Journal of Petrology | Volume 39 | Number 10 | Pages 1721-1764 | 1998
© Oxford University Press 1998
Magmagenesis at Soufriere Volcano, St Vincent, Lesser Antilles Arc
1 Environmental Sciences Division, Iebs, Lancaster University Lancaster La1 4YQ, UK
2 Us Geological Survey Reston, Va 22092, USA
3 Department of Earth Sciences, the Open University Milton Keynes Mk7 6AA, UK
4 Graduate School of Oceanography, University of Rhode Island Narragansett, Ri 02882, USA
Received May 10, 1997; Revised typescript accepted April 16, 1998
 |
ABSTRACT |
|---|
 TOP ABSTRACT IntroductionGeological HistoryPetrography, Mineralogy and...GeochemistryMelt InclusionsPre-Eruptive Water Contents of...DiscussionReferences |
|---|
Soufriere volcano of St Vincent (<0.6 Ma) is composed of basalts and basaltic andesites, the most mafic of which (mg-number 75) may be representative of the parental magmas of the calc-alkaline suites of the Lesser Antilles arc. Parental, possibly primary, magmas at Soufriere had MgO
12.5 wt % and were probably nepheline-normative. They last equilibrated with mantle at 17 kbar pressure, at temperatures of around 1130°C and f(O2) exceeding FMQ (fayalite–magnetite–quartz) + 1. They fractionated, along several liquid lines of descent, through to basaltic andesites and rarer andesites over a range of crustal pressures (5–10 kbar) and temperatures (1000–1100°C), separating initially olivine + Cr-spinel + clinopyroxene + plagioclase ± titanomagnetite and then clinopyroxene + plagioclase + titanomagnetite + orthopyroxene assemblages. The total amount of crystallization was some 76 wt %. Amphibole was apparently not a fractionating phase. Sr and Nd isotopic and trace element systematics show no evidence for significant crustal assimilation. There is conflicting evidence as to the pre-eruptive water contents of Soufriere magmas; compositions of clinopyroxene phenocrysts and melt inclusions suggest H2O >3 wt %, whereas various projections onto phase diagrams are more consistent with relatively anhydrous magmas. Primary magmas at Soufriere were generated by around 15% melting of mid-ocean ridge basalt type mantle sources which had been modified by addition of fluids released from the slab containing contributions from subducted sediments and mafic crust. KEY WORDS: high-MgO arc magmas; geochemistry; magmagenesis; Lesser Antilles; Soufriere St Vincent
|
Introduction |
|---|
|
TOPABSTRACTIntroductionGeological HistoryPetrography, Mineralogy and...GeochemistryMelt InclusionsPre-Eruptive Water Contents of...DiscussionReferences |
|---|
The Lesser Antilles intra-oceanic arc is a 750 km long chain of volcanic islands resulting from the subduction of rocks of Jurassic to Cretaceous age of the American plate beneath the eastern edge of the Caribbean plate (Fig. 1). Plate convergence rates are relatively slow; since the Middle Eocene they have averaged 2.0–2.2 cm/year (Pindell et al., 1988
), compared with the arc average of 6.5 cm/year (Gill, 1981). Brown et al. (1977) showed that the compositions of volcanic rocks vary along the active arc, allowing the islands to be grouped according to three magma series: tholeiitic in the islands north of Montserrat, calc-alkaline in the central islands (Montserrat to St Lucia), and alkaline in the southernmost islands (Grenada and southern Grenadines). Those workers proposed that the volcanic rocks of St Vincent are transitional, in terms of magmatic affinity, between the southern and central island suites, consistent with the geographical position of the island. Thirlwall et al. (1994), on the other hand, suggested that high-MgO basalts from St Vincent are transitional between tholeiitic and calc-alkaline, andSmith et al. (1996) referred to them as being tholeiitic. Although recognizing the transitional nature of the suite, we shall refer to the Soufriere rocks as calc-alkaline.
|
The occurrence of different magma series within a close spatial and temporal context and the common presence of magnesian (MgO >10 wt %) lavas in the southern islands present a rare opportunity to assess the factors which control magma compositions in a modern intra-oceanic arc setting. The more magnesian members (MgO
12.5 wt %) of the Soufriere St Vincent suite represent (near-)primary magmas of a type which acted as parental to the calc-alkaline suites of the central islands. St Vincent is probably the only island in the arc where such rocks are relatively common. We know of only one other occurrence in the calc-alkaline suites, the olivine–clinopyroxene-phyric picritic basalt of Ilet à Ramiers, Martinique (Westercamp & Mervoyer, 1976). There is a compositional continuum between magnesian basalts and andesites on Soufriere. This provides an opportunity to contribute to the debate as to whether generation of calc-alkaline series is by closed-system fractional crystallization of parental basalts or by combined fractional crystallization and crustal contamination. Furthermore, we evaluate the role of water in the evolution of the suite, particularly the issue of whether the mafic magmas were water rich.
|
Geological History |
|---|
|
TOPABSTRACTIntroductionGeological HistoryPetrography, Mineralogy and...GeochemistryMelt InclusionsPre-Eruptive Water Contents of...DiscussionReferences |
|---|
Figure 2 is a generalized geological map of St Vincent; no detailed map exists. The Soufriere stratovolcano dominates the northern half of the island and is the most active subaerial volcano in the arc. K–Ar dating has placed a lower limit of 0.6 Ma on the age of Soufriere (Briden et al., 1979
). There are several other major volcanic centres on the island which are no longer active; the ages of the Richmond Peak–Mt Brisbane centres to the immediate south of Soufriere, and the Grand Bonhomme centre further south are not precisely known, but the pre-Soufriere lavas dated Briden et al. (1979) yielded K–Ar ages of between 1 and 3 Ma.
|
The geological evolution of the volcano has been characterized by four main volcanic formations (Sigurdsson & Carey, 1991
). These represent protracted periods of predominantly effusive or explosive volcanism, which may be controlled more by the geomorphological and hydrological features of the volcano than by the silica and volatile contents of the magmas (Sigurdsson & Carey, 1991). Early activity (0.6 Ma–10 ka) was characterized by the extrusion of basaltic and basaltic andesite lavas from a central vent, with <5% pyroclastic deposits. These are named the Pre-Somma Lavas because they pre-date a probable major structural failure of the volcano's southern flank which created the Somma scarp and generated a thick debris flow. Some Pre-Somma lavas from the southern flanks of Soufriere are unusually magnesian, and may possibly have been erupted from a different centre or centres. A series of well-bedded pyroclastic fall deposits mantles much of the island of St Vincent, and has been tentatively correlated with coarse tephra beds exposed in the crater and on the flanks of Soufriere. The units range from black scoria to yellow lapilli and pumiceous tuff, and are collectively known as the Yellow Tuff Formation, with an estimated volume of 48 km3 (Rowley, 1978). Carbon dating indicates that the formation spans the period 3600–4500 years bp. The rarity of unconformities suggests rapid deposition. The vents feeding the Yellow Tuff Formation have not been identified with certainty; but the sequence was probably erupted from the central Soufriere vent (Sigurdsson & Carey, 1991).
A predominantly effusive phase of activity followed the emplacement of the Yellow Tuff Formation, with the eruption of ponded basaltic and basaltic andesitic Crater Lavas. The most recent phase of activity has been characterized by vulcanian explosive eruptions, generating a thick succession of pyroclastic fall and flow deposits, named the Pyroclastic Formation. New 14C dates (Sigurdsson et al., 1998) suggest that the Pyroclastic Formation may extend back much further than the 1300 years indicated by Sigurdsson & Carey (1991), and possibly overlaps with the Yellow Tuff Formation. There have been at least five major historic eruptions of the Soufriere (1718, 1812, 1902, 1971, 1979); the activity has been characterized by the extrusion of basaltic andesite lava domes in the crater area followed by phreatomagmatic explosions generating pyroclastic flows.
|
Petrography, Mineralogy and Physical Parameters |
|---|
|
TOPABSTRACTIntroductionGeological HistoryPetrography, Mineralogy and...GeochemistryMelt InclusionsPre-Eruptive Water Contents of...DiscussionReferences |
|---|
Petrography
Petrographic descriptions of Soufriere eruptive rocks have been given by Roobol & Smith (1975)
, Carey & Sigurdsson (1978), Shepherd et al. (1979), Graham & Thirlwall (1981), Devine & Sigurdsson (1983), Dostal et al. (1983) and Bardintzeff (1984, 1992). Most of these studies concentrated on the products of the activity since 1902.
Detailed sample and locality descriptions of rocks used in this study are given in Appendix A. Soufriere is formed almost entirely of basalts and basaltic andesites [classification scheme of LeBas et al. (1986)]; andesites are found only as components of mixed magma rocks of the Yellow Tuff Formation. Both basalts and basaltic andesites have been erupted throughout the volcano's history, although basalts were volumetrically at their most abundant in the earlier stages (Pre-Somma and Yellow Tuff Formations). Roobol & Smith (1975) recorded a progressive change from early erupted basaltic andesite to late basalt in the 1902–1903 activity, perhaps pointing to the existence of zoned magma chambers at least spasmodically beneath Soufriere.
Table A1 shows the modal proportions (volume percent, recalculated vesicle- and groundmass-free) of phenocrysts in representative samples from each of the main geological formations of Soufriere. The basalts range from microphyric, fine-grained rocks with abundant (up to 30%) microphenocrysts of ol + spinel ± cpx, to more coarsely porphyritic rocks also containing phyric plagioclase. Basaltic andesites are generally phenocryst rich (35–60%), containing the assemblage cpx + Ti-mag + plag + opx ± ol. The order of appearance of phases was ol + Cr-sp ± Ti-mag, followed by cpx, plag and then opx.
|
Although it forms a core to an augite crystal in STV 303, pigeonite occurs mainly as thin rims around orthopyroxene, clinopyroxene and olivine crystals in the Pre-Somma Lavas (STV 315, 318, 323). Titanomagnetite microphenocrysts are present in some basalts but are always subordinate to Cr-spinel. Ilmenite is present only as a groundmass phase. Amphibole has not been recorded as a phenocryst at Soufriere, though Shepherd et al. (1979)
and Graham & Thirlwall (1981) found corroded amphiboles in products of the 1979 eruption which were assumed to be xenocrystic. We have found rounded crystals of amphibole in STV 354 (from the 1979 eruption) and STV 363 (scoria of unknown age from the Pyroclastic Formation). The amphiboles are 2 mm long, greenish yellow in colour, and contact plagioclase crystals. The aggregates probably represent parts of cumulate blocks. Only one occurrence of phyric apatite has been found, as a microphenocryst included in a plagioclase phenocryst in andesite STV 376(L). There is a marked tendency for the phenocryst phases to form clusters, up to 2 mm across. These vary from monomineralic, clinopyroxenitic clots through olivine–clinopyroxene-rich clusters to titanomagnetite-rich gabbros. STV 358 contains a 4 mm x 3 mm olivine aggregate where the crystals are strained and partially recrystallized.
Melt (glass) inclusions occur in all the phenocryst phases (Graham & Thirlwall, 1981; Devine & Sigurdsson, 1983; Bardintzeff, 1992). Those in olivine tend to be large (50–100 µm) and spheroidal, and generally have large contraction bubbles. There is a tendency for inclusions in olivines in the most magnesian basalts to be at least partly devitrified. Inclusions in orthopyroxene are more abundant and larger (
50 µm) than those in clinopyroxene ( 30 µm), whereas in plagioclase large ( 100 µm) subrectangular to spheroidal inclusions form trains in crystal cores.
Macroscopic evidence (e.g. banded and mingled pumices) for mixing between basaltic andesite and dacite at Soufriere has been documented for the 1902 (Carey & Sigurdsson, 1978) and 1979 (Shepherd et al., 1979; Graham & Thirlwall, 1981; Devine & Sigurdsson, 1983; Bardintzeff, 1992) eruptions. Light-coloured bands of dacite in scoria blocks in the 1979 ejecta contain feldspar, orthopyroxene, quartz and magnetite crystals, and some contain partially fused granitic xenoliths (Graham & Thirlwall, 1981; Devine & Sigurdsson, 1983). They are evidence of at least local crustal fusion beneath Soufriere. We have also found clear evidence of mixing between basalt and andesite and between basaltic andesite and andesite in western outcrops of the Yellow Tuff Formation and in scoria from the Pyroclastic Formation.
Cumulate-textured blocks are found at most volcanoes in the Lesser Antilles, but are particularly abundant and well documented at Soufriere, St Vincent. Mineralogy and textures are highly variable, but hastingsitic amphibole, calcic plagioclase, olivine, titanomagnetite and high-Ca pyroxene are common cumulate phases (Lewis, 1973a, 1973b; Arculus & Wills, 1980; Dostal et al., 1983).
Metavolcanic and calc-silicate sedimentary xenoliths are also common at Soufriere (Devine & Sigurdsson, 1980; Carron & Le Guen de Kerneizon, 1991). We have analysed five metamorphic xenoliths to help constrain the nature of any potential crustal assimilant; the protoliths were tuffs (STV 336, 340), basalt (STV 353), and andesite (STV 337, 339). STV 337 is cordierite bearing.
Phenocryst compositions
In this section, we refer to various mineral–melt partitioning data, using whole-rock compositions as proxies for melt compositions. We appreciate that this creates problems in strongly porphyritic rocks, in that later stages of phenocryst growth may have been from melts widely removed in composition from the bulk rock. We have tried to take account of this effect by using only phenocryst core compositions, which we assume to have crystallized close to the liquidus. Mineral compositional data for Soufriere samples are available from R.M. on request.
