JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 PAGES 1083^1128 2014 doi:10.1093/petrology/egu020 Progressive Interaction between Dry and Wet Mantle during High-temperature Diapiric Upwelling: Constraints from Cenozoic Kita-Matsuura Intraplate Basalt Province, Northwestern Kyushu, Japan TETSUYA SAKUYAMA1*, SHUN’ICHI NAKAI2, MASAKO YOSHIKAWA3, TOMOYUKI SHIBATA3 AND KAZUHITO OZAWA4 1 INSTITUTE FOR RESEARCH ON EARTH EVOLUTION, JAPAN AGENCY FOR MARINE^EARTH SCIENCE AND TECHNOLOGY, YOKOSUKA 237-0061, JAPAN 2 EARTHQUAKE RESEARCH INSTITUTE, UNIVERSITY OF TOKYO, TOKYO 113-0032, JAPAN 3 INSTITUTE FOR GEOTHERMAL SCIENCES, UNIVERSITY OF KYOTO, NOGUCHIHARA, BEPPU CITY, OITA PREFECTURE 874-0903, JAPAN 4 DEPARTMENT OF EARTH AND PLANETARY SCIENCE, GRADUATE SCHOOL OF SCIENCE, UNIVERSITY OF TOKYO, TOKYO 113-0033, JAPAN RECEIVED DECEMBER 25, 2013; ACCEPTED MARCH 31, 2014 Intra-plate Cenozoic volcanism in Kita-Matsuura, northwestern Kyushu, Japan, shows systematic spatio-temporal changes in geochemistry that can be explained by partial melting followed by melt segregation in a region of upwelling mantle. We have examined the thermal and melting history of the upwelling mantle by quantitatively estimating melt water contents and melting conditions. The water content of a spectrum of primary melts is estimated to range from 0·5 to 1·5 wt % based on a combination of a plagioclaseliquid and olivine-saturated liquid geohygrometers and MELTS calculations. The estimated melt segregation temperature ranges from 1330 to 15008C, at pressures from 1·7 to 2·8 GPa under hydrous conditions. Melting temperature and pressure decreased with time, whereas the water content of the primary melts increased. Corresponding temporal decreases in high field strength element (HFSE) abundances and HFSE/large ion lithophile element (LILE) ratios require progressive melt extraction and aggregation KEY WORDS: Kita-Matsuura; Japan; intraplate back-arc volcanism; open-system melting *Corresponding author. Telephone: þ81-46-867-9785. þ81-46-86-9625. E-mail: [email protected] ß The Author 2014. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@ oup.com Fax: from a melting mantle with a continuous and gradually increasing input of H2O-rich fluid or melt into the melting system. The estimated isotope composition of influxed fluid lies on a mixing line between the sediment and altered oceanic crust of the Philippine Sea Plate, with strong affinity to the sediment composition. Based on the temporal variation of the magmas and the melting model, we propose small-scale upwelling (c. 70 km in diameter) of a dry mantle peridotite that interacts progressively with the overlying wet mantle wedge. The wet mantle wedge was previously hydrated by fluids from sediments from the subducted Philippine Sea Plate, whereas the deep and dry mantle could have been derived from the mantle beneath the subducted Pacific Plate through a slab window. JOURNAL OF PETROLOGY VOLUME 55 I N T RO D U C T I O N Cenozoic back-arc volcanism has occurred extensively, but sporadically, along 1000 km of the eastern margin of the Eurasian Plate. The back-arc region, where the subducting slab surface is deeper than 300 km, is characterized by high surface heat flow (Currie & Hyndman, 2006), upper mantle with slow seismic velocity (e.g. Fukao et al., 1992) and high electrical conductivity (Ichiki et al., 2006). Various mantle upwelling models have been postulated to explain the association of these geophysical characteristics with the volcanic activity. These include the following: (1) hot mantle plumes originating from the upper mantle^ lower mantle or core^mantle boundary (Nakamura et al., 1990; Zeng et al., 2011); (2) upwelling of hydrous hot mantle plumes originating from the 410 km discontinuity (Zhao et al., 2007; Kuritani et al., 2009); (3) wet and cold plume upwelling (Richard & Iwamori, 2010); (4) a moving hot region distinct from a hotspot (Miyashiro, 1986); (5) asthenospheric upwelling induced by thickening of a stagnant slab (Zou et al., 2008); (6) lateral mantle extrusion induced by the Indo-Asian continental collision (Liu et al., 2004). A role for mantle upwelling in causing back-arc volcanism, as invoked in these models for eastern Asia, has also been suggested for other areas where backarc volcanism occurs, such as New Zealand (Rafferty & Heming, 1979) and Patagonia (Ignacio et al., 2001). To identify the actual upwelling process among the diverse models that have been proposed, the spatial and temporal relationships between the various processes causing volcanism in each back-arc region need to be quantified. Back-arc volcanoes that occur on continents, especially if they are extinct and young, have an advantage over oceanic back-arc volcanoes, as their entire volcanic sequence can be exposed in gullies formed by erosion of the flanks, and thus can be much more easily sampled at a high temporal resolution. Moreover, continental back-arc volcanoes generally show primitive petrological characteristics, and although the volcanic activity is often spread over a wide area at different eruptive centres, each volcano has a limited extent. As a result, petrological and geochemical data for volcanic rocks in this tectonic setting that are well constrained spatially and temporally can provide spatial and temporal information on mantle upwelling and melting, as long as the effects of crustal processes can be removed. However, only a few studies have exploited this advantage of continental back-arc volcanism (e.g. Iwamori, 1991; Sakuyama et al., 2009). Sakuyama et al. (2009) made systematic geological, geochemical, and chronological investigations of Cenozoic intraplate volcanism in the Kita-Matsuura area, southwestern Japan, which they interpreted in terms of mantle upwelling beneath this region. Their model was, however, far from quantitative as they assumed anhydrous melting without hydrous material input, which caused difficulties NUMBER 6 JUNE 2014 in the estimation of mantle potential temperature and the explanation of the observed fractionation between fluidmobile and fluid-immobile elements. In this study, new mineral chemical data for plagioclase phenocrysts and spinel in the basalts, whole-rock trace element data measured by inductively coupled plasma mass spectrometry (ICP-MS), and whole-rock Sr^Nd^Pb isotope data measured by thermal ionization mass spectrometry (TIMS) are presented. These data are combined with the previous results of Sakuyama et al. (2009) to estimate quantitatively the temporal and spatial changes in melting conditions, including water content, pressure, and temperature. The results successfully place stricter constraints on the melting process and allow us to accurately constrain the mantle potential temperature, which has important geodynamic implications for upwelling in the back-arc region of Kyushu, southwestern Japan. G EOLO GY A N D P R EV I O U S ST U D I E S OF T H E K I TA- M AT S U U R A B A S A LT S Northwestern Kyushu is located between two Cenozoic marginal basins: the Japan Sea and Okinawa Trough. The Pacific Plate is estimated to have been subducting beneath northeastern Asia since 60 Ma (Mu«ller et al., 2008), followed by initiation of subduction of the Philippine Sea Plate at the time of initiation of Japan Sea opening at 25 Ma (Kimura et al., 2005). Since the opening of the Japan Sea between 25 and 15 Ma (Tamaki et al., 1992), the Philippine Sea Plate has been subducting northwestward under southwestern Japan. North of this area the Pacific Plate is subducting westward, and the top of the slab is at a depth of 400^600 km beneath southwestern Japan. The subduction angle of the younger part of the Philippine Sea Plate is shallow (128) beneath the Chugoku area, whereas that of the older part of the Philippine Sea Plate is steep (608) beneath Kyushu. Opening of the Okinawa Trough initiated in Late Miocene times and its northern extension continues to the Beppu-Shimabara graben in western Kyushu (Letouzey & Kimura, 1985), in which the active Unzen volcano is located. The most voluminous Cenozoic basaltic volcanism in southwestern Japan occurred at the end of the Miocene in the Kita-Matsuura area of northwestern Kyushu associated with a compressional stress field (Sakuyama, 2010). The Kita-Matsuura area is located 150 km to the backarc side of the current volcanic front in Kyushu, which is 400 km from the trench. This basaltic volcanism began after a dormant period of 5 Myr following basaltic and andesitic volcanism on Hirado island at 15 Ma during the Middle Miocene when the Japan Sea stopped opening (Jolivet et al., 1994). 1084 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Sakuyama et al. (2009) investigated four volcanic sections that are inferred to have originated from single volcanic centers and together encompass the entire KitaMatsuura basalt succession (from west to east the Hirado, Senryu, Ishimori, and Kunimi sections; Fig. 1). Volcanic activity started at 9·5 Ma in the Hirado section and continued intermittently for 2 Myr in all of the sections (Fig. 2a). In addition to the Kita-Matsuura area, Cenozoic basaltic volcanism also occurred on several of the islands to the west and north of the Hirado section: Takushima, Ikitsuki, and Azuchi-Oshima (Fig. 1a). The Takushima and Ikitsuki basalts range in age from 9·5 to 8·7 Ma and from 8·5 to 6·5 Ma, respectively, and are similar to the basalts in the Kunimi section in terms of their high contents of Na2O þ K2O and low MgO at SiO2 ¼50 wt % (Uto et al., 2004), which suggest that they are cogenetic with the Kita-Matsuura basalts. Below, we summarize the results of Sakuyama et al. (2009) upon which this study builds. Spatial and temporal variations in the Kita-Matsuura basalts After evaluating the effects of crustal processes (magma mixing, crustal assimilation, crystal fractionation), Sakuyama et al. (2009) showed that the diversity of basalt chemistry can be represented by three liquid lines of descent with distinct trends on oxide^oxide variation diagrams: low- (47^50 wt %), medium- (49^52 wt %), and high-SiO2 (51^54 wt %) groups. The groups can also be defined based on their incompatible element ratios, whereby Nb/Y, Zr/Y, and Nb/Th decrease from the lowto high-SiO2 groups (Fig. 2b and c). Most of the samples in the Hirado and Senryu section are classified as belonging to the medium- and high-SiO2 groups (Fig. 2a), whereas low-SiO2 group lava is more common in the eastern sections, such as at Ishimori and Kunimi. In the Hirado, Senryu, and Ishimori sections, lavas sequentially change their composition to the higher-SiO2 group (Figs 1b^d and 2a); in particular, from the medium- to high-SiO2 group in Hirado, from the low-, through medium-, to high-SiO2 group in Senryu, and from the low- to high-SiO2 group in Ishimori. Mantle upwelling model and problems that need to be solved Fig. 1. (a) Simplified map around Kyushu, southwestern Japan. Bold continuous curves represent plate boundaries. Okinawa Trough and Beppu-Shimabara Graben are indicated by bold dotted curves. (b) Distribution of the Kita-Matsuura basalts in northern Kyushu. (c) Geological cross-sections through the Kita-Matsuura basalts along the lines A^B, C^D, E^F, and F^G shown in (b). The vertical scale is twice the horizontal scale. The four cross-sections pass through the four areas studied: Hirado, Senryu, Ishimori, and Kunimi on the A^B, C^D, E^F, and F^G sections, respectively. These are used to clarify the simplified sequence of magmatic groups Sakuyama et al. (2009) found that the melt segregation pressure decreases from 3 GPa for the low- to 1·5 GPa Fig. 1 Continued after Sakuyama et al. (2009). Low^medium- and medium^high-SiO2 groups, which are samples classified into both a low- and mediumSiO2 group and a medium- and high-SiO2 group, respectively, are omitted from this figure for simplicity [see Sakuyama et al. (2009) for the detailed magmatic sequence]. masl, meters above sea level; filled triangle represents 0 masl. 1085 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 for the high-SiO2 group under anhydrous conditions. Following Sobolev & Shimizu (1993) decreases of Nb/Y and Zr/Y from the low-, through medium-, to high-SiO2 groups (Fig. 2b and c) were attributed to an increase of melting degree by progressive near-fractional melting of a primitive or enriched mantle source by applying a melting model open to only output; the temporal decrease in melting pressure and increase of melting degree in every section originated in a mantle upwelling with progressive melting and melt segregation. However, several critical issues were not fully addressed by Sakuyama et al. (2009), including the following: (1) the effect of water in mantle melting was not examined; (2) only the ranges of primary melts were estimated because of uncertainty in the source mantle composition in terms of olivine forsterite content [Fo# ¼ 100 Mg/ (Mg þ Fe2þ)]; (3) melting pressure was only approximately constrained because of the uncertainty in the primary melt compositions; (4) melting temperature (or mantle potential temperature) was not estimated; (5) only fluidimmobile high field strength elements (HFSE) were used in trace element modelling and fluid-mobile elements were not considered. To resolve issues (1), (2), (3), and (4), we have obtained new mineral chemical data; to address issue (5), we obtained a highly accurate trace element and isotope dataset by ICP-MS and TIMS. A N A LY T I C A L M E T H O D S We selected 15 samples from the four stratigraphic sections through the Kita-Matsuura basalts for trace element and isotope analysis: three samples were selected from each of the Kunimi, Ishimori, and Hirado sections and six samples were selected from the Senryu section. Samples were selected to include at least the initial, middle, and terminal basaltic activities in each section. Olivine phenocrysts or microphenocrysts in all the samples selected for analysis were fresh without iddingsite in the core. Totals of 10 major elements determined by X-ray fluorescence (XRF) for all of the samples were 498 wt %, which is suggestive of a minimum effect of subaerial modification on bulkrock chemical compositions. Whole-rock major element contents for most samples in this study were taken from Sakuyama et al. (2009), except for samples 03101930, 03102481, 031027c9, and 031026a8, which were newly analysed by XRF using a Philips PW-1480 system at the Department of Earth and Planetary Fig. 2. Summary of the temporal and spatial variations in the KitaMatsuura basalts based on Sakuyama et al. (2009) and Uto et al. (2004). Fig. 2 Continued (a) Temporal changes of activity in the studied sections according to Sakuyama et al. (2009). Data for Ikitsuki and Takushima are from Uto et al. (2004). Range of eruption ages are indicated by filled bars. (b) Relationship between estimated ranges of segregation pressure and Nb/Y ratio for each group of basalts. (c) Zr/Y vs Nb/Y for each basalt group after Sakuyama et al. (2009). 1086 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Science, University of Tokyo. Rock samples were sliced into chips using a diamond cutter, polished with a diamond, and powdered in an agate mortar. Detailed preparation and analytical methods for major element composition have been described by Sakuyama et al. (2009). Twenty-eight trace elements [Rb, Ba, Th, U, Pb, Ta, Nb, Sr, Zr, Hf, Y, and the rare earth elements (REE)] were determined for 15 samples of the Kita-Matsuura basalt by ICP-MS at the Earthquake Research Institute, the University of Tokyo. Selected rock powders were weighed into Savillex Teflon vials with hydrofluoric acid and perchloric acid. The vials were heated to about 1808C for 2 days. The decomposed samples were then evaporated. The dried samples were dissolved in 20% nitric acid to ensure that there was no precipitation. A part of the solution was diluted in 1·5% nitric acid such that the final dilution factor was 6000. Indium and bismuth were added as internal standards to correct for machine drift and matrix effects. The reproducibility of the analyses for all the elements was better than 5% repeated standard deviation (RSD). Our abundance results on the GSJ standard basalt JB-1a agreed with those of the reference values to within 10%. The results are listed in Table 1. We checked the consistency between the ICP-MS analyses and those determined by XRF by Sakuyama et al. (2009) (see Appendix). Mineral compositions of plagioclase, spinel, and olivine were analysed by electron microprobe (JEOL JCMA733MKII) at the University of Tokyo. The analytical procedures have been described by Nakamura & Kushiro (1970), and used the Bence & Albee (1968) correction method. Accelerating voltage, beam current, and counting time were 15 kV, 12 nA, and 30 s, respectively. The Fo# and NiO content of olivine were determined at 25 kV with a 20 nA beam current and 30 s counting time using a ZAF correction. Rock samples for Sr^Nd^Pb isotope analysis were sliced into chips using a diamond saw, polished with abrasive powder to remove the tracks of the cutter, and washed in an ultrasonic bath. After the chips were crushed to a few millimeter-sized fragments in a tungsten carbide mortar, the fragments were ground in an automatic agate mortar for 2 h. The Sr, Nd, and Pb isotopic compositions of the samples were determined by TIMS at the Institute for Geothermal Sciences, Kyoto University. Details of the analytical procedure for chemical separation and mass spectrometry for Sr, Nd, and Pb isotope determination have been given byYoshikawa et al. (2001) and Shibata et al. (2003). W H O L E - RO C K T R AC E E L E M E N T A N D I S OTOP E A N D M I N E R A L C H E M I C A L DATA Whole-rock trace element composition Figure 3a^d shows trace element patterns normalized to normal mid-ocean ridge basalt (N-MORB) (Pearce & Parkinson, 1993) for the four sections of the KitaMatsuura basalt that were sampled. The three basalt groups (high-, medium-, and low-SiO2 groups) are labeled in each section. The REE patterns are smooth and show enrichment in light REE (LREE) with significant variations in the overall slope and the LREE abundances. There is a smooth decrease from Ba to U, and the relative normalized abundances of Ba, Th, and U are similar (Fig. 3a^d). No basalt sample has an Eu anomaly. Positive Ba, Pb, and Ti anomalies and weak positive Sr anomalies are generally observed. As the concentrations of Nb decrease, the Pb enrichment relative to Ce or Pr and the Nb depletion relative to Th or K increase (Fig. 3e). Generally, the abundances of moderately to highly incompatible trace elements in the eastern sections (Kunimi and Ishimori) are higher than those in the western sections (Senryu and Hirado). The abundance of incompatible trace elements and Nb/ Th and Nb/Y decrease with decreasing age in each section (Sakuyama et al., 2009), although they show relatively wide scatter over a small timescale (less than a few million years). Both the REE abundance and slope of REE normalized to N-MORB decrease with time (Fig. 3a^d). The latter is clearly shown by the temporal variations of Ce/ Yb and Sm/Yb, which decrease with time in each section (Fig. 4c and d). The abundance of incompatible trace elements, and the middle REE (MREE)/heavy REE(HREE) and the LREE/HREE ratios are intimately related to the SiO2 grouping. Weak negative Nb and Ta anomalies occur in all the high-SiO2 group basalts and in some of the medium-SiO2 group basalts, but not in the low-SiO2 group basalts (Fig. 3a^d). The abundance of incompatible trace elements and MREE/HREE, LREE/HREE, Zr/Y, Nb/Th, and Nb/Ce decrease from the low- to high-SiO2 groups (Figs 3 and 4). Chemical compositions of spinel inclusions in olivine phenocrysts and plagioclase phenocrysts or microphenocrysts Olivine phenocrysts in less fractionated basalts (FeO*/ MgO 51·5) contain spinel inclusions 10^30 mm in diameter, although spinel microphenocrysts are rare. Representative major element compositions of spinel inclusions in the least differentiated samples for each SiO2 group are given in Table 2. The spinel Cr# [100 Cr/ (Cr þAl)] ranges between 40 and 60 (Fig. 5a) and the YFe [100 Fe3þ/(Cr þAl þ Fe3þ)] averages 15, except for spinel in Fe-rich olivine (Fig. 5b). As the Fo# of the host olivine decreases, the Cr# of the spinel inclusions is almost constant for the low-SiO2 group and decreases slightly for the high-SiO2 group (Fig. 5a). This relationship between the Cr# of spinel inclusions and the Fo# of host olivine is commonly observed in spinel inclusions in olivine 1087 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Table 1: Major and trace element concentrations for representative Kita-Matsuura basalts, standards and estimated primary melts Section: Hirado Hirado Hirado Senryu Senryu Senryu Senryu Senryu Senryu Ishimori Ishimori Ishimori Basalt group: high high medium high high high medium medium low high high low low 8·1 8·84 6·37 6·48 6·54 6·56 6·59 7·68 5·42 7·33 7·74 7·11 K–Ar age1 (Ma): 7·29 Name: Kunimi 03101930 03101924 03101817 02110723 02110934 02111276 02111173b 03102155 03102145 031029if7 03102481 03102475 031027e4 Major elements (wt %) SiO2 50·86 52·80 50·07 50·83 49·31 50·75 47·47 49·45 46·44 51·41 51·16 48·83 TiO2 1·56 1·32 1·55 1·16 1·20 1·54 1·46 1·91 2·17 1·20 1·47 2·79 1·82 Al2O3 14·85 15·32 17·97 14·85 14·08 16·48 13·96 16·13 17·13 15·73 15·43 16·74 13·84 Fe2O3* 