Journal of Petrology Advance Access published online on December 20, 2008
Journal of Petrology, doi:10.1093/petrology/egn067
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
A Rhyolite Compositional Continuum Governed by Lower Crustal Source Conditions in the Taupo Volcanic Zone, New Zealand
1Department of Geological Sciences, University of Canterbury, PB 4800, Christchurch 8020, New Zealand
2Department of Geological Sciences, Michigan State University, 206 Natural Sciences Bldg., East Lansing, MI 48824-1115, USA
Received May 6, 2008; Revised typescript accepted November 17, 2008
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONANALYTICAL TECHNIQUESGEOCHEMICAL DATASETDATA TREATMENT: POLYTOPIC VECTOR...RHYOLITE PETROGENESISMAJOR- AND TRACE-ELEMENT...CONCLUSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Rhyolites generated in the modern Taupo Volcanic Zone (TVZ), New Zealand, have previously been interpreted as having evolved by a combination of extensive fractional crystallization of mantle-derived mafic magmas and limited crustal assimilation (up to 25%). Polytopic vector analysis (PVA), a form of multivariate statistical analysis, of the major-element compositions of over 475 basaltic to rhyolitic bulk-rock samples, representing over 600 kyr of volcanism within the TVZ, has provided a robust platform for rhyolite characterization and new insights into rhyolite petrogenesis. There is a continuum of compositions between two rhyolite end-member magma types (EM1 and EM2), which have been identified on the basis of the PVA and which have distinct petrological and geochemical characteristics, as follows. EM1: crystal-rich (up to 45%), hydrous phases (± hornblende ± biotite ± cummingtonite), high Aluminum Saturation Index [ASI; molar Al2O3/(CaO + Na2O + K2O)], low FeO*/MgO (calc-alkaline series), depleted abundances of middle rare earth elements (MREE) and Y, and high Sr; EM2: crystal-poor (<10%), anhydrous phases (orthopyroxene ± clinopyroxene), high FeO*/MgO (tholeiitic series), low ASI, less depleted MREE and Y, and low Sr. The range of ASI values, and relative depletion in MREE and Y in the rhyolites is consistent with the results of experiments to constrain the partial melting behaviour of amphibolite at crustal pressures. The major- and trace-element data are also consistent with 50–60% equilibrium crystallization of a crustally contaminated, hornblende-bearing andesite to produce the TVZ rhyolites. Distinct major- and trace-element variations along the continuum between the two rhyolite end-member types can be effectively modelled by simulating changes in the temperature–fO2–fH2O conditions in the lower crust where mantle-derived mafic magmas are stored and differentiate. Low T and high fO2 and fH2O in the crustal magma storage zone promote abundant hornblende crystallization and suppress plagioclase crystallization, which produces the EM1 type rhyolite. By increasing the temperature and/or lowering fO2 and fH2O in the magma storage region, plagioclase becomes more dominant and hornblende crystallization is suppressed, producing more EM2-like rhyolitic magma types.
KEY WORDS: rhyolite; magma genesis; water; Taupo Volcanic Zone; polytopic vector analysis
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOCHEMICAL DATASETDATA TREATMENT: POLYTOPIC VECTOR...RHYOLITE PETROGENESISMAJOR- AND TRACE-ELEMENT...CONCLUSIONSUPPLEMENTARY DATAREFERENCES |
|---|
The New Zealand tectonic regime is characterized by oblique subduction of Pacific oceanic crust beneath the continental crust of the Indo-Australian plate. The central Taupo Volcanic Zone (TVZ) is a region of thinned continental crust located behind the volcanic front to the east (Fig. 1). In this central region, the volcanism is predominantly bimodal, dominated by rhyolitic and basaltic volcanism with subordinate intermediate types. Rhyolite is volumetrically dominant in the modern (1·6 Ma to present) central TVZ, representing over 90% of the deposits (Wilson et al., 2008
). Twenty-five caldera-forming eruptions of rhyodacitic to rhyolitic composition have produced over 6000 km3 of magma during this period. A comprehensive review of the eruptive activity has been given by Wilson et al. (1993, 2008).
|
Early workers (Ewart & Stipp, 1968
; Cole, 1979) proposed that the voluminous rhyolitic magmas in the TVZ originated from anatexis of metasedimentary crust. This view was widely accepted at the time: Reid (1983) concluded that production of rhyolitic magmas by anatexis of metasedimentary crust was more consistent with the results of trace-element modelling than partial melting or fractional crystallization of a primary, mantle-derived basaltic parent. These ideas were later advanced as new isotopic, trace-element and experimental data became available, as summarized by Graham et al. (1995). Those workers concluded that binary mixing between a metasedimentary melt and a mantle-derived basaltic melt, followed by fractional crystallization or a combined assimilation plus fractional crystallization process (AFC), was consistent with the available isotopic (Sr–Nd–Pb–O), major- and trace-element and geophysical evidence. However, details of these processes are not well constrained, with uncertainties regarding the end-member magma compositions, contamination rate, bulk liquid–crystal distribution coefficients and, especially, the differentiation process operating (Graham et al., 1995, p. 81).
To address the above uncertainties we present new evidence based on an expanded bulk-rock geochemical dataset, mineral chemistry and Sr–Nd isotope data, representing c. 600 kyr of the last 1·6 Myr of the modern volcanism, which is broadly consistent with the conclusions of Graham et al. (1995). We develop a revised, integrated model for the evolution of the TVZ rhyolites. To analyze this large geochemical dataset, polytopic vector analysis (PVA) (Johnson et al., 2002; Vogel et al., 2008), a powerful multivariate statistical tool, was used to characterize and evaluate the chemical variation of the TVZ rhyolites. This analysis constrains the rhyolitic end-member compositions and provides a guide for investigating their petrogenesis, based upon major- and trace-element modelling.
In this study, we demonstrate that the TVZ rhyolitic magmas are genetically related to primary, mantle-derived basaltic melts. These basaltic magmas are intruded into the lower continental crust (20–30 km), where they assimilate metasedimentary crust and crystallize in a ‘hot-zone’. Evolved andesitic melts are produced by fractional crystallization of these magmas and accumulate and fractionate at c. 15–20 km depth, forming a mushy accumulation and magma ascent zone at the top of this ‘hot-zone’ under water-saturated or near-saturated conditions. Extraction of evolved low-silica rhyolitic magma, under varying P–T, fH2O, and fO2 conditions, leads to production of a continuum of rhyolitic magma types. We characterize both the end-members and the continuous variations of these rhyolitic magmas, providing new insights into both the petrogenetic evolution of the rhyolites and their geochemical diversity.
|
ANALYTICAL TECHNIQUES |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOCHEMICAL DATASETDATA TREATMENT: POLYTOPIC VECTOR...RHYOLITE PETROGENESISMAJOR- AND TRACE-ELEMENT...CONCLUSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Pumice and lava samples used for this study were trimmed using a diamond grinder and saw to remove any weathered rind or organic matter. Sample fragments were then cleaned in a sonic bath of distilled water for 10 min to remove any surface contaminants from the grinder and saw, then placed in an oven at 100°C for a minimum of 24 h. These dry fragments were rough crushed using a tungsten-carbide pneumatic press. Any additional weathered fragments were removed from the sample aggregate prior to milling in a Frisch Planetary agate mill.
For the X-ray fluorescence spectrometry (XRF) and inductively coupled plasma (ICP) analyses, 3 g of the milled rock powder and 9·0 g of lithium tetraborate (Li2B4O7), along with 0·5 g of ammonium nitrate (NH4NO3; used as an oxidizer), were fused in platinum crucibles at 1000°C for 20–30 min on an orbital mixing stage. The melt was then poured into platinum molds, forming a glass disk that was analyzed by XRF using a Bruker S-4 system. XRF major-element analyses were reduced using a fundamental parameter data reduction method and Bruker Spectra Plus® software, whereas XRF trace-element (Rb, Sr, and Zr) abundances were obtained using standard linear regression techniques using the ratio of the element peak to the Rh Compton peak, which corrects for mass absorption. Prior to any calculations, the background signal was subtracted from the standards and samples. Major- and trace-element concentrations in the samples were calculated based on a linear regression method using BHVO, W-2, STM-1, MRG-1, SY-2, SY-3, DNC-1, PCC-1 JA-2, JA-3, BIR, QLO-1, and RGM-1 standards.
The rare earth elements (REE) Nb, Ta, Hf, Ba, Y, Th, U and Pb were analyzed by laser-ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) on the same glass disks as used for XRF analyses. A Cetac LSX200+ laser ablation system was used, coupled with a Micromass Platform ICP-MS system, using strontium determined by XRF as an internal standard. Trace-element data reduction was done using MassLynx software. Element concentrations in the samples were calculated based on a linear regression method using well-characterized standards. All the major- and trace-element whole-rock analyses reported here were obtained at Michigan State University. Precision and accuracy of both the XRF and LA-ICP-MS chemical analyses have been reported by Vogel et al. (2006).
Hornblende and plagioclase compositions were determined using a Cameca SX 100 electron microprobe at the University of Michigan equipped with five wavelength spectrometers, using an accelerating potential of 15 kV, a focused beam with a 10 µm spot size, counting time of 
3 min per mineral, and a 10 nA beam current. Standards used were natural fluor-topaz (FTOP), natural jadeite (JD-1), natural grossular, Quebec (GROS), natural adularia, St. Gothard, Switzerland (GKFS), synthetic apatite (BACL), synthetic Cr2O3, and synthetic FeSiO3 (FESI).
