Magma Evolution Processes in the Southern Okinawa Trough:Insights from Melt Inclusions

ZHANG Yujiao, and ZHAI Shikui

Magma Evolution Processes in the Southern Okinawa Trough:Insights from Melt Inclusions

ZHANG Yujiao, and ZHAI Shikui*

,,266100,

The Okinawa Trough is an initial back-arc basin that is influenced by the subduction of the Philippine Sea Plate and deve- lops on the continental crust. The Okinawa Trough is a natural laboratory for the study of basin evolution, magmatism, and crust- mantle processes in the early stage of back-arc spreading. Melt inclusions are small droplets of magma that are captured during the mineral crystallization process and can record the geochemical composition changes during magma evolution. In this study, the geochemical compositions of melt inclusions in host plagioclases of two volcanic rock samples at Station nos. 9-1 and 9-2 from the southern Okinawa Trough are systematically analyzed. Based on previous studies, the origin and evolution of magma and the introduction of subducting materials in the study area are discussed. Results show that melt inclusions are characterized by the relative enrichment of large-ion lithophile elements, depletion of high-field-strength elements, and slight enrichment of rare earth elements. Indeed, the subduction of the Philippine Sea Plate introduced sediment-derived melts and fluids into the magma source area of the southern Okinawa Trough. Subsequently, 4% to 5% partial melting of the hydrated mantle produces basaltic magma.The melt inclusions of andesite and dacite investigated in this study were formed by fractional crystallization of basaltic magma. Finally, the crystallization of plagioclase, pyroxene, and magnetite occurred during the late stage of magma evolution. The temperature-pressure data show that the melt inclusions in plagioclase have two capture periods: one is at temperatures above 1250℃ and the other is at temperatures between 1180℃ and 1200℃. The capture pressure of the inclusions at temperatures between 1180℃ and 1200℃ is between 5.6kPa and 6.1kPa, corresponding to the depth of 15–17km below the seafloor. The geochemical characteristics of major and trace elements in inclusions show that the samples from two stations (., 9-1 and 9-2) have similar or identical magma source areas. However, the crystallization differentiation reflected by inclusions in sample 9-1 is more obvious than that in sample 9-2. The inclusions were captured during magma evolution and were not contaminated by crustal materials.

melt inclusions; magma evolution processes; contamination of crustal materials; introduction of subducting materials; the southern Okinawa Trough

1 Introduction

Melt inclusions are small silicate droplets captured by minerals during the mineral crystallization process, withSiO2content usually >50% and a diameter of <100µm (Roedder, 1984). In volcanic rocks, melt inclusions often harden into a glass; however, in plutonic rocks, they often crystallize silicate minerals, metallic phases, and fluid phases (Bodnar and Student, 2006). Compared with whole rock, melt inclusions can record the concentration of original volatiles and metals as they are separated by degassing and fractionation during the magma condensation process. The host minerals remain stable, and the chemical composition of melt inclusions is unaltered by subsequent geolo- gical processes. Thus, the melt inclusions can provide a more detailed window into magma evolution. Because of these characteristics, the melt inclusions are identified as the ideal medium for analyzing the evolution of melt composition, such as magma mixing, fractional crystallization, and fluid exsolution. Furthermore, the melt inclusions can provide key evidence for the composition of silicate melts and related dissolved fluids before the occurrence of geological processes, such as eruption, devolatilization, and crystallization (Lowenstern, 2003) to reconstruct the history of magmatism.

The Okinawa Trough is a young back-arc spreading center (Shinjo and Kato, 2000) formed by the subduction of the Philippine Sea Plate under the Eurasian Plate (Lee., 1980; Kimura, 1985; Letouzey and Kimura, 1986; Sibuet., 1987; Zhai., 1997; Huang., 2006; Guo., 2018a, b). The Okinawa Trough is an ideal area for analyzing magmatism in the early spreading stage of the back-arc basin during modern-style plate subduction. The Okinawa Trough is a rare window into the occurrence and development of the back-arc basin, crust-mantle interaction, melting of mantle materials, and origin of magma. Thus, analyzing the volcanic rocks in the Okinawa Trough to fully understand subduction zone magmatism and deep geodynamic processes and enrich the geological theoretical system of the subduction zone, particularly the back- arc basin, is of significance (Yan and Shi, 2014). Although the petrology and geochemistry of volcanic rocks in the Okinawa Trough have been investigated in many studies (Honma., 1991; Zhai., 1997; Shinjo., 1999; Shu., 2018), divergences between the source and evolution of magma in the trough are still observed (Chen.,2017, 2018, 2019; Li., 2018a, b). Moreover, previous studies of the Okinawa Trough mainly concentrated on the whole-rock geochemistry, which represents the ‘mixture’ of melts formed by various complex factors and processes. These processes include mantle partial melting, magma mixing, fractional crystallization, contamination of crustal materials, and changes after magmatism, which will mask the original characteristics of magma (Ren., 2005; Kent, 2008; Panina., 2018). By contrast, the study of melt inclusions is an effective way to reveal the early mag- matic properties and late evolution processes.

Based on previous research, this study aims to further explore the origin of magma, fractional crystallization, and contamination of crustal materials in the process of magmatism by analyzing the geochemical characteristics of melt inclusions in volcanic rocks from the southern Okinawa Trough.

2 Geological Background

The Okinawa Trough is located on the active continental margin of the western Pacific Ocean (Fig.1) bounded on the north by Kyushu Island in Japan and on the south by Taiwan Island in China. The trough is a back-arc spreadingbasin formed by the subduction of the Philippine Sea Platealong the Ryukyu Trench to the Eurasian Plate (Shinjo., 1999; Li, 2001). Based on petrologic, isotopic chronology, and tectonic studies, the formation and evolution of the Okinawa Trough can be divided into three main stages: 1) uplift denudation stage (mid-Miocene to late Miocene, approximately 6–7Myr), 2) extensional rifting stage (early Pleistocene, approximately 2–0.1Myr), and 3) seafloor spreading stage (late Pleistocene to the present, approximately 0.01–0Myr) (Kimura, 1985). The Okinawa Troughis characterized by high heat flux, intense volcanic activity, development of a large number of normal faults, strong gravity anomaly, and weak negative geomagnetic anomaly, revealing the existence of basement sag and upper mantle uplift in the Okinawa Trough (Zeng., 2010; Lai., 2016). Based on the differences in its topography and tectonic evolution, Shinjo. (1999) proposed that the Okinawa Trough can be divided into three sections, namely, the north, middle, and south sections, bounded by the Tukara and Miyako fault structural belts. The subducted Philippine Sea Plate is approximately 150–200km below the central axis of the trough and 150km below the southern axis of the trough (Sibuet., 1998). The subduction velocity of the Philippine Sea Plate is approximately 7cm yr−1 in the southern part and approximately 5cmyr−1 in themiddle part (Arai., 2017). The thickness of the crust in the southern Okinawa Trough is 13–16km, which gra- dually increases toward the north, reaching up to 30km. The thickness of the crust in the central part is 16–22km (Sibuet., 1987; Liu., 2016).