11·3 11·34 46·46 12·38 10·45 9·13 11·73 10·54 10·08 10·94 9·87 8·82 10·39 12·46 MnO 0·17 0·15 0·16 0·16 0·17 0·16 0·17 0·16 0·16 0·16 0·15 0·14 0·17 MgO 7·38 7·08 6·59 9·0 10·42 6·65 9·62 7·08 7·64 7·63 7·67 4·48 10·81 CaO 9·12 8·23 8·64 8·80 9·10 9·77 10·91 9·79 10·09 9·77 8·36 7·42 11·48 Na2O 2·92 2·70 2·87 2·68 2·39 3·07 2·46 2·74 2·17 2·60 2·81 3·07 1·93 K2O 0·73 1·02 1·64 0·53 0·72 0·81 0·79 1·28 1·15 1·01 1·09 1·86 0·96 P2O5 0·22 0·24 0·41 0·15 0·21 0·31 0·27 0·41 0·41 0·25 0·36 0·51 0·27 Total 99·88 99·31 99·04 99·90 98·13 99·61 98·07 98·82 98·66 98·60 98·88 98·33 99·06 Trace elements (ppm) Rb Ba 16·6 145 26·0 226 42·2 471 12·7 132 15·4 233 11·4 335 16·1 373 27·3 404 23·6 1000 22·7 345 24·1 300 51·4 30·3 484 418 Th 2·19 3·33 6·03 1·75 2·55 3·09 3·92 4·05 3·60 3·35 3·52 6·44 5·46 U 0·41 0·72 0·87 0·38 0·56 0·63 0·81 0·91 0·71 0·69 0·76 1·44 1·07 Ta 0·46 0·81 1·12 0·34 0·58 0·97 0·77 1·51 1·62 0·65 1·12 2·65 Nb 7·73 Pb Sr Zr Hf Y 2·91 346 94·6 2·67 19·6 12·9 4·55 18·9 7·24 308 502 120 151 3·20 19·3 3·80 17·8 5·75 3·63 267 69·9 2·06 18·9 10·1 3·04 336 84·3 2·37 16·6 16·3 4·10 431 14·8 3·28 453 147·1 3·82 19·9 98·8 2·74 19·2 24·8 4·45 26·5 2·82 508 621 173 155 4·51 21·0 4·04 10·9 4·29 449 80·8 2·40 18·7 4·23 44·2 4·43 397 447 131 222 3·46 1·52 25·4 2·76 506 99·8 5·53 2·82 18·8 20·6 19·6 26·6 17·7 La 12·2 17·6 30·9 10·2 14·5 20·9 23·0 26·2 25·2 24·0 22·2 38·1 27·0 Ce 28·6 35·8 65·5 20·4 29·5 43·6 43·3 53·1 54·9 40·6 45·8 76·9 51·4 Pr Nd 3·81 16·5 4·56 18·5 7·48 28·3 2·94 12·7 3·57 14·8 5·25 21·3 5·04 20·3 6·46 26·1 6·63 26·8 5·02 20·3 5·40 21·9 9·03 35·4 5·90 23·6 Sm 4·10 4·37 5·11 3·38 3·48 4·76 4·44 5·67 5·56 4·32 4·86 7·25 4·84 Eu 1·39 1·44 1·63 1·21 1·18 1·62 1·48 1·88 1·83 1·49 1·67 2·28 1·61 Gd 4·55 4·56 4·39 4·05 3·75 4·81 4·55 5·45 5·21 4·41 4·97 6·90 4·59 Tb 0·70 0·70 0·63 0·65 0·57 0·72 0·67 0·80 0·74 0·65 0·72 1·02 0·67 Dy 4·19 4·20 3·73 3·95 3·47 4·23 4·03 4·59 4·29 3·94 4·29 5·87 3·90 Ho 0·83 0·82 0·73 0·79 0·69 0·82 0·79 0·88 0·82 0·80 0·85 1·13 0·75 Er 2·28 2·22 2·03 2·24 1·88 2·28 2·15 2·39 2·14 2·21 2·28 2·99 2·00 Tm 0·31 0·31 0·29 0·31 0·26 0·32 0·30 0·33 0·29 0·30 0·32 0·41 0·27 Yb 1·99 2·00 1·91 2·04 1·69 2·06 1·94 2·08 1·85 1·93 2·05 2·57 1·68 Lu 0·28 0·29 0·28 0·29 0·24 0·30 0·27 0·29 0·26 0·26 0·29 0·29 0·36 (continued) 1088 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Table 1: Continued Section: Kunimi Kunimi Basalt group: low low K–Ar age1 (Ma): 7·26 7·76 Name: 031027c9 031026a8 Standard Estimated primary melt Fo#: JB-1a JB-1a (this study) (reference value) low medium high 90 90 90 Major elements (wt %) SiO2 49·20 47·93 45·89 47·02 TiO2 2·29 3·13 1·56 1·24 49·01 1·07 Al2O3 16·96 16·79 11·89 11·88 12·59 Fe2O3* 10·89 10·71 13·00 11·56 11·38 MnO 0·15 0·17 0·16 0·17 0·17 MgO 6·08 4·72 16·36 16·02 15·16 CaO 9·17 7·12 9·87 9·28 8·14 Na2O 2·73 3·10 1·66 2·10 2·14 0·65 K2O 1·32 2·03 0·82 0·67 P2O5 0·46 0·61 0·23 0·23 Total 99·07 98·61 100 100 0·18 100 Trace elements (ppm) Rb Ba 30·2 464 49·7 618 35·7 39·2 461 504 26·4 262 13·6 316 10·1 106 Th 5·99 7·29 9·53 9·03 3·37 3·32 1·40 U 1·23 1·44 1·81 1·57 0·75 0·68 0·31 Ta 1·73 2·88 1·63 1·93 1·37 0·65 0·28 25·3 26·9 6·70 6·76 Nb Pb 27·7 3·85 48·5 6·03 22·8 2·34 Sr 603 520 423 442 288 Zr 153 246 128 144 118 3·52 3·41 Hf 3·90 6·10 3·05 12·5 2·78 384 83·9 2·32 4·6 2·91 220 56·1 1·66 Y 20·2 25·9 20·9 24·0 15·3 16·3 La 34·6 46·5 43·0 37·6 20·0 19·5 8·1 Ce 65·1 97·3 72·5 65·9 40·6 36·7 16·4 Pr Nd 7·46 29·1 11·06 9·2 7·3 43·0 29·3 26·0 4·84 19·2 4·27 17·2 15·3 2·36 10·3 Sm 5·77 8·29 5·95 5·07 4·03 3·76 2·73 Eu 1·93 2·60 1·47 1·46 1·36 1·25 0·99 Gd 5·34 7·27 5·30 4·67 3·85 3·86 3·28 Tb 0·76 1·04 0·67 0·69 0·57 0·57 0·53 Dy 4·39 5·80 5·37 3·99 3·30 3·42 3·20 Ho 0·84 1·09 0·80 0·71 0·63 0·67 0·64 Er 2·22 2·81 2·90 2·18 1·66 1·82 1·81 Tm 0·31 0·38 0·31 0·33 0·23 0·26 0·25 Yb 1·99 2·42 2·70 2·10 1·43 1·65 1·65 Lu 0·28 0·34 0·29 0·33 0·20 0·23 0·23 *Total iron given as Fe2O3. 1 K–Ar ages are the values that were calculated by regressing separately for each section to estimate ages for lava flows without age determination [see details given by Sakuyama et al. (2009)]. Major element compositions of all the samples in this table were previously reported by Sakuyama et al. (2009) except for 03101930, 03102481, 031027c9, and 031026a8, which were newly analysed for this study by the same analytical method as that of Sakuyama et al. (2009). Fo#, Fo content of equilibrium mantle olivine. 1089 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Representative major element compositions of plagioclase phenocrysts or microphenocrysts are given in Table 3. Plagioclase phenocrysts in the low- and mediumSiO2 groups are euhedral to subhedral and 0·5^3 mm in average size, reaching a maximum of 1cm. The anorthite (An) [100 Ca/(Ca þ Na)] content in the cores varies from 60 to 80, and in the rims from 40 to 70 (Fig. 6a and b). The range of An contents in the groundmass plagioclase is systematically lower than that of the phenocrysts (Fig. 6a and b).In relatively differentiated lava samples (bulk FeO*/MgO41·25) normally zoned plagioclase forms crystal aggregates with olivine and clinopyroxene. The olivine and clinopyroxene in these aggregates have cores with lower Fo# (570) and Mg# [100 Mg/ (Mg þ Fe2þ)] (575), respectively, than those of isolated olivine and clinopyroxene phenocrysts. Large plagioclase phenocrysts are rare in the high-SiO2 group (51% by volume), but euhedral to subhedral normally zoned microphenocrysts are common. These vary from 100 to 200 mm in length and are more than twice the average size of the groundmass plagioclase. The An contents of the microphenocrysts varies from 65 to 87, with a clear mode at 65^70 (Fig. 6c). Some plagioclase microphenocrysts form crystal aggregates with olivine and clinopyroxene. The olivine and clinopyroxene in these aggregates have cores with lower Fo# (570) and Mg# (580), respectively, than the cores of isolated olivine and clinopyroxene phenocrysts. The range of An contents in the groundmass plagioclase is 60^70, which is slightly lower than that of the microphenocrysts. Whole-rock Sr^Nd^Pb isotope compositions Fig. 3. (a^d) Trace element patterns normalized to N-MORB (Pearce & Parkinson, 1993) for the four studied sections of the KitaMatsuura basalts. Areas encircled by broken lines represent the entire range of Kita-Matsuura basalt samples analysed in this study. In each panel, the three basalt groups (high-, medium-, and lowSiO2 groups) are distinguished. (e) Trace element patterns normalized to sample 031026a8 (low-SiO2 basalt) for high- (continuous line), medium- (dashed line), and low-SiO2 (grey field) groups. phenocrysts and can be attributed to fractional crystallization of olivine and spinel (Arai, 1994). The host olivines exhibit bell-shaped Fe^Mg zoning patterns, which indicate that the Fo# around the spinel inclusions was not modified during cooling as the samples were obtained from thin lava flows or the surface of thick lava flows. The Sr^Nd^Pb isotopic compositions are reported in Table 4 and plotted in Fig. 7. These are in the range of 0·70362^0·705479 in 87Sr/86Sr, 0·512584^0·512866 in 143 Nd/144Nd, 18·036^18·451 in 206Pb/204Pb, 15·493^15·715 in 207 Pb/204Pb, and 38·326^39·125 in 208Pb/204Pb. All the samples from this study plot within the range of other Cenozoic basalts from eastern Asia (Fig. 7). The isotopic compositions of sediments on the Pacific Plate (PAC) and the Philippine Sea Plate (PHS) are also plotted in Fig. 7 (Hickey-Vargas, 1991; Plank & Langmuir, 1998; Shimoda et al., 1998; Hauff et al., 2003; Plank et al., 2007). They plot in the more enriched Sr^Nd isotope quadrant and have higher 206Pb/204Pb than the Kita-Matsuura basalts, but similar 208Pb/204Pb to them. The isotopic compositions of the Kita-Matsuura basalts show systematic spatial and temporal variations (Fig. 7b, d, and f). Samples from the easternmost Kunimi section (Figs 1 and 7g), have relatively depleted Sr^Nd isotope compositions, and higher 208Pb/204Pb for a given 206 Pb/204Pb. The Kunimi section lavas become more depleted MORB mantle (DMM)-like with time (Zindler & Hart, 1986). Samples from the Ishimori section (Figs 1 and 7g) plot in the middle of the range of the 1090 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Fig. 4. Relationships between Nb concentration and (a) Nb/Ce and (b) Nb/Th. Variations of Ce/Yb (c) and Sm/Yb (d) are also shown for the three basalt groups as a function of eruption age. In (a) and (b) the symbols define the different basalt groups. In (c) and (d) data from the same section are connected by lines. Kita-Matsuura basalts in Sr^Nd^Pb isotope space. The Ishimori section temporally changes to higher 206Pb/204Pb (Fig. 7d) and 207Pb/204Pb. The three early samples from the Senryu section (Figs 1 and 7g) temporally approach a depleted component in terms of their Sr^Nd^Pb composition, but the three later samples temporally change towards an enriched component in Sr^Nd space and a component with higher 206Pb/204Pb, 207Pb/204Pb, and 208 Pb/204Pb values. Samples from the Hirado section (Figs 1 and 7g), the westernmost section, plot in the most enriched region for the Kita-Matsuura basalts in Sr^Nd space. They show a temporal change to higher 206 Pb/204Pb (Fig. 7d), 207Pb/204Pb, and 208Pb/204Pb (Fig. 7d). There is also a systematic correlation between Pb isotope composition and the SiO2 groups: 206Pb/204Pb for given values of 143Nd/144Nd and 208Pb/204Pb increases in the order of the low-, medium-, and high-SiO2 groups (Fig. 7d). The range of the low-SiO2 group does not overlap that of the high-SiO2 group in the Pb isotope system, whereas the medium-SiO2 group lies between the lowand high-SiO2 groups (Fig. 7d). The 87Sr/86Sr, 143 Nd/144Nd, and 206Pb/204Pb of the low-SiO2 samples in the earliest stages of the Senryu, Ishimori, and Kunimi sections are intermediate within the range of variation of the Kita-Matsuura samples (Fig. 7c, d, and f). Samples from the high-SiO2 group, especially in the latest stages of the Hirado, Senryu, and Ishimori sections, have the highest 206Pb/204Pb among the Kita-Matsuura basalts, except for sample 02110934, which has extremely high 208 Pb/204Pb (Fig. 7c, d, and f). E S T I M AT I O N O F M E LT I N G CON DITIONS In the following section, the three least-fractionated basalts are used to represent the continuous spectrum of parental magmas from which the entire chemical diversity of the Kita-Matsuura basalts is derived (Sakuyama et al., 2009). These are referred to as parental magma 1 [PM1 (¼ 031027e4) hereafter] for the low-SiO2 group, PM2 (¼ 02111173b) for the medium-SiO2 group, and PM3 (¼ 02110934) for the high-SiO2 group. The estimation of melting conditions was performed in four steps. First, the compositions of the primary melts in equilibrium with mantle peridotite were estimated from PM1, PM2, and PM3 by adding olivine (Sakuyama et al., 2009), following Tamura et al. (2000). Addition of clinopyroxene with olivine was investigated to evaluate how much this can affect the results; this did not have any significant effect on the results in the following discussion. The detailed procedure has been described by Sakuyama et al. 1091 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Table 2: Representative chemical compositions of spinel inclusions Host rock: 031027E4 031027E4 031027E4 031027E4 031027E4 031027E4 031027E4 031027E4 Basalt group: low low low low low low low low Section: Kunimi Kunimi Kunimi Kunimi Kunimi Kunimi Kunimi Kunimi Sample: sp1–ol6 sp1–ol22 sp1–ol55 sp1–ol30 sp2–ol2 sp1–ol67 sp2–ol81 sp1–ol86 Major elements (wt %) SiO2 0·05 0·12 Al2O3 14·13 17·96 0·07 14·0 0·08 0·07 0·06 0·02 0·07 19·42 20·73 15·32 16·84 18·04 TiO2 0·75 0·86 0·62 1·05 0·86 0·76 0·95 0·63 FeO* 37·20 25·51 36·11 25·82 23·59 33·04 30·85 27·05 MnO 0·42 0·26 0·39 0·22 0·25 0·32 0·26 0·29 MgO 6·04 10·45 6·59 10·86 12·07 7·14 8·93 9·61 CaO 0·03 0·05 0·04 0·01 0·02 0·00 0·00 0·01 Na2O 0·00 0·01 0·00 0·03 0·00 0·02 0·01 0·01 K2O 0·00 0·01 0·01 0·02 0·01 0·01 0·00 0·00 Cr2O3 38·72 44·03 38·49 41·68 42·37 41·27 40·34 42·48 0·18 V2O3 0·19 0·15 0·22 0·12 0·22 0·21 0·11 NiO 0·14 0·12 0·09 0·13 0·14 0·05 0·09 0·09 Total 97·67 99·54 97·42 99·42 100·31 98·19 98·39 98·44 Cations Si 0·010 0·023 0·016 0·015 0·013 0·012 0·005 0·013 Al 3·538 4·132 3·684 4·440 4·629 3·732 4·015 4·223 Ti 0·119 0·127 0·098 0·153 0·122 0·118 0·144 0·094 Fe2þ 6·611 4·163 6·380 4·188 3·738 5·710 5·220 4·493 Mn 0·075 0·044 0·069 0·036 0·041 0·055 0·044 0·049 Mg 1·914 3·040 2·075 3·139 3·410 2·200 2·693 2·844 Ca 0·007 0·011 0·009 0·003 0·004 0·000 0·000 0·001 Na 0·000 0·005 0·000 0·009 0·000 0·006 0·005 0·002 K 0·000 0·002 0·004 0·004 0·001 0·001 0·000 0·001 Cr 6·503 6·792 6·428 6·391 6·348 6·742 6·453 6·671 V 0·033 0·024 0·037 0·019 0·034 0·035 0·017 0·028 Ni 0·024 0·019 0·015 0·020 0·021 0·008 0·015 0·014 Cation total 1·834 18·380 18·814 18·415 18·360 18·619 18·611 18·434 100Cr3þ/(Cr3þ þ Al3þ) 64·77 62·18 63·57 59·01 57·83 64·37 61·64 61·24 100Cr3þ/(Cr3þ þ Al3þ þ Fe3þ) 48·88 54·72 48·27 51·38 51·22 52·22 50·19 52·96 Host olivine Fo# 77·37 87·23 80·17 86·34 87·88 78·06 82·53 84·74 Host rock: 031027E4 02111173 02111173 02111173 02111173 02111173 02110934 02110934 Basalt group: low medium medium medium medium medium high high Section: Kunimi Senryu Senryu Senryu Senryu Senryu Senryu Senryu Sample: sp1–ol89 sp1–ol4 sp2–ol4 sp1–ol10 sp1–ol15 sp1–ol16 sp4–ol13 sp2–0l10 Major elements (wt %) SiO2 0·03 0·06 0·06 0·03 0·04 0·03 0·10 0·10 Al2O3 15·10 14·62 15·86 18·17 9·48 14·39 22·19 25·64 TiO2 0·91 1·34 1·46 1·24 1·31 1·14 1·19 1·16 FeO* 29·45 39·04 39·29 33·03 47·31 42·21 26·57 26·69 MnO 0·33 0·42 0·39 0·38 0·45 0·38 0·21 0·23 MgO 8·43 6·10 6·51 8·24 4·70 5·91 10·66 11·71 (continued) 1092 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Table 2: Continued Host rock: 031027E4 02111173 02111173 02111173 02111173 02111173 02110934 02110934 Basalt group: low medium medium medium medium medium high high Section: Kunimi Senryu Senryu Senryu Senryu Senryu Senryu Senryu Sample: sp1–ol89 sp1–ol4 sp2–ol4 sp1–ol10 sp1–ol15 sp1–ol16 sp4–ol13 sp2–0l10 CaO 0·01 0·06 0·02 0·02 0·02 0·07 0·01 0·03 Na2O 0·00 0·02 0·03 0·01 0·00 0·00 0·01 0·00 0·00 0·01 0·00 0·01 0·01 0·01 0·01 0·00 Cr2O3 K2O 43·86 33·27 31·90 36·18 32·42 30·78 37·46 31·68 V2O3 0·19 0·18 0·17 0·12 0·20 0·18 0·21 0·16 NiO 0·07 0·12 0·12 0·11 0·14 0·11 0·15 0·16 Total 98·36 95·22 95·81 97·52 96·08 95·20 98·76 97·57 Cations Si 0·006 0·013 0·012 0·005 0·009 0·006 0·020 0·020 Al 3·622 3·762 4·027 4·377 2·565 3·751 5·048 5·794 Ti 0·139 0·220 0·237 0·190 0·226 0·189 0·173 0·167 Fe2þ 5·013 7·128 7·077 5·647 9·080 7·806 4·289 4·279 Mn 0·056 0·078 0·072 0·065 0·087 0·071 0·034 0·037 Mg 2·558 1·984 2·090 2·510 1·606 1·948 3·067 3·347 Ca 0·003 0·013 0·005 0·004 0·006 0·017 0·001 0·006 Na 0·000 0·007 0·013 0·003 0·000 0·000 0·004 0·000 K 0·000 0·002 0·001 0·003 0·004 0·002 0·001 0·001 Cr 7·059 5·743 5·432 5·846 5·881 5·382 5·718 4·802 V 0·030 0·031 0·028 0·19 0·037 0·032 0·033 0·025 Ni 0·011 0·021 0·021 0·018 0·025 0·019 0·023 0·024 Cation total 18·449 19·003 19·015 18·687 19·525 19·224 18·410 18·503 100Cr3þ/(Cr3þ þ Al3þ) 66·06 60·42 57·43 57·19 69·63 58·93 53·11 45·32 100Cr3þ/(Cr3þ þ Al3þ þ Fe3þ) 55·8 42·78 40·47 45·36 40·68 38·60 46·14 38·14 Host olivine Fo# 82·01 78·99 80·32 82·91 75·16 78·27 85·97 85·56 Host rock: 02110934 02110934 02110934 02110934 02110934 02110934 Basalt group: high high high high high high 02110934 high Section: Senryu Senryu Senryu Senryu Senryu Senryu Senryu Sample: sp1–ol18 sp3–ol8 sp1–ol16 sp1–ol30 sp1–ol21 sp1–ol24 sp1–ol20 Major elements (wt %) SiO2 0·05 0·08 0·09 0·08 0·06 0·07 0·08 Al2O3 20·83 24·55 18·25 19·89 21·76 23·52 19·33 TiO2 1·36 0·71 0·67 0·91 0·79 0·90 1·02 FeO* 29·22 20·55 24·94 26·42 24·51 25·84 26·51 MnO 0·33 0·21 0·29 0·26 0·26 0·31 0·28 MgO 9·70 11·82 10·31 10·51 12·06 11·61 10·57 CaO 0·02 0·00 0·03 0·00 0·00 0·00 0·09 Na2O 0·00 0·02 0·00 0·00 0·00 0·00 0·00 K2O 0·00 0·01 0·01 0·00 0·01 0·00 0·00 Cr2O3 36·03 3·98 42·21 37·89 38·55 33·58 39·36 0·17 V2O3 0·24 0·12 0·09 0·13 0·16 0·14 NiO 0·10 0·19 0·16 0·14 0·18 0·18 0·11 Total 97·87 97·23 97·05 96·23 98·34 96·15 97·52 (continued) 1093 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Table 2: Continued Host rock: 02110934 02110934 02110934 02110934 02110934 02110934 02110934 Basalt group: high high high high high high high Section: Senryu Senryu Senryu Senryu Senryu Senryu Senryu Sample: sp1–ol18 sp3–ol8 sp1–ol16 sp1–ol30 sp1–ol21 sp1–ol24 sp1–ol20 Si 0·010 0·015 0·019 0·015 0·011 0·014 0·016 Al 4·859 5·505 4·292 4·695 4·943 5·436 4·515 Ti 0·203 0·101 0·101 0·136 0·114 0·133 0·153 Fe2þ 4·837 3·270 4·162 4·424 3·951 4·237 4·393 Mn 0·055 0·034 0·049 0·044 0·043 0·051 0·047 Mg 2·860 3·352 3·068 3·137 3·464 3·394 3·123 Ca 0·003 0·000 0·006 0·000 0·001 0·001 0·019 Na 0·000 0·008 0·000 0·000 0·000 0·000 0·000 K 0·001 0·001 0·001 0·000 0·003 0·000 0·000 Cr 5·638 5·864 6·659 5·997 5·874 5·206 6·167 V 0·037 0·018 0·015 0·021 0·025 0·022 0·026 Ni 0·016 0·029 0·026 0·023 0·028 0·028 0·018 Cation total 18·520 18·196 18·398 18·492 18·456 18·521 18·477 100Cr3þ/(Cr3þ þ Al3þ) 53·71 51·58 60·81 56·09 54·30 48·92 57·73 100Cr3þ/(Cr3þ þ Al3þ þ Fe3þ) 44·96 48·44 53·26 47·47 46·64 41·04 49·13 Host olivine Fo# 84·54 87·77 86·95 86·02 87·18 86·30 86·84 Cations *Total iron given as FeO. Cation numbers are calculated on the basis of 24 O. Ferric content for spinel was estimated by assuming spinel stoichiometry All Ti was combined with Fe as the ulvöspinel component (Fe2TiO4) in the calculation. All analysed spinel grains are inclusions in olivine phenocrysts. (2009). Second, the fertility of the residual source peridotite was estimated based on the compositional relationship between the spinel inclusions and the host olivine phenocrysts. Third, water contents in the primary melts were estimated using plagioclase^melt and olivine^melt hygrometers and the observed crystallization sequence of the basalts. Fourth, melting pressures and temperatures were estimated under hydrous conditions by using the estimated primary magma compositions and published experimental data. Details of the procedures employed are presented in the following sections. Estimation of primary melt compositions Arai (1987) proposed the use of the olivine^spinel mantle array (OSMA) as an indicator of the degree of melt extraction from a peridotite source on the basis of a positive correlation between the Fo# of olivine and Cr# of chromian spinel in peridotite. Once a primary melt segregates from the source peridotite and starts fractionating olivine phenocrysts, the Fo# of olivine in equilibrium with the residual melt quickly decreases as olivine fractionation progresses; this creates a fractionation trend at a high angle to the OSMA. Because Cr# of both the low- and high-SiO2 groups is 50 at Fo# of 88, olivine in equilibrium with the primary melts of the Kita-Matsuura basalt should have a relatively high Fo#. If the peridotite source mantle beneath northwestern Kyushu follows the OSMA, the Fo# of residual olivine in equilibrium with the primary melt of the Kita-Matsuura basalt is estimated to be 90 or more by linear extrapolation of the observed trend (Fig. 5a). Here, we assume Fo# 90 for olivine in equilibrium with the primary melt of the low-SiO2 group. The melting degree of the source mantle for the lowSiO2 group is estimated to be the lowest of the two groups of the Kita-Matsuura basalt (Sakuyama et al., 2009). The corresponding Fo# for the medium- and high-SiO2 groups are thus inferred to be greater than 90, if they were all derived from the same original source mantle. The rate of increase of Cr# in spinel relative to the increase of Fo# in the host olivine increases with the progress of melting, as shown by the OSMA. The estimated Fo# of olivine in the source peridotite for the high-SiO2 group also has a similar composition to the low-SiO2 group, judging from the intercept at Cr# ¼ 55^60 on the OSMA (Fig. 5a). This suggests a similar degree of depletion of the sources of the Kita-Matsuura primary basalts 1094 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING the partial melt (e.g. Hirose & Kawamoto, 1995). It is thus very important to evaluate the effect of water when estimating melting conditions. Here, we examine water content in the primary melt for each SiO2 group by combining two approaches. First, we estimated water contents in the melt using a combination of the plagioclase^liquid hygrometer of Lange et al. (2009) and a geothermometer for olivine-saturated melts (Sugawara, 2000; Medard & Grove, 2008). This hybrid method utilizes the difference of liquidus temperature decrease between plagioclase and olivine when water is added to the melt. Second, we modelled crystallization trends using MELTS (Ghiorso & Sack, 1995) to test whether the water contents estimated by the geohygrometer can reasonably reproduce the geochemical trend of each SiO2 group. Crystal^liquid hygrometer^thermometer Lange et al. (2009) developed a new thermodynamic model for the plagioclase^liquid exchange reaction between albite and anorthite components, which can be used as a plagioclase^liquid hygrometer. The hygrometer is formulated as wt % H2 O ¼ m0 f ðP,TÞ þ a00 þ Fig. 5. (a) Relationships between the Cr/(Cr þAl) atomic ratio (Cr#) of spinel inclusions and the Fo content (Fo#) of the host olivine in the least fractionated olivine basalt (02110934 ¼ PM3, highSiO2, Senryu) and olivine^clinopyroxene basalt (031027e4 ¼ PM1, low-SiO2, Kunimi) from