The 87Sr/86Sr and 143Nd/144Nd isotopic compositions of selected samples were measured at the Victoria University Geochemistry Laboratory Wellington, New Zealand by multi-collector (MC)-ICP-MS using a Nu-Plasma system. The mass spectrometric and sample preparation techniques used are a modified version of those described in detail by Waight et al. (2002). The isotopic ratios were corrected for mass spectrometric fractionation using an exponential law and normalized to 86Sr/87Sr = 0·710248 (SRM987) and 143Nd/144/Nd = 0·512980 (BHVO-2). Precision (2
) is ± 0·00002.
|
GEOCHEMICAL DATASET |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOCHEMICAL DATASETDATA TREATMENT: POLYTOPIC VECTOR...RHYOLITE PETROGENESISMAJOR- AND TRACE-ELEMENT...CONCLUSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Our dataset consists of 404 bulk-rock major- and trace-element analyses from this study and an additional 1200 analyses from the literature (Table 1). Representative data are given in Table 2, and the complete dataset is available in Electronic Appendix 1 on the Journal of Petrology website at http://www.petrology.oupjournals.org. For simplification and to reduce the clutter in diagrams, the samples have been separated by volcanic centre and appropriate time segment. Table 1 provides a key to the symbols used in the figures for each volcanic centre, including each major magmatic episode.
|
|
The mineralogy (composition and modal per cent) of basalt–rhyolite clasts from the Matahina ignimbrite and post-caldera collapse deposits (
330 ka; Manning, 1995) were analyzed to investigate the physical conditions of magmatic crystallization (e.g. pressure–temperature) and for modelling crystal fractionation processes. The mafic (basalt to andesite) clasts were specifically selected to best represent the potential parental magma compositions. The electron microprobe data for these clasts are given in Electronic Appendix 2.
Secondary hydration and alteration
The samples analyzed in this study were all single pumice clasts, vitric fiamme or lava samples. Loss on ignition (LOI) provides an approximation of the extent of secondary hydration of the glass. Because the samples were fused prior to analysis all the water was driven off during the fusion process. LOI was calculated during the XRF major-element data reduction by determining the initial weight loss and then applying a correction for reduced oxygen concentration to the calculated major-element concentrations. This process was repeated until no changes occurred in the calculated major-element concentrations. In comparing samples where LOI values are measured directly by chemical techniques and calculated based on XRF data, the calculated LOI values are within 10% of the measured values.
Glass and hydrous phenocryst contents are highly variable within the dataset. Hydrous phenocrysts generally make up <1 wt % of the samples; however, some of the more crystal-rich rocks contain up to 10 wt % hydrous phenocrysts. The water contributions from these hydrous minerals are variable, but relatively low, contributing much less than 1 wt % to LOI values in almost all cases. LOI values are dominated by glass hydration. Low crystallinity samples tend to have higher LOI values, consistent with hydrated glass controlling the LOI values. In some cases, secondary alteration (hydrous minerals) has occurred, which will increase the LOI values. These alteration products can be seen in thin section.
Several petrological and geochemical observations are important when considering such a large dataset. First, if the alkali values contain petrological information, the alkali concentrations in the pumice should not be correlated with the measured LOI. Volcanic glass is susceptible to hydration, and once the glass is fully hydrated, ion exchange between the H2O and glass may occur (Noble, 1967). Figure 2a shows the wide range in LOI values relative to the Na/K ratio. A direct relationship is not observed between Na/K and LOI for any of the volcanic suites, although highly variable Na/K ratios are observed. Na2O is correlated with SiO2 for most of the sample suites, and some distinct groups are also observed (Fig. 2b). The more mafic samples are not plotted in Fig. 2b to make this relationship clearer. As Na2O does not correlate with LOI, the variability in Na2O does not reflect secondary hydration processes. Although there has undoubtedly been some alkali mobility, the lack of a strong correlation with LOI within the datasets indicates to us that this exchange is insignificant.
|
A number of samples were removed from the dataset prior to LOI screening. Samples with Ce anomalies (relative to La and Pr) were filtered out from the dataset because this anomaly has previously been associated with weathering and REE mobility (Patino et al., 2003
). Thin sections were screened for the occurrence of zeolites and other secondary alteration minerals. Some pumice-clast and lava samples from published studies were evaluated using X-ray diffraction (XRD) to determine the presence of secondary alteration minerals. These samples were removed from the dataset before our initial screening of the entire dataset. A number of previous studies of the more recent TVZ volcanic rocks (Schmitz & Smith,, 2004; Wilson et al., 2006; Shane et al., 2007) have reported pumice with LOI between 4·0 and 5·0% and glass with up to 6·0 wt % by difference (Fig. 2c). Therefore, on this basis, we have employed a cut-off of analytical totals <95 wt %, regardless of whether or not any secondary alteration minerals were seen in the thin sections.
Bulk-rock and mineral chemistry
Mafic (basalt to andesite) samples
Our study considers the potential genetic relationship between mafic magmas with high water contents (as reflected by the presence of hornblende) and the rhyolites of the TVZ, based on both bulk-rock geochemistry and mineral chemistry. Hornblende-bearing mafic samples are discussed in this section along with data for the currently active basaltic to andesitic volcanoes, hereafter referred to as ‘modern arc’, for comparison (Brown, 1994; Wilson, 2006; this study). These eruptive deposits range from calc-alkaline to tholeiitic in composition (Fig. 3a). Hornblende-bearing samples are similar to modern mafic samples in most major-element oxides (Fig. 3c, e and f); however, several have distinctly different CaO and TiO2 contents (Fig. 3b and d). Trace-element abundances also have a similar range for both hornblende-bearing and modern types (Fig. 4a–e), with the exception of Zr, for which several hornblende-bearing samples contain much higher Zr contents at a similar SiO2 wt % (Fig. 4f) than the modern mafic samples. Hornblende-bearing samples are metaluminous to peraluminous with Aluminum Saturation Index [ASI; molar Al2O3/(CaO + Na2O + K2O)] values ranging between 0·6 and 1·4 (Fig. 5). Published 87Sr/86Sr data are plotted against an index of differentiation (SiO2 wt %) for the modern mafic samples, a few samples in the low-dacite range and the hornblende-bearing mafic samples (Fig. 6). Strontium isotopic compositions range from 0·707905 to 0·702560 and neodymium from 0·513129 to 0·512615.
|
|
|
|
Hornblende-bearing mafic rocks
In this section, we summarize the mineral chemistry data for the hornblende-bearing mafic igneous rocks. The modal abundance, size and shape of the plagioclase and hornblende vary considerably in each of these suites of rocks. Brown (1994
) described the Atiamuri dolerite as medium to fine grained (1–2 mm), highly crystalline (>95%) with strongly zoned plagioclase laths (An83 to An36), abundant euhedral or subhedral hornblende, anhedral clinopyroxene, and rare olivine. Some large Tschermakitic-hornblende grains have partially replaced clinopyroxene. A lack of resorption or other disequilbrium textures led Brown (1994) to suggest that these represent the products of primary magmatic crystallization, not reaction.
A suite of hornblende-bearing mafic samples exhumed as part of the 26·5 ka, 500 km3, caldera-forming Oruanui eruption are also included in our evaluation. These samples were described by Wilson et al. (2006) as fine-grained and crystal-poor (<13%), with a mineral assemblage of plagioclase + clinopyroxene ± orthopyroxene ± olivine ± hornblende + magnetite ± ilmenite. Plagioclase compositions are variable and show signs of disequilibrium.
Hornblende-bearing samples from a post-caldera collapse deposit, which followed the emplacement of the Matahina ignimbrite, sensu stricto, at 330 ka, exhibit chemical disequilibrium in the phenocryst contents. These mafic samples are crystal-rich (up to 50%) with a mineral assemblage of plagioclase + hornblende ± orthopyroxene ± clinopyroxene + magnetite ± ilmenite. Phenocrysts are highly variable in shape, size, and abundance in this suite of rocks. Many show obvious signs of mingling between a mafic and a felsic melt composition. We selected only a few of the most mafic samples to include in this study.
In general, all of the above samples have plagioclase compositions ranging from An94 to An31, with a corresponding increase in FeO from 0·2 to 0·8 wt % (Fig. 7a and b). Amphiboles are magnesio-hastingsite hornblende, Tschermakitic-hornblende, and magnesio-hornblende (Fig. 7c and d), based on the classification scheme of Leake et al. (1997), and range from 0·6 to 2·4 AlT (a.f.p.u.).
|
Rhyodacite–rhyolite samples
The TVZ rhyolites range from tholeiitic to calc-alkaline in composition (Fig. 8a), projecting the Miyashiro (1974
) FeO*/MgO discriminant ratio into the rhyolite field. TVZ rhyolites are slightly to moderately peraluminous to slightly metaluminous with ASI values ranging between 0·9 and 1·75 (Fig. 5). Major-element oxides (TiO2, CaO, MgO, Fe2O3, K2O and Al2O3) display coherent trends correlated with SiO2 (Fig. 8b–h), but with considerable scatter. The Al2O3 data show a widening in the compositional array as SiO2 decreases. Oxides MgO, Na2O, and P2O5 (not shown) show the most scatter with a slight inflection apparent in some of the Na2O data at 75 wt % SiO2.
|
Pumice-clast and lava samples show significant variations in trace-element abundances (Zr 345–61 ppm; Sr 193–33 ppm; Rb 222–56 ppm; Ba 1180–587 ppm) (Fig. 9). Rb, Ba, and Y display an increase with increasing SiO2 (Fig. 9b, d and e). The slight differences in slope indicate the relative degree of incompatibility of these elements; Rb and Ba are more incompatible than Y. Strontium and Zr both decrease with increasing SiO2 and are compatible; however, Zr shows more scatter than Sr. All samples are light REE (LREE) enriched with relatively flat chondrite-normalized middle REE (MREE) and heavy REE (HREE) patterns (Fig. 10).
|
|
The TVZ rhyolites exhibit a narrow range of Nd–Sr isotope compositions (87Sr/86Sr 0·705130–0·706271 and 143Nd/144Nd 0·512744–0·512563) based on both this and previous studies (Fig. 11). Modern mafic samples are plotted for comparison. The overlap of the rhyolite field with some of the andesite compositions should be noted.
|
|
DATA TREATMENT: POLYTOPIC VECTOR ANALYSIS |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOCHEMICAL DATASETDATA TREATMENT: POLYTOPIC VECTOR...RHYOLITE PETROGENESISMAJOR- AND TRACE-ELEMENT...CONCLUSIONSUPPLEMENTARY DATAREFERENCES |
|---|
To understand the petrogenesis of the modern TVZ (600 ka to the present) rhyodacitic–rhyolitic volcanism, we used a subset of the larger database, consisting of more than 400 bulk-rock analyses from both the published literature and this study. This subset was selected to represent the range of rhyolite compositions that have a complete suite of trace-element data available, allowing for a more robust statistical evaluation of the data. The mafic series of samples selected were of hornblende-bearing magmatic rocks, erupted following the caldera collapse associated with the Matahina eruption within the central TVZ where the rhyolitic volcanism occurs. We chose these clasts instead of modern mafic samples, erupted outside the central TVZ, as they represent intermediate compositions more likely to be genetically related to the rhyolites.