Olivine phenocrysts in the most magnesian basalts (e.g. STV 301) take three forms: (1) euhedral to subhedral prisms up to 1 mm in size, with core compositions ranging from Fo89 to Fo87 and substantial zoning (up to 19% Fo); (2) smaller (<0.5 mm) rounded crystals, with slightly more Fe-rich cores (Fo86–83) and zoning <14% Fo; (3) rare, ragged, variably resorbed, probably xenocrystic, grains, with cores of Fo80 and only weak, normal zoning (3% Fo). No reverse zoning has been found; this suggests that crystallization occurred in a system where there was no, or minimal, input of fresh magma. In the basaltic andesites, olivines are small (<1 mm) and commonly embayed through resorption; mantling textures indicate an olivine–orthopyroxene reaction relationship. Core compositions range from Fo85 to Fo55, with rare values of Fo90. NiO contents reach 0.38% and are positively correlated with mg-number.
The most magnesian olivines (mg-number 89) are found as phenocrysts in the most magnesian basalts, e.g. STV 301. This rock has an mg-number of 73 [using an Fe2O3/FeO weight ratio of 0.22, derived from the Kilinc et al. (1983) expression for Fe speciation, and an f(O2) of FMQ (fayalite–magnetite–quartz) + 1.7 (see below)]. Olivine of composition Fo89 could have crystallized from melt with mg-number 73 if KD = 0.33. This value is the same as that determined by Wagner et al. (1995) for olivines grown in 1 kbar, water-saturated experiments on a high-Al2O3 basalt from the Medicine Lake volcano, California, and similar to the average value of 0.34 determined for the 1 kbar water-saturated experiments of Sisson & Grove (1993b). Ulmer (1989) has demonstrated experimentally that Mg–Fe2+ partitioning between olivine and liquid in a calc-alkaline picrobasalt is pressure dependent, and reported a range of KD from 0.315 at 1 bar to 0.365 at 25 kbar. The value for 15 kbar, close to the possible equilibration pressure of 17 kbar for STV 301 (see Figs 8 and 9, below) was 0.341. We conclude, therefore, that STV 301 represents the most primitive melt erupted by Soufriere. Even more primitive rocks have been recorded from other arcs; Eggins (1993) has collated compositions of olivine phenocrysts from several arc systems, some as high as Fo94.
|
|
On plots of Fe2+/Mg ratios of olivine phenocrysts against whole rocks (Fig. 3a), core compositions, in particular, plot on both sides of any reference KD line, that is, they apparently have compositions that are either too evolved (above the line), or too primitive (below) for the whole-rock composition. There are several possible explanations for the scatter:
- The P–T–X dependence of KD (Ulmer, 1989) means that KD should constantly change to lower values as magma evolves, and not remain constant as is commonly assumed.
- Relatively primitive core compositions in basaltic andesites, e.g. Fo86 in SVE 113 (SiO2 54.8 wt %) may represent phenocrysts which did not re-equilibrate from higher-temperature stages of magma evolution, and/or the products of mixing with a more primitive basaltic magma. We note, however, that reverse zoning is uncommon in Soufriere phenocrysts.
- The compositions of the rims of many olivines are more Fe rich than could be predicted from equilibrium relationships (Fig. 3a). This again may reflect magma mixing, this time with a more evolved melt, or differential amounts of re-equilibration of olivine microphenocrysts by solid-state diffusion at magmatic temperatures.
|
In all rock types, a spinel phase fltorms small (<<0.2 mm) euhedral to subhedral inclusions in the cores of olivine and, less frequently, pyroxene phenocrysts, and also occurs more rarely as partially resorbed, discrete microphenocrysts, the size increasing from <0.5 mm in basalts to 0.8 mm in basaltic andesites. There is a continuum of compositions from Cr-spinel to titanomagnetite, although the majority of analyses tend to be bimodal (Fig. 4); in Cr-spinel, cr-number ranges from 35 to 85 (the majority >50), mg-number from 12 to 54 and Fe3+/(Fe3+ + Al + Cr) from 10 to 80. TiO2 varies from 0 to 10 wt %. In titanomagnetites, the same ranges are 0–35 (most <10), 5–26 and 81–98, respectively, and TiO2 varies from 5 to 30 wt % (Fig. 4). Compositional zoning has been measured in only a handful of larger microphenocrysts and shows Mg–Fe, Cr–Al and Fe2+–Fe3+ ratios decreasing, and TiO2 abundances increasing, towards crystal rims. Differentiation within the high-temperature spinels led to ferrite and ulvöspinel enrichment with little change in cr-number (Fig. 4a). This is typical of crystallization under relatively oxidizing conditions (Ballhaus et al., 1991
). There is a crude, positive correlation between Fe2+/Mg ratios in spinels and whole rocks, suggesting an approach to equilibrium crystallization.
|
The Soufriere spinels are similar to spinel compositions recorded from island arcs elsewhere; on a Cr–Al–Fe3+ plot (Fig. 4b), for example, they largely fall within the fields of arc basalts constructed by Eggins (1993)
. The compositional range, and particularly the continuum between Cr-spinels and titanomagnetites, closely matches spinels from picrites of Ambae volcano in the Vanuatu arc (Eggins, 1993). The majority of clinopyroxene phenocrysts are augite, although diopsidic cores are found in olivines and augites in more magnesian rocks, e.g. STV 301 and STV 334. They are typically 0.5–2 mm in size, subhedral, with weak zoning, partially resorbed margins and inclusions of Fe–Ti–Cr oxides. There is a strong tendency in some rocks for pyroxene and olivine crystals to form clusters 2 mm in diameter.
Maximum Cr concentrations ( 0.9 wt %) are found in clinopyroxene cores in some more magnesian basalts (STV 301, 334); values in more evolved rocks are typically <0.1 wt %. The Soufriere clinopyroxenes are notably Al rich, with maximum a.f.u. values >0.25 (Al2O3 > 6 wt %) in some basalts. Variation in Al with mg-number in the clinopyroxenes (Fig. 5a) shows a diffuse peak at mg-number 80. This is similar to the situation in ankaramites from western Epi (Barsdell & Berry, 1990) and in picrites from Ambae (Eggins, 1993), both in the Vanuatu arc. In all three cases, the peak coincides with the commencement of plagioclase crystallization. The clinopyroxenes in Soufriere basalts reach relatively high AlVI values ( 0.13 a.f.u.). Although such high values might indicate high-pressure crystallization (Kennedy et al., 1990), the fact that the distributions of AlIV and AlVI as a function of mg-number are closely similar to that of 
Al suggests that the AlIV/AlVI ratio is composition dependent. Ti behaviour in Soufriere clinopyroxenes (Fig. 5b) fairly closely mimics that of Al, as it does in the Ambae suite (Eggins, 1993), though the abundances in Soufriere basalts tend to be higher.
|
Compositional zoning is most commonly normal, with rimwards enrichment in Fe (<5% Fs) and decrease in Cr, Al and sometimes Ti. Reverse zoning is restricted to some basaltic andesites and is typically at the limit of analytical resolution (Fs
1%).
Clinopyroxene and whole-rock compositions correlate fairly well, suggesting that quasi-equilibrium conditions prevailed during clinopyroxene crystallization (Fig. 3b). Relationships are consistent with an equilibrium KD around 0.4 in the basalts and 0.35 in more evolved rocks. These values are higher than those (0.20–0.25) found experimentally in 1 atm, anhydrous experiments on mid-ocean ridge basalt (MORB) compositions and silica-saturated and -undersaturated arc lavas (Grove & Bryan, 1983; Grove & Baker, 1984; Kennedy et al., 1990) and 1 kbar water-saturated experiments by Sisson & Grove (1993b), but comparable with that of 0.38 determined in an Aleutian high-magnesia basalt at 12 kbar, 1315°C under anhydrous conditions by Johnston & Draper (1992). A possible explanation for the elevated KD values is the relatively aluminous nature of the Soufriere clinopyroxenes. Sisson & Grove (1993a) found experimentally that Al-rich sectors of zoned clinopyroxenes synthesized from natural high-alumina basalts had higher KD (up to 0.31) than less aluminous sectors.
Orthopyroxene crystallized instead of pigeonite in the basaltic andesites presumably because either (1) the melt temperatures were too low for the likely equilibration pressures (
5 kbar) and the mg-number of the rocks (Fig. 6), and/or (2) the f(O2) of the magmas was too high. Pigeonite (Wo7.8 En65.5 Fs26.7) has been found as a core in an augite crystal in STV 303 which may have been relatively water poor and more reduced. We ascribe the occurrence of pigeonite rims (Wo5–11 En48–66 Fs28–45) to the other mafic phases to crystallization duringmagma ascent, as the pigeonite–orthopyroxene inversion is shifted to lower temperatures at lower pressures (Lindsley, 1980).
|
Plagioclase phenocrysts are abundant in all but themost magnesian rocks, ranging in size from 5 mm to microphenocrysts <0.5 mm. Phenocrysts are commonly euhedral, with conspicuous normal or fine (
1 µm) oscillatory zoning. Compositions range from An97 to An46; anorthitic cores are common (An >90), whereas rims tend to be more sodic (An70–50). Average core compositions are shown in Fig. 3d but the range of core compositions in individual specimens can exceed 30% An. The extent of zoning in single grains may exceed 40% An. No reverse zoning has been recorded. Plagioclase inclusions are frequently found within olivine and pyroxene phenocrysts, and these exhibit an even greater range of compositions (An96–An8) than discrete phenocrysts. The most calcic compositions are from inclusions in olivine, whereas the most sodic plagioclases usually occur included in clinopyroxenes.
Ca/Na ratios in plagioclase cores and rims are plotted against whole-rock Ca/Na ratios in Fig. 3a. The spread of core compositions at given whole-rock Ca/Na probably reflects the presence of unre-equilibrated crystals from higher-temperature stages of evolution, mixing between more and less evolved magmas, and crystallization of plagioclase from variably water-rich magmas. Sisson & Grove (1993a), for example, have shown experimentally that KD at 2 kbar varies with water content of the melt, from 1.0 for anhydrous melts to 5.5 for melts with 6% water (Fig. 3d).
Orthopyroxene occurs in Soufriere rocks with >51% SiO2 as (1) small (<2 mm), pink, euhedral to subhedral prisms, (2) anhedral crystals in pyroxene-olivine clusters, (3) reaction rims around olivine, or more rarely clinopyroxene, crystals, and (4) cores to clinopyroxene crystals. The total compositional range is En70–54. Crystals are only weakly, normally zoned, rimward decreases in En usually being 2%. TiO2 and Al2O3 tend to increase rimwards.
The relationship between orthopyroxene and whole-rock Fe2+/Mg ratios (Fig. 3c) is consistent with crystallization of orthopyroxene under close to equilibrium conditions and with KD somewhat greater than 0.4. This is comparable with the value of 0.37 determined for orthopyroxene in a high-magnesia basalt from Umnak island in the Aleutian arc at 12 kbar, 1315°C under anhydrous conditions by Johnston & Draper (1992). Such values are higher than the 0.27 used, for example, by Hunter & Blake (1995), and may, as suggested above for the clinopyroxenes, be related to the relatively aluminous nature of the Soufriere orthopyroxenes (Al2O3 2.6 wt %).
Intensive parameters
Barometry
There is currently no reliable mineral geobarometer which allows us to estimate total lithostatic pressure without an independent assessment of magmatic water contents. We shall return to the question of water contents later. Here we estimate lithostatic pressure using two, experimentally determined, sets of phase relationships, at least one of which is apparently only weakly dependent on the water content of the system.
Sobolev & Danyushevsky (1994) used the Ol–Q–Pl projection (from Di) of Walker et al. (1979) and calibrated ol + opx ± cpx cotectics to determine the pressure of the last equilibration of primary melts with their mantle sources (Fig. 7). The most primitive Soufriere rocks [(ol + sp ± cpx)-phyric] lie close to an interpolated 17 kbar cotectic for both dry and wet (2 wt % H2O) melts, which, if we assume that these rocks were close to saturation with orthopyroxene at higher pressures, corresponds to an equilibration depth of 55 km. Given a crustal thickness of 30 km beneath St Vincent (Boynton et al., 1979), this implies equilibration within the uppermost 30 km or so of the mantle. Rocks in which plagioclase had apparently just become a liquidus phase plot in the range 17–14 kbar. More evolved rocks, including the most mafic opx-phyric specimen (STV 315) plot down-pressure from the 10 kbar cotectic, in an uncalibrated region of the diagram but implying equilibration within the crust.
|
The pressure at which the (olivine ± clinopyroxene)-phyric basalts equilibrated with mantle peridotite can also be estimated using the system diopside (Di)–jadeite + calcium Tschermak's molecule (Jd + CaTs)–olivine (Ol)–quartz (Qz). Figure 8 is a projection from Di into the plane (Jd + CaTs)–Ol–Qz, on which are shown cotectics, defined by experiments on the Tinaquillo lherzolite, representing the locus of liquids in equilibrium with ol + opx + cpx + sp over the pressure range 0.5–4.0 GPa (after Falloon et al., 1988
). The most magnesian rocks project to a pressure of 1.7 GPa (17 kbar). More evolved basalts plot at lower pressures; orthopyroxene apparently became a liquidus phase at 0.5 GPa (5 kbar). Despite the fact that the cotectics shift away from olivine in the hydrous system, the pressure estimates are similar to those from the previous projection.