the Kita-Matsuura basalts. Continuous-line and dashed-line arrows represent fractionation trends defined by clusters of data. (b) Cr^Al^Fe3þ ternary diagram showing the composition of chromian spinel included in olivine phenocrysts within the two least fractionated basalts. and contradicts the previous assumptions (Fo# ¼ 89^91) of Sakuyama et al. (2009). Therefore, we first assumed that the Fo# of olivine in equilibrium with the primary melt of the medium- and high-SiO2 groups was also 90 (the effect of a residue with Fo# higher than 90 will be discussed below). The amount of olivine or olivine þ clinopyroxene added to PM1, PM2, and PM3 was 13, 18, and 16 wt % for olivine addition and 18, 24, and 21wt % for olivine þ clinopyroxene addition, respectively. These values were taken from Sakuyama et al. (2009). The estimated MgO contents in the primary melts are 16·6, 16·2, and 15·3 wt % and FeO* contents are 10·5, 10·4, and 9·9 wt % when olivine was added for the low-, medium-, and highSiO2 groups, respectively (Table 1). Estimation of water content of the magma b00 X 00 þ di X i T ð1Þ where m0, a00, b00, and d00 i are constant parameters that were estimated by fitting during calibration of experimental data, P is the pressure, T the temperature, and Xi is the fraction of the oxide component i in the melt. The water content in a melt is a function of the chemical compositions of the plagioclase and melt, temperature, and pressure. Activity^composition relationships for the plagioclase solid solution were taken from the Holland & Powell (1992) model 4. When we applied this model to natural samples it was necessary to constrain the crystallization pressure and temperature. When the pressure was fixed, the water content could be expressed as a function of temperature. If the plagioclase-saturated melt is also saturated with olivine, the olivine liquidus temperature can be estimated by applying the melt geothermometer equation for an olivine-saturated liquid proposed by Sugawara (2000). The thermometer is formulated as Liq Liq TðKÞ ¼ 1446 1 440XSiO2 0 5XFeO Liq Liq ð2Þ þ 12 32XMgO 3 899XCaO þ 0 0043P where X is mol % and P is the pressure (0·1MPa). The effect of water on the olivine liquidus temperature was evaluated by applying the thermodynamic model proposed by Medard & Grove (2008). This model equates to Water in the source peridotite, even in small amounts, can affect the degree of melting at a given temperature (e.g. Green, 1973). Accordingly, this affects the composition of 1095 melt melt 2 melt 3 2 97ðCH Þ þ 0 0761ðCH Þ TðKÞ ¼ 40 4CH 2O 2O 2O ð3Þ JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Table 3: Representative chemical compositions of plagioclase phenocrysts or microphenocrysts Host rock: 031026A5 031026A5 031026A5 03102045 03102045 03102045 03102147 03102147 03102147 03102151 Basalt group: low low low low low low medium medium medium medium Section: Kunimi Kunimi Kunimi Senryu Senryu Senryu Senryu Senryu Senryu Senryu Sample: pl-1 pl-2 pl-3 pl-1 pl-2 pl-3 pl-1 pl-2 pl-3 pl-1 SiO2 51·19 52·49 55·10 47·37 49·65 52·13 50·07 51·90 54·51 51·89 Al2O3 30·16 29·04 27·70 32·12 30·08 28·85 31·11 29·63 27·56 29·38 TiO2 0·07 0·09 0·04 0·06 0·06 0·13 0·06 0·0 0·14 0·08 FeO* 0·35 0·45 0·35 0·41 0·46 0·72 0·42 0·35 1·06 0·51 MnO 0·00 0·02 0·00 0·04 0·04 0·00 0·00 0·02 0·04 0·01 MgO 0·08 0·08 0·08 0·12 0·08 0·13 0·09 0·13 0·16 0·09 CaO 13·52 12·37 11·04 15·87 14·08 12·33 14·52 12·90 11·05 13·39 Na2O 3·66 4·27 5·07 2·28 3·23 4·23 3·18 3·83 4·96 3·90 K2O 0·26 0·32 0·40 0·15 0·27 0·42 0·15 0·18 0·38 0·30 Cr2O3 0·00 0·00 0·02 0·02 0·00 0·01 0·03 0·00 0·00 0·00 V2O3 0·02 0·00 0·01 0·00 0·04 0·01 0·00 0·00 0·00 0·00 NiO 0·00 0·00 0·00 0·01 0·01 0·00 0·00 0·01 0·05 0·00 Total 99·30 99·13 99·82 98·44 98·00 98·96 99·62 99·00 99·90 99·55 Si 7·042 7·216 7·484 6·629 6·946 7·196 6·887 7·141 7·432 7·127 Al 4·892 4·706 4·436 5·298 4·960 4·695 5·044 4·806 4·429 4·757 Ti 0·007 0·009 0·004 0·006 0·006 0·014 0·007 0·005 0·014 0·008 Fe2þ 0·040 0·052 0·039 0·049 0·054 0·083 0·048 0·040 0·121 0·058 Mn 0·000 0·002 0·000 0·004 0·005 0·000 0·000 0·003 0·004 0·001 Mg 0·017 0·016 0·017 0·026 0·017 0·027 0·018 0·026 0·032 0·019 Ca 1·992 1·823 1·607 2·379 2·111 1·824 2·140 1·902 1·615 1·971 Na 0·977 1·137 1·336 0·620 0·876 1·132 0·848 1·022 1·312 1·037 K 0·045 0·056 0·070 0·026 0·048 0·074 0·027 0·031 0·066 0·053 Cr 0·000 0·000 0·002 0·002 0·000 0·001 0·003 0·000 0·000 0·000 V 0·002 0·000 0·001 0·000 0·004 0·001 0·000 0·000 0·000 0·000 Ni 0·000 0·000 0·000 0·001 0·001 0·000 0·000 0·002 0·005 0·000 Cation total 15·015 15·019 14·995 15·038 15·028 15·045 15·021 14·978 15·029 15·031 An# 66·09 60·43 53·34 79·34 70·68 61·71 71·63 65·06 55·17 65·52 KDðCa=NaÞ 1·18 0·93 0·70 1·49 0·94 0·63 1·38 1·02 0·67 1·29 Host rock: 03102151 03102151 02110607 02110607 02110607 03101923 03101923 Basalt group: medium medium high high high high high high Section: Senryu Senryu Senryu Senryu Senryu Hirado Hirado Hirado Sample: pl-2 pl-3 pl-1 pl-2 pl-3 pl-1 pl-2 pl-3 03101923 SiO2 53·17 56·35 50·88 53·19 56·40 49·86 51·04 51·07 Al2O3 28·84 26·91 30·17 28·83 2·71 31·09 30·19 30·41 TiO2 0·06 0·00 0·01 0·10 0·11 0·05 0·06 0·06 FeO* 0·27 0·25 0·75 0·83 0·68 0·64 0·69 0·62 MnO 0·03 0·00 0·00 0·01 0·01 0·00 0·01 0·00 MgO 0·06 0·03 0·05 0·06 0·04 0·06 0·04 0·06 CaO 12·41 10·07 13·47 11·92 9·66 14·69 13·88 13·39 Na2O 4·56 5·72 3·54 4·40 5·72 3·06 3·38 3·61 K2O 0·31 0·45 0·21 0·27 0·45 0·20 0·22 0·23 Cr2O3 0·00 0·00 0·01 0·00 0·02 0·00 0·00 0·00 (continued) 1096 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Table 3: Continued Host rock: 03102151 03102151 02110607 02110607 02110607 03101923 03101923 Basalt group: medium medium high high high high high 03101923 high Section: Senryu Senryu Senryu Senryu Senryu Hirado Hirado Hirado Sample: pl-2 pl-3 pl-1 pl-2 pl-3 pl-1 pl-2 pl-3 V2O3 0·00 0·00 0·05 0·02 0·05 0·00 0·01 0·03 NiO 0·00 0·00 0·01 0·03 0·00 0·00 0·00 0·00 Total 99·69 99·79 99·15 99·64 99·84 99·65 99·50 99·46 Cations Si 7·265 7·634 7·022 7·272 7·644 6·869 7·020 7·019 Al 4·644 4·298 4·907 4·647 4·267 5·048 4·895 4·927 Ti 0·006 0·000 0·001 0·010 0·011 0·005 0·006 0·006 Fe2þ 0·030 0·029 0·086 0·095 0·077 0·073 0·080 0·071 Mn 0·003 0·000 0·000 0·001 0·001 0·000 0·001 0·000 Mg 0·012 0·006 0·010 0·013 0·009 0·012 0·008 0·012 Ca 1·817 1·461 1·992 1·746 1·404 2·169 2·046 1·972 Na 1·209 1·503 0·947 1·165 1·503 0·818 0·901 0·961 K 0·053 0·078 0·03 0·046 0·078 0·036 0·039 0·040 Cr 0·000 0·000 0·001 0·000 0·002 0·000 0·000 0·000 V 0·000 0·000 0·005 0·002 0·005 0·000 0·000 0·000 0·000 0·000 0·001 0·003 0·000 0·000 0·000 0·000 Cation total Ni 15·038 15·008 15·011 15·000 14·999 15·029 14·996 15·010 An# 60·05 49·29 66·94 59·03 47·03 71·77 68·53 66·34 1·02 0·66 1·27 0·90 0·56 1·57 1·34 1·21 KDðCa=NaÞ *Total iron given as FeO. Cation numbers were calculated on the basis of 24 oxygen atoms. KDðCa=NaÞ was calculated by assuming a whole-rock composition as the melt composition in equilibrium with the plagioclase. melt where CH is the wt % of H2O in the melt. Medard & 2O Grove (2008) experimentally quantified the effects of water on the liquidus of olivine-saturated primitive basaltic and andesitic melts. They estimated Tat a given pressure with a known amount of water added, and successfully separated the effect of water from other potential influences (e.g. melt composition and pressure) over a wide range of olivine-saturated basaltic compositions. Although we do not have a precise constraint on the crystallization pressure, clinopyroxene phenocryst compositions suggest that the crystallization pressure was less than 0·5 GPa (Sakuyama et al., 2009). Therefore, we varied pressure conditions from 0·1 to 0·5 GPa in this study. In contrast to plagioclase, the liquidus temperature drop of olivine is less sensitive to the water content in the melt than that of plagioclase. The temperature^water content in melt relationship for plagioclase and olivine should cross over at a specific temperature and water content, where olivine and plagioclase are both saturated under the same conditions (Fig. 8a). We estimated the water content in the melt from the intersection of the Lange et al. (2009) plagioclase^liquid hygrometer and the olivine-saturated melt geothermometer^hygrometer of Sugawara (2000) and Medard & Grove (2008) (Fig. 8a). For samples in which both olivine and plagioclase are saturated, and equilibrium among olivine, plagioclase, and melt can be assumed, we can estimate the temperature and water content of the melt at a given pressure. To evaluate the accuracy of this approach, we applied this method to the results of experimental studies from the literature under hydrous conditions. Standard deviations (1s) of the difference of the temperature and water content between the estimated melt and experiments were 238C and 0·6 wt %, respectively (see Appendix for details). To ensure equilibrium, we selected aphyric or nearly aphyric samples that contain euhedral olivine and plagioclase phenocrysts, which can be assumed to be in equilibrium with a melt close to the whole-rock composition. We assumed that the phenocryst cores with the highest Fo# and An# in the samples were in equilibrium with the melt. Figure 8b and Table 5 shows the estimated water contents in the melt assuming 0·3 GPa. The error bars 1097 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 respectively. In contrast, the minimum values of the estimated water contents in the samples of the low-, medium-, and high-SiO2 groups were 0·15, 0·26, and 0·35 wt %, respectively. These low values of H2O may reflect crystallization of the plagioclase phenocrysts after degassing during magma ascent to the surface. Because degassing of H2O from the magma before crystallization of plagioclase may have occurred in all the samples, the maximum water contents estimated above are minimum estimates of the water content of the magma. Our estimate of the water content of the primary melts is slightly lower than that of Iwamori (1991); his estimated water contents in alkaline and tholeiitic basalts of the Chugoku district, southwestern Japan, were 0·5^1·5 wt % and 1wt %, respectively. These values were obtained from a melting experiment. Iwamori (1991) also considered the effect of CO2. However, the contribution of CO2 to the generation of the Kita-Matsuura basalt may be limited because the Kita-Matsuura basalt is not as silicaundersaturated as the alkaline basalt in Chugoku. Estimation by MELTS calculations Fig. 6. Histograms of anorthite (An) content of the cores of plagioclase phenocrysts or microphenocrysts for (a) low-SiO2 (b) mediumSiO2 and (c) high-SiO2 basalt groups. Microphenocrysts were analyzed in 03101923 because of the absence of phenocrysts. In (a) and (b) arrows indicate the compositional range of groundmass plagioclase, whereas in (c) they indicate the compositions of the outermost rim of plagioclase microphenocrysts. represent the range of water contents obtained by varying pressure from 0·1 to 0·5 GPa. Because the samples show variable extents of differentiation (Fig. 8b), the water content in the primary melt was then calculated from the fractionation-corrected least differentiated samples (PM1, PM2, and PM3) of each SiO2 group. The maximum estimated water contents in the low-, medium-, and highSiO2 groups are 0·7, 1·3, and 2·1wt %, respectively. Combined with the amount of olivine fractionated from the PM1^PM3, the maximum water contents in the least differentiated melts of low-, medium-, and high-SiO2 groups were estimated to be 0·6, 1·1, and 1·8 wt %, To cross-check the water contents estimated above, we performed MELTS calculations to investigate whether the major element trends for the low-, medium-, and highSiO2 groups can be reproduced by isobaric fractional and equilibrium crystallization starting from parental magmas PM1, PM2, and PM3 (Fig. 9a^c). Water contents assumed in the initial melts were 0·25, 0·5, 1 and 2 wt %, and crystallization pressures were set to 0·1, 0·2 and 0·3 GPa. Isobaric crystallization paths were calculated with increments of 2 K decrease in temperature under oxygen fugacity conditions 1 log unit less than the Ni^ NiO oxygen fugacity buffer (NNO 1) (Nilsson & Peach, 1993; Farley, 1994); the results were then compared with the observed fractionation trends. Although both the fractional and equilibrium crystallization paths are similar, fractional crystallization tends to generate higher TiO2, FeO*, Na2O, K2O, and P2O5, and lower CaO, MgO, and Al2O3. The crystallization paths calculated by MELTS for TiO2, Al2O3, and FeO* during both fractional and equilibrium crystallization are more sensitive to water content than to crystallization pressure (Fig. 9d^f, g^i, m^o). The TiO2 content increases with decreasing water content of the initial melt, particularly during the early crystallization stage (Fig. 9j and m). This behaviour of TiO2 is mainly controlled by the liquidus temperature of plagioclase and Fe^Ti oxides, which is strongly dependent on the water content (Housh & Luhr, 1991; Takagi et al., 2005). As water content in the melt increases, the difference of liquidus temperature between plagioclase and Fe^Ti oxides decreases, whereas the liquidus temperatures of both plagioclase and Fe^Ti oxides decrease. If the melt is enriched in water, the amount of fractionation of plagioclase before the onset of 1098 03101930 Sample: 1099 0·006 38·660 0·002 15·584 0·003 18·378 0·000012 0·512698 0·000013 0·003 38·565 0·001 15·550 0·001 18·296 0·000011 0·512688 0·000015 0·704638 03101924 medium 8·1 Hirado 0·008 38·483 0·003 15·562 0·004 18·286 0·000012 0·512574 0·000013 0·705479 03101817 medium 8·84 Hirado 0·012 38·749 0·005 15·603 0·006 18·353 0·000013 0·512724 0·000016 0·70441 02110723 high 6·37 Senryu 0·018 39·125 0·007 15·715 0·008 18·451 0·000013 0·51277 0·000017 0·704298 02110934 high 6·48 Senryu 0·007 38·506 0·003 15·528 0·003 18·272 0·000010 0·512765 0·000012 0704319 02111276 medium 6·54 Senryu 0·014 38·461 0·006 15·514 0·006 18·209 0·000009 0·512796 0·000013 0·703647 02111173B medium 6·56 Senryu 0·012 38·637 0·005 15·541 0·006 18·237 0·000008 0·512801 0·000015 0·704067 03102155 medium 6·59 Senryu 0·015 38·784 0·006 15·537 0·007 18·248 0·000010 0·512751 0·000013 0·704888 03102145 low 7·68 Senryu 0·003 38·691 0·001 15·576 0·001 18·374 0·000007 0·512744 0·000015 0·704247 031029if7 high 5·42 Ishimori 0·003 38·616 0·001 15·545 0·002 18·289 0·000010 0·51271 0·000014 0·704384 03102481 high 7·33 Ishimori 0·004 38·881 0·002 15·533 0·002 18·216 0·000012 0·512715 0·000013 0·704117 03102475 low 7·74 Ishimori 0·013 38·326 0·005 15·493 0·006 18·036 0·000010 0·512866 0·000012 0·70362 031027e4 low 7·11 Kunimi 0·018 38·413 0·007 15·552 0·009 18·253 0·000010 0·512757 0·000012 0·703845 031027c9 low 7·26 Kunimi 0·003 38·917 0·001 15·565 0·002 18·292 0·000010 0·512697 0·000014 0·704552 031026a8 low 7·76 Kunimi K–Ar ages were calculated by regressing separately for each section to estimate ages for lava flows without age determination [see details given by Sakuyama et al. (2009)]. 1 2s Pb/204Pb 208 2s Pb/204Pb 207 2s Pb/204Pb 206 2s Nd/144Nd 143 2s 0·704973 high SiO2 group: Sr/86Sr 7·29 K–Ar age (Ma)1: 87 Hirado Section: Table 4: Sr^Nd^Pb isotopic compositions of the Kita-Matsuura basalt SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Fig. 7. Sr^Nd^Pb isotopic compositions of the Kita-Matsuura basalts. (a, b) 143Nd/144Nd vs 87Sr/86Sr; (c, d) 208Pb/204Pb vs 206Pb/204Pb; (e, f) 143 Nd/144Nd vs 206Pb/204Pb; (b), (d), and (f) are close-ups of the rectangles in (a), (c), and (e), respectively. Numbers in symbols in (b), (d), and (f) indicate the eruption sequence as shown in (g). AOC, altered oceanic crust. The northern hemisphere reference line (NHRL; Hart, 1984) and magnitude of deviation from the NHRL (8/4 and 7/4) are shown as dashed^dotted lines in (c). Filled inverted triangle represents a 15 Ma dacite sample reported by Uto et al. (2004). Isotopic end-member components for DMM, EM1, EM2, and HIMU are from Zindler & Hart (1986) and Hofmann (1997). PAC, Pacific Plate; PHS, Philippine Sea Plate. 1100 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING 1^2 wt %, respectively. Within these ranges of water contents, a crystallization pressure of 0·2 GPa was predicted for every group. Differentiation paths calculated for pressures greater than 0·2 GPa, assuming both fractional and equilibrium crystallization, predict a decrease in SiO2 during the early stages of crystallization owing to excessive fractionation of clinopyroxene, with the extent of the SiO2 decrease increasing with pressure. Crystal fractionation at pressures higher than 0·2 GPa is, therefore, unlikely as suggested by Sakuyama et al. (2009). The estimates of melt H2O contents based on MELTS calculations are consistent with those estimated using a combination of the plagioclase^liquid hygrometer and olivine-saturated liquid hygrometer. We adopted the maximum values of the H2O contents estimated by the hygrometers as the H2O contents in the least-differentiated samples of each SiO2 group: the values are 0·6 wt %, 1·1wt %, and 1·8 wt % for the low-, medium-, and high-SiO2 groups, respectively. Taking the amount of crystal fractionation from the primary melts into account, the H2O contents in the primary melt of each SiO2 group were estimated to be 0·5 wt %, 0·9 wt %, and 1·5 wt %, respectively. Estimation of melting conditions Melting pressure and the effect of H2O Fig. 8. (a) Relationships between water content estimated using the plagioclase hygrometer of Lange et al. (2009) (dashed and dotted curves) and the olivine-saturated melt thermometer (Sugawara, 2000; Medard & Grove, 2008) (continuous curves) for the mediumSiO2 sample 03102147 from the Senryu section. The intersection of the dashed and continuous curves gives the water content in the melt and crystallization temperature. (b) Relationship between FeO*/ MgO and water content in the melt estimated assuming a pressure of 0·3 GPa. Error bars indicate the range of water contents for pressures from 0·1 to 0·5 GPa. (c) Comparison of estimated water content in the melt for the three basalt groups. Range of water content in the melt estimated by MELTS is indicated by a double-headed arrow. Fe^Ti oxide fractionation decreases. In this case, TiO2 content of the melt does not increase with decreasing MgO. Tatsumi & Suzuki (2009) showed that increased TiO2 content at a given SiO2 content for less hydrous melts was caused by greater fractionation of plagioclase (and lesser fractionation of Fe^Ti oxides). Their conclusion is also consistent with our observations. Plausible ranges of water content in the PM1^PM3 magmas are considered to be 0·25^0·5, 0·25^1, and Although the Mg/Si ratio of a partial melt of peridotite decreases with increasing water content in the system (e.g. Hirose, 1997), the decrease in Mg/Si ratio is suppressed with increasing pressure up to 5 GPa (Kushiro et al., 1968; Inoue & Sawamoto, 1992; Inoue, 1994). In melting experiments in the water-bearing peridotite system at 1GPa, Hirose & Kawamoto (1995) showed that partial melts with less than 2·5 wt % H2O, which formed at temperatures above 12008C, are all within the range of major element compositions of anhydrous melts formed by the same degree of partial melting under pressures from 0·5 to 1·5 GPa. Therefore, it may be concluded that the effect of H2O on the pressure estimation for the Kita-Matsuura basalt is much less than 0·5 GPa, if the magma generation depth ranges from 1·5 to 3 GPa. Further information on the effect of water comes from experiments seeking the conditions at which a primary melt is multiply saturated with four-phase lherzolite or three-phase harzburgite assemblages. Tatsumi et al. (1983) experimentally showed a positive correlation between the water content of the melt and the pressure of multiple saturation up to 2 GPa. According to their experiments, 1·5 wt % water in the melt increases the multiple saturation pressure by 0·2 GPa from 1·5 GPa under anhydrous conditions, and 3 wt % increases the multiple saturation pressure by 0·5 GPa from 1·8 GPa. Linear interpolation of this relationship gives 1101 P wet ¼ P dry þ H2 Oprimary 05 : 30 ð4Þ JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Table 5: Parameters used for estimation of water content in the melt and estimated values of P,Tand H2O Sample Section Group Bulk SiO2 Pl Estimated values FeO*/MgO (wt %) An#1 P T H2O2 (GPa) (8C) (wt %) CFM3 H2O-CF4 (wt %) 031027c6 Kunimi Low-SiO2 1·49 47·9 76 0·3 1173 1·02 0·32 0·70 031027e6 Kunimi Low-SiO2 1·25 47·5 80 0·3 1193 0·83 0·17 0·69 031026a6 Kunimi Low-SiO2 1·90 48·8 64 0·3 1140 0·88 0·32 0·60 031026a5 Kunimi Low-SiO2 1·65 49·3 65 0·3 1191 0·23 0·34 0·15 03102147 Senryu Medium-SiO2 1·89 51·9 72 0·3 1102 1·81 0·28 1·30 03102148 Senryu Medium-SiO2 1·77 52·5 72 0·3 1116 1·54 0·28 1·11 03102597b Ishimori Medium-SiO2 1·77 50·4 73 0·3 1127 1·56 0·30 1·09 03102153 Senryu Medium-SiO2 1·42 50·9 78 0·3 1178 0·86 0·12 0·76 03102152b Senryu Medium-SiO2 1·24 50·6 78 0·3 1203 0·52 0·11 0·46 03102152c Senryu Medium-SiO2 1·15 51·2 78 0·3 1216 0·3 0·12 0·26 031029if7 Ishimori High-SiO2 1·04 52·6 87 0·3 1131 2·29 0·10 2·06 031029if6 Ishimori High-SiO2 1·08 51·7 87 0·3 1143 2·2 0·13 1·92 02111276 Senryu High-SiO2 1·36 51·5 76 0·3 1131 1·68 0·14 1·44 03101503 Senryu High-SiO2 1·19 52·8 85 0·3 1163 1·4 0·11 1·24 03101928 Hirado High-SiO2 1·58 51·5 76 0·3 1141 1·4 0·24 1·06 03101935 Hirado High-SiO2 1·58 51·7 74 0·3 1158 0·98 0·24 0·74 03101924 Hirado High-SiO2 1·33 53·7 77 0·3 1167 0·95 0·30 0·67 03101923 Hirado High-SiO2 1·25 53·7 77 0·3 1182 0·73 0·30 0·51 03101931 Hirado High-SiO2 1·48 51·9 72 0·3 1187 0·44 0·20 02110607 Senryu High-SiO2 1·19 51·6 66 0·3 1229 0 0·35 0·00 *Total iron given as FeO. 1 An# [¼ 100 Ca/(Ca þ Na)] is the highest An# plagioclase core in each sample. 