Unravelling the complex processes that influence the composition of a melt or magma during its rise from source to subvolcanic reservoir, just prior to eruption, can be difficult as many of these processes can operate simultaneously. This has the effect of overprinting the final eruptive products with a number of geochemical ‘signatures’. Recently Vogel et al. (2008) have reviewed the use of polytopic vector analysis (PVA) in separating the effects of magma mixing and fractional crystallization. PVA is a type of principal component procedure that can be used to evaluate mixing or unmixing (fractional crystallization) in geological systems using all of the major- and trace-element data simultaneously. It differs from other principal component techniques in that it not only determines the number of end-members and their compositions, but also partitions the relative proportions of the end-members into each sample. The proportions of end-members can be used as input in other multivariate analyses to evaluate magmatic processes. PVA is used to determine three essential parameters in a mixing system: (1) the minimum number of end-members; (2) the geochemical signature of each end-member; (3) the relative proportions of each end-member in each sample. In PVA, a very useful parameter is the coefficient of determination, which evaluates the predicted concentration versus the actual concentration of an analyte (element) in a sample. This allows the researcher to evaluate and eliminate samples that may have poor analyses or are altered. Data screening involves removal of extreme outlier samples that have experienced secondary alteration or sample contamination, may have element abundances below the detection limit, or are simply samples that are geochemically distinct from the bulk of the dataset. Although these samples can be important in evaluating processes in a single dataset, the purpose of this study is a broad characterization, and extreme outliers were noted, but removed from consideration. A more complete review of the development of PVA has been given by Johnson et al. (2002).
Application of the polytopic vector analysis
Twenty-four major and trace elements obtained by XRF and ICP-MS for >400 basaltic–rhyolitic samples were evaluated using the coefficient of determination (CD) to screen the sample set to determine number of sources or end-members (EM) that are sufficient to explain the variation in the samples. Once the minimum number of end-members was defined, the PVA was applied to characterize the sources and establish the relative proportions of these end-members in each sample. It is important to note that defining the relative proportions of end- members in each sample is just another way of looking at the chemical composition of the samples, except in this manner all of the chemical components in the samples are captured in the proportion of the end-members in each sample.
Results of the PVA are given in Tables 3 and 4 and Electronic Appendix 3 (end-member proportions). Table 3 provides the correlation coefficient (CD) for each element relative to the number of end-members in a solution. As a general rule, CD values >0·5 support a particular end-member solution, >0·7 strong support, and >0·9 very strong support. Correlation coefficients for a three end-member solution provide support to very strong support for nearly all the elements. We observe a wide variation in Na2O in this study and although a four end-member solution would improve the correlation in the dataset, we accept this variability, recognizing that the Na2O content does not have an impact on the overall correlation among end-members. Similarly, the low Zr CD results from a high variability in the amount of accessory zircon in samples and may reflect non-representative sampling. Therefore, a three end-member solution is preferred, as the four end-member solution does not provide a significant improvement in CD for most analytes.
|
|
Polytopic vector analysis characterization of hornblende-bearing mafic rocks and TVZ rhyolites
The compositions and proportions of the three end-members (EM1 and EM2 are rhyolitic and EM3 is basaltic) are given in Table 4 and Electronic Appendix 3, respectively. A summary of the rhyolite end-member geochemical and petrological characteristics are provided in Table 5 and representative eruptions along the geochemical spectrum in Table 6. Hornblende-bearing mafic samples are defined by EM3, which is similar in composition to a high-aluminium basalt with respect to most trace elements, and also similar to the modern mafic samples (Fig. 4) (with the exception of Zr). In plots of the proportions of EM3 vs EM1 and EM2 (Fig. 12a and b), the linear array of hornblende-bearing andesitic–dacitic samples clearly projects towards the EM1 rhyolite type. (Fig. 12b). This linear array defines an actual mixing relationship between the Matahina rhyolite (Table 1) samples and the hornblende-bearing andesite samples to form the dacitic compositions from the post-caldera collapse deposit described above.
|
|
|
Plots of the proportions of EM1 versus the proportions of EM2 in each sample define a continuum of rhyolite compositions ranging between the two end-member types (Fig. 12a), which can be interpreted in one of two ways: (1) the compositional variation reflects a systematic change in the source composition; this can be accomplished by systematically changing the source conditions (P–T–fH2O, fO2) or source composition, which will have the effect of changing the composition of any derivative melts; or (2) a rhyolite–rhyolite magma mixing relationship that results from upper-crustal, vertical or lateral magma migration and mixing. Although both models would require some variation in the source composition, the latter would allow for very distinct end-member type magmas (i.e. EM1 and EM2) to be produced under distinctly different conditions, which are subsequently mixed in the upper crust. In contrast, the former would call for more systematic changes in the lower crustal source zone, prior to melt extraction, which produces the compositional continuum.
|
RHYOLITE PETROGENESIS |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOCHEMICAL DATASETDATA TREATMENT: POLYTOPIC VECTOR...RHYOLITE PETROGENESISMAJOR- AND TRACE-ELEMENT...CONCLUSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Upper crustal mixing of rhyolite genetically related to a primary mantle-derived magma and a metasedimentary rhyolite melt
Previous work on rhyolite petrogenesis by Blattner & Reid (1982
) and McCulloch et al. (1994) focused on Pb–Sr–Nd–O isotope systematics. Those studies concluded that crustal melting of one or both of the two dominant basement lithologies does not produce rhyolites with suitable isotopic compositions. McCulloch et al. (1994) provided evidence to support a model for the generation of the TVZ rhyolites by magmatic differentiation following crustal contamination of a mantle-derived basalt. They demonstrated 15–25% contamination of the parental basalt by Torlesse-type basement rock followed by 70% crystal fractionation with a minimal amount of further contamination. Graham et al. (1995) summarized the interpretations and data from their and other studies, pointing out that the range of rhyolite and potential assimilants was not well represented. Despite these uncertainties, they concluded that either combined assimilation plus fractional crystallization (AFC) of a mantle-derived high-alumina basalt (Graham et al., 1992) or binary mixing followed by prolonged crystal fractionation (McCulloch et al., 1994) was consistent with the available data. Although the new Sr–Nd isotope data presented in this study are within the range for rhyolites from previous studies (Fig. 11), we tested the hypothesis that one of the end-member rhyolite types (EM1, EM2) might be a rhyodacitic–rhyolitic metasedimentary melt that mixed with the rhyolite melt, derived by crystal–liquid differentiation from a parental mantle-derived basalt, to form the mixing continuum of rhyolite compositions. The proportion of EM2 in each of the samples versus 87Sr/86Sr is shown in Fig. 13. EM2 does not correlate with 87Sr/86Sr, precluding simple binary mixing between a crustal melt and a rhyolitic melt genetically related to a primary, mantle-derived basalt. Therefore, we interpret the isotopic variation to be a consequence of variable degrees of crustal assimilation and the type of assimilant. Such AFC processes might occur either in the deep crustal source region or as the rhyolite magma resides in the mid- to upper crust prior to eruption.
|
Rhyolite genetically related to primary, mantle-derived basalt: P–T–fO2–fH2O, phase assemblages, and mafic compositions
The ASI provides an important measure of the relationship between total Al and Ca + Na + K. Because the conditions (fH2O, fO2, P, T) and composition of the source have a significant influence on the ASI of derivative melts, we have evaluated the ASI for the TVZ dataset. Early workers (e.g. Conrad et al., 1988
; Nicholls et al., 1992), on the basis of a relatively small number of samples representing the more recent deposits (22 ka to present), reported that the dominant composition of the TVZ rhyolites is metaluminous (ASI < 1·0). In contrast, the more extensive dataset compiled in this study of over 1500 rhyolite samples representing all of the major eruptions from the TVZ during the past 600 kyr is dominated by peraluminous (ASI > 1·0) compositions (Fig. 5), with only a subordinate population of metaluminous samples.
Experimental studies have provided important constraints on the petrogenesis of evolved peraluminous melts (Kushiro & Yoder, 1972; Conrad et al., 1988; Beard & Lofgren, 1991; Sisson & Grove, 1993a; Müntener et al., 2001; Prouteau & Scaillet, 2003; Sisson et al., 2005), particularly on the source composition and the conditions necessary for the formation of peraluminous magmas. In general, peraluminous melts can be produced in one or a combination of three ways: (1) melting of pelitic crust; (2) hornblende fractionation at low pressure; (3) partial melting or fractional crystallization under H2O near-saturated or saturated conditions.
