content and its isotopic composition can distinguish between the influence of sediments and that of altered oceanic crust (AOC). Boron is highly fractionated in aqueous fluids, leaving a B-depleted solid residue with light ¹¹B values (Bebout et al., 1993; Ryan et al., 1995; Rose et al., 2001) as illustrated in Fig. 10 (dashed arrows 1 and 2). The selective B enrichment recorded by some St. Vincent M.I. (¹¹B between +2· 8 and +11· 8) is consistent with the addition of fluids derived from dehydration of AOC to a modified MORB source already enriched in B and ¹¹B. The B content and isotopic signature of fluids derived from AOC dehydration (curve 4 in Fig. 10) have been calculated on the basis of a Rayleigh model, with D_B^solid/fluid = 50, an isotopic fractionation factor between fluid and solid (_f–s) of +5 (Rose et al., 2001), an initial B content in AOC varying from 17 to 40 ppm and ¹¹B from 6 to 10 (Spivack & Edmond, 1987; Chaussidon & Libourel, 1993; Chaussidon & Jambon, 1994). M.I. with selective B enrichment plot on a mixing curve between two end-members identified as the modified mantle melt and fluids from AOC dehydration. Based on this model and the M.I. data, we estimate that up to 40% of a fluid end-member with 130 ppm B and light ¹¹B (–8·5), implying 4·5% dehydration of the AOC, is necessary to account for the compositions of the B-enriched M.I. The contribution of fluids derived from AOC dehydration is in agreement with the ¹⁸O (3–7) values of these M.I. that are selectively enriched in boron (Fig. 11a). For example, the ¹⁸O of a fluid extracted from AOC at between 200 and 300°C could be between 2 and 8, assuming equilibrium with a metabasalt having ¹⁸O = 8–11 and an equal amount of chlorite and albite (Alt & Shanks, 2006).

The light ⁷Li compositions (down to –10) of M.I. selectively enriched in boron also support the idea of AOC-derived fluid input, assuming no Li diffusion from the M.I. towards the host olivine and no ⁷Li fractionation (see Analytical Methods section). Indeed, it is thought that ⁷Li preferentially partitions into aqueous fluids relative to ⁶Li during mineral–fluid or silicate melt–fluid interactions (Peacock & Hervig, 1999; Pistinier & Henderson, 2003). As a result, the Li isotopic composition of dehydrated rock residues becomes lighter as dehydration progresses. Dehydration fluids would also become lighter.

Zack et al. (2003) suggested that light ⁷Li values (down to –11) measured in eclogites were produced by isotope fractionation through Rayleigh distillation during dehydration of clays and/or chlorite at an early stage of metamorphism. In this case, dehydration fluids issued from the eclogites (e.g. dehydrated oceanic crust) could have ⁷Li –11, depending on the equilibrium of the expelled fluids with the rocks, the chosen values for D_Li^rock/fluids and the fractionation factor (Marschall et al., 2007). In contrast, Marschall et al. (2007) proposed that dehydration of the subducted slab accounts for only –3 of the ⁷Li variation of the subducted AOC composition, based on a Rayleigh model with variable D_Li^rock/fluid depending on the P–T path and mineral composition of the dehydrated rocks (AOC: ⁷Li from 0 to 14, Marschall et al., 2007, and references therein). They proposed, contrary to Zack et al. (2003), that the light Li isotopic compositions measured in eclogites are due to kinetic fractionation with the surrounding sediment pile. In this case, assuming fast ⁶Li diffusion and no interaction with other slab layers, fluids issued from the dehydration of the altered oceanic crust would not have ⁷Li < –3.

In this study the M.I. do not display very light ⁷Li with high Li contents: Li contents are relatively constant, or slightly decrease with decreasing ⁷Li. The negative ⁷Li, measured in M.I. (down to –10) could be ascribed to the contribution of fluid affected by kinetic fractionation during slab dehydration and fluid extraction and/or when dehydration fluids from AOC fluxed the sediments pile (Richter et al., 2006). In fact, the input of fluids derived from AOC dehydration as a source of the negative ⁷Li of the M.I. is reinforced by their selective enrichment in boron (B up to 53 ppm, Fig. 11), and their high ¹¹B (from +2·8 to +11· 8) and ¹⁸O (6) compositions. Moreover, we emphasize here that such light ⁷Li compositions have not been reported for arc lavas (⁷Li from +3 to +7, Elliott et al., 2004; Ryan & Kyle, 2004). We thus propose that the M.I. record initial melt heterogeneities that are lost in the bulk magmas (lavas) in which Li can be easily homogenized because of its high diffusivity (Lundstrom et al., 2000).

In summary, the selective B enrichment recorded in a small number of M.I., as well as their high Cl contents, their light ⁷Li and their ¹⁸O (3–7) signatures are readily explained by the addition of fluids derived from AOC dehydration to an already modified MORB-type mantle source. This fluid had a medium solute content, evident from its higher B and Cl contents compared with the aqueous fluids responsible for the first stage of mantle wedge metasomatism. We thus suggest that this AOC-fluid component is most probably released later (i.e. at greater depth) than the fluids with a seawater signature that modified the MORB source at an earlier stage (Hermann et al., 2006).

