Lithological structure of western Pacific lithosphere reconstructed from mantle xenoliths in a petit-spot volcano

Abstract

The lithospheric mantle, formed at the mid-ocean ridge as a residue of crustal production, comprises theoretically depleted peridotite, but more fertile components (e.g., lherzolite and pyroxenite) have been reported, creating an enigmatic picture of the lithosphere. The oceanic lithosphere has also been found to be locally modified by intraplate magmatism as proposed from geochemistry of mantle xenolith. Petit-spot xenoliths are particularly notable as direct evidence of old lithospheric mantle and expected to retain essential information about oceanic lithosphere prior to its subduction. In this study, we report on the lithological structure of Pacific lithosphere aged at 160 Ma, just subducting into Mariana Trench, based on petrology and chemistry of ultramafic xenoliths from a petit-spot knoll, and then, we suggest the occurrence of petit-spot melt infiltration resulting in mantle metasomatism and formation of pyroxene-rich vein. Our petit-spot ultramafic xenoliths can be divided into three main types: a depleted peridotite as a residue of crust production, an enriched peridotite, and fertile pyroxenites as the product of melt–rock interactions prior to entrapment. Geothermobarometry also suggests that the depleted peridotite was derived from the uppermost lithospheric mantle, whereas the enriched peridotite and Al-augite pyroxenites were obtained from deeper layers of the lithosphere. Moreover, thermal gradient of the lithosphere estimated from these data is considerably hotter than pristine geotherm estimated on the basis of plate age. Hence, we could illustrate that the oldest portion of the Pacific lithosphere (160 Ma), which was not observed before, was locally fertilized and heated by prior multiple petit-spot magmatic events, and pyroxene-rich metasomatic veins penetrated from the base to the middle/upper lithosphere. Such local lithospheric fertilization is plausible at the plate-bending field, and the nature of Pacific Plate subducting into Mariana Trench may be partly different from what has been assumed so far.

1 Introduction

Pacific lithosphere, formed at the triple-ridge junction in Early Jurassic, is composed of the basaltic crust and lithospheric mantle (e.g., Müller et al. 2008; Boschman and Van Hinsbergen 2016). In theory, a high degree of melting is expected in the uppermost mantle column as a consequence of basaltic crust production in the mid-ocean ridge and upper lithospheric mantle supposedly comprises of depleted peridotite (e.g., Langmuir et al. 1992).

Studies on the oceanic lithospheric mantle have been conducted in limited areas such as ophiolites (e.g., Meckel et al. 2003; Anma et al. 2009; Akizawa and Arai 2009; Dilek and Morishita 2009; Dilek and Furnes 2014; Goodenough et al. 2014), abyssal peridotites exposed along transform faults in mid-ocean ridges (e.g., Dick and Natland 1996; Niu et al. 1997; Mallick et al. 2015; Regelous et al. 2016; Warren 2016), oceanic core complexes at slow-spreading ridges (e.g., Ohara et al. 2001; Escartín et al. 2008; Blackman et al. 2009; Olive et al. 2010; Pressling et al. 2012; Akizawa et al. 2021; Sen et al. 2021; Harigane et al. 2022), and mantle xenoliths in intraplate volcanoes (e.g., Clague 1988; Hauri and Hart 1994; Sen et al. 2005; Bizimis et al. 2007, 2013; Jackson et al. 2016; Snortum et al. 2019). These studies revealed that the lithospheric mantle mainly comprises depleted peridotites, but a significant number of fertile components crosscutting the depleted peridotites were also recognized. Although the compositional diversity can be caused by mid-ocean ridge magmatism as a function of multiple melt infiltrations or the vertical variations in degree of partial melting (Plank and Langmuir 1992; Akizawa et al. 2012; Rochat et al. 2017), local fertilizations can occur via intraplate volcanism (e.g., Hauri et al. 1993).

A number of studies on mantle xenoliths, brought by hotspot magmatisms on Pacific Plate, have proved the occurrence of chemical, thermal, and petrological modification of lithospheric mantle by intrusive magma (Tracy 1980; Fodor et al. 1982; Clague 1988; Hauri and Hart 1994; Qi et al. 1994; Dieu 1995; Burnard et al. 1998; Sen et al. 2005; Bizimis et al. 2007, 2013; Jackson et al. 2016; Snortum et al. 2019). Such local magma infiltration in the lithosphere has also been suggested at the older Pacific Plate just subducting into a trench based on fresh mantle xenoliths brought by petit-spot magma (Yamamoto et al. 2009, 2014; Pilet et al. 2016).

Mantle xenoliths from petit-spot volcanoes can provide direct information of the Pacific lithosphere below an abyssal plain. Petit-spot magma is derived from the lithosphere–asthenosphere boundary, and its xenoliths certainly originate from the lithosphere (Hirano et al. 2006). For example, the geochemistry and geothermobarometry of petit-spot mantle xenoliths revealed that the old portion (130–135 Ma) of the Pacific lithosphere is thermally and/or compositionally modified by prior petit-spot magmatism events, and it has been suggested that the nature of the subducting lithosphere might be different from what we had previously assumed (Yamamoto et al. 2009, 2014; Pilet et al. 2016). Here, we report the various small ultramafic xenoliths derived from the oldest part of Pacific lithosphere (ca. 160 Ma) to discuss the existence of residual mantle and the occurrence of local metasomatism resulted from intrusive petit-spot magma.

2 Background

2.1 Petit-spot volcanic field in the western Pacific

Petit-spot is a form of small scale magmatism caused by lithospheric stresses arising from plate-subduction; it is believed to originate from the lithosphere–asthenosphere boundary (Hirano et al. 2006; Machida et al. 2015, 2017). The petit-spot volcanic cluster in this study, first reported by Yamamoto et al. (2018), is surrounded by Cretaceous seamounts and oceanic islands of the Western Pacific Seamount Province (WPSP; Koppers et al. 2003) (Fig. 1a). The study area is in the abyssal plain southeast of Minamitorishima (Marcus) Island, corresponding to the oldest portion of the Pacific Plate, at the northern tip of the Jurassic Quiet Zone (Fig. 1b) (Tivey et al. 2006). Young and fresh petit-spot lava samples were collected for the first time from a knoll in this cluster by two submersible divings of 6K#1203 and 6K#1206 using a SHINKAI 6500 deep-submergence vehicle (DSV) during the YK10-05 research cruise of the R/V Yokosuka from the Japan Agency for Marine-Earth Science and Technology (Fig. 1c). According to geochemical features and estimated eruption ages (< 3 Ma), this knoll is expected to have erupted owing to a swell in the plate that developed in response to subduction beneath the Mariana Trench (Hirano et al. 2019). Another petit-spot knoll was newly investigated by submersible diving of 6K#1466 during the YK16-01 research cruise (Figs. 1c, 2a). Notably, the basalt samples include ultramafic xenoliths, as stated later in the sample description and petrography section.