As has been shown previously (Graham et al., 1995), melting of pelitic crust is unsuitable for producing the TVZ rhyolites on the basis of Sr–Nd–Pb–O isotopic evidence. Fractionation of hornblende will produce only mildly peraluminous melts, which would not account for the range in ASI observed in the TVZ rhyolites. Therefore, we consider the critical parameter to be H2O, involving either partial melting or fractional crystallization under H2O near-saturated or saturated conditions.
Müntener et al. (2001) concluded from experiments that high H2O contents in calc-alkaline melts are a prerequisite for producing peraluminous derivative magmas. They also demonstrated that even small variations in the H2O content of the melt could have significant effects on the crystallization sequence, suppressing, for example, the crystallization of plagioclase and thus resulting in peraluminous liquids. A necessary consequence of these experiments is that the differentiating magma would leave a hornblende-bearing or pyroxenite ultramafic plutonic restite, which is commonly found in subduction zone settings.
Other experimental studies (Beard & Lofgren, 1991; Rapp et al., 1991; Wolf & Wyllie, 1994; Rapp, 1995; Sisson et al., 2005) have also shown that a wide range of rhyolitic, peraluminous melt compositions can be produced by partial melting of a gabbroic (amphibolite) protolith under varying P–T–fO2–fH2O conditions. In general, these experiments highlight the importance of H2O saturation and K/Na in the starting material; these variables have a significant impact on the ASI and K/Na of the derivative melts, respectively. Figure 14 compares the experimental results of Beard & Lofgren (1991) and Sisson et al. (2005) with the composition of the TVZ rhyolites. Noteworthy features of the figure are the ASI range of the dacitic–rhyolitic melts produced by melting hornblende-bearing protoliths of varying compositions and the agreement with the observed ASI range of the TVZ rhyolites.
|
The Beard & Lofgren (1991
) experiments performed under H2O-saturated conditions produce the most peraluminous melts. The residues in equilibrium with the H2O-saturated melts always contained hornblende and clinopyroxene in varying abundances at pressures >3·0 kbar. In contrast, H2O under-saturated experiments only contain hornblende at pressures >3·0 kbar and temperatures <900°C. Water-saturated runs also display an overall lower modal abundance of plagioclase when compared with the dehydration runs, reflecting the suppression of plagioclase crystallization at high water contents (Fig. 15a). Furthermore, variations in temperature have the effect of changing the modal hornblende/clinopyroxene ratio in the residue (Fig. 15b and c). Sisson et al. (2005) controlled fO2, using different oxygen buffers, in their dehydration melting experiments. A correlation between fO2 and the Mg-number of the derivative melt (glass) demonstrates the effects of fO2 on melt composition (Fig. 15d). In summary, variations in the compositions of the derivative melts and the equilibrium phases in these experiments are affected by T, fO2, and fH2O: (1) changes in T affect the modal proportions of the residual phases; (2) changes in fO2 affect the Mg-number of the derivative melts; (3) changes in fH2O produce variability in the ASI of the derivative melts (Fig. 14).
|
NNO (nickel–nickel oxide buffer)]; different symbols represent different starting materials.Lower crustal magma source zone
Geophysical considerations
Experimental studies, discussed above, provide important constraints for our petrogenetic model, indicating that lower crustal P–T–H2O conditions can explain the observed geochemical variations seen in the rhyolites. This is important for two key reasons: (1) crystal fractionation at the base of the crust alleviates the spatial problem of accommodating a large amount of residual material from fractional crystallization processes in the mid- to upper crust, for which there is no evidence; (2) higher temperatures in the lower crust provide sufficient heat for crustal assimilation (Jackson et al., 2003
; Dufek & Bergantz, 2005; Annen et al., 2006).
Harrison & White (2004) presented a crustal cross-section of the TVZ, interpreted from wide-angle seismic experiments, that defines the lower 15–30 km as a zone of heavily intruded mafic crust with 
2% partial melt. Other studies, by Ogawa et al. (1999) and Bannister et al. (2004), interpreted low S-wave velocities in terms of a highly conductive zone between 15 and 20 km depth in the central TVZ. Numerical modelling of rhyolitic melt generation and segregation by Jackson et al. (2003) concluded that repeated injection of basaltic sills into the lower crust would facilitate the development of an accumulation and magma ascent zone. Subsequently, similar models involving a lower crustal ‘hot-zone’ that controls the composition of the evolved melts (Dufek & Bergantz, 2005; Annen et al., 2006) were developed. We suggest that these numerical models of intermediate to evolved melt production in lower crustal ‘hot zones’ are consistent with geophysical observations of the TVZ and that basaltic intrusions at between 20 and 30 km depth may be capped by a mushy accumulation and magma ascent zone between 15 and 20 km depth. Jackson et al. (2003) also demonstrated that significant volumes of rhyolitic melt could be produced within geologically short timescales (104–106 years), which is a requisite in the TVZ, where the rate of rhyolite melt production is very high (0·1–0·2 m3/s; Wilson, 1993).
Petrological observations
Determining the water content of subduction-related magmas has been the focus of many studies (e.g. Sisson & Layne, 1993; Blatter & Carmichael, 1998, 2001; Moore & Carmichael, 1998; Benjamin et al., 2007). Although the importance of the role of water in subduction zone environments is well established, few high-H2O content (as indicated by the presence of hornblende) mafic eruptions have been reported worldwide (e.g. Cerro La Pilita, Western Mexico, Luhr & Carmichael, 1985; Mount Lamington, Papua New Guinea, Arculus et al., 1983. In a comprehensive study of the Izu–Bonin arc, Tamura & Tatsumi (2002) concluded that the negative slope of the water-saturated liquidi for calc-alkaline melts could have caused these ‘wet’ mafic magmas to stall in the mid-crust. This arresting of ‘wet’ magmas is consistent with the observations that hornblende-bearing gabbros are commonly found as part of exhumed suprasubduction-zone lithologies in the lower crust (Blundy & Sparks, 1992; Sisson et al., 1996). Barclay & Carmichael (2004) concluded, based on water-saturated fractionation experiments on hornblende-bearing basalts from Western Mexico, that the ascent of hydrous basaltic magmas from the base of the crust is retarded by their high crystallinity, resulting in stalling and subsequent ‘stockpiling’ of these magmas.
The TVZ has erupted mafic magmas over the past 600 kyr; however, the current focus of mafic volcanism occurs outside the central TVZ. Basalts represent <1% of the total erupted volume within the central TVZ, and even fewer examples of hornblende-bearing mafic samples have been documented (Brown, 1994; Leonard et al., 2002; Wilson et al., 2006; this study). Interestingly, these are all co-eruptive pumice clasts associated with major caldera-forming rhyolitic eruptions. Hornblende-bearing gabbro or diorite has also been reported as exhumed xenoliths from several eruptions within the TVZ (Brown, 1994; Beresford, 1997; Burt et al., 1998). The depletion in MREE relative to chondrite observed in all of the TVZ rhyolites (Fig. 16) strongly supports the presence of hornblende in the source mineral assemblage and suggests a potential genetic relationship between these hornblende-bearing mafic magmas and the rhyolites.
|
Mineral chemistry
The chemical compositions of hornblende and plagioclase from the Matahina post-caldera collapse deposits (this study) and the Oruanui calc-alkaline mafic pumice clasts (Wilson et al., 2006
) are unique within the TVZ. Some of the samples have high-Al2O3 (>9·5 wt %) and Mg-number (>65) hornblende, and high-An plagioclase (>An85) (Fig. 7).
Al in hornblende has been used by some as a geobarometer under SiO2-saturated conditions (e.g. Johnson & Rutherford, 1989). Although we are not seeking a quantitative measure of the pressure for depth estimates, a semi-quantitative estimate can be obtained by comparing the hornblende from the crystal-rich, mafic samples of the Matahina post-caldera collapse deposits with those from the rhyolitic ignimbrites.
Experimental studies that have focused on the Al content of hornblende as a geobarometer (Johnson & Rutherford, 1989; Thomas & Ernst, 1990; Schmidt, 1992; Ernst & Liu, 1998) have demonstrated that increasing pressure favours a coupled Al-Tschermak substitution (TSi + M1–M3Mg = TAl + M1–M3Al). In contrast, temperature variations do not favour the Al-Tschermak substitution (Spear, 1981). Instead, temperature favours the edenite exchange [TSi + A
= TAl + A(Na + K)] (Spear, 1981; Blundy & Holland, 1990) and also correlates with Ti content. Figure 17a shows the positive correlation between Ti content and TAl in the Matahina rhyolite hornblendes and in associated andesitic samples. This suggests that some of the hornblende variation is the result of changes in temperature. However, the plot of TAl vs M1–M3Al also shows a positive correlation, suggesting that the variation in Altotal is also governed by changes in pressure (Fig. 17b). Calculations using the methods of Schmidt (1992) yield pressures for hornblende cores consistently greater than 4·0 kbar, which are consistent with the semi-quantitative study of calcic-amphibole in mid-ocean ridge basalt (MORB) by Ernst & Liu (1998). High Al2O3 (>9·5 wt % Al2O3; >1·75 Altotal a.p.f.u.) in hornblende has also been reported in numerous crystallization–melting experiments under variable T–fO2–fH2O conditions (Conrad et al., 1988; Beard & Lofgren, 1991; Prouteau & Scaillet, 2003; Sisson et al., 2005), consistent with pressures greater than 4·0 kbar. Hence, we estimate that these TVZ magmas crystallized at or close to 15–20 km depth, which corresponds to the location of our proposed accumulation and magma ascent zone.
|
Geochemical evaluation of basalt and andesite as parental magmas to the rhyolites
Price et al. (2005
) evaluated the potential genetic relationship between the modern TVZ andesites and the rhyolites. They demonstrated that the andesites have Sr–Nd isotope compositions that extend into the rhyolite field as shown in Fig. 11 and that rhyolitic melt inclusions in some were similar to TVZ rhyolite compositions. This led them to conclude that a ‘pre-conditioning’ of the lower crust with andesite was a prerequisite for rhyolite production.
