Abstract

Since the early Cenozoic, the West Philippine Basin (WPB) and the whole Philippine Sea Plate (PSP) has undergone a complex geological evolution. In this study, we presented K-Ar ages, in situ trace element, and major element compositions of minerals of basalts collected from the Benham Rise and the Central Basin Fault (CBF) in the WPB, to constrain their magmatic process and regional geological evolution. Olivine phenocrysts and microlites in the alkali basalts (20.9 Ma) from the Benham Rise have forsterite (Fo) contents of 56.90%–76.10% and 53.13%-66.41%, respectively. The clinopyroxenes in the tholeiites (29.1 Ma) from the CBF is predominantly diopside and augite, and it is depleted in light rare earth elements (LREEs) (LaN/YbN=0.133.40) and large-ion lithophile elements (LILEs). The plagioclases in the basalts from both of the Benham Rise and the CBF are predominantly labradorite and andesine, with a minor amount of bytownite, and it is enriched in LREEs, Ba, Sr, and Pb and exhibits strong positive Eu anomalies. However, there exist obvious differences in plagioclase compositions between these two tectonic sites. The source lithology of the Benham Rise basaltic rocks could be garnet pyroxenite, and yet that of the CBF could be spinel-lherzolite. The calculated mantle potential temperature beneath the Benham Rise is 1439°C–1473°C, which is significantly higher than that beneath the CBF (1345°C–1381°C), suggesting there existed thermal anomaly beneath the Rise during basaltic magmatism. This study also calculated the temperature and pressure of the clinopyroxenes and plagioclases, which have been used to indicate magmatic processes. Finally, we suggest that the Benham Rise basaltic rocks may be related to a mantle plume (e.g., the Oki-Daito mantle plume), and the CBF was once located in a back-arc spreading center behind an active subduction zone. The extinction of the Oki-Daito mantle plume activity might be at about 20.9 Ma, and cessation of the back-arc spreading of WPB was at about 29.1 Ma or younger.

1. Introduction

The Philippine Sea is one of the largest marginal seas in the world, and it has been studied extensively. The Philippine Sea Plate (PSP) lies at the intersection of the Eurasian Plate, the Pacific Plate, and the Indo-Australian Plate, and it is surrounded by a series of subduction zones, with the Ryukyu trench to the north, the Mariana and Ogasawara Trench to the east, the Yap and Palau trenches to the south, and the Philippine Trench to the west. The Philippine Sea Plate consists of the West Philippine Basin (WPB), the Shikoku Basin, the Parece Vela Basin, the Mariana Trough, and two remnant arcs (the Kyushu-Palau Ridge and the West Mariana arc) [15]. Tectonic reconstructions of the Philippine Sea Plate suggest that the plate was located near the equator before 50 Ma, and it migrated northward to its present position, with a nearly 90° clockwise rotation [1].

At present, there are three models for the origin of the West Philippine Basin. (1) The WPB is a segment of trapped oceanic crust [3, 6]. In this model, the WPB is a marginal basin formed along the Kula-Pacific accretionary plate boundary. (2) The WPB was developed by a back-arc spreading system [711]. Hall [1] detailed the tectonic evolution of the western Pacific, including the PSP since 55 Ma. The PSP was originally located on the equator, and it has gradually migrated northward since the Early Cenozoic. During its northward migration, the West Philippine Basin formed (50–30 Ma) and the proto-Izu-Bonin-Mariana (IBM) (35–30 Ma) was rifted in turn. The rifting of the proto-IBM leads to the formation of the Shikoku Basin (25–17 Ma) and the Parece Vela Basin (28–23 Ma), and the subduction region of the Pacific Plate is backward to the east of the Izu-Bonin-Mariana arc [15, 7, 12]. In addition, a clockwise rotation of nearby 90° occurred as the PSP migrated northward [1, 4]. (3) The development of the WPB is influenced by both back-arc spreading and a mantle plume [1, 7, 13, 14]. In the West Philippine Basin, scientists have collected various rock types, including N-MORBs (Normal Mid-Ocean Ridge Basalts) from the Central Basin Fault (CBF); OIBs (Ocean Island Basalts) from the Benham Rise, the Urdaneta Plateau, the Amami Plateau, the Daito Ridge, and the Oki-Daito Ridge; and arc volcanic rocks from the Kyushu-Palau Ridge, and their geodynamics settings appear to be related to back-arc spreading, a mantle plume, and volcanic arc processes, respectively [2, 4, 5, 13, 1517]. The basalts erupted from the spreading centers on the Philippine Sea Plate have the isotopic characteristics of Indian Mid-Ocean Ridge basalts, rather than Pacific Mid-Ocean Ridge basalts [2]. Previous studies have shown that the Amami Plateau, the Daito Ridge, and the Oki-Daito Ridge are related to arc-plume interaction. The Benham Rise and the Urdaneta Plateau may have been affected by a mantle plume (e.g., the Oki-Daito mantle plume) [5, 13, 16]. Until now, many studies have focused on the WPB, but few basement rock samples have been collected from the Benham Rise, and the CBF and the magmatic processes of the WPB have not been studied in detail, which hinders our understanding of the tectonic evolution of the WPB and the PSP.

In this study, we obtained in situ trace element (using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS)) and major element (using EMPA) compositions of the clinopyroxene, plagioclase, and olivine in the basalts and investigated the source lithology, mantle potential temperature and primary melt compositions, and magmatic processes of the basalts from the Benham Rise and the CBF in the West Philippine Basin.