Sediment-derived fluids
Each selective element enrichment and/or isotopic composition that deviates from an enriched-MORB magma signature is poorly explained by only addition of AOC-derived fluids to the mantle source and implies addition of another component. Several inclusions show a wide range in B isotopic composition (down to –10) and B content (8–18 ppm) that could be explained by the contribution of sediment-derived fluids (Fig. 10) as opposed to AOC-derived fluids. Moreover, these negative B isotopic compositions could not be explained by variation in the degree of source melting, because it would imply <1% batch melting (Fig. 10). In the Lesser Antilles arc, sediments have been dredged SE of Grenada at 9°N in the western tropical Atlantic (ODP Site Leg 144-207, Demerara Rise). These represent a suitable proxy for the sediments subducted in the southern part of the arc, where the sediment series is composed of (from top to bottom) two Foraminifera and nannofossil-rich units, a black shale unit and a more silica-rich unit with zircon-bearing sand horizons (Carpentier, 2007, and ODP reports). Each sediment unit was analysed as a bulk sample at SARM-CRPG and found to contain from 11 to 65 ppm B. However, the isotopic compositions of the Lesser Antilles sediments are still poorly known. An average representative boron isotopic composition of ¹¹B = –10 was hypothesized for the Martinique continental and marine sediments (Smith et al., 1997), close to that of the sediments used in the Rose et al. (2001) model. Moreover, ODP Leg 144 contains many carbonaceous units, which should have high ¹¹B values (between +15·3 and +39·8; Deylhe et al., 2001), resembling that of seawater, as carbonates are expected to be strongly influenced by seawater. Nevertheless, the influence of such sediments cannot explain the large range of compositional variation found in M.I. with ¹¹B values as low as –10.

We have calculated the concentration and isotopic composition of boron in the fluids produced by dehydration of Lesser Antilles terrigenous sediments (assuming ¹¹B –10) during continuing dehydration. Following the Rose et al. (2001) model, we used a fractionation coefficient _{solid–fluid} of +5. The St. Vincent M.I. intersect the calculated trend of residual dehydrated sediments using a Rayleigh model (Fig. 10). They cannot result from mixing between modified MORB-source mantle and a direct melt of the residual dehydrated sediments. M.I. with 16·8 ppm B and ¹¹B of –12·0 would imply more than 80% of an hypothetical dehydrated sediment melt that should be Si-rich (Nichols et al., 1994; Prouteau et al., 2001) contrary to the main chemical characteristics of the St. Vincent M.I. We propose an alternative explanation involving fluids released from the dehydration of terrigenous sediments (B = 32 ppm and ¹¹B = –34) for the origin of the light ¹¹B recorded by the M.I. The latter data plot on the calculated mixing curve between the modified MORB source and fluids produced by 6% sediment dehydration (Fig. 10). The fluid contribution is estimated to represent up to 45% of the B.

Several additional observations support this hypothesis, as follows.

1. M.I. with light ¹¹B have a heavy oxygen isotopic signature (¹⁸O up to +10) consistent with the ¹⁸O of oceanic clays or clastic sediments (¹⁸O = 12–16 and 10–20, respectively; Eiler et al., 2000, 2005). Assuming no ¹⁸O fractionation between fluids and silicates, melt inclusions slightly enriched in Li, with negative ¹¹B and ¹⁸O from +6 to +10 would record a contribution of up to 50% of a fluid deriving from the dehydration of sediments having ¹⁸O 15 (taken as an average of both clays and clastic sediment compositions) to a modified mantle source having ¹⁸O from +4 to +5.
   
2. A subset of these M.I. (nine, including seven ‘normal’ inclusions) have negative ³⁴S (–2·0 to –4·8; Fig. 11b), characteristic of organic-rich sediments. Indeed, pelagic sediment ³⁴S composition varies with the amount of organic carbon such that microbial sulphate reduction diminishes ³⁴S (up to –20; Alt & Shanks, 2006). Black shales from the Demerara Rise in the Lesser Antilles have sulphur isotopic compositions as low as –35, and S contents up to 6 wt % (Cruse & Lyons, 2004), whereas carbonaceous and siliceous sediments can have positive ³⁴S and lower S contents (up to 14 and 850 ppm S, respectively; for example, at the ODP Site 800-802 in the Pacific; Alt & Shanks, 2006). Recycling of a black shale unit in the southern part of the Lesser Antilles arc has been proposed by Carpentier (2007). The influence of organic matter-rich sediments is also suggested on the basis of the ¹³C signature of hydrothermal fluids in the southern part of the arc (Van Soest et al., 1998).
   
3. These M.I. have Li contents ranging from 2·7 to 9·6 ppm (Fig. 8b), and variable ⁷Li (from –9·6 to +4·7). The highest Li contents (7–9·6 ppm) could be explained by the influence of sediments that have Li contents from 10 to 74 ppm (Bouman et al., 2004; Carpentier, 2007). The slightly negative ⁷Li would also be consistent with Li isotopic fractionation during slab dehydration, which would result in lighter ⁷Li in the residual sediments. As an example, Teng et al. (2007) calculated dehydrated metapelite ⁷Li values in the range –2 to –10 using a Rayleigh distillation model and different fluid/rock isotopic fractionation factors (from 1· 001 to 1· 004). However, there is no strong influence of sediment observed in the ⁷Li and Li abundance data, with only a few Li-enriched inclusions. This minimal influence possibly could be due to the close Li and ⁷Li compositions of the sediments and the AOC (Elliott, 2007).
   

We thus argue for late-stage fluid extraction from subducted sediments that had experienced 6% dehydration, which is required to explain the negative ¹¹B signature of the M.I., and a black shale contribution, to account for the negative ³⁴S. These fluids have Cl and Li contents as high as AOC dehydration fluids, but lower B contents, and higher S content as a result of the black shale input.