Modern TVZ basalts and andesites show evidence of crustal contamination, which can be evaluated based on incompatible element behaviour. If mafic magmas are contaminated in the lower crust via binary mixing, followed by equilibrium crystallization and batch melt removal prior to eruption, we expect incompatible element abundances (K2O, Rb) to define a mixing hyperbola. Figure 18a shows a simple binary mixing model between a least-evolved basaltic composition and an average metasedimentary crust composition (from Reid, 1983). The K2O/Rb ratios of the mafic samples decrease significantly with small amounts of change in Rb, indicating that the first 20% of assimilation, accompanied by only a negligible amount of crystal fractionation, can explain the data array. Compositions that fall off of the mixing hyperbola, dominantly andesite, may be derived from a combination of assimilation and fractional crystallization (AFC) (see Graham et al., 1995).
|
The hornblende-bearing andesite clasts investigated in this study have similar K2O/Rb ratios, plotting close to the binary mixing line (Fig. 18a), and isotopic compositions to the rhyolites, suggestive of open-system assimilation with a limited degree of crystal fractionation (Fig. 18a and b). The nearly flat K2O/Rb data array of the hornblende-bearing andesite to rhyolite series of the TVZ is consistent with near closed-system crystallization as the dominant process producing the rhyolites from a contaminated andesite source magma (Fig. 18a). Trace-element (La, Y, Yb) variations (not shown) with increasing silica content are also consistent with lower–middle crustal differentiation of hornblende-bearing mafic magmas, rather than amphibolite melting, in accordance with a REE modelling study by Brophy (2008
). Therefore, we prefer a fractional crystallization, as opposed to a partial melting, petrogenetic model for the rhyodacite–rhyolite melts.
Annen et al. (2006) calculated the fraction of crustal (e.g. pelite, greywacke) assimilation by basaltic magmas at lower crustal temperatures. Their numerical models show that the temperature gradient above the uppermost basaltic sill emplacement decreases rapidly, and, as a consequence, the amount of assimilation decreases substantially. This suggests that very little assimilation will occur once the evolved melt has been removed. The small range in rhyolite isotope compositions and their overlap with the modern andesite field (Fig. 18b) supports the interpretation that the isotopic overlap is governed by the thermal limitations on crustal melting.
Bachmann & Bergantz (2004) investigated the chemical and physical constraints on the generation of rhyodacitic–rhyolitic melts by expulsion from an intermediate composition crystal mush. Their model suggests that once the intermediate magma reaches 45–50 vol. % crystals chaotic convection ceases, which allows the interstitial rhyodactic–rhyolitic melt to be extracted along a horizon above the solidifying base. The results of these numerical models suggest that thermal, chemical and physical conditions require (1) the bulk of the wall-rock assimilation to occur in the lower crust, and (2) the rhyodacitic–rhyolitic melt to remain in mechanical and chemical equilibrium with the crystallizing andesite prior to its extraction.
Consequently, we propose a two-stage process: (1) open-system contamination of intruded basalt by the surrounding metasedimenatary crust with concomitant fractional crystallization and andesitic melt extraction; (2) near closed-system (with respect to metasedimentary crustal assimilation, but open to primary andesite recharge) equilibrium crystallization and batch rhyodacitic–rhyolitic melt extraction.
|
MAJOR- AND TRACE-ELEMENT MODELLING |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOCHEMICAL DATASETDATA TREATMENT: POLYTOPIC VECTOR...RHYOLITE PETROGENESISMAJOR- AND TRACE-ELEMENT...CONCLUSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Major elements
The distinct characteristics of the two end-member rhyolite types provide a guide for geochemical modelling of their derivation from a relatively primitive mafic parent magma. EM1 rhyolite is defined by a high ASI, low FeO*/MgO, and an abundance of hydrous ferromagnesian phases. In contrast, EM2 is defined by a low ASI, high FeO*/MgO and an abundance of anhydrous ferromagnesian phases. The experimental analogs discussed above provide important constraints on the modal mineral assemblages that are likely to be in equilibrium with melts of these two end-member types. Based on these constraints, we chose to model crystal fractionation using a mineral assemblage of plagioclase + hornblende + orthopyroxene + magnetite + ilmenite ± clinopyroxene. This mineral assemblage is consistent with the residual phases in the experiments at appropriate pressures (3·0–7·0 kbar), considered reasonable estimates for the lower crust. The least-evolved mineral compositions that occur in the selected parental rock type (UC1100) (hornblende + orthopyroxene + clinopyroxene + magnetite + ilmenite) were used for the modelling (Electronic Appendix 4). These compositions were fixed for major-element modelling runs. As the composition of plagioclase can vary greatly during a given melt evolution, we chose to use generic plagioclase end-member compositions (An98 and An2; Deer et al., 1992
), which allowed the regression calculation to find a ‘best-fit’ composition. Samples used in modelling were chosen to represent eruptive products most similar to the two rhyolitic end-member types along the compositional continuum. The least-evolved calc-alkaline, hornblende-bearing andesite (UC1100) that was used as the source rock meets all of our criteria for a potential parent: (1) its K2O/Rb ratio is similar to that of the TVZ rhyolites; (2) its Sr–Nd isotopic composition is similar to that of the TVZ rhyolites and to that of the hornblende-bearing andesitic rocks. In addition, this rock has a crystallinity, estimated to be c. 50% based on modal analysis, a peraluminous rhyodacitic–rhyolitic glass composition, and was erupted with an EM1 rhyolite magma type.
Equilibrium crystallization of the primitive andesite to produce the representative EM1 and EM2 rhyolites was evaluated using multiple linear regression analysis of major-element data. An important feature of this modelling is that it describes equilibrium crystallization conditions, where the melt remains in contact with the crystals until conditions (e.g. buoyancy, deformation) are appropriate to allow the melt to separate. Rhyolite magma compositions could be generated by either crystallization or melting processes; the results would be identical. The sum of the squares (
r2) of the residuals provides a measure of the similarity between the measured and theoretical chemical parameters tested. In modelling equilibrium crystallization directly from an andesite parent to representative rhyolite types, the r2 residual was set to <0·1 as a cut-off.
Results from the model calculations are presented in Table 7. These demonstrate that both EM1 and EM2 could be produced by 50–60% liquid–crystal differentiation from a crustally contaminated andesite parent magma by hornblende-dominant fractionation and orthopyroxene–clinopyroxene-dominant fractionation, respectively. Although we first modelled equilibrium fractionation for the EM2 magma type using a hornblende–orthopyroxene assemblage, the r2 residual was greater than our cut-off so we tried modelling with an orthopyroxene–clinopyroxene assemblage. The respective assemblages modelled produced liquid compositions consistent with the TVZ rhyolite end-member types. For simplification, apatite was not used in these calculations so that P2O5 disparity was expected. K2O has the highest residual, as a result of an apparently low parental K2O magma content (1 wt %). We tested this using an artificially higher K2O (1·25 wt %) content in the parent and were able to improve the results for the samples analyzed. This suggests that the actiual parental magma had a K2O content slightly greater than the parent melt composition used here for the rhyolite modelling. However, modern andesites and those found within the Matahina post-caldera collapse deposits have similar K2O contents to those that we modelled.
|
Several important observations can be made based on the modal mineral assemblages predicted by the regression analyses. First, the hornblende/orthopyroxene ratio predicted for the EM1 type rhyolite is similar to that observed in the water-saturated, low-temperature amphibolite melting experiments of Beard & Lofgren (1991
) and in the actual sample (UC1100) chosen to represent the parent magma. The variations in the hornblende/orthopyroxene ratio are consistent with changes in the experimental phase proportions discussed above, which are controlled by the T–P–fH2O conditions. Second, the EM2 rhyolite type model assemblage is clinopyroxene–orthopyroxene dominant. Only amphibolite dehydration experiments at high temperatures (>925°C) produce residual melts with anhydrous phase assemblages, suggesting that the EM2 melts were produced under similar conditions. Third, the lower modal proportions of plagioclase in the predicted EM1 assemblage are suggestive of high H2O pressures. This is consistent with the abundant hydrous phases and higher ASI characterizing EM1.
Trace elements
The trace-element concentrations of melts are controlled by the solid-phase assemblage in equilibrium with the melt and the original concentrations of trace elements in their source, as well as other intensive parameters (P, T, fO2, fH2O). Therefore, we chose to apply trace-element modelling to test the validity of our major-element fractionation model. Our objective was to determine if the modal mineral assemblages predicted by the major-element regression analyses could produce the distinct concentrations of key trace elements (Rb, Sr, Ba, Y, REE) in the EM1 and EM2 rhyolites. An average of the trace-element concentrations in several of the mafic compositions was used to represent the trace-element abundances in the potential parental magma. In these calculations we used the predicted modal mineral assemblages obtained from the major-element regression modelling.
Trace-element concentrations in the melt were calculated for equilibrium crystallization using published partition coefficients (GERM online database and Rollinson, 1993) reported in Electronic Appendix 5. Equilibrium crystallization was preferred rather than Rayleigh fractionation because, as discussed above, the major-element modelling requires the residue to remain in equilibrium with the melt until the melt is extracted. Low-silica rhyolite or rhyolitic liquid partition coefficients (Kd) were preferred over high-silica rhyolite to be consistent with the andesite–rhyolite major-element modelling. In addition, experimental or calculated Kd values, which account for T, P, and/or composition dependence, were preferred over those determined from phenocryst–matrix partitioning.