2. Geologic Setting and Sample Descriptions

The West Philippine Basin is bounded by the Philippine Trench to the west, the Ryukyu Trench to the north, and about 3000 km of the Kyushu-Palau Ridge to the east (Haraguchi et al. 2011). The WPB consists of a Mesozoic arc terrane in the north, including Gagua Ridge (123-105 Ma), the Amami Plateau (~115 Ma), the Daito Ridge, and the Oki-Daito Ridge (44.4–40.5 Ma); the Benham Rise (39.8–35.9 Ma) and the Urdaneta Plateau (41.6–37.2 Ma) in the west; and the CBF (49.0–35.0 Ma) in the middle (Figure 1) [3, 13, 16]. Since 50 Ma, there has been widespread magmatism in the West Philippine Basin (e.g., [2, 16, 17]; Haraguchi et al. 2011; [5];Yan and Shi 2013; [13]). The igneous rocks of the Amami-Daito province are basalts, dacites, and tonalites. They have the geochemical characteristics of intraoceanic island arc rocks, and they may have been affected by a mantle plume [4, 5, 16]. The igneous rocks in the Huatung Basin are basalts and gabbros, with E-MORB trace element signatures and Indian MORB Sr, Nd, and Hf isotopic signatures [18]. The igneous rocks of the Kyushu-Palau Ridge are tholeiitic, and they have the geochemical characteristics of oceanic arc tholeiites (Haraguchi et al. 2011). The igneous rocks of the Benham Rise and the Urdaneta Plateau are basalts, and they have the geochemical characteristics of OIBs [2, 13, 16]. The igneous rocks of the CBF are basalts, and they have the geochemical characteristics of MORBs [2, 17].

The eleven basaltic rock samples in this study were collected from two dredge stations (100DS-Vinogradov Seamount in the Benham Rise and 94DS in the CBF) during China-Germany joint leg no. SO-57 cruise of R/V Sonne (Figure 1) [19]. Samples from 100DS are alkali basalts and are of oceanic island basalt- (OIB-) like characteristics, and those from 94DS are tholeiites and similar to back-arc basin basalts (BABBs) [20]. The six samples from site 100DS exhibit porphyritic textures and contain plagioclase and olivine phenocrysts, with some microlites (olivine and plagioclase, <0.1 mm grain size) in the groundmass (Figures 2(a) and 2(b)). The phenocryst content of the samples from site 100DS is 5%–10%, and the plagioclase content accounts for 70%–90% of the total phenocrysts. The grain size of the olivine phenocrysts in the basalts from site 100DS is mostly 0.2–0.4 mm, and some are even as large as 1.5 mm (sample 100DS7). The olivine phenocrysts are idiomorphic and hypidiomorphic, and some have been altered to iddingsite. The grain size of the plagioclase phenocrysts ranges from 0.04×0.20 to 0.2×2.00mm2. The five samples from site 94DS exhibit porphyritic textures and contain plagioclase and clinopyroxene phenocrysts, with some microlites (clinopyroxene and plagioclase) in the groundmass (Figures 2(c) and 2(d)). The phenocryst content of the samples from site 94DS is 10%–20%, and the plagioclase content accounts for 50%–70% of the total phenocrysts. The grain size of the clinopyroxene in the basalts from site 94DS varies from 0.2 to 0.3 mm. The clinopyroxene phenocrysts are idiomorphic and hypidiomorphic. The grain size of the plagioclase phenocrysts ranges from 0.03×0.20 to 0.3×1.50mm2, and most are polysynthetic twins.

3. Analytical Methods

3.1. Dating Methods

Based on detailed petrographic observations, we selected two samples (one for each dredge location) with almost no containing olivine phenocrysts to be dated. The analytical work was carried out in the State Key Laboratory of Earthquake Dynamic (SKLED), Institute of Geology, China Earthquake Administration. The detailed method of potassium–argon dating is as follows, argon isotopes were measured in MM1200 mass spectrometer using the isotopic dilution method. Argon-38, whose purity is better than 99.9%, is used as the diluent, and the standard sample ZBH biotite (whose age is 132.6 Ma) was used during the calibration of diluent. Potassium content was measured using a G-3 flame photometer. Parameters for age calculation are as follows, 40Ar/36Ar=295.5; λ=5.543×1010a1; λβ=4.962×1010/a; λe=0.581×10l0/a; 40K/K=1.167×104mmol/mol.

3.2. In Situ Compositional Analytical Methods for Minerals

The major elements were analyzed using the JXA-8230 Electron Microprobe Analyzer (EMPA) at the Key Laboratory of Marine Geology and Metallogeny, First Institute of Oceanography, Ministry of Natural Resources (MNR), China. The working conditions of the instrument were as follows: the acceleration voltage was 15 kV, the electron beam current was about 2×108A, the electron beam spot was 1 μm, and the quantitative detection limit was about 100 ppm. The standards used in these analyses were albite for Na and Si, orthoclase for K, diopside for Ca, olivine for Mg, hematite for Fe, garnet for Al, chromium oxide for Cr, rutile for Ti, rhodonite for Mn, and nickel silicide for Ni. The analytical results were corrected using the ZAF method (Z, A, and F represent the atomic number, absorption, and fluorescence, respectively). The precision of the major element analyses (SiO2, Al2O3, and CaO) is better than 1%, and that of the minor element analyses (Na2O, K2O, TiO2, P2O5, MnO, Cr2O3, and MgO) is better than 5%. During the EMPA analysis, backscattering electron (BSE) images of typical minerals were obtained. The data were processed using the Geokit geochemical software tool [21].

The in situ trace element analyses of the minerals in the Late Cenozoic basaltic rocks from Thailand were conducted at the Beijing Createch Testing International Co. Ltd. The laser sampling was performed using an ESI NWR 193 nm excimer laser ablation system, and an Analytik Jena Plasma Quant MS instrument was used to acquire the ion-signal intensities. The spot diameter of the laser beam was 35 μm, and the ablation frequency was 10 Hz. Helium was used as carrier gas. Each spot analysis incorporated approximately 20 s of background acquisition from the sample. The element contents were calibrated against multiple reference materials (NIST SRM612 and NIST SRM610). The data calibration was conducted using suitable internal standards and Glitter (Si was the internal standard for this experiment). The data for most of the elements have an accuracy of less than 5% and a precision of greater than 10%.

