Lattice-Preferred Orientation and Seismic Anisotropy of Minerals in Retrograded Eclogites from Xitieshan, Northwestern China, and Implications for Seismic Reﬂectance of Rocks in the Subduction Zone

: Various rock phases, including those in subducting slabs, impact seismic anisotropy in subduction zones. The seismic velocity and anisotropy of rocks are strongly affected by the lattice-preferred orientation (LPO) of minerals; this was measured in retrograded eclogites from Xitieshan, northwest China, to understand the seismic velocity, anisotropy, and seismic reﬂectance of the upper part of the subducting slab. For omphacite, an S-type LPO was observed in three samples. For amphibole, the <001> axes were aligned subparallel to the lineation, and the (010) poles were aligned subnormal to foliation. The LPOs of amphibole and omphacite were similar in most samples. The misorientation angle between amphibole and neighboring omphacite was small, and a lack of intracrystalline deformation features was observed in the amphibole. This indicates that the LPO of amphibole was formed by the topotactic growth of amphibole during retrogression of eclogites. The P-wave anisotropy of amphibole in retrograded eclogites was large (approximately 3.7–7.3%). The seismic properties of retrograded eclogites and amphibole were similar, indicating that the seismic properties of retrograded eclogites are strongly affected by the amphibole LPO. The contact boundary between serpentinized peridotites and retrograded eclogites showed a high reﬂection coefﬁcient, indicating that a reﬂected seismic wave can be easily detected at this boundary.

In addition, owing to the various seismic velocity structures of subduction zones, seismic waves tend to be reflected at the boundary of rock phases with varying seismic velocities. In particular, the upper parts of subducting slabs have a fast seismic velocity [6], which can cause a difference in seismic velocity from surrounding rocks and may reflect seismic waves due to this difference in seismic velocity [6,[50][51][52][53]. In previous studies, the reflection coefficient (R C ) was suggested to quantify the degree of seismic wave reflection [6,50,51]. Calculating the R C of the boundaries between retrograded eclogites and surrounding rocks, using their P-wave velocities, and comparing the degree of retrogression of eclogites and R C of these boundaries may explain the degree of reflection of seismic waves at these boundaries.
Garnet is a seismically nearly isotropic mineral with a cubic shape that develops little LPO [54]. The LPOs of other minerals in retrograded eclogites-amphibole and omphacite-have been studied in both naturally and experimentally deformed rocks. Four types of LPOs of amphibole have been reported: type I, type II, type III, and type IV [18,36]. A previous experimental study on amphibolite, which was conducted at a pressure of 1 GPa and temperatures of 480-700 • C [18], reported that the type I amphibole LPO was formed under relatively low-stress and low-temperature conditions and is characterized as the <100> axes aligned subnormal to foliation and <001> axes aligned subparallel to lineation. The type II amphibole LPO was formed under relatively high-stress and moderate-temperature deformation conditions and is characterized as <100> axes aligned subnormal to foliation and (010) poles aligned subparallel to lineation. The type III amphibole LPO was formed under relatively low-stress and high-temperature deformation conditions and is characterized as <100> axes aligned subnormal to foliation, with both <001> axes and (010) poles forming a girdle shape subparallel to foliation [18]. A recent experimental study reported that the type IV amphibole LPO can be formed owing to high strain [36] and is characterized as <001> axes aligned subparallel to lineation, with both <100> axes and (110) poles forming a girdle shape subnormal to the shear direction [36]. Previously, the LPOs of amphibole, similar to the type IV amphibole LPO, have been reported in naturally deformed rocks, which were observed in metabasites from southwestern Spain and northwest Scotland, retrogressed Limo harzburgites from northwestern Spain, and amphibolites from Spain [44,[55][56][57][58]. On the other hand, an amphibole LPO has also been reported in which the (010) poles are aligned subnormal to foliation and the <001> axes are aligned subparallel to lineation [44,59,60].
Three types of omphacite LPOs have been reported: L-, S-, and LS-type omphacite LPOs [61]. The S-type omphacite LPO is characterized as the <001> axis of omphacite forming a girdle parallel to foliation and the (010) pole of omphacite aligned subnormal to foliation. The L-type omphacite LPO is characterized as the <001> axis of omphacite aligned subparallel to lineation and the (010) pole of omphacite forming a girdle subnormal to lineation. The LS-type omphacite LPO is characterized as the <001> axis of omphacite aligned subparallel to lineation and the (010) pole of omphacite aligned subnormal to foliation, indicating a (010) <001> slip system [20,61].
The seismic reflection coefficient (R C ) was previously calculated only for the boundary between fresh eclogite and neighboring rocks [6,50]. In this study, the omphacite and amphibole LPOs in the retrograded eclogites from Xitieshan, northwest China, which represent various degrees of retrogression (35-90%), were measured. The Xitieshan eclogites are useful for studying the relationship between the degree of retrogression and seismic velocity, anisotropy, and reflectance. The seismic velocity and anisotropy of retrograded eclogites were calculated using the LPOs of minerals, and the effect of the amphibole LPO on the seismic velocity and anisotropy of retrograded eclogites was also investigated. R C was calculated using the P-wave velocity, and R C was examined to understand the seismic wave reflectance of the boundaries between the retrograded eclogites and neighboring rocks. The results of the present study provide a valuable opportunity to understand the seismic velocity, anisotropy, and reflectance of retrograded eclogites and to identify the boundary between the subducting slab and mantle wedge in various subduction zones.

