Skip Navigation


Journal of Petrology Advance Access originally published online on December 8, 2005
Journal of Petrology 2006 47(3):595-629; doi:10.1093/petrology/egi087
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
47/3/595    most recent
egi087v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (5)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by TATSUMI, Y.
Right arrow Articles by YOSHIKAWA, M.
Right arrow Search for Related Content
GeoRef
Right arrow GeoRef Citation
Social Bookmarking
 Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2005. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org

The Petrology and Geochemistry of Oto-Zan Composite Lava Flow on Shodo-Shima Island, SW Japan: Remelting of a Solidified High-Mg Andesite Magma

Y. TATSUMI1,*, T. SUZUKI1, H. KAWABATA1, K. SATO1, T. MIYAZAKI1, Q. CHANG1, T. TAKAHASHI1, K. TANI1, T. SHIBATA2 and M. YOSHIKAWA2

1 INSTITUTE FOR RESEARCH ON EARTH EVOLUTION (IFREE), JAPAN AGENCY FOR MARINE-EARTH SCIENCE AND TECHNOLOGY (JAMSTEC), YOKOSUKA 237-0061, JAPAN
2 INSTITUTE FOR GEOTHERMAL SCIENCES, KYOTO UNIVERSITY, BEPPU 974-0907, JAPAN

RECEIVED DECEMBER 16, 2004; ACCEPTED OCTOBER 20, 2005


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 GEOLOGICAL SETTINGS
 ANALYTICAL AND EXPERIMENTAL...
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The Oto-Zan lava in the Setouchi volcanic belt is composed of phenocryst-poor, sparsely plagioclase-phyric andesites (sanukitoids) and forms a composite lava flow. The phenocryst assemblages and element abundances change but Sr–Nd–Pb isotopic compositions are constant throughout the lava flow. The sanukitoid at the base is a high-Mg andesite (HMA) and contains Mg- and Ni-rich olivine and Cr-rich chromite, suggesting the emplacement of a mantle-derived hydrous (~7 wt % H2O) HMA magma. However, Oto-Zan sanukitoids contain little H2O and are phenocryst-poor. The liquid lines of descent obtained for an Oto-Zan HMA at 0·3 GPa in the presence of 0·7–2·1 wt % H2O suggest that mixing of an HMA magma with a differentiated felsic melt can reasonably explain the petrographical and chemical characteristics of Oto-Zan sanukitoids. We propose a model whereby a hydrous HMA magma crystallizes extensively within the crust, resulting in the formation of an HMA pluton and causing liberation of H2O from the magma system. The HMA pluton, in which interstitial rhyolitic melts still remain, is then heated from the base by intrusion of a high-T basalt magma, forming an H2O-deficient HMA magma at the base of the pluton. During ascent, this secondary HMA magma entrains the overlying interstitial rhyolitic melt, resulting in variable self-mixing and formation of a zoned magma reservoir, comprising more felsic magmas upwards. More effective upwelling of more mafic, and hence less viscous, magmas through a propagated vent finally results in the emplacement of the composite lava flow.

KEY WORDS: high-Mg andesite; sanukitoid; composite lava; solidification; remelting


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 GEOLOGICAL SETTINGS
 ANALYTICAL AND EXPERIMENTAL...
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The majority of andesites that typify subduction zone magmatism are derived from parental basaltic magmas through variable differentiation processes such as fractional crystallization, mixing with felsic magmas, crustal contamination and anatexis of pre-existing basaltic crustal materials (e.g. Gill, 1981Go; Hildreth, 1981Go; Sakuyama, 1981Go; Hunter, 1998Go; Temel et al., 1998Go; Couch et al., 2001Go; Tatsumi et al., 2002Go; Grove et al., 2003Go). On the other hand, mantle-derived, primary andesitic magmas also exist in arcs. Such unusual andesites are referred to as high-Mg andesites (HMAs), as they are characterized by high MgO contents and/or Mg-number (= 100 x Mg/(Mg + Fe)) and are in equilibrium with Mg-rich mantle minerals. The occurrence of HMAs has been reported, for example, from the Bonin Islands (Kuroda et al., 1978Go), Baja California (Saunders et al., 1987Go), Piip volcano in W Aleutians (Yogodzinski et al., 1994Go), and the Setouchi volcanic belt in SW Japan (Tatsumi & Ishizaka, 1981Go).

High-pressure melting experiments on both simple and natural peridotite systems (e.g. Kushiro, 1969Go; Hirose, 1997Go) have established that partial melting of upper mantle peridotite under hydrous conditions is a possible mechanism for HMA magma production. It has been further demonstrated that some HMAs are multiply saturated with peridotitic phases under hydrous conditions at mantle pressures (e.g. Kushiro & Sato, 1978Go; Tatsumi, 1981Go, 1982Go; Umino & Kushiro, 1989Go; van der Laan et al., 1989Go). These experimental results led Crawford et al. (1989)Go and Tatsumi & Maruyama (1989)Go to the conclusion that the direct overprinting of aqueous fluids released from the subducting lithosphere onto the mantle wedge is a likely mechanism for HMA magma generation. However, this rather simple process is not the only way to attain equilibration between hydrous HMA magmas and mantle peridotite. A process involving partial melting of subducting lithosphere and subsequent reaction of such hydrous felsic slab melts with mantle wedge peridotite has been widely accepted as a likely mechanism for the generation of hydrous HMA magmas (Kay, 1978Go; Pearce et al., 1992Go; Yogodzinski et al., 1994Go; Kelemen, 1995Go; Shimoda et al., 1998Go; Tatsumi, 2001Go; Hanyu et al., 2002Go). Geochemical modelling suggests that the observed Sr–Nd–Pb–Hf isotopic signatures of HMAs can be reasonably and quantitatively understood by this process (Tatsumi & Hanyu, 2003Go).

HMA magmas contain larger amounts of H2O than basalt magmas when the magma left the mantle (e.g. Tatsumi, 1982Go), and will be oversaturated with H2O during ascent, causing extensive crystallization. It is thus likely that HMAs are rich in phenocrysts and contain a larger amount of H2O than other arc basaltic magmas, as observed for boninites (e.g. Dobson & O'Neil, 1987Go; Umino & Kushiro, 1989Go; Sobolev & Danyushevsky, 1994Go). In contrast, HMAs in the Setouchi volcanic belt are rather aphyric and poor in H2O (e.g. Tatsumi & Ishizaka, 1981Go, 1982aGo, 1982bGo). This apparent paradox needs to be assessed.

In this paper, the petrography and geochemistry of a single composite lava flow that contains HMA at its base are presented. The origin of the composite lava flow and a likely mechanism that may overcome the above paradox will be discussed.