4. Results

4.1. Dating Results

The potassium–argon ages of basalts from the Benham Rise and the CBF are listed in Table 1. The apparent age of basalt (100DS16) from the Benham Rise is 20.9 Ma, which is obviously younger than the ages of basalts from IODP site 292 [2]. The apparent age of basalt (94DS1) from the CBF is 29.1 Ma, which is similar to the previous K-Ar ages of basalts from CBF [12, 22].

4.2. In Situ Mineral Compositions

4.2.1. Olivine

(1) Major Element Compositions. The analytical results of the olivine in the basalts from the West Philippines Basin are reported in Supplementary Table 1. For sample 100DS18, the Fo contents of the olivine phenocrysts and the microlites range from 57.80% to 76.10% and 56.90% to 60.00%, respectively. The Fo contents of the olivine microlites are significantly lower than those of the phenocryst rims, while those of the other samples are similar to those of the phenocryst edges. In general, FeO, MnO, Al2O3, Ti2O3, and CaO increase and NiO decreases with decreasing Fo content. The CaO contents of the olivine range from 0.144 to 0.51 wt%, which is within the range of basaltic phenocrysts (>0.1%) but is much higher than that of mantle peridotite xenoliths (<0.1%) [23]. The grain size of the olivine is relatively small, and no reaction rims or deformation structures, which are common to olivine xenocrysts, were observed, indicating that the olivine phenocrysts in this study are magmatic in origin [24]. In addition, NiO decreases with decreasing Fo content, which differs from the mantle olivine array (Figure 3).

For sample 100DS7, the Fo contents of big olivine grain (>1.5 mm) range from 61.90% to 90.80%, and the Fo contents of olivine microlites are about 58.60%, similar to the rim of big olivine grain. The oxide contents (e.g., TiO2, FeO, MnO, and CaO) of big olivine grain vary significantly from core to rim. The olivine has relative high Fo (%) (core: 90.30–90.80, rim: 61.90–62.80) and NiO contents (core: 0.25–0.36, rim: 0.09–0.12) and low CaO contents (core: 0.07–0.11, rim: 0.55–0.57) and MnO contents (core: 0.12–0.17, rim: 0.45–0.51), indicating that the cores of olivine are xenocrysts from the mantle, and the rim of the olivine is formed in a magma chamber over the course of a few years prior to eruption (Figure 3). The presence of mantle-derived xenoliths indicates rapid magmatic ascent [25, 26].

(2) Trace Element Compositions. The trace element concentrations of the olivine from the West Philippines Basin are listed in Supplementary Table 2. The total rare earth element concentrations (ΣREE) of the olivine xenocrysts are very low, ranging from 0.05 to 0.43 ppm, with LaN/YbN of 1.78–22.29, CeN/YbN of 2.67–17.13, SmN/YbN of 0.29–4.76, LaN/NdN of 1.24–5.90, and LaN/SmN of 4.68–6.24. Based on their rare earth element concentrations, the olivine microlites can be divided into two groups (Figures 4(a) and 4(b)). Group 1 has relatively high REE concentrations ranging from 100.82 to 156.01 ppm and exhibits light rare earth element (LREE) enrichment, with relatively high LREE/HREE (3.46–5.06) and LaN/YbN (9.39–19.30) ratios, and negative Ce anomalies (Ce/Ce=0.450.86, with an average of 0.57) or no obvious Ce anomalies (Ce/Ce=CeN/LaNPrN0.5). Group 2 has very low REE concentrations ranging from 1.85 to 3.28 ppm and exhibits LREE enrichment, with relatively high LREE/HREE ratios (2.26–4.81) and obvious negative Ce (Ce/Ce=0.25) and Eu (Eu/Eu=0.58) anomalies (Eu/Eu=EuN/SmNGdN0.5) (Supplementary Table 2).

On the primitive mantle-normalized trace element spider diagrams (Figure 4(b)), most of the group 1 samples are generally enriched in Nb, La, and Pb and depleted in Ta, Ce, Sr, and Ti; while the group 2 samples are enriched in Ba, Nb, Ta, Pb, Zr, Hf, and Ti and are depleted in Th, La, Ce, Pr, Sr, Sm, and Eu (Figure 4).

4.2.2. Clinopyroxene

(1) Major Element Compositions. The analytical results of the pyroxene are reported in Supplementary Table 3. The pyroxene from sample 94DS0 are augite, and those from sample 94DS1 are diopside and augite [31] (Figure 5). The SiO2 contents of the clinopyroxene from sample 94DS0 are relatively high and vary slightly, ranging from 50.86 to 53.89 wt%, while the SiO2 contents of the clinopyroxene from sample 94DS1 are relatively low and vary considerably, ranging from 43.77 to 50.22 wt%. Several chemical differences between samples 94DS0 and 94DS1 were identified in this study. For example, the TiO2, FeO, and Na2O contents of the clinopyroxene from sample 94DS are lower than those from sample 94DS1, whereas the Cr2O3 contents of the former are higher than those of the latter, indicating that during its evolution, the magma became depleted in Fe, Na, and Ti. The Al2O3 contents of the clinopyroxene from sample 94DS (ranging from 1.75 to 4.87, with an average of 3.46) are lower than those from sample 94DS1 (ranging from 3.73 to 8.11, with an average of 6.27). The Mg#s (Mg#=Mg/Mg+Fe) of the clinopyroxene from sample 94DS1 (Mg# =56–75) are significantly lower than those from sample 94DS0 (Mg#=8088). The oxide contents of the clinopyroxene from sample 94DS1 are similar to those of the low Mg# clinopyroxene from the Okinawa Trough, while the oxide contents of the clinopyroxene from sample 94DS0 are similar to those of the high Mg# clinopyroxene from the Okinawa Trough [32] (Figure 6).