Geological Settings
Xitieshan is located in the central part of the North Qaidam ultra-high-pressure (UHP) metamorphic belt, which is located at the northeastern margin of the Tibetan Plateau in northwestern China (Figures 1 and 2). The North Qaidam UHP metamorphic belt is a typical Alpine-type continental collision/subduction zone, containing the characteristic continental rock association of granitic and pelitic gneiss with intercalated eclogite and peridotite [33,34,[62][63][64]. From the southeastern part, four individual UHP terranes were observed: the Dulan eclogite-bearing terrane, the Xitieshan eclogite-bearing terrane, the Lulingshan garnet peridotite-bearing terrane, and the Yuka eclogite-bearing terrane. Coesite and diamonds were found as inclusions in zircon and garnet in all four terranes, proving that these four terranes had been under UHP metamorphic conditions, similar to other continental UHP metamorphic terranes in China [64].  The eclogites in the Xitieshan terrane can be divided into two groups: bimineralic eclogite and phengite-bearing eclogite. Bimineralic eclogite is the major component of the Xitishan terrane, whereas phengite-bearing eclogite concentrations are minor and can only be seen in the Huangyanggou area. Most eclogites in the Xitishan terrane were retrograded into garnet amphibolite or amphibolite. The original eclogite can be found in the center of the large bodies.
The Xitieshan eclogites experienced two stages of retrogression. According to previous studies [66,67], the peak pressure and temperature conditions of metamorphism of the Xitieshan eclogite were P = 2.71-3.17 GPa and T = 751-791 • C. In the first stage (isothermal decompression), the proportion of omphacite and garnet in the sample decreased, while the proportion of amphibole and plagioclase increased at a temperature of T~750 • C when pressure decreased from P~2.3 GPa to P~1.5 GPa. In the second stage (symplectite formation), symplectite of diopside, plagioclase, and edenite appeared under the condition of P = 8-12 kbar and T = 660-720 • C [67].
In addition to the Xitieshan area, geochronology data reported that there were two major different ages measured using zircon: 750-800 Ma and 877 ± 8 Ma were calculated in the cores of the zircon, and 440-460 Ma and 433 ± 3 Ma were calculated at the rim of the zircon. Based on the ages calculated in the core of the zircon, some of the Xitieshan terrane eclogites were produced in Neoproterozoic protoliths much before the Paleozoic subduction or collision event occurred in the North Qaidam UHP metamorphism. Eclogite metamorphism occurred in the age calculated at the rim of zircon [67][68][69][70][71].