Fig. 1

Bathymetric and seafloor age map around the study area. a The bathymetry of the western Pacific near the Mariana Trench. The red box shows the study area, to the southeast of Minamitorishima (Marcus) Island. Bathymetric data are from ETOPO1 (NOAA National Geophysical Data Center; http://www.ngdc.noaa.gov/). b The seafloor age map of same area as in (a). This study area is within a 160–170 Ma region called the Jurassic Quiet Zone (JQZ) (Tivey et al. 2006). The absolute motion of the Pacific Plate in this area is from Gripp and Gordon (1990). The seafloor age data are from Müller et al. (2008). c A detailed bathymetric map of the study area. The onboard multibeam data are from the YK10-05 cruise by JAMSTEC. The petit-spot knoll explored in this study, surveyed by the 6K#1466 dive, is displayed in a white box; that surveyed by dives 6K#1203 and 6K#1206 was reported by Hirano et al. (2019)

Fig. 2

Petit-spot knoll and sample descriptions. a The small conical petit-spot knoll and sampling point of the 6K#1466 dive by SHINKAI 6500 (pink stars). Three basaltic lavas, 6K#1466R6-001, R7-001, and R7-003, were collected from locations near the western slope of the knoll. b, c The representative basalt sample (6K#1466R6-001) and its photomicrograph. d–g The occurrence of ultramafic xenoliths: d MPSX-PR1, e MPSX-PR2, f MPSX-PX1, and g MPSX-PX5

2.2 Xenoliths from Pacific lithosphere

A number of studies reported on mantle xenoliths in the Pacific Plate: Hawaiian Islands, Samoan Islands, Cook Islands, Tahiti, seamounts on the WPSP, and petit-spot volcanoes (e.g., Tracy 1980; Fodor et al. 1982; Clague 1988; Hauri and Hart 1994; Qi et al. 1994; Dieu 1995; Burnard et al. 1998; Sen et al. 2005; Hirano et al. 2006; Bizimis et al. 2007, 2013; Jackson et al. 2016; Snortum et al. 2019). Xenoliths from the Hawaiian Island chain are mainly harzburgite, lherzolite, wehrlite, dunite, and crustal rocks, principally from Kilauea lava, Honolulu Volcanics, and Loihi seamount (e.g., Jackson and Wright 1970; Clague 1988; Wagner and Grove 1998). In the Honolulu Volcanics, diverse types of xenoliths including harzburgite, lherzolite, pyroxenite (with and without garnet), and dunite have been reported mainly in rejuvenated-stage lava (Clague and Frey 1982; Roedder 1983; Clague 1988; Sen et al. 2005; Bizimis et al. 2005, 2007). The Samoan xenoliths are harzburgite and dunite, which show various degrees of light-rare-earth element (LREE) concentrations in correlation with equilibrium temperatures, implying that the carbonatite and/or silicate metasomatism was due to the early stage of the Samoan mantle plume rather than to host magmatism (Hauri and Hart 1994; Burnard et al. 1998; Ashley et al. 2020; Byerly et al. 2021). Olivine nephelinite on Aitutaki of Cook Islands and basalts in Tahiti include ultramafic xenoliths of both Cr-diopside series and Al-augite series, suggesting previous melt infiltrations before entrapment as well (Tracy 1980; Fodor et al. 1982; Qi et al. 1994). Dieu (1995) classified ultramafic xenoliths from Limalok guyot and MIT guyot in the WPSP into type 1 (residual mantle) and type 2 (cumulate from basaltic melt), suggesting the occurrence of carbonatite metasomatism in the residual mantle. It is almost certain that several portions of the Pacific lithosphere below the hotspot seamounts/islands were modified by plume-related magmatism, although some depleted xenoliths were reported.

Xenoliths from petit-spot volcanoes are unique samples having information about the older part of Pacific lithosphere just before subduction. Xenoliths from previously reported petit-spots are basalt (crust origin), dolerite, gabbro, and peridotite (Hirano et al. 2006). Studies on ultramafic xenoliths and concomitant olivine xenocrysts in the petit-spot basalts have been limited to those in the NW Pacific (Hirano et al. 2004; Harigane et al. 2011; Yamamoto et al. 2009, 2014; Pilet et al. 2016). Peridotite xenoliths from the northwestern (NW) Pacific petit-spot showing a deep origin (< ~ 45 km: Yamamoto et al. 2014) have contributed to the understanding of deeper portions of the lithosphere. In contrast, the fact that no xenoliths originating from layers deeper than ~ 45 km have been reported has helped to clarify the petit-spot magma eruption process (i.e., magma stagnation with fractionation in the middle of the lithosphere; Machida et al. 2017). Based on geochemical results and a numerical simulation, Pilet et al. (2016) suggested that the lithospheric mantle was locally refertilized with formation of metasomatic veins because of multiple petit-spot magmatism events before entrapment. This was supported by noble gas isotopic ratios and geothermobarometry of petit-spot mantle xenoliths (Yamamoto et al. 2009, 2014). Such magma percolation in the oceanic lithosphere could occur ubiquitously because petit-spot magmatism is plausibly a universal geological phenomenon at the plate-bending fields and/or where the small amount of volatiles exists below lithosphere (e.g., Hirano et al. 2008; Hirano 2011).

3 Sample description and petrography

Basalt samples were collected at near the summit of the small conical knoll located approximately 100 km south of Minamitorishima Island (Figs. 1c, 2a). The conical knoll is ca. 160 m in height and ca. 1 km in diameter, with a summit water depth of ca. 5300 m (Fig. 2a). The volume of the volcanic edifice is estimated at approximately 0.04 km³, which is comparable to those of other petit-spot knolls in the NW Pacific (0.001–1 km³; Hirano et al. 2008). The image of outcrop is shown in Machida et al. (2021) which explored the petit-spot lava flows by combining a sub-bottom profiler with a multibeam echo sounder. Three basaltic lava samples, named 6K#1466R6-001, 6K#1466R7-001, and 6K#1466R7-003, contain small (< 1 cm in diameter) xenoliths and xenocrysts. Seven xenoliths, one xenocryst, and some isolated (handpicked) clinopyroxene xenocrysts were selected from the three samples.

3.1 Host basalt

The three host basalts are highly vesicular (20–40 vol.% vesicularity) and mantled by 4.3–5.2-mm-thick manganese crusts over 5.5–6.6-mm-thick palagonite rinds (hydrated quench glass) (Fig. 2b). The basalts contain needle-shaped Ti-augite (50–400 µm in size), subhedral olivine with aureoles of iddingsite (up to 100 µm in size), opaque minerals (up to 10 µm in size), glass, and crystallite, without remarkable phenocrysts (Fig. 2c). Young eruption ages were estimated, ranging from 5.2 to 0.4 Ma using the average thickness of manganese crusts and their deposition rate at the seafloor (1–10 mm/Myr; Hein et al. 1999) or ranging from 2.2 to 0.2 Ma using the average thickness of palagonite and its hydration rate (0.03–0.3 mm/Myr; Moore et al. 1985). These are based on macroscopic and microscopic observations.

3.2 Xenoliths and xenocrysts

As the xenoliths are too small to determine lithology from the modal composition (< 1 cm in diameter; Fig. 2d–g and Additional file 1: Fig. S1), we simply use the terms peridotite for olivine-rich xenoliths and pyroxenite for pyroxene-rich xenoliths. The samples are identified by sequential numbers with the abbreviations of MPSX-PR (Minamitorishima petit-spot xenolith peridotite) or MPSX-PX (Minamitorishima petit-spot xenolith pyroxenite). Entire scanning electron microscope images of the xenoliths are shown in Additional file 1: Fig. S1.