Rb–Sr–Zr–Ba–Y
The predicted concentrations of Rb do not vary significantly between the two end-member rhyolite compositions, but are slightly lower than in the actual rhyolites (Table 7). These values are at the lower end of the spectrum of compositions observed in the TVZ rhyolites (Rb 56–222 ppm); however, few low-silica rhyolites (SiO2 < 74·0 wt %) exceed 120 ppm Rb. As Rb is an incompatible element and there is such a wide range of concentrations observed in the natural rhyolites, we suggest that a wide range of Rb source contents is dictating this variation, probably controlled by early crustal assimilation. Modern andesites in the TVZ display a wide range of Rb concentrations (Fig. 4b) at similar SiO2 contents to our andesite parent, which is also consistent with considerable variation in the source. Furthermore, the higher Rb concentrations (>120 ppm) are restricted to high-silica rhyolites and are interpreted to represent upper crustal fractionation–assimilation following removal from the source zone. Predicted Sr concentrations are highly dependent on the modal proportion of plagioclase in the fractionating assemblage. Our modelling closely estimates Sr concentrations in the liquids for both end-member rhyolite types. A comparison of Sr contents between the two end-members shows that EM1 is enriched in Sr relative to EM2. This disparity is consistent with the c. 10% modal variation in plagioclase between the two assemblages modelled. Available partition coefficients for Zr in hornblende in equilibrium with a rhyolite melt vary considerably (0·27–4·0), but are always greater than those from the same studies relative to pyroxene. As a result, we expect that in melts in which hornblende is dominant over pyroxene in the fractionating assemblage the melts will be depleted in Zr relative to those in which pyroxene is the dominant fractionating phase. A significant disparity between the predicted and observed concentrations of Zr in the EM2 model suggests that either the starting composition does not accurately represent that for the rhyolites modelled (similar to Rb a large variation exists in andesite Zr concentrations) or the partition coefficient used is lower than the actual value for this system. Nevertheless, the model appropriately predicts the variation in Zr concentration between EM1 and EM2 rhyolites.
The concentrations of Ba and Y are particularly useful in this modelling as Ba behaves incompatibly whereas Y is highly compatible in hornblende, behaving similarly to the MREE. Figure 19 shows the Ba and Y concentrations predicted by our model, compared with the compositions of the TVZ rhyolites. The model predictions are in agreement with the actual rhyolite data. The separate, but positive sloping, trends observed within the rhyolite datasets may represent similarities in the fractionating mineral assemblage as the magma is removed from the source zone and stored in the upper crust (Fig. 15).
|
); Mangaone (
). Arrows projecting from EM3 (source rock) towards the Ba–Y values calculated from trace-element modelling for each end-member type; either hornblende (hb)-dominant fractionation or pyroxene (px)-dominant fractionation.Rare earth elements
The predicted melt concentrations of the REE are in excellent agreement with the REE patterns of the natural rhyolites (Fig. 20a). Hornblende-dominant fractionation effectively produces a more depleted MREE–HREE trend over orthopyroxene-dominant fractionation, with an enrichment in LREE. Conversely, orthopyroxene-dominant fractionation shows a relative enrichment in MREE and enrichment in LREE. Another distinct difference between the two modelled liquids is the Eu anomaly (Eu/Eu*), which is much more pronounced in the EM2 samples. This depletion in Eu is consistent with the higher modal proportion of plagioclase in equilibrium with this end-member. Figure 20b shows several examples of EM1 and EM2 magma types (EM1: Rotoiti and Lake Okataina pyroclastic deposits; EM2: Mangaone and Mamaku pyroclastic deposits) with REE patterns plotted separately for clarity. The MREE depletion, high Sr, and smaller Eu anomaly of the EM1 rhyolites emphasize the importance of the coupled nature of hornblende crystallization under high-H2O conditions, which consequently suppresses plagioclase crystallization.
|
).Petrological evidence for restites
The petrogenetic model presented here predicts the occurrence of both hornblende-dominant and pyroxene-dominant gabbroic restites in the crust. Over the modern volcanic history of the TVZ, both of these types of xenolith have been reported in lithic lag deposits and within host pyroclastic deposits. Hornblende-rich gabbros have been described from the Kaingaroa ignimbrite (Beresford, 1997
), and Whakamaru ignimbrite (Brown, 1994) and pyroxene-rich gabbros occur as lithics in the more recent Kaharoa pyroclastic deposits (Leonard et al., 2002). Brown (1994) reported strongly zoned An83–36 plagioclase in gabbros containing rhyolitic, peraluminous glass. Plagioclases analyzed in this study have even more An-rich cores (An94). The strongly zoned plagioclases observed by Brown (1994) reflect in situ changes in melt composition as the magma crystallized and the high-An cores require high H2O activity, both consistent with the proposed model. |
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOCHEMICAL DATASETDATA TREATMENT: POLYTOPIC VECTOR...RHYOLITE PETROGENESISMAJOR- AND TRACE-ELEMENT...CONCLUSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Using polytopic vector analysis (PVA) of our extensive TVZ dataset we have been able to identify a continuum of rhyolite compositions between two distinct end- member types. Variations in the major- and trace-element compositions of the rhyolite end-members (EM1 and EM2) are best explained by differences in lower crustal conditions (P–T–fO2–fH2O) that dictate the nature of the fractionating mineral assemblage from the parental magma and, as a result, produce a continuum of evolved melt compositions between the two contrasting rhyolite end-members. Saturated to near-saturated H2O conditions and/or hornblende in the source are important in producing the observed range of metaluminous to peraluminous rhyolites. In experiments under water-saturated conditions hornblende is always present at pressures
3 kbar, whereas in dehydration experiments it is present only in trace amounts (Beard & Lofgren, 1991). In addition, the modal assemblage of orthopyroxene–clinopyroxene–hornblende changes with variations in fO2, fH2O, P and T. Specifically, increasing the temperature increases the modal abundance of orthopyroxene–clinopyroxene, and in dehydration experiments at T > 925°C hornblende is completely absent. In contrast, hornblende is always present in 7·0 kbar (Sisson et al., 2005) dehydration experiments, regardless of variations in T and fO2, thereby emphasizing the importance of the starting composition in determining the phases in equilibrium with the melt. These experiments highlight the importance of source conditions in producing distinct melt compositions.
The distinct geochemical and petrological characteristics of the TVZ rhyolites can be summarized as follows in the context of a continuum between two end-members: EM1 magma types are produced under low P–T and high fO2–fH2O conditions, promoting the fractionation of hornblende and suppressing that of plagioclase, resulting in: (1) the depletion of MREE and Y; (2) an increase in Sr; (3) a decrease in Zr; (4) less pronounced depletion of Eu; (5) oxidizing conditions, which reduce the compatibility of Fe and produce liquids with lower FeO*/MgO (calc-alkaline); (6) abundant crystallization of hydrous phases; (7) moderately peraluminous melts. In contrast, EM2 magma types are produced under higher P–T and/or lower fO2–fH2O, or more reducing conditions, promoting the fractionation of clinopyroxene–orthopyroxene over hornblende, resulting in: (1) an enrichment of MREE and Y; (2) a decrease in Sr; (3) an increase in Zr; (4) large Eu anomaly; (5) reducing conditions producing high FeO*/MgO (tholeiitic) ratios; (6) low crystallinity, dominated by anhydrous phases; (7) slightly to moderately peraluminous melts. Subtle changes in the source P–T conditions (
1·0–2·0 kbar, 50–100°C) can change the modal proportions in equilibrium with the evolved melt. Consequently, subtle changes in melt composition will be produced. More dramatic changes in the source P–T or fH2O–fO2 conditions (i.e. exhausting the H2O) produce a distinctly different melt composition closer to the extreme EM2.
We have shown that the diversity of magma compositions found within the TVZ is primarily governed by differences in the source conditions and variations in the composition or isotopic character of the surrounding metasedimentary crust that is assimilated into the crystallizing basalt. The overall amount of crustal contamination is dictated by the amount of heat input into the crust by intruded basaltic magma. As the temperature and composition of the mantle are not likely to have changed over this short time frame, a near-constant assimilation rate is sufficient to explain the observed isotopic shifts. To satisfy the thermal, physical, and geochemical constraints identified in this study, we suggest that a lower crustal ‘hot-zone’ is continuously rejuvenated by successive basalt emplacement.
Our model is similar to those conceptualized by Charlier et al. (2005) and Smith et al. (2005) for TVZ rhyolite production; however, we emphasize here the importance of the lower crustal environment in governing the compositions of the rhyodacite–rhyolites. We propose a scenario, illustrated in Fig. 21, in which primitive basaltic magmas produced in the mantle wedge intrude into the lower metasedimentary crust as a series of sills (20–30 km depth) (Jackson et al., 2003; Dufek & Bergantz, 2005; Annen et al., 2006). These hot, water-rich basaltic intrusions assimilate metasedimentary basement lithologies (probably greywacke), then undergo rapid crystallization. Batches of andesitic melt are removed and coalesce in the mid-crust (15–20 km depth), forming an accumulation and ascent zone. Initially, in the andesitic melts, hornblende dominates the fractionating assemblage under water-saturated, high-fO2 conditions, and, consequently, the melt composition is driven towards a peraluminous EM1 rhyolite type [low FeO*/MgO (calc-alkaline), depleted MREE and Y, abundant hydrous phases]. Once crystallinity reaches 45–50 vol. %, in the fractionating andesite, batches of evolved rhyolitic melt ascend and are stored in the mid- to upper crust prior to eruption. However, if melt production continues within the mantle wedge, a progressive inverted stacking of the basaltic sills can continue in the source zone. This will have the effect of raising the temperature in the source zone, changing the ratio of hydrous to anhydrous phases in the intruding basaltic sills, as well as in the intermediate melt in the accumulation and ascent zone. Consequently, the composition of the evolved melts will shift towards the EM2 rhyolite type [high FeO*/MgO (tholeiitic), non-depleted MREE and Y, anhydrous phases]. If the H2O flux from the mantle wedge decreases, the change will become more dramatic until crystallization conditions are anhydrous in the source zone. One potential consequence of the proposed model is that there could be a temporal geochemical progression as successive basalts are emplaced and/or the flux of H2O changes over time.
|
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOCHEMICAL DATASETDATA TREATMENT: POLYTOPIC VECTOR...RHYOLITE PETROGENESISMAJOR- AND TRACE-ELEMENT...CONCLUSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
ACKNOWLEDGEMENTS |
|---|
We would like to thank the Marsden Fund administered by the Royal Society of New Zealand (UOC0508) and the Department of Geological Sciences, University of Canterbury, Mason Trust Fund for financial support. Electron microprobe analysis at the University of Michigan was supported by NSF grant EAR-9911352. The thorough and constructive reviews from Ian Graham, Kurt Knesel, and an anonymous reviewer are greatly appreciated.