    GEOLOGICAL SETTINGS
 TOP
 ABSTRACT
 INTRODUCTION
 GEOLOGICAL SETTINGS
 ANALYTICAL AND EXPERIMENTAL...
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The Oto-Zan lava flow is located on Shodo-Shima Island, SW Japan (Fig. 1). The current SW Japan arc is being built by subduction of the Philippine Sea plate from the Nankai Trough beneath the Eurasian plate (Fig. 1). The Quaternary volcanic front is located ~100 km above the top of the subducting slab, as is the case for most arc–trench systems (Tatsumi & Eggins, 1995Go). Miocene igneous rocks are distributed in the present fore-arc region of the SW Japan arc. In the near-trench region of this arc (the Outer Zone), felsic volcano-plutonic complexes were emplaced at 14 ± 1 Ma (Shibata, 1978Go; Sumii, 2000Go) into a Cretaceous to Miocene accretionary prism or subduction complex (Fig. 1). Synchronous with this near-trench magmatism, i.e. at 13·7 ± 1·0 Ma (Tatsumi et al., 2001Go, 2003Go), volcanism took place in the Setouchi volcanic belt (Fig. 1). The zonal arrangement of Miocene magmatism parallel to the arc–trench system suggests a contribution from the plate subduction to the formation of these magmatic belts.


Figure 1
View larger version (64K):
[in this window]
[in a new window]
 
Fig. 1. Tectonic setting of the Setouchi volcanic belt (inset) and a simplified geological map of Shodo-Shima Island after Tatsumi (1983)Go. The Setouchi volcanic belt (open circles) and coeval Outer Zone felsic complexes (hatched circles) are distributed to the trench side of the Quaternary volcanic front (QVF) of the SW Japan arc. Setouchi volcanic rocks on Shodo-Shima Island, which cover granitic basement (4), are divided into two groups, the Kankakei (1, porphyritic andesite; 2, sanukitoid) and the Uchinomi (3) formations in descending order. The Oto-Zan lava flow forms the lowermost part of the Kankakei Formation. Bathymetric contours are given in 2000 m intervals.

 
Miocene magmatism, both in the Setouchi region and the Outer Zone of the SW Japan arc, was largely synchronous with the timing of a 40–50° clockwise rotation of the arc sliver at 14–16 Ma that was caused by the opening of the Sea of Japan back-arc basin (Otofuji et al., 1991Go). The Shikoku Basin, situated to the south of the SW Japan arc (Fig. 1), is also a back-arc basin that was created behind the Izu–Bonin–Mariana arc by rifting at 30–15 Ma (Okino et al., 1994Go, 1998Go, 1999Go). It is thus inferred that the southward drift of the SW Japan arc in association with the clockwise rotation of the SW Japan arc sliver resulted in subduction of the young (<16 my) oceanic lithosphere of the Shikoku Basin. Such an unusual tectonic setting and accompanying thermal conditions could have caused slab melting to produce magmatism unusually close to the trench in the Miocene SW Japan arc (Furukawa & Tatsumi, 1999Go; Tatsumi & Hanyu, 2003Go).

Rocks distributed in the Setouchi volcanic belt include phenocryst-poor, sparsely plagioclase-phyric andesites (including HMAs) and basalts referred to as sanukitoids (Tatsumi & Ishizaka, 1981Go), porphyritic and plagioclase-phyric ‘normal’ calc-alkalic andesites, garnet-bearing dacites/rhyolites, and pitchstones. These volcanic rocks are collectively distributed on Shodo-Shima Island, which is located to the NE of Shikoku (Fig. 1). The older Setouchi volcanic rocks on the island (Uchinomi Formation), which cover basement granites and gneisses, are composed of felsic lava flows, lava domes, sheets, dykes and volcaniclastic rocks, whereas the younger group (Kankakei Formation) consists of volcanic rocks of intermediate to mafic composition (Fig. 1). K–Ar dates for Setouchi volcanics on Shodo-Shima Island (Tatsumi et al., 2001Go) yield mean ages of 12·82 ± 0·12 and 13·78 ± 0·17 Ma for the Kankakei and Uchinomi Formations, respectively, which is consistent with the stratigraphic relations. Tatsumi (1983)Go suggested that most lavas and volcaniclastics on Shodo-Shima Island were emplaced under subaqueous conditions, as they show typical water-chilled structures with cracks perpendicular to the block surface.

The Oto-Zan lava flow, which is located in the western part of the island, forms the lowermost part of the Kankakei Formation and directly covers basement rocks (Fig. 1). The maximum thickness of this lava flow is ~100 m. Although the lowermost part of the lava flow forms a lava clinker consisting of water-chilled volcanic breccias and volcaniclastics, the main part of the lava flow is massive. Importantly, no flow unit boundary is observed within the lava flow, suggesting that the Oto-Zan lava is a single lava flow.


    ANALYTICAL AND EXPERIMENTAL METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 GEOLOGICAL SETTINGS
 ANALYTICAL AND EXPERIMENTAL...
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Chemical analyses
Major and trace element (Ni, Cu, Zn, Rb, Sr, Y, Zr, Nb, Ba, Pb and Th) compositions were measured using RIGAKU® Simaltics 3512 and Rix 3000 X-ray fluorescence (XRF) spectrometers on fused glass beads and pressed powder pellets, respectively. Detailed analytical procedures have been described by Goto & Tatsumi (1994Go, 1996Go) and Tani et al. (2005)Go.

Rock samples for trace elements (ICP-MS) and Sr–Nd–Pb isotope analysis were crushed to coarse chips (<0·5 mm3) and fresh pieces were hand-picked. To avoid surface contamination, the rock chips were washed with ethanol and then leached with 0·5M HCl at room temperature for 1 h. Finally, the chips were rinsed three times with Milli-Q water. The chips were ground to less than 200 mesh size using a vibration mill made of alumina ceramic.

Concentrations of rare earth and 14 other trace elements (Sc, Rb, Sr, Y, Zr, Nb, Cs, Ba, Hf, Ta, Tl, Pb, Th and U) were determined using a VG Elemental® PQ3 inductively coupled plasma mass spectrometer (ICP-MS) enhanced with a chicane lens system, following the procedures described by Chang et al. (2003)Go. Trace element data, except for high-field-strength elements (HFSE: Zr, Nb, Hf and Ta), were obtained after HF–HClO4–HNO3 digestions. For HFSE analysis, alkali fusion (LiBO2/Li2B4O7, Spectroflux® 100B of Johnson Matthey) was applied to ensure a complete decomposition of refractory minor phases. Analytical accuracy and precision for ICP-MS analyses, estimated from repeated measurements of international reference rocks, were better than 10 and 2–5%, respectively.

