Journal of Petrology Advance Access published online on December 24, 2008
Journal of Petrology, doi:10.1093/petrology/egn066
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Assimilation of Plutonic Roots, Formation of High-K ‘Exotic’ Melt Inclusions and Genesis of Andesitic Magmas at Volcán De Colima, Mexico
Department Of Earth Sciences, University Of Bristol, Wills Memorial Building, Bristol Bs8 1Rj, Uk
Received June 17, 2008; Revised typescript accepted November 14, 2008
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONPhenocryst-hosted melt inclusions from the 1998–2005 andesite eruptions of Volcán de Colima (Mexico) show broad ranges of major and trace element contents that do not overlap with the bulk-rock compositions and indicate that melt inclusions can be formed by and record a range of processes involved in the genesis of andesites. The melt inclusions that demonstrably record the evolution of the melt feeding the eruption indicate low-pressure (130–10 MPa) crystallization of a dacitic melt despite the monotonous bulk andesitic composition of historical magmas at Volcán de Colima. Mingling of dacite melt with gabbroic fragments in the shallow sub-volcanic system is the process responsible for generating the bulk andesitic composition of the magmas. A significant proportion of the melt inclusions have distinctive high large ion lithophile element (LILE) signatures. These ‘exotic’ high-K melt inclusions in pyroxenes are thought to result from incongruent melting of interstitial biotite during assimilation of gabbroic fragments in the dacitic melt. A second group of exotic high-K melt inclusions found in plagioclase are likely to result from dissolution of higher-pressure (>200 MPa) amphibole, plagioclase, magnetite and biotite cumulates during assimilation in the ascending dacitic melt. Although they are not volumetrically abundant, high-K melts formed during assimilation of plutonic fragments and crystal cumulates made a significant contribution to the LILE contents of the magmas and represent a potential source for this group of elements. The range of melt inclusion compositions in Volcán de Colima magmas emphasizes the importance of mixing between ascending evolved melts and crystal populations formed during previous episodes of magmatism over a range of pressures, temperatures and volatile contents. Cannibalization of plutonic roots appears to be a fundamental process in the genesis of andesite magmas and melt inclusions at continental arc volcanoes.
KEY WORDS: andesite; arc; magma mingling; melt inclusion; Mexican volcanic belt
|
INTRODUCTION |
|---|
Andesites typically show complicated petrography with multiple crystal populations and complex mineral zoning, indicating that open-system magmatic processes (magma mixing and mingling, assimilation of crystal cumulates, crustal contamination) are predominant at arc volcanoes (e.g. Anderson, 1976
; Eichelberger, 1978; Gamble et al., 1999; Dungan et al., 2001; Reubi et al., 2003; Humphreys et al., 2006a). Consequently, bulk-rock compositions, although useful to constrain the overall evolution of volcanic systems, provide limited information on the composition of the melts in the magmatic system and the processes that control their evolution. Knowledge of these parameters is fundamental both to evaluating the hazard at active volcanoes, as they ultimately control the eruption style and intensity, and to understanding the formation and evolution of continental crust.
Melt inclusions (MIs) trapped in phenocrysts may be used to document the composition and evolution of the crystallizing melt (e.g. Roedder, 1984; Luhr, 2001; Lowenstern, 2003; Blundy & Cashman, 2005; Wallace, 2005; Witter et al., 2005; Portnyagin et al., 2007). However, interpretations from MIs are valid only if they accurately record and preserve the original melt composition. Post-entrapment crystallization of the host phase and/or daughter crystals and diffusive re-equilibration with the host and/or external magma are composition-modifying processes known to affect MIs (e.g. Roedder, 1984; Danyushevsky et al., 2000; Cottrell et al., 2002; Spandler et al., 2007). In addition, ‘exotic’ MIs with unusual major and trace element compositions have been reported in phenocrysts from primitive mantle-derived magmas, which suggest that some MIs may reflect grain-scale processes and do not represent geologically significant melts (e.g. Danyushevsky et al., 2004). Although MIs are powerful tools to document the evolution of melts, careful evaluation of their origin, significance and representativeness is required before they can be used to interpret the evolution of subvolcanic magmatic systems.
Here we present a detailed study of melt inclusions in andesitic magmas erupted at Volcán de Colima (VDC), Mexico, between 1998 and 2005. The aim is to combine the major and trace element chemistry of MIs with detailed petrographic observations to document the processes that controlled the evolution of the melts, and to establish a petrogenetic model for the andesites. We show that a significant proportion of the MIs formed during assimilation of plutonic fragments and crystal cumulates and trap exotic melts. In addition to their importance as petrological recorders, these exotic melts are potential sources of large ion lithophile elements (LILE) in the magmas. We also show that the MIs that demonstrably record the evolution of the melt crystallizing in the magmatic system are dacitic in composition, in contrast to the andesitic compositions of their host rocks. We suggest that this discrepancy is not a bias of the MI record but reflects a compositional gap likely to be a fundamental aspect of arc magmatism (Reubi & Blundy, in preparation).
|
ANALYTICAL METHODS |
|---|
Whole-rocks were analysed for major elements (Table 1) by X-ray fluorescence (XRF) at the University of Auckland, using a Siemens SRS 3000 instrument. Rock chips and single phenocrysts separated from crushed samples were mounted in epoxy resin and polished to expose the melt inclusions. Each mount was examined using back-scattered electron (BSE) images to identify suitable melt inclusions and to establish the absence of daughter crystals. BSE images were also used to characterize the compositional zoning of phenocrysts and to perform point counts taking into account the composition and textures of the phenocrysts.
|
A total of 162 inclusions from eight samples was first analysed for major and minor elements by electron microprobe analysis (EMPA). A subset of 77 inclusions was then analysed for H2O and trace elements by secondary ion mass spectrometry (SIMS). Humphreys et al. (2006b
) have shown that electron-beam irradiation during EMPA causes permanent damage to the top 1 µm of hydrous, silica-rich glasses even at the analytical conditions described below. To avoid this problem, the mounts were lightly polished after EMP analysis to remove a few micrometres of the surface layer of the inclusions prior to SIMS analysis.
Minerals and glasses were analysed at the University of Bristol using a CAMECA SX-100 five-spectrometer wavelength-dispersive spectrometry (WDS) instrument. Plagioclases and pyroxenes were analysed using a 20 kV accelerating voltage, a 1 µm beam diameter, and 10 nA and 20 nA beam current, respectively. Melt inclusions and groundmass glasses were analysed using a 20 kV accelerating voltage, 2 nA beam current and a 12–15 µm beam diameter, with Na and Si analysed first to reduce the effects of alkali migration (Humphreys et al., 2006b). Data reduction was performed online using a stoichiometric PAP correction model (Pouchou & Pichoir, 1984).
Ion-microprobe (SIMS) analyses were carried out using a Cameca IMS-4f instrument at the University of Edinburgh with 10 kV (nominal) O– primary beam and 4·5 kV positive secondary ions following the procedures described by Blundy et al. (2008). Kinetic filtering with a 75 ± 20 V offset was used to minimize transmission of molecular ions. Conventional peak-stripping was used to remove persistent molecular ion interferences (e.g. 29Si16O on 45Sc). Light element isotopes (1H, 7Li, 9Be, 11B, 26Mg, 30Si, 45Sc and 47Ti) in melt inclusions were analysed first using 6 nA current at the sample surface, corresponding to an 
15 µm sputter area. A subset of inclusions were analysed for heavier element isotopes (30Si, 47Ti, 85Rb, 88Sr, 89Y, 90Zr, 93Nb, 133Cs, 138Ba, 139La, 140Ce, 141Pr, 143Nd, 149Sm, 157Gd, 159Tb, 161Dy, 165Ho, 171Yb, 178Hf, 181Ta, 232Th and 238U) using a 3 nA current, corresponding to a 10 µm sputter area. Because of the small size of the inclusions, the light and heavy trace element analyses were conducted on the same spot; repeat analysis of 47Ti was used to check for internal consistency. For all analyses, 30Si was used as an internal standard and values were corrected for their SiO2 contents as measured by EMPA. Trace element calibration was carried out using NIST SRM610 multi-element glass. For plagioclase analysis, the calibration was corrected for the ion yield of NIST SRM610 relative to Lake County plagioclase using ratios determined by R. Hinton (personal communication). H2O analysis was calibrated against a set of glass standards of known H2O content, following the method of Blundy & Cashman (2005). Typical uncertainties, based on counting statistics, are 4% relative for H2O and 15% relative for trace elements.