Measurement of Lattice-Preferred Orientation (LPO) of Minerals
Foliation of a rock was defined by compositional layering of amphibole and garnet, and lineation was determined by the shape-preferred orientation of minerals on foliation, using the projection function method [72]. Thin sections were made in the XZ plane to measure the LPO, the X axis was set parallel to lineation, and the Z axis was set normal to foliation.
The LPOs of the samples were measured using electron back-scattered diffraction (EBSD) using a JEOL-JSM-7100 field emission scanning electron microscope (FE-SEM) with AZTEC and HKL Channel 5 software (5.12.74.0) installed at the School of Earth and Environmental Sciences (SEES), Seoul National University (SNU). Detailed settings of the EBSD mapping are provided in Table 1. An accelerating voltage of 20 kV and a working distance of 20 mm with a tilting angle of 70 • were the main settings of the FE-SEM with EBSD. The samples were observed using automatic mapping mode, with a step size of 23 to 35 µm, depending on the grain size of most of the samples. To eliminate the oversampling of large grains, pole figures were constructed from one point per grain [73]. The omphacite LPOs, which were in direct contact with amphibole single crystals, were measured using a smaller step size (1 µm). The misorientation index (M-index) [74] was calculated to estimate the fabric strength of the sample using the uncorrelated grain pairs obtained from the orientation data. The M-index ranged from 0 (random) to 1 (single grain). The misorientation angle was measured using HKL Channel 5 software (Tango).  [76]). * Sample 882 is the most retrograded eclogite, and the composition of omphacite was changed to diopside. X: lineation, Z: normal to foliation, and Y: normal to X and Z.

Calculation of Seismic Velocity and Anisotropy
Seismic velocity and anisotropy were calculated using the orientation data of the minerals and the single-crystal elastic constants of amphibole [77] and omphacite [29] in sample 885, 880, and 881; diopside [78] in sample 882; garnet [54]; and plagioclase [79] using the Fortran program [80]. The whole-rock seismic velocity and anisotropy were calculated by combining the calculated results of individual minerals, including amphibole, garnet, and omphacite in samples 885, 880, and 881; diopside in sample 882; and all minor minerals, quartz [81], ilmenite [82], and mica [83], with weighted values for the number of grains.
AV P indicates the anisotropy of the P-wave velocity in all propagation directions, calculated using Equation (1). AV S indicates the anisotropy of split fast and slow S-wave velocity in the same propagation direction, calculated using Equation (2) [80].

Micro-Raman Spectroscopy
A micro-Raman spectrometer located at the Tectonophysics Laboratory in SEES, SNU, was used to identify minerals clearly. This spectrometer was equipped with a 532 nm laser (10 mW power) and an optical microscope with a 50× objective lens. It had a resolution of 0.01 cm −1 over the wavenumber range of 50 to 3550 cm −1 , with a beam size of 0.67 µm. The spectrum was overlapped 32 times to reduce noise.

Analysis of Chemical Composition and Temperature Estimation
The chemical composition of the minerals in the samples was measured using a JEOL JXA-8530F Plus electron hyper probe installed at the Center for Research facility at Gyeongsang National University and at the National Center for Inter-University Research Facilities at Seoul National University, South Korea, using an acceleration voltage of 15 kV, a beam current of 10 nA, and a beam size of 5 µm in diameter. The amphibole phase was classified using the Excel spreadsheet provided by Locock [101] based on the classification of Hawthorne et al. [102].
The equilibrium temperature of the samples was estimated by thermometry of garnet-clinopyroxene Fe-Mg exchange [75,76], and the measured oxide mass percentages of garnet, clinopyroxene, amphibole, and minor minerals were used to calculate the equilibrium temperature.

Sample Description and Microstructures
All samples were identified as retrograded eclogites, and the degree of retrogression of the samples varied depending on the sample. The samples had a high proportion of amphibole and plagioclase, and half of the clinopyroxene grains exhibited a symplectite structure of omphacite, diopside, and plagioclase, indicating that our samples were retrograded. The modal composition of rocks and the aspect ratio of minerals are shown in Table 1, and photomicrographs of the samples are shown in Figure 3. Sample 885 was a relatively fresh eclogite, whereas sample 882 was the most retrograded eclogite. Samples 881 and 880 showed an intermediate degree of retrogression, between that of samples 885 and 882. The phase of the minerals in sample 885 (the relatively fresh eclogite) was identified using micro-Raman spectroscopy. Clinopyroxene was identified as omphacite, with Raman peaks at 142, 338, 674, and 1016 cm −1 , whereas garnet in the same sample showed Raman peaks at 356 and 911 cm −1 . An unknown mineral found in the relatively fresh eclogite showing Raman peaks at 175, 300, and 1099 cm −1 was identified as zircon ( Figure 4).  Amphibole and omphacite in the samples showed well-developed foliation (Figure 3b-f), but garnet showed less elongated grains than other minerals, indicating that garnet grains are less affected by deformation (Figure 3a,b,f). All mineral grains, except garnet, showed curved or angular-shaped grain boundaries, whereas garnet grains showed straight grain boundaries. Most of the amphibole grains showed cleavage in grains, with elongated grain shapes (Figure 3b,d). However, amphibole grains showed a lack of intracrystalline deformation features, such as undulose extinction or subgrains (Figure 3b,d).