MPSX-PR1 peridotite is approximately 7 mm in size. Its constituent minerals are olivine, clinopyroxene, orthopyroxene, and spinel, without visible reaction rims or mineral overgrowths derived from reaction with the host basalt (Fig. 3a). Orthopyroxenes with lamellae are larger (< 1.5 mm in diameter) than the olivines and clinopyroxenes (0.1–0.7 mm in diameter). MPSX-PR2 peridotite is approximately 7 mm in size, and composed of olivine, clinopyroxene, and orthopyroxene. The olivines and orthopyroxenes range in size up to 3 mm and from 0.5 to 0.7 mm, respectively, whereas clinopyroxenes with orthopyroxene lamellae range in size from 0.2 to 1.2 mm (Fig. 3b). Thin (< 10 µm in width) reaction rims are observed all around the contact part with the host basalt. A spongy texture occurs along the grain boundaries of clinopyroxenes with lamellae (Fig. 3b).

Fig. 3

Scanning electron microscope images of the samples. a MPSX-PR1 peridotite, b MPSX-PR2 peridotite, c MPSX-PX1 pyroxenite, d MPSX-PX2 pyroxenite, e MPSX-PX3 pyroxenite, f MPSX-PX4 pyroxenite, g MPSX-PX5 pyroxenite, and h MPSX-PX6 clinopyroxene xenocryst

MPSX-PX1 pyroxenite (~ 8 mm in size) consists of coarse-grained (0.5–3 mm in diameter) clinopyroxene and orthopyroxenes with thin reaction rims in contact with the groundmass. These are surrounded partly by micrometer-sized clinopyroxenes (mainly Ti-augite of a few tens of µm), resembling those in the groundmass of the host basalt (Fig. 3c). The clinopyroxene shows spongy texture as well. MPSX-PX2 and MPSX-PX3 pyroxenites are 3 and 4.5 mm in diameter, respectively. They are mostly monocrystalline orthopyroxene accompanied by small (0.1–0.5 mm in diameter) clinopyroxenes. Needle-shaped micrometer-sized Ti-augite embedded in a glass matrix completely surrounds both samples (Fig. 3d, e). MPSX-PX4 and MPSX-PX5 pyroxenites are approximately 5 and 9 mm in size, respectively. They are composed mainly of orthopyroxene and clinopyroxene. The clinopyroxenes are spongy textured with orthopyroxene lamellae and/or blebs (Fig. 3f, g). Ti-magnetite occur in porosities of spongy texture, and thin reaction rims observed as well (Fig. 3f, g). MPSX-PX6 clinopyroxene xenocryst (3–4 mm in diameter) has a spongy texture and a micron-sized Ti-augite rim as well (Fig. 3h).

As a whole, the petrographic characteristics of xenoliths are summarized as the following three points: (1) There are no significant petrographic differences between the two peridotite xenoliths; (2) the pyroxenites are divided into three types: clinopyroxene- and orthopyroxene-bearing aggregates, orthopyroxene lamellae/blebs in clinopyroxene, and clinopyroxene grains in orthopyroxene; and (3) the spongy texture occurred predominantly in the outer part of clinopyroxene grains, except for MPSX-PR1.

4 Analytical methods

Major-element compositions of minerals were determined using an electron probe microanalyzer (EPMA; JXA-8900R) at Atmosphere and Ocean Research Institute, the University of Tokyo. The analyses were performed using an accelerating voltage of 15 kV, a beam current of 12 nA, and a beam diameter of 3 µm. A peak counting time of 20 s and a background counting time of 10 s were used, except for Ni, for which a peak counting time of 30 s and a background counting time of 15 s were used. Natural and synthetic minerals were used as standards. The elemental mapping was also conducted using an accelerating voltage of 15 kV, a beam current of 100 nA, and a probe diameter of 15 µm dwelled at 80 ms. Each map was acquired in 1–5 h with a step size of 15 or 20 µm at resolutions ranging between 220 and 430 pixels in X by 220–500 pixels in Y. The profile analyses were performed for selected clinopyroxene grains with a step size of 5–60 µm.

The trace-element compositions of minerals were determined using a laser ablation-inductively coupled plasma mass spectrometry (LA-ICP-MS; New Wave Research UP-213 and Agilent 7500s) at Kanazawa University. The Nd: YAG deep UV (ultraviolet) laser’s wavelength is 213 nm. The analyses were conducted with 100 µm spot size (Fig. 3). A repetition frequency of 6 Hz and a laser energy density of 8 J cm⁻² were used. NIST612 glass (distributed by National Institute of Standards and Technology) was employed for calibration, using the preferred values of Pearce et al. (1997). Data reduction was undertaken with ²⁹Si as the initial standard, and SiO₂ concentrations obtained by an electron microprobe analysis (Longerich et al. 1996). BCR-2G (distributed by the United States Geological Survey) was used as a secondary standard to assess the precision of each analytical session (Jochum and Nohl 2008).

Whole-rock major- and trace-element compositions of two basalt samples were analyzed at Activation Laboratories Ltd., Canada, using Code 4Lithoresearch Lithogeochemistry Package. The package uses lithium metaborate/tetraborate fusion with inductively coupled plasma optical emission spectrometry (FUS-ICPOES) and inductively coupled plasma mass spectroscopy (FUS-ICPMS) for the major- and trace-element analyses, respectively.

Raman spectra of the CO₂ fluid inclusion in pyroxene were obtained using a micro-Raman spectrum analysis system set up at the Hokkaido University Museum. Raman spectra were gained using diode-pumped solid-state laser (532 nm, Gem 532; Laser Quantum) excitation, a spectrometer with 75 cm focal length (Acton SP-2750; Princeton Instruments, Inc.), and CCD camera (1650 × 200 pixels, 16 µm width, iVac; Andor Technology). A 50 µm core diameter multimode fiber served as the pinhole for confocality and as the entrance slit to the spectrometer. The laser was focused through a 50 × objective (N.A. = 0.8, LUPlan; Nikon Corp.) with working distance of approximately 1.0 mm. The laser output was 50 mW at the source and approximately 8.0 mW at the sample surface. The wavenumber dispersion for each pixel of the present Raman system when using a grating of 1,800 lines/mm is 0.29 cm⁻¹/pixel at around 1334 cm⁻¹. Raman spectra were collected with two to four accumulations of 100 s.

A Raman CO₂ densimeter was applied to estimate the density of CO₂ fluid inclusions. This analysis used the density dependence of Fermi diad splits (Δ) of CO₂ fluid, and we applied the Δ–density relation of Hagiwara et al. (2020). To measure Δ with high accuracy, real-time wavenumber calibration was applied to reduce the deviation of Raman frequency (Odake et al. 2008). As described by Hagiwara et al. (2020), the measured Δ was calibrated by using the measured and known (227.46 cm⁻¹) distances between the Ne lines according to the following expression:

$${\Delta }_{\text{real}}^{\text{corrected}}= {\Delta }_{\text{mes}}^{\text{CO}_2} \times \left(\frac{227.46}{{\Delta }_{\text{mes}}^{\text{Ne}}}\right)$$

