Peridotites with back-arc basin affinity exposed at the southwestern tip of the Mariana forearc

Peridotites at water depths of 3430 to 5999 m have been discovered using the submersible Shinkai6500 (dives 6K-1397 and 6K-1398) on the southwestern slope of the 139°E Ridge (11°12′N, 139°15′E), a small ridge at the southwesternmost tip of the Mariana forearc near the junction with the Yap Trench and Parece Vela Basin. The peridotites studied consist of 17 residual harzburgites and one dunite and show various textures with respect to their depths. Peridotites with coarse-grained (> 1 mm) textures were sampled from the shallowest part (3705–4042 m) of the dive area, and peridotites with fine-grained (< 0.5 mm) textures were sampled deeper (5996 m). Olivine crystal-fabrics vary with grain size, with (010)[100] A-type patterns for the coarse-grained peridotites, {0kl}[100] D-type patterns for the fine-grained peridotites, and various indistinct patterns in samples of variable grain sizes. Fine-grained peridotites with D-type olivine crystal-fabrics could result from deformation under relatively higher flow stresses, suggesting that a ductile shear zone in the lithospheric mantle could occur in the deepest part of 139°E Ridge. Spinel Cr# range from relatively low (0.36) to moderately high (up to 0.57), and correlate with Ti contents (0.07–0.45 wt.%). The trace element patterns of clinopyroxene similarly exhibit steepening slopes from the middle to the light REEs regardless of textural variations. These mineralogical and geochemical features would result from melt-rock interactions under conditions of relatively shallow lithospheric mantle, which are much more comparable with the Parece Vela Basin peridotites than the Mariana forearc peridotites. Consequently, the Parece Vela Basin mantle is more likely exposed on the inner slope of the westernmost Mariana Trench, presumably due to the collision of the Caroline Ridge.


Introduction
The southern Mariana forearc is a non-accretionary convergent plate margin, where the rock suites cropped out on the landside slope are similar to those found in many ophiolites (Bloomer and Hawkins 1983;Reagan et al. 2013;Stern et al. 2020). However, mantle peridotites in the southern Mariana forearc are only partly understood (Ohara and Ishii 1998;Michibayashi et al. 2009;Sato and Ishii 2011;Michibayashi et al. 2016a;Reagan et al. 2018). Previous geological expeditions to the eastern side of the Challenger Deep have found trench peridotites at depths as shallow as 4500 m below sea level (mbsl) (Stern et al. 2020). These peridotites could have been derived from either forearc (Ohara and Ishii 1998;Michibayashi et al. 2016a) or backarc (Michibayashi et al. 2009;Sato and Ishii 2011) mantle, suggesting that their geological and petrological characteristics may be complicated in relation to the structural evolution of the southern Mariana Trench. In contrast, a few dredge expeditions have been conducted

Progress in Earth and
Planetary Science to the western side of the Challenger Deep (Hawkins and Batiza 1977;Beccaluva et al. 1980; Fig. 1). Rocks dredged from a small ridge running parallel to the Mariana trench axis on Scripps Institution expedition INDOPAC, Leg 4, in 1976 included a serpentinized peridotite that is highly sheared, so that the "ridge" was thought to be a tectonic sliver of sheared ultramafic rock (Hawkins and Batiza 1977). However, previous studies were limited in scope, and more detailed and extensive geological studies of the mantle peridotites are necessary to understand the tectonic history of the southern Mariana Trench.
In this paper, we present some petrophysical features of newly sampled peridotites from the small ridge on the western side of the Challenger Deep, where Hawkins and Batiza (1977) have found serpentinized peridotites. We tentatively named the small ridge '139°E Ridge' . We show details of their crystallographic fabrics along with other petrological and geochemical data and use these to explain briefly how they have been formed.