The complete dataset for plagioclase, pyroxenes, amphibole, melt inclusion major elements and melt inclusion trace elements are presented in Electronic Appendices A1, A2, A3, A4 and A5, respectively, which are available for downloading from the Journal of Petrology website at http://petrology.oxfordjournals.org.
|
VOLCAN DE COLIMA |
|---|
Volcán de Colima (VDC), Mexico is one of the most historically active volcanoes in North America (Luhr & Carmichael, 1980
). It is located in the western part of the Trans-Mexican Volcanic Belt within the north–south-trending Colima graben. VDC is the youngest andesitic cone of a larger volcanic complex, which also includes Nevado de Colima and Volcán Cantaro (Fig. 1). The age of volcanism decreases progressively southward from Volcán Cantaro (1·52–0·95 Ma) to Nevado de Colima (0·53–0·14 Ma) and to the currently active VDC, also known as Fuego de Colima (Cortés et al., 2005). A group of 11 cinder cones and associated lava flows with ages between 1·2 Ma and 60 ka erupted NW and NE of Nevado de Colima (Luhr & Carmichael, 1981; Carmichael et al., 2006; Fig. 1). Nine of the cinder cones produced basaltic alkaline lavas and scoria (basanite, leucite-basanite, and minette). The two remaining cones erupted calc-alkaline high-Al basalt and basaltic andesite, respectively. Luhr & Carmichael (1981) considered these high-Al basalts as plausible parental magmas to VDC andesites.
|
Volcán de Colima is located inside a horseshoe-shaped caldera that formed by sector collapse of an earlier cone 4·3 kyr ago (Luhr & Prestegaard, 1988
). VDC has been persistently active over the last 500 years, producing effusive and vulcanian explosive episodes, with major Plinian explosive eruptions in 1818 and 1913. Detailed accounts of the historical eruptive activity have been presented by Gonzalez et al. (2002) and Luhr (2002). The current eruptive episode started in November 1998 after 7 years of repose. The volcanic activity is characterized by effusive periods of dome growth and lava flows (in 1998–1999, 2001–2003 and 2004), alternating with periods of intermittent explosive vulcanian events. Several major vulcanian explosions resulting in the formation of pyroclastic flows occurred in 1999, 2001, 2003, and 2005. We examined a suite of samples that cover the temporal evolution since the onset of activity in 1998. All investigated magmas are andesitic in composition with 59·0–61·4 wt% SiO2 (H2O-free) (this study, Table 1; Luhr, 2002; Mora et al., 2002), as are all historical lavas (Luhr & Carmichael, 1980; Luhr, 2002). |
PETROLOGY OF 1998–2005 ANDESITES |
|---|
Like most andesitic magmas, Colima andesites show complex petrography comprising a large range of mineral textures and zoning patterns indicative of complex magmatic histories. All the 1998–2005 andesites show similar petrographic features with 30–38% vol.% phenocryst-sized minerals (i.e. irrespective of their origin). Plagioclase is the dominant phase (24–34% of phenocrysts), followed by orthopyroxene (6–9%), clinopyroxene (4–5%), titanomagnetite (0·5–1·5%), and rare, resorbed hornblende and olivine (<0·5%). The groundmass comprises a rhyolitic glass (72·8–77·1 wt% SiO2, H2O-free) with microlites of plagioclase, orthopyroxene, clinopyroxene, titanomagnetite and very rare ilmenite.
Plagioclase compositions show a similar range in all samples with broad main populations around An60–40 and subordinate Ca-rich plagioclases (to An82) (Fig. 2; Table 2). The majority of plagioclases show weak oscillatory zoning (Fig. 3a), but occasional crystals with textural evidence for major resorption events are also observed and represent about 5% of the plagioclases. The majority of resorbed plagioclases have corroded high-An (An82–65) cores filled in by low-An plagioclase (An55–45) and melt inclusions, forming a patchy texture on BSE images (Fig. 4e and f). These patchy plagioclases have oscillatory-zoned and inclusion-free rims with the same range of XAn as the oscillatory-zoned phenocrysts. In addition, rare plagioclases showing a concentric resorption zone associated with an increase in XAn from core to rim are also observed (Fig. 4d). Melt inclusions are often present along these resorption zones. Microlite plagioclases range from An52 to An33.
|
|
|
|
As for plagioclase, pyroxene compositions show a similar range in all samples with a broad main population around Mg-number 76–70 and Mg-rich outliers to Mg-number 88 (Fig. 2; Table 3). The majority of crystals show weak oscillatory zoning (Fig. 3b), but pyroxenes displaying complex zoning with several successive reverse (to Mg-number 88) and normal zones are ubiquitous (Fig. 3c). These pyroxenes always have a low-Mg-number rim compositionally similar to the weakly zoned pyroxenes. The high-Mg-number cores are always truncated and often show rounded to wavy contours (Fig. 3c) indicative of a resorption event prior to crystallization of their low-Mg-number rims. Pyroxene microlites are orthopyroxenes Mg-number 73–62.
|
Resorbed amphibole with Mg-number 71–63 (Table 4) is observed in all samples, but is rare (<0·5%). The degree of resorption ranges from thick reaction rims to complete pseudomorphs of fine- to medium-grained pyroxene, plagioclase and titanomagnetite. Resorbed amphiboles are generally mantled by clots of coarse-grained plagioclase, clinopyroxene, orthopyroxene and titanomagnetite and interstitial glass (Fig. 3d).
|
The presence of gabbroic crystal clots, up to a few millimetres in size, and showing a range of textures from glomerocrysts with vesiculated interstitial glass to glass-free plutonic fragments is a ubiquitous feature of these andesites (Fig. 3e). These clots contain orthopyroxene (Mg-number 87–70), clinopyroxene (Mg-number 88–70), calcic plagioclase (An82–50), titanomagnetite, and olivine (Fo73–70) (Tables 2 and 3). Olivine is surrounded by reaction coronas of pyroxenes and titanomagnetite. Pyroxenes in the clots commonly show complex zoning with multiple high-Mg-number (to Mg-number 82) zones, whereas coexisting plagioclase shows normal zoning. Pyroxenes in the clots generally have a distinct low-Mg (Mg-number
70) rim surrounding a resorbed high-Mg-number core (Fig. 3f). In addition, a distinct type of gabbroic crystal clot with slightly more evolved crystal compositions (Mg-number 80–75 and An75–51 for pyroxene and plagioclase, respectively) and higher modal proportion of plagioclase is observed in the 1998 lava flow only. These clots show clear signs of melting with sieve-textured plagioclases and dark interstitial glasses. Pyroxenes in the clots show strong zoning patterns and compositions (up to Mg-number 88) distinct from the main population of pyroxenes (Mg-number 76–70, weak oscillatory zoning). Instead, they have clear textural and compositional similarity to the Mg-rich outliers (Fig. 2), suggesting that disaggregation of the clots contributed a significant portion of the free phenocryst-sized crystals.
The largest gabbroic clots were hand picked from crushed samples and melted for 15 min in a furnace at 1450°C and 1 atm. The quenched homogeneous glasses show a range of composition mostly indicative of orthopyroxene, clinopyroxene or plagioclase accumulation. However, the average composition (n = 18), as well as the fragments with the least evidence of accumulation, has between 50 and 55 wt% SiO2 (Table 1) and is similar in composition to the basalts erupted at Pleistocene cinder cones surrounding the stratovolcanoes (Luhr & Carmichael, 1981; Carmichael et al., 2006). It is therefore likely that many gabbroic clots represent partially to fully crystallized basaltic to basaltic andesite melts, rather than true crystal cumulates.