In this equation, ${\Delta }_{\text{real}}^{\text{corrected}}$ denotes the corrected splitting of the Fermi diad, ${\Delta }_{\text{mes}}^{\text{Ne}}$ represents the measured separation between the 1,449.19 and 1221.73 cm⁻¹ Ne emission lines, and ${\Delta }_{\text{mes}}^{\text{CO}2}$ stands for the measured splitting of Δ in the Raman spectrum. The measured spectra for CO₂ and Ne lines were fitted with the GRAMS AI program, respectively, using a Gaussian and Lorentzian mixture and a Lorentzian function. More than 500 counts of the Fermi diad band at 1386 cm⁻¹ are necessary to obtain a delta (Δ) value with high precision (within 1σ of 0.03 cm⁻¹) for the present system (Kawakami et al. 2003), and then, we use the data more than 500 counts. To correct the discrepancy in the Δ–density relations derived in earlier studies, we applied the correction method proposed by Hagiwara et al. (2020). In this method, a correction term δ is added to ${\Delta }_{\text{real}}^{\text{corrected}}$ to estimate the density accurately using the Δ-density relation of Hagiwara et al. (2020). To calculate the correction term δ, the two standard fluid inclusions with densities of 0.914 and 1.167 g/cm³ were used. See Hagiwara et al. (2020) for detailed methods. The conversion from obtained density to pressure of CO₂ fluid inclusion was conducted using the equation [2] and [4] from Pitzer and Sterner (1995). Note that the peaks except for CO₂ (i.e., N₂, H₂S, CO, SO₂, and CH₄) were below the detection limit. Moreover, considering the room temperature (23 ℃), laser power (8 mW), and the temperature increase in inclusions per unit laser power (1 ℃/mW for pyroxene at 532 nm) (Hagiwara et al. 2021), the inclusion temperature exceeded the supercritical temperature of CO₂, indicating that CO₂ was in a single phase during the analysis. See the supplementary information of Yamamoto et al. (2014) for the detailed description about this geobarometry.

5 Results

5.1 Basalt chemistry

The host basalts are silica-undersaturated foidite (i.e., 39.7–40.2 wt% SiO₂, and 3.9–4.4 wt% total alkalis [Na₂O + K₂O]) (Fig. 4a). Trace-element patterns of the foidites are similar to petit-spot basalts reported by Hirano et al. (2006), Machida et al. (2015), and Sato et al. (2018) (Fig. 4b) in particular Zr, Hf, and Ti negative anomalies. Detailed values are provided in Table 1.

Fig. 4

Geochemistry of the petit-spot basalts. a Total alkali silica diagram. Triangles and circles show the whole-rock and quenched glass compositions, respectively. The whole-rock composition of 6K#1466 basalts (pink triangles) is highly silica-undersaturated. The compositions previously reported at petit-spots from the NW Pacific are represented as open triangles and circles (Hirano et al. 2006; Machida et al. 2015; Sato et al. 2018). b Primitive mantle normalized trace-element compositions of basalts. The pink lines are 6K#1466 basalts in this study, and the blue lines are those reported by Hirano et al. (2019). The gray lines describe previously reported petit-spots from the NW Pacific (Hirano et al. 2006; Machida et al. 2015; Sato et al. 2018). Dotted lines describe the values of representative OIBs (Oceanic Island Basalts) and primitive mantle, from Sun and McDonough (1989). The numerical data are provided in Table 1

Table 1 Major and trace element composition of petit-spot basalts

5.2 Major-element composition of minerals

Olivines in MPSX-PR1 peridotite show forsterite content (Fo#: Mg# = 100 ×Mg²⁺/[Mg²⁺ + Fe²⁺] for olivine) of 90.5–91.0 and NiO concentrations of 0.36–0.40 wt% (Fig. 5a). Clinopyroxenes and orthopyroxenes of MPSX-PR1 display the highest Mg# of 92.7–93.4 and 90.7–91.6, respectively, within ultramafic xenoliths in this study (Fig. 5b, c). The TiO₂ and Al₂O₃ concentrations of clinopyroxene in MPSX-PR1 are the lowest, 0.05–0.16 wt% and 1.7–2.0 wt%, respectively, with a high Cr₂O₃ content of 0.94–1.06 wt% (Fig. 5c, d). The orthopyroxene Al₂O₃ contents are low, 1.4–1.7 wt%, whereas the CaO concentrations have a wide range of 0.73–2.5 wt%. Chromian spinels are found only in MPSX-PR1, showing Cr# (100 × Cr³⁺/[Cr³⁺ + Al³⁺]) of 60.1–62.7 and Mg# of 52.9–64.8. As for MPSX-PR2 peridotite, olivines represent lower Fo# (88.1–88.7) and similar NiO content (0.33–0.41 wt%) relative to those in MPSX-PR1 (Fig. 5a). Clinopyroxenes in MPSX-PR2 have lower Mg# and higher TiO₂ concentrations of 81.9–87.0 and 1.00–1.62 wt%, respectively, than those of MPSX-PR1 (Fig. 5c). Although the Al₂O₃ content of clinopyroxenes is higher than that in MPSX-PR1 (4.6–6.9 wt%), Cr₂O₃ concentrations have similar values of 0.87–1.14 wt% (Fig. 5d). The Mg#, Al₂O₃, and CaO contents of orthopyroxene are 88.2–89.2, 2.3–4.9 wt%, and 1.1–1.2 wt%, respectively.

Fig. 5

Major-element compositions of minerals in ultramafic xenoliths. The circles describe the core parts of peridotite xenoliths, and the rhombuses describe pyroxenite xenoliths. Fo# and Mg# equal 100 × Mg²⁺/[Mg²⁺ + Fe²⁺] for olivine and clinopyroxene. Mantle olivine arrays are quoted from Takahashi et al. (1987). a Major-element compositions of olivine. b Ca–Mg–Fe pyroxene classification diagram, from Morimoto (1988). Symbols are the same as in (a). c TiO₂ versus Mg# clinopyroxene classification from Wass (1979). Cpx rind (triangles) is microcrystals surrounding each xenolith, and groundmass Ti-augite (box with x) is in 6K#1466R6-001 basalt. d Clinopyroxene major-element composition, of Cr₂O₃ versus Al₂O₃. The detailed values are shown in Additional file 4: Table S2. The mineral compositions of abyssal peridotite (East Pacific Rise; Dick and Natland 1996; Regelous et al. 2016), peridotite xenoliths from NW Pacific petit-spots (site A, B; Yamamoto et al. 2014; Pilet et al. 2016), peridotite xenoliths from the Pali vent and Salt Lake Crater, Hawaii (Bizimis et al. 2007), and Tahitian harzburgite (Akizawa et al. 2017) are shown here

The geochemical characteristics of pyroxenite xenoliths and pyroxene xenocrysts (MPSX-PX1 to MPSX-PX6 and isolated xenocrysts) are similar to each other. Clinopyroxenes have lower Mg# and higher TiO₂ concentrations of 72.6–79.8 and 0.59–1.61 wt%, respectively, compared to the peridotite xenoliths in this study (Fig. 5c). The Al₂O₃ content of clinopyroxene shows a wide range, of 3.7–7.5 wt%, resembling MPSX-PR2, whereas Cr₂O₃ contents of 0–0.27 wt% are generally lower than those of the two peridotites (Fig. 5d). The Mg#, Al₂O₃, and CaO contents of orthopyroxene are 76.3–80.3, 2.5–5.1 wt%, and 1.4–2.1 wt%, respectively. Orthopyroxene lamellae and indeterminate-shaped orthopyroxenes in the clinopyroxene host phase of MPSX-PX4 and MPSX-PX5 all have similar compositions. The average compositions of isolated clinopyroxene xenocrysts were 72.7 Mg#, 1.46 wt% TiO₂, 7.2 wt% Al₂O₃, and 0.06 wt% Cr₂O₃.