Chemical Composition and Equilibrium Temperature
The chemical compositions of the minerals in the samples are listed in Table 2. Muscovite was found in two samples, 882 and 885, whereas the other samples, 880 and 881, did not contain muscovite. By analyzing the amphibole phase by chemical composition, amphibole in all the samples was found to be magnesiohornblende. Based on the chemical composition of the analyzed minerals and the calculation of the temperature conditions of the samples using geothermometry [75,76], the samples were found to be equilibrated at a temperature of~750 • C. The equilibrium temperatures of samples are presented in Table 1.

Lattice-Preferred Orientations of Minerals
The amphibole LPOs are shown in Figure 5, and the strengths of the LPOs of the samples are described in Table 3. For amphibole in retrograded eclogites with relatively little retrogression (samples 880, 881, and 885), the <001> axes of amphibole were strongly aligned subparallel to lineation (L), with a weak girdle distribution subparallel to foliation (S). The amphibole (010) poles were strongly aligned subnormal to foliation, and the amphibole (110) poles formed a girdle distribution subnormal to lineation. The most retrograded eclogite (sample 882) showed that the amphibole <001> axes were strongly aligned subparallel to lineation, with a girdle distribution subparallel to foliation (S), but both amphibole (010) and (110) poles were weakly aligned subnormal to foliation ( Figure 5).
The omphacite LPOs are shown in the lower part of Figure 5, and the strengths of the LPOs of the samples are listed in Table 3. Omphacites in samples 880, 881, and 885 showed that the <001> axes were strongly aligned subparallel to lineation, with a girdle distribution subparallel to foliation, and the (010) poles were strongly aligned subnormal to foliation. This omphacite alignment is known as the S-type omphacite LPO [61]. The most retrograded sample 882 contained little clinopyroxene, and the number of clinopyroxene grains in the pole figure was too small (<150 grains) to say LPO [74]. The pole figure did not provide a reliable indication.
The garnet and plagioclase LPOs are shown in Figure 6. They showed a weak LPO, which was considerably dispersed compared to the amphibole and omphacite LPO. The garnet LPO was almost random, independent of the degree of retrogression. In the case of the plagioclase LPO, two samples (885 and 880) showed (010) poles aligned subparallel to lineation.

Misorientation between Amphibole and Omphacite
The misorientation angles between omphacite and amphibole in direct contact with omphacite were measured and are shown in Figure 7. The misorientation angles were significantly low-lower than 3 • in both boundaries between omphacite grains, and amphibole in direct contact with omphacite (Figure 7a-c, <0.5 • ) and the boundary between omphacite and amphibole, which were trapped and surrounded by omphacite (Figure 7d-f, <2 • ). The symplectites in Figure 7 consist of omphacite, amphibole, diopside, and plagioclase, which were confirmed by measuring the chemical compositions of minerals using the electron probe micro analyzer installed at the National Center for Inter-University Research Facilities at Seoul National University, South Korea.