CaO-rich inclusions
About 10% of the M.I. selected for SIMS analyses have high CaO/Al₂O₃ (>0·8). They are characterized by H₂O contents from 1·2 to 3·9 wt %, SiO₂ < 45·5 wt %, and K₂O between 0·3 and 0·45 wt %. These inclusions cannot be distinguished from the other ‘normal’ basaltic M.I. (CaO/Al₂O₃ < 0·8), on the basis of their Cl, B or H₂O contents (Fig. 9a–c) and isotopic signatures (⁷Li, ¹¹B, ¹⁸O, ³⁴S). The significance of the Ca-rich M.I. hosted in olivine is strongly debated, as considered below. Although the origin of Ca-rich melts is beyond the scope of this work, we emphasize some interesting points that could be developed in future. Danyushevsky et al. (2004) proposed that M.I. in high-Fo olivine (Fo >85 mol %) having anomalously high CaO and low SiO₂ contents, already described in MORB, OIB and island arc basalts, result from dissolution–reaction–mixing (DRM) in the magmatic plumbing system. Fast cooling of the primitive magma should favour entrapment of numerous large M.I. in high-Fo olivines. Danyushevsky et al. (2004) argued that the magma bodies with which the melts react are often formed during differentiation of earlier batches of the same magma type, and thus no obvious isotopic anomalies should be observed. We cannot totally exclude a DRM process. However, we emphasize that (1) there is no relationship between the size, shape and composition of the St. Vincent M.I.; (2) in contrast to what was described by Danyushevsky et al. (2004), the M.I. in St. Vincent olivines have a rather small size and are rare; (3) St. Vincent M.I. that clearly depart from the compositional domain of St. Vincent lavas in terms of their high CaO/Al₂O₃ still preserved the influence of different slab-derived fluids, as the ‘normal’ basaltic M.I. do, and thus track the mantle source signatures.

Ca-rich, ne-normative melts are commonly described in arc settings (e.g. Schiano et al., 2000; Médard et al., 2004). Some workers have argued that such CaO-rich melts cannot be derived from fertile mantle peridotite sources (Della-Pasqua & Varne, 1997; Schiano et al., 2000; Kogiso & Hirschmann, 2001), but instead could be explained by: (1) melting of carbonated or (CO₂ + H₂O) fluxed lherzolite (Della-Pasqua & Varne, 1997); (2) interaction between picritic melts and clinopyroxene-rich lithologies (Kamenetsky et al., 1998); (3) melting of pyroxenitic or wehrlitic lithologies (Schiano et al., 2000; Kogiso & Hirschmann, 2001); (4) melting of depleted lherzolite (Kogiso & Hirschmann, 2001); (5) melting of amphibole-bearing, wehrlite cumulates in the arc crust at 0·2 GPa (2 kbar) and 1220°C (Médard et al., 2004). Actually, CaO-rich lavas (C-Series; Thirlwall & Graham, 1984) already exist in Grenada in the southern part of Lesser Antilles arc and could be derived from the melting of wehrlite veins in a mantle source that experienced successive metasomatism events (Vannucci et al., 2007). St. Vincent CaO-rich, ne-normative melts trapped as M.I. in olivine could represent melting of amphibole- and pyroxene-bearing veins. Assuming such an hypothesis, the Ca-rich melts cannot be derived from a specific horizon in the mantle wedge beneath St. Vincent but instead from a mineralogically heterogeneous mantle wedge with dispersed veins, to explain the presence of CaO-rich melts that are isotopically indistinguishable from typical basaltic melts. A definitive conclusion about the origin of these specific M.I. in Lesser Antilles arc requires further investigation.

Water contents and magma extraction conditions
Water contents of M.I. representative of St. Vincent basaltic magmas range from 0·9 to 5·5 wt % (Fig. 9a–c), indicating an excess of water in these magmas compared with MORB (0·1 to 0·6 wt % Newman et al., 2000; 0·12 wt % for N-MORB to 0·5 wt % for E-MORB, Sobolev & Chaussidon, 1996). Variable aqueous fluid inputs from the slab and progressive fluid transfer to the overlying mantle wedge with increasing depth of subduction were previously highlighted by studies across different arcs (Ryan et al., 1995; Walker et al., 2003; Grove et al., 2006; Portnyagin et al., 2007). Similarly, decoupling between water and elements such as Cl, B and Li is consistent with differential fractionation of water and fluid-mobile elements as reported in arc magmas (Kent & Elliott, 2002).

The thermal conditions during mantle partial melting, and thus the thermal state of the mantle wedge beneath St. Vincent, can be estimated by combining our water data with the available high-pressure experimental data on St. Vincent basalts (Pichavant et al., 2002; Pichavant & Macdonald, 2007). Based on experimental phase equilibria, the high-magnesia basalts erupted at St. Vincent are thought to have been produced by partial melting of a spinel lherzolite mantle source under relatively dry conditions (Pichavant et al., 2002), in agreement with Devine (1995), who suggested a dissolved water concentration of 2 wt % in the primary magmas of Grenada and Kick'em Jenny.

A mantle source fluxed by a hydrous slab-derived component in the Lesser Antilles arc would produce hydrous melts (Pichavant & Macdonald, 2003). High-pressure experiments on phase equilibria have demonstrated that basaltic magmas containing up 6·5 wt % of H₂O can be generated by partial melting of the mantle wedge beneath St. Vincent, at 1160°C and 18·5 kbar (Pichavant & Macdonald, 2007). Combining the P–T–H₂O experimental conditions of basaltic melt generation at St. Vincent (Pichavant et al., 2002; Pichavant & Macdonald, 2007) and the range of water contents (between 2·2 and 3·6 wt %) of our primitive ‘normal’ M.I. (hosted in Fo_>87), we consider that partial melting occurred at between 13 and 14·5 kbar (50 km), and temperatures of 1220°C and 1195°C, respectively. In such a model, magma extraction would take place under upper mantle conditions; however, the crustal thickness and structure beneath St. Vincent are poorly known (Macdonald et al., 2000, and references therein).

Under such pressure–X_H2O conditions, St. Vincent-type magmas could have been generated by 10 to 20% partial melting of a mantle source containing between 0·2 and 0·9 wt % H₂O (assuming that H₂O was perfectly incompatible during melting). These estimates are in agreement with the degrees of melting calculating in the B model (10–20%; see ‘Seawater-like signature’ section) and those that have previously been published for erupted HMB at St. Vincent (14–18%; Heath et al., 1998; Pichavant et al., 2002). St. Vincent parental magmas would therefore have MgO content between 11 and 14 wt %.