In summary, pyroxenite xenoliths and pyroxene xenocrysts in this study have more fertile compositions than MPSX-PR1 peridotite, and some elements resemble the clinopyroxene concentrations in MPSX-PR2 peridotite. Detailed values are given in Additional file 4: Tables S1 to S4.

We focused on Ca and Al variation in the line profile and the elemental mapping for clinopyroxene (Fig. 6). The clinopyroxene rim in MPSX-PR1 shows slightly higher Ca and lower Al than the homogeneous inner part (Fig. 6a). From MPSX-PR2 Cpx#1 to MPSX-PX6, clinopyroxenes tend to show an increase for Ca and a decrease for Al in spongy textures, while both Ca and Al tend to increase at the outermost rims (where they are in contact with host groundmass), but these areas are heterogeneous (Fig. 6b–i). The clinopyroxene grain of MPSX-PX2 is surrounded by orthopyroxene and does not represent the reaction texture, but gradual zonings for Al are identified (Fig. 6e). The reaction rim including the outermost rim and the spongy texture of MPSX-PX5 is relatively thin (< 90 µm), but Ca becomes more pronounced at the rim, while Al is variable (Fig. 6h). Simultaneously, although MPSX-PX5 shows compositional heterogeneity according to the distribution of orthopyroxene lamellae, the core has a different composition from the reacted portions (Fig. 6h).

Fig. 6

Elemental mappings of xenolith and line profiles of clinopyroxene for Ca and Al. The red lines on the backscattered electron image and white lines on the elemental map are the analyzed lines. Triangle symbols indicate the starting points. In the a.p.f.u (atoms per formula unit) vs. distance diagram, the red and blue diamonds display Ca and Al cation numbers, respectively. The cation numbers for Al are multiplied by a constant for ease of viewing. a MPSX-PR1. b, c MPSX-PR2. d MPSX-PX1. e MPSX-PX2. f MPSX-PX3. g MPSX-PX4. h MPSX-PX5. i MPSX-PX6

5.3 Trace-element composition of clinopyroxenes

Primitive mantle normalized trace-element patterns for the clinopyroxene core in this study are shown in Fig. 7a. The previously reported data of abyssal peridotites and xenoliths in Pacific hotspots and NW Pacific petit-spot lavas are also available in Fig. 7a, b. The MPSX-PR1 peridotite generally resembles abyssal peridotite from the East Pacific Rise (EPR) Hess Deep, with a low LREE/heavy-rare-earth element (HREE) ratio (Dick and Natland 1996; Regelous et al. 2016; Warren 2016). In contrast, clinopyroxene patterns in the MPSX-PR2 peridotite, pyroxenites, and pyroxene xenocrysts (MPSX-PX1, MPSX-PX3, MPSX-PX4, and isolated clinopyroxenes) show higher LREE/HREE ratios than that of MPSX-PR1 (i.e., high enrichment of LREE). The absolute concentrations of incompatible elements between the fertile series (MPSX-PR2 peridotite and pyroxenite xenoliths) and depleted peridotite (MPSX-PR1) are clearly different as well. The MPSX-PR2 peridotite notably shows negative anomalies in Zr, Hf, and Ti, which are not shown in the pyroxenite series and pyroxene xenocrysts. The concentrations of U, Th, Nb, and Ta are higher in the MPSX-PR2 clinopyroxenes than in the pyroxenite xenoliths and pyroxene xenocrysts. Detailed values are provided in Additional file 4: Tables S5 and S6.

Fig. 7

Trace-element compositions of clinopyroxene. a A compilation of this study’s data. The depleted peridotite (MPSX-PR1) is shown as green lines. Red lines display the enriched peridotite (MPSX-PR2). Pyroxenite data (blue lines) include all pyroxenite xenoliths, xenocrysts (MPSX-PX1 to PX6), and isolated clinopyroxene xenocrysts. Gray solid lines show metasomatized peridotite xenoliths from the NW Pacific petit-spot reported by Pilet et al. (2016). Gray dotted lines represent the abyssal peridotite from East Pacific Rise (Dick and Natland 1996; Regelous et al. 2016). b Compilation of the other peridotite xenoliths from Pacific. The trace-element compositions of the Hawaiian peridotite xenoliths (Pali vent; black lines, and Salt Lake Crater; azure lines) are from Bizimis et al. (2007). REE concentrations of Hawaiian peridotite xenoliths (Ka'au vent; green diamonds, Pali vent; white diamonds) are also compiled here (Sen et al. 1993). Tahitian metasomatized harzburgite xenolith is from Akizawa et al. (2017). The data for the primitive mantle are from Sun and McDonough (1989). The detailed values are shown in Additional file 4: Table S5

5.4 Geothermobarometry

Equilibrium temperatures were obtained using the two-pyroxene thermometer formulated by Brey and Köhler (1990) for samples including both orthopyroxene and clinopyroxene. The equilibrium temperatures at core part of two-pyroxenes are 977 ± 62 °C for MPSX-PR1 peridotite, 1069 ± 85 °C for MPSX-PR2 peridotite. Most pyroxenites show similar temperatures of 1076 ± 41 °C for MPSX-PX1, 1053 ± 16 °C for MPSX-PX2, 1073 ± 25 °C for MPSX-PX3, and 1093 ± 17 °C for MPSX-PX4, but MPSX-PX5 shows higher temperature of 1157 ± 50 °C. We use these values to calculate the residual pressures of CO₂ inclusion for two peridotites and two pyroxenites (MPSX-PX1 and PX5). The K_D values [= (X^cpx_Fe/X^cpx_Mg)/(X^opx_Fe/X^opx_Mg)] are within the equilibrium range of 1.09 ± 0.14 proposed by Putirka (2008), but MPSX-PR1 has a small offset (K_D = 0.762).

Residual pressures were estimated by adopting the densimeter of CO₂ fluid inclusions in the pyroxenes. The pressures calculated from the density of CO₂ fluid inclusions with two-pyroxene temperatures are 0.410 ± 0.006 GPa for MPSX-PR1 peridotite, 0.902 ± 0.005 GPa for MPSX-PR2 peridotite, 0.620 ± 0.008 GPa for MPSX-PX1 pyroxenite, and 1.197 ± 0.002 GPa for MPSX-PX5 pyroxenite. The pressures obtained were converted to subseafloor depths; the resulting temperature–depth and pressure diagram is shown as Fig. 8a. A detailed account of the validation of this geobarometry for the mantle xenolith is given later (the first paragraph in Sect. 6.3). Imagery of a fluid inclusion and Raman shift is provided in Additional file 2: Fig. S2. Detailed parameters and all obtained values are given in Table 2 and S7. The equilibrium temperatures and pressures for ultramafic xenoliths were also calculated using the two-pyroxene geothermobarometer of Putirka (2008). The calculated temperatures and pressures are 1011 ± 56 °C and 0.37 ± 0.08 GPa for MPSX-PR1 peridotite, and 1100 ± 21 °C and 0.76 ± 0.16 GPa for MPSX-PR2 peridotite. The range of temperature and pressure for five pyroxenite xenoliths is 1078–1109 °C and 0.82–1.07 GPa. The depth–temperature diagram for these values is also shown as Fig. 8b. We use the average core composition of pyroxenes. In particular, the temperature of MPSX-PX5 pyroxenite was applied to the average composition of the most homogeneous part of the clinopyroxene and orthopyroxene blebs (Fig. 6h). Detailed values are given in Table 2. The method of error calculation for this thermobarometry is described in Additional file 3.