Seismic Velocity and Anisotropy
The seismic velocities and anisotropies of the samples are summarized in Table 4. The seismic velocity and anisotropy of the samples are shown in the stereonet for the single-mineral amphibole, omphacite, and garnet and for the whole rock consisting of all minerals, including minor minerals such as plagioclase and ilmenite (Figures 8-10). In the case of amphibole, P-wave velocity (V P ) was in the range of 6.59 to 7.09 km/s, and P-wave anisotropy (AV P ) was in the range of 3.7% to 7.3%, whereas the maximum S-wave anisotropy (max. AV S ) was in the range of 2.72% to 3.61%. In the case of omphacite, the V P of omphacite was in the range of 8.31 to 8.54 km/s, which is faster than that of amphibole. However, the AV P of omphacite was in the range of 2.1% to 2.6% and the maximum AV S was in the range of 1.23% to 1.56%, which is smaller than that of amphibole. In the case of garnet, V P was the fastest among the minerals, ranging from 9.31 to 9.34 km/s. However, garnet AV P and maximum AV S were the smallest compared to other minerals, ranging from 0.1% to 0.3% and 0.16% to 0.55%, respectively (Table 4, Figure 9).     The seismic velocity and anisotropy of whole rock showed a value between the seismic velocity and anisotropy of omphacite and amphibole. The V P of whole rock was in the range of 6.70 to 7.66 km/s, AV P was in the range of 1.2% to 6.6%, and the maximum AV S of whole rock was in the range of 1.40% to 2.46%. The seismic velocity and anisotropy profile of whole rock (Figure 10, color-coded and contour lines in pole figures) were much more similar to those of amphibole than those of other minerals (Figures 8 and 9).

Reflection Coefficient of Samples
The reflection coefficient (R C ) of the boundaries between the studied retrograded eclogites and other rocks in subduction zones is shown in Table 5. The R C of the boundary between serpentinites and the most retrograded eclogite (sample 882) was in the range of 0.024 to 0.147. The R C of the boundary between blueschists and the most retrograded eclogite was in the range of 0.011 to 0.05. In addition, the R C of the boundary between dunite and the most retrograded eclogite was in the range of 0.060 to 0.092. Finally, the R C of the boundary between peridotites and the most retrograded eclogite was in the range of 0.067 (for amphibole peridotite) to 0.129 (for garnet peridotite). Furthermore, the maximum R C (R C = 0.204) was observed at the contact boundary between the serpentinized peridotite [97] and relatively fresh eclogite (sample 885), and the minimum R C (R C = 0.002) was observed at the contact boundary between dunite [91] and relatively fresh eclogite.

Lattice-Preferred Orientations of Minerals
Omphacite in retrograded eclogites (samples 880, 881, and 885) from Xitieshan, northwestern China, showed that the <001> axes are strongly aligned subparallel to lineation, with a girdle distribution subparallel to foliation, and the (010) poles are strongly aligned subnormal to foliation. This is called S-type omphacite LPO [61]. The observed S-type omphacite LPOs were consistent with those reported in previous studies [8,22]. The S-type omphacite LPO has been reported in eclogites from the eastern Alps [103] and Cabo Ortegal eclogites from northwestern Spain [104].
Amphibole in retrograded eclogites from Xitieshan, northwest China, showed that the amphibole (010) poles were aligned subnormally to foliation, and the maximum concentration of amphibole <001> axes were aligned subparallel to lineation, which is referred to as type V amphibole LPO. Previously, four other types of amphibole LPOs have been reported in both naturally and experimentally deformed amphibole [12,18,36]. The type V amphibole LPO in this study is different from the amphibole LPOs (types I, II, III, and IV) reported previously by high-pressure experimental studies. Similar amphibole LPOs to type V LPO have been reported only in naturally deformed amphibole from the Malpica-Tui allochthonous complex, northwestern Spain, and Cabo Ortegal, Spain [44,59], and in retrograded eclogites from the Sanbagawa metamorphic belt, central Shikoku, southwestern Japan [60]. To verify that there was no error in the reference frame (X: lineation, Z: normal to foliation) of samples in our study, the aspect ratios of amphibole and omphacite in the XZ, YZ, and XY planes were determined and compared (Table 1). Grains on the XZ plane were more elongated compared to those on the other planes. This result indicates that the reference frame is correct.
The amphibole LPO (type V) in the studied samples (samples 880, 881, and 885) has only been reported in retrograded eclogites, and the amphibole LPO was similar to that of omphacite in the same samples ( Figure 5). The lack of intracrystalline deformation in amphibole indicates that the samples were not deformed after the retrogression of eclogite (Figure 3b,d). In addition, comparison of the misorientation angle between amphibole and the contacting omphacite (Figure 7) showed small angles in both the contact boundary (<2 • ) and the grain itself (<2.5 • ), suggesting that type V amphibole LPOs are strongly affected by omphacite LPOs [105,106]. Our results indicate that the amphibole LPO is formed due to the topotactic growth of amphibole during the retrogression of eclogite. Previous studies have also reported that amphibole LPOs are similar to those of omphacite in the same sample, and they suggested that amphibole is formed due to the topotactic growth during retrogression of eclogite [44,59,60].