The analytical procedure used for chemical separation and mass spectrometry for Sr, Nd and Pb isotope determinations was outlined by Yoshikawa et al. (2001)Go, Shibata et al. (2003)Go and Miyazaki et al. (2003)Go. Total procedural blanks for Sr, Nd and Pb were less than 10, 10 and 5 pg, respectively. Mass spectrometry was performed on a Thermo-Finnigan® Triton TI equipped with nine Faraday cups, using a static multi-collection mode. Normalizing factors used to correct for isotopic fractionation in the Sr, Nd and Pb isotope analyses were 86Sr/88Sr = 0·1194, 146Nd/144Nd = 0·7219, and 0·145 % per atomic mass unit, respectively. Measured isotopic ratios for standard materials were 87Sr/86Sr = 0·710268 ± 19 (2{sigma}) for NIST 987 (n = 10), 143Nd/144Nd = 0·511844 ± 11 (2{sigma}) for La Jolla (n = 10), and 208Pb/204Pb = 36·712 ± 11 (2{sigma}), 207Pb/204Pb = 15·495 ± 4 (2{sigma}), 206Pb/204Pb = 16·939 ± 3 (2{sigma}) for NIST 981 (n = 35).

Mineral compositions were analysed using JEOL JXA-8800 and -8900 electron-probe micro-analysers following the method described by Shukuno (2003)Go. The excitation potential, specimen current, and analytical time were: 15 kV, 15 nA and 20 s (25 kV, 20 nA and 100 s for Mn, Ca and Ni analyses) for olivine; 15 kV, 12 nA and 20 s for spinel; 15 kV, 15 nA and 20 s for pyroxene and plagioclase. ZAF correction procedures were employed.

Melting experiments
Starting material
An HMA (SD-249), which forms a part of the Oto-Zan lava flow, was used as the starting material for the experiments (Table 1). This HMA sample possesses a composition almost identical to the sample from the lowermost part of the Oto-Zan lava flow (OTO-1 in Table 1), but with lower Na2O and higher K2O contents. Sample SD-249 was selected because it is completely fresh, whereas OTO-1 contains altered glass in the groundmass. In order to adjust the above compositional differences and to apply the experimental results to the Oto-Zan composite lava compositions, we used ‘corrected’ melt compositions, as described later.


View this table:
[in this window]
[in a new window]
 
Table 1: Compositions of Oto-Zan HMA samples

 
Experimental
Experiments were performed in a Kobelco 500 MPa type internally heated pressure vessel in which pure Ar gas was used as the pressure medium. Three glasses with different water contents were used as starting materials, and were prepared as follows. The powdered sample was first sealed in a Pt capsule (4·7 mm inner diameter, 0·15 mm wall) with a small amount of distilled water. The sample was then heated to 1300–1350°C at 0·2 GPa for 20–30 min, and quenched isobarically. The amounts of H2O in the recovered glass are 0·7, 1·5 and 2·1 wt %, measured using an FT-IR micro-spectrometer. Although slight Fe-loss occurred (from 5·6 to 5·2–4·9 wt % FeO* from the original sample to the recovered glasses, respectively), no change was observed for other elements.

The glass sample was set in an Au25Pd capsule (2·0 mm inner diameter, 0·15 mm wall), and was hung by Mo wire in the hotspot of a Mo furnace within the pressure vessel and held for 6–75 h at 0·3 GPa and a set temperature. The experimental pressure corresponds to middle-crustal depths beneath the Setouchi volcanic belt. Pressures were measured with a strain-gauge pressure transducer. Temperatures were monitored with two W5Re–W26Re thermocouples spaced vertically 5 mm apart, and the observed temperature gradient across the sample was less than 10°C. At the end of the run, the hanging wire was cut with a surging current, thereby letting the capsule fall to the cold (<250°C) bottom of the vessel and quench isobarically.

Oxygen fugacity during melting experiments was estimated based on a solution model for coexisting magnetite and ilmenite (Spencer & Lindsley, 1981Go). Although the observed compositions of magnetite and ilmenite are beyond the upper limit of their oxygen barometer, we tentatively show these estimates in Fig. 2. The experimental device generated redox conditions within a well-defined range at ~NNO+3, which is ~2 log units higher than the oxygen fugacity estimated for the Oto-Zan HMA magma based on olivine–spinel compositions (see below). If this is the case, then the difference in oxygen fugacity will have significant impacts on the stability of oxide minerals and, thereby, could influence the variation of FeO*/MgO and TiO2 during magmatic differentiation.


Figure 2
View larger version (24K):
[in this window]
[in a new window]
 
Fig. 2. fO2 relative to the NNO and QFM oxygen buffers during melting experiments and of the HMA sanukitoid OTO-1 inferred from magnetite–ilmenite compositions (based on method of Spencer & Lindsley, 1981Go) and spinel compositions based on methods by Ballhaus et al. (1990Go, 1991)Go, and Ballhaus (1993)Go, respectively.

 
Mineral and glass compositions
Mineral and glass compositions were analysed using JEOL JXA-8800 electron-probe micro-analyser. The excitation potential, specimen current, and analytical time were 15 kv, 12 nA and 20 s, respectively. A defocused electron beam of 10 µm in diameter was used for glass, whereas a focused beam was employed for the measurement of the crystals. ZAF correction procedures were employed.

Modal compositions were calculated by mass balance based on compositions of the starting material, minerals and glasses.


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 GEOLOGICAL SETTINGS
 ANALYTICAL AND EXPERIMENTAL...
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Petrography
Modal proportions of phenocrysts and representative mineral compositions are shown in Tables 26. Rocks forming the Oto-Zan lava flow are distinct from typical orogenic andesites in two important ways. First, they are poor in phenocrysts (<15 vol. %). Second, they do not contain plagioclase as a phenocryst phase, which is the major phenocryst phase in typical andesites. These petrographic signatures characterize sanukitoids occurring in the Setouchi volcanic belt. An additional petrographic feature of sanukitoid is its compact nature, i.e. the absence of vesicles.


View this table:
[in this window]
[in a new window]
 
Table 2: Modal mineralogy of the Oto-Za lava flow as a function of stratigraphic height above the base of the flow

 

View this table:
[in this window]
[in a new window]
 
Table 3: Representative compositions of olivine phenocrysts

 

View this table:
[in this window]
[in a new window]
 
Table 4: Representative compositions of spinel inclusions

 

View this table:
[in this window]
[in a new window]
 
Table 5: Representative compositions of clinopyroxene phenocrysts

 

View this table:
[in this window]
[in a new window]
 
Table 6: Representative compositions of orthopyroxene phenocrysts

 
One petrographic signature observed for the Oto-Zan lava is that both the amount and the assemblage of phenocryst phases change gradually within the single lava flow (Fig. 3), indicating that the lava is a composite flow. The basal part of the lava flow is composed of augite–olivine andesite, with the amount of olivine and augite decreasing upwards. Orthopyroxene, instead of olivine, appears as a phenocryst phase ~20 m from the base. The upper part of the lava flow comprises augite–orthopyroxene (bronzite to hypersthene) andesite.