	CONCLUSIONS: CONSTRAINTS ON THE GENESIS OF ST. VINCENT HIGH-MAGNESIA BASALTS

TOP ABSTRACT INTRODUCTION SAMPLE LOCATION AND DESCRIPTION ANALYTICAL METHODS MINERAL CHEMISTRY MELT INCLUSIONS DISCUSSION CONCLUSIONS: CONSTRAINTS ON THE... SUPPLEMENTARY DATA REFERENCES

 
Melt inclusions in St. Vincent olivines record heterogeneities in the parental magma compositions at the mantle source depth that are lost during their transfer to the surface at the whole-rock scale. Indeed, they represent melts variably enriched in water and other elements that were produced by partial melting of the mantle wedge under variable pressure– temperature conditions.

At least three stages of contamination of an initial MORB-like mantle source are required to explain the compositions observed in the St. Vincent M.I. Figure 12 summarizes the successive inputs of fluids having an increasingly higher solute content: (1) an initial aqueous fluid with a low solute content and a seawater-like signature, rich in H₂O, B and Cl, with a heavy B isotopic composition; (2) a second fluid derived from dehydration of altered oceanic crust with an intermediate solute content, which was mainly responsible for selective enrichment of B in the melt; (3) finally, a late-stage fluid with a high solute content and the boron and sulphur isotopic signatures of dehydrated terrigenous sediments, probably involving black shales in agreement with other findings (Carpentier, 2007). The water-rich fluids interaction could have resulted in the formation of amphibole- and pyroxene-bearing lithologies in the mantle wedge beneath St. Vincent. Partial melting of a mineralogically heterogeneous mantle, with amphibole-bearing veins, could adequately explain the small volumes of CaO-rich melts trapped during olivine crystallization, which are otherwise isotopically or chemically indistinguishable from the remaining inclusions. By combining our data on the dissolved water concentrations of M.I. with experimental data (Pichavant et al., 2002), we propose that the St. Vincent magmas could have been successively extracted from the upper mantle at a pressure between 13 and 14·5 kbar and temperatures from 1220 to 1195°C.

View larger version (52K): [in this window] [in a new window] [Download PowerPoint slide]   Fig. 12. Schematic illustration summarizing the various stages of contamination of the mantle wedge beneath St. Vincent and the depth and temperature of generation of basaltic magmas. Chemical and isotopic compositions of olivine-hosted M.I. lead us to identify at least three different fluids that were progressively enriched in solutes: first an aqueous fluid with a seawater isotopic signature, a second fluid produced by dehydration of AOC, and a late-stage, solute-rich fluid extracted from partially dehydrated sediments. Pressures and temperatures of melt extraction are estimated on the basis of the water content of St. Vincent M.I. (2·2–3·6 wt %) and from P–T conditions of experimental phase equilibria published by Pichavant et al. (2002) and Pichavant & Macdonald (2007) (see text for details).

 

	SUPPLEMENTARY DATA

TOP ABSTRACT INTRODUCTION SAMPLE LOCATION AND DESCRIPTION ANALYTICAL METHODS MINERAL CHEMISTRY MELT INCLUSIONS DISCUSSION CONCLUSIONS: CONSTRAINTS ON THE... SUPPLEMENTARY DATA REFERENCES

 
Supplementary data for this paper are available at Journal of Petrology online.

	ACKNOWLEDGEMENTS

 
We thank Richard Robertson (Geological Survey of West Indies) and Aisha Samuels (Soufrière monitoring unit) for their kind help in sampling. We are grateful to R. Macdonald, V. Kamenetsky and an anonymous reviewer for their constructive criticisms, which greatly improved our manuscript. For technical assistance, we are grateful to Denis Mangin and Claire Rollion-Blard for ion probe measurements on the Cameca SIMS 1270, and Olfa Belhaj for her help in sample preparation and analysis. Michel Pichavant, Marion Carpentier, Marion Le Voyer, Estelle Rose, Pierre Schiano, Benoit Welsch and Pete Burnard are also acknowledged for their constructive discussions and suggestions. This study was funded by CNRS-INSU DyETI and ANR UD-Antilles.

---

*Corresponding author. Fax: +33 3 83 51 17 98. E-mail: abouvier{at}crpg.cnrs-nancy.fr

	REFERENCES

TOP ABSTRACT INTRODUCTION SAMPLE LOCATION AND DESCRIPTION ANALYTICAL METHODS MINERAL CHEMISTRY MELT INCLUSIONS DISCUSSION CONCLUSIONS: CONSTRAINTS ON THE... SUPPLEMENTARY DATA REFERENCES

 
Alt FC, Shanks WC. Stable isotope compositions of serpentine seamounts in the Mariana forearc: Serpentinization processes, fluid sources and sulphur metasomatism. Earth and Planetary Science Letters (2006) 242:272–285.[CrossRef][Web of Science]

Baker MB, Grove TL, Price R. Primitive basalts and andesites the Mt Shasta region, N. California: products of varying melt fraction and water content. Contributions to Mineralogy and Petrology (1994) 118:111–129.[CrossRef][Web of Science]

Bebout GE, Ryan J, Leeman W. B–Be systematics in subduction-related metamorphic rocks: Characterization of the subducted component. Geochimica et Cosmochimica Acta (1993) 57:2227–2237.[CrossRef][Web of Science]

Beck P, Chaussidon M, Barrat JA, Gillet, Ph., Bohn M. Diffusion induced Li isotopic fractionation during the cooling of magmatic rocks: The case of pyroxene phenocrysts from nakhlite meteorites. Geochimica et Cosmochimica Acta (2006) 70:4813–4825.[CrossRef][Web of Science]

Bouman C, Elliott T, Vroon PZ. Lithium inputs to subduction zones. Chemical Geology (2004) 212:59–79.[CrossRef][Web of Science]

Brown GM, Holland JG, Sigurdsson H, Tomblin JF, Arculus RJ. Geochemistry of the Lesser Antilles island arc. Geochimica et Cosmochimica Acta (1977) 41:785–801.[CrossRef][Web of Science]

Carpentier M. Composition chimique des sédiments entrant dans la zone de subduction des Petites Antilles. (2007) PhD thesis: Grenoble University. 354.