*Corresponding author. Telephone: (+64) 3 364 2987 ext. 7713. Fax: (+64) 3 364 2769. E-mail: cdd21{at}student.canterbury.ac.nz
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONANALYTICAL TECHNIQUESGEOCHEMICAL DATASETDATA TREATMENT: POLYTOPIC VECTOR...RHYOLITE PETROGENESISMAJOR- AND TRACE-ELEMENT...CONCLUSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Annen C, Blundy JD, Sparks RSJ. The genesis of intermediate and silicic magmas in deep crustal hot zones. Journal of Petrology (2006) 47:505–539.
Arculus RJ, Johnson RW, Chappell BW, McKee CO, Sakai H. Ophiolite-contaminated andesites, trachy-basalts and cognate inclusions of Mount Lamington, Papua New Guinea: anhydrite–amphibole-bearing lavas and the 1951 cumulodome. Journal of Volcanology and Geothermal Research (1983) 18:215–258.[CrossRef][Web of Science]
Bachmann O, Bergantz GW. On the origin of crystal-poor rhyolites: extracted from batholithic crystal mushes. Journal of Petrology (2004) 45:1565–1582.
Bannister S, Bryan CJ, Bibby HM. Shear wave velocity variation across the Taupo Volcanic Zone, New Zealand, from receiver function inversion. Geophysical Journal International (2004) 159:291–310.[CrossRef][Web of Science]
Barclay J, Carmichael ISE. A hornblende basalt from western Mexico: water-saturated phase relations constrain a pressure–temperature window of eruptibility. Journal of Petrology (2004) 45:485–506.
Beard JS, Lofgren GE. Dehydration Melting and Water-Saturated Melting of Basaltic and Andesitic Greenstones and Amphibolites at 1, 3, and 6·9 kb. Journal of Petrology (1991) 32:365–401.
Benjamin ER, Plank T, Wade JA, Kelley KA, Hauri EH, Alvarado GE. High water contents in basaltic magmas from Irazú Volcano, Costa Rica. Journal of Volcanology and Geothermal Research (2007) 168:68–92.[Web of Science]
Beresford SW. Volcanology and geochemistry of the Kaingaroa Ignimbrite, Taupo Volcanic Zone, New Zealand. (1997) University of Canterbury. PhD thesis.
Bignall G, Browne PRL, Kyle PR. Geochemical characterisation of hydrothermally altered ignimbrites in active geothermal fields from the central Taupo Volcanic Zone, New Zealand. Journal of Volcanology and Geothermal Research (1996) 73:79–97.[CrossRef][Web of Science]
Blatter DL, Carmichael ISE. Plagioclase-free andesites from Zitacuaro (Michoacan), Mexico: petrology and experimental constraints. Contributions to Mineralogy and Petrology (1998) 132:121–138.[CrossRef][Web of Science]
Blatter DL, Carmichael ISE. Hydrous phase equilibria of a Mexican high-silica andesite: a candidate for mantle origin? Geochimica et Cosmochimica Acta (2001) 65:4043–4065.[CrossRef][Web of Science]
Blattner P, Reid F. The origin of lavas and ignimbrites of the Taupo Volcanic Zone, New Zealand, in the light of oxygen isotope data. Geochimica et Cosmochimica Acta (1982) 46:1417–1430.[CrossRef][Web of Science]
Blundy JD, Holland TJB. Calcic amphibole equilibria and a new amphibole–plagioclase geothermometer. Contributions to Mineralogy and Petrology (1990) 104:208–224.[CrossRef][Web of Science]
Blundy JD, Sparks RSJ. Petrogenesis of mafic inclusions in granotoids of the Adamello Massif, Italy. Journal of Petrology (1992) 33:1039–1104.
Brophy JG. A study of rare earth element (REE)–SiO2 variations in felsic liquids generated by basalt fractionation and amphibolite melting: a potential test for discriminating between the two different processes. In: Contributions to Mineralogy and Petrology (2008) doi:10.1007/s00410-008-0289-x.
Brown SJA. Geology and geochemistry of the Whakamaru Group ignimbrites, and associated rhyolite domes, Taupo Volcanic Zone, New Zealand. (1994) University of Canterbury. PhD thesis.
Brown SJA, Wilson CJN, Cole JW, Wooden JL. The Whakamaru Group ignimbrites, Taupo volcanic zone, New Zealand: evidence for reverse tapping of a zoned silicic magmatic system. Journal of Volcanology and Geothermal Research (1998) 84:1–37.[CrossRef][Web of Science]
Burt RM, Brown SJA, Cole JW, Shelley D, Waight TE. Glass-bearing plutonic fragments from ignimbrites of the Okataina caldera complex, Taupo Volcanic Zone, New Zealand: remnants of a partially molten intrusion associated with preceding eruptions. Journal of Volcanology and Geothermal Research (1998) 84:209–237.[CrossRef][Web of Science]
Charlier BLA, Wilson CJN, Lowenstern JB, Blake S, van Calsteren PW, Davidson JP. Magma generation at a large, hyperactive silicic volcano (Taupo, New Zealand) revealed by U–Th and U–Pb systematics in zircons. Journal of Petrology (2005) 46:3–32.
Cole JW. Structure, petrology and genesis of Cenozoic volcanism, Taupo Volcanic Zone, New Zealand—a review. New Zealand Journal of Geology and Geophysics (1979) 22:631–657.
Conrad WK, Nicholls IA, Wall VJ. Water-saturated and undersaturated melting of metaluminous and peraluminous crustal compositions at 10 kb; evidence for the origin of silicic magmas in the Taupo Volcanic Zone, New Zealand, and other occurrences. Journal of Petrology (1988) 29:765–803.
Deer WA, Howie RA, Zussman J. An Introduction to the Rock-Forming Minerals (1992) Harlow: Longman. 438–439.
Dufek J, Bergantz GW. Lower crustal magma genesis and preservation: a stochastic framework for the evaluation of basalt–crust interaction. Journal of Petrology (2005) 46:2167–2195.
Ernst WG, Liu J. Experimental phase-equilibrium study of Al- and Ti-contents of calcic amphibole in MORB—A semiquantitative thermobarometer. American Mineralogist (1998) 83:952–969.[Abstract]
Ewart A, Stipp JJ. Petrogenesis of the volcanic rocks of the Central North Island, New Zealand, as indicated by a study of 87Sr/86Sr ratios, and Sr, Rb, K, U and Th abundances. Geochimica et Cosmochimica Acta (1968) 32:699–736.[CrossRef][Web of Science]
Gamble JA, Smith IEM, McCulloch MT, Graham IJ, Kokelaar BP. The geochemistry and petrogenesis of basalts from the Taupo Volcanic Zone and Kermadec Island Arc, S. W. Pacific. Journal of Volcanology and Geothermal Research (1993) 54:265–290.[CrossRef][Web of Science]
Graham IJ, Gulson BL, Hedenquist JW, Mizon K. Petrogenesis of late Cenozoic volcanic rocks from the Taupo volcanic zone, New Zealand, in the light of new lead isotope data. Geochimica et Cosmochimica Acta (1992) 56:2797–2819.[CrossRef][Web of Science]
Graham IJ, Cole JW, Briggs RM, Gamble JA, Smith IEM. Petrology and petrogenesis of volcanic rocks from the Taupo Volcanic Zone: a review. Journal of Volcanology and Geothermal Research (1995) 68:59–87.[CrossRef][Web of Science]
Gravley DM. The Ohakuri pyroclastic deposits and the evolution of the Rotorua–Ohakuri volcanotectonic depression. (2004) University of Canterbury. PhD thesis.
Harrison AJ, White RS. Crustal structure of the Taupo Volcanic Zone, New Zealand; stretching and igneous intrusion. Geophysical Research Letters (2004) 31:1–4.
Jackson MD, Cheadle MJ, Atherton MP. Quantitative modelling of granitic melt generation and segregation in the continental crust. Journal of Geophysical Research (2003) 108:1–21.
Johnson GW, Ehrlich R, Full W. Principal components analysis and receptor models in environmental forensics. In: An Introduction to Environmental Forensics—Murphy BL, Morrison RD, eds. (2002) San Diego, CA: Academic Press. 461–515.
Johnson MC, Rutherford MJ. Experimental calibration of the aluminium-in-hornblende geobarometer with application to Long Valley Caldera (California) volcanic rocks. Geology (1989) 17:837–841.
Karhunen RA. The Pokai and Chimp Ignimbrites of NW Taupo Volcanic Zone. (1993) University of Canterbury. PhD thesis.
Kushiro I, Yoder HS Jr. Origin of calc-alkalic peraluminous andesites and dacites. Carnegie Institution of Washington Yearbook (1972) 71:411–413.
Leake BE, Wooley AR, Arps CES, et al. Nomenclature of amphiboles: Report of the Subcommittee on Amphiboles of the International mineralogical Association, Commission on New Minerals and Mineral Names. American Mineralogist (1997) 82:1019–1037.[Abstract]
Leonard GS. The evolution of Maroa Volcanic Centre, Taupo Volcanic Zone, New Zealand. (2003) University of Canterbury. PhD thesis.