Figure 3
View larger version (17K):
[in this window]
[in a new window]
 
Fig. 3. Variation of modal minerology throughout the Oto-Zan lava flow as a function of height above the base. The phenocryst assemblage changes from olivine (ol) + clinopyroxene (cpx) in the basal part, via olivine + clinopyroxene + orthopyroxene (opx), to orthopyroxene + clinopyroxene in the upper part of the lava flow, indicating that the Oto-Zan lava forms a composite lava flow. The absence of plagioclase phenocrysts, which characterizes sanukitoids in the Setouchi volcanic belt, is significant.

 
Olivine phenocrysts at the base of the lava flow (OTO-1) are more magnesian and characterized by a narrower compositional range than those in the sample at 19 m above the base (OTO-7; Fig. 4). Olivine in OTO-1 has a composition in equilibrium with the bulk rock in terms of Fe–Mg exchange partitioning, whereas that in OTO-7 has Mg-numbers lower than that of inferred equilibrium olivine (Fig. 4). The NiO content at a given Mg-number of olivine phenocryst in OTO-7 is higher than that in OTO-1. The compositional trend for olivine in OTO-1 can be explained by simple olivine fractionation inferred on the basis of Fe–Mg–Ni exchange partitioning between olivine and silicate melts (Roeder & Emslie, 1970Go; Kinzler et al., 1990Go) and an assumption of Fe2+/(Fe2+ + Fe3+) = 0·9 in the magma (Fig. 4), whereas some additional processes may be needed for understanding the unusually nickeliferous trend for OTO-7 olivine. Such unusually nickeliferous olivine phenocrysts have been found in some basalts and andesites (e.g. Sato & Banno, 1983Go; Nabelek & Langmuir, 1986Go; Nakamura, 1995Go; Tatsumi et al., 2003Go, 2004Go). Nakamura (1995)Go examined the compositional zoning of olivine phenocrysts in calc-alkalic andesites from the Yatsugatake volcano, Central Japan, by using a growth and diffusion model in the Mg–Fe–Ni system. He showed that the composition of nickeliferous olivine phenocrysts could be explained by diffusion processes within normally zoned olivine, causing Fe-enrichment without a marked depression in the Ni content. This is due to a greater Fe–Mg inter-diffusion coefficient than the Ni tracer-diffusion coefficient in olivine. Tatsumi et al. (2003)Go favored this explanation for the occurrence of unusually nickeliferous olivine in rather olivine-rich (~20 vol. %) sanukitoids (Fig. 4). Although OTO-7 contains only 7 vol. % of olivine phenocrysts, it may be inferred that a long residence time of olivine phenocrysts in the magma may cause such unusual compositions.


Figure 4
View larger version (15K):
[in this window]
[in a new window]
 
Fig. 4. Compositions of olivine phenocrysts in two samples (OTO-1, OTO-7) from the Oto-Zan composite lava flow. Stars indicate compositions of olivine in equilibrium with the host rock, calculated based on Fe–Mg–Ni exchange partitioning and an inferred Fe3+/Fe2+ ratio in the magma. An olivine perfect fractionation trend is also shown by the dashed line (see text). Olivine phenocrysts in OTO-7, as well as those in olivine-rich HMAs from other regions of the Setouchi volcanic belt, have compositions in disequilibrium with the magma. Compositional ranges of olivine in mantle peridotites (Sato, 1977Go; Takahashi, 1990Go) and olivine-rich Setouchi HMAs (Tatsumi et al., 2003Go) are shown.

 
Many olivine phenocrysts contain chromite inclusions. In a sample from the base of the lava flow (OTO-1), these spinels show a limited compositional range, whereas those in OTO-7 range from chromite to chromian titanomagnetite (Fig. 5a). The following compositional differences in chromite inclusions should be stressed: (1) chromites in OTO-7 are enriched in Fe2+, Fe3+, Ti and depleted in Mg, Al, Cr compared with those in OTO-1 (Fig. 5a–c); (2) such compositional differences are also observed for chromite included in individual olivine phenocrysts in OTO-7 as a function of the distance from the rim of the olivine crystal. These compositional changes are documented for spinel inclusions in rather phenocryst-rich HMAs from the Setouchi volcanic belt (Fig. 5a–c). Scowen et al. (1991)Go documented similar chemical variability of chromite inclusions in olivine phenocrysts from the 1959 Kilauea Iki lava lake and proposed that the compositional changes may be caused by re-equilibration with the residual melt by cationic diffusion (Mg, Al, Cr outwards and Fe2+, Fe3+, Ti inwards) through the olivine crystals. Re-equilibration of chromite inclusions, but not of the Ni content in the host olivine as described before, implies a significantly larger flux of Al3+,Cr3+, Fe3+ and Ti4+ through the olivine crystal than Ni2+. Although the diffusivities of such elements having high charge have not been measured experimentally, they are likely to be significantly slower than the divalent cations such as Ni. If so, then the explanation for the elevated Ni content in olivine, involving a slow cooling rate, could not be valid for the variability in the composition of chromite. Further work is needed to understand the compositional relationship between chromite inclusions and the host olivine.


Figure 5
View larger version (26K):
[in this window]
[in a new window]
 
Fig. 5. Compositions of chromite inclusions in olivine phenocrysts (a, b and c) and spinel–olivine compositional relations (d, e and f) in sanukitoids from the Oto-Zan composite lava flow and other regions in the Setouchi volcanic belt. The unusual Fe3+-rich compositional trend for chromites in OTO-7, as well as those characteristically observed for olivine-rich HMA sanukitoids, can be explained by elemental diffusion through the olivine crystals associated with the long residence time and slow rate of cooling of the magmas (see text). Data for HMAs are from Tatsumi & Ishizaka (1981Go, 1982aGo, 1982bGo) and Tatsumi et al. (2003)Go. OSMA, olivine spinel mantle array (Arai, 1994Go).

 
Ballhaus et al. (1990Go, 1991)Go and Ballhaus (1993)Go examined Cr–Al-rich spinel compositions and demonstrated that Fe3+ in spinel may provide a reasonable estimate of the fO2 relative to the FMQ buffer for a magma. Figure 2 indicates fO2 estimates based on spinel crystals in the Oto-Zan lava flow and suggests that the fO2 of the Oto-Zan magma may be close to, or one log-unit higher than, the NNO buffer, which is a typical value for subduction zone magmas (Ballhaus, 1993Go).