Fig.1 Okinawa Trough topography and sampling locations.

Although researchers agree on the nature of the back- arc basin of the Okinawa Trough, controversy about the crustal nature of the trough exists. Shang. (2014) pro-posed that only thinned continental crust exists in the Okinawa Trough, indicating that the trough is in the initial continental rift stage. Based on the geological features and mantle density anomalies, Li. (1998) and Zhou. (2001) determined that the Okinawa Trough hasreached the mature continental rift stage and back-arc spreading has begun. Liang. (2001) and Gao. (2009) proposed that oceanic crust has developed along the central rift of the Okinawa Trough. According to the gravity anomaly, the NW fault of the Okinawa Trough is considered a transform fault (Lee., 1980; Jin and Yu, 1988).

3 Materials and Methods

3.1 Sample Collection and Preparation

The rock samples in this study were obtained from the southern Okinawa Trough and by TV grab of the survey ship ‘’ during the HOBAB4 cruise in 2016 (Table 1).

To prepare the thin sections, the samples were first cut to standard slide size and then ground to the required thickness at the Geological Exploration Technology Co., Langfang City, China. Petrologic studies were performed at the Laboratory of Rock and Mineral Identification and Sediment Analysis, Ocean University of China, Qingdao, China, using a polarizing microscope to obtain detailed mineralogical and petrographic observations of the thin sections. The volcanic rock types, mineral types, and rock structures were preliminarily identified. The microscopic phenomena were observed, marked, and photographed in detail. Subsequently, the target area was selected to provide a working basis for further geochemical analysis.

Table 1 Sampled locations

3.2 Analytical Methods

Homogenization experiment of melt inclusions was con- ducted at the State Key Laboratory of Mineralization Me- chanism Research of Endogenous Metal Deposits, NanjingUniversity, Nanjing, China, using the Linkam TS1500 high-temperature stage. Before heating the inclusions, pure NaCl crystals were used as the standard for calibration purposes. The melting point of NaCl measured in this study was approximately 799℃, compared with the standard meltingpoint of sodium chloride crystals of 800.4℃ (Ninane., 2004), resulting in a temperature measurement error of 5℃, which is consistent with international standards. In this research, the change characteristics of melt inclusions during the heating process were investigated using the tem- perature gradient method to achieve a balance between too fast and too slow heating of the melt inclusion. Indeed, too fast heating of the melt inclusion can lead to inaccurate temperature measurements or even inclusion rupture, whereas too slow heating of the melt inclusion can lead to H2O diffusion (Danyushevsky., 2002; Kent, 2008). Hence, the samples were heated to 600℃at a heating rate of 10℃min−1. After reaching 600℃, the samples were heated at a heating rate of 50℃(2h)−1. Finally, when the inclusions became transparent and bubbles appeared, the inclusions were heated at 5℃ until the bubbles disappeared. The homogeneous inclusions were immediately quenched for further study.

Major element analysis of melt inclusions and their host minerals was conducted at the Key Laboratory of Submarine Geosciences and Prospecting Techniques, MOE, China, using the JEOL JXA-8230 electron probe (Table 2). The working conditions included the acceleration voltage of 15kV, probe current of 20nA, and beam spot size of 5µm. The standard samples were provided by the company SPI, and analytical procedures were obtained from the study ofWang and Gaetani (2008). The accuracy of major elementanalysis was higher than 1%. Consequently, three points were selected from each appropriate inclusion to obtain the average value of the components of each inclusion. Meanwhile, the host minerals of inclusions were sampled for further study.

Table 2 Chemical compositions (wt%) of plagioclase phenocrysts analyzed by an electron probe microanalyzer

Notes: FeO*=FeO+Fe2O3; ‘−’denotes not detected or the calculated value is less than the valid value; P, point.

Analyses of trace elements and rare earth elements (REEs) in melt inclusions were conducted at the State Key Laboratory of Mineral Deposit Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Beijing, China using the laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS; LA system: GeoLasPro 193nm ArF excimer laser; ICP-MS: Agilent 7900 model). During the LA process, helium was used as carrier gas and argon was used as compensation gas, and a small amount of nitrogen was added to the system to improve sensitivity. The three gasses were mixed through a T-shaped connector before being introduced into the ICP. The sample chamber was a standard ablation cell. A resin-made mold was used to obtain a small volume of sampling space to reduce memory effects and improve washing efficiency. Thesignal acquisition of a single sample included a blank signal of approximately 20s, the sampling time is 50s, and the time of signal attenuation to the background value is approximately 20s. During the analysis, the working parameters of the laser are as follows: a frequency of 9–10Hz, energy of 10–11Jcm−2, and beam spot size of 24–44m, which were adjusted according to the inclusion size. The USGS reference glass (such as NIST 610, BCR-2G, BIR-1G, and BHVO-2G) was used as the calibration standard, and the element content was quantitatively calculated by the double internal standard method. NIST610, BCR-2G, BIR-1G, and BHVO-2G were analyzed twice for every ten inclusions. The offline processing of analysis data, including sample and blank signal selection, instrument sensitivity drift correction, and element content calculation, was performed using the software SILLS (Guillong., 2008), with an accuracy higher than 10%.