Detailed point counting was done using BSE images for two representative samples (130305, 220101), to estimate the proportion of gabbroic clots, isolated mafic crystals texturally and compositionally similar to the crystals in gabbroic clots, and isolated crystals with weak oscillatory zoning. Considering their relatively simple zoning and the fact that they constitute the dominant unimodal compositional populations, the last group of crystals are likely to be true phenocrysts. Judging by the systematic presence of low Mg-number and XAn overgrowths around the isolated mafic crystals, it is likely that these crystals and the clots from which they derive represent antecrysts and ‘anteclasts’ entrained in the melt from which the phenocrysts grew. Results indicate that the gabbroic clots represent 5–15% of the total volume, the isolated mafic crystals 3–10%, and the weakly zoned crystals 23–25%, with plagioclase being the dominant phase (19–22%).
|
INTENSIVE PARAMETERS |
|---|
Temperature and fO2
Ilmenite is very rare in VDC andesites. Only two touching ilmenite–magnetite pairs were found as part of this study, giving temperatures of 918 and 931°C, and log(fO2) values of –10·76 and –10·47 using the method of Andersen & Lindsley (1988
). A single ilmenite–magnetite pair reported by Mora et al. (2002) in the 1998 lava flow gives a temperature of 901°C and a log(fO2) value of –11·38 according to the same calculation method.
Temperatures calculated for touching orthopyroxenes and clinopyroxenes pairs with Mg-number 76–70 range from 960 to 1031°C (average 995°C, n = 19) according to the method of Andersen et al. (1993), confirming the temperatures obtained by Luhr (2002) and Mora et al. (2002) for the 1998–1999 lava flow. Coexisting pyroxene pairs with Mg-number 86–78 in the gabbroic clots give temperatures between 1055 and 1103°C (average 1073°C, n = 6).
The applicability of the Putirka (2005) plagioclase–melt thermometer to melt inclusions and host-plagioclase has been discussed by Blundy et al. (2006). To ensure equilibrium between melt and plagioclase, only compositions of plagioclase in-fill within the MIs (see textural description of MIs below and Fig. 4e and f) were used in the calculation. This thermometer applied to VDC andesites using H2O contents in MIs obtained by SIMS (see below) gives temperatures between 959 and 1015°C (n = 6).
In summary, the two-pyroxene and plagioclase–melt thermometers record temperatures of crystallization of 996 ± 35°C. In contrast, the gabbroic clots, from which the isolated mafic crystals and plagioclase cores appear to derive, record a range of temperatures that extends to 1103°C, consistent with their crystallization from (hydrous) basaltic magmas. On account of their relatively rapid re-equilibration (Venezky & Rutherford, 1999), the Fe–Ti oxides thermometers are likely to provide an estimate of pre-eruptive temperature, 901–931°C.
Volatile content and pressure
Volatile contents in MIs obtained by SIMS indicate a range from 2·5 to 0·1 wt% H2O and 800 to 0 ppm CO2 in VDC 1998–2005 magmas. Assuming volatile-saturated conditions, as inferred on the basis of experimental petrology data by Moore & Carmichael (1998), the trapping pressures range from 150 to 10 MPa [calculated at 996°C using the method of Newman & Lowenstern (2002)].
|
MELT INCLUSION MORPHOLOGY AND OCCURRENCE |
|---|
MIs are found in plagioclase, orthopyroxene and clinopyroxene phenocrysts. In pyroxenes, MIs are mostly observed in weakly oscillatory-zoned phenocrysts, where they are randomly distributed from core to rim (Fig. 4a). MIs are also observed in pyroxenes from the gabbroic clots. In this latter case, the MIs are located within the low-Mg-number rims. These inclusions occasionally show a narrow neck separating them from pools of glass inside the clots (Fig. 4b and c), indicating that some of the MIs trapped interstitial melts in the gabbroic clots rather than the melt crystallizing in the wider magmatic system. Considering the textural and compositional indications of disaggregation of the clots, it is likely that some MIs in isolated pyroxene crystals have a similar origin.
In plagioclase, the large majority of MIs are associated with major resorption zones; inclusions in weakly oscillatory-zoned crystals are rare. MIs in resorbed plagioclases are located either along concentric resorption zones associated with a sharp increase in XAn (Fig. 4d) or more commonly in patchy-textured cores (Fig. 4e and f). In patchy cores, MIs are associated with low-An plagioclase (An55–45) infilling the resorbed high-An cores. Detailed observations of these textures indicate that the MIs are relics of larger pools of melt that have been replaced by low-An plagioclase, suggesting that extensive resorption of high-An crystals formed a network of large pools of melt that penetrated deeply into the resorbed crystal and subsequent partially crystallized, hence forming the intricate patchy textures. The MIs in plagioclase have negative crystal shapes and show low-An plagioclase along their margins, indicating crystallization of the host into the inclusion, in a fashion similar to that described by Blundy et al. (2006) in plagioclase-hosted MIs from Mount St. Helens.
MIs with one or more daughter crystals of Fe–Ti oxide, plagioclase, orthopyroxene or clinopyroxene are observed in all samples and all types of host crystal. These MI were not analysed as part of this study; only glassy MIs, free of daughter crystals and bubbles, at the exposed surface, were analysed. MIs in plagioclases were analysed despite textural evidence for crystallization of the host mineral within the inclusions, as previous studies have shown that this could occur before sealing of the inclusion, therefore not necessarily altering the ability of the inclusion to record the evolution of the surrounding melt (Blundy & Cashman, 2005; Blundy et al., 2006).
|
CHEMISTRY OF MELT INCLUSIONS AND GLASSES |
|---|
Melt inclusions
MIs show a broad range of SiO2 contents from 62 to 76 wt% (H2O-free) and large variations in all major and trace elements at a given SiO2 content (Figs 5 and 6
; Table 5). On the basis of their composition and host phase, the MIs can be divided into four groups: (1) low-K in pyroxenes; (2) high-K in pyroxenes; (3) low-K in plagioclase; (4) high-K in plagioclase (Fig. 5).
|
|
Low-K MIs in pyroxenes have between 64·0 and 74·5 wt% SiO2. For most major elements this group of MI plots between the bulk-rock and groundmass glass compositions (Fig. 5). SiO2 correlates negatively with CaO, MgO, FeO, Al2O3 and NaO, and positively with K2O. There is a compositional gap between the least evolved MI (dacite) and the most evolved bulk-rock (andesite). Compared with the bulk-rock, the low-K MIs have lower CaO, MgO, FeO and Sr, and higher SiO2, K2O and incompatible trace elements. These MIs have similar incompatible element ratios to the bulk-rocks (Fig. 6).
|
High-K melt inclusions in pyroxenes are characterized by higher K2O and lower CaO contents at a given SiO2 content compared with the low-K MIs and show a slightly higher range of SiO2 extending to 75·0 wt%. Al2O3, FeO and MgO contents in the two groups are similar. The high-K MIs show ranges of trace elements that overlap with the low-K MIs but extend to significantly higher Ba, Rb, Cs, Zr, Nb and Hf, and to lower Sr contents (Fig. 6). The high-K MIs, low-K MIs and the bulk-rocks have similar element ratios for all other incompatible elements (e.g. La/Nd, La/Sm, Nd/Th), but show a clear fractionation for Ba and Sr.
Low-K melt inclusions in plagioclase are rare; only two were found out of 39 inclusions analysed in plagioclase (Fig. 5). These MIs have similar compositions to the low-K MIs in pyroxenes, although they have slightly lower CaO and Al2O3, and higher FeO and MgO contents, suggesting modification by up to 13% post-entrapment crystallization of the host plagioclase. Hereafter, we assume that these MIs are part of the same group as the low-K MIs in pyroxenes.
High-K melt inclusions in plagioclase have high MgO, FeO and TiO2, and low Al2O3 contents (Fig. 5), plotting on trends clearly distinct from the other MIs and bulk-rocks. These MIs have similar features to the high-K group in pyroxenes (i.e. high K2O, Ba and Cs, and low Sr; Fig. 6), but are distinguished by generally high incompatible element contents and Rb/La, Zr/La and Nb/La ratios similar to the low-K MIs in pyroxenes.