Clinopyroxene phenocrysts in the lower part of the Oto-Zan lava flow are oscillatory zoned and show a wide range in composition (e.g. OTO-1 in Fig. 6). On the other hand, those in olivine-free pyroxene andesites in the upper part of the lava flow (Fig. 3) include both normally zoned and reversely zoned crystals (Table 5). Orthoyroxene-phyric rocks also contain both normally zoned and reversely zoned orthopyroxene phenocrysts (Fig. 6 and Table 6), suggesting some disequilibrium processes operated during magmatic differentiation, such as mixing of magmas with different compositions.


Figure 6
View larger version (28K):
[in this window]
[in a new window]
 
Fig. 6. Compositions of pyroxene phenocrysts in three samples from the Oto-Zan composite lava flow. Both reversely zoned (RZ) and normally zoned (NR) orthopyroxene phenocrysts are observed in Oto-Zan sanukitoids, suggesting mixing of compositionally different magmas during magmatic differentiation processes.

 
In the middle to upper part of the lava flow, i.e. typically at >20 m above the base and higher, pegmatitic streaks, which consist of cristobalite and minor alkali feldspar and biotite, are developed along platy joints.

Bulk chemical compositions
Major and trace element compositions for all samples, obtained by XRF analysis, are listed in Table 7. Trace element concentrations (ICP-MS) and Sr–Nd–Pb isotopic compositions for representative samples are shown in Tables 8 and 9.


View this table:
[in this window]
[in a new window]
 
Table 7: Major and trace element compositions measured by XRF of the Oto-Zan lava flow

 

View this table:
[in this window]
[in a new window]
 
Table 8: Trace element concentrations by ICP-MS

 

View this table:
[in this window]
[in a new window]
 
Table 9: Sr–Nd–Pb isotopic compositions

 
Sanukitoids from the Oto-Zan lava flow exhibit more of a calc-alkalic trend than andesites from Quaternary volcanoes in the NE Japan arc and other sanukitoids in the Setouchi volcanic belt (Fig. 7a). Oto-Zan sanukitoids have N-MORB-normalized incompatible element patterns identical to those in the other regions of the Setouchi volcanic belt (Fig. 7b), which typify both subduction zone magmas and bulk continental crust compositions, i.e. higher concentrations of elements with higher incompatibility during mantle melting and relative depletions and enrichment in Nb–Ta and Pb, respectively. In contrast, boninite, another type of HMA, is more depleted in incompatible trace elements and REEs than sanukitoids (Fig. 7b and c).


Figure 7
View larger version (24K):
[in this window]
[in a new window]
 
Fig. 7. Geochemical characteristics of sanukitoids from the Oto-Zan lava flow and other regions of the Setouchi volcanic belt (Shimoda et al., 1998Go; Tatsumi & Ishizaka, 1982aGo, 1982bGo; Tatsumi et al., 2003Go), boninites (Pearce et al., 1992Go; Taylor et al., 1994Go) and Quaternary calc-alkalic volcanic rocks from the NE Japan arc (NEJ CA; Tatsumi & Kogiso, 2003Go). The liquid line of descent obtained for an HMA from the Shasta region, California (Grove et al., 2003Go) is shown in (a). Sanukitoids comprising the Oto-Zan lava flow show an extreme calc-alkalic trend and have trace element characteristics identical to those from other regions of the Setouchi volcanic belt and are distinct from boninites. TH, tholeiitic series; CA, calc-alkalic series (Miyashiro, 1974Go). Normalization values from Sun & McDonough (1989)Go. (a) FeO*/MgO vs wt% SiO2; (b) N-MORB normalized incompatible element patterns; (c) chondrite-normalized REE patterns. Normalization values from Sun & McDonough (1989)Go.

 
The Sr–Nd–Pb isotopic compositions of Oto-Zan sanukitoids are shown in Fig. 8, together with previously reported HMAs and basement rocks from Shodo-Shima Island (Shimoda et al., 1998Go; Tatsumi et al., 2002Go). Oto-Zan sanukitoids exhibit more enriched isotopic signatures than other HMAs from the Mito Peninsula (Fig. 1), only ~5 km away from the Oto-Zan.


Figure 8
View larger version (10K):
[in this window]
[in a new window]
 
Fig. 8. Sr–Nd–Pb isotopic compositions of the Oto-Zan lava flow, HMAs (SDHMA; Shimoda et al., 1998Go), and basement rocks (Tatsumi et al., 2002Go) from Shodo-Shima Island. The isotopic compositions of the Oto-Zan sanukitoids differ from those of the Shodo-Shima HMAs and cannot be explained by crustal contamination of those HMA magmas, suggesting the distribution of highly heterogeneous magma sources beneath Shodo-Shima Island.

 
The compositional variation in the Oto-Zan lava flow as a function of height from the base of the lava flow is shown in Figs 9 and 10. Concentrations of most elements change with increasing height, synchronously with the changes in modal composition (cf. Fig. 3), although the abundance of some elements such as Ba, Pb, Sr, Y and Th, and Nd and Pb isotopic ratios remain essentially constant throughout the lava flow (Figs 9 and 10). Two samples (SD303 at the top and OTO-7 at 19 m from the base) possess higher 87Sr/86Sr (>0·7057) than the other samples (~0·7055). This is unlikely to have resulted from selective crustal contamination, because of rather constant 143Nd/144Nd compositions for all the samples. An unusual ‘spike’ in those samples is also documented for Ba (Fig. 9). One possible explanation for such spikes in Sr isotopic composition and Ba abundance may be selective transport of these elements in gas-rich phases that could extensively react with crustal components; there is a higher concentration of Ba and Sr in the pegmatitic streaks that are ubiquitous in the middle to upper part of the lava flow.


Figure 9
View larger version (29K):
[in this window]
[in a new window]
 
Fig. 9. Variations in element concentrations in the Oto-Zan lava flow as a function of height from the base. Concentrations of most elements change with increasing height synchronously with the changes in modal minerology (cf. Fig. 2), although the abundance of some elements such as Ba, Pb, Sr, Y and Th remains constant throughout the lava flow.

 

Figure 10
View larger version (12K):
[in this window]
[in a new window]
 
Fig. 10. Variations in Sr–Nd–Pb isotopic ratios in the Oto-Zan lava flow as a function of height from the base. Although some ‘spikes’ are documented, these isotopic ratios are broadly constant throughout the lava flow (shaded), suggesting a rather closed-system differentiation from a common parental magma for the Oto-Zan sanukitoid samples.