Fig. 8

Depth–temperature diagram from geothermobarometry of ultramafic xenoliths. The detailed data are provided in Table 2. Each age-equivalent geotherm is drawn using the plate model of GDH1 (Stein and Stein 1992). a Depth and temperature of xenoliths acquired by the density of CO₂ inclusions and two-pyroxene thermometer (Brey and Köhler 1990). b Depth and temperature calculated using the two-pyroxene geothermobarometer of Putirka (2008). The estimated P–T condition at which ultramafic xenoliths are hotter than the 160 Ma-geotherm of GDH1 model. Symbols are same as for Fig. 5. Error ranges for depth are within the sizes of the symbols. The pressures were converted to depths assuming the density of the seawater, oceanic crust, and mantle to be 1.023 g/cm³, 2.86 g/cm³, and 3.30 g/cm³, respectively (Carlson et al. 1988; Nayar et al. 2016; Weller et al. 2019), and thickness of the crust to be 6 km based on the seismic investigation near this study area (Kaneda et al. 2010). The depth of seafloor (i.e., thickness of seawater) was assumed to be 5600 m from each seafloor survey of research cruises in this study area

Table 2 Geothermobarometry of ultramafic xenoliths

6 Discussion

6.1 The origins of xenolith

Ultramafic xenoliths in intraplate volcanoes can be divided into Cr-diopside series or Al-augite series, depending on the clinopyroxene composition (Wilshire and Shervais 1975; Wass 1979). The former is interpreted as a residue after crust production at the spreading center, including Cr-rich spinels, Mg-rich orthopyroxenes, and Mg-rich olivines. Conversely, the latter has three conceivable origins as metasomatic vein formed by melt–peridotite reaction, phenocryst crystallized at high pressure, or a cumulate from alkali basaltic melt, composed of Al-rich spinels, Fe-rich orthopyroxenes, and Fe-rich olivines. Cr-diopside series xenoliths generally show the depletion of incompatible elements, but sometimes have enriched composition modally or cryptically (e.g., a high LREE/HREE ratio; Dawson 1984). In addition, ultramafic xenoliths often represent higher equilibrium temperatures than those expected by the underlying pristine oceanic lithosphere (Hauri and Hart 1994; Qi et al. 1994). Such features often suggest local lithospheric modifications resulted from intraplate magma infiltrations.

MPSX-PR1 and MPSX-PR2 peridotites have high Cr₂O₃ contents in clinopyroxenes (Fig. 5d). However, other chemical features of these two peridotites are significantly different. In the pyroxene quadrilateral diagram (Fig. 5b), the clinopyroxenes in MPSX-PR1 peridotite are classified as diopside and those in MPSX-PR2 peridotite are classified as augite. The fayalite component of olivine and the TiO₂, Al₂O₃, and iron contents of clinopyroxene are higher in MPSX-PR2 peridotite than those in MPSX-PR1 peridotite (Fig. 5b–d). The Mg# and Al₂O₃ contents of the orthopyroxenes also differ: The higher Al₂O₃ and lower Mg# of MPSX-PR2 as compared with MPSX-PR1 indicate that MPSX-PR2 peridotite is more fertile than MPSX-PR1. The mineral compositions of MPSX-PR1 peridotite and the other Pacific oceanic peridotites are similar to each other with the exception of the Al₂O₃ and Cr₂O₃ contents of clinopyroxene (Fig. 5). The variation of clinopyroxene Al₂O₃ contents in the other Pacific peridotite xenoliths could be explained by the degree of fertilization (Bizimis et al. 2007; Yamamoto et al. 2014; Pilet et al. 2016; Akizawa et al. 2017). Based only on the above, we can estimate that MPSX-PR1 peridotite originates from residual mantle with highly depleted Cr# of chromian spinels (60.1–62.7; Additional file 4: Table S4) (cf., < 60.6 Cr# in Hess Deep peridotites: Arai and Matsukage, 1996; Dick and Natland, 1996). By contrast, MPSX-PR2 peridotite is derived from fertilized mantle and its protolith might be Cr-diopside series depleted peridotite.

The primitive mantle normalized trace-element patterns of the clinopyroxenes in MPSX-PR1 peridotite shows low LREE/HREE ratios, overlapping the composition of residual harzburgites from the Hess Deep, EPR (Dick and Natland 1996; Regelous et al. 2016; Warren 2016) (Fig. 7a). Even some LREE-depleted peridotites reported from the Ka'au vent and the Pali vent in Hawaii (Sen et al. 1993) are not as depleted in LREE as those of the EPR (Fig. 7b). In contrast, clinopyroxenes in MPSX-PR2 peridotite are marked by relative enrichment in LREE, resembling those of metasomatized peridotites reported from the NW Pacific petit-spot (Pilet et al. 2016) (Fig. 7a). Additionally, all xenoliths in this study are suggested to originate from spinel-stability field, based on their trace-element patterns (i.e., the absence of HREE depletion; Fig. 7a) and thermobarometry (Fig. 8). This finding is conflict with NW Pacific mantle xenoliths where the trace-element patterns indicate both garnet- and spinel-stability field origins (Pilet et al. 2016) (Fig. 7a). Moreover, enriched MPSX-PR2 peridotite shows the negative anomalies of Zr, Hf, Ti, Sr, and Nb as well as the metasomatized Tahitian harzburgite xenolith and some enriched Hawaiian peridotite xenoliths (Fig. 7b; Bizimis et al. 2005, 2007; Akizawa et al. 2017). This also suggests that the enriched peridotite (MPSX-PR2) is of metasomatic origin. As discussed below, we expect that pyroxenites, which show a similar trace-element composition to those of enriched peridotite (MPSX-PR2), were also formed by magma–lithospheric mantle interaction.

In summary, we suggest that MPSX-PR1 peridotite corresponds to a depleted harzburgite left after melt extraction, whereas MPSX-PR2 peridotite is a metasomatized mantle fragment by intraplate magmatism. We propose this metasomatic agent could be previous petit-spot magmatisms as discussed later.