Groundmass and interstitial glasses
Groundmass glasses have the most evolved compositions with a SiO2 (anhydrous) range from 72·8 to 77·1 (Fig. 7). They form elongated trends that overlap and extend the trends defined by low-K MIs in pyroxenes, suggesting a genetic relationship with this group of MI. At high SiO2 contents, there is also an overlap with high-K MIs in pyroxenes; however, the CaO and K2O vs SiO2 trends formed by these high-K MIs are oblique to the groundmass glasses trends, suggesting that they were produced by distinct magmatic processes. The abundance of plagioclase microlites prevented analyses of groundmass glasses by ion-microprobe; consequently, no trace element data are available for these glasses.
|
The interstitial glasses in the gabbroic clots have a broad range of silica contents from 65·5 to 75·6 wt% (anhydrous) and large ranges of K2O and CaO. Most of these glasses have high K2O and low CaO, and plot within the field defined by high-K MIs in pyroxenes (Fig. 7). Trace element contents, particularly high Rb and extension to low Sr, further suggest that some of the interstitial glasses in gabbroic clots have a chemical affinity with high-K MIs in pyroxenes (Fig. 6). The interstitial glasses in clots surrounding breaking-down amphiboles have 70·3–72·8 wt% SiO2 (anhydrous) and are chemically similar to the interstitial glasses in the gabbroic clots. The glasses from the amphibole clots define trends that plot along the prolongation of the high-K MIs in plagioclase (Fig. 7).
|
DISCUSSION |
|---|
Formation and modification of melt inclusions
Melt inclusions can be trapped during either crystallization or resorption of the host crystal (Roedder, 1984
). In both cases, they represent snapshots of the melt surrounding the host and, as such, are useful tracers of the magmatic evolution. MIs trapped during crystallization should, in principle, accurately represent the melt crystallizing in the magmatic system. However, there have been concerns that MIs may trap chemically distinct boundary layers developing at the surface of growing crystals (e.g. Lowenstern, 2003). Numerous studies have shown that the most MIs have major and trace element compositions comparable with their host-rock, irrespective of their host crystal, and consequently that they effectively record the composition of the crystallizing melt rather than the boundary-layer melts (e.g. Blundy & Cashman, 2005; Humphreys et al., 2006a; Portnyagin et al., 2007). On the other hand, MIs formed during resorption of the host may trap melts modified by dissolution of the host and/or surrounding crystals, hence recording grain-scale processes. Danyushevsky et al. (2004) suggested that compositionally anomalous MIs in primitive olivine may form by dissolution–reaction–mixing processes in semi-solidified crystal mush zones. A careful evaluation of the origin and significance of MIs is required to fully appreciate their relevance as tracers of magmatic evolution. In particular, identifying MIs that can be assumed to accurately record the melts crystallizing in the subvolcanic magmatic system, despite the difference in scale between these two entities, is fundamental to making any inference on the genesis and evolution of the magmas.
MIs may be modified after trapping by crystallization of the host phase on the wall or daughter crystals in the inclusion (e.g. Roedder; 1984). The MIs investigated in this study were carefully checked to ascertain the absence of daughter crystals. Crystallization on the inclusion walls may be difficult to evaluate on BSE images, particularly for pyroxenes, if there is no significant variation in composition between the host and the crystal forming around the walls of the inclusion. However, MIs in clinopyroxenes and orthopyroxenes show the same range of composition that spans the low-K and high-K groups. Consequently, the observed variations in composition and the two MI groups in pyroxenes cannot be the result of post-entrapment crystallization of the host crystal. In addition, the variations in K2O contents between the groups of MIs as well as the variations observed at a given SiO2 content within each group cannot be produced by crystallization of the host or of the daughter crystals (Fig. 5). The observed variations in CaO, Ba, Rb and Zr also cannot be accounted for by crystallization, as other compatible and incompatible elements show similar and restricted ranges of compositions (Figs 5 and 6). Consequently, crystallization of the host or daughter crystals can be ruled out as a cause of the compositional variations observed within and between the groups of MIs in pyroxenes.
High-K MIs in plagioclase show compositional trends distinct from the inclusions in pyroxenes (Fig. 5) and textural evidence for crystallization of the host into the inclusion (Fig. 4d–f). The major elements, with the exception of MgO, of the least evolved MIs of this group could be produced by 45–50% plagioclase crystallization from a melt similar in composition to the bulk-rock. However, the Sr contents of these MIs, although low, are still too high to be consistent with this possibility. Moreover, the trends shown by this group of MIs are perpendicular to the trends produced by crystallization of plagioclase alone (Fig. 5), suggesting that crystallization of the host, although likely to have modified the composition of these MIs, is not responsible for the trends observed and, consequently, another process is involved. Diffusion re-equilibration can clearly be ruled out, as this process cannot account for the observed variations in major elements (e.g. high MgO in plagioclase-hosted MIs and high K2O in some of the pyroxene-hosted MIs). In addition, the trace element contents in MIs show no systematic correlation with the mineral–melt partition coefficient of the host phase as would be expected if diffusion was the cause of the trace element variations (Cottrell et al., 2002). We conclude that MIs in pyroxenes essentially record true melts present in the magmatic system. MIs in plagioclase have potentially been affected to some extent by post-entrapment modification; however, the trends shown are mostly produced by genuine melts.
Significance of the melt inclusion groups
The occurrence of three distinct groups of MIs in VDC andesites suggests either the presence of distinct batches of magma crystallizing simultaneously or that the MIs were formed by distinct processes, raising the possibility that some of the MIs may not record the composition of the melts feeding the current eruption.
Low-K melt inclusions in pyroxenes and plagioclase
The linear relationship between the low-K MIs in pyroxenes and plagioclase, and the groundmass glasses (Figs 5 and 8) suggests that the process that produced the low-K MI trends ultimately produced the residual melts (i.e. groundmass glasses). Moreover, on most major elements vs SiO2 diagrams, the bulk-rocks plot along the low-SiO2 extension of these low-K MI–groundmass glass trends (Fig. 5), indicating that this group of MIs are representative of the bulk of the melt present in the magmatic system. The low-K MI trends are consistent with crystallization of an assemblage similar to the assumed ‘true’ phenocryst assemblage estimated on the basis of BSE images (Fig. 8). Least-squares models reproduce the range of compositions shown by the majority of low-K MIs (64–70 wt% SiO2) by 27·3% crystallization of the following assemblage (in wt%): 71·8% plagioclase + 10·1% clinopyroxene + 10·7% orthopyroxene + 7·3% Fe–Ti oxides (
r2 = 0·33). This corresponds closely to modal proportions of ‘true’ phenocrysts (23–25%). However, the overall range of compositions requires about 50% crystallization, suggesting that some of the MIs became isolated from the groundmass melt only after the onset of microphenocryst and microlite crystallization.
|
To further test the possibility that the low-K MIs represent the composition of the crystallizing melt, we calculated melt compositions in equilibrium with the main compositional populations of phenocrysts; that is, Mg-number 76–70 and An60–40 for pyroxenes and plagioclase, respectively (i.e. corresponding to the weakly zoned crystals assumed to be ‘true’ phenocrysts). These melt compositions were calculated on the basis of pyroxene Mg-number and Sr concentration measured in plagioclase by ion-microprobe, and using Fe/MgKD min/melt = 0·26–0·30 (Sisson & Grove, 1993
) for pyroxenes and the SrKD plag/melt expression from Blundy & Wood (1991). The calculated compositional ranges show good agreement with the low-K MIs (Fig. 9), indicating that this group of inclusions indeed record the evolution of the melt crystallizing in the magmatic system and consequently that this melt was dacitic and not andesitic as would be expected from the bulk-rock compositions.
|
High-K melt inclusions in pyroxenes
The trends toward high K2O and low CaO contents, clearly distinct from the bulk-rock compositions, shown by the least evolved high-K melt inclusions in pyroxenes, together with their distinctive trace element characteristics (Fig. 6), suggest that this group of MIs does not record a melt that is volumetrically significant in the magmatic system. The mismatch between the calculated melt compositions in equilibrium with the main compositional populations of phenocrysts and the high-K MIs (Fig. 9) further suggests that these inclusions do not record the composition of the melt from which the phenocrysts were crystallizing. Some of the interstitial glasses in the gabbroic clots show compositional similarities to these high-K MIs (i.e. high K2O and Rb, low CaO and Sr; Figs 6 and 7). In addition, there is clear textural evidence that some MIs formed within the clots, trapping these high-K interstitial melts (Fig. 4b and c). It is therefore apparent that the high-K group of MIs record grain-scale processes within the gabbroic clots and that subsequent disaggregation of the clots dispersed the hosts of these MIs throughout the magma.