 
Melting phase relations and liquid compositions
The phase assemblages and the average compositions of minerals and glasses are listed in Table 10, and the modal compositions of experimental products in the presence of 0·7, 1·5 and 2·1 wt % H2O are shown in Fig. 11. In order to apply the melt compositions obtained in the experiments to the Oto-Zan lava compositions, ‘corrected’ melt compositions were calculated by adjusting the Na2O and K2O contents in the starting material (SD-249) to those in the OTO-1 (Tables 1 and 10) and recalculating to 100% total. This should provide reasonable estimates for K2O and at least for Na2O in rather high-temperature melts, as these elements are not strongly partitioned into the mineral phases present (e.g. pyroxene; Table 10) and behave largely as incompatible elements.


Figure 11
View larger version (15K):
[in this window]
[in a new window]
 
Fig. 11. Modal minerology of experimental run products estimated from mass balance calculations and phase compositions. Although amphibole and biotite are present in the low-T run products, modal proportions could not be estimated because of the absence of phases of analysable size. Plag, plagioclase; opx, orthopyroxene; cpx, clinopyroxene; Fe–Ti, magnetite and/or ilmenite; ol, olivine; amph, amphibole; bt, biotite.

 

View this table:
[in this window]
[in a new window]
 
Table 10: Results of experiments for Oto-Zan HMA

 
An important criterion for equilibrium is the achievement of regular and consistent partitioning of major elements between crystalline phases and melt. The calculated orthopyroxene–melt and clinopyroxene–melt Fe–Mg exchange distribution coefficients (KDFe–Mg) are 0·24 ± 0·02 (1{sigma}) and 0·31 ± 0·03 (1{sigma}), respectively. The plagioclase–melt Ca–Na exchange distribution coefficient is 2·1 ± 0·2 (1{sigma}). Although the clinopyroxene–melt KDFe–Mg is slightly higher, the other exchange distribution coefficients are close to those previously reported (e.g. Gaetani et al., 1993Go; Sisson & Grove, 1993Go; Grove et al., 2003Go). This, together with constant melt and mineral compositions within each run product (Table 10), suggests a sufficiently close approach to equilibrium.

The ‘corrected’ melt compositions are plotted in Fig. 12, together with the compositional range of the Oto-Zan composite lava flow. In the FeO*/MgO vs SiO2 diagram, the liquid line of descent obtained for the Oto-Zan HMA is close to that for an HMA from the Mt Shasta region (Grove et al., 2002Go). In a strict sense, this crystallization path, especially in its early stage, does not follow a calc-alkalic differentiation trend because the degree of increasing FeO*/MgO against SiO2 content is greater than that defined by Miyashiro (1974)Go. It should be stressed that differences in the H2O content of the starting materials (0·7, 1·5 and 2·1 wt %) do not yield significant changes in the liquid line of descent, although small but systematic differences are recognized for TiO2 and Al2O3 (Fig. 12).


Figure 12
View larger version (18K):
[in this window]
[in a new window]
 
Fig. 12. Comparison of the compositions of the quenched experimental melts and the Oto-Zan lava. The liquid line of descent for the Oto-Zan HMA (continuous lines for 0·7 wt % H2O experiments), which is similar to that of a Mt Shasta HMA (grey arrow in the FeO*/MgO diagram) from Grove et al. (2002)Go, cannot explain the differentiation trend of the Oto-Zan magmas. Instead, mixing between an HMA magma and a highly differentiated rhyolitic magma (star) can reproduce the Oto-Zan trend. However, a felsic magma with a much lower concentration of K2O, which can be produced in the presence of biotite, is required for the end-member component (see text). TH, tholeiitic series; CA, calc-alkalic series (Miyashiro, 1974Go); FeO*, total iron as FeO.

 
The experimental product after a run at 900°C in the presence of 0·7 wt % H2O contains only a small amount of glass, suggesting that this temperature is close to the solidus (Fig. 11). At this temperature, amphibole and biotite are also observed in the run products, suggesting that the subsolidus mineral assemblage includes amphibole and biotite in addition to plagioclase, Fe–Ti oxide and pyroxene. Although the modal abundance of these phases could not be determined from the present experiments, the total amount of H2O held in the subsolidus phases can be constrained to less than 1 wt %. It is worth pointing out that even if the subsolidus modal assemblage were 100% hornblende, it would still only contain ~2 wt % H2O. As the experimental starting material has a composition consistent with a mantle-derived HMA that originally contains ~7 wt % H2O (Tatsumi, 1982Go), this result indicates that H2O can be effectively extracted from a mantle-derived hydrous HMA magma through solidification.


    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 GEOLOGICAL SETTINGS
 ANALYTICAL AND EXPERIMENTAL...
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The reasons for interest in the origin of the Oto-Zan composite lava flow are twofold. First, there is a paradox concerning the origin of Setouchi HMA magmas; andesites forming the composite lava flows are poor in both phenocrysts and H2O, although the HMA compositions found at the base of the lava flows may have originally contained a large amount of H2O during their final equilibration in the uppermost mantle. Second, HMAs in the Setouchi volcanic belt often form composite lava flows and dykes, suggesting that some particular processes may have contributed to the formation of Setouchi HMAs.

Mantle-derived hydrous HMA magma
HMA at the base of the Oto-Zan lava flow
Andesites comprising the Oto-Zan lava flow possess FeO*/MgO ratios smaller than unity (Fig. 7a and Table 7) and are referred to as HMA magmas. In particular, those forming the basal part of the lava flow (e.g. OTO-1) contain Mg-rich olivine (up to Mg-number of 90) and Cr-rich chromite (Cr/(Cr + Al + Fe3+) > 0·7; Figs 4 and 5, and Tables 3 and 4), which typify HMAs in the Setouchi volcanic belt and other subduction zones (e.g. Tatsumi & Ishizaka, 1981Go, 1982aGo, 1982bGo). However, the relationship between Mg-number and NiO content in olivine from OTO-1 (Fig. 4) suggests that this HMA is differentiated, as the olivine that is in equilibrium with the bulk OTO-1 is less nickeliferous and less magnesian than the most Ni- and Mg-rich olivine phenocrysts in the sample.