The 17 kbar pressure estimates for the (olivine ± clinopyroxene)-phyric basalts refer to the final pressure of equilibration with mantle peridotite. It is possible that even the most primitive Soufriere magmas were themselves derived from more olivine-rich parents; for example, Thirlwall et al. (1996) have inferred that the parental magmas to the basaltic suites of Grenada were picritic basalts with MgO contents of 16 wt %. Such parental magmas may have been an integration of melt fractions or batches from various levels in the mantle wedge. There may even have been mantle fusion at these relatively shallow depths, especially if the isotherms beneath St Vincent have been perturbed upwards by the passage of magma batches through reused conduits, in the manner proposed for Umnak island in the Aleutians by Johnston & Draper (1992).
Thermometry
Previous temperature estimates for Soufriere rocks are summarized in Table 1. There are some inconsistencies in these estimates, particularly the fact that the basaltic andesites yield higher temperatures than the basalts. We have used, therefore, the olivine–spinel exchange thermometer of Ballhaus et al. (1991) and the two-pyroxene thermometer of Lindsley (1983) to try to determine more precisely the crystallization temperatures of Soufriere rocks. Unfortunately, the absence of phenocrysts of ilmenite precluded use of the coexisting oxides geothermometer.
|
Successful application of the Ballhaus et al. (1991)
thermometer relies, inter alia, on the correct choice of (1) equilibration pressure, (2) olivine–spinel pairs coexisting in equilibrium, and (3) oxidation ratio of the spinel phase. We have used pressures estimated from Figs 7 and 8 (ranging from 1.7 GPa for STV 301 to 0.5 GPa for STV 315), and oxidation ratios based on stoichiometry. Furthermore, we have used only basalts (i.e. SiO2 <52 wt %), as olivine is in reaction relationship with orthopyroxene in the basaltic andesites. The most magnesian olivine and Cr-spinel compositions were used, but the results were found to be extremely sensitive to slight variations in input compositions (particularly of olivine). To illustrate this, a range of temperatures (and oxygen fugacities) are shown in Table 2, corresponding to a range of olivine mg-numbers. It is usually difficult to determine simply on textural grounds which highly magnesian olivine composition most closely approaches equilibrium with respect to the most magnesian Cr-spinel inclusion found in each rock.
|
Calculated temperatures range from 1026 to 1147°C (Table 2). The higher end of this range is more likely to represent realistic crystallization temperatures for basalts, and compares well with previously determined temperatures for Soufriere rocks (Table 1).
We applied the two-pyroxene thermometer to the nine samples for which we have data for coexisting pyroxenes. In most samples, the pyroxenes apparently coexisted close to equilibrium, in the sense that tie-lines do not cross (Fig. 9) and mg-numbers are similar. Only four, however, gave consistent temperatures for both pyroxenes, in the sense that the ranges for each pyroxene overlap. All four are basaltic andesites within the narrow compositional range SiO2 54–56%. Temperatures (for pressures of 5 kbar) are 1050–1060°C (±50°C), a little lower than most of those quoted in Table 1, and overlapping with the temperature ranges for the basalts which were estimated using the olivine–spinel geothermometer.
Oxygen fugacity
f(O2) has been estimated using the oxygen geobarometer of Ballhaus et al. (1991), and the results are presented in Table 2. Temperatures were taken from Table 2 and pressures were inferred from Figs 8 and 9. We use only basalts, as the barometer is most applicable to mafic rocks. No correction was made for the absence of orthopyroxene in the basalts, but the correction is small for rocks of this composition (a few tenths of a log unit; Ballhaus et al., 1991).
Using the Ballhaus et al. (1991) formulation, f(O2) values in the basalts range from FMQ + 0.9 to + 1.9. Such values are comparable with those estimated from coexisting oxides in rocks from Statia and from Mt Pelée, Martinique (Smith & Roobol, 1990). They are also comparable with f(O2) estimates for olivine–spinel pairs in other island arc lavas (e.g. FMQ + 1 to + 3; Eggins, 1993), which are generally more oxidizing than values obtained for MORB and intra-plate basalts (FMQ –1 to FMQ + 1; Ballhaus et al., 1991). This suggests that the mantle wedge over subduction zones is more oxidizing than the mantle sources of other basalt types, perhaps as a result of introducing fluids derived from the subducting slab into the wedge (Ballhaus et al., 1991; Wood, 1991).
|
Geochemistry |
|---|
|
TOPABSTRACTIntroductionGeological HistoryPetrography, Mineralogy and...GeochemistryMelt InclusionsPre-Eruptive Water Contents of...DiscussionReferences |
|---|
Analytical data for Soufriere rocks have previously been presented by Tomblin (1968)
, Baker (1972), Pushkar et al. (1973), Roobol & Smith (1975), Brown et al. (1977), Rowley (1978), Graham & Thirlwall (1981), Devine & Sigurdsson (1983), Dostal et al. (1983), White & Patchett (1984), White & Dupré (1986), Bardintzeff (1992), Thirlwall et al. (1994) and Turner et al. (1996). Our data set essentially covers the compositional range as established in these papers and for the sake of internal consistency we use, with some exceptions (specified in the text), only our data in subsequent sections. We present major and X-ray fluorescence (XRF) trace element data for 49 eruptive rocks and five metamorphic xenoliths ( Tables A2 and A3), instrumental neutron activation analysis (INAA) data for 18 eruptive rocks (Table A4), Sr isotopic ratios in 19 eruptive rocks and five xenoliths, Nd isotopic ratios in 13 eruptive rocks and two xenoliths, and Pb isotopic ratios in 14 eruptive rocks (Table A5). Analytical techniques are described in Appendix B.
|
|
|
Liquid compositions?
Before the geochemical evolution of Soufriere can be discussed, it is necessary to assess the extent to which whole-rock compositions represent liquid compositions. Modal proportions of individual phenocryst phases are as high as 47 vol. % (e.g. plagioclase in STV 315; Table A1), which might reflect accumulation. If such porphyritic rocks represent liquid compositions, they must preserve the equilibrium proportions of crystallizing phenocryst phases. Furthermore, there is abundant evidence for disequilibrium conditions during crystallization (complexly zoned phenocrysts, resorbed mineral textures). Despite the large range in mineral compositions observable in individual rocks and between different rocks, the phenocryst assemblages and average compositions vary with whole-rock composition in a fairly predictable manner and tie-lines between the coexisting mineral phases generally do not cross.
Whereas Crawford et al. (1987) have suggested that arc basalts with >18 wt % Al2O3 form by plagioclase accumulation, Sisson & Grove (1993a) have shown that liquids saturated with ol + pl + cpx + H2O at 2 kbar contain 20 wt % Al2O3. A 20 wt % value has been used by Davidson et al. (1993) for identifying plagioclase accumulation in Lesser Antilles rocks. At Soufriere, Al2O3 concentrations increase overall with increasing differentiation, as marked by decreasing MgO abundances (see Fig. 14a, below), but not beyond the 20 wt % threshold value. The lack of a pronounced positive Eu anomaly in chondrite-normalized REE diagrams (see Fig. 15, below) also argues against significant accumulation of plagioclase.
|
|
As for the possibility of olivine accumulation in some of the most magnesian lavas, we noted earlier that the olivine in the most primitive basalt (STV 301) was apparently in, or close to, equilibrium with the whole-rock composition. Furthermore, there are no simple correlations in the basalts between modal olivine or clinopyroxene and mg-number or Ni and Cr contents in the whole rocks. There seems to be no good evidence for olivine accumulation, a conclusion also reached for C-series lavas of Grenada by Thirlwall & Graham (1984)
. Finally, the compositions of melt inclusions in phenocrysts fairly closely match those of the whole rocks, at least in the range covered by the whole rocks; Fig. 10 shows CaO–SiO2 relationships as an example. We suggest, therefore, that the Soufriere bulk rocks represent (near-)liquid compositions.
|
Magmatic affinity
Figure 11 is a plot of MgO vs degree of silica saturation. The Soufriere St Vincent rocks form an array stretching from silica saturation (Q
0) to silica oversaturation (Q 20), overlapping at the higher-Q end with the calc-alkaline and low-K tholeiite fields. The slope of the array, and the slightly nepheline-normative character of STV 301, seem to indicate that more primitive Soufriere magmas, if they existed, would have been nepheline-normative, a feature which they would share with the silica-undersaturated mafic end-members of the M- and C- series of Grenada (Fig. 11). The slope of the Soufriere array in Fig. 11 is related to fractionation of silica-undersaturated cpx + spinel, in addition to ol + plag.
|
When plotted in standard discrimination diagrams (Fig. 12), the St Vincent suite is not unequivocally assigned to either tholeiitic or calc-alkaline category. Miyashiro (1974)
used FeO*/MgO–SiO2 relationships to discriminate between the two series. He proposed that each series should have progressed into the relevant field at intermediate degrees of differentiation (i.e. 2 < FeO*/MgO < 5), that is, the discriminating feature is the slope of the compositional trend. The Soufriere suite would be more likely to be classified as tholeiitic on this basis, though the data spread across the boundary (Fig. 12a). On the K2O–SiO2 plot (Fig. 12b; after Rickwood, 1989) the rocks scatter across the boundaries between the two magma series. In an AFM diagram (Fig. 12c; after Irvine & Baragar, 1971), the mafic members of the suite plot in the tholeiite field, whereas more evolved members trend into the calc-alkaline.
|
We feel that trace element and isotope data provide a rationale for selecting calc-alkaline affinity for the Soufriere rocks. Figure 13, for example, presents MgO–(87Sr/86Sr)i relationships for low-K tholeiites, calc-alkaline and C- and M-series rocks from the arc. The Soufriere data, with almost constant 87Sr/86Sr and a large range in MgO contents, are clearly distinct from the tholeiitic and M- and C- series rocks but overlap with the calc-alkaline field and extend it towards more more primitive compositions. Trace element ratios, such as Nb/Yb and Ba/La (not shown) also distinguish low-K tholeiites and Soufriere eruptives. We suggest, therefore, that to distinguish them from the tholeiitic rocks of the northern islands, the rocks from Soufriere are best referred to as having calc-alkaline character, and that some may represent the mafic end-member of the calc-alkaline magma series in the Lesser Antilles.
|
The Soufriere rocks are close compositionally to a suite of lavas in the neighbouring island of Bequia named the isotopically homogeneous suite (IHS) by Smith et al. (1996)
, who pointed out that the similarities suggest that the two suites were derived from similar sources and have similar evolutionary histories.
Major and trace element variations
Major and selected trace element variations in Soufriere rocks are displayed on MgO plots in Fig. 14. Six Yellow Tuff Formation samples, five of which were separated from palpably mixed rocks, define a linear, mixing, trend. The end-members are a basalt with MgO = 10 wt % and a silicic andesite (MgO = 2 wt %), which represents the most evolved rock erupted from the Soufriere. In the other rocks, SiO2, Na2O, K2O, P2O5, Ba, Rb, Sr, Th, U, Zr, Hf and rare earth elements (REE) increase, and CaO, Fe2O3*, Co, Cr, Ni, Sc and V decrease, with decreasing MgO. The distribution of TiO2, and possibly also Cs and Nb, is more complex and seems to show a decrease to 6 wt % MgO and then an increase with further decrease in MgO. A rather similar TiO2 pattern seems to be shown by the IHS on Bequia (Smith et al., 1996, fig. 6). V abundances in the IHS, however, increase with increasing differentiation.
Chondrite-normalized REE patterns (Fig. 15) are gen- erally flat to slightly light REE (LREE) enriched ([La/Yb]N 0.9–2.2), with weak Eu anomalies which range from slightly positive in the most magnesian basalts (Eu/Eu* 1.07) to slightly negative in the basaltic andesites ( 0.96) and andesite (0.89). With one exception, the rocks show a Ce anomaly, Ce/Ce* ranging from 0.87 to 0.98. Flat patterns like those from Soufriere have previously been recorded from calc-alkaline basalts and basaltic andesites from St Lucia and Guadeloupe, the low-K tholeiites of St Kitts and certain silica-undersaturated basalts of Grenada (White & Dupré, 1986; Thirlwall et al., 1994).