Pyroxenite xenoliths (MPSX-PX1 to MPSX-PX5), the clinopyroxene xenocryst (MPSX-PX6), and isolated (handpicked) clinopyroxene xenocrysts are similar composition and all categorized as Al-augite series, considering that clinopyroxene Cr₂O₃ contents are low (0–0.28 wt%) and Al₂O₃ contents are high (Fig. 5b, d) and orthopyroxenes are rich in Fe and Al (Fig. 5b). The chemical characteristics are obviously different from the two peridotites, as shown in Fig. 5a–d. The cores of the MPSX-PX6 augite xenocryst and other isolated clinopyroxene xenocrysts are similar in compositions to clinopyroxenes in pyroxenite xenoliths (Figs. 5b–d, 7), implying their same origin. Among the three conceivable origins of Al-augite series pyroxenite xenoliths as described above, this study samples plausibly originate from metasomatic veins to explain their petrographic features as follows (Wass 1979) (Fig. 5c): (1) The existence of bleb-like orthopyroxene lamellae in the clinopyroxene grains (Fig. 3f, g) implies a growth of exsolved orthopyroxene for a relatively long period under an isothermal condition and cooling (e.g., Toh and Shimobayashi 2000; Liang et al. 2018). (2) The spongy texture in rims or grain boundaries of clinopyroxene (Fig. 3) suggests disequilibrium with the host magma (e.g., Shaw et al. 2006; Su et al. 2011). (3) Micro-Ti-augite grains around xenoliths (Fig. 3), having the similar composition as those in the groundmass, are crystallized after entrapment in low-pressure conditions (Fig. 5c). The compositional modification of the lithospheric mantle due to melt–rock interactions is indicated by Al-augite series xenoliths as seen in various oceanic islands/seamounts (e.g., Tracy 1980; Fodor et al. 1982; Qi et al. 1994). Moreover, fertile pyroxene-rich veins formed by interactions between alkali basaltic melt and lithospheric mantle have been proposed (e.g., Wilshire and Shervais 1975). Given the absence of olivine in our pyroxenites, it is expected that the pyroxenites were formed by the process of olivine dissolution and pyroxene formation (i.e., $Liq1+Ol\to Cpx+Opx+Liq2$). This pyroxenite-formation process is proposed to have taken place in oceanic upper mantle (Laukert et al. 2014). The diagram of Mg# vs. Ca/Al and Ti/Eu vs. (La/Yb)_N for clinopyroxenes suggests strongly that the metasomatized mantle xenoliths in this study were metasomatized by silicate magma rather than carbonatitic magma (Fig. 9a, b). Given that it has been proposed that silicate metasomatism transforms peridotite to websterite (pyroxenite) and orthopyroxenite by consuming olivine, our olivine-free pyroxenites are expected to have been formed by silicate melt–mantle rock interaction instead of carbonatite metasomatism with orthopyroxene consumption and formation of clinopyroxene-rich lherzolite or wehrlite (e.g., Laurora et al. 2001). In addition, such fertile pyroxenite xenoliths often show higher equilibrium temperatures than might be expected from pristine geotherm on the basis of plate age in correlation with LREE concentrations as found in our samples (e.g., Hauri and Hart 1994; Burnard et al. 1998).

Fig. 9

a Diagram of Mg# versus Ca/Al. b Diagram of Ti/Eu versus (La/Yb)_N. (La/Yb)_N is the chondrite normalized value. The data for the chondrites are from Sun and McDonough (1989). The ranges of silicate metasomatism and carbonatite metasomatism are from Coltorti et al (1999) and Zong and Liu (2018). Symbols are as for Fig. 5

All of these lines of evidence indicate that the pyroxenite xenoliths in this study originated from metasomatic veins in the lithospheric mantle.

6.2 Identification of the metasomatic agent

In our study area, there are several candidates for the triggering of the metasomatic event: (1) off-axis magmatism near the mid-ocean ridge (e.g., Seyler et al. 2001; Rochat et al. 2017), (2) Cretaceous intraplate magmatism (100–120 Ma, Koppers et al. 2003; Aftabuzzaman et al. 2021), (3) irregular volcanism on the Cretaceous edifice of Minamitorishima Island at approximately 40 Ma (Hirano et al. 2021), (4) petit-spot magmatism before the xenoliths-host magma event, and (5) host magma. Among these, previous (prior to xenolith entrapment) petit-spot magmatism most reasonably explains our results because of the high equilibrium temperature of the metasomatized xenoliths (~ 1100–1200 °C).

According to the study of numerical simulation about relationships between heating/cooling rates and elemental diffusion in two-pyroxene of mantle xenolith, it takes 5 to 6 million years for a clinopyroxene of ~ 700 °C (the condition when the elemental diffusion of clinopyroxene is halted) to reach 1300 °C (near the estimated potential temperature of the petit-spot; Machida et al. 2017) in a heating process at the rate of 10⁻⁴ °C/yr with the creation of gradual compositional zoning or without zoning (Yamamoto et al. 2017). Moreover, a clinopyroxene of 1300 °C takes only 1–2 million years to reach 1100 °C without creating notable compositional zoning even assuming a slow cooling rate of 10⁻⁴ °C/year (Yamamoto et al. 2017). As for the clinopyroxenes in the metasomatic pyroxenites, showing an equilibrium temperature of ~ 1100 °C without compositional zoning or with gentle zoning (Fig. 6b–i), these are expected to have experienced a similar heating/cooling processes to that described above. Additionally, if the pyroxenites in this study were formed in such a process, it would be required a total heating/cooling time of 6–8 million years just prior to entrapment. Given the relation of the cooling period to the timing of the host magmatism (0.2–5.2 Ma), volcanism at ~ 40 Ma, Cretaceous hotspots, and off-axis magma infiltration are unsuitable as a metasomatic agent. The ascending time of general petit-spot melts (a few hundred days; Sato et al. 2018) also rules out the host magma as a metasomatic agent because our fertile xenoliths (MPSX-PR2 and MPSX-PX1–6) have no steep elemental zoning. In addition, they show a remarkable compositional difference between the core and the spongy-textured part, including grain boundaries (where reacted with host magma; Fig. 6b–i) that support the occurrence of pre-entrapment fertilization. The comparatively recent magmatism at the lithosphere–asthenosphere boundary, generated by small amounts of volatiles and/or plate flexure (i.e., petit-spot), is the most probable metasomatic agent. As such, we argue that the metasomatic agent was petit-spot magma continuously percolated in the lithosphere prior to the host magma eruption. This inference is supported by the fact that the trace-element composition of the clinopyroxene in fertile xenoliths shows the same specific high field strength elements (i.e., HFSEs) anomaly as the petit-spot magma (Figs. 4b, 7a). Additionally, the MPSX-PR2 peridotite shows more pronounced anomalies of HFSEs than pyroxenites without notable Hf and Ti anomalies (Fig. 7a). This compositional relationship is similar to those between pyroxenite and enriched peridotite from Salt Lake Crater in Hawaii (Bizimis et al. 2003, 2005).

Pilet et al. (2016) suggested that the trace-element compositions of the clinopyroxene in peridotite xenoliths brought by the NW Pacific petit-spot volcano were exemplified by the metasomatic enrichment process of the petit-spot melt–depleted peridotite interaction. Those researchers also conducted a forward modeling exercise, which showed that the focused and porous melt from the lithosphere–asthenosphere boundary could form metasomatic veins in lithospheric mantle. The pyroxenites in the present study are good candidates for identification as metasomatic veins of the type just described. Yamamoto et al. (2009) provide the noble gas isotopic compositions of petit-spot peridotite xenoliths in NW Pacific petit-spots. The low ⁴He/⁴⁰Ar* (an asterisk indicates correction for atmospheric contribution) of peridotite xenoliths in the NW Pacific petit-spots indicates the injection of petit-spot magma into the lithospheric mantle prior to entrapment by the host magmatism (Yamamoto et al. 2009). Those researchers concluded that petit-spot volcanism should be recognized as a polygenetic magmatism with a long active period. Yamamoto et al. (2014) provided the residual pressure and two-pyroxene temperature recorded in peridotite xenoliths in NW Pacific petit-spot, and the obtained geothermal gradient is hotter than that of the GDH1 plate model and the half-space cooling model. Such a hot geotherm is interpreted as a localized feature restricted to the petit-spot region because widespread thermal anomalies are impossible to explain on the basis of present water depths and their associated isostatic adjustments (Turcotte and Schubert 2002; Yamamoto et al. 2014). In the western Pacific petit-spot region, the same phenomenon was observed.