Geological setting
The Mariana Trench is where the Pacific plate subducts beneath the Philippine Sea plate and its strike changes from N-S in the north to E-W in the south. The Challenger Deep, part of the southern Mariana Trench southwest of Guam, is the deepest trench in the world (Fujioka Deep-tow camera dive tracks  are shown by red lines. D-1 is the dredge line reported by Hawkins and Batiza (1977). 1438D1 is the dredge line reported by Beccaluva et al. (1980). c Lithologies recovered along the dive tracks plotted on a bathymetric map. The bathymetric dataset used in each panel is a compilation from the following bathymetric data: ETOPO1, U.S. Extended Continental Shelf Cruise (Armstrong 2010), and Yokosuka ship-board bathymetry et al. 2002;Fryer et al. 2003). The Mariana forearc narrows markedly along the E-W in the south (Fig. 1a), so that the Mariana trench axis has an arcuate shape due to the collision of the Ogasawara Plateau in the north and the Caroline Ridge in the south (Miller et al. 2006). The southern Mariana Trench from south of Guam to the Yap Trench junction shows a characteristic morphology where the trench axis runs across both the Mariana volcanic arc and a backarc basin ( Fig. 1a; e.g., Sleeper et al. 2021). The Mariana and Yap Trenches intersect at a nearly perpendicular angle, forming a typical trenchtrench junction (Ohara et al. 2002a;Chen et al. 2019).
The Mariana Arc consists of an active and two fossil volcanic arcs separated by the Parece Vela Basin (29.0-7.9 Ma) and the Mariana Trough (6 Ma to present), both of which are backarc basins (Fig. 1a;Okino et al. 1998;Ohara et al. 2002a, b;Ohara et al. 2003;Tani et al. 2011). The Southeast Mariana forearc rift (SEMFR) is an unusual volcanic rift in the Mariana forearc, which extends from the trench to the southernmost Mariana volcanic arc (Ribeiro et al. 2013a(Ribeiro et al. , 2013b. West Santa Rosa Bank Fault (WSRBF) separates older rocks of the Santa Rosa Bank from the SEMFR younger rocks.

The 139°E ridge
The study site is located on the southwestern slope of the 139°E Ridge (11°12′N, 139°15′E), which is a small ridge at the southwesternmost tip of the Mariana Trench inner slope at the junction with the Yap Trench (~ 11°20′N, ~ 139°20′E; Fig. 1a, b). Detailed geological surveys of the study area were conducted using the submersible Shinkai6500 (dives 6K-1397 and 6K-1398) and a deep-tow camera (dives YKDT-170 and YKDT-171) as part of R/V Yokosuka cruises in 2014 (Ohara et al. 2015). Each dive site is shown in Fig. 1b, c.
The southwestern slope of the 139°E Ridge differs from the northeastern slope in both bathymetry and geology. The southwestern slope is rugged, whereas the northeastern slope is smooth with small knolls. The northeastern slope is contiguous with the Parece Vela Basin. It is noted that Hawkins and Batiza (1977) dredged serpentinized peridotites from the southwestern slope at depths of 2350-1566 mbsl (D-1 in Fig. 1b).
Several outcrops of plutonic rock were observed on the deeper part of the southwestern slope (Fig. 2a). During our dive surveys, 48 samples (35 peridotites, 2 olivine gabbros, 3 clinopyroxenites, 5 basalts, and 3 carbonates) were recovered at depths of 5999-3430 mbsl (Fig. 1c). Peridotites with varying degrees of serpentinization were found at all sampling sites. Moreover, the YKDT-170 dive revealed that the shallow western part of the ridge (from 1985 to 1740 mbsl) consists of volcanic breccias and carbonate rocks (Fig. 2b). During the YKDT-171 dive, Mncoated old volcanic rubble was observed covering a small knoll on the northern flank of the ridge (Fig. 2c).