With decreasing MgO content of Soufriere whole rocks, LREE/MREE (medium REE) (e.g. La/Sm) and LREE/HREE (heavy REE) (La/Yb) ratios decrease to rocks with MgO around 3 wt %, with a sharp increase into the andesite STV 376(L) (Fig. 16). This is unusual behaviour in a suite of arc rocks, where LREE enrichment with increasing differentiation is the norm, as for example, in the IHS of Bequia (Smith et al., 1996). There is, however, apparent LREE depletion in more differentiated members of the isotopically diverse suites (IDS) of lavas on Bequia.
|
Figure 17 presents chondrite-normalized element abundance plots for three basalts with MgO >7 wt %. The patterns are characteristic of arc magmas, namely, high LILE/HFSE (large ion lithophile elements/high field strength elements) and LREE/HFSE ratios, Nb depletion relative to La, and relatively high Ba/La and Sr/Nd ratios. The LREE sections of the patterns are relatively flat, their abundance varying from 14 to 18 times chondritic. The HFSE and HREE sections are also rather flat, with 8–12 times enrichment over chondrites. The rocks are enriched in the LILE (Rb, Th, K and Ba), but not Sr, compared with the LREE. Average MORB is also plotted in Fig. 17. As is typical of arc magmas, the Soufriere rocks show LILE and LREE enrichment and HFSE depletion relative to MORB.
|
Isotopic data
Figures 18 and 19 show Sr–Nd–Pb isotopic variations in Lesser Antilles rocks, including our new data from Soufriere St Vincent, and also fields for Atlantic Ocean sediments and typical MORB. 87Sr/86Sr ratios (0.703752–0.704407) of Soufriere rocks are somewhat higher than those of MORB, whereas 143Nd/144Nd ratios (0.512976–0.512843) are slightly lower. The new Pb data extend the range established by previous workers (White & Dupré, 1986
; M. F. Thirlwall, unpublished data figured in Smith et al., 1996) to slightly less radiogenic compositions (206Pb/204Pb = 18.644–19.328). On Pb–Pb plots (Fig. 19), the Soufriere data (and the closely similar IHS of Bequia, not shown) form linear trends which extrapolate to intersect the fields of MORB and locally subducting sediments, as determined from Deep Sea Drilling Project (DSDP) Hole 543. This strongly suggests that at least two components have been involved in their petrogenesis—depleted mantle (MORB) and a subduction component at least partly derived from the sediments (White & Dupré, 1986; Smith et al., 1996).
|
|
The relative homogeneity of the Soufriere and Bequia IHS rocks compared with calc-alkaline rocks of the central islands (St Lucia, Martinique, Dominica) is mainly a function of MgO content; isotopic diversity within series is restricted to rocks with MgO <5 wt % (Fig. 13). However, at given MgO, Soufriere St Vincent and other calc-alkaline rocks of the Lesser Antilles arc typically have higher 87Sr/86Sr, 206Pb/204Pb and 207Pb/204Pb ratios, and lower 143Nd/144Nd than the tholeiitic lavas of the northern islands. The M- and C-series lavas of Grenada typically have higher 87Sr/86Sr and lower 143Nd/144Nd at given MgO than the calc-alkaline rocks. As these features characterize even the most primitive rocks, they are likely to indicate isotopically distinct mantle sources.
Magmatic lineages
As a result of scatter within the rocks of each formation, it is difficult to distinguish between the different formations at Soufriere on the basis of major or trace element compositions. There are, though, some minor differences. For example, the Pre-Somma lavas generally have slightly higher K2O abundances at given MgO than the Crater Lavas and rocks of the Pyroclastic Formation (Fig. 14a). On a more detailed scale, Graham & Thirlwall (1981) noted that the 1971 lava is depleted in Al, Ca and Sc, enriched by 10% in Na, K, Rb, Zr and Zn, and higher in Cr, compared with the 1979 lava. They suggested that the lavas represent two entirely different magma batches. Magnesian basalt STV 301 has higher Ti, K, P, La, Nb, Th and U, and lower Ni, abundances, and lower Rb/K ratios, than Pre-Somma rocks of similar MgO content. In contrast, STV 315 (MgO = 7.8 wt %) has low abundances of La, Sr and Th compared with rocks of similar MgO content. As there are no systematic correlations between trace element ratios, or between them and any isotopic ratio, we infer that the mantle sources were compositionally heterogeneous on a small scale.
Thus, although we discuss the Soufriere rocks as describing a cogenetic suite, in reality they probably represent many different liquid lines of descent, a point already made for the recent pyroclastic deposits of Mt Pelée, Martinique (Smith & Roobol, 1990) and the M-series rocks of Grenada (Devine, 1995; Thirlwall et al., 1996).
A few generalizations may be made concerning the distribution of rock types in time. The most primitive lavas (SiO2 <50 wt %) are restricted to the Pre-Somma Formation and the average composition of Pre-Somma rocks is accordingly less SiO2 rich than younger rocks. The widest compositional range, basalt to andesite, occurs in the Yellow Tuff Formation and we can infer that the magma reservoirs were more mature by that stage. More recent activity (the Crater Lavas and Pyroclastic Formation) has been dominated by basaltic andesites but eruption of basalt during the compositionally zoned 1902 eruption suggests that basalt magma has been available at depth in the system, although normally unable to erupt through more evolved overlying magmas.
|
Melt Inclusions |
|---|
|
TOPABSTRACTIntroductionGeological HistoryPetrography, Mineralogy and...GeochemistryMelt InclusionsPre-Eruptive Water Contents of...DiscussionReferences |
|---|
Graham & Thirlwall (1981)
, Devine & Sigurdsson (1983) and Bardintzeff (1992) provided compositional data on melt inclusions in products of the 1979 eruption. We add here (Table A6) electron microprobe data for a further eight rocks, including representatives of all the major stratigraphic units. The following discussion incorporates material from the earlier papers.
|
Melt inclusions are found in all the main phenocryst phases. Compositions range from basalt (47 wt % SiO2) to high-silica rhyolite (77 wt % SiO2). A considerable part of the range may be found in the products of single eruptions. Thus, whole-rock analyses of the products of the 1979 eruption show very little major element variation (Graham & Thirlwall, 1981
, table 1), whereas melt inclusions in phenocrysts range in SiO2 content from 48.6 to 62.0 wt % (Devine & Sigurdsson, 1983, table V).
The melt inclusions extend to high-silica compositions unknown among the eruptive products of Soufriere and relatively uncommon in the Lesser Antilles arc. It is also important to note that inclusions with much higher silica contents than the host rocks occur in every sample analysed, e.g. inclusions of rhyolitic composition occur in basalt STV 315. It is possible that melt compositions in such cases have been significantly modified by post-entrapment crystallization. This effect is likely to have been small at Soufriere, however. Graham & Thirlwall (1981) noted, for example, that the melt inclusions in olivine do not show the distribution of compositions along an olivine control line expected from further crystallization of olivine after inclusion. Similar comments can be made for the absence of plagioclase control among inclusions in plagioclase phenocrysts (see Devine & Sigurdsson, 1983, fig. 2). From a detailed analysis of differences in inclusion composition in different phenocryst hosts, Devine & Sigurdsson (1983) concluded that, on average, no more than 8 wt % of any inclusion could have precipitated out of the host mineral during quenching. These observations, plus the absence of optical and electron probe evidence for crystallization of the inclusion rinds, lead us to conclude that post-entrapment crystallization has not significantly altered the compositions of the melt inclusions.
The following observations suggest, on the other hand, that the inclusions represent trapped liquids which were cognate to the host phenocrysts:
- There is an overall positive correlation between the SiO2 contents of inclusions and host rocks (Tables A2 and A6).
- There is also a relationship between inclusion composition and the order of crystallization of phenocryst phases. Thus, olivine phenocrysts tend to have inclusions which are more magnesian than those in plagioclase, whereas those in orthopyroxene are almost exclusively more evolved (Devine & Sigurdsson, 1983; Bardintzeff, 1992).
- Major element variations in the inclusions mimic those in the whole rocks (Fig. 10).
The matrix compositions of Soufriere basaltic andesites and andesites are invariably silicic (Graham & Thirlwall, 1981; Devine & Sigurdsson, 1983; Bardintzeff, 1992) and in the only rocks for which melt inclusion and matrix glass data are available, those of the 1979 eruption, the melt inclusions extend to much more magnesian compositions than the matrix glass (e.g. Devine & Sigurdsson, 1983, table V). We suggest, therefore, that growing phenocrysts incorporated increasingly evolved residual melts.
In any given Soufriere sample, the highest Cl abundances are found in melt inclusions in clinopyroxene phenocrysts. If we take these to be closest to magmatic Cl abundances, they indicate contents mainly between 0.2 and 0.3 wt % and up to 0.6 wt %. Such values seem to be rather high for arc magmas; for example, Perfit et al. (1980) reported a range of 460–2000 ppm Cl in calc-alkaline magmas, and Sisson & Layne (1993) found, by ion microprobe, Cl abundances <0.16 wt % in melt inclusions in rocks of the Fuego 1974 eruption, Guatemala. Devine (1995) reported, however, Cl abundances between 0.34 and 0.38 wt % in melt inclusions in M-series lavas from Grenada. Relatively high Cl contents may be a feature of Lesser Antilles magmas.
|
Pre-Eruptive Water Contents of the Magmas |
|---|
|
TOPABSTRACTIntroductionGeological HistoryPetrography, Mineralogy and...GeochemistryMelt InclusionsPre-Eruptive Water Contents of...DiscussionReferences |
|---|
In this section, we estimate the pre-eruptive water contents of Soufriere magmas using three techniques: (1) the ‘difference’ method from published electron probe analyses; (2) comparison with phase equilibrium experimental results; (3) clinopyroxene compositions.
The difference method
The difference between 100% and the analytical total for melt inclusion compositions derived from electron probe analysis may be used as an indicator of the volatile content of the inclusions (Anderson, 1979). Devine & Sigurdsson (1983) and Bardintzeff (1992) have applied the technique to products of the 1979 eruption of Soufriere, the former workers giving a detailed analysis of the potential problems of interpretation. They concluded that the difference method can give reliable estimates of volatile content (inferred to be dominated by water), a conclusion supported by Sisson & Layne (1993), who showed that summation deficits in inclusions from various subduction-related volcanoes agree with the water contents determined by ion microprobe with an average error of ±0.65 wt %.
Devine & Sigurdsson (1983) reported that the volatile content of olivine-hosted inclusions in the 1979 Soufriere rocks ranged from 2 to 8 wt %, averaging 2.9 wt %. In contrast, the volatile content of inclusions in plagioclase and pyroxenes was probably <2 wt %, suggesting that the more evolved magmas had degassed before eruption. Bardintzeff (1992) also found a negative, but very scattered, correlation between inferred water content and SiO2 content of the glasses in 1979 products. Rather than using data from inclusions in phenocrysts as indicators of water contents in the basaltic magmas, he chose instead data for mafic interstitial glasses in cumulate blocks, giving an average of 2.5 wt %, comparable with the Devine & Sigurdsson estimate. For more evolved rocks, Bardintzeff (1992) chose, somewhat arbitrarily, a range of 3–5 wt %, with an average of 4 wt %, i.e. water contents increased with differentiation.
Phase equilibrium studies
We now compare magma compositions with experimentally determined phase equilibria in the system olivine (Ol)–high-Ca pyroxene (Cpx)–plagioclase (Pl)–silica (Qz). This analysis is based mainly on the data and discussion of Sisson & Grove (1993a, 1993b), who determined the phase relations of a series of natural aphyric high-alumina basalts and intrusive equivalents through melting experiments at 2 kbar, under water-saturated conditions and with f(O2) buffered at NNO (nickel-nickel oxide), i.e. at somewhat less oxidizing conditions than prevailed at Soufriere.
Figure 20 is the pseudoternary Ol–Cpx–Pl, projected from Qz–magnetite (Mt)–orthoclase (Or)–apatite (Ap). It shows the hydrous 2 kbar multiple saturation boundary for ol + cpx + pl, deduced from experiments on natural basalts and projected back towards the piercing point in the forsterite–diopside–anorthite system. Sisson & Grove (1993a) noted that the compositions of high-alumina basalts, basaltic andesites and andesites from the Aleutians and Fuego volcano in Guatemala plot close to the water-saturated multiple saturation boundary, and suggested that the close correspondence indicates that the magmas had pre-eruptive water contents comparable with those of the experimental liquids (up to 6 wt %).
|
Basalts and basaltic andesites from Soufriere St Vincent are also plotted in Fig. 20. A first interpretation would be that the series evolved by ol + cpx crystallization to the hydrous 2 kbar multiple saturation boundary, where plagioclase also began to crystallize. Such an interpretation is not, however, totally supported by petrographic evidence. Almost all the Soufriere suite is plagioclase-phyric, not just those plotting on the multiple saturation boundary.
Similar discrepancies are seen in other projections in the same system. Figure 21 is a projection from Pl–Mt–Or–Ap onto Ol–Cpx–Qz. Liquids saturated with ol + high-Ca cpx + pl (± spinel) define a multiple saturation boundary which projects from the silica-poor side of the Ol–Cpx sideline to the Ol–Qz sideline. Sisson & Grove (1993a) noted that, as in the last projection, there is a close correspondence between the position of the boundary and the fields of high-alumina basalts and andesites from the Aleutians and Fuego volcano, and again took this as evidence that the natural magmas had evolved under water-rich conditions. Figure 21 also shows the multiple saturation boundary defined by liquids saturated with ol + pl + hb (± mt), produced in melting experiments on hornblende-bearing intrusive rocks (Sisson & Grove, 1993a). The boundary in this case is a reaction boundary along which olivine and liquid react to produce hornblende. When all the olivine is consumed, the liquids follow a pl + hb + mt coprecipitation trend towards the Qz apex.
|
Whole-rock data from Soufriere are projected onto the Ol–Cpx–Qz plane in Fig. 21. They start on the silica-poor side of the Ol–Cpx sideline, cross the multiple saturation boundary and then trend towards the Qz apex. The more evolved part of the trend thus appears similar to the hb + pl + mt control line. There are at least two problems with this interpretation: (1) amphibole is not a phenocryst phase in less evolved rocks, as would be required if the liquid trend were controlled by pl + hb crystallization; and (2) Sisson & Grove (1993a
) confirmed the results of Kushiro (1969) that high contents of dissolved water destabilize low-Ca pyroxene relative to olivine, and they suggested that low-Ca pyroxene will not crystallize from water-rich basaltic or basaltic andesite melts. At Soufriere, however, orthopyroxene starts crystallizing from basaltic andesite liquids whose projected compositions lie on the multiple saturation boundary ol + cpx + pl + liquid. Crystallization of the assemblage ol + pl + cpx + opx + Ti-mag then drove liquids down towards the quartz apex. Thus, although the broad spread of analyses around the 2 kbar multiple saturation boundary is consistent with high water contents in the Soufriere magmas, the early crystallization of orthopyroxene points to water-poor conditions.