4 Results

The samples are grayish white, with obvious cross-sec-tional, pore, and scoria structures. The sizes of the pores vary, and no sediment filling is observed in the pores. Microscopically, a porphyritic texture is observed in the samples. The matrix in sample 9-1 consists of plagioclase (Pl, approximately 55%), orthopyroxene (Opx, approxima- tely 3%), and glass (approximately 42%). The matrix of sample 9-2 consists of plagioclase (Pl, approximately 30%),orthopyroxene (Opx, approximately 5%), olivine (Ol, approximately 1%), and glass (approximately 64%). The size of plagioclase phenocrysts is approximately 0.1–10mm, and most of them are self-shaped to semi-self-shaped plates with cleavage. The plagioclase phenocrysts generally show polysynthetic twin or zoning textures and good transparency under the lens. The inclusions in this study exist in plagioclase phenocrysts, which are abundant, well preserved, and clearly defined. The melt inclusions in plagioclase phenocrysts have two main types. The first type of melt inclusions is distributed along the crystal growth zones of minerals (Fig.2a), which are formed because of the interphase of the rapid and slow growth of mineral crystals. Most of these inclusions have a diameter of less than 5µm, which makes them difficult to analyze. The second type of melt inclusions is randomly distributed within the host minerals (Fig.2b), with a diameter ranging from 5µm to 120µm, mostly between 10µm and 40µm. Therefore, the second type of melt inclusions, which has a larger size than the first type of melt inclusions, is selected for analysis and research. Most of the inclusions are rounded or elliptical; some are triangularand polygonal in shape. In addition to the afore mentionedtwo types of melt inclusions, some sieve-like melt inclusions are detected in plagioclase (Fig.2c), which may be caused by partial melting of the host minerals (Nakamura and Shimakita, 1998). Therefore, this type of melt inclusions cannot represent the signatures of the melt and is excluded from our study. Most melt inclusions contain one or se- veral bubbles, whereas a small number of them do not contain bubbles. In contrast to that of sample 9-1, the bubbles in the melt inclusions of sample 9-2 are large, have a darker color, and do not have daughter crystals (Fig.2d). The melt inclusions selected for analysis in this study are isolated or exist as small clusters in host mineral crystals, far from host mineral fractures, and regarded as primary inclusions.

Fig.2 Representative transmitted images of minerals and melt inclusions in volcanic rocks collected from the southern Okinawa Trough. (a), melt inclusions in the growth zone of plagioclase; (b), primary inclusions randomly distributed within plagioclase; (c), plagioclase phenocrysts with a sieve texture; (d), dark melt inclusions.

The homogenization experiment was conducted on 65 melt inclusions of 6 plagioclase phenocrysts from samples 9-1 and 9-2. As shown in Fig.3, no notable change was observed in the homogenization process at temperatures below 1000℃. At temperatures between 1000℃and 1100℃, some brown inclusions began fading, the lines decreased or even disappeared, and the transparency increased.At temperatures above 1100℃, the bubbles gra- dually shrank and began to move. Subsequently, at temperatures above 1180℃, the inclusions became completely homogeneous and the bubbles disappeared. At temperatures above 1200℃, the inclusions began to burst. Thus, the homogenization temperatures of 1180℃to 1200℃were selected for the inclusions. For some inclusions, no homogenization was observed, even though the temperature reached 1250℃, which indicates two capture periods for melt inclusions: one is at temperatures between 1180℃and 1200℃, and the other is at temperatures above 1250℃.

The chemical compositions of plagioclase phenocrysts and melt inclusions are presented in Tables 2 and 3, respectively. Furthermore, the major and trace element com- positions of rock samples from the study area have been used from previously published by Chen.,1995; Shinjo.,1999; Wang., 2000; Guo., 2016a, b; Shu., 2017; Chen., 2018. Finally, Tables 4 and 5 present the trace element compositions of melt inclusions in host plagioclases from samples 9-1 and 9-2. Based on the results of electron microprobe analysis, the melt inclusions in two volcanic rock samples from the southern Okinawa Trough include a wide range of major element contents, namely, Al2O3(9.94–17.43wt%), SiO2(55.56–62.47wt%), MgO (1.67–5.05wt%), CaO (5.36–8.22wt%), FeO* (6.77–12.27wt%), K2O (0.27–1.00wt%), Na2O (0.19–1.67wt%), MnO (0.09–0.25wt%), and TiO2(0.66–1.13wt%). Fig.4 shows the total alkali-silica diagram of inclusions and rock samples. The diagram shows that all melt inclusions belong to the subalkaline series, with the composition varying from andesite to dacite. As illustrated in the K2O-SiO2diagram (Fig.4b), the composition of melt inclusions belongs to the low-K tholeiitic series. Based on the major element content diagram, the content of SiO2is negatively correlated with the contents of Al2O3, FeO*, MgO, CaO, and MnO and positively correlated with the content of TiO2; meanwhile, K2O, Na2O, and Cr2O3show no obvious linear relationship with SiO2(Fig.5). The published data show that the volcanic rocks in the southern Okinawa Trough include basalt, basaltic andesite, andesite, dacite, and rhyolite, indicating a complete basic-acid evolution sequence (Fig.4a).

Fig.3 Homogenization process of melt inclusions in plagioclase.

The primitive-mantle-normalized multi-trace element spi- dergram (Fig.6) shows that the melt inclusions of the two volcanic rock samples exhibit similar patterns, such as large-ion lithophile element (LILE) enrichment, high-field- strength elements (HFSE) depletion, and notable Pb positive anomalies and Nb-Ta negative anomalies. Moreover, the melt inclusions are characterized by slight enrichment of light REEs relative to heavy REEs and marked Eu negative anomalies in the chondrite-normalized REE patterns. However, significant differences in the absolute concentrations of trace elements are observed despite their similar patterns. The trace element and REE contents of melt inclusions in sample 9-2 are significantly lower than those in sample 9-1; however, similar trace element contents are observed in two melt inclusions (., 9-1-5 and 9-1-11) in sample 9-1 (Fig.6).

Table 3 Chemical compositions (wt%) of melt inclusions analyzed by an electron probe microanalyzer

Note: Same as Table 1.