Two HMAs (TGI from the Osaka region of the Setouchi volcanic belt and SD-261 on Shodo-Shima Island; Table 7) have been proposed as least-differentiated, near primary andesite magmas that were produced in the upper mantle leaving harzburgitic and lherzolitic residues, respectively, based on both petrographic examination and melting phase relations at high pressures (Tatsumi, 1981Go, 1982Go; Tatsumi & Ishizaka, 1981Go, 1982aGo, 1982bGo). Although the composition of the HMA OTO-1 may be modified a little by olivine fractionation, as mentioned above, the compositional similarity between those HMAs and OTO-1 suggests that the parental magma for the Oto-Zan lava flow was andesitic and possessed a composition close to the OTO-1 HMA. Furthermore, the Oto-Zan HMAs have incompatible trace element and REE characteristics identical to those of HMAs in other regions of the Setouchi volcanic belt (Fig. 7b and c), providing compelling evidence that the Oto-Zan HMA magma was produced in the upper mantle.

Although experimental results indicate that Setouchi HMA magmas can equilibrate with mantle peridotites solely under hydrous conditions, this does not necessarily require the direct addition of slab-derived aqueous fluids to the mantle wedge. Many researchers (e.g. Pearce et al., 1992Go; Yogodzinski et al., 1994Go; Kelemen, 1995Go) have favoured a mechanism for HMA magma production that involves partial melting of the subducting lithosphere and interaction of these slab-derived, hydrous, silicic melts with the overlying mantle wedge peridotite, as originally proposed by Kay (1978)Go. The geochemical characteristics of Setouchi HMAs can be quantitatively explained by this mechanism (Tatsumi, 2001Go; Hanyu et al., 2002Go; Tatsumi & Hanyu, 2003Go).

If such a mechanism was the case for generation of Oto-Zan and other Setouchi HMA magmas, then the Oto-Zan HMA magma must have contained a large amount of H2O when it was in final equilibrium with mantle peridotite. Experimental results for Setouchi HMAs under H2O-undersaturated conditions (Tatsumi, 1981Go, 1982Go) suggest that ~7 wt % H2O in the magma is needed for equilibration at 1·0 GPa—a pressure equivalent to the depth immediately below the Moho beneath the Setouchi volcanic belt. It should be stressed, however, that the Oto-Zan HMA, as well as other Setouchi HMAs, contain little H2O as supported by the present XRF data (Table 7); a HMA from the Oto-Zan lava flow (SD-249) contains 0·89 wt % H2O(+) (Tatsumi & Ishizaka, 1982aGo), much smaller than boninites (2–5 wt %; Dobson & O'Neil, 1987Go; Sobolev & Danyushevsky, 1994Go). Boninites are also distinct from Setouchi HMAs in containing much larger amounts of phenocrysts, up to ~40% (e.g. Umino & Kushiro, 1989Go), which is consistent with extensive crystallization caused by H2O-oversaturation during magmatic ascent. It may, thus, be suggested that some particular processes may operate for producing phenocryst-poor and H2O-poor Setouchi HMAs that are originally rich in H2O and must become saturated with H2O within the crust.

Source heterogeneity
The Sr–Nd–Pb isotopic compositions of the Oto-Zan lava samples are clearly different from those of other HMAs from Mito Peninsula on Shodo-Shima island (Figs 1 and 8). This is also the case for Hf isotopes (Hanyu et al., 2002Go); an Oto-Zan HMA sample (SD-249) possesses an {varepsilon}Hf of 6·5, significantly lower than other Shodo-Shima HMAs (7·6–8·9). One possible explanation for the difference in isotopic composition, especially in terms of Sr–Nd isotopes, would be contamination by crustal materials, as basement rocks, including granite, gneiss and gabbro, have much higher 87Sr/86Sr and lower 143Nd/144Nd than the HMAs (Fig. 8). The presence of xenocrystic minerals such as quartz and Na-rich plagioclase in the Oto-Zan lava flow (Table 2) may be consistent with this mechanism. However, Pb isotopic compositions for the Shodo-Shima basement rocks do not support the involvement of crustal materials in the magmatic differentiation of the Oto-Zan sanukitoid magma (Fig. 8). Therefore, source heterogeneity beneath Shodo-Shima Island in terms of Sr–Nd–Pb isotopic composition, is suggested to be responsible for the observed isotopic spatial variations.

There is additional evidence suggesting that the HMA magma source beneath Shodo-Shima Island is heterogeneous. It is well established that olivine in the upper mantle possesses rather constant NiO contents of 0·35–0·5 wt %, whereas its Mg-number can be variable (Fig. 4; Sato, 1977Go; Takahashi, 1990Go). Some olivines that crystallized from primitive HMAs in the Setouchi volcanic belt, including those on Shodo-Shima Island, have Mg-number and NiO compositions consistent with a mantle signature (Tatsumi & Ishizaka, 1981Go, 1982aGo, 1982bGo; Tatsumi et al., 2003Go). Olivine phenocrysts embedded in the Oto-Zan HMA (OTO-1) are characterized by significantly higher NiO contents, up to 0·6 wt % at Mg-number of ~90 (Fig. 4), suggesting the presence of unusually nickeliferous olivine in its source.

The depleted chemical signature, in terms of high Ni contents in olivine, for the Oto-Zan HMA source is reinforced by the composition of spinel inclusions in the olivine. Two types of HMAs are recognized in the Setouchi volcanic belt (Tatsumi, 1982Go): one with a crystallization sequence of olivine -> orthopyroxene (opx-HMA) and the other with olivine -> clinopyroxene (cpx-HMA). The results of high-pressure melting experiments (Tatsumi, 1981Go, 1982Go) indicate that harzburgitic and lherzolitic peridotites are the melting residues for opx- and cpx-HMA magmas, respectively, leading Tatsumi (1982)Go to the conclusion that the opx-HMA magmas are produced either by higher degrees of partial melting or from a relatively depleted mantle peridotite source compared with the cpx-HMA magmas. This is consistent with the difference observed in the spinel compositions, with higher Cr/(Cr + Al) for the opx-HMA magmas (Fig. 5a). Although the Oto-Zan HMA is classified, in terms of its phenocryst assemblage and crystallization sequence, as a cpx-HMA, its spinel inclusions have compositions identical to those in opx-HMAs (Fig. 5a and f).

Magmatic differentiation
Fractional crystallization and assimilation
Continuous sampling and analysis of the Oto-Zan lava flow clearly document a variation in mineralogical and chemical compositions within the single lava flow as a function of height from the base (Figs 3 and 9). It may thus be suggested that magmas having different chemistry and mineralogy have contributed to the lava formation.