This lack of correspondence between projected composition and phenocryst assemblage has been noted by Feeley & Davidson (1994), who, in a study of calc-alkaline lavas and magmatic inclusions of the Volcán Ollagüe in the central Andes, found that basaltic andesite lavas scatter around 2 kbar ol + hb + pl and ol + cpx + pl saturation boundaries in the Ol–Cpx–Qz pseudoternary, despite the facts that they are not plagioclase or amphibole saturated and that olivine phenocrysts are rare or absent in the magmatic inclusions. Those workers suggested that the correspondence is coincidental, the trend probably reflecting crustal assimilation or mixing with or without fractional crystallization.
We now use an alternative set of phase relationships. Figure 22 shows pseudoternary phase diagrams in the system Ol–Di–Pl–Mt–SiOr, where SiOr is a ‘magmaphile’ component consisting of combined normative quartz and orthoclase. Phase relationships were determined from a suite of high-alumina basalts and andesites from Atka island in the Aleutians (Baker & Eggler, 1987). Experimental conditions varied from anhydrous, at pressures from 1 atm to 8 kbar, to hydrous (2 wt % H2O in the melt) at 2 and 5 kbar.
|
Soufriere rocks are plotted using the algorithm of Baker & Eggler (1983)
, with Fe2O3/FeO = 0.22. In the projection from Di–Mt (Fig. 22a), presented by Baker & Eggler (1987) as an effective geohygrometer, the basaltic andesites plot close to the 2–5 kbar hydrous liquid lines of multiple saturation (LLMS) but the basalts clearly do not lie on any reasonable extrapolation of the trends. The phase relationships seem to point to relatively dry basaltic magmas at Soufriere. In the Pl–Mt projection (Fig. 22b), which is more useful as a geobarometer (Baker & Eggler, 1987), the basaltic andesites and andesites of Soufriere St Vincent lie close to the 5 kbar hydrous LLMS for ol + pl + aug and pl + aug + opx. The distribution is consistent with an origin from basaltic parents by fractionation of ol + pl + aug, to produce derivative melts with 2 wt % H2O. The array of Soufriere basalts is again more consistent with crystallization under water-poor conditions, at pressures exceeding 8 kbar.
Clinopyroxene compositions
Experiments by Gaetani et al. (1993) on the crystallization of basalts containing up to 6 wt % dissolved water have provided some evidence on the effects of water on the compositions of high-Ca pyroxenes. The effect of increasing pressure up to 15 kbar in anhydrous systems is to decrease the Wo content of clinopyroxene while increasing (CaTs + CrTs), the Al- and Cr-rich components (Fig. 23). In the experiments of Gaetani et al., conducted at 2 kbar, the effect of 6 wt % dissolved water was to increase the Wo content of the clinopyroxenes at nearly constant CaTs + CrTs. The fields of clinopyroxenes in ol- and cpx-phyric lavas from Soufriere and from selected arcs are shown in Fig. 23. The similarity to the clinopyroxenes produced experimentally at 2 kbar led Gaetani et al. (1993) to suggest that the arc lavas had evolved through hydrous crystallization at crustal pressures. A similar conclusion can be made for the Soufriere rocks, despite the fact that they probably equilibrated at higher pressures.
|
In summary, the various methods for estimating magmatic water contents have given conflicting results. Clinopyroxene compositions are consistent with water contents exceeding 3 wt %, whereas the early crystallization of plagioclase and orthopyroxene is more consistent with relatively anhydrous basaltic magmas generating hydrous, but water-undersaturated, andesitic magmas. The absence of amphibole phenocrysts may also point to water-undersaturated conditions. This is unlikely to be due solely to a compositional effect, as many Soufriere basaltic andesites contain >3 wt % Na2O, the level at which amphibole should be stabilized, given high enough water contents (Cawthorn & O'Hara, 1976
; Sisson & Grove, 1993a). Given these uncertainties, it is still unclear whether magmatic evolution at Soufriere was accompanied by continuous (or episodic) degassing, or by increasing water contents in residual melts.
Magmatic evolution
Many previous studies of the evolution of primary magmas in Lesser Antilles volcanic suites have recognized the polybaric nature of the fractionation processes and the role on many islands of crustal assimilation (Thirlwall & Graham, 1984; White & Patchett, 1984; Davidson, 1985, 1986, 1987; White & Dupré, 1986; Davidson & Harmon, 1989; Smith & Roobol, 1990; Bardintzeff, 1992; Thirlwall et al., 1994, 1996; Smith et al., 1996; Turner et al., 1996). Below, we comment on the nature of the primary magmas, assess the degree of crustal contamination in the Soufriere rocks and then attempt to model magma evolution by fractional crystallization.
First, we should comment on the role of magma mixing in the evolution of the Soufriere suite. As noted in the Petrography section, macroscopically mixed rocks have been found at a few localities on Soufriere and this must raise the possibility that other members of the suite represent more thoroughly hybridized materials. The spread of analyses in MgO variation diagrams (Fig. 14) would allow for this possibility. We note, however, the following points:
- Reverse zoning in phenocrysts is, however, rare to absent at Soufriere, except as part of oscillatory zoning in plagioclase. This is not consistent with incorporation into the magmas of large volumes of melt of different temperature.
- The non-linear nature of compositional variation within the suite (e.g. MgO–Na2O and MgO–CaO, Fig. 14a) is more consistent with evolution by fractional crystallization.
- There is an overall correlation between phenocryst and whole-rock compositions.
Although, therefore, there may have been a minor role for mixing, or for heat transfer from new influxes of basalt lower into the magma reservoirs, we suggest that magma mixing has not been a major differentiation process at Soufriere.
Primary magma compositions
Identification of primary magmas is critically dependent on the composition chosen for the mantle source rocks, in particular the mg-number of the olivine. Some simple whole-rock compositional criteria are widely used, including mg-number >70, FeO*/MgO <1, Ni >200 ppm and Cr >400 ppm (Tatsumi & Eggins, 1995). These criteria are met by the more magnesian lavas of the Pre-Somma group (e.g. STV 301, 310 and 358) and we have already shown that olivine phenocrysts in STV 301 were in (near-)equilibrium with the whole-rock composition. These rocks may therefore be close to primary magma compositions. It has already been inferred that the parental rocks of St Vincent suites and of the IHS of Bequia contained >12 wt % MgO (Smith et al., 1996), although normative compositions (Fig. 11) may indicate that the melts were nepheline-normative. In comparison, Thirlwall et al. (1996) argued that all Grenada magmas were ultimately derived by fractional crystallization from undersaturated picrites with 16 wt % MgO.
Degrees of partial melting
Geochemical studies have established that there have been three components in the mantle sources of Lesser Antilles magmas: (1) the mantle wedge, which is similar to, or slightly enriched in HFSE relative to, the N-MORB source; (2) a fluid phase added to the wedge from the subducting slab, with contributions from both sediments and altered basaltic crust; (3) a component formed by partial melting of subducted sediments (Thirlwall & Graham, 1984; White & Patchett, 1984; Davidson, 1986, 1987; White & Dupré, 1986; Davidson & Harmon, 1989; Thirlwall et al., 1994, 1996; Smith et al., 1996; Turner et al., 1996; Hawkesworth et al., 1997). Although it is accepted that the proportions of each component vary along the arc, there is as yet no consensus on how the variation relates to individual islands.
There are major difficulties associated with estimating degrees of partial melting of mantle sources, related, inter alia, to uncertainties about mantle composition and the effects of water content on phase equilibria. For example, Stolper & Newman (1994) have suggested, in a study of glasses from the Mariana trough, that a 0.2 wt % increase in water content in the source of the melts caused the degree of melting to rise by some 12 wt %.
In the case of the Lesser Antilles, La/Yb has been used qualitatively by Thirlwall et al. (1994) as a measure of degree of partial melting, higher ratios signifying smaller degree melts. Those workers suggested that St Vincent rocks, on the basis of relatively low La/Yb ratios, represent fairly high-degree melt fractions, which they saw as consistent with their transitional tholeiitic–calc-alkaline character.
Pearce & Parkinson (1993) have used an Nb–Yb plot to study the melting process in arcs, and argued that the element–element plot allowed them to separate source depletion from degree of melting. To minimize the effects of fractional crystallization, which may be significant for HFSE even in magmas with MgO between 10 and 6 wt % (Thirlwall et al., 1994), Pearce & Parkinson (1993) recommended normalizing Nb and Y values to an MgO content of 9 wt %. Relevant values for Soufriere rocks are Nb 3 ppm and Yb 2.2 ppm, which suggests a degree of partial melting of 15%. Given the rather dubious assumption in such plots that a given MgO content signals equal amounts of fractionation of the primary magmas (Thirlwall et al., 1994), the value of 15% can be taken as no more than an indicator of the degree of melting.
An important result from the study by Thirlwall et al. of HFSE anomalies is that there is no evidence for residual minor phases in the source mantle of the Soufriere St Vincent magmas. For example, near-constant Ti/Eu, Zr/Sm and Nb/La ratios during partial melting, as inferred from La/Yb ratios, preclude a role for residual amphibole, as amphibole should fractionate HFSE from REE. If such phases existed in the mantle sources, they disappeared during the melting process.
Crustal contamination
Various geochemical criteria have been used to assess the effects of crustal contamination during magma evolution in the Lesser Antilles. These include increases, with advancing differentiation, in 87Sr/86Sr and 206Pb/204Pb, decreases in 143Nd/144Nd, and increases in Rb/Ba, Rb/K, K/La, La/Yb, Pb/Sr and U/La (Thirlwall & Graham, 1984; White & Dupré, 1986; Davidson, 1987; Davidson & Harmon, 1989; Smith et al., 1996; Thirlwall et al., 1996).
Sr and Nd isotopic ratios in Soufriere rocks have only a limited range, are close to MORB in composition and show no systematic change with degree of fractionation (Figs 13 and 18). This suggests that the magmas evolved by fractional crystallization in a closed system. It is also possible that open system assimilation–fractional crystallization (AFC) occurred where the magmas and assimilant(s) had similar isotopic characteristics. In this regard, we note that some of the xenoliths analysed during this study are isotopically indistinguishable from the lavas (Table A5). However, incompatible trace element ratios in evolved rocks show the same range as the most magnesian basalts, as demonstrated, for example, by Rb/K–SiO2 relationships (Fig. 24; after Davidson, 1987), suggesting that the range has been derived from the mantle source(s). The spread in the data means that we cannot preclude a contamination component, but we equally can conclude that there is no compelling evidence that the Soufriere magmas experienced significant crustal contamination.
|
Modelling fractionation trends
Major element variations (except P) within the suite have been modelled using the Mix ’n’ Mac petrological mixing program by D. R. Mason, which is based on the least-squares method of Bryan et al. (1969)
. Sums of squared residuals (r2) should be <0.2 for an acceptable solution. Modelling was done in three steps: magnesian basalt STV 310 to basalt STV 334 to basaltic andesite STV 79–70 to andesite STV 376(L). The whole-rock compositions are joined by continuous lines in Fig. 14. The mineral data used in the modelling were selected to reflect typical equilibrium compositions with respect to the parent magma at each stage. Input Fe2O3/FeO ratios varied with the silica content of the whole rock (Middlemost, 1989).
The best solutions (i.e. those with low r2 and reasonable degrees of crystallization) were achieved by fractionating 14% ol + 8% cpx + 14% plag + 1% Ti-mag ± 0.1% Cr-sp from basalt STV 310, then 8% ol + 10% cpx + 9% plag + 0.5% Ti-mag from basalt STV 334, followed by 7% cpx + 26% pl + 4% Ti-mag + 7% opx from basaltic andesite STV 79–70. These are the observed phenocryst assemblages, and the relative proportions of fractionating phases are similar to those found in the rocks (Table A1). Given the range of phenocryst compositions and proportions in the rocks and the fact that the suite comprises many liquid lines of descent, such models cannot be considered robust [a point made by Devine (1995) in modelling the evolution of Grenada rocks]. Nevertheless, the results are at least consistent with evolution of the Soufriere suite by closed-system fractionation of the observed phenocryst phases, the most evolved rock representing some 76 wt % crystallization of the parental basalt.
We have not attempted to model the derivation of the rhyolitic melt inclusions from andesitic compositions. However, Devine & Sigurdsson (1983) found that the interval 60–75 wt % SiO2 could be satisfactorily modelled using assemblages close to 5% ol + 13% cpx + 74% pl + 6% Ti-mag + 2% opx. Because of the reaction relationship between them, it is unlikely that olivine and orthopyroxene coprecipitated in the fractionating assemblage. The important observation, though, is that compositional variation in the suite can apparently be explained by removal or addition of the observed phenocryst phases.
Apart from some aspects of REE distribution, we have not attempted to model trace element variations in the Soufriere rocks because of the shortage of partition coefficient data for the phenocrysts. Trace element distributions are, however, qualitatively consistent with fractional crystallization of the observed phases. The rapid decrease of Ni and Cr, and the slower decrease in Co and Sc, with slowly increasing Zr is typical of fractionation of olivine and clinopyroxene, a point made for the IHS of Bequia by Smith et al. (1996). The systematic increases in ITE from most to least magnesian rocks are also consistent with an evolution by fractional crystallization. The enrichment factors for the elements most likely to be strongly incompatible, such as Ba, Cs and Rb, are some 4–5 (Fig. 14b), which is consistent with the 76 wt % crystallization figure found in the major element modelling.