Table 4 Trace and rare earth element contents(×10−6) of melt inclusions analyzed by LA-ICP-MS of sample 9-1

()

()

ElementP1P2P3P4P5P6P7P8P9P10P11P12P13P14 Ho1.901.902.181.410.740.901.802.192.070.380.991.210.651.29 Er5.806.146.375.801.921.964.964.766.532.312.682.892.093.61 Tm0.910.890.900.830.290.261.321.101.030.170.370.540.280.54 Yb5.846.336.565.472.541.862.699.607.261.492.222.91.633.28 Lu0.860.871.010.620.370.370.921.010.890.210.300.540.250.43 Hf3.874.564.773.611.481.864.742.583.951.532.052.621.052.93 Ta0.300.200.250.140.110.060.21−0.25−0.080.190.030.12 Pb18.6417.8119.9216.312.808.1520.4640.0419.63−3.525.593.076.64 Th5.325.096.043.711.191.695.504.255.451.402.112.691.102.75 U1.511.241.590.940.310.611.511.382.110.350.480.740.330.61

Table 5 Trace and rare earth element contents(×10−6) of melt inclusions analyzed by LA-ICP-MS of sample 9-2

Fig.4 Plot of silicate melt inclusions and rock samples from the southern Okinawa Trough. (a), alkali-silica diagram of melt inclusions and investigated rock samples (Le Maitre et al., 1989), with the boundary between alkalinity and subalkalinity derived from Irvine and Baragar (1971); (b), K2O-SiO2 diagram (wt%) of melt inclusions.

Fig.5 Two-component correlation diagram of melt inclusions.

Fig.6 Multi-trace element spidergram normalized by the primitive mantle and rare earth element pattern normalized by chondrite of melt inclusions. The normalization data of primitive mantle and chondrite are fromSun and McDoungh (1989).

Given that the two samples were collected from two adjacent areas and had similar major and trace element contents, samples 9-1 and 9-2 may have the same magma source, and the differences in their absolute element contents may have resulted from their different degrees of magma evolution.

5 Discussion

5.1 Crustal Assimilation

Whether and to what extent there are assimilation and contamination of crustal materials in the Okinawa Trough magma have remained controversial. Zhang. (2018) proposed that a certain degree of crustal assimilation was present in the magmas of the Okinawa Trough. However, Li. (2017, 2018a, b) proposed that no significant crustal assimilation occurred during magma evolution in the Okinawa Trough. Because the mantle-derived melts pass through the crust during their ascent, the extent of the role that crustal assimilation plays in magma evolution needs to be assessed (Hong., 2013). The incompatible element ratio of Ti/V of the mid-ocean ridge basalt (MORB) is 102 (Sun and McDonough, 1989) and that of the crust is 20 (Rudnick and Gao, 2003). The Ti/V ratios of melt inclusions in samples 9-1 and 9-2 are between 81.0 and 109.7, which are similar to that of MORB melts. Thus, the assimilation and contamination of crustal materials play an insignificant role in the magma evolution of the study area.

The proportion of incompatible elements remains unchanged during mantle melting but changes significantly during assimilation and contamination of continental crust materials. Therefore, the ratios of incompatible elements can be used to trace mantle-derived magma and subsequent processes (Green, 2006). As shown in the Nb/Yb-Th/Yb diagram (Fig.7), the melt inclusions of samples 9-1 and 9- 2 are near the MORB and far from the crust, which further indicates that the role of the assimilation and contamination of crustal materials in the magma evolution of the study area is insignificant.

Fig.7 Nb/Yb-Th/Yb diagram of melt inclusions in samples 9-1 and 9-2 (base map according toPearce, 2008).

5.2 Petrogenesis of the Andesitic Melt Inclusions

Thus far, no consensus on the formation of andesites in the Okinawa Trough has been reached. Ishizuka. (1990) proposed that the andesites in the Okinawa Trough were formed through magma mixing between rhyolitic andbasaltic magmas. By contrast, Shinjo and Kato (2000) pro- posed that the andesites in the Okinawa Trough were form- ed by crustal assimilation of basaltic magma.

The rock samples (Station nos. 9-1 and 9-2) in this study were collected from the southern Okinawa Trough. The whole rock (basalt) and its melt inclusions belong to the low-K rock series (Fig.4). Because the low-K and high-K basaltic magmas are derived from different magma source areas and crystallization differentiation can only form more acidic rocks of the same series (Nandedkar., 2014), the andesite melt inclusions in this study can be considered the product of the crystallization evolution of low-K basaltic magma. Moreover, according to the continuous va- riation of the main elements in rocks, fractional crystallization may be the main process of magma evolution. The correlation diagram of the trace element content (ratio) in the melt inclusions is consistent with that of the evolution trend of crystallization differentiation or magma mixing (Fig.8).

5.3 Addition of Subducting Materials

In general, identifying magma sources for volcanic rocks in the subduction zones is a complex task because of the influences of altered oceanic crust, pelagic sediments, slab-derived melts and fluids, and crust (Plank, 2005). The geo- chemical signatures of volcanic rocks are dependent on the addition of subducting materials if the mantle of the magma sources is stable (Li., 2015). Furthermore, distinguishing between crustal assimilation and the addition of subducting materials based only on the trace elements and isotopes of melt inclusions is difficult. However, based on the previously presented results, the assi- milation and contamination of crust materials to melt inclusions have been excluded from the magma evolution process. Therefore, assessing the influences of subduction on the magma in the southern Okinawa Trough based on the major and trace element compositions of melt inclusions and the previously published whole-rock data of basalts is viable.

Fig.8 (La/Sm)N-La diagram of melt inclusions in samples 9-1 and 9-2.

The melt inclusions in samples 9-1 and 9-2 are characterized by notable Nb-Ta negative anomalies and LILE enrichments on the primitive-mantle-normalized trace element diagram, indicating the influences of subduction- related components (Yang., 2017). In the binary diagrams of a subduction-immobile element (., Zr)mobile elements (., Th, U, Pb, Rb, and Ba; Fig.9), the investigated samples and previously investigated basalts from the southern Okinawa Trough are plotted above the enriched and normal MORBs, indicating the addition of subducting materials to the magma source.

HFSEs are only mobile in the melt phase, whereas LILEs are mainly mobile in the fluid phase (Class., 2013). Thus, magma with high Th/Nb ratios may have resulted from the addition of slab-derived melt (Class., 2013), whereas magma with high Ba/Th ratios may be modified by slab-derived fluids (Pearce and Peate, 1995; Elliott., 1997). The melt inclusions of samples 9-1 and 9-2 have Th/Yb ratios of 0.44–1.15 (with an average of 0.78) and Th/Nb ratios of 0.95–1.96 (with an average of 1.17). Furthermore, the melt inclusions of samples 9-1 and 9-2 have relatively low Ba/La ratios (., 17.32–35.80, with an average of 23.92) and Ba/Th (., 54.91–124.63, with an average of 81.07; Fig.10). Therefore, the magma source in the southern Okinawa Trough is modified by the subducted sediment-derived melts, rather than the slab- derived fluids, which resulted in the high Pb contents and low Ce/Pb ratios of the melt inclusions.