Chan LH, Edmond JM, Thompson G. Lithium isotopic composition of submarine basalts: implication for the lithium cycle in the oceans. Earth and Planetary Science Letters (1992) 108:151–160.[CrossRef][Web of Science]

Chaussidon M, Jambon A. Boron content and isotopic composition of oceanic basalts: geochemical and cosmochemical implications. Earth and Planetary Science Letters (1994) 121:277–291.[CrossRef][Web of Science]

Chaussidon M, Libourel G. Boron partitioning in the upper mantle—an experimental and ion probe study. Geochimica et Cosmochimica Acta (1993) 57:5053–5062.[CrossRef][Web of Science]

Chaussidon M, Sheppard SMF, Michard A. Hydrogen, sulphur and neodymium isotope variations in the mantle beneath the EPR at 12°50'N. In: Stable Isotope Geochemistry: a Tribute to Samuel Epstein. Geochemical Society, Special Publication.—Taylor HP, O’Neil JR, Kaplan IR, eds. (1991) 3:325–337.

Cruse AM, Lyons TW. Trace metal records of regional paleoenvironmental variability in Pennsylvanian (Upper Carboniferous) black shales. Chemical Geology (2004) 206:319–345.[CrossRef][Web of Science]

Danyushevsky LV, Della-Pasqua FN, Sokolov S. Re-equilibration of melt inclusions trapped by magnesian olivine phenocrysts from subduction-related magmas: petrological implications. Contributions to Mineralogy and Petrology (2000) 138:68–83.[CrossRef][Web of Science]

Danyushevsky LV, McNeill AW, Sobolev A. Experimental and petrological studies of melt inclusions in phenocrysts from mantle-derived magmas: an overview of techniques, advantages and complications. Chemical Geology (2002) 183:5–24.[CrossRef][Web of Science]

Danyushevsky LV, Leslie RAJ, Crawford AJ, Durance P. Melt inclusions in primitive olivine phenocrysts: the role of localized reaction processes in the origin of anomalous compositions. Journal of Petrology (2004) 45(12):2531–2553.[Abstract/Free Full Text]

Davidson J. Mechanisms of contamination in Lesser Antilles island arc magmas from radiogenic and oxygen isotope relationships. Earth and Planetary Science Letters (1985) 72:163–174.[CrossRef][Web of Science]

Davidson J, Harmon S. Oxygen isotope constraints on the petrogenesis of volcanic arc magmas from Martinique, Lesser Antilles. Earth and Planetary Science Letters (1989) 95:255–270.[CrossRef][Web of Science]

Decitre S. Variations de la composition isotopique du lithium dans les périotites océaniques serpentinisées et dans le manteau—Implications pour le cycle du lithium. (2000) PhD thesis: INPL, Nancy. 154.

Decitre S, Deloule E, Reisberg L, James R, Agrinier P, Mevel C. Behavior of lithium and its isotopes during serpentinization of oceanic peridotites. Geochemistry, Geophysics, Geosystems (2002) doi:10.1029/2001GC000178.

De Hoog JCM, Taylor BE, Van Bergen MJ. Sulfur isotope systematics of basaltic lavas from Indonesia: implications for the sulfur cycle in subduction zones. Earth and Planetary Science Letters (2001) 189:237–252.[CrossRef][Web of Science]

Della-Pasqua F, Varne R. Primitive ankaramitic magma in volcanic arc: a melt-inclusion approach. Canadian Mineralogist (1997) 35:291–312.[Web of Science]

Della-Pasqua FN, Kamenetsky VS, Gasparon M, Crawford AJ, Varne R. Al-rich spinels in primitive arc volcanics. Mineralogy and Petrology (1995) 53:1–26.[CrossRef][Web of Science]

Devine JD. Petrogenesis of the basalt–andesite–dacite association of Grenada, Lesser Antilles island arc, revisited. Journal of Volcanology and Geothermal Research (1995) 69:1–33.[CrossRef][Web of Science]

Deyhle A, Kopf A, Eisenhauer A. Boron systematics of authigenic carbonates: a new approach to identify fluids processes in accretionary prisms. Earth and Planetary Science Letters (2001) 187:191–205.[CrossRef][Web of Science]

Draper DS, Johnston AD. Anhydrous PT phase relations of an Aleutian high-MgO basalt: an investigation of the role of olivine–liquid reaction in the generation of arc-alumina basalts. Contributions to Mineralogy and Petrology (1992) 112:501–519.[CrossRef][Web of Science]

Eiler JM, Crawford A, Elliott T, Farley KA, Valley JW, Stolper E. Oxygen isotope geochemistry of oceanic-arc lavas. Journal of Petrology (2000) 41:229–256.[Abstract/Free Full Text]

Eiler JM, Carr MJ, Reagan M, Stopler E. Oxygen isotope constraints on the sources of central American arc lavas. Geochemistry, Geophysics, Geosystems (2005) 6:Q07007.[CrossRef]

Elliott T. Tracers of the slab. In: Inside the Subduction Factory. Geophysical Monograph, American Geophysical Union.—Eiler J, ed. (2007) 138:23–43.

Elliott T, Jeffcoate A, Bouman C. The terrestrial Li isotope cycle: light-weight constraints on mantle convection. Earth and Planetary Science Letters (2004) 220:231–245.[CrossRef][Web of Science]

Faure F, Schiano P. Experimental investigation of equilibration conditions during forsterite growth and melt inclusion formation. Earth and Planetary Science Letters (2005) 236:882–898.[CrossRef][Web of Science]

Gaetani GA, Grove TL. Experimental constraints on melt generation in the mantle wedge. In: Inside the Subduction Factory. Geophysical Monograph Series, American Geophysical Union.—Eiler J, ed. (2007) 138:107–131.