Although trace element distribution is generally consistent with evolution of the suite by closed-system fractional crystallization, the LREE depletion relative to MREE and HREE during fractionation from basalts to basaltic andesites creates a problem (Fig. 15). We have modelled variations in La/Yb ratios by Rayleigh fractionation, using, as proxies for phenocryst partition coefficient, data for mineral phases (and glass) separated from a Soufriere cumulate xenolith by Dostal et al. (1983). The fractionating assemblages are those estimated from major element modelling. The calculated trend shows the expected LREE enrichment, from [La/Yb]N = 1.63 at 12.5 wt % MgO to 1.80 at 2.3 wt % MgO, and does not provide a ready explanation for the observed LREE depletion.
We noted earlier that the Soufriere rocks represent several magmatic lineages. Rather than a crystal fractionation control, the LREE depletion in more evolved rocks may reflect the fact that these rocks were derived from parental magmas with lower LREE/HREE ratios than the magnesian rocks in our analysed suite. This might also explain the TiO2 decrease, followed by an increase, in increasingly evolved rocks.
A role for amphibole?
Amphibole is a common phenocryst phase in several Lesser Antilles suites, where it is restricted to evolved rocks (andesites and dacites), e.g. the low-K tholeiitic series of Mt Misery (now Mt Liamuiga) volcano, St Kitts (Baker, 1968), the calc-alkaline rocks of Montserrat (Rea, 1974) and Mt Pelée, Martinique (Smith & Roobol, 1990) and the M-series association of Grenada (Devine, 1995). It has not, however, been found as a phenocryst phase in any Soufriere St Vincent rock. The occurrence of amphibole in cumulate blocks and as occasional xenocrysts in the 1979 eruption products may, however, be evidence of high magmatic water contents at depth, and the question is raised whether all Soufriere magmas went through a higher-pressure phase of amphibole crystallization, which was followed by amphibole resorption during ascent. Compositional variation in the eruptive suite may therefore reflect a cryptic amphibole control, as has been suggested for other calc-alkaline suites (Foden & Green, 1992; Miller et al., 1992; Romnick et al., 1992).
Amphibole stability is dependent on temperature. Even at high pressures and water contents, amphibole is unlikely to be stable above 1000°C (Helz, 1982; Allen & Boettcher, 1983; Spulber & Rutherford, 1983), with the possible exception of fluor-richterites, which might be stable above 1200°C (Arculus, 1994). The basalts and basaltic andesites of Soufriere St Vincent may simply have been too hot (T 1050°C) to crystallize amphibole as a near-liquidus phase.
Amphibole stability is also sensitive to melt composition. Cawthorn & O'Hara (1976) and Cawthorn (1976) showed that in the system CMAS–Na2O–H2O at pH2O = 5 kbar, hydrous basaltic melts needed 3 wt % Na2O to crystallize pargasitic amphibole. Sisson & Grove (1993a) confirmed, from melting experiments on natural high-alumina basalts at 2 kbar (water saturated), that hornblende stability is sensitive to Na2O content, as well as H2O content, of the melt, and pointed out that although hornblende can form as a (near-)liquidus phase in hydrous, Na-rich magmas, it will not be stable in moderate-Na systems until melt compositions are andesitic. Na2O concentrations 3 wt % tend to be restricted, in the Lesser Antilles, to rocks with 5 wt % MgO; the restriction of amphibole to more evolved rocks is thus in line with the experimental evidence.
In addition to the experimental evidence, we showed earlier that production of basalts with MgO 8 wt % from more magnesian (MgO = 12.5 wt %) parental basalts at Soufriere could be satisfactorily modelled using combinations of the crystallizing assemblage ol + cpx + pl + Cr-sp + Ti-mag + opx. Nevertheless, an equally good fit (r2 = 0.03), with about the same inferred degree of crystallization, is achieved using the assemblage 10% ol + 4% cpx + 11% pl + 0.1% Cr-sp + 17% hb, which is matched by assemblages in cumulate blocks (Arculus & Wills, 1980). However, adding amphibole to the high-pressure crystallizing assemblage has little effect on trace element distribution, as likely partition coefficients, except for Ba and Rb, are similar to those for clinopyroxene (Dostal et al., 1983). In particular, amphibole does not provide a solution to the problem of LREE depletion relative to HREE, as partition coefficients for HREE are much larger than those for LREE (Sisson, 1994).
We suggest that amphibole was not involved in the evolution of the Soufriere suite for the following reasons:(1)there is a compositional and mineralogical continuum from magnesian basalts, last equilibrated in the mantle, to andesites; amphibole is not present as a phenocryst phase;(2)major element modelling indicates that compositional variation in the suite can be explained without recourse to amphibole fractionation;(3)the magmas may simply have been too hot and too Na poor to stabilize amphibole.
|
Discussion |
|---|
|
TOPABSTRACTIntroductionGeological HistoryPetrography, Mineralogy and...GeochemistryMelt InclusionsPre-Eruptive Water Contents of...DiscussionReferences |
|---|
A summary of the evolutionary sequence of Soufriere magmas is shown schematically in Fig. 25. By analogy with magmatism on Grenada, the most primitive magmas may have been derived from more olivine-rich (picritic) magmas; however, their compositions are equally compatible with their being primary magmas in their own right. These most primitive Soufriere magmas may have been generated by
15% partial melting of a mantle source which resembled the N-MORB source before addition of a subduction-related component which contained contributions from subducted sediments and basaltic crust. The most primitive magmas equilibrated with mantle at 50–60 km depth; temperatures were around 1130° and f(O2) was more oxidizing than FMQ + 1. A period of olivine + spinel ± clinopyroxene fractionation was followed by the appearance of plagioclase as a phenocryst phase. The pre-eruptive water contents of these magmas are uncertain but the levels were insufficient to prevent relatively early crystallization of plagioclase, such that the assemblage ol + sp + cpx + plag is found in rocks with MgO >10 wt %. Fractionation of this assemblage resulted in modest increases in Al2O3 content and FeO*/MgO ratios of residual liquids, that is, in a trend transitional between tholeiitic and calc-alkaline.
|
In the early stages of Soufriere's evolution, some of the more primitive magmas reached the surface relatively unmodified; rocks with the highest mg-numbers are restricted to the earliest stratigraphic unit. As the plumbing system matured, magmas tended to reside for longer periods at the base of, and within, the crust and the most recent products have tended to be basaltic andesites (T
1050°C). Judging from our pressure estimates (Figs 7 and 8), magma reservoirs existed at depths of 30 km (equivalent to 1 GPa) when ol + sp + cpx + plag were the fractionating phases and magma compositions evolved through towards basaltic andesite. The depth of higher-level magma reservoirs is difficult to determine without an accurate knowledge of magmatic water contents. However, basaltic andesites may have resided at mid-crustal depths (10–15 km; Figs 7 and 8). This estimate is consistent with ground tilt measurements which, over a 12 year period including the 1979 eruption, were interpreted to point to the presence of a magma chamber at a depth of >10 km (Fiske & Shepherd, 1982, 1990). Evidence relating to the water contents of the (basaltic) andesite magmas is equivocal but is consistent with the possibility that they were hydrous but water undersaturated.
There is no isotopic or trace element evidence for significant crustal contamination of the Soufriere magmas, despite the ubiquity of the process in other calc-alkaline suites in the Lesser Antilles. The efficiency of AFC processes is related to such factors as the rate of magma supply, the temperature gradient between magma and wall rocks, and the thickness of crystal mush on the chamber walls and roof. Soufriere eruptive products carry unusually abundant cumulate nodules (Arculus & Wills 1980), which may indicate thick, marginal crystal deposits. These in turn may have armoured the magmas from reaction with, and contamination by, the country rocks. The development of thick marginal facies may be a result of low magma supply rates at Soufriere, which would be consistent with the relative rarity of mixed magma rocks. Data for volumetric volcanic production for the last 0.1 my show that the islands with the strongest isotopic evidence for crustal contamination have had the highest volumetric production (>5 km3; Wadge, 1986, fig. 10).
In addition, volcanic activity may have started on St Vincent much more recently than on neighbouringislands; in contrast to St Lucia, Martinique, Grenada and many of the Grenadine islands, St Vincent lacks exposed Miocene rocks. The lack of deposits older than 2.8 Ma may simply reflect incomplete sampling or complete burial by post-Pliocene volcanic deposits. However, there is some support for the relative youth of St Vincent from seismic data, as discussed by Wadge (1986). The seismic refraction horizon between the upper- and middle-crustal layers rises beneath most of the Lesser Antillean islands, which has been interpreted to reflect large volumes of intruded material. Wadge (1986) suggested that large andesitic plutons underlying the islands usually intercept rising basaltic magmas, which then evolve towards andesitic and dacitic compositions through AFC and mixing. However, the seismic boundary is flatter and less distinct beneath St Vincent, which may indicate a briefer history of magmatism, and less well-developed magma reservoirs, which would be con- sistent with the relative abundance of basaltic rocks and limited amounts of mixing and contamination.
We suggest that the parental magmas to other calc-alkaline suites in the Lesser Antilles were similar to the magnesian basalts of Soufriere and that the development of a more strongly calc-alkaline character in evolved rocks was a result of AFC processes in the crust resulting from higher magma supply rates and more mature magmatic plumbing systems.
Appendix A: Sample Descriptions
Abbreviations used: BA, basaltic andesite; PSL Pre-Somma Lava; YTF, Yellow Tuff Formation; CL, Crater Lava; PF, Pyroclastic Formation.
STV 301: 13°15.65'N, 61°07.05,'W. Basaltic PSL from Black Point.
STV 303: 13°15'N, 61°07'W. Basaltic pumice from YTF fall deposit, near Mangrove.
STV 306: 13°18.01'N, 61°07.24'W. BA scoria from PF, lower Rabacca.
STV 307: 13°18.07'N, 61°07.54'W. BA pumice from lapilli fall layer in YTF, lower Rabacca.
STV 309: 13°18.3'N, 61°08.5'W. PSL block from flow breccia overlain by YTF, Rabacca River, below Chainspout.
STV 310: 13°18.2'N, 61°08.3'W. Basaltic PSL from Rabacca River, below Chainspout.
STV 312: 13°18.45'N, 61°07.14'W. BA scoria in PF pyroclastic flow, Waterloo sea cliff, Oven Gully.
STV 313: 13°20.33'N, 61°07.40'W. BA pumice from grey fall unit close to base of roadside YTF succession, Robin Rock.
STV 314: 13°20.2'N, 61°07.4'W. BA PSL from flow underlying YTF, Robin Rock Point.
STV 315: 13°20.59'N, 61°07.66'W. Basaltic PSL, London Point.
STV 316: 13°21.21'N, 61°07.87'W. BA PSL, Chibarabu Point.
STV 317: 13°21.38'N, 61°07.97'W. Basaltic scoria from YTF, Sion Hill.
STV 318: 13°22.52'N, 61°08.45'W. BA PSL from Cow-and-calves flow, Owia salt pond.
STV 319: 13°22.59'N, 61°09.24'W. Basaltic PSL from Commantawana Bay.
STV 320: 13°22.6'N, 61°09.3'W. Basaltic scoria from >1 m thick YTF fall, Jumby Point.
STV 323: 13°21.54'N, 61°07.89'W. BA PSL, south Sandy Bay.
STV 324: 13°19.66'N, 61°10.71'W. BA Crater Lava, Soufriere Table flow.
STV 325: 13°19.82'N 61°10.56'W. BA pumice from 1979 fall deposit, Windward Trail.
STV 326: 13°20.8'N, 61°10.75'W. BA PSL, basal flow from Soufriere Mountains.
STV 330: 13°18.52'N, 61°07.87'W. BA pumice from lowest fall layer, Lot 14.
STV 331: 13°18.52'N, 61°07.87'W. BA pumice from (1718?) fall layer, Lot 14.
STV 332: 13°18.52'N, 61°07.87'W. BA pumice from (1812?) eruption, Lot 14.
STV 333: 13°18.997'N, 61°09.169'W. Basaltic crystal-rich fall unit, pre-1718, Bamboo.
STV 334: 13°18.64'N, 61°08.81'W. Basaltic scoria clasts from 1902 surge deposit? Lot 14.
STV 335: 13°18.64'N, 61°08.81'W. BA tephra clasts from fall deposit (1812?), Lot 14.
STV 345: 13°17.44'N, 61°08.27'W. BA PSL from quarry, Indian Estate.
STV 349: 13°18.68'N, 61°07.99'W. BA scoria from PF pyroclastic flow, Waribishy River.
STV 351: 13°18.25'N, 61°08.43'W. BA PSL, lowermost lava near road, Waribishy.
STV 354: 13°18.721'N, 61°09.795'W. BA scoria from PF, upper Rabacca River.
STV 356: 13°18.8'N, 61°09.9'W. Basaltic PSL, lower flow at waterfall cliff, top of valley, upper Rabacca River.
STV 357: 13°18.8'N, 61°10.0'W. BA PSL, upper cliff-forming flow, upper Rabacca River.
STV 358: 13°18.25'N, 61°08.43'W. Basaltic PSL, lava bed at falls, Rabacca River.
STV 359: 13°17.969'N 61°08.794'W. BA scoria, 1902 pyroclastic flow, Dry Rabacca.