6.3 The profile of Pacific lithosphere aged at 160 Ma

Combined with the barometric data by CO₂ fluid inclusions in pyroxene using Raman spectroscopy, we obtained a lithological structure of the oldest portion of the Pacific lithosphere. Although the two-pyroxene geobarometer of Mercier et al. (1984) and the geobarometer based on the pressure dependence of Ca partition between olivine and clinopyroxene (Köhler and Brey 1990) have been used so far, limitations of accuracy and precision mean they should be used with caution (Laurora et al. 2001; Yamamoto et al. 2014). For that reason, we used a form of geobarometry based on the density of CO₂ fluid inclusions, first proposed by Roedder (1965) and improved and updated by subsequent studies (Miller and Richter 1982; Roedder 1983; Rosso and Bodnar 1995; Seitz et al. 1996; Andersen and Neumann 2001; Yamamoto et al. 2002, 2007; Hagiwara et al. 2020, 2021). Since fluid inclusions in a mineral constituting a mantle xenolith should have existed in the mantle before being entrapped by magma, they can be used as a geobarometric probe for the host xenolith if the internal pressure of the fluid inclusions was in equilibrium with surrounding mantle. Even if the internal pressure of the fluid inclusion temporarily differed from that of its surroundings for some reason, this difference would be erased in a geochronologically short time (i.e., several days to years) in view of the physical properties (e.g., strain rate) of olivine and pyroxene (Yamamoto et al. 2008). The internal pressure of fluid inclusions in a xenolith may also respond to external pressure reduction during the transportation by magma. Under those conditions, olivine (having a low plastic deformation strength) may permit fluid inclusions within it to expand (e.g., Yamamoto et al. 2008). On the other hand, considering the plastic deformation strength of pyroxene or spinel and the eruption timescale of petit-spot magma (a few hundred days), the expansion of fluid inclusions during ascent and cooling may have a negligible in the original depth estimation (Sato et al. 2018; Yamamoto et al. 2011). All the same, the density (pressure) retained in CO₂ inclusions in natural samples does display variation and may have been affected by creep, decrepitation, or bursting during transport and cooling. In the face of such uncertainties, the measured values of inclusion density or pressure should be treated as limit indicators of minimum depth of formation, in the sense that the pressure and density of such an inclusion may diminish on ascent but should never increase.

The depth calculated by CO₂ inclusion of the MPSX-PR1 depleted harzburgite (>11.9 km) is the shallowest among the ultramafic xenoliths in this study, whereas that of the enriched peridotite (MPSX-PR2) is deeper, at 27.4 km (Fig. 8). Two pyroxenites (MPSX-PX1 and PX5) have equilibrium temperatures higher than depleted harzburgite (MPSX-PR1), representing residual depths of 18.5 km and 36.6 km, respectively, deeper than the depleted harzburgite. Additionally, upon applying the geothermobarometer of Putirka (2008) to estimate the origin of xenoliths, an increasing order of depth (from shallowest to deepest) was obtained as follows: the depleted peridotite (MPSX-PR1; 10.7 km), fertile peridotite (MPSX-PR2; 22.0 km), and pyroxenites with MPSX-PX5 (32.6 km); this order was consistent with that obtained from CO₂ inclusions (Fig. 8b). However, the MPSX-PX1 pyroxenite displayed a deeper value using Putirka’s method (26.1 km) than that obtained for MPSX-PR2 peridotite, which differed from the depth-order obtained for these two rocks by the analysis of their CO₂ inclusions. It follows that the depth of MPSX-PX1 estimated using CO₂ inclusion should indeed be considered as a minimum depth only. Overall, these findings suggests that the origin of metasomatic pyroxene-rich veins and fertilized peridotite lies in a deeper part of the lithosphere than the depleted residual mantle (Fig. 8). Most of the petit-spot magma is expected to have stagnated at the stress-rotation field in the lithosphere, where the crystal fractionation and the melt-rock interaction occur, as was made evident by the multiple-saturation experiment (Machida et al. 2017; Hirano and Machida 2022). The fertile components in this study, the pyroxenites in particular, could be derived from the mid-depth of the lithosphere assuming the about 80 km-thick lithosphere (i.e., ~40 km in depth; Kawakatsu et al. 2009) where petit-spot magma previously stagnated (Fig. 10). These pyroxenites, originating from metasomatic veins, consist with the so-called green core clinopyroxene xenocrysts as seen in the relicts of metasomatic veins in the Costa Rican petit-spot lavas (Buchs et al. 2013; Pilet et al. 2016). 

Fig. 10

Schematic model of the Pacific lithosphere at 160 Ma. The plate thickness of the study area (~ 80 km) is quoted from Kawakatsu et al. (2009). The formation ages of each volcanic edifice are quoted from Koppers et al. (2003) (Cretaceous seamounts), Hirano et al. (2019) (Petit-spot knoll 6K#1203, 1206), Aftabuzzaman et al. (2021) (Minamitorishima Island), and Hirano et al. (2021) (Rejuvenated volcanoes of Minamitorishima Island)

As proposed by Yamamoto et al. (2014), petit-spot volcanoes erupted on the ocean floor should be considered to be just ‘the tip of the iceberg,’ and petit-spot melts could account for metasomatic veining in the lithosphere. As described above (Sect. 6.2), if the large thermal modification enough to change the overall geothermal structure of oceanic plate occurs, the water depth will become shallower (oceanic plate will float up), and then, this thermal modification would be a very local phenomenon. However, since it takes hundreds of thousands to millions of years to convert the geotherm, it would be difficult to find a magma reservoir sufficiently large to supply magma for such a long time. The partial melts on the lithosphere–asthenosphere boundary (i.e., petit-spot magma) are therefore suggested to be the only underlying cause that can explain all phenomenon reported in this study. The oldest portion of the Pacific lithosphere has been modified, thermally and compositionally, by multiple petit-spot magmatisms prior to entrapment. In contrast, the depleted harzburgite sample, analogous to the pristine oceanic lithosphere, is a fundamental discovery that can yield insights into the chemistry of the deep oceanic lithosphere in its unaltered state. Although our data only allow us to see the local phenomena, such localized metasomatism in lithosphere could occur below the seafloor without hotspots, and we have to reappraise the nature of subducting plate.

7 Conclusions

A petit-spot knoll was newly discovered on the abyssal plain south of the Minamitorishima (Marcus) Island. The silica-undersaturated petit-spot basalts erupted on the concave surface of the outer-rise swell because of subduction into the Mariana Trench. Ultramafic xenoliths of the oldest Pacific lithosphere could be divided into residual peridotite, fertile peridotite, and metasomatism-origin pyroxenites. Geochemistry and geothermobarometry of ultramafic xenoliths illustrated that metasomatic pyroxenite veins, formed via petit-spot magma–lithospheric peridotite interaction during the petit-spot magma stagnation, distribute in the oldest Pacific lithosphere. Moreover, we proved that depleted residual harzburgite practically exists in the uppermost portion of lithospheric mantle. The geotherm of the underlying lithosphere acquired from the equilibrium temperatures of xenoliths is much hotter than crustal age-appropriate temperature of oceanic lithosphere, which is probably caused by the relatively recent petit-spot magmatism. Although the following observation could not be verified from our samples, an identical phenomenon may occur not only in known petit-spot regions, but also in the plate-bending subduction fields in general, and/or where small amounts of volatiles are present.