Fig.9 Plots of mobile elements versus Zr for the investigated samples and previously investigated basalts. Data sources for basalts from Chen et al. (1995), Shinjo et al. (1999), Wang et al. (2000), Guo et al. (2016a, b), Shu et al. (2017), Chen et al. (2018).

Fig.10 (a) Th/Yb-Ba/La plot (Dokuz, 2011) and (b) Th/Nb-Ba/Th plot (Elliott et al., 1997) for the investigated samples 9-1 and 9-2.

As discussed previously, the melt inclusions of volcanic rocks from the southern Okinawa Trough are affected by subduction sediments or sediment-derived melts. To as-sess the effect of such fluids or melts on magma sources, immobile element ratios relatively fixed during crystalliza-tion differentiation or partial melting can be used to constrain the variable contributions of subduction-related meltsand fluids (Hawkesworth., 1993; Stolper and New-man, 1994). Based on the current literature, Th content in sediments is higher than that in the mantle, and Th is highly immiscible relative to HFSE and REE in fluids (Singer, 2007). Moreover, Ba content in sediments or altered oceanic crust is higher than that in the mantle, but Ba is highly soluble in fluids (Brenan., 1998; Scam- belluri., 2004; Kessel., 2005). Therefore, high Th/Nb ratios indicate the involvement of sediment melts in the magma (Eliiott., 1997; Class., 2013), whereas high La/Nb and Ba/Nb ratios indicate the involvement of subducted sediment-derived fluids in the magma (Brenan., 1998; Scambelluri., 2004; Kessel., 2005). As shown in Fig.11, the investigated samples and previously investigated basalts are affected by both fluids and melts.

To further evaluate the influences of subducted sedi-ment-derived fluids and melts in the magma source, the MORB-ocean island basalt (OIB) series and island arc volcanic rock diagrams (Pearce, 2008) based on the Th, Nb,and Yb contents of basalts are used to model the proportion of the involved subducting materials. The melt inclusions of the volcanic rock samples are plotted between the MORB- OIB array and the island arc basalt series and are probably the products of 4% to 5% partial melting of the mantle wedge, according to the model of Pearce (2008). Moreover, the previously investigated basalts from the same area of the southern Okinawa Trough are plotted (Fig.12), whichindicates that these basalts are derived from 4% to 5% partial melting of the mantle wedge, and their magma evo- lution led to the formation of the melt inclusions in the study area.

Fig.11 (a) La/Nb-Th/Nb and (b) Ba/Nb-Th/Nb plots of basalts and the investigated samples (after Zamboni et al., 2016). Data sources for basalts from Chen et al. (1995), Shinjo et al. (1999), Wang et al., (2000), Guo et al. (2016a, b), Shu et al. (2017), Chen et al. (2018).

Fig.12 Nb/Yb-Th/Yb diagram for melt inclusions (after Pearce, 2008). Data sources of basalts from Chen et al. (1995), Shinjo et al. (1999), Chung et al., (2000), Guo et al. (2016a, b), Shu et al. (2017), Chen et al. (2018).

5.4 Fractional Crystallization

The fluid-mobile element Rb is usually considered a proxy of the magma evolution process because of its incompatibility in crystallized minerals (Zajacz and Halter, 2007; Sun., 2013). Ba and Rb contents in the volcanic rock samples from the study area are well correlated, indicating the significant role of fractional crystallization during magma evolution (Fig.13). The differences in the absolute contents of Ba and Rb between samples 9-1 and 9-2 indicate that sample 9-1 experienced a higher degree of fractional crystallization than sample 9-2. Meanwhile, the total REE content of sample 9-1 is higher than that of sample 9-2, which indicates that the magma evolution de-gree of sample 9-1 is higher than that of sample 9-2. The linear relationships shown in the Harker variation diagramsare generally the results of partial melting and crystal frac-tionation. The SiO2content of the melt inclusions in sam-ples 9-1 and 9-2 is negatively correlated with the Al2O3, FeO*, MgO, CaO, and MnO contents and positively corre-lated with the TiO2content but not correlated with the K2O, Na2O, and Cr2O3contents (Fig.5). The negative cor-relationbetween CaO and SiO2indicates that CaO-rich minerals, such as hornblende, plagioclase, and pyroxene, fractionate during magma evolution (Zhang., 2010). Furthermore, the notable Eu and Sr negative anomalies in Fig.6 indicate the prominent fractionation of plagioclase (Cai., 2015). The crystallization of clinopyroxene ge- nerally results in the remarkable decrease of the SiO2content and CaO/Al2O3ratio (Zhang., 2017). The MgO- CaO/Al2O3diagram (Fig.13b) shows that the CaO/Al2O3ratios of the melt inclusions in samples 9-1 and 9-2 decrease as the MgO content decreases, indicating the fractionation of clinopyroxene. Moreover, the FeO* content decreases as the SiO2content increases, indicating the fractionation of accessory minerals, such as magnetite (Guo., 2014).

Fig.13 (a) Ba-Rb diagram (after Sun et al., 2013) and (b) CaO/Al2O3-MgO plot (Zhang et al., 2017) of melt inclusions.

5.5 P-T Conditions of the Formation of the Melt Inclusions

In the past decades, the plagioclase geothermometer hasbeen successfully developed (Kudo and Weill, 1970; Mathez,1973; Michael, 1976). For plagioclase phenolcrysts with crystallization temperatures higher than 1100℃,the model of Sugawara (2001) is accurate. The formulation is expressed as follows:

The crystallization temperatures of the melt inclusions are between 1180℃ and 1200℃, as mentioned previously. Thus, based on these temperatures, we derive the following expression:

In this study, the calculated capture pressures of the homogenized melt inclusions range from 5.6 to 6.1kPa, with an analytical error of 1.8kPa. With the assumed cru- stal density of 3300kgm−3(Zhao., 2011) and seawater density of 1000kgm−3, the capture depth can be inferred to be between 15km and 17km (Fig.14), which is similar to the depth of the Moho discontinuity (Lai., 2016).