Figure 6 shows the compositional spectrum of the clinopyroxene. The clinopyroxene is characterized by low Al2O3 contents (less than 8.11%). Na2O and MnO do not systematically correlate with Mg#. The Cr2O3 and SiO2 contents are positively correlated with Mg#. The FeO, TiO2, Al2O3, and CaO contents are negatively correlated with Mg#. The inverse FeO-Mg#Cpx trend, which is self-correlated because the Mg#=100×Mg2+/Mg2++Fe2+, is purposely plotted to show the data’s coherence [33]. The Cr2O3 contents of the clinopyroxene decrease with decreasing Mg#Cpx. Cr is highly compatible in clinopyroxene [33, 34]. The decrease in CaO is due to the fractionation of clinopyroxene since clinopyroxene fractionation can reduce the CaO content of the residual melt [29].

(2) Trace Element Composition. The trace element concentrations of the clinopyroxene in the basalts from site 94DS are listed in Supplementary Table 4.

The ΣREE values of the clinopyroxene from sample 94DS0 range from 15.98 to 17.39 ppm, with an average of 17.32 ppm, which are significantly lower than those of the clinopyroxene from the Late Cenozoic basalts from the South China Sea (average of 62 ppm) and Thailand (average of 84 ppm) [35, 36]. Compared to the clinopyroxene from the Late Cenozoic basalts from the South China Sea and Thailand [35, 36], the clinopyroxene from sample 94DS0 exhibits LREE depletion (Figure 7), with relatively low ratios of LREE/HREE (ranging from 0.57 to 0.66, with an average of 0.60), LaN/YbN (0.13 to 0.17, average of 0.16), CeN/YbN (0.22 to 0.27, average of 0.25), SmN/YbN (0.87 to 0.91, average of 0.90), LaN/NdN (0.27 to 0.33, average of 0.29), and LaN/SmN (0.15 to 0.19, average of 0.17). The clinopyroxene from 94DS exhibits significant negative Eu anomalies (Eu/Eu=0.770.79, with an average of 0.78) but no obvious Ce anomalies, indicating the fractionation of plagioclase during the magmatic processes.

The ΣREE values of clinopyroxene from sample 94DS1 range from 35.33 to 112.07 ppm, with an average of 59.67 ppm, which is slightly lower than those of clinopyroxene from the Late Cenozoic basalts from the South China Sea (average of 62 ppm) and Thailand (average of 84 ppm) [35, 36]. Compared to the clinopyroxene from the Late Cenozoic basalts from the South China Sea and Thailand, the clinopyroxene from sample 94DS1 exhibits LREE depletion, with relatively low ratios of LREE/HREE (1.11–3.40, average of 1.70), LaN/YbN (0.38–3.40, average of 1.26), CeN/YbN (0.55–1.70, average of 0.89), SmN/YbN (1.25–1.76, average of 1.54), and LaN/NdN (0.32–2.24, average of 0.72). In addition, the clinopyroxene from sample 94DS1 exhibits slight negative Ce anomalies (Ce/Ce=0.440.94, average of 0.80) but no obvious Eu anomalies.

On the primitive mantle-normalized trace element spider diagrams (Figure 7), most of the samples are generally depleted in Large-ion lithophile elements (LILEs) (such as Ba and Sr), indicating the fractionation of plagioclase during the magmatic processes since Ba and Sr are mainly incorporated into K-rich and Ca-rich minerals in the form of isomorphism. The trace element and rare earth element characteristics of the clinopyroxene from sample 94DS1 are similar to those of the low Mg# clinopyroxene from the Okinawa Trough, while the trace element and rare earth element characteristics of the clinopyroxene from sample 94DS0 are similar to those of the high Mg# clinopyroxene from the Okinawa Trough, which is consistent with results of the major element compositions [32] (Figures 6 and 7).

4.2.3. Plagioclase

(1) Major Element Compositions. The analytical results of the plagioclases are reported in Supplementary Table 5. The plagioclase in this study is predominantly labradorite and bytownite, with minor andesine (Figure 8). The An values of the plagioclase from samples 100DS7 and 100DS18 range from 60.63 to 66.42 (average of 63.00) and 61.53 to 68.11 (average of 63.67), respectively. While the An values of the plagioclase from samples 94DS0 and 94DS1 range from 75.20 to 94.91 (average of 83.29) and 62.40 to 79.62 (average of 75.04), respectively. The chemical compositions of some of the plagioclase phenocrysts analyzed in this study show no obvious variations from core to rim, suggesting that the chemical compositions of the plagioclase phenocrysts are relatively uniform. In addition, the An contents of the plagioclase microlites are similar to those of the phenocryst rims.

(2) Trace Element Compositions. The trace element concentrations of the plagioclases are reported in Supplementary Table 6. On the chondrite-normalized REE patterns (Figure 9), all of the plagioclases are enriched in LREEs. Except for those from sample 94DS0, most of the plagioclase samples exhibit strong positive Eu anomalies; LREE, Ba, U, Sr, Pb, and Eu enrichments; and Th, Nb, Ce, Zr, and Ti depletions. While the plagioclase phenocrysts from sample 94DS0 exhibit negative Eu anomalies; Th, U, Pb, and Hf enrichments; and Ba, Nb, Sr, and Ti depletions (Figure 9). The plagioclase phenocrysts from sample 94DS0 are similar to the MATA-1 plagioclase xenocrysts from the Late Cenozoic basalts from Thailand [36], indicating that the plagioclase phenocrysts may be xenocrysts, rather than magmatic in origin, which were trapped by sample 94DS0 during its ascent to the surface. In addition, the trace element and rare earth element contents of the plagioclase in the basalts from site 100DS are significantly higher than those from site 94DS, and the trace element and rare earth element characteristics of the plagioclase from site 100DS are similar to those from Thailand [36].