Gladney ES, Roelandts I. Compilation of elemental concentration data for USGS BIR-1, DNC-1, and W-2. Geostandards Newsletter (1987) 12:63–118.[CrossRef][Web of Science]

Grove T, Chatterjee N, Arman SW, Médard E. The influence of H₂O on mantle wedge melting. Earth and Planetary Science Letters (2006) 249:74–89.[CrossRef][Web of Science]

Gurenko AA, Chaussidon M. Boron concentrations and isotopic composition of the Icelandic mantle: evidence from glass inclusions in olivine. Chemical Geology (1997) 135:21–34.[CrossRef][Web of Science]

Hawkesworth CJ, O’Nions RK, Arculus RJ. Nd and Sr isotope geochemistry of island arc volcanics, Grenada, Lesser Antilles. Earth and Planetary Science Letters (1979) 45:237–248.[CrossRef][Web of Science]

Heath E, Macdonald R, Belkin H, Hawkesworth C, Sigurdsson H. Magmagenesis at Soufrière Volcano, St. Vincent, Lesser Antilles Arc. Journal of Petrology (1998) 39:1721–1764.[CrossRef][Web of Science]

Hermann J, Spandler C, Hack A, Korsakov AV. Aqueous fluids and hydrous melts in high-pressure and ultra-high pressure rocks: Implications for element transfer in subduction zones. Lithos (2006) 92:399–417.[CrossRef][Web of Science]

Jeffcoate AB, Elliott T, Kasemann SA, Ionov D, Cooper K, Brooker R. Li isotope fractionation in peridotites and mafic melts. Geochimica et Cosmochimica Acta (2007) 71:202–218.[CrossRef][Web of Science]

Kamenetsky VS, Eggins SM, Crawford AJ, Green DH, Gasparon AJ, Falloon TJ. Calcic melt inclusions in primitive olivine at 43°N MAR: evidence for melt–rock reaction/melting involving clinopyroxene-rich lithologies during MORB generation. Earth and Planetary Science Letters (1998) 160:115–132.[CrossRef][Web of Science]

Kent AJR, Elliott TR. Melt inclusions from Marianas arc lavas: implications for the composition and formation of island arc magmas. Chemical Geology (2002) 183:263–286.[CrossRef][Web of Science]

Kogiso T, Hirschmann MM. Experimental study of clinopyroxenite partial melting and the origin of ultra-calcic melt inclusions. Contributions to Mineralogy and Petrology (2001) 142:347–360.[Web of Science]

Kress VC, Carmichael ISE. The compressibility of silicate liquids containing Fe₂O₃ and the effect of composition, temperature, oxygen fugacity and pressure on their redox states. Contributions to Mineralogy and Petrology (1991) 108:82–92.[CrossRef][Web of Science]

Lassiter JC, Hauri EH, Nikogosian IK, Barsczus HG. Chlorine–potassium variations in melt inclusions from Rivavae and Rapa, Austral Islands: constraints on chlorine recycling in the mantle and evidence for brine-induced melting of oceanic crust. Earth and Planetary Science Letters (2002) 202:525–540.[CrossRef][Web of Science]

Leeman WP, Sisson VB. Geochemistry of boron and its implications for crustal and mantle processes. In: Boron: Mineralogy, Petrology and Geochemistry. Mineralogical Society of America, Reviews in Mineralogy.—Grew ES, Anovitz LM, eds. (1996) 33:645–708.

Lundstrom CC, Chaussidon M, Hsui AT, Kelemen P, Zimmerman M. Observations of Li isotopic variations in the Trinity Ophiolite: Evidence for isotopic fractionation by diffusion during mantle melting. Geochimica et Cosmochimica Acta (2005) 69:735–751.[CrossRef][Web of Science]

Macdonald R, Hawkesworth CJ, Heath E. The Lesser Antilles volcanic chain: a study in arc magmatism. Earth-Science Reviews (2000) 49:1–76.

Mackwell SJ, Kohlstedt DL. Diffusion of hydrogen in olivine: implication for water in the mantle. Journal of Geophysical Research (1990) 95:5079–5088.

Marschall H, Pogge von Strandmann PAE, Seitz H.-M, Elliott T, Niu Y. The lithium isotopic composition of orogenic eclogites and deep subducted slabs. Earth and Planetary Science Letters (2007) 262(3–4):563–580.[CrossRef][Web of Science]

Maurel C, Maurel P. Etude expérimentale de l’équilibre Fe²⁺–Fe³⁺ dans les spinelles chromifères et les liquides silicatés basiques coexistants. Comptes Rendus de l’Académie des Sciences (1982) 285:209–215.

Médard E, Schmidt MW, Schiano P. Liquidus surfaces of ultra-calcic primitive melts: formation conditions and sources. Contributions to Mineralogy and Petrology (2004) 148:201–215.[CrossRef][Web of Science]

Mosbah M, Métrich N, Massiot P. PIGME Fluorine determination using a nuclear microprobe with application to glass inclusions. Nuclear Instruments and Methods in Physics Research (1991) B58:227–231.[CrossRef]

Newman S, Stolper EM, Stern RJ. H₂O and CO₂ in magmas from Mariana arc and back arc systems. Geochemistry, Geophysics, Geosystems (2000) doi:1:1999GC000027.