6 Conclusions

In this study, the geochemical compositions of melt in clusions in host plagioclases from volcanic rock samples at Station nos. 9-1 and 9-2 in the southern Okinawa Trough are systematically analyzed. Combined with the characteristics of major and trace elements of basalts from the study area, the magma material source and magma evolution are discussed.

Fig.14 Temperature-pressure model for samples 9-1 and 9- 2 (after Sugawara, 2001).

1) The melt inclusions were captured by plagioclase phe- nocrysts at temperatures above 1250℃ and between 1180℃ and 1200℃. The pressure of the latter is 5.6–6.1kPa, corresponding to the depth of 15–17km below the seafloor.

2) The melt inclusions were not contaminated by crustal materials when they were captured. The ratio of major and trace elements of the melt inclusions shows that the subduction of the Philippine Sea Plate introduced sediment-derived melts and fluids into the magma source area of the southern Okinawa Trough. Subsequently, 4% to 5% partial melting of the hydrated mantle produces basaltic magma. The melt inclusions of andesite and dacite investigated in this study were formed by fractional crystallization of basaltic magma.

3) The Harker variation diagrams of inclusions in samples 9-1 and 9-2 show that the main minerals (such as plagioclase and pyroxene) and accessory minerals (such as magnetite and amphibole) of magma have mainly undergone crystallization differentiation.

Acknowledgements

All authors are thankful to two anonymous reviewers for the thorough revision of the manuscript and their insightful comments. This research was funded by the National Program on Key Basic Research of China (973 Program) (No. 2013CB429702).

Arail, R., Kodaira, S., Kaiho, Y., Takahashi, T., Miura, S., Kaneda, Y.,., 2017. Crustal structure of the southern Okinawa Trough: Symmetrical rifting, submarine volcano, and potential mantle accretion in the continental back-arc basin.–, 122 (1): 622-641.

Bodnar, R. J., and Student, J. J., 2006. Melt inclusions in plutonic rocks: Petrography and microthermometry. In:, vol. 36. Webster, J. D., ed., Mineralogical Association of Canada, Montreal, 1-25.

Brenan, J. M., Ryerson, F. J., and Shaw, H. F., 1998. The role of aqueous fluids in the slab-to-mantle transfer of boron, beryllium, and lithium during subduction: Experiments and models., 62 (19-20): 3337-3347.

Cai, Y. C., Fan, H. R., Santosh, M., Hu, F. F., Yang, K. F., Liu, X.,., 2015. Silicate melt inclusions in clinopyroxene phe- nocrysts from mafic dikes in the eastern North China Craton: Constraints on melt evolution., 97: 150-168.

Chen, C., Lee, T., Shieh, Y. N., and Hsu, W. Y., 1995. Magmatism at the onset of back-arc basin spreading in the Okinawa Trough., 69 (3): 313-322.

Chen, Z., Zeng, Z., Wang, X., Yin, X., Chen, S., Guo, K.,., 2018. U-Th/He dating and chemical compositions of apatite in the dacite from the southwestern Okinawa Trough: Implication for petrogenesis.,161: 1-13.

Chen, Z., Zeng, Z., Wang, X., Zhang, Y., Yin, X., Chen, S.,., 2017. Mineral chemistry indicates the petrogenesis of rhyolite from the southwestern Okinawa Trough., 16 (6): 1097-1108.

Chen, Z., Zeng, Z., Yin, X., Wang, X., Zhang, Y., Chen, S.,., 2019. Petrogenesis of highly fractionated rhyolites in the sou- thwestern Okinawa Trough: Constraints from whole-rock geochemistry data and Sr-Nd-Pb-O isotope., 54 (1): 316-332.

Class, C., Miller, D. M., Goldstein, S. L., and Langmuir, C. H., 2013. Distinguishing melt and fluid subduction components in Umnak Volcanics, Aleutn Arc., 1 (6): 1004.

Danyushevsky, L. V., McNeill, A. W., and Sobolev, A. V., 2002. Experimental and petrological studies of melt inclusions in phenocrysts from mantle-derived magmas: An overview of techniques, advantages and complications., 183: 5-24.

Dokuz, A., 2011. A slab detachment and delamination model for the generation of Carboniferous high-potassium I-type magmatism in the eastern Pontides, NE Turkey: The Köse compo- site pluton., 19 (4): 926-944.

Elliott, T., Plank, T., Zindler, A., White, W., and Bourdon, B., 1997. Element transport from slab to volcanic front at the Mariana arc., 102 (B7): 14991.

Gao, J. Y., Zhang, T., Fang, Y. X., Yang, C. G., and Mei, S., 2009. Faulting, magmatism and crustal oceanization of the Okinawa Trough., 28 (3): 40-49.

Guo, F., Li, H., Fan, W., Li, J., Zhao, L., and Huang, M., 2016a. Variable sediment flux in generation of Permian subduction- related mafic intrusions from the Yanbian region, NE China., 261: 195-215.

Guo, K., Shu, Y., Wang, X., Liu, Q., Yin, X., Chen, S.,., 2018a. Different magma sources and evolutions of white and black pumice from the middle Okinawa Trough: Evidence from major, trace elements and Sr-Nd-Pb isotopes., 54 (1): 206-220.

Guo, K., Zhai, S. K., Wang, X. Y., Yu, Z. H., Lai, Z. Q., Chen, S.,., 2018b. The dynamics of the southern Okinawa Trough magmatic system: New insights from the microanalysis of the An contents, trace element concentrations and Sr isotopic com- positions of plagioclase hosted in basalts and silicic rocks., 497: 146-161.

Guo, K., Zhai, S. K., Yu, Z., Wang, S., Zhang, X., and Wang, X., 2016b. Geochemical and Sr-Nd-Pb-Li isotopic characteristics of volcanic rocks from the Okinawa Trough: Implications for the influence of subduction components and the contaminationof crustal materials., 180: 140-151.

Guillong, M., Meier, D. L., Allan, M. M., Heinrich, C. A., and Yardley, B. W. D., 2008. SILLS: A MATLAB-based program for the reduction of laser ablation ICP-MS data of homogeneous materials and inclusions.. Vancouver, 328-333.

Green, N. L., 2006. Influence of slab thermal structure on basalt source regions and melting conditions: REE and HFSE constrains from the Garibaldi volcanic belt, northern Cascadia sub- duction system., 87 (1-2): 23-49.