2 H2O content is a value estimated by assuming the bulk-rock composition as the liquid composition, which is in equilibrium with olivine with the highest Fo# and plagioclase with the highest An# in the sample. 3 CFM represents the fractionated mass ratio from the least fractionated samples of each SiO2 group. 4 H2O-CF represents the water content in the melt; values were corrected for crystal fractionation from the least differentiated samples (PM1, PM2, and PM3) for each group. We use equation (4) to estimate the melting pressure under hydrous conditions by comparing the estimated KitaMatsuura primary melts with the results of anhydrous high-pressure melting experiments in a CIPW normative nepheline^olivine^quartz (Ne’^Ol’^Qtz’) ternary projection (Irvine & Baragar, 1971), following Sakuyama et al. (2009) (Fig. 10). The estimated melting pressure is in the range of 2·5^2·8 GPa for the low-SiO2 group, 2^2·3 GPa for the medium-SiO2 group, and 1·5^1·6 GPa for the highSiO2 group; these results apply if the Fo# of olivine in equilibrium with the primary melts was 90 (Fig. 10). Taking the effect of water into account by applying equation (3), the melting pressures of the primary magmas of the three basalt groups are now estimated to be 2·5^2·9 GPa, 2·1^2·4 GPa, and 1·7^1·9 GPa for the low-, medium-, and high-SiO2 groups, respectively. If we assume a higher degree of melting, resulting in a Fo# of 91 in the residue, for the high-SiO2 group as an extreme case, the melting pressure of the high-SiO2 group primary melt is estimated to be 1·8^2·1GPa, which is distinctly lower than that of the low-SiO2 group (Fig. 10). It should be noted that the estimation of melting pressure in this study and in that of Sakuyama et al. (2009) does not depend on source peridotite composition, because the isobaric compositional trends of the partial melts in the Ne’^ Ol’^Qtz’ ternary do not shift significantly (Hirose & Kushiro, 1993) even if the peridotite contains 50% of a basaltic component (Kogiso et al., 1998). 1102 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Melting temperature There is a simple relationship between temperature, pressure, and MgO content in liquids saturated with olivine and/or pyroxene (Ramsay et al., 1984; Sugawara, 2000; Maaloe, 2004; Herzberg et al., 2007). We recompiled the results of anhydrous melting experiments saturated with olivine and pyroxene under mantle melting conditions (0·5^4 GPa) and performed a linear regression analysis of the dependence of the temperature (T in 8C) as a function of both pressure (P in GPa) and the MgO content (wt %). We obtained the following relationship for anhydrous conditions (see Appendix for details): T ¼ 1080 7 þ 54 75 0 27074P 2 ð5Þ þ2 21634P 0 99731 þ 15 08MgO: The olivine liquidus temperature estimated by equation (5) represents the maximum temperature at which the basaltic melt was generated because water in the melt decreases the olivine liquidus temperature. By combining equations (3) and (5), we have estimated the segregation Fig. 9. Variation diagrams for TiO2, Al2O3, and FeO* vs MgO for groundmass compositions in the Kita-Matsuura basalts (a^c) and fractional (d^f, j^l) and equilibrium (g^i, m^o) crystallization paths modeled by the MELTS program. Calculated MELTS trends are shown for several different initial melt water contents at 0·2 GPa, and at different crystallization pressures for fixed amounts of water (0·25, 0·50, 1 and 2 wt % for the low-, medium-, and high-SiO2 groups) in (d^i). (continued) 1103 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Fig. 9. (Continued). temperature of each Kita-Matsuura basalt group from the segregation pressure and primary melt compositions, including the water content estimated above. Estimated conditions are listed in Table 6 and shown in Fig. 11. The most plausible ranges of melt segregation conditions are summarized in Table 6. The segregation temperatures were estimated to be 1450^15008C, 1400^14608C, and 1350^14008C for the low-, medium-, and high-SiO2 groups, respectively. Addition of just olivine resulted in higher temperatures and pressure than when both olivine and clinopyroxene were added to estimate the primary melt composition. However, even though clinopyroxene was added with olivine, the temperature and pressure of melting clearly decrease from low-, through medium-, to the high-SiO2 group (Fig. 11). Therefore, the uncertainty surrounding deep cryptic fractionation of clinopyroxene does not affect the following discussion. Mantle potential temperature Mantle potential temperature is one of the most important parameters in an area of upwelling mantle (McKenzie, 1984; McKenzie & Bickle, 1988; Putirka et al., 2007; Lee et al., 2009, and references therein). To estimate the potential temperature of the upwelling mantle from volcanic rocks, either the source mantle composition or the degree of melting at which the estimated primary melt was generated are prerequisites, as the adiabat of melting mantle is different from that of solid mantle (McKenzie, 1984): 10^208C GPa1 for solid mantle and 708C GPa1 for melting mantle (e.g. McKenzie, 1984; Iwamori et al., 1995). The primary melt of the low-SiO2 group is most likely to be the closest to the initial melt from the upwelling mantle beneath Kita-Matsuura. Therefore, provided that the primary melt of the low-SiO2 group was an initial sample of the very first melt generated from the upwelling mantle, we can estimate the minimum potential temperature of the upwelling mantle to be 14508C by extrapolating the adiabat of a solid mantle (Fig. 11a). We argue that the potential temperature of the source mantle for the Kita-Matsuura basalt was higher than 14508C, which is 1008C higher than the mantle potential temperatures inferred beneath normal mid-ocean ridges, the Japan Sea, and the Chugoku district in southwestern Japan 1104 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Table 6: Estimated melting conditions for the three basalt groups Group Fig. 10. Normative compositions of estimated primary melts in the ternary nepheline^olivine^quartz. Isobaric curves defining the compositions of partial melts of anhydrous peridotite are from a compilation by Sakuyama et al. (2009). The projection scheme is after Irvine & Baragar (1971): Ne’ ¼Ne þ 0·6Ab; Qtz’ ¼Qtz þ 0·4Ab þ 0·25Opx; Ol’ ¼Ol þ 0·75Opx. (McKenzie, 1984; Yamashita & Fujii, 1992; Tatsumi et al., 1994; Iwamori et al., 1995; Lee et al., 2009), 0^1008C higher than that beneath back-arc basins (Wiens et al., 2006), and 50^2008C lower than that of a hotspot (Putirka, 2005; Lee et al., 2009). Temporal and spatial changes in melting conditions Now that we have estimated the melting conditions for the primary melts, the temporal and spatial changes of the magma groups shown in Fig. 1b can be interpreted as reflecting temporal and spatial variations in the melting conditions (Fig. 11b). Basaltic volcanism began at the Hirado section in the westernmost part of the Kita-Matsuura area. Beneath each section, the estimated melting pressure and temperature decrease over time, whereas the water content of the primary melt increases until eruption of the high-SiO2 group, which is generated at the shallowest level in the mantle. Magma eruption ceased slightly earlier (7·5 Ma) in the Hirado section than in the other sections (Sakuyama et al., 2009). Meanwhile, the Kunimi section, in the easternmost part of Kita-Matsuura, 35 km east of the Hirado section, produced only melt generated at high pressure and temperature under virtually anhydrous conditions (low-SiO2 group) and the duration of the basaltic volcanism was shorter than at the other sections (Fig. 11b). Basalt generated at higher pressure and temperature under nearly anhydrous conditions is more abundant in the marginal sections such as Ishimori and Kunimi than in the Hirado and Senryu sections. This spatial variation can be extended to the west of the Hirado section, as the basalt in the Ikitsuki section is more similar to those in sections to the east of the Hirado section than that in the Hirado section in terms of eruption age and major element chemical composition (Fig. 11b). Assuming that melting Fo# H2O Phase* P (GPa) T (8C) High-SiO2y 90 1·5 ol 1·9 1370 High-SiO2 90 0·6 ol 1·7 1398 High-SiO2 90 0·0 ol 1·7 1418 High-SiO2 90 1·5 ol þ cpx 1·7 1346 High-SiO2 90 0·6 ol þ cpx 1·6 1373 High-SiO2 90 0·0 ol þ cpx 1·5 1392 High-SiO2 91 1·5 ol 2·1 1424 High-SiO2 91 0·6 ol 2·0 1451 High-SiO2 91 1·5 ol þ cpx 1·9 1387 High-SiO2 91 0·6 ol þ cpx 1·8 1412 Medium-SiO2y 90 0·9 ol 2·4 1454 Medium-SiO2 90 0·3 ol 2·3 1460 Medium-SiO2 90 0·0 ol 2·3 1470 Medium-SiO2 90 0·9 ol þ cpx 2·2 1418 Medium-SiO2 90 0·3 ol þ cpx 2·1 1424 Medium-SiO2 90 0·0 ol þ cpx 2·0 1433 Low-SiO2y 90 0·5 ol 2·9 1483 Low-SiO2 90 0·0 ol 2·8 1499 Low-SiO2 90 0·5 ol þ cpx 2·6 1449 Low-SiO2 90 0·0 ol þ cpx 2·5 1465 *Phase indicates added crystal(s) to estimate primary melt compositions. Ol and ol þ cpx represent olivine and olivine þ clinopyroxene addition, respectively. yThe most plausible melting conditions. Melting conditions assuming Fo# ¼ 91 for the high-SiO2 group are also included to discuss the effects of the extent of source depletion. (See main text for details.) Melting pressure and temperature were estimated by the method described in the main text assuming Fo# of olivine in equilibrium with the primary melt and H2O content in the primary melt. pressures under hydrous conditions changed from 2·9 GPa (low-SiO2) to 1·9 GPa (high-SiO2) from 7·7 to 6·5 Ma in the Senryu section, we propose mantle upwelling at a velocity of 2 cm a1. DISCUSSION Sakuyama et al. (2009) proposed that the Kita-Matsuura basalts could be separated into low-, medium- and highSiO2 groups that exhibit separate liquid lines of descent. By applying the melting model of Ozawa & Shimizu (1995) to fluid-immobile elements, Sakuyama et al. (2009) concluded that the degree of melting systematically increased from the low-, through medium-, to high-SiO2 groups. The volcanism started at the Hirado section with an eruption of medium-SiO2 group basalt followed by high-SiO2 group basalt. The volcanism at the Senryu and 1105 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Fig. 11. (a) Melting temperatures and pressures estimated for primary melts with various water contents for the low- (open and filled circles), medium- (open and filled triangles), and high-SiO2 (open and filled squares) basalt groups in equilibrium with olivine of Fo# 90. The results for a residual olivine composition of Fo# 91 are also shown for the high-SiO2 group defined by the diagonally striped rectangle. The water content is represented by the size of the filled symbols, and dry conditions are defined by open symbols. Higher-pressure primary melts were estimated from olivine addition, whereas lower-pressure melts were estimated from olivine þ clinopyroxene addition. Experimentally determined solidi (KLB1 and HK66; Takahashi & Kushiro, 1983; Takahashi et al., 1993) and compiled solidi for variable water contents (Hirschmann, 2000; Katz et al., 2003) are also shown. Adiabatic P^T paths for solid peridotite (178C GPa1) and melting peridotite (708C GPa1) for a mantle potential temperature of 14508C are shown by the thick gray lines after McKenzie (1984). The horizontal lines with double arrowheads indicate the range of potential temperatures for the Chugoku district, southwestern Japan (Iwamori, 1991) and mid-ocean ridges (MOR) (Lee et al., 2009). The bold dotted-line arrow indicates a hypothetical melting adiabat for the low-SiO2 group primary magma. (b) Estimated melting pressures and temperatures are shown for the five Kita-Matsuura basalt localities; bold dashed and fine dotted lines define isochrons to clarify the spatio-temporal changes. The conditions for Takushima and Ikitsuki were estimated using the whole-rock analyses of Uto et al. (2004). Gray areas represent interpolated melting paths. 1106 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Ishimori sections followed, with eruption of low-SiO2 group magma progressing to medium- and high-SiO2 group magma, whereas the Kunimi section produced low-SiO2 group magma all the way up to the uppermost horizon. Using new data, we have further constrained the magma generation conditions beneath Kita-Matsuura: (1) the water contents of the primary magma in the low-, medium-, and high-SiO2 groups are estimated to be 0·5, 0·9, and 1·5 wt %, respectively; (2) the melting pressure decreased from 2·5^2·9 to 2·1^2·4, and 1·7^1·9 GPa and the temperature decreased from 1450^1500 to 1400^1460 and 1350^14008C for the low-, medium-, and high-SiO2 groups, respectively; (3) the estimated mantle potential temperature was 14508C or higher, even after taking into account the effects of water in the peridotite system. The Kita-Matsuura basalts have the following geochemical characteristics: (1) HFSE/LREE decreases with decreasing HFSE abundance (Fig. 4a); (2) LREE/HREE, MREE/HREE, Zr/Y (Fig. 4c), and Nb/Y (Fig. 4d), and Nb/Th decrease from low-, through medium-, to the high-SiO2 group. Combining these petrological and geochemical results with previously published geological data (Sakuyama et al., 2009), we can now constrain the melting processes beneath northwestern Kyushu from 9·5 Ma until c. 6 Ma. Melting of a homogeneous mantle source The decrease of HFSE/large ion lithophile element (LILE) and HFSE/LREE ratios with time and with decreasing HFSE abundance are common features in all the studied sections in the Kita-Matsuura area. The relative depletion of HFSE with respect to LILE and LREE has been generally attributed to (1) the presence of residual accessory minerals (e.g. rutile, titanite, zircon) that retain HFSE in the melting residue (e.g. Wood et al., 1979; Saunders et al., 1980), (2) interaction of the ascending magma with surrounding harzburgitic peridotite (e.g. Kelemen et al., 1990), or (3) melting of a mantle source composition modified by the influx of LILE-enriched slab-derived fluids or melts (e.g. Brenan et al., 1994). Melting models that assume a homogeneous mantle source in a system closed to input (Fig. 12b, model 1), however, fail to reproduce the systematic coupling of an increase in the water content in the primary magma and the increase in the magnitude of the HFSE depletion relative to LILE and LREE with a decrease in the trace element abundance, because water and HFSE, LILE and REE are all incompatible elements. Thus, either a heterogeneous source mantle (Fig. 12b, model 2) or the involvement of an agent rich in water and fluid-mobile components (Fig. 12b, model 3) is required. Melting of internally heterogeneous upwelling mantle The source mantle may be chemically heterogeneous on a scale much smaller than 10 km (Fig. 12b, model 2), which is the distance between the studied sections in the KitaMatsuura area. In this model, each SiO2 group would have formed by the melting of a specific portion of the heterogeneous mantle and segregation of the resultant melt at an appropriate depth. However, melting and melt segregation from a certain portion of the mantle will suppress further melting of this portion at shallower levels. The important basic premise of this model is that the cross-over of the solidus in upwelling mantle is coupled with the timing of melt segregation without further melt generation. The onset of melting of a hydrous mantle peridotite source that rises adiabatically in a closed system is deeper than that for the equivalent anhydrous mantle because the solidus temperature is lowered in the presence of water. Therefore, melt production in the deeper portion of the upwelling mantle should be higher for hydrous peridotite. Because the degree of melting increases from the low- to the high-SiO2 group (Sakuyama et al., 2009), as well as the water content in the primary magma, the water content in the source mantle should also increase from the low- to the high-SiO2 group (Hirose & Kawamoto, 1995; Katz et al., 2003). It is thus difficult to explain why the high-SiO2 group magma with a high H2O content was derived from the shallowest depth. Although a refractory mantle peridotite composition can delay the onset of melting, the estimated highest degree of melting of the highSiO2 group of the Kita-Matsuura basalts cannot be achieved by this model. This is because the degree of melting at a shallow level is more strongly controlled by the source depletion rather than its H2O content. This is due to the decreased role of H2O in lowering the solidus temperature of a peridotite at a greater degree of melting by the dilution of H2O in the partial melt. This suggests that water must have been introduced into the source mantle of the high-SiO2 group after the segregation of the medium-SiO2 melt, violating the assumption of a closed system. We therefore conclude that the essential features of the Kita-Matsuura basalts cannot be produced by any internally heterogeneous mantle model, but instead that they require a supply of aqueous fluid into the melting system at each melting stage. Progressive melting of a homogeneous upwelling mantle source and addition of fluid from a chemically stratified mantle Another plausible melting model is the progressive addition of H2O-bearing fluid to an upwelling of mantle source (Fig. 12b, model 3). This is the most plausible model that can be used to explain the progressive increases in H2O and LILE contents, and the degree of melting with simultaneous decreases in pressure and temperature over time. There are multiple mechanisms to achieve the process described above. The first model involves a 1107 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Fig. 12. (a) Schematic illustration of the diapiric upwelling model for the petrogenesis of the Kita-Matsuura basalts. (b) Schematic illustrations of the melting models: model 1, progressive decompression melting of a single source mantle in a system closed to input; model 2, decompression melting of an internally heterogeneous mantle, each portion of which provides melt of low-, medium-, and high-SiO2 basalt composition independently (see the main text for details); model 3, progressive decompression melting open to both input and output. (c) Schematic illustration of a homogeneous diapiric upwelling mantle that is interacting with a geochemically distinct hydrous ambient mantle for each instant of time shown in (b). progressive supply of fluids from the downgoing Philippine Sea slab to adiabatically upwelling anhydrous mantle, as occurs in the mantle wedge above a subduction zone. The second model involves the interaction of an ascending mantle diapir with a geochemically distinct shallower upper mantle layer as proposed by Sakuyama et al. (2009). Although recent seismic topography studies have revealed the presence of a downgoing slab beneath northwestern Kyushu (Zhao et al., 2012), as is the case with our previous model, we prefer the second model. This is because the imaged high Vp anomaly beneath northwestern Kyushu is not as clear as those imaged beneath other subduction zones, and the chemical compositions of Cenozoic basalts in northwestern Kyushu are clearly distinct from those of typical subduction zone volcanism. Upwelling of the mantle in this region is suggested to be related to the initiation of the opening of the Beppu-Shimabara Rift at the northern extension of the Okinawa Trough, which occurred before 6 Ma. The presence of pre-metasomatized mantle beneath north Kyushu is also reasonable because the area had long been a supra-subduction zone setting, related to the subduction of the Pacific Plate slab since the Cretaceous up to the Middle Miocene before opening of the Japan Sea (60^15 Ma; e.g. Kimura et al., 2005). When part of the cooler ambient mantle is mechanically incorporated into the thermally buoyant mantle diapir, it will be heated and decompress concurrently as it rises within the diapir, forming a mushroom-like morphology (Griffiths, 1986). If the incorporated mantle is heated to temperatures high enough to melt or dehydrate, it will provide a melt or fluid to the overlying original diapiric mantle (Fig. 12b, model 3 and Fig. 12c). This process may promote further melting of the original upwelling diapir, even if the upwelling mantle becomes refractory after melt segregation. Below, we discuss a mantle melting model open to both outputs and inputs to explain the petrological, geochemical, and spatiotemporal variations in the Kita-Matsuura basalts. 