5. Discussion

5.1. Timing for Benham Rise and CBF Volcanic Activity

The apparent age of basalt from the Benham Rise is 20.9 Ma, which is similar to the ages of basalts from Vinogradov Seamount (22.0 Ma), but is obviously younger than the ages of basalts from IODP site 292 (35.6-36.2 Ma) [2]. Thus, the age of 20.9 Ma could represent the extinction time of the Oki-Daito mantle plume, and the younger age (relative to those for IODP site 292 basalts) could represent late-stage volcanic activity that followed the main stage of volcanism on the Benham Rise.

The apparent age of basalt from the CBF is 29.1 Ma, which is similar to the previous K-Ar ages of basalts from CBF (27.4 Ma-28.1 Ma) [12, 22], suggesting the seafloor spreading of WPB was still active during this time, and this age could represent the time of the cessation of the seafloor spreading in WPB. The K-Ar ages of basalts from CBF show a decreasing trend from west-north to east-south, which indicates the time of the cessation of the seafloor spreading in WPB is gradually decreasing from west-north to east-south. Based on the reliable K–Ar and Ar–Ar ages of basalts from WPB, the half-spreading rate of WPB can be calculated as 4.8 cm/y between 29.1 and 44.4 Ma, and 5.2 cm/y between 29.1 and 36.2 Ma, respectively. The newly obtained values are slightly faster than the estimate of 4.4 cm/y [7], but slightly lower than the estimate of 5.5 cm/y [13].

5.2. Nature of the Source Lithology and Partial Melting

5.2.1. Constraints from Whole Rock Compositions

Basalts have long been regarded to be partial melts of peridotite, while a great deal of experimental studies have shown that silica-deficient alkali basalts can also be produced from silica-deficient eclogite and garnet pyroxenite, hornblendrite, and carbonated peridotite [35, 3841]. In recent decades, researchers have successfully used the Fe/Mn ratio, FC3MS parameter (FeOT/CaO3×MgO/SiO2), CaO content, Dy/Yb ratio, and Yb content to determine the nature of the source lithology [29, 35, 4245]. The basalts of Benham Rise have relatively low CaO contents (6.32%-9.36%) and high FC3MS parameter (0.72-1.14) and Fe/Mn ratio (51.23-70.92), which indicate that the source lithology of Benham Rise basalts is pyroxenite. While basalts of CBF have relatively high CaO contents (9.66%-9.99%) and low FC3MS parameter (0.39-0.46) and Fe/Mn ratio (34.82-37.50), which indicate that the source lithology of Benham Rise basalts is peridotite. Compared to spinel, garnet preferentially incorporates HREEs over LREEs, and basaltic rocks associated with garnet in the source have a high Dy/Yb ratio. The basalts of Benham Rise have a relatively high Dy/Yb ratio (1.98-2.40), indicating the mantle source of Benham Rise containing garnet. However, the basalts of CBF have relatively low Dy/Yb ratio (1.46–1.74), indicating the mantle source of Benham Rise containing spinel. Therefore, the source lithology of the basalts from the Benham Rise is garnet pyroxenite, and the source lithology of the basalts from the CBF is spinel-peridotite.

The estimated primary compositions of the basalts from the CBF and sites DSDP447 and ODP1201 plot within the experimental field are defined by the partial melting of peridotite, while the estimated primary compositions of the Benham Rise plot within the silica-deficient eclogite field (Figure 10) [38]. Furthermore, most of the basalts from the Benham Rise have high Fe/Mn (51.23-70.92) ratios and FC3MS values (>0.5), indicating that the mantle source of the basalts from the Benham Rise is pyroxenite rather than peridotite. In addition, the basalts from the Benham Rise have low SiO2 (44.13-48.54) and high FeO (8.62-9.34) contents, with no obvious depletions or enrichments in Zr and Hf, suggesting that neither carbonated peridotite nor hornblendite is a suitable mantle source [4648]. The effective melting pressure and melting temperature based on the primary melts are Pf=23.827.1kbar (with the depth of about 70-80 km) and T=1439°C1473°C (Table 2), which is in agreement with a source lithology of garnet lherzolite/pyroxenite, suggesting that the basalts of the Benham Rise were derived from a garnet-bearing mantle source. Thus, the bulk rock major and trace element compositions give a consistent result, i.e., that the source lithology of the basalts from the Benham Rise is garnet pyroxenite, and the source lithology of the basalts from the CBF and sites DSDP447 and ODP1201 is spinel-peridotite.

The LREE concentrations more likely reflect the extent of the degree of partial melting. Compared to the CBF, the basalts from the Benham Rise have higher La/Sm ratios (7.12-7.97), indicating that the extent of the degree of partial melting of the basalts from the Benham Rise is lower than that of those from the CBF. Calculated by the primary melt composition, the degree of melting of the samples from the CBF ranged from 10% to 11%, while the degree of melting of the samples from the Benham Rise ranged from 5% to 7%, which is consistent with the result of whole-rock composition.

In conclusion, the source lithology of the basalts from the Benham Rise is garnet pyroxenite, the source lithology of the basalts from the CBF and sites DSDP447 and ODP1201 is spinel-peridotite, and the degree of partial melting of the basalts from the Benham Rise is lower than that of those from the CBF.

5.2.2. Constraints from Olivine Compositions

Olivine is the first silicate mineral that crystallizes from all mantle-derived magma as they ascent to the surface. Thus, the olivine can be used to fingerprint the mantle source of basaltic magma. Sobolev et al. [49] showed that the presence of pyroxenite in mantle source can be identified by the existence of olivines containing high Ni contents and Fe/Mn ratios but low Ca and Mn contents. In addition, the correlations of Mn/Zn versus Ni, Zn versus Mn, and 10000Zn/Fe versus 100Mn/Fe in olivines are reliable indicators of pyroxenite versus peridotite sources [50]. Howarth and Harris [50] showed that olivines crystallized from pyroxenite source melts have low Mn/Zn ratios (<13) while those from peridotite source melts have high Mn/Zn ratios (>15). As shown in Figure 11, except olivine xenocryst, all the olivines of sample 100DS7 trend toward pyroxenite sources, which are consistent with the results of whole rock. However, the olivine xenocryst of sample 100DS7 plotted in the field of peridotite source, suggesting that the olivine xenocryst originated from peridotite source.