The characteristically high K, Cs, Rb, Ba, Nb and Hf in this group of MIs and the interstitial glasses in the gabbroic clots strongly suggest melting of biotite in their genesis (Fig. 10). Low CaO and Sr are probably fingerprints of plagioclase crystallization, whereas the generally low MgO could indicate crystallization of pyroxenes. We propose that reaction of interstitial phases ± melt in near-solidus gabbroic fragments driven by heating to 950–1000°C as these fragments became entrained in the dacitic melt could produce the high-K MIs in pyroxenes and the interstitial melts in the gabbroic fragments. Dehydration melting experiments on tonalites comprising biotite, plagioclase and quartz produce orthopyroxene, clinopyroxene, plagioclase and melt at 900°C and 5 kbar (Singh & Johannes, 1996). At temperatures <850°C K-feldspar is an additional reaction product and biotite is stable up to 830°C. Progressive melting of biotite + melt ± plagioclase ± quartz will therefore produce an assemblage with K-feldspar at the onset of melting that gives way to plagioclase as the temperature increases. In addition, the An% of the produced plagioclase will increase as temperature increases further (Singh & Johannes, 1996). Consequently, the melt produced early during melting will be depleted in Ba and only once the limit of the K-feldspar stability is approached will the melt become enriched in Ba. Conversely, Sr will be strongly depleted in the early melts and become less depleted as the product plagioclase becomes more An-rich and DSr decreases (Blundy & Wood, 1991). This may account for the negative correlation observed for Ba and Sr vs SiO2, as well as the decoupling between Rb and Ba (not all high-Rb composition MIs have high Ba) in the most evolved MIs (Fig. 6). It should be noted that despite careful searching, biotite and K-feldspar have not been observed in any of the gabbroic clots, which may be expected considering that the temperature of the entraining dacitic melt is well above the stability field of both minerals. Resorption of high-Mg-number cores and crystallization of low-Mg-number rims in the gabbroic fragments (Fig. 3f) is likely to be a consequence of re-equilibration at 950–1000°C, as suggested by the occurrence of the high-K MIs in these rims.
|
) and Philpotts & Schnetzler (1970
), Bea et al. (1994
), Ewart & Griffin (1994
).High-K melt inclusions in plagioclase
The high-K MIs in plagioclase are chemically distinct from the bulk-rock compositions and the melts in equilibrium with the phenocrysts (Figs 5 and 9), indicating that, like the high-K MIs in pyroxenes, they do not represent a volumetrically dominant melt in the volcanic plumbing system. As discussed above, it is unlikely that the trends formed by these MIs are entirely due to post-entrapment crystallization of the host plagioclase, although the composition of the MIs may have been modified to some extent. The major element chemistry suggests that amphibole exercised a strong control on these trends (Fig. 11). Crystallization of amphibole is unlikely to produce the observed trends, as it would require a starting melt compositionally unlike any magma erupted or identified elsewhere in the Colima graben. Melting or dissolution of amphibole and mixing with melts similar to the residual melt in the andesitic magmas is a more likely scenario. The high Ba and Rb contents of these MIs cannot be accounted for by melting amphibole alone, as these elements are incompatible in this phase (e.g. Luhr & Carmichael, 1980
), suggesting that biotite is also a melting phase. Least-squares models reproduce well (r2 = 0·65) the composition of the least evolved MIs in plagioclase by adding 50·4% of an assemblage comprising 42·1% amphibole + 6·1% magnetite + 2·2% biotite and subtracting 12·1% plagioclase from a groundmass glass composition with 73·3 wt% SiO2. The amount of subtracted plagioclase could indicate either post-entrapment crystallization of the host plagioclase or a peritectic reaction in which plagioclase was crystallizing as amphibole was melting. On the other hand, the chemistry of the MIs indicates that pyroxenes, generally a product of amphibole breakdown experiments (e.g. Rutherford & Hills, 1993; Rutherford & Devine, 2003), were not produced during melting of the amphibole, suggesting that melting occurred outside the pyroxene stability field. There are several lines of experimental evidence to support this possibility.
|
The experimental phase diagram for a Mascota andesite compositionally similar to VDC andesites (Moore & Carmichael, 1998
) indicates that amphibole + plagioclase + magnetite are the stable phases under water-saturated conditions in melts with >5·5 wt% H2O (corresponding to P > 200 MPa). We present below several lines of evidence that the andesites are in fact magma mingling products. Consequently, the starting compositions used in these experiments are not fully representative of the melt crystallizing. In dacitic melts, the above assemblage plus biotite is stable at temperatures <800°C, pressures of 200–300 MPa and in melts with >6 wt% H2O (Scaillet & Evans, 1999; Costa et al., 2004; Holtz et al., 2005). For example, the 220 MPa experiments of Scaillet & Evans (1999) on a Pinatubo dacite show that isobaric heating under H2O-undersaturated conditions leads to the breakdown of amphibole ± biotite without concomitant precipitation of pyroxenes. It is therefore likely that the high-K MIs in plagioclase record melting of cumulates formed by crystallization of hydrous dacitic melts at pressure 
200 MPa. The temperatures recorded by the plagioclase–melt and two-pyroxene thermometers indicate a temperature of 1000°C for the dacitic melt in which the cumulates were entrained. For amphibole to melt without producing pyroxenes, a melt H2O content >6 wt% is required at these conditions. This is much higher than the values indicated by the MIs in the 1998–2005 samples (<2·5 wt%), although H2O contents up to 6·2 wt% in MIs of the 1913 eruption (J. F. Luhr, personal communication) clearly indicate the occurrence of hydrous melts at VDC.
Implications for Volcán de Colima
An andesitic volcano without andesitic melt
One of the striking features of Figs 5 and 8 is the compositional gap between the bulk-rocks and the low-K MIs (59–61 wt% vs 64–75 wt% SiO2, respectively), and the marked inflection between the bulk-rock and MI trends. This inflection could reflect a change from higher-pressure crystallization that generated the silicic melt from basaltic parents to lower-pressure crystallization of the silicic melt itself (e.g. Annen et al., 2006; Blundy et al., 2008; Humphreys et al., 2008). The experimentally produced liquid lines of descent for hydrous basaltic melts over a range of crustal pressures (Grove et al., 1997; Muntener et al., 2001; Sisson et al., 2005; Pichavant & Macdonald, 2007) reproduce the MI compositions (Fig. 12), suggesting that fractionation of mafic magmas within the crust is capable of producing the silicic liquids present at VDC. However, the curved experimental liquid lines of descent are inconsistent with the andesite bulk-rock compositions, which lie on linear trends that form chords across the curved liquid lines of descent. No experiments using basaltic starting compositions generate melts with such linear chemistry, most notable for their high MgO contents (Fig. 12). The location of the VDC andesites on straight lines between the basalts and the dacitic liquids is diagnostic of a mixing process. This is in agreement with Luhr & Carmichael's (1980) observation that the compatible trace element contents of VDC andesites cannot be reproduced by fractional crystallization alone and require mixing or mingling. The majority of phenocrysts in the studied samples show weak oscillatory zoning without evidence for abrupt changes in melt composition, indicating that mixing between liquids of contrasted composition is not the dominant process. A more plausible explanation, and one supported by the petrology, is that the andesite bulk-rock compositions are controlled by mechanical incorporation of gabbroic fragments into more evolved melts. The composition of the entraining melt is characterized graphically by the intersection between the individual linear trends defined by the bulk-rocks and by the MIs. This point of intersection coincides with the least evolved MIs (Figs 8 and 12), corroborating our earlier proposal that the low-K MIs accurately record the evolution of the melt from the onset of crystallization. It follows that the magmatic system feeding VDC comprises crystallizing dacitic magmas (64 wt% SiO2) and a pre-existing gabbroic plutonic body (50–55 wt% SiO2).