Petrography
18 peridotite samples have been chosen for our study: 12 from dive 6K-1397 and 6 from dive 6K-1398. In the laboratory, using bleached and saw-cut samples, the foliation and lineation were identified on the basis of the alignment of spinel and pyroxene grains (Fig. 2d, e). Four harzburgites (samples 6K-1397-R03, 6 K-1397-R04, 6K-1397-R08, and 6K-1398-R14) show clear centimeterscale pyroxene-rich or plagioclase-rich layering parallel to the harzburgite foliation (Fig. 2d, e). We have made thin sections cut perpendicular to the foliation and parallel to the lineation (i.e., XZ sections), except for one dunite sample (6K-1398-R11) for which we did not identify any foliation or lineation. All thin sections were polished using 1 μm diamond paste and colloidal silica for > 5 h for microstructural observations and analyses.
Mineral abundances were determined by microscope point-counting. About 2000 points per thin sections (28 × 48 mm) were measured at 0.2 mm intervals. The minerals identified include both primary minerals such as olivine, orthopyroxene, clinopyroxene, spinel, and plagioclase, and secondary minerals such as serpentine, bastite, and magnetite.

Crystal-preferred orientations (CPOs)
The crystal-preferred orientations (CPOs) of the olivine and orthopyroxene for harzburgite samples were measured on polished thin sections using a scanning electron microscope equipped with an electron back scatter diffraction (EBSD) system (HITACHI S-3400N with HKL Channel 5) at Shizuoka University (now at Nagoya University). EBSD patterns were produced by interaction between an electron beam and the crystals in thin sections tilted at 70° to the horizontal plane, and indexation of the diffraction patterns was confirmed manually for each orientation. We determined the crystal orientations of ~200 olivine grains and ~100 orthopyroxene grains as one point per grain, and visually checked the computerized indexation of each diffraction pattern.
To characterize the CPOs, we determined the fabric strength and distribution density of the principal crystallographic axes. We used the J-index (Mainprice and Silver 1993) to quantify the intensity of a given CPO. The J-index has a value of unity for a random distribution and a value of infinity for a single crystal (Mainprice and Silver 1993;Michibayashi and Mainprice 2004). The fabric strength was also determined using the M-index technique (Skemer et al. 2005). Using this approach, the distribution of random-pair misorientation angles of a sample (Wheeler et al. 2001) is compared with the distribution of misorientation angles from a theoretical random fabric. The M-index is scaled from zero to one, where M = 0 represents a random fabric and M = 1 represents a single crystal. In addition, we also computed the pfJ index. The pfJ index has a value of unity for a random distribution and a maximum value for olivine of ~ 60 for a single crystal of olivine. In the present case of olivine, the [100], [010], and [001] axes are all two-fold rotation axes,

P-wave velocities estimated from the olivine CPOs and Vp-Flinn diagram
A single crystal of olivine contains intrinsic elastic anisotropies. Therefore, olivine aggregates have characteristic seismic properties in accordance with the distribution of crystallographic axes. For quantifying CPOs, the P-wave velocities of a virtual olivine aggregate were estimated from the crystallographic orientation data measured by EBSD and the elastic property of a single crystal of olivine (Abramson et al. 1997) using the MTEX Toolbox (v. 5.8.0) for MATLAB ® (Mainprice et al. 2011). Since P-wave velocities can be calculated in all directions, the V P anisotropy ( AV P ) is given by where V 1 and V 3 are the maximum and minimum velocities, respectively. Olivine CPO types were quantified using the V P -Flinn Diagram (Michibayashi et al. 2016a, b). The angle of inclination between the point of origin (1, 1) and a point in the Flinn diagram can be used as a quantitative measure of the olivine CPO pattern, which was introduced as the Fabric-Index Angle (FIA) by Michibayashi et al. (2016a), as follows: By using the V P structure of a virtual olivine aggregate as a framework, V X , V Y , and V Z can be rephrased to V 1 , V 2 , and V 3 , respectively. We use three P-wave velocities (V 1 , V 2 , and V 3 ) in a Flinn diagram with V 2 /V 3 for the horizontal axis and V 1 /V 2 for the vertical axis, and the origin at (1, 1). The Fabric-Index Angle, using variable V 1 , V 2 , and V 3 , can be used to determine whether the olivine fabric is AG type, D type, or a point maximum A type (see also Kakihata et al. 2022).