Leonard GS, Cole JW, Nairn IA, Self S. Basalt triggering of the c. AD 1305 Kaharoa rhyolite eruption, Tarawera Volcanic Complex, New Zealand. Journal of Volcanology and Geothermal Research (2002) 115:461–486.[CrossRef][Web of Science]
Luhr JF, Carmichael ISE. Jorullo Volcano, Michocoan, Mexico (1759–1774): the earliest stages of fractionation in calc-alkaline magmas. Contributions to Mineralogy and Petrology (1985) 90:142–161.[CrossRef][Web of Science]
Manning DA. Late Pleistocene tephrostratigraphy of the Eastern Bay of Plenty region, New Zealand. (1995) Victoria University. PhD thesis.
McCulloch MT, Kyser TK, Woodhead JD, Kinsley L. Pb–Sr–Nd–O isotopic constraints on the origin of rhyolites from the Taupo volcanic zone of New Zealand; evidence for assimilation followed by fractionation from basalt. Contributions to Mineralogy and Petrology (1994) 115:303–312.[CrossRef][Web of Science]
Milner DM, Cole JW, Wood CP. Mamaku Ignimbrite: a caldera-forming ignimbrite erupted from a compositionally zoned magma chamber in Taupo Volcanic Zone, New Zealand. Journal of Volcanology and Geothermal Research (2001) 122:243–264.
Miyashiro A. Volcanic rock series in island arcs and active continental margins. American Journal of Science (1974) 274:321–355.[Abstract]
Moore GM, 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]
Müntener O, Kelemen PB, Grove TL. The role of H2O during crystallisation of primitive arc magmas under upper-most mantle conditions and genesis of igneous pyroxenites: an experimental study. Contributions to Mineralogy and Petrology (2001) 141:643–658.[Web of Science]
Nairn IA, Shane PR, Cole JW, Leonard GJ, Self S, Pearson N. Rhyolite magma processes of the AD 1315 Kaharoa eruption episode, Tarawera volcano, New Zealand. Journal of Volcanology and Geothermal Research (2004) 131:265–294.[CrossRef][Web of Science]
Nicholls IA, Oba T, Conrad WK. The nature of primary rhyolitic magmas involved in crustal evolution: Evidence from an experimental study of cummingtonite-bearing rhyolites, Taupo Volcanic Zone, New Zealand. Geochimica et Cosmochimica Acta (1992) 56:955–962.[CrossRef][Web of Science]
Noble DC. Sodium, potassium, and ferrous iron contents of some secondarily hydrated natural silicic glasses. American Mineralogist (1967) 52:280–286.[Web of Science]
Ogawa Y, Bibby HM, Caldwell TG, Takakura S, Uchida T, Matsushima N, Bennie SL, Tosha T, Nishi Y. Wide-band magnetotelluric measurements across the Taupo volcanic zone, New Zealand; preliminary results. Geophysical Research Letters (1999) 26:3673–3676.[CrossRef][Web of Science]
Patino LC, Velbel MA, Price JR, Wade JA. Trace element mobility during spheroidal weathering of basalts and andesites in Hawaii and Guatemala. Chemical Geology (2003) 202:343–364.[CrossRef][Web of Science]
Price RC, Gamble JA, Smith IEM, Stewart RB, Eggins S, Wright IC. An integrated model for the temporal evolution of andesites and rhyolites and crustal development in New Zealand's North Island. Journal of Volcanology and Geothermal Research (2005) 140:1–24.[CrossRef][Web of Science]
Prouteau G, Scaillet B. Experimental constraints on the origin of the 1991 Pinatubo dacite. Journal of Petrology (2003) 44:2203–2241.
Rapp RP. Amphibole-out phase boundary in partially melted metabasalt, its control over liquid fraction and composition and source permeability. Journal of Geophysical Research (1995) 100:15601–15610.[CrossRef]
Rapp RP, Watson EB, Miller CF. Partial melting of amphibolite/eclogite and the origin of Archean trondhjemites and tonalites. Precambrian Research (1991) 51:1–25.[CrossRef][Web of Science]
Reid FW. Origin of the rhyolitic rocks of the Taupo Volcanic Zone, New Zealand. Journal of Volcanology and Geothermal Research (1983) 18:191–214.[CrossRef][Web of Science]
Rollinson HR. Using Geochemical Data: Evalution, Presentation, Interpretation (1993) Englewood Cliffs, NJ: Prentice Hall. 108–111.
Schmidt MW. Amphibole composition in tonalite as a function of pressure: an experimental calibration of the Al-in-hornblende barometer. Contributions to Mineralogy and Petrology (1992) 110:304–310.[CrossRef][Web of Science]
Schmitz MD, Smith IEM. The petrology of the Rotoiti eruption sequence, Taupo Volcanic Zone: an example of fractionation and mixing in a rhyolitic system. Journal of Petrology (2004) 45:2045–2066.
Shane P, Martin SB, Smith VC, Beggs KF, Darragh MB, Cole JW, Nairn IA. Multiple rhyolite magmas and basalt injection in the 17·7 ka Rerewhakaaitu eruption episode from Tarawera volcanic complex, New Zealand. Journal of Volcanology and Geothermal Research (2007) 164:1–26.[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 (1993a) 113:143–166.[CrossRef][Web of Science]
Sisson TW, Layne GD. H2O in basalt and basaltic andesite glass inclusions from four subduction-related volcanoes. Earth and Planetary Science Letters (1993) 117:619–635.[CrossRef][Web of Science]
Sisson TW, Grove TL, Coleman DS. Hornblende gabbro sill complex at Onion Valley, California, and a mixing origin for the Sierra Nevada Batholith. Contributions to Mineralogy and Petrology (1996) 126:81–108.[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]
Smith VC, Shane P, Smith IEM. Tephrostratigraphy and geochemical fingerprinting of the Mangaone Subgroup tephra beds, Okataina Volcanic Centre, New Zealand. New Zealand Journal of Geology and Geophysics (2002) 45:207–219.
Smith VC, Shane P, Nairn IA. Reactivation of a rhyolitic magma body by new rhyolitic intrusion before the 15·8 Rotorua eruptive episode: implications for magma storage in the Okataina Volcanic Centre, New Zealand. Journal of the Geological Society, London (2004) 161:757–772.
Smith VC, Shane P, Nairn IA. Trends in rhyolite geochemistry, mineralogy, and magma storage during the last 50 kyr at Okataina and Taupo volcanic centres, Taupo volcanic zone, New Zealand. Journal of Volcanology and Geothermal Research (2005) 148:372–406.[CrossRef][Web of Science]
Spear FS. An experimental study of hornblende stability and compositional variability in amphibolite. American Journal of Science (1981) 281:697–734.
Sun S, McDonough WF. Chemical and isotopic systematics of oceanic basalts: implications for mantle compositions and processes. In: Saunders, A. D. & Norry, M. J. (eds). Magmatism in the Ocean Basins. Geological Society, London, Special Publications (1989) 42:313–345.
Sutton AN, Blake S, Wilson CJN. An outline geochemistry of rhyolite eruptives from Taupo volcanic centre, New Zealand. Journal of Volcanology and Geothermal Research (1995) 68:153–175.[CrossRef][Web of Science]
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.
Thomas WM, Ernst WG. The aluminium content of hornblende in calc-alkaline granitic rocks: a mineralogic barometer calibrated experimentally to 12 kbars. In: Fluid-mineral interactions; a tribute to H. P. Eugster. Geochemical Society, Special Publications—Spenser RJ, Chou IM, eds. (1990) 2:59–63.
Vogel TA, Flood TP, Patino LC, Wilmot MS, Maximo RPR, Arpa CB, Arcilla CA, Stimac JA. Geochemistry of silicic magmas in the Macolod Corridor, SW Luzon, Philippines: evidence of distinct, mantle-derived, crustal sources for silicic magmas. Contributions to Mineralogy and Petrology (2006) 151:267–281.[CrossRef][Web of Science]
Vogel TA, Hidalgo PJ, Patino LC, Tefend KS, Ehrlich R. Evaluation of magma mixing and fractional crystallization using whole-rock chemical analyses: polytopic vector analyses. (2008) Geochemistry, Geophysics, Geosystems 9, doi:10.1029/2007GC001790.
Waight TE, Baker JA, Peate DW. Sr isotope ratio measurements by double-focusing MC-ICPMS: techniques, observations and pitfalls. International Journal of Mass Spectrometry (2002) 221:229–244.[CrossRef][Web of Science]
Wilson CJN. Stratigraphy, chronology, styles and dynamics of late Quaternary eruptions from Taupo Volcano, New Zealand. Philosophical Transactions of the Royal Society of London, Series A (1993) 343:205–306.
Wilson CJN, Blake S, Charlier BLA, Sutton AN. The 26·5 ka Oruanui eruption, Taupo Volcano, New Zealand: development, characteristics and evacuation of a large rhyolitic magma body. Journal of Petrology (2006) 27:35–69.
Wilson CJN, Gravley DM, Leonard GS, Rowland JV. Volcanism in the central Taupo Volcanic Zone, New Zealand: tempo styles and controls. In: Studies in Volcanology: The Legacy of George Walker—Thordarson T, et al, eds. (2008) Volume 2. IAVCEI Proceedings in Volcanology. (in press).
Wolf MB, Wyllie PJ. Dehydration-melting of amphibolite at 10 kbar: the effects of temperature and time. Contributions to Mineralogy and Petrology (1994) 115:369–383.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. D. Deering Cannibalization of an amphibole-rich andesitic progenitor induced by caldera-collapse during the Matahina eruption: Evidence from amphibole compositions American Mineralogist, August 1, 2009; 94(8-9): 1162 - 1174. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||