Hawkesworth, C. J., Gallagher, K., Hergt, J. M., and McDermott, F., 1993. Mantle and slab contributions in arc magmas., 21 (1): 175-204.

Hong, L. B., Zhang, Y. H., and Qian, S. P., 2013. Constrains from melt inclusions and their host olivines on the petrogenesis of Oligocene–early Miocene Xindian basalts, Chifeng area, NorthChina Craton., 165 (2): 305-326.

Honma, H., Kusakabe, M., Kagami, H., Iizumi, S., Sakai, H., Ko- dama, Y.,., 1991. Major and trace element chemistry and D/H,18O/16O,87Sr/86Sr and143Nd/144Nd ratios of rocks from the spreading center of the Okinawa Trough, a marginal black- arc basin., 25 (2): 121-136.

Huang, P., Li, A., and Jiang, H., 2006. Geochemical features and their geological implications of volcanic rocks from the nor- thern and middle Okinawa Trough.,22 (6): 1703-1712.

Irvine, T. N. J., and Baragar, W. R. A., 1971. A guide to the che- mical classification of the commom volcanic rocks., 8 (5): 523-548.

Jin, X., and Yu, P., 1988. Structure and tectonic evolution of Okinawa Trough., 31 (5): 106-115.

Kent, A. J. R., 2008. Melt inclusions in basaltic and related volcanic rocks., 69: 273- 331.

Kessel, R., Schmidt, M. W., Ulmer, P., and Pettke, T., 2005. Trace element signature of subduction-zone fluids, melts and supercritical liquids at 120–180km depth., 437 (7059): 724- 727.

Kimura, M., 1985. Back-arc rifting in the Okinawa Trough., 2 (3): 222-240.

Kudo, A. M., and Weill, D. F., 1970. An igneous plagioclase thermometer.,25 (1): 52-65.

Lai, Z., Zhao, G., Han, Z., Huang, B., Li, M., Tian, L.,.,2016. The magma plumbing system in the Mariana Trough back-arc basin at 18˚N., 18: 132- 139.

Lai, Z., Zhao, G., Han, Z., Liu, B., Bu, X., and Leng, C., 2016. Back-arc magma processes in the Okinawa Trough: New insights from textural and compositional variations of plagioclase in basalts.,51: 346-356.

Le Maitre, R., 1989.. Blackwell, Oxford, 1-193.

Lee, C. S., Shor Jr., G. G., Bibee, L. D., Lu, R. S., and Hilde, T. W. C., 1980. Okinawa Trough: Origin of a back-arc basin., 35 (1): 219-241.

Letouzey, J., and Kimura, M., 1986. The Okinawa Trough: Genesis of a back-arc basin developing along a continental margin., 125 (1): 209-230.

Li, C. S., Arndt, N. T., Tang, Q., and Riple, E. M., 2015. Trace element indiscrimination diagram., 232 (7): 76-83.

Li, N. S., 2001. On tectonic problems of the Okinawa Trough., 19 (3): 255- 264.

Li, N. S., Jiang, L. L., and Li, C. Z., 1998. A study on the nature of crustal structures of the Okinawa Trough., 10 (5): 443-450 (in Chinese with English abstract).

Li, X., Zeng, Z., Shuai, C., Yao, M., Yang, H., Zhang, Y.,, 2018a. Geochemical and Sr-Nd-Pb isotopic compositions of vol- canic rocks from the Iheya Ridge, the middle Okinawa Trough: Implications for petrogenesis and a mantle source.,37 (1): 73-88.

Li, X. H., Zeng, Z. G., Wang, X. Y., Chen, S., Ma, Y., Yang, H.,, 2018b. Petrogenesis of basalt from the middle Okinawa Trough: New insights from olivine-hosted melt inclusions., 53 (6): 1-18.

Liang, R. C., Wang, S. G., and Wu, J. L., 2001. Preliminary study on geophysical field and the new crust in the middle of the Okinawa Trough., 21: 57-64 (in Chinese with English abstract).

Liu, B., Li, S. Z., Suo, Y. H., Li, G. X., Dai, L. M., Somerville, I. D.,., 2016. The geological nature and geodynamics of the Okinawa Trough, western Pacific., 51: 416- 428.

Lowenstern, J. B., 2003. Melt inclusions come of age: Volatiles, volcanoes and Sorby’s legacy., 5: 1-21.

Mathez, E.A., 1973. Refinement of the kudo-weill plagioclase thermometer and its application to basaltic rocks.y, 41 (1): 61-72.

Michael, J. D., 1976. Plagioclase-melt equilibria., 40 (4): 457-465.

Nandedkar, R. H., Ulmer, P., and Muntener, O., 2014. Fractional crystallization of primitive, hydrous arc magmas: An experimental study at 0.7Gpa., 167 (6): 1015.

Nakamura, M., and Shimakita, S., 1998. Dissolution origin and syn-entrapment compositional change of melt inclusions in plagioclase., 161: 119- 133.

Ninane, L., Humblot, C., Gerard, P., and Solvay, 2004. Method for producing sodium chloride crystals. US Patent.

Panina, L. I., Isakova, A. T., and Sazonov, A. M., 2018. The Olivinite of the Krestovskaya intrusion–The product of larnite- normative alkali ultramafic magma: Melt inclusion data.,26 (2): 167-180.

Pearce, J. A., 2008. Geochemical fingerprinting of oceanic basalts with applications to ophiolite classification and the search for Archean oceanic crust., 100 (1-4): 14-48.

Pearce, J. A., and Peate, D. W., 1995. Tectonic implications of the composition of volcanic arc magmas., 23 (1): 251-285.

Plank, T., 2005. Constraints from thorium/lanthanum on sediments recycling at subduction zones and the evolution of the continents., 46 (5): 921-944.

Ren, Z. Y., Ingle, S., Takahashi, E., Hirano, N., and Hirata, T., 2005. The chemical structure of the Hawailian mantle plume., 436: 837-840.

Roedder, E., 1984. Fluid inclusions. In:. Mineralogy Society of America, Washington D. C., 1- 644.

Rudnick, R. L., and Gao, S., 2003. Composition of the continentalcrust. In:. 2nd edition. Holland, H. D., and Turekian, K. K., eds., Elsevier, Amsterdam, 1-64.