Major element compositions
The major element compositions of olivine, orthopyroxene, clinopyroxene, spinel, and plagioclase were measured using an electron probe micro-analyzer (EPMA: JEOL JXA-8900R) at Shizuoka University, Japan. The operating conditions were an accelerating voltage of (1) 20 kV, a specimen current of 12 nA, and a beam diameter of 5 µm. Natural and synthetic JEOL mineral standards were used for data calibration. The X-ray peak of Ni was counted for 30 s, whereas those of the other elements were counted for 20 s. The conventional ZAF matrix correction was used. The Fe 3+ contents of spinel were calculated on the basis of stoichiometry. We estimated temperature conditions assuming a pressure of 15 kbar and using the Ol-Sp thermometer established by Ballhaus et al. (1990) as follows: Ti is the number of Ti cations in spinel on the basis of 4 oxygens; and K . Oxygen fugacity was calculated from coexisting olivine and spinel compositions using the oxygen barometer of Ballhaus et al. (1990) as follows: where P is in GPa, T is in K, and X sp Al is the Al/ (Fe 3+ + VI Al + Ti + Cr) ratio in spinel.

Trace element compositions
Rare earth element (REE) and trace-element contents of clinopyroxene and orthopyroxene in harzburgite samples were determined using laser ablation-inductively coupled plasma-mass spectrometry (Thermo Scientific Element XR) at Academia Sinica, Taiwan. 43 Ca was used as an internal standard for data reduction based on elemental concentrations obtained by EPMA. NIST SRM 612, BHVO-1, and BCR-1 were used as external calibration standards and were analyzed at the beginning of each batch of no more than 12 unknowns. NIST SRM 612 was analyzed again at the end of each batch. The diameter of ablation spots was 60 μm. After each analysis, data reduction was carried out using GLITTER software (Griffin et al. 2008). (3)

Modal compositions
The peridotites consist of harzburgite and dunite. Results of the modal compositions are shown in Additional file 1: Table S1. Modal abundances of olivine range from 61.1 to 83.0%, and those of orthopyroxene and clinopyroxene range from 11.5 to 32.5% and 0 to 3.4%, respectively. Abundances of spinel range from 0.1 to 3.2%, whereas those of plagioclase range from 0 to 5.7%. These modal data were subsequently re-calculated to estimate the modal compositions of primary minerals in the peridotite, so that mesh-textured serpentine and associated magnetite were assigned to olivine, bastite and associated talc were assigned to orthopyroxene, and bastite with clinopyroxene relics was assigned to clinopyroxene. Magnetite at the margins of spinel was assigned to spinel. The re-calculations show that the 18 peridotites include 17 harzburgites and 1 dunite (Fig. 3).
Twelve harzburgite and one dunite samples have heterogeneous intermediate textures that are characterized by heterogeneous distributions of both coarse and fine olivine grains. Coarse olivine grains show undulose extinction and kink bands (Fig. 5c). In general, olivine grains show polygonal textures with triple junction grain boundaries (Fig. 5d). Orthopyroxenes show undulose extinction.  (Fig. 5e). Plagioclase appears as fresh interstitial grains (Fig. 5e) or altered phases as irregular diffuse dark brown patches of hydrogrossular (Fig. 5f ).
Orthopyroxene CPOs are also shown in Fig. 6. The J-index values ranged from 2.35 to 5.05 and M-index ranged from 0.0092 to 0.0821, although the number of grains measured are mostly less than 100. A few heterogeneous textures correspond to (100)[001] patterns such as 6K-1397-R21 and 6K-1398-R12, whereas fine-grained textures correspond to indistinct patterns with weak fabric intensities. We do not present any data for the patterns of orthopyroxenes in the coarse-grained texture because of the small number of measurements.