1108 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Multistage open-system melting model To account for the addition of fluid into the melting system, we used the open-system melting model of Ozawa (2001), which can model simultaneous influx of melt or fluid and melt segregation in multiple stages. The equation used is from equation (59) of Ozawa (2001): for f5fc ¼ ac/[1 þ b(1 ac)], lq,in lq Cj ¼ Cj0 þ Cj bf Pj0 þ f ð1 þ b Pj Þ ð6Þ and for f fc ¼ ac/[1 þ b(1 ac)], h i9 8 lq,in,k lq,in,k lq,0,k > b k Cj bk Cj Cj ð1 Pjk þ bk Þ > > > > > > > < = k þbk 1 1P lq j " # Cj ¼ k k k k > f k ð1 Pjk þ b gk Þ 1Pj þb g 1 Pjk þ bk > > > > > > > : 1 þ 0,k ; 0,k 0,k Pj ð1 a Þ þ a ð7Þ lq Cj is the concentration of the jth component in where lq,in is the concentration of the jth component the melt, Cj lq,0 in the influxing fluid, Cj is the initial concentration of the jth component in the melt, a is the melt fraction in the initial system, ac is the critical melt fraction, b is the mass influx rate (influxing mass fraction of the initial solid mass divided by the degree of melting), f is the extent of solid^melt reaction relative to the initial solid mass, Pj is the weighted average of the jth partition coefficient between the solid and melt, P j0 is the initial bulk concentration of the jth component normalized to the initial melt composition, and g is the mass separation rate (separated mass fraction of the initial solid mass divided by the degree of melting). Parameters with superscript k are the parameters for the kth stage. In this model, influx begins simultaneously with the initiation of melting. A schematic illustration of the key parameters is shown in Fig. 13. In our previous model, only single-stage critical melting was examined for Nb/Yand Zr/Y (Sakuyama et al., 2009). However, the necessity of a melting model also open to input is required to explain the temporal changes in pressure, temperature, and water content in the melt consistently (see above). The open-system critical melting (OSCM) model of Ozawa & Shimizu (1995), which assumes constant influx and melt separation rates during the entire melting event, can handle more general opensystem problems than the model of Sobolev & Shimizu (1993). The assumption of constant influx and melt separation rate is, however, inappropriate for the KitaMatsuura basalts, particularly because the HFSE/LILE ratio of the samples decreases with decreasing HFSE abundance, together with a temporal increase in water content in the estimated primary melts (Table 6). This strongly suggests increases of water-rich fluid or melt influx into the Fig. 13. Schematic illustration of the open-system melting model used in this study. (See text for details.) melting system as melting proceeds. We therefore employed the multistage melting model represented by equation (7), in which the extent to which the system is open can be varied (by adjusting b and g), although the model requires more parameters (e.g. number of stages) that are difficult to constrain. However, this choice as a conceptual basis should be more reasonable compared with other models. We separated the melting process into six stages (Table 7 and Fig. 14a). The first three stages involve garnet peridotite melting, the fourth stage melting during the transition from garnet to spinel peridotite, and the last two stages spinel peridotite melting. The first stage allows only input into the source to investigate the degree of enrichment of the original source before melting. Therefore, we set up two stages for both the garnet and spinel stages and one stage for the transition from garnet to spinel. Because melting started at the garnet stage and finished at the spinel 1109 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Table 7: Estimated set of parameters for open-system melting at each melting stage and assumed mineral and melting mode Stage Enriched mantle source (Workman et al., 2004) Initial value at each melting stage f Lithology Mode (wt fraction) Ol Opx Cpx Sp Parameters for open-system melting a b g Input Output Gt 1 0·000 0·550 0·200 0·150 0 0·100 Gt lhz 0·000 0·020 0·00 Closed Closed 2 0·010 0·555 0·204 0·143 0 0·098 Gt lhz 0·010 0·020 0·50 Open Open 3 0·025 0·562 0·210 0·133 0 0·095 Gt lhz 0·018 0·040 0·90 Open Open 4 0·045 0·572 0·218 0·119 0 0·091 Gt/Sp lhz 0·021 0·070 1·10 Open Open 5 0·063 0·564 0·296 0·159 0·021 0 Sp lhz 0·021 0·100 1·10 Open Open 6 0·100 0·599 0·292 0·132 0·019 0 Sp hzb 0·020 0·140 1·20 Open Open Melting mode Ol Gt stage Sp–Gt transition Sp stage Opx Cpx Sp Gt 0 0·3 0·08 –0·19 0·81 1·00 –1·90 –3·70 –1·07 –0·30 0·40 0·82 0·08 4·67 0 Range of melting degree and melting stage for accumulated melt to reproduce each group Z Degree Stage Lithology Low 0–0·065 1–5 Gt-lhz–Gt/Sp-lhz 0·14 (wt %) Medium 0·025–0·090 3–5 Gt-lhz–Sp-lhz 0·27 (wt %) High 0·045–0·120 4–6 Sp-lhz–Sp-hzb 0·49 (wt %) Primitive mantle source (Sun & McDonough, 1989) Stage Initial value at each melting stage f Lithology Mode (wt fraction) Ol Opx Cpx Sp Parameters for open system melting a b g Input Output Gt 1 0·000 0·550 0·200 0·150 0 0·100 Gt lhz 0·000 0·010 0·00 Closed Closed 2 0·015 0·557 0·206 0·140 0 0·097 Gt lhz 0·015 0·025 0·50 Open Open 3 0·020 0·560 0·208 0·137 0 0·096 Gt lhz 0·018 0·035 0·80 Open Open 4 0·035 0·567 0·214 0·126 0 0·093 Gt/Sp lhz 0·021 0·050 1·10 Open Open 5 0·054 0·558 0·293 0·167 0·022 0 Sp lhz 0·021 0·080 1·10 Open Open 6 0·100 0·602 0·288 0·134 0·019 0 Sp hzb 0·021 0·090 1·10 Open Open Melting mode Ol Gt stage Sp–Gt transition Sp stage Opx Cpx Sp 0·08 –0·19 0·81 1·00 –1·90 –3·70 –1·07 0 –0·30 0·40 0·82 0·08 Gt 0·3 4·67 0 Range of melting degree and melting stage for accumulated melt to reproduce each group Z Degree Stage Lithology Low 0–0·041 1–4 Gt-lhz–Gt/Sp-lhz 0·10 (wt %) Medium 0·020–0·058 3–5 Gt-lhz–Sp-lhz 0·20 (wt %) High 0·035–0·185 4–6 Sp-lhz–Sp-hzb 1·31 (wt %) f, a, b, g, and Z are the degree of melting, the mass fraction of melt in the current system of the melting stage, the influx rate, the separation rate, and the ratio of total influxed material relative to the initial solid mass, respectively. Ol, Opx, Cpx, Sp, Gt, lhz, and hzb are olivine, orthopyroxene, clinopyroxene, spinel, garnet, lherzolite, and harzburgite, respectively. Low, medium, and high represent low-, medium-, and high-SiO2 groups, respectively. Estimated Z is the value at the highest degree of melting for each magma group. 1110 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Fig. 14. Results of the open-system melting model. (a) Optimum model parameters for the mineral mode, melt influx rate (b) and total influxed mass relative to the initial mass as a function of the degree of melting. The range of accumulated melt in the primary melts of each basalt group and the degree of melting at which the garnet^spinel transition occurs are also indicated. (b) Primitive mantle normalized trace element patterns of altered oceanic crust (AOC; gray field), sediments (between continuous lines), average of Pacific Plate sediments (filled diamonds), and average of Pacific Plate AOC (open diamonds). Also shown is the calculated fluid composition derived from AOC. Modelled primitive mantle normalized trace element patterns for the primary melts of (c) high-SiO2, (d) medium-SiO2, and (e) low-SiO2 basalt groups from a PMTL source (filled diamonds) and an EM source (open diamonds). In these panels the gray fields define the trace element patterns of the estimated primary melts. The instantaneous melt composition from PMTL source mantle is shown by a bold dotted line. Each primary melt targeted for use in the model calculations is shown by a dashed line. stage, we had to take into account the effect of the phase transition during modelling. Although the number of stages for the multistage melting model is arbitrary, this configuration is rather simple to treat melting from the garnet to the spinel stage comprehensively. Mineral modes and melting parameters such as a, b, and g can be different for each stage, but are assumed to be constant within each stage. The initial mineral mode and the melting modes for the garnet and spinel peridotite melting stages are listed in Table 7, and the initial composition is assumed to be primitive mantle (PMTL; Sun & McDonough, 1989) and enriched mantle (EM; Workman et al., 2004), the compositions of which are given in Table 8. Sakuyama et al. (2009) have already shown that a depleted MORB-source mantle (DMM) cannot reproduce the HFSE compositions of the Kita-Matsuura basalts. The melting stoichiometries for the garnet and spinel peridotite melting are the same as those adopted by Sakuyama et al. (2009) and the initial bulk mineral mode follows Johnson et al. (1990). Reaction stoichiometry for the transition from garnet to spinel peridotite is according to Walter et al. (1995). We ignored any temperature and pressure dependence of partition coefficients between mineral and fluid or melt, the effect of which will be discussed below. We explored other melting parameters (e.g. F, a, b, and g) that could reproduce the estimated primary melt compositions by accumulated melting. First, we estimated an optimum parameter set at the degree of melting estimated for the low-SiO2 group (lowest degrees of melting). Then, we estimated the sets of parameters for successive melting stages corresponding to the medium- and high-SiO2 groups (at higher melting degrees) starting from the residue of the previous melting stage. This treatment is based on our model concept that assumes progressive melting of 1111 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Table 8: Partition coefficients of minerals and chemical compositions of initial mantle and influxed material assumed in open-system melting calculations Ol Opx Cpx Sp Gt PMTL EM2 Rb 0·000179 0·0006 0·00175 0 0·0007 0·635 1·456 261·4 Ba 0·000032 0·0035 0·0006 0 0·0007 6·989 15·857 2051·8 Th 0·000052 0·013 0·00531 0 0·001 0·085 0·177 22·7 U 0·000018 0·0017 0·00361 0 0·00918 0·02 0·040 4·5 Ta 0·00007 0·003 0·0102 0·02 0·02 0·041 0·063 0·6 Nb 0·0001 0·003 0·0077 0·02 0·015 0·71 1·087 17·2 K 0·000177 0·0003 0·0072 0 0·0007 La 0·000028 0·0025 0·0536 0·01 0·0015 0·687 0·895 Ce 0·000038 0·005 0·0858 0·01 0·008 1·775 1·923 62·8 Pb 0·000479 0·0013 0·072 0 0·0005 0·185 0·144 109·3 Pr 0·0008 0·0048 0·15 0·01 0·054 Sr 0·0015 0·007 0·1283 0 0·006 Nd 0·00042 0·0068 0·1873 0·01 0·087 Zr 0·007 0·021 0·123 0·02 0·5 Hf 0·0038 0·01 0·256 0·02 0·24 Sm 0·0013 0·01 0·291 0·008 0·7 Eu 0·0016 0·013 0·31 0·007 Gd 0·0055 0·016 0·3 Tb 0·0041 0·019 Dy 0·01 Ti Y 250 Influx 0·276 21 1·354 50431·5 33·8 0·251 2·5 20·044 253·0 1·140 23·4 8·835 109·4 0·31 0·238 2·2 0·444 0·347 4·9 0·9 0·17 0·128 1·6 0·006 1·19 0·6 0·445 4·9 0·31 0·009 1·5 0·108 0·082 0·7 0·022 0·33 0·01 2·2 0·74 0·569 5·2 0·006 0·024 0·384 0·048 0·65 0·007 0·06 0·421 0·0023 2·8 4·55 3·655 28·2 Ho 0·007 0·026 0·31 0·009 3·3 0·164 0·127 1·0 Er 0·0087 0·03 0·29 0·01 3·6 0·48 0·378 3·2 Tm 0·009 0·12 0·255 0·01 3·5 0·074 11·2 1300 900·5 5427·5 Yb 0·017 0·049 0·28 0·008 3·88 0·49 0·387 3·2 Lu 0·02 0·06 0·28 0·02 3·79 0·07 0·061 0·5 Ol, Opx, Cpx, Sp, Gt, PMTL, and influx represent olivine, orthopyroxene, clinopyroxene, spinel, garnet, primitive mantle (Sun & McDonough, 1989), and assumed influx fluid, respectively. EM2 is from Workman et al. (2004). Values in the columns for minerals are the assumed partition coefficients, and those for PMTL, EM2, and influx are concentrations in ppm. a common source mantle that consecutively produces the three basalt groups as the mantle ascends. Composition of the influxing material As shown by Sakuyama et al. (2009) and from the newly obtained geochemical data for the Kita-Matsuura basalts, LILE and LREE enrichment relative to HFSE is more evident for the medium- and high-SiO2 basalt groups, which suggests that the contribution of a fluid released from the subducted slab should be considered along with water added to the source mantle beneath Kita-Matsuura. Because the upper mantle beneath northwestern Kyushu was located above the subduction zone before the opening of the Japan Sea (Kimura et al., 2005), the upper mantle must have been affected by fluid released from the subducting slab. When subducting slabs dehydrate and release aqueous fluid, large fractions of the LILE, LREE, Th, U, and Pb are efficiently removed from the subducted sediments and altered oceanic crust. The chemical composition of the released fluid depends on the source material and mobility of each element: mobility ¼ CSTM CRP 100ð%Þ CSTM ð8Þ where CSTM and CRP are an element concentration of the starting material and run products, respectively (Tatsumi et al., 1986). Primitive mantle normalized trace element patterns of fluids from subducting sediments generally show a relatively flat pattern from Rb to U, with strong positive Pb and K anomalies, and a weak positive Ti anomaly (Plank 1112 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING & Langmuir, 1998; Plank et al., 2007; Fig. 14b) similar to the patterns of the Kita-Matsuura basalts, whereas fluids derived from dehydration of altered oceanic crust show strong depletion in Ba, Th, and Pb (Alt & Teagle, 2003; Kelley et al., 2003; Nakamura et al., 2007; Fig. 14b). Thus fluids derived from subducting sediment may dominate the upper mantle signature beneath the back-arc region of Kyushu. There are two candidates for the origin of the sedimentderived fluids, one from the Pacific Plate slab and the other from the Philippine Sea Plate slab. Although the trace element compositions of subducted sediments are different between the two oceanic plates (e.g. Plank & Langmuir, 1998), the difference is much smaller than the difference between sediment and altered oceanic crust. Thus, it is difficult to distinguish the origin of the fluid without additional isotopic data. Therefore, first, we assumed that sediment with an average composition of sediments from the Pacific Plate and Philippine Sea Plate [reported by Plank & Langmuir (1998)] was the hypothetical sediment that releases fluid. As a first-order approximation, we used a slab sediment fluid composition derived by 1% dehydration of Pacific Plate sediment in accordance with the mobilities of trace elements determined by Aizawa et al. (1999). There are a number of ways to transport the sediment-derived fluid component into the upwelling anhydrous mantle. These include (1) direct addition of the fluid through channel flow from the Pacific Plate slab to the upwelling mantle and (2) entrainment of fluid-metasomatized, water-bearing mantle (either lithospheric or asthenospheric) in the upwelling dry mantle. The latter step needs an additional process that includes sediment fluid metasomatism, fluid release from the metasomatized mantle, and introduction of the secondary fluid into the dry mantle. It is impossible to treat such a complex process quantitatively. We therefore used the immediate sediment fluid composition and tested whether or not such a fluid can account for the opensystem melting model. The calculated chemical composition of this influxing material into upwelling mantle is given in Table 8. Below we discuss the calculation results for open-system melting. The effects of the composition of the influxed material on the estimation of b are also discussed below. Modelling results The parameters used in the modelling are listed in Table 8 and the calculated trace element compositions are shown in Fig. 14. Sakuyama et al. (2009) proposed that the lowSiO2 group can be reproduced by the lowest degree of melting whereas the high-SiO2 group can be reproduced by the highest degree of melting. The calculated degree of melting generally agrees in order, but with slightly different values (Table 8 and Fig. 14). The origin of this difference is related to the difference of the adopted melting models, which is discussed further below. Optimized trace element compositions of the partial melt from both the PMTL and EM source reproduce the primary melt for each group reasonably well. In particular, the LILE- and LREE-enriched patterns of the lowSiO2 group and enriched patterns with depletions in HFSE for the medium- and high-SiO2 groups are reproduced by both PMTL and EM source compositions. Differences between these source materials can be partially compensated for by different degrees of melting and the amount of influxed material. The PMTL source gives relatively better results (Fig. 14c^e). For the PMTL source the degrees of melting for the low- and medium-SiO2 groups need to be lower and that for the high-SiO2 group higher than those for the EM source in order for the PMTL source to produce a melt more enriched in LILE and LREE and depleted in HREE than the EM source at the garnet stage. A higher degree of melting is required for the PMTL source to produce a melt with more depleted HREE at the spinel stage than for the EM source. Melting of the PMTL source requires more fluid influx than the EM source because LILE and LREE are more enriched in the EM source than in the PMTL source. The EM source gives a lower MREE abundance for the lowSiO2 group, and lower LREE and MREE for the medium-SiO2 group. Both the PMTL and EM sources yielded slightly higher HREE abundances for the high-SiO2 groups because the contribution of the melt generated at the spinel stage is greater at higher degrees of melting. The abundance of Pb for every SiO2 group (Fig. 14c^e), which strongly depends on the assumed composition of the influx material, is not reproduced well by either PMTL or EM melting. We will discuss the results for the PMTL source in more detail below. To reproduce the trace element patterns of the low-SiO2 group primary melt, a minor influx of material is required (b ¼0·010^0·035) during the garnet peridotite melting stages, up to 4·5% melting, and a further small influx of material (b ¼0·05) is necessary when melting occurs during the transition from garnet to spinel peridotite (Table 7a). Initially, once spinel peridotite is stable (stage 5), a moderate amount of material influx (b ¼0·08) is required during melting to reproduce the trace element pattern of the medium-SiO2 group primary melt at f ¼ 5·8% (Table 7a). Later, during spinel peridotite melting (stage 6), a higher influx of material (b ¼0·09) is required to reproduce the trace element pattern of the high-SiO2 group primary melt at f ¼ 18·5%. The optimized influx rate (b) increases from 0·01 to 0·09 between melting to generate the low-SiO2 group and the highSiO2 group primary melts (Table 7a), which suggests that the fluid or melt influx steadily increased as the diapir ascended. The total amount of material that was added to 1113 JOURNAL OF PETROLOGY VOLUME 55 the diapir for each stage of the melting was 0·10 wt % for the low-, 0·20 wt % for the medium-, and 1·3 wt % for the high-SiO2 groups relative to the initial mass of the source peridotite, which was 1·4 wt % for the low-, 4·8 wt % for the medium-, and 7·6 wt % for the high-SiO2 group relative to the accumulated melt mass. The increases in the rate of influx and the mass of material that influxed into the system during the generation of the low-, through the medium-, to the high-SiO2 primary melts are consistent with the increase in the water content of the primary melts estimated above, if the influxing material contained tens of weight per cent water. Identification of isotopic components involved in melting and their relative contribution The Pb isotope compositions of volcanic rocks, especially for subduction-related volcanism, have been used as a tracer to investigate the contribution of fluid in the magma generation process (e.g. Nakamura et al., 2008; Straub et al., 2009), as Pb is considered to be highly partitioned into the fluid phase released from the igneous and sedimentary layers of the subducting slab (Kogiso et al., 1997; Aizawa et al., 1999; Kessel et al., 2005). The observed systematic relationships among melting pressure, trace element composition, and Pb isotope composition for the Kita-Matsuura basalts allow us to estimate the contribution of appropriate source materials identified as endmember components in 206Pb/204Pb^207Pb/204Pb^208Pb/ 204 Pb isotopic space. Because the Pb isotope compositions of the KitaMatsuura basalts are highly variable (Fig. 7c^f), at least three isotopic end-member components are required. As mentioned above, the low-SiO2 group lavas that erupted in the earliest stage in each section were produced by the smallest degree of melting and were the least affected by the influx of fluid, which allows us to assume that the Pb isotope composition of these lavas should not have been significantly modified from the composition of the original upwelling mantle. However, even within the low-SiO2 group, there are systematic differences in Pb isotope composition. Therefore, we used the average composition of the three low-SiO2 group samples with the highest 208 Pb/204Pb (Kunimi, 031026a8; Ishimori, 03102475; Senryu, 03102145) among the Kita-Matsuura basalts as the initial composition of the upwelling mantle: component C1 (Fig. 15). A sample with the lowest 208Pb/204Pb (Kunimi, 031027e4) was also classified as belonging to the low-SiO2 group, even though it has a considerably different Pb isotope composition from those of the low-SiO2 group lavas in the earliest stage (component C1). This suggests that the original upwelling mantle may have already been isotopically heterogeneous before fluid influx. Extension of NUMBER 6 JUNE 2014 the trend from component C1 to 031027e4 points towards the DMM composition (Fig. 7b, d, and f), which is inferred to be the dominant asthenospheric mantle component. Thus, we assume that DMM is another end-member component for the Kita-Matsuura basalts: component C2 (Fig. 15). The original upwelling mantle before it was affected by fluid is, therefore, assumed to be a mixture of components C1 and C2. Samples of the medium-SiO2 and high-SiO2 groups that erupted in the middle stage (03101930, 03101924, 