|
The high-K MIs in plagioclase further suggest the presence of an amphibole-, plagioclase- and biotite-bearing assemblage indicative of crystallization of hydrous silicic melts within the plumbing system of VDC. Moore & Carmichael (1998
) suggested that VDC andesites originally had 6·5 wt% H2O, started crystallizing upon reaching volatile-saturated conditions at 250 MPa, and subsequently re-equilibrated at low pressure. They suggested that high-An (65–85 mol%) plagioclases and resorbed amphiboles represent relicts of the onset of crystallization at even higher H2O pressure. Our data support crystallization of hydrous melts at 200–300 MPa. However, progressive re-equilibration to low pressure should produce abundant amphibole with reaction rims texturally similar to the decompression-driven breakdown textures obtained experimentally (e.g. Rutherford & Hills, 1993; Rutherford & Devine, 2003) and plagioclase should show progressive decrease in XAn (Housh & Luhr, 1991; Moore & Carmichael, 1998). Considering the very low proportion (<0·5%) of amphibole with reaction rims, the evidence for complete dissolution of amphibole and the strong resorption of plagioclase, we suggest instead that the high-pressure crystal cumulates were produced by earlier pulses of magmatism and that entrainment in the dacite melt resulted in dissolution of amphibole and resorption of calcic plagioclase as a result of heating. Annen et al. (2006) proposed that hydrous silicic melts that ascend adiabatically from the lower crust will cross the H2O-undersaturated liquidus and become superheated. This has two important consequences that may account for the range of textures and melt compositions observed in VDC andesites; (1) any entrained crystals (plutonic fragments or cumulates) will be resorbed as long as the ascending melt is superheated; (2) the melt attains H2O saturation at a super-liquidus temperature (assuming limited resorption-related cooling). H2O exsolution is therefore not accompanied by crystallization until the melt intersects the H2O-saturated liquidus. A dacitic melt with 6 wt% H2O, extracted at 1010°C and ascending adiabatically will reach volatile saturation at pH2O = 230 MPa but will not start crystallizing until it intersects the H2O-saturated liquidus at 70 MPa (Fig. 13), at which point melt inclusions may become trapped and will contain <3 wt% H2O. Any crystal or rock fragment entrained at pressures >70 MPa will be resorbed (Fig. 13) before crystallizing a rim in equilibrium with a dacitic melt (i.e. low An% and Mg-number for plagioclase and pyroxenes, respectively) once crystallization resumes at lower pressure, 
70 MPa. Patchy-textured plagioclase (Fig. 4e and f) and resorbed high-Mg-number pyroxenes (Fig. 3c and f) are likely to form by this process. We suggest that assimilation and partial dissolution of gabbroic fragments and high-pressure crystal cumulates is a consequence of superheated, hydrous silicic melts ascending through the plutonic roots formed by long-lived magmatic activity at VDC.
|
An alternative interpretation is that the gabbroic clots are restitic fragments resulting from incomplete separation of liquidus phases from residual silicic melts in the source regions (Chappell & White, 2001
; Tamura & Tatsumi, 2002). However, the high-K compositions of the interstitial melts in the gabbroic clots are clearly distinct from the crystallizing low-K dacitic melt. Thus the fragments are unlikely to be residues (restites) involved in the production of the silicic melts, although, as discussed below, segregation of the high-K melts from the gabbroic fragments did contribute to the bulk compositions of the magmas.
LILE-rich melts produced by assimilation of gabbros
The high-K MIs demonstrate that LILE-rich melts can be produced by partial melting of biotite-bearing gabbros. Considering the importance of LILE as tracers of fluids in subduction zones and the strong fluid-like signatures of these melts, assessing their possible contribution to the LILE budget of the bulk-rock compositions is important. On LILE/LREE and LILE/HFSE (high field strength element) diagrams, the bulk-rocks define trends oblique to the crystallization path defined by the low-K MIs (Figs 6 and 14a). These trends are consistent with a mingling line between the gabbroic fragments and melts with higher LILE/LREE ratios than the majority of low-K MIs in pyroxenes (Fig. 14), suggesting that the bulk melt composition before mingling is a mixture of low-K and high-K melts. The exact proportion of mixing depends on the bulk composition of the high-K melt and the gabbroic clots. Assuming that the average compositions of the gabbroic clots and the high-K MIs are representative of these values, about 30% mixing with high-K melts is required. The spread in CaO, K2O and incompatible trace elements shown by the low-K MIs (Figs 5, 6 and 14) could indicate either amalgamation of batches of dacitic melts with distinct compositions or, more likely, progressive input of high-K (LILE-rich) melts during crystallization. In any case, it is clear that the influx of LILE-rich melts produced by partial melting of biotite-bearing gabbros occurred late during the evolution of the dacitic melt and contributed a significant amount of the LILE budget of the bulk-rocks
|
Implications for the petrogenesis of andesites
Compositional gaps between melt inclusions and bulk-rock compositions are also documented at Mount St. Helens and Shiveluch volcanoes (Blundy & Cashman, 2005
; Humphreys et al., 2006a). In both cases, as for VDC, there are clear indications that the absence of andesitic melts is not an artefact of the melt inclusion record but reflects the fact that these andesites are mixtures of dacitic to rhyolitic melts and mafic fragments and crystals within the subvolcanic plumbing system. The absence of andesitic melts at these three typical andesitic stratovolcanoes raises questions regarding the true abundance of andesitic melts in subduction zones, a longstanding debate in igneous petrology (see Gill, 1981; Reubi & Blundy, in preparation).
Xenoliths of mafic cumulates and/or gabbros are common in andesitic rocks (Eichelberger, 1978; Kelemen, 1986; Heliker, 1995; Costa et al., 2002; Dungan & Davidson, 2004; Bacon et al., 2007; Streck et al., 2007). It is likely that the roots of many andesitic volcanoes contain gabbros and cumulate rocks left behind by previous pulses of magmatism. Field exposures of plutonic rocks thought to represent sub-volcanic systems, such as the Tertiary-age Blumone Complex in the Adamello Massif, Italy (Ulmer et al., 1983; John & Blundy, 1993), contain abundant gabbros and cumulate rocks. It is apparent from the above discussion that mingling between ascending silicic melts and these cumulates and/or gabbros is an important aspect of the petrogenesis of andesitic magmas.
Exotic high-K MIs compositionally similar to VDC high-K MIs in pyroxenes have been documented at Mount St. Helens (Blundy et al., 2008), where gabbroic fragments with high-K interstitial melts also occur (Heliker, 1995). A significant proportion of high-K MIs chemically distinct from any erupted magma have also been reported in a compilation of MIs from Kurile–Kamchatka volcanoes (Tolstykh et al., 2007). It appears that exotic high-K MIs potentially of similar origin to those at VDC may not be rare in arc magmas. In addition, Dungan & Davidson (2004) suggested grain-boundary melting of hydrous phases and plagioclase in mafic cumulates as a source of elevated alkaline and incompatible trace elements in dacite magmas from Tatara–San Pedro, Chile. These examples provides a stream of evidence that LILE-rich melts produced during assimilation of the mafic plutonic roots beneath arc volcanoes may have a significant effect on the LILE budget of the magmas. This process may be significant but difficult to identify in mafic magmas, as the assimilated crystals will tend to blend perfectly with true phenocrysts. The extent to which this process contributes to the LILE budget of arc magmas is an open question that needs to be considered before making assumptions on the flux of volatiles in subduction zones using this group of elements as a proxy for the fluid component.
|
CONCLUSION |
|---|
Melt inclusions in Volcán de Colima andesites show broad ranges of major and trace element contents that pinpoint multiple petrogenetic processes for their origin and demonstrate that the andesites are mixtures of crystals and melts formed over a range of pressure, temperature and volatile contents within the sub-volcanic plumbing system. Detailed petrographic and chemical investigations indicate that a significant proportion of the melt inclusions formed by partial melting of plutonic fragments and crystal cumulates during assimilation in a silicic melt. The melts formed by this assimilation process have ‘exotic’ high-LILE signatures that are clearly distinct from the bulk composition of the magmas. The exotic melt inclusions formed during assimilation have limited significance regarding the evolution of the melt feeding Volcán de Colima, but provide important petrogenetic information that would be difficult to establish on the basis of the petrography alone. The melt inclusions that demonstrably record the evolution of the melt crystallizing in the magmatic system are dacitic, in contrast to the andesitic composition of the bulk-rocks. This discrepancy is not due to a biased sample set or to post-entrapment crystallization of the melt inclusions, but reflects a compositional gap observed at several andesitic volcanoes. This gap is likely to be a fundamental aspect of arc magmatism. Assimilation of plutonic roots left behind by previous episodes of mafic magmatism largely masks the gap in the bulk-rock record. This is particularly true beneath large stratovolcanoes, where ascending batches of magma must repeatedly encounter and interact with plutonic rocks and cumulates left behind by precursor magmas. In this context it is striking that andesite magmas are much more abundant at continental arcs than at island arcs. We propose that cannibalization of ancestral igneous rocks is an inevitable consequence of the repeated fluxing of melts through long-lived sub-volcanic conduits of the type that characterize continental arc stratovolcanoes.