Nichols GT, Wyllie PJ, Stern CR. Subduction zone melting of pelagic sediments constrained by melting experiments. Nature (1994) 371:785–788.[CrossRef][Web of Science]

Parkinson IJ, Arculus RJ, Eggins SM. Peridotite xenoliths from Grenada, Lesser Antilles Island Arc. Contributions to Mineralogy and Petrology (2003) 146:241–262.[CrossRef][Web of Science]

Peacock SM, Hervig RL. Boron isotopic composition of subduction-zone metamorphic rocks. Chemical Geology (1999) 160:281–290.[CrossRef][Web of Science]

Pearce JA, Peate DW. Tectonic implications of the composition of volcanic arc magmas. Annual Review of Earth and Planetary Sciences (1995) 24:251–285.[CrossRef]

Pichavant M, Macdonald R. Mantle genesis and crustal evolution of primitive calc-alkaline basaltic magmas from the Lesser Antilles arc. In: Intra-Oceanic Subduction Systems: Tectonics and Magmatic Processes. Geological Society, London, Special Publications.—Larter RD, Leat PT, eds. (2003) 219:pp. 239–254.

Pichavant M, Macdonald R. Crystallization of primitive basaltic magmas at crustal pressures and genesis of the calc-alkaline igneous suite: experimental evidence from St Vincent, Lesser Antilles arc. Contributions to Mineralogy and Petrology (2007) 154:535–558.[CrossRef][Web of Science]

Pichavant M, Mysen BO, Macdonald R. Source and H₂O contents of high-MgO magmas in island arc settings: An experimental study of a primitive calc-alkaline basalt from St. Vincent, Lesser Antilles arc. Geochimica et Cosmochimica Acta (2002) 66:2193–2209.[CrossRef][Web of Science]

Pistinier JA, Henderson GM. Lithium-isotope fractionation during continental weathering. Earth and Planetary Science Letters (2003) 214:327–339.[CrossRef][Web of Science]

Portnyagin M, Hoernle K, Plechov P, Mirinov N, Khubunaya S. Constraints on mantle melting and compositions and nature of slab components in volcanic arcs from volatiles (H₂O, S, Cl, F) and trace elements in melt inclusions from the Kamchatka Arc. Earth and Planetary Science Letters (2007) doi:10.1016/j.epsl.2006.12.005.

Poussineau S. Dynamique des magmas andésitiques: approches expérimentales et pétrostructurale, application à la Soufrière de Guadeloupe et à la Montagne Pelée. (2005) PhD thesis: Orléans University. 299.

Prouteau G, Scaillet B, Pichavant M, Maury R. Evidence for mantle metasomatism by hydrous silicic melt derived from subducted oceanic crust. Nature (2001) 410:197–200.[CrossRef][Web of Science][Medline]

Richter FM, Mendybaev RA, Christensen JN, Hutcheon ID, Williams RW, Sturchio NC, Beloso A. D. Jr. Kinetic isotopic fractionation during diffusion of ionic species in water. Geochimica et Cosmochimica Acta (2006) 70:277–289.[CrossRef][Web of Science]

Rio S, Métrich N, Mosbah M, Massiot P. Lithium, boron and beryllium in volcanic glasses and minerals studied by nuclear microprobe. Nuclear Instruments and Methods in Physics Research B (1995) 100:141–148.[CrossRef]

Rose EF, Shimizu N, Layne GD, Grove TL. Melt production beneath Mt. Shasta from boron data in primitive melt inclusions. Science (2001) 293:281–283.[Abstract/Free Full Text]

Rosner M, Wiedenbeck M, Ludwig T. Composition-induced variations in SIMS instrumental mass fractionation during boron isotope ratio measurements of silicate glasses. Geostandards and Geoanalytical Research (2008) 32(1):27–38.[CrossRef][Web of Science]

Ryan JG, Kyle PR. Lithium abundance and lithium isotope variations in mantle sources: insights from intraplate volcanic rocks from Ross Island and Marie Byrd Land (Antarctica) and other oceanic islands. Chemical Geology (2004) 212:125–142.[CrossRef][Web of Science]

Ryan JG, Morris J, Tera F, Leeman WP, Tsvetkov A. Cross-arc geochemical variations in the Kurile arc as a function of slab depth. Science (1995) 270:625–627.[Abstract/Free Full Text]

Schiano P, Eiler JM, Hutcheon ID, Stopler EM. Primitive CaO-rich, silica-undersaturated melts in islands arcs: Evidence for the involvement of clinopyroxene-rich lithologies in the petrogenesis of arc magmas. Geochemistry, Geophysics, Geosystems (2000) doi:1:1999GC000032.

Smith HJ, Leeman WP, Davidson J, Spivack AJ. The B isotopic composition of arc lavas from Martinique, Lesser Antilles. Earth and Planetary Science Letters (1997) 146:303–314.[CrossRef][Web of Science]

Sobolev AV, Chaussidon M. H₂O concentrations in primary melts from supra-subduction zones and mid-ocean ridges: Implications for H₂O storage and recycling in the mantle. Earth and Planetary Science Letters (1996) 137:45–55.[CrossRef][Web of Science]

Spilliaert N, Métrich N, Allard P. S–Cl–F degassing pattern of water-rich alkali basalt: Modelling and relationship with eruption styles on Mount Etna volcano. Earth and Planetary Science Letters (2006) 248:772–786.[CrossRef][Web of Science]

Spivack AJ, Edmond JM. Boron isotope exhange between seawater and ocean crust. Geochimica et Cosmochimica Acta (1987) 51:1033–1043.[CrossRef][Web of Science]

Straub SM, Layne GD. The systematics of boron isotopes in Izu arc front volcanic rocks. Earth and Planetary Science Letters (2002) 198:25–39.[CrossRef][Web of Science]

Straub SM, Layne GD. The systematics of chlorine, fluorine, and water in Izu arc front volcanic rocks: Implications for volatile recycling in subduction zones. Geochimica et Cosmochimica Acta (2003) 67:4179–4203.[CrossRef][Web of Science]

Taylor HP, Sheppard SMF. Igneous rocks: I. Processes of isotopic fractionation and isotope systematics. In: Stable Isotopes in High Temperature Geological Processes. Mineralogical Society of America, Reviews in Mineralogy.—Valley JW, Taylor HP, O’Neil JR, eds. (1986) 16:227–271.