03102155, 02111173B, 02111276, and 03102481) are higher in 143Nd/144Nd and lower in 87Sr/86Sr and 208Pb/204Pb than those of component C1. This suggests that the contribution of C2 temporally increased. Because the source of the high-SiO2 basalt is estimated to have been affected by fluid from the subducted slab, as discussed above, the third component should represent the fluid influxed to the upwelling mantle. High-SiO2 samples that erupted in the latest stage of the Hirado, Senryu, and Ishimori sections (03101817, 02110934, 02110723, and 031029if7) are lower in 143Nd/144Nd and higher in 87 Sr/86Sr, 206Pb/204Pb, and 208Pb/204Pb than the samples erupted in the middle stage. This suggests that the effects from the third component, which was richer in radiogenic Pb and Sr and poorer in radiogenic Nd, became much stronger in the latest stage of the Kita-Matsuura volcanism. The Pb isotope compositions of altered oceanic crust and sediment on the Philippine Sea Plate and Pacific Plate are shown in Figs 7 and 15. Subducting materials to represent the Philippine Sea Plate were chosen from sites in its northern part [sites 442, 443, and 444 of Deep Sea Drilling Project (DSDP) Leg 58 and site 582 of DSDP Leg 87]. The sediments are enriched in radiogenic Pb and Sr and depleted in radiogenic Nd, which is consistent with the characteristics of the third component required in the petrogenesis of the Kita-Matsuura basalts. Altered oceanic crust and terrigenous sediment of the Philippine Sea Plate have higher 8/4 and 7/4 than those of the Pacific Plate (Hickey-Vargas, 1991; Plank & Langmuir, 1998; Shimoda et al., 1998; Hauff et al., 2003; Plank et al., 2007). Here, we assumed an average of the high-SiO2 group samples with the highest 206Pb/204Pb as a hypothetical product produced by mixing between a partial melt of the original upwelling mantle with C1 þC2 composition and an influxed fluid. We refer to the fluid as component C3 (Fig. 15a). In the open-system modelling, the degree of melting is mostly constrained by fluid-immobile elements such as the HFSE and HREE, and the amount of influxed material by fluid-mobile elements such as Rb, Ba, Th, U, K and the LREE. Good constraints are not provided by Pb and Sr because the assumed abundance of these elements in the influxed material may not be appropriate. Consequently, we estimate the concentrations of Pb, Sr, and Nd and the isotope composition of the influxed fluid 1114 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Fig. 15. (a) 208Pb/204Pb vs 206Pb/204Pb for the Kita-Matsuura basalts and illustration of the method used to estimate the composition of the influxed fluid. The open star represents the estimated composition of the influxed fluid. The other symbols are the same as in Fig. 2d. (b) 208 Pb/204Pb vs 206Pb/204Pb, (c) 143Nd/144Nd vs 87Sr/86Sr, and (d) 143Nd/144Nd vs 206Pb/204Pb for the Kita-Matsuura basalts shown by filled diamonds. The compositional ranges for sediments (SED) and altered oceanic crust (AOC) from the Pacific slab (PAC) and the Philippine Sea slab (PHS) are shown by dark gray areas (Hickey-Vargas, 1991; Plank & Langmuir, 1998; Shimoda et al., 1998; Hauff et al., 2003; Plank et al., 2007). The altered oceanic crust and sediment compositions of the Philippine Sea Plate and Pacific Plate used in the mixing calculations are indicated by symbols: crosses, plus signs, open triangles, and squares. Mixing trajectories marked off every 25 wt % between the materials of the Philippine Sea Plate are shown in (d) as continuous curves and that for the Pacific Plate as dashed curves. by using the degree of melting and influxed mass estimated by trace element modelling. First, the partial melt composition from the C1 mantle was calculated by accumulating the instantaneous melts over a range of degrees of melting from 0·045 to 0·12 for the enriched mantle source and from 0·035 to 0·185 for the primitive mantle without any slab input. Trace element (Sr, Nd, and Pb) concentrations in the initial solid composition of component C1 were assumed as those of enriched mantle (Workman et al., 2004) and primitive mantle (Sun & McDonough, 1989). Next, we estimated the fluid composition by assuming that the average value of the high-SiO2 group samples with the highest 206Pb/204Pb represents a mixture between the melt estimated in the first step and fluid. By applying the mass of influxed material relative to the initial solid mass [0·35 wt % (¼ 0·49^0·14 wt %) for the enriched mantle and 1·21wt % (¼ 1·31^0·10 wt %) for the primitive mantle], which were estimated by trace element modelling and conducting mass-balance calculations (Table 7), we estimated the trace element concentrations and the Sr^ Nd^Pb isotopic composition of the fluid derived from component C3 (Table 9). If we adopt these three mantle components (C1, C2, and C3), all the other samples of the Kita-Matsuura basalts can be reproduced by mixing (Fig. 15a). The estimated Pb isotope composition of the fluid (component C3) for both the enriched and primitive mantle sources plots close to a mixing line between altered oceanic crust and sediment on the Philippine Sea Plate with a strong affinity to the sediment composition (Fig. 15). This suggests that the fluid was originally derived from the sediment layer of the Philippine Sea Plate and that it was 1115 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Table 9: Assumed parameters and estimated fluid isotopic composition Mass (initial solid as 1) Sr (ppm) Sr/86Sr Melt from initial Mixed melt solid PMTL 03101930* 0·15y 108y 0·162z 87 0·704519 7·1y 143 0·512721 0·512734 Pb (ppm) 0·3y 3·1 Pb/204Pb 0·012y 214 Nd (ppm) Nd/144Nd Influxed fluid 1528z 0·704482 0·704450z 13·0 86·1z 0·512747z 23·8z Melt from initial Mixed melt solid EM 03101930* 0·075 99·8y 0·704519 6·7y 0·0035 214 0·704482 13·0 0·512721 0·512734 0·57y 2·1 Influxed fluid 0·0785 2661 0·704452 147·1 0·512747 34·8 206 18·2522 18·3887 18·4137z 18·2522 18·3887 18·4374 207 15·5447 15·6195 15·6332z 15·5447 15·6195 15·6461 208 38·8606 38·8063 38·7964z 38·8606 38·8063 38·7870 Pb/204Pb Pb/204Pb *Concentrations of Sr, Nd, and Pb were corrected to estimated values of the primary melt composition in equilibrium with residual olivine with Fo# ¼ 90, assuming the olivine maximum fractionation model. Sr–Nd–Pb isotopic compositions of initial solid for both the PMTL and EM sources were assumed to be average compositions of the three low-SiO2 group samples (Fig. 15a). yValues estimated by applying the open-system melting model in this study. Compositions for elements in the melt produced from the initial solid were calculated by the accumulated melting model without input. zValues estimated in this study. involved in the generation of the medium- and high-SiO2 group basalts in the later stages of the Kita-Matsuura basalt activity, provided that the subducted sediment composition of the Philippine Sea Plate was not much different from the current sediment on the Philippine Sea Plate. In addition, we estimate the relative contribution of C2 and C3 to C1 during the generation of the Kita-Matsuura basalts (Table 10) based on the composition of C3. We can constrain only the relative contribution of the endmember components, as the absolute value of the mixing ratio depends on the assumed composition of the endmember components. The fraction of C2 relative to C1 temporally increased in the Kunimi, Ishimori, and Senryu sections, except during the latest activity (02110723) in the Senryu section; however, proportions were almost constant in the Hirado section. The contribution of sediment-derived fluid (C3) temporally increased in every section, except for the Kunimi section. As the SiO2 content in the primary melt increased and the estimated melting pressure deceased, the contribution of the sediment-derived (C3) component increased. The average contribution of the C3 fluid to the petrogenesis of the low-, medium-, and high-SiO2 group lavas was 0·15, 0·41, and 1·05 wt %, respectively. These values are reasonably consistent with the total influxed fluid mass estimated above on the basis of incompatible trace elements. Comparison with the previous model and implications for polybaric melting The degree of melting estimated in this study (4·1^18% for PMTL) is systematically higher than that obtained by Sakuyama et al. (2009) (3^9%). The discrepancy is due to the melt segregation process; accumulated melts were adopted in this study, whereas Sakuyama et al. (2009) used instantaneous melts. An accumulated melt derived by high degrees of fractional melting has incompatible trace element abundances similar to those of an instantaneous melt formed by lower degrees of melting (Fig. 14c^e). Although it is difficult to constrain the melt segregation process just from trace elements, the results of this study may be more plausible in that they are consistent with the major element chemistry. An increase in the degree of melting is a linear function of the H2O content in the partial melt (Hirose & Kawamoto, 1995). The presence of 1wt % H2O in the melt, which corresponds to the estimated H2O content in the primary melt of the high-SiO2 group, increases the extent of melting at 1GPa by 20% (Hirose & Kawamoto, 1995) in comparison with anhydrous conditions. Provided that this relationship can also be applied at 2 GPa, the degree of melting for the high-SiO2 group should be as high as 20%, which is consistent with the degree of melting estimated in this study. Furthermore, the melting model that Sakuyama et al. (2009) adopted fails to reproduce the systematic variation of both fluid-immobile and fluid-mobile elements. We therefore conclude that the results of this study are more plausible than those of Sakuyama et al. (2009). However, it should be noted that this new melting model does not invalidate the essential conclusions of Sakuyama et al. (2009), which show that the melting degree increased from the low- to the high-SiO2 group, as the results of the previous study are valid as long as only fluid-immobile HFSE are considered. 1116 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Table 10: Estimated amount of fluid component for samples in each section Section Sample and group K–Ar age* C2/(C1 þ C2) Fluid (Ma) (fraction) component (wt %) Hirado High 03101930 7·29 0·7 1·8 High 03101924 8·1 0·65 0·65 Medium 03101817 8·84 0·84 0·70 High 02110723 6·37 0·22 0·85 High 02110934y 6·48 – – Medium 02111276 6·54 0·72 0·55 Medium 02111173B 6·56 0·63 0·25 Medium 03102155 6·59 0·34 0·15 Low 03102145 7·68 0·11 0·05 High 031029if7 5·42 0·50 1·45 High 03102481 7·33 0·57 0·50 Low 03102475 7·74 0·17 0·0 Low 031027e4y 7·11 0·66 0·0 Low 031027c9x 7·26 0·89 0·55 Low 031026a8 7·76 0·01 0·15 Senryu Ishimori polybaric mantle melting (e.g. Hirose & Kushiro, 1998) suggest that the MgO content of an accumulated partial melt is almost constant for variable melting degrees (6·4^21%). There is no significant variation in MgO content among the presumed primary melt compositions of the three Kita-Matsuura basalt groups. Even in the case that the Fo# of olivine in the residual peridotite in equilibrium with the high-SiO2 primary melt was 91, the MgO content of the primary melt only becames 17·6 wt %, which is 1wt % higher than that of the low-SiO2 melt. However, this difference in MgO content is still smaller than expected from the estimated difference in the degree of melting under isobaric conditions. Slight decreases in the FeO* contents of the estimated primary melts of the low- to high-SiO2 groups can also be explained by a combination of effects of pressure and degree of melting. Therefore, small differences in the MgO contents of the primary melts as well as FeO* among the three KitaMatsuura basalt groups are consistent with accumulation of a partial melt generated by polybaric incremental melting. Comparison with mantle potential temperatures estimated in other studies Kunimi Low, medium, and high represent low-SiO2, medium-SiO2, and high-SiO2 groups, respectively. *K–Ar ages were calculated by regressing separately for each section to estimate ages for lava flows without age determination [see details given by Sakuyama et al. (2009)]. ySamples outside the mixing triangle between C1, C2, and C3. The MgO content of a partial melt of peridotite increases with an increase in the degree of melting under isobaric conditions: MgO content generally increases by 42 wt % with an increase of 10% in the degree of melting (e.g. Hirose & Kushiro, 1993; Baker & Stolper, 1994; Kushiro, 1996; Pickering-Witter & Johnston, 2000; Schwab & Johnston, 2001). Therefore, the estimated difference of 410% in the degree of melting between the low- and high-SiO2 groups may indicate that the MgO contents of the estimated primary melt compositions should increase in the order of low-, medium-, and high-SiO2 groups. However, this expectation is not necessarily the case for polybaric incremental melting because the MgO content of the partial melt decreases with decrease of pressure at a given melting degree. This suggests that the MgO content of the primary melt does not necessarily increase with an increase in the degree of melting, especially if melting was polybaric. Indeed, results of high-pressure experiments on Putirka (2005) estimated the mantle potential temperature to be 14508C beneath mid-ocean ridges, which is higher than other estimates (1280^14008C; McKenzie & Bickle, 1988; Iwamori et al., 1995; Lee et al., 2009). This difference is mainly due to the fact that the Putirka (2005) study was based on the most magnesian phenocrysts (Fo# ¼ 91^92) found in basalts. In contrast, the Fo# of residual olivine in peridotite assumed in many of the other studies is 89^ 90 (e.g. Lee et al., 2009). Lee et al. (2009) obtained a mantle potential temperature of 1300^14008C beneath mid-ocean ridges by assuming Fo# ¼ 90. Because the Fo# of olivine in peridotite increases with the degree of melting, olivine in peridotite at the lowest pressure and temperature during adiabatic melting is expected to have the highest Fo#, if we assume a single episode of peridotite upwelling. Because the melting degree of the high-SiO2 group is higher than that of the low-SiO2 group, the Fo# of olivine in equilibrium with the primary melt of the high-SiO2 group could be higher than 90; however, in this study a value of 90 was assumed for all three SiO2 groups. Accordingly, the primary magma for the high-SiO2 group is most likely to be in equilibrium with olivine having the highest Fo# in northwestern Kyushu. If we assume an Fo# of 91 for the residual peridotite of the high-SiO2 group, which is also the highest Fo# of olivine in ultramafic xenoliths from northwestern Kyushu, the mantle potential temperature beneath Kita-Matsuura is estimated to be 414008C. This estimate is consistent with the result obtained above for the low-SiO2 group primary melt assuming Fo# ¼ 90 for the residual olivine in peridotite. 1117 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 If one erroneously used Fo# ¼ 91 for the low-SiO2 primary melt, which was generated at the initial melting stage, the mantle potential temperature would be overestimated at 415008C. Therefore, it is important to constrain the appropriate Fo# of the olivine in the residual peridotite for each stage of melt segregation for accurate estimation of the mantle potential temperature in an adiabatically upwelling mantle. Our estimate of the potential temperature beneath Kita-Matsuura is based on the simple assumption that the low-SiO2 basalt magma was produced under near-anhydrous conditions and at a lower degree of melting. Therefore, a simple estimation method for potential temperature is appropriate based on the olivine fractionation-corrected primary basalt composition, as noted above. The estimated mantle potential temperature (14508C) may be slightly higher than in the mid-ocean ridge setting; however, it is lower than that for ocean island basalts (Tp416008C; e.g. Putirka, 2005). Our estimate would be reasonable for a back-arc setting and is consistent with the melting regime discussed above. the lower mantle may have been able to rise through the rupture, cross the mantle Transition Zone (410^660 km), and continue upwelling (Fig. 16b). Alternatively, the rupturing of the slab may have triggered upwelling of hotter mantle from beneath the slab (Fig. 16c) in a similar way to the slab window model (e.g. Thorkelson & Taylor, 1989). Cenozoic basaltic back-arc volcanism in northwestern Kyushu is aligned ENE^WSW. This distribution may be caused by the orientation of the rupture beneath southwestern Japan, although it can also be explained by near-surface tectonics (Sakuyama et al., 2009). The high 3 He/4He ratio, up to 16 RA, in a mantle xenolith from the Higashi-Matsuura basalt, which erupted at 3 Ma to the east of the Kita-Matsuura basalts (Fig. 1), also supports our model (Sumino et al., 2000). Although we have no age constraints for the initiation of rupturing or for how long such a rupture can exist, these models are currently the most plausible mechanisms to explain the high mantle potential temperatures beneath Kita-Matsuura, which are higher than those beneath mid-ocean ridges. Geodynamic implications of a hot mantle diapir upwelling beneath Kyushu CONC LUSIONS Plume models proposed for the eastern margin of the Eurasian Plate have assumed that the upwelling originates at either the upper^lower mantle or core^mantle boundary (e.g. Nakamura et al.,1990).These models, however, may not entirely rule out the possibility of a shallower origin, as they were based only on trace element and isotope data.The high mantle potential temperature estimated for the KitaMatsuura basalt (414508C) suggests that the mantle upwelling responsible for the Kita-Matsuura volcanism originates from depths that, at the very least, are deeper than the source mantle of MORB. Obayashi et al. (2009) observed a discontinuity in the subducting Pacific Plate beneath Japan; this trends in an east^west direction from southwestern Japan to the Yellow Sea at a depth of 300^700 km (Fig. 16a). They interpreted this discontinuity to be a rupture in the Pacific slab, which is expected based on the geometry of plate motions. Obayashi et al. (2006) also observed a low P-wave velocity region under the Pacific Plate on the oceanward side of northern Honshu extending from 660 km depth, which can be observed in the seismic tomography of Huang & Zhao (2006) (Fig. 16a). They attributed this low-velocity anomaly to a high-temperature anomaly associated with a small amount of melt related to hot mantle upwelling from the lower mantle. If this is the case, the hot upwelling through the lower mantle may have stopped rising once it reached the stagnant slab at the 660 km discontinuity, marking the base of the upper mantle. The upwelling may then have split and spread horizontally to the east along the bottom of the slab (Fig. 16b). If the rupture in the stagnant slab was already present below northwestern Kyushu at 9 Ma, some of the material upwelling from As noted above, there are several mechanisms that can account for the temporal and spatial changes in the chemistry, pressure, temperature, degree of melting, and H2O content of the Kita-Matsuura basalts. Three possible mechanisms can introduce fluid progressively into the hot and dry upwelling mantle. These include (1) the progressive introduction of fluid from the deep-seated (mantle Transition Zone, MTZ) Pacific Plate slab, (2) interaction of upwelling deep, hot mantle with pre-existing metasomatized, hydrous shallow mantle lithosphere, and (3) entrainment of hydrous wedge mantle material by the upwelling mantle diapir. The first model is problematic because the sediment on the stagnant Pacific Plate slab in the MTZ may have a different composition from that sampled. In particular, the water content of the slab sediment would probably be much lower because of dehydration during subduction beneath the volcanic arc in Japan (Kimura et al., 2010) and the Marianas (Kelley et al., 2010). Instead, fluids may have been introduced from subducted serpentinite within the oceanic lithosphere or nominally anhydrous minerals within the MTZ (Richard & Bercovici, 2009). Alternatively, they could be related to melting of Khollandite in subducted slab sediments in the MTZ (Rapp et al., 2008), which released fluid to metasomatize the source of the intra-plate basalts above the stagnant Pacific slab as proposed for northeastern China (Kuritani et al., 2011). However, such a model would release the water at depth and does not readily explain the progressive addition of water inferred for the Kita-Matsuura basalts. The second ‘wet mantle lithosphere’ model maybe a possible candidate. As the onset of melting and water introduction begins at a pressure of 3 GPa and ends at 1·5 GPa, as 1118 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Fig. 16. (a) P-wave velocity anomalies around the Japan Sea at depths from 480 to 550 km (Obayashi et al., 2009) and an east^west vertical cross-section showing P-wave velocity anomalies along a latitude of 338N (Huang & Zhao, 2006). Red and blue colours denote low and high seismic velocity anomalies, respectively. Green dashed line indicates a plausible rupture of the Pacific Plate. Red dashed line represents 338N. (b) Schematic illustration of active upwelling from the lower mantle and penetration of the upwelling; (c, d) schematic illustration of passive upwelling from the mantle Transition Zone induced by tearing of the subducted Pacific Plate to generate the Late Miocene basaltic activity in northwestern Kyushu. Distribution of isotopic end-member components C1, C2, and C3 is shown in (c). discussed above, the depth range may correlate to the thickness of the continental lithosphere (Zhu, 2007). However, the high Tp ¼14508C of the Kita-Matsuura basalts throughout the melting regime precludes involvement of cool mantle lithosphere (T512008C or less under hydrous conditions). The third model was proposed by Sakkuyama et al. (2009) to explain the spatio-temporal variations in chemistry, pressure, and the degree of melting. In this study, we have further clarified the progressive addition of water and fluid-mobile elements through time. This new information does not violate the original model that we proposed in 2009, which still provides the most plausible mechanism to explain the geochemical data. In combination with the possibility of deep^hot mantle upwelling, as noted above, we propose again the mechanical entrainment of metasomatized asthenosphere by a rising mantle diapir. AC K N O W L E D G E M E N T S We are deeply grateful to Yoshiyuki Tatsumi and Hikaru Iwamori for scientific guidance and constructive discussions. Sincere thanks are extended to Hiroko Nagahara for extensive advice. Hideto Yoshida is also thanked for assistance with the electron microprobe. We greatly appreciate thoughtful and constructive reviews by Eiichi Takahashi, Jun-Ichi Kimura, Erin Todd, John Gamble, Marjorie Wilson and two anonymous reviewers. FU N DI NG Part of this work was supported by funds from Ministry of Education, Culture, Sports, Science and Technology of Japan (21540495 to M.Y. and 23740398 to T.S.). 