Trace element compositions of the pyroxenes
The results of the clinopyroxene analyses are shown in Additional file 1: Table S13. Primitive-mantle-normalized incompatible element patterns for clinopyroxene are shown in Fig. 9a, b, and chondrite-normalized REE patterns are shown in Fig. 9c. The patterns exhibit steepening slopes from the middle to the light REEs. Nb and Zr show positive and negative anomalies relative to the LREEs, respectively.
The results of orthopyroxene analyses are listed in Additional file 1: Table S14. Primitive-mantle-normalized incompatible element patterns for orthopyroxene are shown in Fig. 9d. The patterns slope down from the HREEs to the LREEs. There are positive anomalies of Ti, Hf, and Nb.

Discussion
The shallow (from 1985 to 1740 mbsl) western part of the 139°E Ridge consists of volcanic breccias and carbonate rocks (YKDT-170 in Fig. 1b), whereas Hawkins and Batiza (1977) have found serpentinized peridotites at the similar depth (D-1 in Fig. 1b). Moreover, Mncoated volcanic rubble was observed at a small knoll on the northern flank of the ridge (YKDT-171 in Fig. 1b). In contrast, the deeper (5999 to 3430 mbsl) part of the southwestern slope consists dominantly peridotites along with some volcanic rocks and carbonates (Fig. 1b,  c), suggesting that the 139°E Ridge could consist of the shallower basaltic crust and the deeper mantle peridotites in a cross section (Fig. 11). Moreover, Crawford et al. (1986) suggested that some samples from the deepest north-western flank of the ridge (1438D1 in Fig. 1b) may represent backarc basin affinity of the initial stage of the Mariana Trough opening, as Beccaluva et al. (1980) have obtained ~ 7.8 Ma in age from the volcanic rocks. It suggests that the 139°E Ridge could be abruptly terminated on the west by a tectonic boundary such as a fault. Here, we use our new data to explore Red symbols indicate coarse-grained texture, blue symbols indicate fine-grained texture, and green symbols indicate samples with heterogeneous texture following two questions: (1) What was the origin of peridotites in the 139°E Ridge? and (2) What is the dominant olivine crystal-fabric type in these peridotites and how were these modified by structural development of the 139°E Ridge?