LILE-rich melts produced during assimilation of gabbros contribute to the high concentrations of these elements in the rocks and provide an alternative source to fluids from the slab. Evidence for exotic high-K melt inclusions at several arc volcanoes suggests that the potential contribution of this process should be carefully evaluated before using LILE as a proxy of fluid signature in arc magmas.
|
SUPPLEMENTARY DATA |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
ACKNOWLEDGEMENTS |
|---|
This work was supported by a Marie Curie Fellowship (to O.R.) and a NERC Senior Research Fellowship (to J.B.). Samples were provided by J. C. Mora and N. Varley, who also provided valuable help during field work in Colima. We are very grateful to Richard Hinton, Simone Kasemann and John Craven (Ion Microprobe Facility, University of Edinburgh) for assistance with ion microprobe analysis, and to Stuart Kearns (University of Bristol) for assistance with EMPA. The manuscript was improved by reviews by Yoshi Tamura, Suzanne Straub and Bill Leeman, and editorial comments by John Gamble.
*Corresponding author. Present address: Institute of Isotope Geochemistry and Mineral Resources, ETH Zurich, Claussiusstrasse 25, NW, 8092 Zurich. E-mail: olivier.reubi{at}erdw.ethz.ch
|
REFERENCES |
|---|
Andersen DJ, Lindsley DH, Davidson PM. Quilf—a Pascal program to assess equilibria among Fe–Mg–Mn–Ti oxides, pyroxenes, olivine, and quartz. Computers and Geosciences (1993) 19:1333–1350.[CrossRef]
Andersen DJ, Lindsley DH. Internally consistent solution models for Fe-Mg-Mn-Fe-Ti oxides. American Mineralogist (1988) 73:714–726.[Abstract]
Anderson AT. Magma mixing—petrological process and volcanological tool. Journal of Volcanology and Geothermal Research (1976) 1:3–33.[CrossRef][Web of Science]
Annen C, Blundy JD, Sparks R. SJ. The genesis of intermediate and silicic magmas in deep crustal hot zones. Journal of Petrology (2006) 47:505–539.
Bacon CR, Sisson TW, Mazdab FK. Young cumulate complex beneath Veniamin of caldera, Aleutian arc dated by zircon in erupted plutonic blocks. Geology (2007) 35:491–494.
Bea F, Pereira MD, Stroh A. Mineral leucosome trace-element partitioning in a peraluminous migmatite (a laser ablation-ICP-MS study). Chemical Geology (1994) 117:291–312.[CrossRef][Web of Science]
Blundy J, Cashman K. Rapid decompression-driven crystallisation recorded by melt inclusions from Mount St. Helens volcano. Geology (2005) 33:793–796.
Blundy JD, Wood BJ. Crystal-chemical controls on the partitioning of Sr and Ba between plagioclase feldspar, silicate melts, and hydrothermal solutions. Geochimica et Cosmochimica Acta (1991) 55:193–209.[CrossRef][Web of Science]
Blundy J, Cashman K, Humphreys M. Magma heating by decompression-driven crystallisation beneath andesite volcanoes. Nature (2006) 443:76–80.[CrossRef][Medline]
Blundy J, Cashman K, Berlo K. Evolving magma storage conditions beneath Mount St. Helens inferred from chemical variations in melt inclusions from the 1980–1986 and current eruptions. A volcano rekindled; the renewed eruption of Mount St. Helens, 2004–2006. US Geological Survey Professional Papers—Sherrod DR, Scott WE, Stauffer PH, eds. (2008) 1750:00–00.
Carmichael ISE, Frey HM, Lange RA, Hall CM. The Pleistocene cinder cones surrounding Volcan Colima, Mexico re-visited: eruption ages and volumes, oxidation states, and sulfur content. Bulletin of Volcanology (2006) 68:407–419.[CrossRef][Web of Science]
Chappell BW, White A. JR. Two contrasting granite types: 25 years later. Australian Journal of Earth Sciences (2001) 48:489–499.[CrossRef][Web of Science]
Cortés A, Garduno-Monry VH, Navarro-Ochoa C, Komorowski JC, Saucedo R, Macias JL, Gavilanes JC. Carta geologica del complejo volcanico de Colima. In: Cartas Geologicas y Mineras 10 (2005) Mexico City: UnaD.
Costa F, Dungan MA, Singer BS. Hornblende- and phlogopite-bearing gabbroic xenoliths from Volcan San Pedro (36°S), Chilean Andes: Evidence for melt and fluid migration and reactions in subduction-related plutons. Journal of Petrology (2002) 43:219–241.
Costa F, Scaillet B, Pichavant M. Petrological and experimental constraints on the pre-eruption conditions of Holocene dacite from Volcan San Pedro (36°S, Chilean Andes) and the importance of sulphur in silicic subduction-related magmas. Journal of Petrology (2004) 45:855–881.
Cottrell E, Spiegelman M, Langmuir CH. Consequences of diffusive reequilibration for the interpretation of melt inclusions. Geochemistry, Geophysics, Geosystems (2002) 3:U1–U26.
Danyushevsky LV, Della-Pasqua FN, Sokolov S. Re-equilibration of melt inclusions trapped by magnesian olivine phenocrysts from subduction-related magmas: petrological implications. Contributions to Mineralogy and Petrology (2000) 138:68–83.[CrossRef][Web of Science]
Danyushevsky LV, Leslie R. AJ, Crawford AJ, Durance P. Melt inclusions in primitive olivine phenocrysts: The role of localized reaction processes in the origin of anomalous compositions. Journal of Petrology (2004) 45:2531–2553.
Dungan MA, Davidson J. Partial assimilative recycling of the mafic plutonic roots of arc volcanoes: An example from the Chilean Andes. Geology (2004) 32:773–776.
Dungan MA, Wulff A, Thompson R. Eruptive stratigraphy of the Tatara–San Pedro complex, 36°S, Southern Volcanic Zone, Chilean Andes: Reconstruction method and implications for magma evolution at long-lived arc volcanic centers. Journal of Petrology (2001) 42:555–626.
Eichelberger JC. Andesitic volcanism and crustal evolution. Nature (1978) 275:21–27.[CrossRef]
Ewart A, Griffin WL. Application of proton-microprobe data to trace-element partitioning in volcanic rocks. Chemical Geology (1994) 117:251–284.[CrossRef][Web of Science]
Gamble JA, Wood CP, Price RC, Smith I. EM, Stewart RB, Waight T. A fifty year perspective of magmatic evolution on Ruapehu Volcano, New Zealand: verification of open system behaviour in arc volcano. Earth and Planetary Science Letters (1999) 170:301–314.[CrossRef][Web of Science]
Gill JB. Orogenic Andesites and Plate Tectonics (1981) Berlin: Springer.
Gonzalez MB, Ramirez JJ, Navarro C. Summary of the historical eruptive activity of Volcan de Colima, Mexico 1519–2000. Journal of Volcanology and Geothermal Research (2002) 117:21–46.[CrossRef][Web of Science]
Grove TL, DonnellyNolan JM, Housh T. Magmatic processes that generated the rhyolite of Glass Mountain, Medicine Lake volcano, N California. Contributions to Mineralogy and Petrology (1997) 127:205–223.[CrossRef][Web of Science]
Heliker C. Inclusions in Mount St. Helens Dacite Erupted from 1980 through 1983. Journal of Volcanology and Geothermal Research (1995) 66:115–135.[CrossRef][Web of Science]
Higuchi H, Nagasawa H. Partition of trace elements between rock-forming minerals and host volcanic rocks. Earth and Planetary Science Letters (1969) 7:281–287.[CrossRef][Web of Science]
Holtz F, Sato H, Lewis J, Behrens H, Nakada S. Experimental petrology of the 1991–1995 Unzen dacite, Japan. Part I: Phase relations, phase composition and pre-eruptive conditions. Journal of Petrology (2005) 46:319–337.
Housh TB, Luhr JF. Plagioclase–melt equilibria in hydrous systems. American Mineralogist (1991) 76:477–492.[Abstract]
Humphreys MCS, Blundy JD, Sparks R. SJ. Magma evolution and open-system processes at Shiveluch Volcano: Insights from phenocryst zoning. Journal of Petrology (2006a) 47:2303–2334.
Humphreys MCS, Kearns SL, Blundy JD. SIMS investigation of electron-beam damage to hydrous, rhyolitic glasses: Implications for melt inclusion analysis. American Mineralogist (2006b) 91:667–679.