STV 362: 13°19.05'N 61°13.57'W. BA scoria from 1902? pyroclastic flow, south bank of Dry Wallibou.
STV 363: 13°19'N, 61°13'W. BA scoria from PF pyroclastic flow, Dry Wallibou.
STV 365: 13°19'N, 61°13'W. BA banded scoria from PF pyroclastic flow, Wallibou–Ronde.
STV 369: 13°22.2'N, 61°11.7'W. BA PSL from upper flow near concrete bridge, Baleine Falls.
STV 371: 13°20.3'N, 61°13.0'W. BA scoria from PF pyroclastic flow, Larikai beach.
STV 372: 13°20.2'N, 61°12.7'W. BA PSL, platy lava beneath YTF, Larikai River.
STV 373: 13°17.3'N, 61°14.4'W. BA pumice from fall unit in YTF, south Chateaubelair.
STV 374: 13°17.2'N, 61°15.1'W. Basaltic lapilli from fall layer in YTF, Dark View.
STV 376: 13°17.1'N, 61°15.2'W. Pumice clasts from mixed YTF fall unit, Rose Bank. Separated into dark basaltic (D) and light andesitic (L) components.
STV 377: 13°17.1'N, 61°15.4'W. Banded pumice clasts from mixed YTF fall unit, Rose Bank. Separated into dark BA (D), intermediate BA (I) and light andesitic (L) components.
STV 79-70: 13°20'N, 61°11'W. BA lava from 1979 crater dome.
SVE 113: 13°20'N, 61°11'W. BA CL from thickest flow in NW (Larikai) wall of crater.
SVE 114: 13°20'N, 61°11'W. Andesitic CL from Nose Dike, cutting SE interior of crater wall.
SVE 115: 13°20'N, 61°11'W. BA CL, 15 m thick lava, SW wall of crater.
SVE 116: 13°20'N 61°11'W. BA CL prominent in W S and SW of crater wall.
|
Appendix B: Analytical Methods
Major and trace element compositions were measured at Lancaster University using a Phillips 1400 XRF spectrometer, and are quoted in wt % and ppm, respectively. Standards were run at frequent intervals and calibrations were regularly drift corrected using a monitor buffer. Relative 1
precision is <0.5% for most major elements (2% for Na2O, MnO and P2O5) and <3% for most minor elements (4% for Ba and Sc, and 5.5% for Cr and Nb). Concentrations of the REE plus Ta, Hf and Cs were determined by INAA, and are quoted in ppm. Samples were irradiated with neutron flux monitors and standards at the Imperial College Reactor Centre, and counted twice (to ensure good precision for both short-lived and longer-lived isotopes) by gamma-ray spectrometry at the Open University. Relative 1 precision is <3.5% for all elements except Cs (8%) and Ta (6%). Compositions of phenocrysts and silicate melt inclusions were measured by electron microprobe at the US Geo- logical Survey in Virginia. Sodium was analysed first to minimize its loss. Major and minor elements were counted for 20 and 40 or 60 s, respectively. Relative 1 precision is estimated to be 1–2% for major elements and 5–10% for minor elements. Sr, Nd and Pb isotopic ratios and Th and U concentrations (in ppm) were determined by solid-source thermal ionization mass spectrometry (TIMS) using Finnegan MAT 261 and 262 instruments at the Open University. Chemical separation procedures for Sr/Nd, Pb and Th/U were undertaken in separate clean chemistry laboratories, and total procedure blanks indicated negligible contamination. Machine standards and rock standards were also analysed to ensure the high quality and reproducibility of the data. 87Sr/86Sr ratios are relative to an average value of 0.710279 ± 19 for the NBS 987 standard, and replicate analyses of the J&M Nd standard yielded 143Nd/144Nd = 0.511794 ± 9. Pb data were corrected for mass fractionation relative to the NBS 981 standard. Average 1 uncertainties on Th and U concentrations are ±0.005 and ±0.001 ppm, respectively.
|
Acknowledgements |
|---|
We thank Richard Robertson of the Seismic Research Unit, University of the West Indies, for invaluable help in the field. Isotopic analyses were carried out by E.H. at the Open University; the untiring assistance of Mabs Johnston, Peter van Calsteren and Simon Turner is gratefully acknowledged. Thanks are also due to Nick Rogers (Open University) for INAA data, and Vicky Burnett, Paul Wheeler and Stuart Black (Lancaster University) for help with XRF analyses. We thank the Natural Environment Research Council for support, through a research studentship to E.H. R.M. and H.S. thank NATO for support for field-work on St Vincent. Finally, we thank Drs S. Eggins and I. Parkinson for very helpful journal reviews.
* Corresponding author. Telephone: 01524 593934. Fax: 01524 593985.
|
References |
|---|
|
TOPABSTRACTIntroductionGeological HistoryPetrography, Mineralogy and...GeochemistryMelt InclusionsPre-Eruptive Water Contents of...DiscussionReferences |
|---|
Allen J. C., Boettcher A. L. Amphiboles in andesite and basalt. II. Stability as a function of P–T–fH2O–fO2. American Mineralogist (1983) 63:307–314.
Anderson A. T. Water in some hypersthenic magmas. Journal of Geology (1979) 87:509–531.[Web of Science]
Arculus R. J. Aspects of magma genesis in arcs. Lithos (1994) 33:189–208.[Web of Science]
Arculus R. J., Wills K. J. A. The petrology of plutonic blocks and inclusions from the Lesser Antilles island arc. Journal of Petrology (1980) 21:743–799.
Baker P. E. Petrology of Mt. Misery volcano, St. Kitts, West Indies. Lithos (1968) 1:124–150.
Baker P. E. The Soufriere volcano, St. Vincent and its 1971–72 eruption. Journal of Earth Sciences, Leeds (1972) 8:205–217.
Baker D. R., Eggler D. H. Fractionation paths of Atka (Aleutians) high-alumina basalts: constraints from phase relations. Journal of Volcanology and Geothermal Research (1983) 18:387–404.[Web of Science]
Baker D. R., Eggler D. H. Compositions of anhydrous and hydrous melts coexisting with plagioclase, augite, and olivine or low-Ca pyroxene from 1 atm to 8 kbar-application to the Aleutian volcanic center of Atka. American Mineralogist (1987) 72:12–28.[Abstract]
Ballhaus C., Berry R. F., Green D. H. High pressure experimental calibration of the olivine–orthopyroxene–spinel oxygen geobarometer: implications for the oxidation state of the mantle. Contributions to Mineralogy and Petrology (1991) 107:27–40. [Correction (1994) Contributions to Mineralogy and Petrology 118, 109.].[Web of Science]
Bardintzeff J.-M. Les pyroxènes et leurs inclusions, marqueurs privilégiés des nuées ardentes (Saint-Vincent, Antilles, 1979). Bulletin Minéralogique (1984) 107:41–54.
Bardintzeff J.-M. Magma mixing processes in volcanic contexts, a thermodynamic approach with the examples of St. Vincent Soufriere volcano, West Indies and Cerro Chiquito, Guatemala. Terra Nova (1992) 4:553–566.[Web of Science]
Barsdell M., Berry R. F. Origin and evolution of primitive island arc ankaramites from western Epi, Vanuatu. Journal of Petrology (1990) 31:747–777.
Boynton C. H., Westbrook G. K., Bott M. H. P., Long R. E. A seismic refraction investigation of crustal structure beneath the Lesser Antilles island arc. Geophysical Journal of the Royal Astronomical Society (1979) 58:371–393.[Web of Science]
Briden J. C., Rex D. C., Faller A. M., Tomblin J. F. K-Ar geochronology and palaeomagnetism of volcanic rocks in the Lesser Antilles island arc. Philosophical Transactions of the Royal Society of London, Series A (1979) 291:485–528.
Brown G. M., Holland J. G., Sigurdsson H., Tomblin J. F., Arculus R. J. Geochemistry of the Lesser Antilles volcanic island arc. Geochimica et Cosmochimica Acta (1977) 41:785–801.[Web of Science]
Bryan W. B., Finger L. W., Chayes F. Estimating proportions in petrographic mixing equations by least-squares approximation. Science (1969) 163:926–927.
Carey S. N., Sigurdsson H. Deep-sea evidence for distribution of tephra from the mixed magma eruption of the Soufriere on St Vincent, 1902: ash turbidites and air fall. Geology (1978) 6:271–274.
Carron J.-P., Le Guen de Kerneizon M. Carbon dioxide release through metamorphism of tuffitic limestones in a magma chamber: the case of Soufriere volcano, St. Vincent, Lesser Antilles. Bulletin de la Societé Géologique de France (1991) 162:1029–1035.[Abstract]
Cawthorn R. G. Melting relations in part of the system CaO–MgO–Al2O3–SiO2–Na2O–H2O under 5 kb pressure. Journal of Petrology (1976) 17:44–72.
Cawthorn R. G., O'Hara M. J. Amphibole fractionation in calc-alkaline magma series. American Journal of Science (1976) 276:309–329.
Crawford A. J., Falloon T. J., Eggins S. The origin of island arc high-alumina basalts. Contributions to Mineralogy and Petrology (1987) 97:417–430.[Web of Science]
Davidson J. P. Petrogenesis of Lesser Antilles island arc magmas: isotopic and geochemical constraints. (1984) Leeds University. Ph.D. Thesis.
Davidson J. P. Mechanisms of contamination in Lesser Antilles island arc magmas from radiogenic and oxygen isotopic relationships. Earth and Planetary Science Letters (1985) 72:163–174.[Web of Science]
Davidson J. P. Isotopic and trace element constraints on the petrogenesis of subduction-related lavas from Martinique, Lesser Antilles. Journal of Geophysical Research (1986) 91:5943–5962.
Davidson J. P. Crustal contamination versus subduction zone enrichment: examples from the Lesser Antilles and implications for mantle source compositions of island arc volcanic rocks. Geochimica et Cosmochimica Acta (1987) 51:2185–2198.[Web of Science]
Davidson J. P., Harmon R. S. Oxygen isotopic constraints on the petrogenesis of volcanic arc magmas from Martinique, Lesser Antilles. Earth and Planetary Science Letters (1989) 95:255–270.[Web of Science]
Davidson J. P., Boghossian N. D., Wilson M. The geochemistry of the igneous rock suite of St. Martin, Northern Lesser Antilles. Journal of Petrology (1993) 34:839–866.
Devine J. D. Petrogenesis of the basalt–andesite–dacite association of Grenada, Lesser Antilles island arc, revisited. Journal of Volcanology and Geothermal Research (1995) 69:1–33.[Web of Science]
Devine J. D., Sigurdsson H. Garnet–fassaite calc-silicate nodule from La Soufriere, St. Vincent. American Mineralogist (1980) 65:302–305.[Abstract]
Devine J. D., Sigurdsson H. The liquid composition and crystallization history of the 1979 Soufriere magma, St. Vincent, W.I. Journal of Volcanology and Geothermal Research (1983) 16:1–31.[Web of Science]
Dostal J., Dupuy C., Carron J. P., Le Guen de Kerneizon M., Maury R. C. Partition coefficients of trace elements: application to volcanic rocks of St. Vincent, West Indies. Geochimica et Cosmochimica Acta (1983) 47:525–533.[Web of Science]
Eggins S. M. Origin and differentiation of picritic arc magmas, Ambae (Aoba), Vanuatu. Contributions to Mineralogy and Petrology (1993) 114:79–100.[Web of Science]
Falloon T. J., Green D. H., Hatton C. J., Harris K. L. Anhydrous partial melting of a fertile and depleted peridotite from 2 to 30 kbars and application to basalt petrogenesis. Journal of Petrology (1988) 29:1257–1282.
Feeley T. C., Davidson J. P. Petrology of calc-alkaline lavas at Volcán Ollagüe and the origin of compositional diversity at central Andean stratovolcanoes. Journal of Petrology (1994) 35:1295–1340.
Fiske R. S., Shepherd J. B. Deformation studies on Soufriere, St. Vincent, between 1977 and 1981. Science (1982) 216:1125–1126.
Fiske R. S., Shepherd J. B. Twelve years of ground-tilt measurements on the Soufriere of St. Vincent, 1977–1989. Bulletin Volcanologique (1990) 52:227–241.[Web of Science]
Foden J. D., Green D. H. Possible role for amphibole in the origin of andesite: some experimental and natural evidence. Contributions to Mineralogy and Petrology (1992) 109:479–493.[Web of Science]
Gaetani G. A., Grove T. L., Bryan W. B. The influence of water on the subduction-related igneous rocks. Nature (1993) 365:332–334.
Gill J. Orogenic Andesites and Plate Tectonics (1981) Berlin: Springer-Verlag. 390.
Graham A. M. Genesis of the igneous rock suites of Grenada, Lesser Antilles. (1980) Edinburgh University. Ph.D. Thesis.
Graham A. M., Thirlwall M. F. Petrology of the 1979 eruption of Soufriere volcano, St. Vincent, Lesser Antilles. Contributions to Mineralogy and Petrology (1981) 76:336–342.[Web of Science]
Grove T. L. Origin of calc-alkaline series lavas at Medicine Lake volcano by fractionation, assimilation and mixing-corrections and clarifications. Contributions to Mineralogy and Petrology (1993) 114:422–424.
Grove T. L., Baker M. B. Phase equilibrium controls on the tholeiitic versus calc-alkaline differentiation trends. Journal of Geophysical Research (1984) 89:3253–3274.