In summary, we conclude that the source lithology of the basalts from the Benham Rise is garnet pyroxenite, and the source lithology of the basalts from the CBF and sites DSDP447 and ODP1201 is spinel-peridotite.

5.3. Primary Melt Compositions and Mantle Potential Temperatures

The primary melt composition refers to the partial melts that equilibrate with the source rocks, which can be used to constrain the thermal state of the mantle source and the source lithology [42, 5158]. The magmatic rocks usually undergo variable crystallization and assimilation, which modify the composition of the primary melts. Previous studies have proposed that mantle-derived rock samples with MgO>7.5wt% are generally believed to have only experienced the addition or subtraction of olivine [59, 60]. Therefore, only the CBF basalts with MgO contents of >8 wt% are chosen as the starting composition. In addition, the predominant phenocrysts in the Benham Rise basalts are plagioclase, with minor olivine. We used the reverse crystallization model in Petrolog to minimize the influence of plagioclase, and then, we used the reversed melt as the starting composition. The primary melt compositions were estimated by adding small (0.1%) increments of olivine (Fo=90.1) to the samples, assuming an olivine-melt distribution coefficient KDFe/Mgol/liq=0.31 and Fe2+/Fe2++Fe3+=0.9 in the melt [61].

Because the Fo contents of olivine phenocrysts are relatively low (<76.10%), we do not use olivine phenocrysts to calculate the mantle potential temperatures of Benham Rise. The melting conditions and mantle potential temperatures were estimated based on the primary melt compositions [29, 42, 51, 55, 56, 62]; (Herzberg 2007) (Table 2). The mantle potential temperature beneath the CBF varies from 1345°C to 1381°C (average of 1363°C), which is similar to those from sites 447 (1315°C–1364°C, average of 1334°C) and 1201 (1301°C–1363°C, average of 1323°C) (Table 2). The mantle potential temperature beneath the Benham Rise varies from 1439°C to 1473°C (average of 1454°C), which is significantly higher than that of the basalts from the CBF (T=58°C128°C) and sites 447 (T=75°C158°C) and 1201 (T=76°C172°C), indicating that the mantle source of the basalts from the Benham Rise is related to a mantle plume. Based on the primary melts, the effective melting pressure (Pf) and melting temperature (T) of the CBF basalts are Pf=10.011.4kbar (average of 11.1 kbar, with the depth about 30-35 km) and T=1313°C1342°C (average of 1328°C), which are similar to those of the basalts from sites 447 (Pf=9.312.4kbar; T=1306°C1337°C) and 1201 (Pf=9.414.0kbar; T=1279°C1350°C) but are lower than those of the Benham Rise basalts (Pf=23.827.1kbar, average of 24.7 kbar; T=1409°C1440°C, average of 1421°C). The degree of melting of the samples from the CBF ranged from 10% to 11%, while the degree of melting of the samples from the Benham Rise ranged from 5% to 7%. In conclusion, the mantle potential temperature beneath the CBF is lower than that beneath the Benham Rise, suggesting there existed thermal anomaly during basaltic magmatism in the Benham Rise, which may relate to a mantle plume (Table 2).

Figure 12 shows a plot of the effective melting pressure vs. temperature. We conclude that the temperatures and pressures of the primary melt compositions of the CBF basalts plot above the spinel-garnet lherzolite transition (~50–60 km), but the temperatures and pressures of the primary melt compositions of the Benham Rise basalts plot below the spinel-garnet lherzolite/pyroxenite field [6366].

5.4. Physical Conditions of Crystallization

5.4.1. Physical Conditions of the Crystallization of the Basalts from Site 94DS

(1) Crystallization Pressure and Temperature of the Clinopyroxene. In recent decades, many researchers have proposed various thermobarometers for calculating the crystallization pressures and temperatures of clinopyroxene [62, 67, 68]. Putirka et al. [68] established a series of thermodynamic equations based on experimental work that relates the temperature and pressure to the equilibrium constants and allows for the construction of clinopyroxene-liquid thermobarometers. However, these thermobarometers are only applicable to basaltic magma temperature and pressure calculations, with an application range of 1110°C–1475°C and 0.4–0.9 GPa. Putirka [62] experimentally recalibrated the thermobarometers and established a new equilibrium temperature and pressure formula that can be applied to mafic magmas and intermediate-silicic magmas. In this study, we used the Putirka [62] equation to calculate the crystallization pressures and temperatures of the clinopyroxene (Table 3). The KD (Fe-Mg)cpx-liq value can be used to estimate whether the clinopyroxene is in equilibrium with the host rock (the equilibrium KD (Fe-Mg)cpx-liq ranging from 0.20 to 0.36), and the KD (Fe-Mg) of the clinopyroxene in this study varies from 0.25 to 0.29, indicating that all of the clinopyroxene are in equilibrium with the host rock. The calculated results show that the temperature and pressure of the basalts from sample 94DS0 are 1170°C–1225°C (average of 1196°C) and 0.5–1.0 GPa (average of 0.8 GPa), respectively, while the temperature and pressure of the basalts from sample 94DS1 are 1112°C–1157°C (average of 1127°C) and 0.1–0.5 GPa (average of 0.3 GPa), respectively. This suggests that the crystallization depth of the clinopyroxene from sample 94DS0 was deeper than that of sample 94DS1, which is consistent with the clinopyroxene having higher Mg#s (Table 3). In addition, the crystallization temperatures and pressures of the clinopyroxene in the basalts from site 94DS are similar to those from the middle of the Okinawa Trough (1121°C–1212°C and 0.1–1.2 GPa) [32]. Based on the calculated pressures, the depth of the magma chambers of samples 94DS0 and 94DS1 was calculated from the equilibrium temperatures and pressures of the clinopyroxene, and melt can be inferred to be 8–25 and 6–17 km, respectively.