Humphreys MCS, Blundy JD, Sparks R. SJ. Shallow-level decompression crystallisation and deep magma supply at Shiveluch Volcano. Contributions to Mineralogy and Petrology (2008) 155:45–61.[CrossRef][Web of Science]
John BE, Blundy JD. Emplacement-related deformation of granitoid magmas, southern Adamello Massif, Italy. Geological Society of America Bulletin (1993) 105:1517–1541.
Kelemen PB. Assimilation of ultramafic rock in subduction-related magmatic arcs. Journal of Geology (1986) 94:829–843.[Web of Science]
Lowenstern JB. Melt inclusions come of age: volatiles, volcanoes and Sorby's legacy. In: Melt Inclusions in Volcanic Systems: Methods, Applications and Problems—De Vivo B, Bodnar RJ, eds. (2003) Amsterdam: Elsevier. 1–21.
Luhr JF. Glass inclusions and melt volatile contents at Paricutin Volcano, Mexico. Contributions to Mineralogy and Petrology (2001) 142:261–283.[Web of Science]
Luhr JF. Petrology and geochemistry of the 1991 and 1998–1999 lava flows from Volcan de Colima, Mexico: implications for the end of the current eruptive cycle. Journal of Volcanology and Geothermal Research (2002) 117:169–194.[Web of Science]
Luhr JF, Carmichael I SE. Colima Volcanic Complex, Mexico. 1. Post-caldera andesites from Volcan Colima. Contributions to Mineralogy and Petrology (1980) 71:343–372.[CrossRef][Web of Science]
Luhr JF, Carmichael ISE. The Colima Volcanic Complex, Mexico. 2. Late-Quaternary cinder cones. Contributions to Mineralogy and Petrology (1981) 76:127–147.[CrossRef][Web of Science]
Luhr JF, Prestegaard KL. Caldera formation at Volcano Colima, Mexico, by a large Holocene volcanic debris avalanche. Journal of Volcanology and Geothermal Research (1988) 35:335–348.[CrossRef][Web of Science]
Moore G, Carmichael ISE. The hydrous phase equilibria (to 3 kbar) of an andesite and basaltic andesite from western Mexico: constraints on water content and conditions of phenocryst growth. Contributions to Mineralogy and Petrology (1998) 130:304–319.[CrossRef][Web of Science]
Mora JC, Macias JL, Saucedo R, Orlando A, Manetti P, Vaselli O. Petrology of the 1998–2000 products of Volcan de Colima, Mexico. Journal of Volcanology and Geothermal Research (2002) 117:195–212.[CrossRef][Web of Science]
Muntener O, Kelemen PB, Grove TL. The role of H2O during crystallisation of primitive arc magmas under uppermost mantle conditions and genesis of igneous pyroxenites: an experimental study. Contributions to Mineralogy and Petrology (2001) 141:643–658.[Web of Science]
Newman S, Lowenstern JB. VOLATILECALC: a silicate melt–H2O–CO2 solution model written in Visual Basic for Excel. Computers and Geosciences (2002) 28:597–604.[CrossRef]
Philpotts JA, Schnetzler CC. Phenocryst–matrix partition coefficients for K, Rb, Sr and Ba, with applications to anorthosite and basalt genesis. Geochimica et Cosmochimica Acta (1970) 34:307–322.[CrossRef][Web of Science]
Pichavant M, Macdonald R. Crystallisation of primitive basaltic magmas at crustal pressures and genesis of the calc-alkaline igneous suite: experimental evidence from St Vincent, Lesser Antilles arc. Contributions to Mineralogy and Petrology (2007) 154:535–558.[CrossRef][Web of Science]
Portnyagin M, Hoernle K, Plechov P, Mironov N, Khubunaya S. Constraints on mantle melting and composition and nature of slab components in volcanic arcs from volatiles (H2O, S, Cl, F) and trace elements in melt inclusions from the Kamchatka Arc. Earth and Planetary Science Letters (2007) 255:53–69.[CrossRef][Web of Science]
Pouchou JL, Pichoir F. PAP correction procedure for improved quantitative microanalysis. In: Microbeam Analysis—Armstrong JT, ed. (1984) San Francisco, CA: San Francisco Press. 104–106.
Putirka KA. Igneous thermometers and barometers based on plagioclase plus liquid equilibria: Tests of some existing models and new calibrations. American Mineralogist (2005) 90:336–346.
Reubi O, Nicholls IA, Kamenetsky VS. Early mixing and mingling in the evolution of basaltic magmas: evidence from phenocryst assemblages, Slamet Volcano, Java, Indonesia. Journal of Volcanology and Geothermal Research (2003) 119:255–274.[CrossRef][Web of Science]
Roedder E, ed. Fluid Inclusions. Mineralogical Society of America. In: Reviews in Mineralogy (1984) 14.
Rutherford MJ, Devine J. Magmatic conditions and magma ascent indicated by hornblende phase equilibria and reactions in the 1995–2002 Soufrière Hills magma. Journal of Petrology (2003) 44:1433–1454.
Rutherford MJ, Hills PM. Magma ascent rate from amphibole breakdown: an experimental study applied to the 1980–1986 Mount St. Helens eruptions. Journal of Geophysical Research (1993) 98:19667–19685.[CrossRef]
Scaillet B, Evans BW. The 15 June 1991 eruption of Mount Pinatubo. I. Phase equilibria and pre-eruption P–T–fO2–fH2O conditions of the dacite magma. Journal of Petrology (1999) 40:381–411.[CrossRef][Web of Science]
Singh J, Johannes W. Dehydration melting of tonalites. 2. Composition of melts and solids. Contributions to Mineralogy and Petrology (1996) 125:26–44.[CrossRef][Web of Science]
Sisson TW, Grove TL. Experimental investigations of the role of H2O in calc-alkaline differentiation and subduction zone magmatism. Contributions to Mineralogy and Petrology (1993) 113:143–166.[CrossRef][Web of Science]
Sisson TW, Ratajeski K, Hankins WB, Glazner AF. Voluminous granitic magmas from common basaltic sources. Contributions to Mineralogy and Petrology (2005) 148:635–661.[CrossRef][Web of Science]
Spandler C, O’Neill H. SC, Kamenetsky VS. Survival times of anomalous melt inclusions from element diffusion in olivine and chromite. Nature (2007) 447:303–306.[CrossRef][Web of Science][Medline]
Streck MJ, Leeman WP, Chesley J. High-magnesian andesite from Mount Shasta: A product of magma mixing and contamination, not primitive mantle melt. Geology (2007) 35:351–354.
Tamura Y, Tatsumi Y. Remelting of an andesitic crust as a possible origin for rhyolitic magma in oceanic arcs: An example from the Izu–Bonin arc. Journal of Petrology (2002) 43:1029–1047.
Tolstykh ML, Naumov VB, Babansky AD. Composition, volatile components, and trace elements in andesitic melts of the Kurile–Kamchatka region. Acta Petrologica Sinica (2007) 23:93–104.
Ulmer P, Callegari E, Sonderegger UC. Genesis of the mafic and ultramafic rocks and their genetical relations to the tonalitic–trondhjemitic granitoids of the southern part of the Adamello Batholith (Northern Italy). Memorie della Società Geologica Italiana (1983) 26:171–222.
Venezky DY, Rutherford MJ. Petrology and Fe–Ti oxide reequilibration of the 1991 Mount Unzen mixed magma. Journal of Volcanology and Geothermal Research (1999) 89:213–230.[CrossRef][Web of Science]
Wallace PJ. Volatiles in subduction zone magmas: concentrations and fluxes based on melt inclusion and volcanic gas data. Journal of Volcanology and Geothermal Research (2005) 140:217–240.[CrossRef][Web of Science]
Witter JB, Kress VC, Newhall CG. Volcan Popocatepetl, Mexico. Petrology, magma mixing, and immediate sources of volatiles for the 1994–present eruption. Journal of Petrology (2005) 46:2337–2366.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  H. E. Frimmel, A. Zeh, B. Lehrmann, D. Hallbauer, and W. Frank Geochemical and Geochronological Constraints on the Nature of the Immediate Basement next to the Mesoarchaean Auriferous Witwatersrand Basin, South Africa J. Petrology, December 1, 2009; 50(12): 2187 - 2220. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||