Grove T. L., Bryan W. B. Fractionation of pyroxene-phyric MORB at low pressure: an experimental study. Contributions to Mineralogy and Petrology (1983) 84:293–309.[Web of Science]
Hawkesworth C. J., O'Nions R. K., Arculus R. J. Nd- and Sr- isotope geochemistry of island arc volcanics, Grenada, Lesser Antilles. Earth and Planetary Science Letters (1979) 45:237–248.[Web of Science]
Hawkesworth C. J., Turner S. P., Peate D., McDermott F., van Calsteren P. Elemental U and Th variations in island arc rocks: implications for U-series isotopes. Chemical Geology (1997) 139:207–221.[Web of Science]
Helz R. T. Phase relations and compositions of amphiboles produced in studies of the melting behaviour of rocks. In: Amphiboles: Petrology and Experimental Phase Relations. Mineralogical Society of America, Reviews in Mineralogy—Veblen D. R., Ribbe P. H., eds. (1982) 9B:279–353.
Hunter A. G., Blake S. Petrogenetic evolution of a transitional tholeiitic-calc-alkaline series: Towada volcano, Japan. Journal of Petrology (1995) 36:1579–1605.
Irvine T. N., Baragar W. R. A. A guide to the chemical classification of the common volcanic rocks. Canadian Journal of Earth Sciences (1971) 8:523–548.
Johnston A. D., Draper D. S. Near-liquidus phase relations of an anhydrous high-magnesia basalt from the Aleutian Islands: implications for arc magma genesis and ascent. Journal of Volcanology and Geothermal Research (1992) 52:27–41.[Web of Science]
Kennedy A. K., Grove T. L., Johnson R. W. Experimental and major element constraints on the evolution of lavas from Lihir Island, Papua New Guinea. Contributions to Mineralogy and Petrology (1990) 104:722–734.[Web of Science]
Kersting A. B., Arculus R. J. Klyuchevskoy volcano, Kamchatka, Russia: the role of high-flux recharged, tapped, and fractionated magma chamber(s) in the genesis of high-Al2O3 from high-MgO basalt. Journal of Petrology (1994) 35:1–41.
Kilinc A., Carmichael I. S. E., Rivers M. L., Sack R. O. The ferric–ferrous ratio of natural silicate liquids equilibrated in air. Contributions to Mineralogy and Petrology (1983) 83:136–140.[Web of Science]
Kudo A. M., Weill D. F. An igneous plagioclase thermometer. Contributions to Mineralogy and Petrology (1970) 25:52–65.[Web of Science]
Kushiro I. The system forsterite–diopside–silica with and without water at high pressures. American Journal of Science (1969) 267A:269–294.[Web of Science]
LeBas M. J., LeMaitre R. W., Streckeisen A., Zanettin B. A chemical classification of volcanic rocks based on the total alkali–silica diagram. Journal of Petrology (1986) 27:745–750.
Lewis J. F. Petrology of the ejected plutonic blocks of the Soufriere volcano, St. Vincent, West Indies. Journal of Petrology (1973a) 14:81–112.
Lewis J. F. Mineralogy of the ejected blocks of the Soufriere volcano, St. Vincent: olivine, pyroxene, amphibole and magnetite paragenesis. Contributions to Mineralogy and Petrology (1973b) 38:197–220.[Web of Science]
Lindsley D. H. Phase equilibria of pyroxenes at pressures >1 atmosphere. In: Pyroxenes. Mineralogical Society of America, Reviews in Mineralogy—Prewitt C. T., ed. (1980) 7:289–306.
Lindsley D. H. Pyroxene thermometry. American Mineralogist (1983) 68:477–493.[Abstract]
Martin-Lauzer F. R., Ingrin J., Porier J. P. Transmission electron microscopic study of the immiscibility in natural and synthetic rhyolitic glasses. Earth and Planetary Science Letters (1986) 79:168–178.[Web of Science]
McDonough W. F., Sun S. S., Ringwood A. E., Jagoutz E., Hofmann A. W. K, Rb and Cs in the Earth and Moon and the evolution of the mantle of the Earth. Geochimica et Cosmochimica Acta (1992) 56:1001–1012.[Web of Science]
Middlemost E. A. K. Iron oxidation ratios, norms and the classification of volcanic rocks. Chemical Geology (1989) 77:19–26.[Web of Science]
Miller D. M., Langmuir C. H., Goldstein S. L., Franks A. L. The importance of parental magma composition to calc-alkaline and tholeiitic evolution-evidence from Umnak island in the Aleutians. Journal of Geophysical Research-Solid Earth (1992) 97:321–343.
Miyashiro A. Volcanic rock series in island arcs and active continental margins. American Journal of Science (1974) 274:321–355.[Abstract]
Nakamura N. Determination of REE, Ba, Fe, Mg, Na and K in carbonaceous and ordinary chondrites. Geochimica et Cosmochimica Acta (1974) 38:757–775.[Web of Science]
Pearce J. A. Role of the sub-continental lithosphere in magma genesis at active continental margins. In: Continental Basalts and Mantle Xenoliths—Hawkesworth C. J., Norry M. J., eds. (1983) Nantwich, UK: Shiva. 230–249.
Pearce J. A., Parkinson I. J. Trace element models for mantle melting: applications to volcanic arc petrogenesis. In: Magmatic Processes and Plate Tectonics. Geological Society, London, Special Publication—Prichard H. M., Alabaster T., Harris N. B. W., Neary C. R., eds. (1993) 76:373–403.
Perfit M. R., Gust D. A., Bence A. E., Arculus R. J., Taylor S. R. Chemical characteristics of island arc basalts: implications for mantle sources. Chemical Geology (1980) 30:227–256.[Web of Science]
Pindell J. L., Cande S. C., Pitman W. C. III, Rowley D. B., Dewey J. F., LaBrecque J., Haxby W. A plate-kinematic framework for models of Caribbean evolution. Tectonophysics (1988) 155:121–138.[Web of Science]
Pushkar P., Steuber A. M., Tomblin J. F., Julian G. M. Strontium isotopic ratios in volcanic rocks from St. Vincent and St. Lucia, Lesser Antilles. Journal of Geophysical Research (1973) 78:1279–1287.
Rea W. J. The volcanic geology and petrology of Montserrat, West Indies. Journal of the Geological Society, London (1974) 130:341–366.
Rickwood P. C. Boundary lines within petrologic diagrams which use oxides of major and minor elements. Lithos (1989) 22:247–263.[Web of Science]
Romnick J. D., Mahlburg Kay S., Kay R. W. The influence of amphibole fractionation on the evolution of calc-alkaline andesite and dacite tephra from the central Aleutians, Alaska. Contributions to Mineralogy and Petrology (1992) 112:101–118.[Web of Science]
Roobol M. J., Smith A. L. A comparison of the recent eruptions of Mt. Pelée, Martinique and Soufriere, St. Vincent. Bulletin Volcanologique (1975) 39:214–240.
Rowley K. Late pyroclastic deposits of Soufriere volcano, West Indies. Geological Society of America Bulletin (1978) 89:825–835.
Saunders A. D., Tarney J. Geochemical characteristics of basaltic volcanism within back-arc basins. In: Marginal Basin Geology. Geological Society, London, Special Publication—Kokelaar B. P., Howells M. F., eds. (1984) 16:59–76.
Saunders A. D., Norry M. J., Tarney J. Origin of MORB and chemically depleted mantle reservoirs: trace element constraints. In: Journal of Petrology (1988) Special Lithosphere Issue. 415–445.
Shepherd J. B., Aspinall W. P., Rowley K. C., Pereira J., Sigurdsson H., Fiske R. S., Tomblin J. F. The eruption of Soufriere volcano, St. Vincent, April–June 1979. Nature (1979) 282:24–28.[Web of Science]
Sigurdsson H., Carey S. Caribbean volcanoes: a field guide. In: Geological Association of Canada, Mineralogical Association of Canada, Society of Economic Geologists (1991) Joint Annual Meeting, Toronto 1991, Field Trip B1:: Guidebook. 101.
Sigurdsson H., Macdonald R., Harkness D. D., Pringle M. S., Robertson R. A. E., Heath E. Volcanic history of the Soufriere of St Vincent. In: Bulletin of Volcanology (1998) submitted.
Sisson T. W. Hornblende–melt trace element partitioning measured by ion microprobe. Chemical Geology (1994) 117:331–344.[Web of Science]
Sisson T. W., Grove T. L. Experimental investigations of the role of H2O in calc-alkaline differentiation and subduction zone magmatism. Contributions to Mineralogy and Petrology (1993a) 113:143–166.[Web of Science]
Sisson T. W., Grove T. L. Temperatures and H2O contents of low-MgO high-alumina basalts. Contributions to Mineralogy and Petrology (1993b) 113:167–184.[Web of Science]
Sisson T. W., Layne G. D. H2O in basalt and basaltic andesite glass inclusions from four subduction-related volcanoes. Earth and Planetary Science Letters (1993) 17:619–635.
Smith A. L., Roobol M. J. Mt. Pelée, Martinique; a study of an active island-arc volcano. Geological Society of America, Memoir (1990) 175:105.
Smith T. E., Thirlwall M. F., MacPherson C. Trace element and isotope geochemistry of the volcanic rocks of Bequia, Grenadine Islands, Lesser Antilles arc: a study of subduction enrichment and intra-crustal contamination. Journal of Petrology (1996) 37:117–143.
Sobolev A. V., Danyushevsky L. V. Petrology and geochemistry of boninites from the north termination of the Tonga trench: constraints on the generation conditions of primary high-Ca boninite magmas. Journal of Petrology (1994) 35:1183–1211.
Spulber S. D., Rutherford M. J. The origin of rhyolite and plagiogranite in oceanic crust: an experimental study. Journal of Petrology (1983) 24:1–25.
Stolper E., Newman S. The role of water in the petrogenesis of Mariana trough magmas. Earth and Planetary Science Letters (1994) 121:293–325.[Web of Science]
Sun S. S. Lead isotopic study of young volcanic rocks from mid-ocean ridges, ocean islands and island arcs. Philosophical Transactions of the Royal Society of London, Series A (1980) 297:409–445.
Tatsumi Y., Eggins S. Subduction Zone Magmatism (1995) Cambridge, MA: Blackwell Science, Inc. 211.
Thirlwall M. F., Graham A. M. Evolution of high-Ca, high-Sr C-series basalts from Grenada, Lesser Antilles: contamination in the arc crust. Journal of the Geological Society, London (1984) 141:427–445.
Thirlwall M. F., Smith T. E., Graham A. M., Theodorou N., Hollings P., Davidson J. P., Arculus R. J. High field strength element anomalies in arc lavas: source or process? Journal of Petrology (1994) 35:819–838.
Thirlwall M. F., Graham A. M., Arculus R. J., Harmon R. S., MacPherson C. G. Resolution of the effects of crustal assimilation, 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.[Web of Science]
Tomblin J. F. Chemical analyses of volcanic rocks from the Lesser Antilles. Special Publication of the University of the West Indies Seismic Research Unit (1968) 15:23.
Tormey D. R., Grove T. L., Bryan W. B. Experimental petrology of normal MORB near the Kane Fracture Zone: 22°-25°N, Mid-Atlantic Ridge. Contributions to Mineralogy and Petrology (1987) 96:121–139.[Web of Science]
Turner S., Hawkesworth C. J., 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.[Web of Science]
Ulmer P. The dependence of the Fe2+–Mg cation-partitioning between olivine and basaltic liquid on pressure, temperature and composition: an experimental study to 30 kbars. Contributions to Mineralogy and Petrology (1989) 101:261–273.[Web of Science]
Wadge G. The dykes and structural setting of the volcanic front in the Lesser Antilles island arc. Bulletin Volcanologique (1986) 48:349–372.
Wagner T. P., Donnelly-Nolan J. M., Grove T. L. Evidence of hydrous differentiation and crystal accumulation in the low-MgO, high-Al2O3 Lake Basalt from Medicine Lake volcano, California. Contributions to Mineralogy and Petrology (1995) 121:210–216.
Wakita H., Rey P., Schmitt R. A. Abundances of the 14 rare-earth elements and twelve other trace elements in Apollo 12 samples: five igneous and one breccia rocks and four soils. In: Proceedings of the 2nd Lunar Science Conference (1971) Oxford: Pergamon Press. 1319–1329.
Walker D., Shibara T., DeLong S. E. Abyssal tholeiites from the Oceanographer fracture zone. II. Phase equilibria and mixing. Contributions to Mineralogy and Petrology (1979) 70:111–125.[Web of Science]
Westercamp D., Mervoyer B. Les séries volcaniques de la Martinique et de la Guadeloupe (F.W.I.). Bulletin du Bureau de Recherches Géologiques et Minières, Section IV (1976) 4:229–242.
White W. M., Dupré B. Sediment subduction and magma genesis in the Lesser Antilles: isotopic and trace element constraints. Journal of Geophysical Research (1986) 91:5927–5941.
White W. M., Patchett J. 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.[Web of Science]
Wills K. J. A. Geological history of southern Dominica. (1974) Durham University. Ph.D. Thesis.
Wood B. J. Oxygen barometry of spinel peridotites. In: Oxide Minerals: Petrologic and Magmatic Significance. Mineralogical Society of America, Reviews in Mineralogy—Lindsley D. H., ed. (1991) 25:417–443.
Wood B. J., Banno S. Garnet–orthopyroxene and orthopyroxene–clinopyroxene relationships in simple and complex systems. Contributions to Mineralogy and Petrology (1973) 42:109–124.[Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||