Origin of the 139°E Ridge peridotites
The western side of the Challenger Deep is bounded by the 139°E Ridge. The ridge is located at the westernmost tip of the Mariana forearc as well as at the southern tip of the Parece Vela Rift. Therefore, it is not clear whether the Fig. 8 Compositional data for dunite (red) and harzburgite (green). Error bar is the deviation of average values. Green filled circles indicate harzburgite, and red filled circles indicate dunite. a Average Mg# of olivine vs. average Cr# of spinel in the 139°E Ridge peridotites. The region between the two solid lines is the olivine-spinel mantle array (OSMA), where mantle-derived spinel peridotites plot (Arai 1994). b Average Mg# vs. average Cr# for spinel grains in the 139°E Ridge peridotites. c Average TiO 2 values vs. average Cr# for spinels in the 139°E Ridge peridotites. d Average TiO 2 contents of spinel vs. ΔFMQ values for the 139°E Ridge peridotites. ΔFMQ values were calculated from the major element compositions of olivine and spinel (Ballhaus et al. 1990). ΔFMQ values for abyssal peridotites (Bryndzia and Wood 1990) and forearc peridotites (Parkinson and Arculus 1999) are shown in the vertical lines. e Average TiO 2 values vs. Y Fe values for spinel in the 139°E Ridge harzburgites. The compositional field for the Godzilla Megamullion is shown by green domain (Loocke et al. 2013), the field for the Parece Vela Basin is shown by orange domain , the field for the Mariana Trough is shown by blue domain (Ohara et al. 2002a), the field for the "Mariana Trough" outcropped in Mariana Trench is shown by cyan domain (Michibashi et al. 2009), the field for the Yap Trench is shown by magenta domain (Ohara et al. 2002b;Chen et al. 2019), and the field for the Mariana forearc is shown by a dashed line (Ishii et al. 1992;Ohara and Ishii 1998) peridotites exposed on the ridge were derived from the Mariana forearc or from the mantle of the Parece Vela Basin.
The relationships between spinel Cr# and olivine Mg# are consistent with the olivine-spinel mantle array (OSMA; Fig. 8a; Arai 1994), showing that the peridotites are exposed mantle. Spinel Cr# is generally an indicator of the degree of melting and melt-peridotite reaction, since spinel Cr# correlates with the degree of melting (Dick and Bullen 1984;Keleman et al. 1992). Chemical compositions of spinel in our samples are similar to those of the Parece Vela Rift peridotites Loocke et al. 2013) rather than those of Mariana forearc peridotites (Ishii et al. 1992;Ohara and Ishii 1998) (Fig. 8b, c), suggesting that the peridotites were derived from a relatively fertile mantle with spinel Cr# as low as 0.36. Ti contents can record the influence of trapped or transient MORB-like melts (Dick and Bullen 1984). The elevated TiO 2 contents at medium values of Cr# (~ 0.5) of the harzburgite spinel and presence of interstitial plagioclase are characteristics, like those of Parece Vela Rift peridotites (Fig. 8c), suggesting that melt reacting with  McDonough and Sun (1995). c Chondrite-normalized rare earth element patterns for clinopyroxenes. Chondrite values are from Anders and Grevesse (1989). Black dots indicate clinopyroxenes of the 139°E Ridge peridotites. Red dots indicate calculated clinopyroxenes in equilibrium with Parece Vela Basin basalt . Yellow field is clinopyroxene from Parece Vela Rift peridotites , whereas green field is clinopyroxene from Mariana Trough peridotites (Ohara et al. 2002a, b). The blue field is clinopyroxene from Mariana forearc (Ishii et al. 1992). The partition coefficients between clinopyroxene and melt are from Hart and Dunn (1993). d Primitive-mantle-normalized incompatible element patterns for orthopyroxenes of the 139°E Ridge peridotites. Primitive mantle values are from McDonough and Sun (1995) the peridotites would be compositionally compatible with Parece Vela Basin basalts. Moreover, our peridotite samples show a heterogeneous distribution of clinopyroxene (Fig. 2), and these form veins in sample 6K-1397-R08. The trace element patterns for clinopyroxene are similar regardless of their occurrence, indicating a common origin (Fig. 9a, b). Their middle to heavy REE contents are more depleted than those of the Parece Vela Basin peridotites  but are more fertile than those of the Mariana forearc peridotites (Ishii et al. 1992) (Fig. 9c). It is noted that several clinopyroxene patterns show strongly negative Sr anomalies, which are probably related to the presence of plagioclase, which has a high partition coefficient compared to clinopyroxene (Blundy and Wood 1994). These mineralogical and geochemical features could result from melt-rock interactions under conditions of the shallow lithospheric mantle. As a result, although the 139°E Ridge peridotites were collected from the Mariana Trench inner slope (Fig. 1a-c), we argue that the 139°E Ridge peridotites had been formed in a tectonic environment such as a back-arc basin rather than a subduction initiation/forearc environment, so that the Parece Vela Basin mantle could be exposed on the southwestern slope of the 139°E Ridge (Fig. 1b). Hellebrand et al. (2002) established the following equation for the degree of melting: where F is the degree of fractional melting (%). Accordingly, the degrees of melting of the uppermost mantle are estimated to have been ~ 15%, using plagioclase free samples.
Spinel Ti contents correlate with the redox state (Y Fe and ΔFMQ) (Fig. 8d, e), and such a correlation has been documented for the Parece Vela Basin peridotites by Ohara et al. (2003). The values of ΔFMQ are higher than those of abyssal peridotites but in the range of arcperidotites (Fig. 8d). Since the oxidizing condition of the mantle could have resulted from a hydrous environment (5) F =10 ln (Spinel Cr#) + 24,   (Sato 1978), peridotites with high-TiO 2 spinel may have been oxidized by basaltic melt.