(2) Crystallization Pressure and Temperature of the Plagioclase. Kudo and Weill [69] created the first plagioclase-liquid geothermometer. Subsequently, many workers have various thermobarometers for calculating the crystallization pressures and temperatures of plagioclase formation. In addition, since the geothermometer of Kudo and Weill [69] does not consider whether the melt and plagioclase reached equilibrium or the effect of water on plagioclase crystallization, we used the pressure and temperature calculations of Putirka [62]. The KD (Ab-An)pl-liq of the plagioclase in the basalts from site 94DS indicates that most of the plagioclases are in equilibrium with the basalts composition (the equilibrium KD (Ab-An)pl-liq ranging from 0.16 to 0.38), and the plagioclase, which is not in equilibrium with the basalts composition, was not considered in this study. The H2O content of the basalts from samples 94DS1 and 94DS0, which were calculated using the thermobarometer of Putirka [62], is 1.25–1.51 wt% (average of 1.37 wt%) and 1.40–2.43 wt% (average of 1.65 wt%). The crystallization temperatures and pressures of the plagioclase from samples 94DS1 and 94DS0 are 1165°C–1180°C and 0.2–0.4 GPa and 1174°C–1179°C and 0.1–0.2 GPa, respectively (Table 4), which are slightly lower than those of the South Okinawa Trough (1218°C–1239°C and 0.6–0.9 GPa) but are similar to those of the Mariana Trough (975°C–1212°C) [37, 70].

Compared with the crystallization pressures and temperatures of the plagioclase and clinopyroxene from basalt sample 94DS0, the crystallization pressures and temperatures of the clinopyroxene from sample 94DS1 are slightly lower than those of the plagioclase, while the crystallization pressures and temperatures of the clinopyroxene from sample 94DS0 are slightly higher than those of the plagioclase. Thus, the results indicate that the order of crystallization of sample 94DS1 is plagioclase and clinopyroxene, while the clinopyroxene from sample 94DS1 has no obvious negative Eu anomalies (Figure 7), suggesting that few or no plagioclase crystallized before the formation of the clinopyroxene. In addition, the obvious negative Eu anomalies in the microlite clinopyroxene in sample 94DS0 are attributed to plagioclase crystallization before the formation of the clinopyroxene.

5.4.2. Physical Conditions of the Crystallization of the Basalts from Site 100DS

The crystallization temperatures and pressures of the plagioclase from basalt sample 100DS18 are 1141°C–1145°C (averagevalue=1143°C) and 1.3–1.4 GPa (averagevalue=1.4GPa), respectively, and the crystallization temperatures and pressures of the plagioclase from basalt sample 100DS7 are 1133°C–1140°C (average of 1136°C) and 1.4–1.5 GPa (average of 1.4 GPa), respectively. This indicates that the plagioclase from samples 100DS7 and 100DS18 crystallized under similar physical conditions. The crystallization temperatures and pressures of the plagioclase from the Benham Rise can be comparable to those from Thailand (1145°C–1214°C and 0.4–0.9 GPa) and the South China Sea basin (927°C–1179°C) [36, 71]. In addition, the temperatures of the plagioclase from the Benham Rise basalts are similar to those from the CBF, while the pressures of the plagioclase from the Benham Rise basalts (1.3–1.5 GPa) are significantly higher than those from the CBF (0.1–0.4 GPa), suggesting that the plagioclase from the WPB formed at a higher pressure. The H2O content of the melt under the Benham Rise, which was calculated using the data for the plagioclase, is significantly higher than that of the CBF, indicating that the magma under the Benham Rise had a higher H2O content (Table 4). Though the mantle potential temperature beneath the Benham Rise is significantly higher than that of the CBF, the temperatures of the plagioclase from the Benham Rise basalts are similar to those of the CBF, suggesting that the temperature conditions of the mantle source had little or no influence on the temperature conditions of the magma crystallization differentiation at shallow depths.

5.5. Influence of the Subduction Component

Previous studies have shown that the start of the subduction under the Izu-Bonin-Mariana arc, the onset of the rifting and spreading of the WPB, and the effect of the mantle plume occurred in almost the same period [13]. Thus, it is important to discuss whether the magmatism in the WPB is affected by the subduction component. La is mobile in high-temperature hydrous melts [72], Sr and Ba are mobile in low-temperature aqueous fluids, and HREEs, including Y, are immobile during subduction. Using the method of Sun et al. [73] and the trace element compositions of the plagioclase of basalts from site 94DS, we calculated the Sr/Y, Ba/Y, and La/Y ratios of the hypothetical parental melt. In Figure 13, the Sr/Y and Ba/Y ratios of the calculated melt of sample 94DS1 and the basalt composition of sample 94DS1 (similar to the calculated melt based on the composition of plagioclase from the South Okinawa Trough) are higher than that of MORB, which indicates the effect of a subduction component on the mantle source of sample 94DS1 [30, 37]. The Sr/Y ratio of the calculated melt of sample 94DS0 and the basalt composition of sample 94DS0 are similar to those of MORB, and the Ba/Y ratio of the calculated melt of sample 94DS0 and the basalt composition of sample 94DS0 are higher than those of MORB, indicating the contribution of a subduction component to the magma source of sample 94DS0. In addition, the Sr/Y and Ba/Y ratios of the basalts from site DSDP447 plot within the MORB field, while the Sr/Y ratios of the basalts from site ODP1201 are slightly higher than those of MORB, suggesting that the subduction component had little or no influence on the magma source of the basalts from sites DSDP447 and ODP1201 [17].