1119 JOURNAL OF PETROLOGY VOLUME 55 R EF ER ENC ES Aizawa,Y., Tatsumi,Y. & Yamada, H. (1999). Element transport by dehydration of subducted sediments: Implication for arc and ocean island magmatism. Island Arc 8, 38^46. Alt, J. & Teagle, D. (2003). Hydrothermal alteration of upper oceanic crust formed at a fast-spreading ridge: mineral, chemical, and isotopic evidence from ODP Site 801. Chemical Geology 201, 191^211. Arai, S. (1987). An estimation of the least depleted spinel peridotite on the basis of olivine^spinel mantle array. Neues Jahrbuch fu«r Mineralogie, Monatshefte 8, 347^354. Arai, S. (1994). Compositional variation of olivine^chromian spinel in Mg-rich magmas as a guide to their residual spinel peridotites. Journal of Volcanology and Geothermal Research 59, 279^293. Baker, D. & Eggler, D. (1987). Compositions of anhydrous and hydrous melts coexisting with plagioclase, augite, and olivine or low-Ca pyroxene from 1 atm to 8 kbarçapplication to the Aleutian volcanic center of Atka. American Mineralogist 72, 12^28. Baker, M. B. & Stolper, E. M. (1994). Determining the composition of high-pressure mantle melts using diamond aggregates. Geochimica et Cosmochimica Acta 58, 2811^2827. Bartels, K. S., Kinzler, R. J. & Grove, T. L. (1991). High pressure phase relations of primitive high-alumina basalts from Medicine Lake volcano, northern California. Contributions to Mineralogy and Petrology 108, 253^270. Bence, A. E. & Albee, A. L. (1968). Empirical correction factors for the electron microanalysis of silicates and oxides. Journal of Geology 76, 382^403. Berndt, J., Koepke, J. & Holtz, F. (2005). An experimental investigation of the influence of water and oxygen fugacity on differentiation of MORB at 200 MPa. Journal of Petrology 46, 135^167. Brenan, J. M., Shaw, H. F., Phinney, D. L. & Ryerson, F. J. (1994). Rutile^fluid partitioning of Nb, Ta, Zr, U and Th: Implications for high-field-strength element depletions in island-arc basalts. Earth and Planetary Science Letters 128, 327^339. Currie, C. & Hyndman, R. (2006). The thermal structure of subduction zone back arcs. Journal of Geophysical Research 111, B08404, doi: 08410.01029/02005jb004024. Falloon, T. J., Green, D. H., Danyushevsky, L. V. & Faul, U. H. (1999). Peridotite melting at 1·0 and 1·5 GPa: an experimental evaluation of techniques using diamond aggregates and mineral mixes for determination of near-solidus melts. Journal of Petrology 40, 1343^1375. Farley, K. N. (1994). Oxidation state and sulfur concentrations in Lau Basin basalts. In: Hawkins, J., Parson, L. & Allan, J. et al. (eds) Proceedings of the Ocean Drilling Program, Scientific Results 135. College Station, TX: Ocean Drilling Program, pp. 603^613. Feig, S., Koepke, J. & Snow, J. (2006). Effect of water on tholeiitic basalt phase equilibria: an experimental study under oxidizing conditions. Contributions to Mineralogy and Petrology 152, 611^638. Feig, S. T., Koepke, J. & Snow, J. E. (2010). Effect of oxygen fugacity and water on phase equilibria of a hydrous tholeiitic basalt. Contributions to Mineralogy and Petrology 160, 551^568. Fram, M. & Longhi, J. (1992). Phase-equilibria of dikes associated with Proterozoic anorthosite complexes. American Mineralogist 77,605^616. Freise, M., Holtz, F., Nowak, M., Scoates, J. S. & Strauss, H. (2009). Differentiation and crystallization conditions of basalts from the Kerguelen large igneous province: an experimental study. Contributions to Mineralogy and Petrology 158, 505^527. Fukao, Y., Obayashi, M., Inoue, H. & Nenbai, M. (1992). Subducting slabs stagnant in the mantle transition zone. Journal of Geophysical Research 97, 4809^4822. Ghiorso, M. S. & Sack, R. O. (1995). Chemical mass transfer in magmatic processes IV. A revised and internally consistent NUMBER 6 JUNE 2014 thermodynamic model for the interpolation and extrapolation of liquid^solid equilibria in magmatic systems at elevated temperatures and pressures. Contributions to Mineralogy and Petrology 119, 197^212. Green, D. H. (1973). Experimental melting studies on a model upper mantle composition of high pressure under H2O-saturated and H,O-undersaturated conditions. Earth and Planetary Science Letters 19, 37^45. Griffiths, R. W. (1986). Particle motions induced by spherical convective elements in Stokes flow. Journal of Fluid Mechanics 166, 139^159. Grove, T., Elkins-Tanton, L., Parman, S., Chatterjee, N., Muntener, O. & Gaetani, G. (2003). Fractional crystallization and mantle-melting controls on calc-alkaline differentiation trends. Contributions to Mineralogy and Petrology 145, 515^533. Grove, T. L. & Bryan, W. B. (1983). Fractionation of pyroxene-phyric MORB at low pressure: An experimental study. Contributions to Mineralogy and Petrology 84, 293^309. Grove, T. L. & Juster, T. (1989). Experimental investigations of low-Ca pyroxene stability and olivine^pyroxene liquid equilibria at 1-atm in natural basaltic and andesitic liquids. Contributions to Mineralogy and Petrology 103, 287^305. Grove,T. L., Gerlach, D. C. & Sando,T.W. (1982). Origin of calc-alkaline series lavas at Medicine Lake Volcano by fractionation, assimilation and mixing. Contributions to Mineralogy and Petrology 80,160^182. Grove, T. L., Kinzler, R. J. & Bryan, W. B. (1990). Natural and experimental phase relations of lavas from Serocki volcano. In: Detrick, R. S. & Honnorez, J. et al. (eds) Proceedings of the Ocean Drilling Program, Scientific Results 106/109. College Station, TX: Ocean Drilling Program, pp. 9^17. Grove,T. L., Kinzler, R. J. & Bryan,W. B. (1992). Fractionation of midocean ridge basalt (MORB). In: Morgan, J. P., Blackman, D. K. & Sinton, J. M. (eds) Mantle Flow and Melt Generation at Mid-Ocean Ridges. American Geophysical Union, Geophysical Monograph 71,281^309. Hamada, M. & Fujii, T. (2007). H2O-rich island arc low-K tholeiite magma inferred from Ca-rich plagioclase^melt inclusion equilibria. Geochemical Journal 41, 437^461. Hamada, M. & Fujii, T. (2008). Experimental constraints on the effects of pressure and H2O on the fractional crystallization of highMg island arc basalt. Contributions to Mineralogy and Petrology 155, 767^790. Hart, S. R. (1984). A large-scale isotope anomaly in the SouthernHemisphere mantle. Nature 309, 753^757. Hauff, F., Hoernle, K. & Schmidt, A. (2003). Sr^Nd^Pb composition of Mesozoic Pacific oceanic crust (Site 1149 and 801, ODP Leg 185): Implications for alteration of ocean crust and the input into the Izu^Bonin^Mariana subduction system. Geochemistry, Geophysics, Geosystems 4, doi:10.1029/2002GC000421. Herzberg, C., Asimow, P. D., Arndt, N., Niu, Y. L., Lesher, C. M., Fitton, J. G., Cheadle, M. J. & Saunders, A. D. (2007). Temperatures in ambient mantle and plumes: Constraints from basalts, picrites, and komatiites. Geochemistry, Geophysics, Geosystems 8, Q02006, doi:10.1029/2006GC001390. Hickey-Vargas, R. (1991). Isotope characteristics of submarine lavas from the Philippine Sea: implications for the origin of arc and basin magmas of the Philippine tectonic plate. Earth and Planetary Science Letters 107, 290^304. Hirose, K. (1997). Melting experiments on lherzolite KLB-1 under hydrous conditions and generation of high-magnesian andesitic melts. Geology 25, 42^44. Hirose, K. & Kawamoto, T. (1995). Hydrous partial melting of lherzolite at 1 GPaçThe effect of H2O on the genesis of basaltic magmas. Earth and Planetary Science Letters 133, 463^473. 1120 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Hirose, K. & Kawamura, K. (1994). A new experimental approach for incremental batch melting of peridotite at 1·5 GPa. Geophysical Research Letters 21, 2139^2142. Hirose, K. & Kushiro, I. (1993). Partial melting of dry peridotites at high pressures: Determination of compositions of melts segregated from peridotite using aggregates of diamond. Earth and Planetary Science Letters 114, 477^489. Hirose, K. & Kushiro, I. (1998). The effect of melt segregation on polybaric mantle melting: Estimation from the incremental melting experiments. Physics of the Earth and Planetary Interiors 107, 111^118. Hirschmann, M. M. (2000). Mantle solidus: Experimental constraints and the effects of peridotite composition. Geochemistry, Geophysics, Geosystems 1, doi:10.1029/2000GC000070. Hofmann, A. W. (1997). Mantle geochemistry: The message from oceanic volcanism. Nature 385, 219^229. Holland, T. & Powell, R. (1992). Plagioclase feldsparsçActivity^composition relations based upon Darken quadratic formalism and Landau theory. American Mineralogist 77, 53^61. Honma, U. (2012). Hydrous and anhydrous melting experiments of an alkali basalt and a transitional tholeiite from the Oginosen volcano, Southwest Japan: The possible influence of melt depolymerization on Ca^Na partitioning between plagioclase and the melt. Journal of Mineralogical and Petrological Sciences 107, 8^32. Housh, T. B. & Luhr, J. F. (1991). Plagioclase^melt equilibria in hydrous systems. American Mineralogist 76, 477^492. Huang, J. & Zhao, D. (2006). High-resolution mantle tomography of China and surrounding regions. Journal of Geophysical Research 111, doi:10.1029/2005JB004066. Ichiki, M., Baba, K., Obayashi, M. & Utada, H. (2006). Water content and geotherm in the upper mantle above the stagnant slab: Interpretation of electrical conductivity and seismic P-wave velocity models. Physics of the Earth and Planetary Interiors 155, 1^15. Ignacio, C., Lopez, I., Oyarzun, R. & Marquez, A. (2001). The northern Patagonia Somuncura plateau basalts: a product of slab-induced, shallow asthenospheric upwelling? Terra Nova 13, 117^121. Inoue, T. (1994). Effect of water on melting phase-relations and melt composition in the system Mg2SiO4^MgSiO3^H2O up to 15 GPa. Physics of the Earth and Planetary Interiors 85, 237^263. Inoue, T. & Sawamoto, H. (1992). High pressure melting of pyrolite under hydrous condition and its geophysical implications. In: Syono, Y. & Manghnani, M. H. (eds) High-Pressure Research: Application to Earth and Planetary Sciences. American Geophysical Union, Geophysical Monograph 67, 323^331. Irvine, T. N. & Baragar, W. R. (1971). A guide to the chemical classification of the common volcanic rocks. Canadian Journal of Earth Sciences 8, 523^548. Iwamori, H. (1991). Zonal structure of Cenozoic basalts related to mantle upwelling in southwest Japan. Journal of Geophysical Research 96, 6157^6170. Iwamori, H., McKenzie, D. & Takahashi, E. (1995). Melt generation by isentropic mantle upwelling. Earth and Planetary Science Letters 134, 253^266. Jaques, A. L. & Green, D. H. (1980). Anhydrous melting of peridotite at 0^15 kb pressure and the genesis of tholeiitic basalts. Contributions to Mineralogy and Petrology 73, 287^310. Johnson, K. T. M., Dick, H. J. B. & Shimizu, N. (1990). Melting in the oceanic upper mantle: an ion microprobe study of diopsides in abyssal peridotites. Journal of Geophysical Research 95, 2661^2678. Jolivet, L., Tamaki, K. & Fournier, M. (1994). Japan Sea, opening history and mechanismça synthesis. Journal of Geophysical Research 99, 22237^22259. Juster, T., Grove, T. L. & Perfit, M. (1989). Experimental constraints on the generation of Fe^Ti basalts, andesites, and rhyodacites at the Galapagos Spreading Center, 858W and 958W. Journal of Geophysical Research 94, 9251^9274. Katz, R. F., Spiegelman, M. & Langmuir, C. H. (2003). A new parameterization of hydrous mantle melting. Geochemistry, Geophysics, Geosystems 4(9), doi:10.1029/2002GC000433. Kelemen, P., Johnson, K. T. M., Kinzler, R. J. & Irving, A. J. (1990). High-field-strength element depletions in arc basalts due to mantle^magma interaction. Nature 345, 521^524. Kelemen, P., Shimizu, N. & Dunn, T. (1993). Relative depletion of niobium in some arc magmas and the continental crust; partitioning of K, Nb, La and Ce during melt/rock reaction in the upper mantle. Earth and Planetary Science Letters 120, 111^134. Kelley, K., Plank, T., Ludden, J. & Staudigel, H. (2003). Composition of altered oceanic crust at ODP Sites 801 and 1149. Geochemistry, Geophysics, Geosystems 4(6), 2002GC000435. Kelley, K. A., Plank, T., Newman, S., Stolper, E. M., Grove, T. L., Parman, S. & Hauri, E. H. (2010). Mantle melting as a function of water content beneath the Mariana Arc. Journal of Petrology 51, 1711^1738. Kessel, R., Schmidt, M. W., Ulmer, P. & Pettke, T. (2005). Trace element signature of subduction-zone fluids, melts and supercritical liquids at 120^180 km depth. Nature 437, 724^727. Kimura, J.-I., Stern, R. J. & Yoshida, T. (2005). Reinitiation of subduction and magmatic responses in SW Japan during Neogene time. Geological Society of America Bulletin 117, 969^986. Kimura, J. I., Kent, A. J. R., Rowe, M. C., Katakuse, M., Nakano, F., Hacker, B. R., van Keken, P. E., Kawabata, H. & Stern, R. J. (2010). Origin of cross-chain geochemical variation in Quaternary lavas from the northern Izu arc: Using a quantitative mass balance approach to identify mantle sources and mantle wedge processes. Geochemistry, Geophysics, Geosystems 11, doi:10.1029/2010GC003050. Kinzler, R. J. & Grove, T. L. (1992). Primary magmas of mid-ocean ridge basalts. 1. Experiments and methods. Journal of Geophysical Research 97, 6885^6906. Koepke, J., Feig, S. T., Snow, J. & Freise, M. (2004). Petrogenesis of oceanic plagiogranites by partial melting of gabbros: an experimental study. Contributions to Mineralogy and Petrology 146, 414^432. Kogiso, T., Tatsumi, Y. & Nakano, S. (1997). Trace element transport during dehydration processes in the subducted oceanic crust. 1. Experiments and implications for the origin of ocean island basalts. Earth and Planetary Science Letters 148, 193^205. Kogiso, T., Hirose, K. & Takahashi, E. (1998). Melting experiments on homogeneous mixtures of peridotite and basalt: application to the genesis of ocean island basalts. Earth and Planetary Science Letters 162, 45^61. Kuritani, T., Kimura, J. I., Miyamoto, T., Wei, H. Q., Shimano, T., Maeno, F., Jin, X. & Taniguchi, H. (2009). Intraplate magmatism related to deceleration of upwelling asthenospheric mantle: Implications from the Changbaishan shield basalts, northeast China. Lithos 112, 247^258. Kuritani, T., Ohtani, E. & Kimura, J.-I. (2011). Intensive hydration of the mantle transition zone beneath China caused by slab stagnation. Nature Geoscience 4, 713^716. Kushiro, I. (1996). Partial melting of a fertile mantle peridotite at high pressures: an experimental study using aggregates of diamond. In: Basu, A. & Hart, S. (eds) Earth Processes: Reading the Isotopic Code. American Geophysical Union, Geophysical Monograph 95, 109^122. Kushiro, I., Yoder, H. S. & Nishikawa, M. (1968). Effect of water on the melting of enstatite. Geological Society of America Bulletin 79, 1685^1692. 1121 JOURNAL OF PETROLOGY VOLUME 55 Lange, R. A., Frey, H. M. & Hector, J. (2009). A thermodynamic model for the plagioclase^liquid hygrometer/thermometer. American Mineralogist 94, 494^506. Lee, C. T. A., Luffi, P., Plank, T., Dalton, H. & Leeman, W. P. (2009). Constraints on the depths and temperatures of basaltic magma generation on Earth and other terrestrial planets using new thermobarometers for mafic magmas. Earth and Planetary Science Letters 279, 20^33. Letouzey, J. & Kimura, M. (1985). Okinawa trough genesis: structure and evolution of a backarc basin developed in a continent. Marine and Petroleum Geology 2, 111^130. Liu, M., Cui, X. & Liu, F. (2004). Cenozoic rifting and volcanism in eastern China: a mantle dynamic link to the Indo-Asian collision? Tectonophysics 393, 29^42. Maaloe, S. (2004). The solidus of harzburgite to 3 GPa pressure: the compositions of primary abyssal tholeiite. Mineralogy and Petrology 81, 1^17. McKenzie, D. (1984). The generation and compaction of partially molten rock. Journal of Petrology 25, 713^765. McKenzie, D. & Bickle, M. (1988). The volume and composition of melt generated by extension of the lithosphere. Journal of Petrology 29, 625^679. Medard, E. & Grove, T. (2008). The effect of H2O on the olivine liquidus of basaltic melts: experiments and thermodynamic models. Contributions to Mineralogy and Petrology 155, 417^432. Meen, J. K. (1987). Formation of shoshonites from calcalkaline basalt magmasçGeochemical and experimental constraints from the type locality. Contributions to Mineralogy and Petrology 97, 333^351. Meen, J. K. (1990). Elevation of potassium content of basaltic magma by fractional crystallization: the effect of pressure. Contributions to Mineralogy and Petrology 104, 309^331. Miyashiro, A. (1986). Hot regions and the origin of marginal basins in the western Pacific. Tectonophysics 122, 195^216. Mu«ller, R. D., Sdrolias, M., Gaina, C., Steinberger, B. & Heine, C. (2008). Long-term sea-level fluctuations driven by ocean basin dynamics. Science 319, 1357^1362. Muntener, O., Kelemen, P. B. & Grove, T. L. (2001). The role of H2O during crystallization of primitive arc magmas under uppermost mantle conditions and genesis of igneous pyroxenites: an experimental study. Contributions to Mineralogy and Petrology 141, 643^658. Nakamura, E., Campbell, I. H. & McCulloch, M. T. (1990). Chemical geodynamics in the back-arc region of Japan based on the trace element and Sr^Nd isotopic compositions. Tectonophysics 174, 207^233. Nakamura, H., Iwamori, H. & Kimura, J. I. (2008). Geochemical evidence for enhanced fluid flux due to overlapping subducting plates. Nature Geoscience 1, 380^384. Nakamura, K., Kato, Y., Tamaki, K. & Ishii, T. (2007). Geochemistry of hydrothermally altered basaltic rocks from the Southwest Indian Ridge near the Rodriguez Triple Junction. Marine Geology 239, 125^141. Nakamura, Y. & Kushiro, I. (1970). Compositional relations of coexisting orthopyroxene, pigeonite and augite in a tholeiitic andesite from Hakone volcano. Contributions to Mineralogy and Petrology 26, 265^275. Nilsson, K. & Peach, C. L. (1993). Sulfur speciation, oxidation-state, and sulfur concentration in backarc magmas. Geochimica et Cosmochimica Acta 57, 3807^3813. Obayashi, M., Sugioka, H., Yoshimitsu, J. & Fukao, Y. (2006). High temperature anomalies oceanward of subducting slabs at the 410km discontinuity. Earth and Planetary Science Letters 243, 149^158. Obayashi, M., Yoshimitsu, J. & Fukao, Y. (2009). Tearing of stagnant slab. Science 324, 1173^1175. NUMBER 6 JUNE 2014 Ozawa, K. (2001). Mass balance equations for open magmatic systems: Trace element behavior and its application to open system melting in the upper mantle. Journal of Geophysical Research 106, 13407^13434. Ozawa, K. & Shimizu, N. (1995). Open system melting in the upper mantle: Constraints from the Hayachine-Miyamori ophiolite, northeastern Japan. Journal of Geophysical Research 100, 22315^22335. Parman, S. W., Grove, T. L., Kelley, K. A. & Plank, T. (2011). Alongarc variations in the pre-eruptive H2O contents of Mariana Arc magmas inferred from fractionation paths. Journal of Petrology 52, 257^278. Pearce, J. A. & Parkinson, I. J. (1993). Trace element models for mantle melting: Application to volcanic arc petrogenesis. In: Prichard, H. M., Alabaster, T., Harris, N. B. W. & Neary, C. R. (eds) Magmatic Processes and Plate Tectonics. Geological Society, London, Special Publications 76, 373^403. Pertermann, M. & Lundstrom, C. (2006). Phase equilibrium experiments at 0·5 GPa and 1100^13008C on a basaltic andesite from Arenal volcano, Costa Rica. Journal of Volcanology and Geothermal Research 157, 222^235. Pickering-Witter, J. & Johnston, A. D. (2000). The effects of variable bulk composition on the melting systematics of fertile peridotitic assemblages. Contributions to Mineralogy and Petrology 140, 190^211. Plank, T. & Langmuir, C. (1998). The chemical composition of subducting sediment and its consequences for the crust and mantle. Chemical Geology 145, 325^394. Plank, T., Kelley, K., Murray, R. & Stern, L. (2007). Chemical composition of sediments subducting at the Izu^Bonin trench. Geochemistry, Geophysics, Geosystems 8(4), 2006GC001444. Putirka, K. D. (2005). Mantle potential temperatures at Hawaii, Iceland, and the mid-ocean ridge system, as inferred from olivine phenocrysts: Evidence for thermally driven mantle plumes. Geochemistry, Geophysics, Geosystems 6, Q05108, doi:10.1029/ 2005GC000915. Putirka, K. D., Perfit, M., Ryerson, F. J. & Jackson, M. G. (2007). Ambient and excess mantle temperatures, olivine thermometry, and active vs. passive upwelling. Chemical Geology 241, 177^206. Rafferty, W. J. & Heming, R. F. (1979). Quaternary alkalic and sub-alkalic volcanism in south Auckland, New Zealand. Contributions to Mineralogy and Petrology 71, 139^150. Ramsay, W. R. H., Crawford, A. J. & Foden, J. D. (1984). Field setting, mineralogy, chemistry, and genesis of arc picrites, New Georgia, Solomon Islands. Contributions to Mineralogy and Petrology 88, 386^402. Rapp, R. P., Irifune, T., Shimizu, N., Nishiyama, N., Norman, M. D. & Inoue, J. (2008). Subduction recycling of continental sediments and the origin of geochemically enriched reservoirs in the deep mantle. Earth and Planetary Science Letters 271, 14^23. Richard, G. C. & Bercovici, D. (2009). Water-induced convection in the Earth’s mantle transition zone. Journal of Geophysical Research 114, doi:10.1029/2008JB005734. Richard, G. C. & Iwamori, H. (2010). Stagnant slab, wet plumes and Cenozoic volcanism in East Asia. Physics of the Earth and Planetary Interiors 183, 280^287. Robinson, J. A. C., Wood, B. J. & Blundy, J. D. (1998). The beginning of melting of fertile and depleted peridotite at 1·5 GPa. Earth and Planetary Science Letters 155, 97^111. Robinson, P., Townsend, A. T., Yu, Z. S. & Munker, C. (1999). Determination of scandium, yttrium and rare earth elements in rocks by high resolution inductively coupled plasma-mass spectrometry. Geostandards Newsletter 23, 31^46. 