Teng F-Z, McDonough WF, Rudnick RL, Wing BA. Limited lithium isotopic fractionation during progressive metamorphic dehydration in metapelites: A case study from the Onawa contact aureole, Maine. Chemical Geology (2007) 239:1–12.[CrossRef][Web of Science]

Thirlwall MF, Graham AM. Evolution of high-Ca, high-Sr C-Series basalts from Grenada Lesser Antilles: the effects of intra-crustal contamination. Journal of the Geological Society, London (1984) 141:427–445.[Abstract/Free Full Text]

Thirlwall MF, Smith TE, Graham AM, Thodorou N, Hollings P, Davidson JP, Arculus RJ. High field strength element anomalies in arc lavas: source or process? Journal of Petrology (1994) 35:819–838.[Abstract/Free Full Text]

Thirlwall MF, Graham AM, Arculus RJ, Harmon RS, Macpherson CG. Resolution of the effects of crustal contamination, sediment subduction, and fluid transport in island arc magmas: Pb–Sr–Nd–O isotope geochemistry of Grenada, Lesser Antilles. Geochimica et Cosmochimica Acta (1996) 60:4785–4810.[CrossRef][Web of Science]

Toplis MJ. The thermodynamics of iron and magnesium partitioning between olivine and liquid: criteria for assessing and predicting equilibrium in natural experimental systems. Contributions to Mineralogy and Petrology (2005) 149:22–39.[CrossRef][Web of Science]

Turner S, Hawkesworth C, Van Calsteren P, Heath E, Macdonald R, Black S. U-series isotopes and destructive plate margin magma genesis in the Lesser Antilles. Earth and Planetary Science Letters (1996) 142:191–207.[CrossRef][Web of Science]

Vannucci R, Tiepolo M, Defant MJ, Kepezhinskas P. The metasomatic record in the shallow peridotite mantle beneath Grenada (Lesser Antilles arc). Lithos (2007) 99:25–44.[CrossRef][Web of Science]

Van Soest MC, Hilton DR, Kreulen R. Tracing crustal and contributions to arc magmatism in the Lesser Antilles island arc using helium and carbon relationships in geothermal fluids. Geochimica and Cosmochimica Acta (1998) 62:3323–3335.[CrossRef]

Van Soest MC, Hilton DR, Macpherson CG, Mattey DP. Resolving sediments subduction and crustal contamination in the Lesser Antilles Island Arc: a combined He–O–Sr isotope approach. Journal of Petrology (2002) 43:143–170.[Abstract/Free Full Text]

Walker JA, Roggensack K, Patino LC, Cameron BI, Matias O. The water and trace element contents of melt inclusions across an active subduction zone. Contributions to Mineralogy and Petrology (2003) 146:62–77.[CrossRef][Web of Science]

Wallace PJ. Volatiles in subduction zone magmas: concentrations and fluxes based on melt inclusion and volcanic gas data. Journal of Volcanology and Geothermal Rresearch (2005) 140:217–240.[CrossRef]

White WM, Patchett PJ. Hf–Nd–Sr isotopes and incompatible element abundances in island arcs: implications for magma origins and crust–mantle evolution. Earth and Planetary Science Letters (1984) 67:167–185.[CrossRef][Web of Science]

White WM, Dupré B, Vidal P. Isotope and trace element geochemistry of sediments from the Barbados Ridge–Demerara Plain region, Atlantic Ocean. Geochimica et Cosmochimica Acta (1985) 49:1875–1886.[CrossRef][Web of Science]

Woodland SJ, Pearson DG, Thirlwall MF. A platinium group element and Re–Os isotope investigation of siderophile element recycling in subduction zone: comparison of Grenada, Lesser Antilles Arc and the Izu–Bonin Arc. Journal of Petrology (2002) 43:171–198.[Abstract/Free Full Text]

You CF, Castillo PR, Gieskes JM, Chan LH, Spivack AJ. Trace-element behaviour in hydrothermal experiments: Implications for fluid processes at shallow depths in subduction zones. Earth and Planetary Science Letters (1996) 140:41–52.[CrossRef][Web of Science]

Zack T, Tomascak PB, Rudnick RL, Dalpe C, Mcdonough WF. Extremely light Li in orogenic eclogites: The role of isotope fractionation during dehydration in subducted oceanic crust. Earth and Planetary Science Letters (2003) 208:279–290.[CrossRef][Web of Science]

CiteULike    Connotea    Del.icio.us    What's this?

This article has been cited by other articles:

				N. Metrich and P. J. Wallace Volatile Abundances in Basaltic Magmas and Their Degassing Paths Tracked by Melt Inclusions Reviews in Mineralogy and Geochemistry, January 1, 2008; 69(1): 363 - 402. [Full Text] [PDF]

This Article Abstract FREE Full Text (PDF) Supplementary Data All Versions of this Article: 49/8/1427    most recent egn031v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (3) Request Permissions Google Scholar Articles by Bouvier, A.-S. Articles by Deloule, E. Search for Related Content GeoRef GeoRef Citation Social Bookmarking          What's this?

Online ISSN 1460-2415 - Print ISSN 0022-3530

Copyright © 2010 Oxford University Press

Oxford Journals Oxford University Press

• Site Map
• Privacy Policy
• Frequently Asked Questions

Other Oxford University Press sites: Oxford University Press Oxford Journals China Oxford Journals Japan American National Biography Booksellers' Information Service Children's Fiction and Poetry Children's Reference Corporate & Special Sales Dictionaries Dictionary of National Biography Digital Reference English Language Teaching Higher Education Textbooks Humanities International Education Unit Law Medicine Music Online Products Oxford English Dictionary Reference Rights and Permissions Science School Books Social Sciences Very Short Introductions World's Classics