Andesites forming the lower part of the lava flow are more depleted in Si, Na, K, Rb and Zr, and more enriched in Fe, Mg, Ca and Ni than those from the upper part (Fig. 9). This, together with constant Nd–Pb isotopic compositions throughout the lava flow (Fig. 10) would suggest the derivation of more felsic magmas through fractional crystallization of a hydrous HMA magma such as OTO-1, which occurs at the base of the lava flow. Grove et al. (2003)Go determined the phase relations of an HMA from the Mt Shasta region over a range of pressure and temperature conditions under H2O-saturated conditions and demonstrated the liquid line of descent through fractional crystallization of such a hydrous HMA magma (Fig. 7a). Their results may be applicable for examining possible magmatic differentiation paths for the Oto-Zan HMA magma. The reasons for believing so are twofold. First, the experimental starting composition was similar to the Oto-Zan HMA composition. Second, the original H2O content of the Setouchi HMAs (~7 wt %) would cause H2O-oversaturation in a shallow crustal magma reservoir, which is the condition reproduced by the Grove et al. (2003)Go experiments. A marked increase in FeO*/MgO with increasing SiO2 content in the differentiated melt was observed in their crystallization experiments (Fig. 7a), which is a trend quite different from that of the Oto-Zan magma differentiation.

In order to further examine the crystallization differentiation process, a simple mass-balance calculation was made assuming an initial composition equal to the HMA OTO-1 and a final composition equal to one of the most differentiated samples within the lava flow (OTO-9). The results indicate that fractionation of plagioclase and magnetite is needed to produce the differentiated magma from the parental HMA (Table 11). However, neither plagioclase nor magnetite is present as a phenocryst phase in the Oto-Zan samples. Petrographic evidence also does not support this simple process. The presence of both normally zoned and reversely zoned pyroxene phenocrysts in a single sample, especially in differentiated rocks, indicates an open-system differentiation process such as mixing of two compositionally different magmas. It may thus be concluded that a fractional crystallization process was not responsible for differentiation of the Oto-Zan magma.


View this table:
[in this window]
[in a new window]
 
Table 11: Results of mass-balance calculation

 
Simultaneous assimilation of crustal materials and fractional crystallization (e.g. DePaolo, 1981Go; Hildreth & Moorbath, 1988Go; Dias & Leterrier, 1994Go) is another possibility for producing differentiated magmas from an HMA magma in the Oto-Zan lava flow, because plagioclase and quartz, which are major constituent minerals in basement granitic rocks, occur ubiquitously, although not voluminously, as xenocrysts in the lava flow (Table 2). A simple mass-balance calculation based on compositions of Oto-Zan lava samples, phenocryst phases and a granite from Shodo-Shima Island (Tatsumi et al., 2002Go) yields results consistent with assimilation coupled with fractional crystallization (Table 11). However, the isotopic compositions of the Oto-Zan samples and the basement rocks, especially for Pb, do not support this (Fig. 8); a contaminant having a much higher 207Pb/204Pb and 208Pb/204Pb composition than the Shodo-Shima basement rocks is required for the assimilation process, and there is no evidence for such materials.

Magma mixing
Many calc-alkalic andesites are characterized by the following petrographic and chemical signatures: (1) the presence of reversely zoned pyroxene phenocrysts with rounded cores mantled by rims with higher Mg-number; (2) the presence of plagioclase phenocrysts with a wide range of compositions and with a dusty zone containing fine melt inclusions; (3) the presence of disequilibrium mineral assemblages such as Mg-rich olivine and quartz; (4) linear chemical trends typically demonstrated for MgO and FeO against SiO2. These observations led several authors (e.g. Eichelberger, 1975Go; Sakuyama, 1979Go, 1981Go; Cribb & Barton, 1997Go; Hunter, 1998Go; Rorolo & Castorina, 1998Go; Feeley et al., 2002Go; Tatsumi et al., 2002Go) to the conclusion that mixing of mafic and felsic magmas plays a significant role in the production of calc-alkalic andesite magmas. Orthopyroxene-bearing andesites from the Oto-Zan lava flow, i.e. samples from 19 m above the flow base, contain ortho- and clino-pyroxenes that exhibit both normal and reverse zoning (Fig. 6, Tables 5 and 6). Furthermore, the differentiation trends observed for this lava flow are generally linear (Fig. 13). It may thus be suggested that mixing of compositionally different magmas may have played a role in the evolution of the Oto-Zan magma.


Figure 13
View larger version (17K):
[in this window]
[in a new window]
 
Fig. 13. Chemical differentiation trends for the Oto-Zan sanukitoid samples. The highly linear trends observed can be explained by mixing of two magmas: one having a composition close to the least differentiated Oto-Zan HMA and the other with a rhyolitic composition (open circle). Rhyolites on Shodo-Shima Island, also belonging to the Setouchi volcanic rocks, are not a suitable candidate for the felsic end-member, as the rhyolite compositions do not match those of the inferred rhyolite (see text).

 
A possible, and the most plausible, candidate for the mafic end-member magma would be the HMA OTO-1 sample that forms the basal part of the lava flow. Rhyolites do exist on Shodo-Shima Island and could represent a possible felsic end-member. The compositions of such rhyolites are shown in Fig. 13, together with those of the Oto-Zan lava flow. Linear chemical trends observed for Oto-Zan sanukitoids, such as those for SiO2 vs MgO, FeO and Th, can be reasonably explained by the contribution of a Shodo-Shima rhyolite magma as the felsic end-member in a magma mixing process. However, chemical trends for other elements do not support this (Fig. 13).

Alternatively, the relatively constant Sr–Nd–Pb isotopic compositions throughout the lava flow would be consistent with derivation of the felsic end-member magma from the parental Oto-Zan HMA magma, which can be referred as to ‘internal mixing’ or ‘back mixing’. Formation processes of such a differentiated felsic magma are hereafter discussed based on the present experimental results.

Although the compositional trends of melts produced at high temperatures, especially those for TiO2 and Al2O3, are governed by the amount of H2O in the system, the composition of melts produced at lower temperature (~1000°C) are broadly similar (Fig. 12 and Table 10). This may be attributed to H2O-saturation in such small-volume, low-temperature, rhyolitic melts, even with different H2O contents in the overall system, because 4–7 wt % H2O in a rhyolitic melt (0·7, 1·5 and 2·1 wt % H2O in the system with 90, 60 and 50% solids and 10, 40 and 50% melt, respectively) may be close to or above the solubility limit of H2O (Burnham & Nekcasil, 1986Go; Moore et al., 1998Go; Yamashita, 1999Go). It thus seems reasonable to assume a melt with a composition that plots on an extended portion of the liquid line of descent as a possible felsic melt derived from the Oto-Zan HMA magma (Fig. 12). If we accept this as the end-member felsic magma, then the differentiation trends of the Oto-Zan andesites, including the unusually linear trend for MgO and the decrease in TiO2 with increasing SiO2, can be successfully explained by mixing of the basal HMA magma with this felsic magma (Fig. 12). However, the behaviour of alkali elements, especially K2O, is less clear; a felsic magma with a lower concentration