Dominant fabric type and its modification by structural development of the 139°E Ridge
Detailed geological surveys using the submersible Shinkai6500 (dives 6 K-1397 and 6 K-1398) revealed that the fresh peridotites occur on the 139°E Ridge. Peridotites on the 139°E Ridge show somehow systematic textural variations as a function of their exposed depths. Coarse-grained peridotites have been found from the shallowest sites (3705-4042 mbsl), whereas fine-grained peridotites occur at the deepest site (5996 mbsl) (Fig. 10a,  b). These textural variations could have resulted from various deformation conditions such as deformation at near-solidus temperatures for the coarse-grained peridotites and ductile shearing at lower temperatures for the fine-grained peridotites (e.g., Nicolas and Poirier 1976;Michibayashi and Mainprice 2004). There is a relationship between olivine crystal-fabrics and their textures. A-type crystal-fabrics occur in coarsegrained peridotites (Fig. 10a), and this is consistent with experimental studies demonstrating that (010)[100] slip system is the most active among {0kl}[100] slip systems under the highest-temperature conditions (e.g., Nicolas et al. 1973;Avé Lallemant et al. 1975;Zhang and Karato 1995). Therefore, the A-type crystal-fabrics in the coarsegrained peridotites on the 139°E Ridge may represent the dominant fabric of the lithospheric mantle below the Parece Vela Basin.
The peridotites with fine-grained textures from the deepest site on the ridge show higher FIA values for their D-type crystal-fabrics than the other harzburgites ( Fig. 10a, b). Since D-type crystal-fabrics are considered to result from deformation under relatively higher flow stresses than those for A-type crystal-fabrics (Mainprice and Nicolas 1989;Jung et al. 2006;Michibayashi et al. 2006), a ductile shear zone could occur in the deepest part of the 139°E Ridge (Fig. 11). Therefore, the 139°E Ridge may have been uplifted along an underlying ductile shear zone, where the Caroline Ridge collided with the southern Mariana forearc (Altis 1999;Miller et al. 2006). It is likely that the development of a ductile shear zone in the deepest part of the ridge would have been induced by this collision (Fig. 11).
The temperatures ranged from ca. 600-750 °C, regardless of textures and sampling depths (Fig. 10c), suggesting that the estimated temperatures may not reflect deformation conditions. Instead, we argue that the heterogeneous textures between the coarse-grained and the fine-grained textures along the depth profile (Fig. 10a, b) could correspond to different degrees of deformation rather than differences in deformation conditions such as temperature ( Fig. 11; e.g., Warren et al. 2008;Michibayashi et al. 2006Michibayashi et al. , 2009.

Conclusions
Based on new geochemical and crystal-fabric data for peridotites of the 139°E Ridge (11°2′N, 139°3′E) in the westernmost side of Challenger Deep, we came to the following conclusions.
Peridotites consist of mantle harzburgite and dunite. Chemical compositions of spinel in the peridotites indicate a relatively fertile mantle with Cr# as low as 0.36. Furthermore, the elevated TiO 2 contents at medium values of Cr# (~ 0.5) of spinel and presence of interstitial plagioclase as well as the trace element compositions of pyroxene are characteristics closer to those of Parece Vela Rift peridotites rather than those of the Mariana forearc peridotites, suggesting that the Parece Vela Basin mantle could be exposed on the southwestern slope of the 139°E Ridge.
Peridotites have either coarse-grained texture (> 1 mm), heterogeneous intermediate texture, or fine-grained texture (< 0.5 mm). Coarse-grained peridotites are sampled from the shallowest part (3705-4042 m) of the dive area, whereas the fine-grained peridotites came from the deepest part (5996 m). Olivine crystal-fabrics vary with peridotite textures: two coarse-grained textures are associated with (010)[100] patterns, three fine-grained textures with {0kl}[100] patterns, and heterogeneous textures with various indistinct fabric patterns. The variations of the olivine textures and crystal-fabrics with depth suggest variations in the deformation process with depth. As a consequence, it is likely that a ductile shear zone could occur in the deepest part of the 139°E Ridge, by which the 139°E Ridge may have been uplifted at its base where the Caroline Ridge collided with the southern Mariana forearc.
Additional file 1. A collection of petrophysical and geochemical data for the 139°E Ridge peridotites shown in this study.