1122 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Ryerson, F. J. & Watson, E. B. (1987). Rutile saturation in magmas: implications for Ti^Nb^Ta depletion in island-arc basalts. Earth and Planetary Science Letters 86, 225^239. Sakuyama, T. (2010). Cenozoic tectonics and volcanism in northern Kyushu: Significance for studies on tectonic magma provinces. Journal of Geography 119, 224^234. Sakuyama, T., Ozawa, K., Sumino, H. & Nagao, K. (2009). Progressive melt extraction from upwelling mantle constrained by the Kita-Matsuura basalts in NW Kyushu, SW Japan. Journal of Petrology 50, 725^779. Sano, T. & Yamashita, S. (2004). Experimental petrology of basement lavas from Ocean Drilling Program Leg 192: implications for differentiation processes in Ontong^Java Plateau magmas. In: Fitton, J. G., Mahoney, J. J., Wallace, P. J. & Saunders, A. D. (eds) Origin and Evolution of the Ontong^Java Plateau. Geological Society, London, Special Publications 229, 185^218. Sano, T., Fujii, T., Deshmukh, S. S., Fukuoka, T. & Aramaki, S. (2001). Differentiation processes of Deccan Trap basalts: contribution from geochemistry and experimental petrology. Journal of Petrology 42, 2175^2195. Saunders, A. D., Tarney, J. & Weaver, S. D. (1980). Transverse geochemical variations across the Antarctic Peninsulaçimplications for the genesis of calc-alkaline magmas. Earth and Planetary Science Letters 46, 344^360. Schwab, B. E. & Johnston, A. D. (2001). Melting systematics of modally variable, compositionally intermediate peridotites and the effects of mineral fertility. Journal of Petrology 42, 1789^1811. Scoates, J. S., Cascio, M. L., Weis, D. & Lindsley, D. H. (2006). Experimental constraints on the origin and evolution of mildly alkalic basalts from the Kerguelen Archipelago, Southeast Indian Ocean. Contributions to Mineralogy and Petrology 151, 582^599. Shibata, T., Yoshikawa, M. & Tatsumi, Y. (2003). An analytical method for determining precise Sr and Nd isotopic compositions and results for thirteen rock standard materials. Frontier Research on Earth Evolution 1, 363^367. Shimoda, G., Tatsumi, Y., Nohda, S., Ishizaka, K. & Jahn, B. (1998). Setouchi high-Mg andesites revisited: geochemical evidence for melting of subducting sediments. Earth and Planetary Science Letters 160, 479^492. Sisson, T. W. & Grove, T. L. (1993). Experimental investigations of the role of H2O in calc-alkaline differentiation and subduction zone magmatism. Contributions to Mineralogy and Petrology 113, 143^166. Snyder, D., Carmichael, I. S. E. & Wiebe, R. A. (1993). Experimental study of liquid evolution in an Fe-rich, layered mafic intrusion: constraints of Fe^Ti oxide precipitation on the T^fO2 and T^P paths of tholeiitic magmas. Contributions to Mineralogy and Petrology 113, 73^86. Sobolev, A. V. & Shimizu, N. (1993). Ultra-depleted primary melt included in an olivine from the Mid-Atlantic Ridge. Nature 363, 151^154. Straub, S. M., Goldstein, S. L., Class, C. & Schmidt, A. (2009). Midocean-ridge basalt of Indian type in the northwest Pacific Ocean basin. Nature Geoscience 2, 286^289. Sugawara, T. (2000). Empirical relationships between temperature, pressure, and MgO content in olivine and pyroxene saturated liquid. Journal of Geophysical Research 105, 8457^8472. Sumino, H., Nakai, S. i., Nagao, K. & Notsu, K. (2000). High 3 He/4He ratio in xenolith from Takashima: evidence for plume type volcanism in southwestern Japan. Geophysical Research Letters 27, 1211^1214. Sun, S.-S. & McDonough, W. F. (1989). Chemical and isotopic systematics of oceanic basalts: implications for mantle composition and processes. In: Saunders, A. D. & Norry, M. J. (eds) Magmatism in the Ocean Basins, Geological Society, London, Special Publications 42, 313^345. Takagi, D., Sato, H. & Nakagawa, M. (2005). Experimental study of a low-alkali tholeiite at 1^5 kbar: optimal condition for the crystallization of high-An plagioclase in hydrous arc tholeiite. Contributions to Mineralogy and Petrology 149, 527^540. Takahashi, E. & Kushiro, I. (1983). Melting of a dry peridotite at high pressures and basalt magma genesis. American Mineralogist 68, 859^879. Takahashi, E., Shimazaki, T., Tsuzaki, Y. & Yoshida, H. (1993). Melting study of a peridotite KLB-1 to 6·5 GPa, and the origin of basaltic magmas. Philosophical Transactions of the Royal Society of London, Series A 342, 105^120. Tamaki, K., Suyehiro, K., Allan, J., Ingle, J. & Pisciotto, K. A. (1992). Tectonic synthesis and implications of Japan Sea ODP drilling. In: Tamaki, K., Suyehiro, K. & Allan, J. (eds) Proceedings of the Ocean Drilling Program, Scientific Results 127/128. College Station, TX: Ocean Drilling Program, pp. 1333^1348. Tamura, Y., Yuhara, M. & Ishii, T. (2000). Primary arc basalts from Daisen volcano, Japan: Equilibrium crystal fractionation versus disequilibrium fractionation during supercooling. Journal of Petrology 41, 431^448. Tatsumi, Y. & Suzuki, T. (2009). Tholeiitic vs calc-alkalic differentiation and evolution of arc crust: constraints from melting experiments on a basalt from the Izu^Bonin^Mariana Arc. Journal of Petrology 50, 1575^1603. Tatsumi, Y., Sakuyama, M., Fukuyama, H. & Kushiro, I. (1983). Generation of arc magmas and thermal structure of the mantle wedge in subduction zones. Journal of Geophysical Research 88, 5815^5825. Tatsumi, Y., Hamilton, D. L. & Nesbitt, R. W. (1986). Chemical characteristics of fluid phase released from a subducted lithosphere and origin of arc magmasçevidence from high-pressure experiments and natural rocks. Journal of Volcanology and Geothermal Research 29, 293^309. Tatsumi, Y., Furukawa, Y. & Yamashita, S. (1994). Thermal and geochemical evolution of the mantle wedge in the northeast Japan arc 1. Contribution from experimental petrology. Journal of Geophysical Research 99, 22275^22283. Thorkelson, D. J. & Taylor, R. P. (1989). Cordilleran slab windows. Geology 17, 833^836. Toplis, M. & Carroll, M. (1995). An experimental study of the influence of oxygen fugacity on Fe^Ti oxide stability, phase relations, and mineral^melt equilibria in ferro-basaltic systems. Journal of Petrology 36, 1137^1170. Tormey, D., Grove, T. & Bryan, W. (1987). Experimental petrology of normal MORB near the Kane Fracture Zone: 228^258N, Mid-Atlantic Ridge. Contributions to Mineralogy and Petrology 96, 121^139. Uto, K., Hoang, N. & Matsui, K. (2004). Cenozoic lithospheric extension induced magmatism in Southwest Japan. Tectonophysics 393, 281^299. Walter, M. J. (1998). Melting of garnet peridotite and the origin of komatiite and depleted lithosphere. Journal of Petrology 39, 29^60. Walter, M. J., Sisson, T. W. & Presnall, D. C. (1995). A mass proportion method for calculating melting reactions and application to melting of model upper-mantle lherzolite. Earth and Planetary Science Letters 135, 77^90. Wiens, D. A., Kelley, K. A. & Plank, T. (2006). Mantle temperature variations beneath back-arc spreading centers inferred from seismology, petrology, and bathymetry. Earth and Planetary Science Letters 248, 30^42. 1123 JOURNAL OF PETROLOGY VOLUME 55 Wood, D. A., Joron, J. L., Treuil, M., Norry, M. & Tarney, J. (1979). Elemental and Sr isotope variations in basic lavas from Iceland and the surrounding ocean-floorçnature of mantle source inhomogeneities. Contributions to Mineralogy and Petrology 70, 319^339. Workman, R. K., Hart, S. R., Jackson, M., Regelous, M., Farley, K. A., Blusztajn, J., Kurz, M. & Staudigel, H. (2004). Recycled metasomatized lithosphere as the origin of the enriched mantle II (EM2) end-member: Evidence from the Samoan volcanic chain. Geochemistry, Geophysics, Geosystems 5, Q04008, doi:04010.01029/ 02003GC000623. Yamashita, S. & Fujii, T. (1992). Experimental petrology of basement basaltic rocks from Sites 794 and 797, Japan Sea. In: Tamaki, K., Suyehiro, K. & Allan, J. et al. (eds) Proceedings of the Ocean Drilling Program, Scientific Results 127/128. College Station, TX: Ocean Drilling Program, pp. 891^898. Yang, H.-J., Kinzler, R. J. & Grove, T. L. (1996). Experiments and models of anhydrous, basaltic olivine^plagioclase^augite saturated melts from 0·001 to 10 kbar. Contributions to Mineralogy and Petrology 124, 1^18. Yoshikawa, M., Shibata, T. & Tatsumi, Y. (2001). The Sr, Nd and Pb isotopic ratios of GSJ standard rocks. Annual Report of Beppu Geothermal Research Laboratory, Kyoto University FY2000, 30. Zeng, G., Chen, L.-H., Hofmann, A. W., Jiang, S.-Y. & Xu, X.-S. (2011). Crust recycling in the sources of two parallel volcanic chains in Shandong, north China. Earth and Planetary Science Letters 302, 359^368. Zhao, D., Maruyama, S. & Omori, S. (2007). Mantle dynamics of Western Pacific and East Asia: Insight from seismic tomography and mineral physics. Gondwana Research 11, 120^131. Zhao, D. P., Yanada, T., Hasegawa, A., Umino, N. & Wei, W. (2012). Imaging the subducting slabs and mantle upwelling under the Japan Islands. Geophysical Journal International 190, 816^828. Zhu, J. (2007). The structural characteristics of lithosphere in the continent of Eurasia and its marginal seas. Earth Science Frontiers 14, 1^20. Zindler, A. & Hart, S. (1986). Chemical geodynamics. Annual Review of Earth and Planetary Sciences 14, 493^571. Zou, H. B., Fan, Q. C. & Yao, Y. P. (2008). U^Th systematics of dispersed young volcanoes in NE China: Asthenosphere upwelling caused by piling up and upward thickening of stagnant Pacific slab. Chemical Geology 255, 134^142. A P P E N D I X A : C O M PA R I S O N O F T R AC E E L E M E N T C O N C E N T R AT I O N S A N A LY Z E D B Y I C P- M S A N D X R F Results of the comparison are shown in Fig. 17. Rb, Ba, Th, Nb, and La show good agreement. Sr, Zr, and Y are higher by 5^10% in the XRF analyses, but still show reasonable correlation. The discrepancy in Y between XRF and ICP-MS could have originated in errors in the standard values (Robinson et al., 1999). Pb shows a weak correlation between the XRF and ICP-MS data, as the precision of Pb measured by XRF is much less than that achieved using ICP-MS, resulting in up to 50% difference. Elemental ratios used in this study (Zr/Y, Nb/Y, and Nb/Th) show good correlation and the NUMBER 6 JUNE 2014 deviations are smaller than 15%. These differences are much smaller than the variation in each SiO2 group. For example, if we used ICP-MS data instead of XRF data for modelling of the Nb/Y^Zr/Y variations, similar to Sakuyama et al. (2009), the degree of critical melting based on ICP-MS data would be systematically at most 0·5% lower than that for the XRF data, which is negligibly small. APPENDIX B: G E O T H E R M O M E T RY AT H I G H PRESSU RE Relationships are observed amongst temperature, pressure and melt MgO content in high-pressure anhydrous peridotite melting experiments where the melt is in equilibrium with olivine and pyroxenes (Fig. 18). The data plotted are taken from the published literature (Jaques & Green, 1980; Hirose & Kushiro, 1993; Takahashi et al., 1993; Baker & Stolper, 1994; Hirose & Kawamura, 1994; Kushiro, 1996; Hirose & Kushiro, 1998; Robinson et al., 1998; Walter, 1998; Falloon et al., 1999; Pickering-Witter & Johnston, 2000; Schwab & Johnston, 2001). Regression lines for each pressure calculated using equation (3) from this study are shown in Fig. 18a and b. The relationship between melt fraction and temperature as a function of pressure is shown in Fig. 18c. Regression lines at each pressure are almost parallel to each other with good correlation coefficients, most of which are greater than 0·9, as suggested by Maaloe (2004). Because the uncertainty of pressure estimation is 0·3 GPa and that of the MgO content of the melt is 1·0 wt %, the uncertainty on the temperature is 308C. Temperatures calculated following the method of Sugawara (2000) are consistent with our estimates according to equation (3). The largest and average differences obtained between equation (3) and Sugawara (2000) were 118C and 38C, respectively. By using the MgO content in experimentally produced partial melts of a peridotite^basalt hybrid source (KG1; Kogiso et al., 1998) and the melting pressure estimated by normative projection of the partial melts, we calculated a temperature from equation (3) to compare with the actual experimental temperature. The calculated temperature is lower than the experimental temperature for KG1 at 3·0 GPa by up to 358C, whereas the estimated temperatures for pressures lower than 3·0 GPa are lower than the experimental temperature by only 108C on average and the 1s is 158C. The estimation method for melting temperature according to equation (3) is, therefore, applicable to peridotite systems with a basaltic component less than 50%. 1124 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING Fig. 17. Comparison of trace element concentrations and ratios analyzed by inductively coupled plasma mass spectrometry (ICP-MS) and X-ray fluorescence (XRF). 1125 JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 JUNE 2014 Fig. 18. (a) Temperature (8C) vs MgO content (wt %) of olivine- and clinopyroxene-saturated melt in anhydrous peridotite melting experiments. Continuous lines are linear regression lines for each pressure. (b) Temperature (8C) vs MgO content (wt %) of experimental melts plotted in (a) calculated using equation (3). (c) Melt fraction vs temperature at different pressures for the experiments plotted in (a). 1126 SAKUYAMA et al. MANTLE INTERACTION DURING DIAPIRIC UPWELLING A PPEN DI X C: POSSI BI L I T Y FOR A M E LT I N G M O D E L I N A S Y S T E M C LOS E D TO I N P U T Fig. 19. (a) Na2O þ K2O vs SiO2 for the Kita-Matsuura basalts compared with melt compositions from the high-pressure experiments that were used for testing the geothermometer from this study. (b) Comparison of temperature predicted by the geothermometer with that measured in the experiments. (c) Comparison of H2O contents in the melt predicted by the hygrometer used in this study with those reported in the experiments. The presence of HFSE-rich minerals as residual minerals in the melting system produces a melt with negative HFSE anomalies. Any titanium-bearing minerals, however, cannot remain as a residual phase in the melting system at high degrees of melting (7 wt % as estimated above), as their solubility is high in a basaltic melt (Ryerson & Watson, 1987). Even if a titanium-bearing phase remained in the melting system, the magnitude of the HFSE depletion relative to LILE and LREE rapidly decreases as the degree of melting increases because highly incompatible LILE and LREE decrease faster than HFSE. In the Kita-Matsuura basalts, the extent of relative depletion in HFSE increases as the abundance of HFSE decreases. This relationship between HFSE and the degree of melting does not depend on the nature of the melting process (e.g. batch or fractional melting). Therefore the possibility of the existence of a residual HFSE-rich mineral is rejected as being the cause of the geochemical characteristics of the Kita-Matsuura basalts. Melt^rock reactions in the upper mantle may have the potential to generate HFSE-depleted high-magnesian andesite and calc-alkaline magma series rocks (e.g. Kelemen et al., 1993; Grove et al., 2003). To increase HFSE depletion as their abundance decreases, the assimilated mass must be greater than the crystallized mass (Kelemen et al., 1993); however, this process also decreases the volatile content of the magma as the reaction proceeds. This is not the case for the Kita-Matsuura basalts, as the estimated water content of the primary magmas increases as the HFSE content of the magma decreases. In addition, crystal dissolution from the surrounding peridotite by a percolating melt, which results in an increase in the melt mass, would be enhanced along an inverted geothermal gradient in the mantle, such as in the mantle wedge above a subducting slab. When the Kita-Matsuura basaltic magmatism was active, the KitaMatsuura area was situated in the back-arc region, where the thermal gradient in the upper mantle must be different from that near the wedge corner. Melt^rock reactions, which may preferentially occur in the wedge corner where the thermal gradient is inverted, therefore, may not be effective. If a series of mafic lithologies with identical compositions to PM1, PM2 and PM3 were present in the peridotite host (heterogeneous mantle) and if they completely melted to produce each melt of the Kita-Matsuura basalts, the SiO2-poor basalts should have formed first followed by the SiO2-rich basalts within each section in order to explain the observed temporal major element variations. 1127 JOURNAL OF PETROLOGY VOLUME 55 It is, however, difficult for a heterogeneous mantle source to reproduce such temporal variations, as both the solidus and liquidus temperatures of a mafic lithology with the PM3 chemical composition and a higher water content are estimated to be lower than those of lithologies with a PM1 chemical composition and lower water content according to pMELTS calculations. This is inconsistent with the observed systematic temporal variation of the least-fractionated basalts of the Kita-Matsuura basalt, as the source region for each section was homogeneous in temperature. A PPEN DI X D: A PPLICA BI LI T Y OF M AG M A T H E R M OM ET E R A N D P L A G I O C L A S E ^ M E LT H Y G RO M E T E R AT L O W PRESSU RE We compiled data from experimental studies conducted under hydrous conditions and applied our method to estimate the temperature and the water content. Experimental studies that we used are as follows: Grove et al. (1982, 1990, 1992); Grove & Bryan (1983); Baker & Eggler (1987); Meen (1987, 1990); Tormey et al. (1987); Grove & Juster (1989); Juster et al. (1989); Bartels et al. (1991); Fram & Longhi (1992); Kinzler & Grove (1992); Sisson & Grove (1993); Snyder et al. (1993); Toplis & Carroll (1995); Yang NUMBER 6 JUNE 2014 et al. (1996); Muntener et al. (2001); Sano et al. (2001); Koepke et al. (2004); Sano & Yamashita (2004); Berndt et al. (2005); Takagi et al. (2005); Feig et al. (2006, 2010); Pertermann & Lundstrom (2006); Scoates et al. (2006); Hamada & Fujii (2007, 2008); Freise et al. (2009); Tatsumi & Suzuki (2009); Parman et al. (2011); Honma (2012). The experimental pressure and temperature and the water contents in the melt varied from 0 to 1·5 GPa, from 943 to 12958C, and from 0 to 6·0 wt %, respectively. The range of the experimental melt compositions was subalkalic to alkalic (Fig. 19a) with SiO2446·9 wt %. However, the minimum value of the SiO2 content of the experimental melts under hydrous conditions was 48·5 wt %, which is 1wt % higher than that of the Kita-Matsuura basalts (47·5 wt %). To check how reliably this method can be applied to the Kita-Matsuura basalts, the experiments were filtered by An# [¼ 100 Ca/(Ca þ Na)] from 60 to 90 and SiO2 contents less than 60 wt %. Calculated temperature and H2O contents in the melt were compared with experimental temperature and H2O content in melt (Fig. 19b and c). Temperature estimates show a good agreement with experimental temperatures (Fig. 19b); the regression line is y ¼ 0·9951x and R2 ¼ 0·8975. In contrast, the estimated H2O contents in the melt showed a slight shift to values lower than the experimental H2O contents (y ¼ 0·873x), although a fairly good correlation was obtained (R2 ¼ 0·873). 1128
© Copyright 2026 Paperzz