In conclusion, the magma source of site 94DS was affected by a subduction component, while the subduction component had little or no influence on the magma source of the basalts from sites DSDP447 and ODP1201.

5.6. Tectonic Setting and Geological Evolution

The West Philippine Basin contains various rock types, including N-MORB, OIB, and arc volcanic rocks, and its geodynamics background is related to back-arc spreading, a mantle plume, and a volcanic arc [2, 4, 5, 13, 16, 17]; (Haraguchi et al. 2011). Thus, the tectonic setting of the samples in this study needs to be further constrained. In the plot of Y3-Ti/10-Zr (Figure 14(a)) [74], most of the samples from site 100DS and sites IODP 292, 294, and 446 plot in the within plate basalt field (WPB), while all of the samples from site 94DS and sites IODP 447 and ODP1201 plot within the ocean-floor basalts or cal-alkali to low-potassium tholeiite fields. On the plot of Zr vs. Zr/Y (Figure 14(b)) [75], most of the samples from site 100DS and sites IODP 292, 294, and 446 plot in the WPB field, while all of the samples from sites 94DS, DSDP447, and ODP1201 plot within the overlap between the MORB and IAB (island arc basalt) fields. Based on the results described above, the basalts from sites 94DS, DSDP447, and ODP1201 were formed in a similar tectonic setting, which is related to a mid-ocean Ridge setting. The basalts from site 100DS and sites IODP 292, 294, and 446 have a similar geodynamic background, which was affected by a mantle plume.

Recent studies have shown that magma with a surface area of about 0.11 Mkm2 and a volume of 0.13 Mkm3 extruded within 5 Myr, and the geochemical characteristics of the basalts from the Benham Rise are similar to OIB, indicating that the Benham Rise is likely a large igneous province [76]. In addition, the morphology of the Benham Rise and the thicker crust (about 15 km) calculated from the gravity anomalies beneath the Benham Rise supports the viewpoint that the volcanism is attributed to the interaction between a hotspot (or mantle plume) and a spreading ridge.

Based on the reliable K-Ar and Ar-Ar ages of basalts from WPB, we proposed a tectonic evolution model for WPB (Figure 15).

Before 52 Ma, there existed some late Mesozoic terranes (including Gagua Ridge, Daito Ridge Group, Daito Ridge, and Amami Plateau), and there was no volcanic activity in the WPB (Figure 15(a)) [13].

Between 51-45 Ma, the Oki-Daito plume arrived at the spreading center of WPB about 45 Ma, resulting in widespread OIB-like volcanic edifices and the spreading of WPB until 35 Ma [13]. The spreading of the WPB may lead to the split of Daito Ridge Group, and push part of Daito Ridge Group northward to form Daito Ridge in the north-east of CBF (Figure 15(b)). During this period, Benham Rise was formed on the south-west side of the spreading center, and a chain of age-progressive oceanic Plateaus (Urdaneta Plateau) was formed on the north-east side of the spreading center of WPB (Figure 15(c)) [13].

At about 35 Ma, the spreading center jumped to north-east of Oki-Daito plume, forming a chain of Seamounts (e.g., Toog Seamount, Santos Seamount, and Akle Seamount) in the south-west of the spreading center (Figure 15(d)) [76]. However, the ages of these Seamounts are unknown, which hinder us to understand the effect of the Oki-Daito plume.

Based on the age of basalts from CBF, the cessation time of the spreading of WPB was about 29 to 27 Ma (Figure 15(e)). After the cessation of the spreading of WPB, The Vinogradov Seamount formed near the spreading center about 22 to 20 Ma (Figure 15(f)), which represents the extinction time of the Oki-Daito mantle plume.

6. Conclusions

  • (1)

    This study presents compositional characteristics of minerals in basalts from the Benham Rise and the CBF as follows, Olivines from the Rise have forsterite contents of 56.90%–90.80% and have high ΣREE values (0.05–0.43 ppm). The clinopyroxenes from the CBF are diopside and augite, and they are depleted in LREEs (LaN/YbN=0.133.40) and LILEs. The plagioclases are predominantly labradorite and andesine, with minor bytownite, and there exist obvious differences in plagioclase compositions between these two tectonic sites

  • (2)

    The source lithology of the Benham Rise basaltic rocks could be garnet pyroxenite, and yet that of the CBF could be spinel-lherzolite. The degree of melting of the samples from the CBF ranged from 10% to 11%, while the degree of melting of the samples from the Benham Rise ranged from 5% to 7%. The magma source of sample 94DS was affected by a subduction component

  • (3)

    The calculated mantle potential temperature beneath the Benham Rise is 1439°C–1473°C, which is significantly higher than that of the CBF (1345°C–1381°C), suggesting there existed thermal anomaly beneath the Rise during basaltic magmatism, which may relate to a mantle plume

  • (4)

    This study developed a conceptual model for WPB geological evolution and suggested the extinction of the Oki-Daito mantle plume activity might be at about 20.9 Ma, and cessation of the back-arc spreading of WPB was at about 29.1 Ma or younger

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Acknowledgments

We thank Prof. Sanzhong Li and two anonymous reviewers for their constructive comments and suggestions, and Songjian Ao for editorial handling. This work was supported by the National Key Research and Development Program of China (Grant No. 2017YFC0602305), the National Natural Science Foundations of China (grants nos. 41776070, U1606401, 41276003, and 41322036), AoShan Talents Program Supported by the Qingdao National Laboratory for Marine Science and Technology (No. 2015ASTP-ES16), and Taishan Scholarship from Shandong Province. We thank LetPub (http://www.letpub.com) for its linguistic assistance during the preparation of this manuscript.

This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.