Scambelluri, M., Muntener, O., Ottolini, L., Pettke, T. T., and Vannucci, R., 2004. The fate of B, Cl and Li in the subducted oceanic mantle and in the antigorite breakdown fluids.,222 (1): 217-234.

Shang, L. N., Zhang, X. H., Wu, Z. Q., and Han, B., 2014. Crustal structure of the Okinawa Trough: Evidence from wide-angle reflection and seismic refraction surveys., 34: 75-84.

Shinjo, R., and Kato, Y., 2000. Geochemical constraints on the origin of bimodal magmatism at the Okinawa Trough, an incipient back-arc basin., 54 (3): 117-137.

Shinio R., Chung, S. L., Kato, Y., and Kimura, M., 1999. Geochemical and Sr-Nd isotopic characteristics of volcanic rocks from the Okinawa Trough and Ryukyu Arc: Implications for the evolution of a young, intracontinental back arc basin.,104 (B5): 10591-10608.

Shu, Y., Nielsen, S. G., Zeng, Z., Shinjo, R., Blusztajn, J., Wang, X.,., 2017. Tracing subducted sediment inputs to the Ryukyu arc-Okinawa Trough system: Evidence from thallium iso- topes., 217: 462-491.

Sibuet, J. C., Deffontaines, B., Hsu, S. K.,Thareau, N., Le Formal, J. P., and Liu, C., 1998. Okinawa trough backarc basin: Early tectonic and magma evolution.–,103 (B12): 30245-30267.

Sibuet,J. C., Letouzey, J., Barbier, F., Charvet, J., Foucher, J. P., Hilde, T. W. C.,, 1987. Back arc extension in the OkinawaTrough.–, 92 (B13): 14041-14063.

Singer, B. S., Jicha, B. R., Leeman, W. P., Rogers, N. W., Thirl- wall, M. F., Ryan, J.,., 2007. Along-strike trace element and isotopic variation in Aleutian Island arc basalt: Subduction melts sediments and dehydrates serpentine., 112 (B6): B06206.

Stolper, E., and Newman, S., 1994. The role of water in the petrogenesis of Mariana trough magmas., 121 (3-4): 293-325.

Sugawara, T., 2001. Ferric iron partitioning between plagioclase and silicate liquid: Thermodynamics and petrological applications., 141: 659-686.

Sun, H., Xiao, Y., Gao, Y., Lai, J., and Wang, Y., 2013. Fluid and melt inclusions in the Mesozoic Fangcheng basalt from North China Craton: Implications for magma evolution and fluid/melt-peridotite reaction.,165 (5): 885-901.

Sun, S. S., and McDonough, W. F., 1989. Chemical and isotopic systematics of oceanic basalts: Implications for mantle composition and processes. In:. Saunders, A. D., and Norry, M. J., eds., Geological Society Special Publication, London, 313-345.

Wang, C., Yang, M., Chou, C., Chang, Y. C., and Lee, C. S., 2000. Westward extension of the Okinawa Trough at its western end in the northern Taiwan area: Bathymetric and seismo- logical evidence.,372 (2): 459-480.

Wang, Z., and Gaetani, G. A., 2008. Partitioning of Ni between olivine and siliceous eclogite partial melt: Experimental constraints on the mantle source of Hawaiian basalts., 156 (5): 661-678.

Yan, Q., and Shi, X., 2014. Petrologic perspectives on tectonic evolution of a nascent basin (Okinawa Trough) behind Ryukyu arc: A review., 33 (4): 1-12.

Yang, Y. Z., Wang, Y., Ye, R. S., Li, S. Q., He, J. F., Siebel, W.,., 2017. Petrology and geochemistry of early Cretaceous A-type granitoids and late Mesozoic mafic dikes and their relationship to adakitic intrusions in the lower Yangtze River belt, Southeast China., 59 (1): 62-79.

Zajacz, Z., and Halter, W., 2007. LA-ICPMS analyses of silicate melt inclusions in co-precipitated minerals: Quantification, data analysis and mineral/melt partitioning., 71 (4): 1021-1040.

Zamboni, D., Gazel, E., Ryan, J. G., Cannatelli, C., Lucchi, F., Atlas, Z. D.,., 2016. Contrasting sediment melt and fluid signatures for magma components in the Aeolian arc: Implications for numerical modeling of subduction systems., 17 (6): 2034-2053.

Zeng, Z., Yu, S., Wang, X., Fu, Y., Yin, X., Zhang, G.,., 2010.Geochemical and isotopic characteristics of volcanic rocks from the northern East China Sea shelf margin and the Okinawa Trough., 29 (4): 48-61.

Zhai, S., Chen, L., Wang, Z., and Gan, X., 1997. Primary analysis on pumice magmatism model of the Okinawa Trough.,17 (1): 59-66 (in Chinese with English abstract).

Zhang, G. L., Jiang, S. Q., Ouyang, H. G., Chen, D. G., Yin, X. B., Wang, X. Y.,., 2010. Magma mixing in upper mantle: Evidence from high Mg# olivine hosted melt inclusions in MORBs near East Pacific Rise 13˚N., 55 (16): 1643-1656.

Zhang, Y. H., Ren, Z. Y., Hong, L. B., Zhang, L., Qian, S. P., Wu, Y. D.,., 2017. Petrogenesis of Cenozoic Zhoucun basalts, western Shandong.,41 (1): 183-201 (in Chinese with English abstract).

Zhang, Y., Zeng, Z., Li, X., Yin, X., Wang, X., Chen, S.,., 2018. High-potassium volcanic rocks from the Okinawa Trough: Implications for a cryptic potassium-rich and DUPAL-like source., 53 (5): 1755-1766.

Zhao, Z., Niu, Y., Christensen, N. I., Zhou, W., Hou, Q., Zhang, Z.,., 2011. Delamination and ultra-deep subduction of continental crust: Constrains from elastic ware velocity and density measurement in ulreahigh-pressure metamorphic rocks., 29: 781-801.

Zhou, Z., Liao, Z., Jin, X., and Jia, J., 2001. Okinawa Trough: The highest stage of continental tension rifting in back-arc setting.,21 (1): 51-55 (in Chinesewith English abstract).

April 18, 2020;

July 20, 2020;

July 23, 2021

© Ocean University of China, Science Press and Springer-Verlag GmbH Germany 2021

. Tel: 0086-532-66782865

E-mail: zhai2000@ouc.edu.cn

(Edited by Chen Wenwen)