Seismic Velocity, Anisotropy, and Reflection Coefficient (R C )
The amphibole P-wave velocity (V P ) was slow (in the range of 6.59 to 7.09 km/s), and amphibole P-wave anisotropy (AV P ) was large (in the range of 3.7% to 7.3%) ( Table 4). The amphibole S-wave velocity (V S ) was also slow (in the range of 3.64 to 3.79 km/s) and amphibole S-wave anisotropy (max. AV S ) large (in the range of 2.72% to 3.61%). When the omphacite seismic velocity and anisotropy were compared with those of amphibole, the seismic velocity of omphacite was much faster than that of amphibole in both P-and S-waves, but the seismic anisotropy of omphacite was much smaller than that of amphibole in both P-and S-waves (Table 4). This is consistent with previous studies on the seismic velocity and anisotropy of these minerals in eclogites [32,107] and the seismic velocity and anisotropy analysis of mineral mixtures in eclogite [22].
The garnet P-wave velocity was the fastest in the studied samples, whereas the amphibole P-wave velocity was the slowest. The P-wave and S-wave anisotropies of garnet were the slowest in the samples, similar to previous studies [22]. The trends in seismic velocity and anisotropy of whole retrograded eclogites (color coded in pole figures in Figure 10) were similar to those of amphibole ( Figure 8). This result indicates that the seismic properties of retrograded eclogites are strongly affected by the seismic properties of amphibole due to the high fraction of amphibole (33-91.5%) in retrograded eclogites.
When the seismic velocity and anisotropy of relatively fresh eclogite (sample 885) were compared with those of the most retrograded eclogite (sample 882), the P-wave velocity decreased with increasing retrogression (Figures 8-10 and Table 4), but the P-and S-wave anisotropies increased with increasing retrogression (Figures 8-10 and Table 4). This result indicates that when eclogite increases to the eclogite instability zone and gets retrograded, the seismic velocity of retrograded eclogite decreases, but the seismic anisotropy of retrograded eclogite increases.
The R C of the studied samples was in the range of 0.002 to 0.204 (Table 5). The contact boundary between relatively fresh eclogite and serpentinized peridotites showed a high R C (up to 0.204), which was high enough to reflect the seismic wave. However, at the contact boundary between the studied eclogites and the dunites and blueschists, the R C was small (0.002-0.05), independent of the degree of retrogression of eclogite (Table 5). A small R C indicates that the seismic wave reflects little at this boundary, and the seismic wave reflected by this boundary is difficult to detect [6].
In Figure 11, the R C between various rocks and the studied eclogites is compared with the P-wave velocity. In the case of serpentinized peridotites, which have low V P (V P < 6.8 km/s), the R C was high for the contact boundary with relatively fresh eclogite (sample 885), indicating that the boundary between serpentinized peridotites and the relatively fresh eclogites reflects a large seismic wave, resulting in this boundary being easily detected compared to the other contact boundaries of rocks. In contrast, in the case of dry mantle rocks (spinel peridotites and garnet peridotites), which have a high V P (V P > 7.5 km/s), the R C was high for the contact boundary with the most retrograded eclogite (sample 882). This result indicates that the contact boundary between dry mantle rocks and the most retrograded eclogite reflects more seismic waves, resulting in the boundary being easily detected compared to the other contact boundaries of rocks, similar to the contact boundary between serpentinized peridotites and relatively fresh eclogite. However, in the case of blueschists, which have an intermediate V P (6.8 km/s < V P < 7.5 km/s), the R C was low for the contact boundary with the studied eclogites, whether eclogites were retrograded or not, indicating that the boundaries between blueschists and eclogites reflected little seismic waves, independent of the degree of retrogression of eclogite, indicating that these boundaries may be difficult to detect. Figure 11. Relationship between P-wave velocity (V P ) of a surrounding rock and reflection coefficient (R C ) of the contact boundary between the studied retrograded eclogite and the surrounding rock (Table 5).

Conclusions
The LPOs of minerals in retrograded eclogites from Xitieshan, northwestern China, were studied to understand the seismic anisotropies and seismic reflection coefficients of the rocks that are in contact with other surrounding rocks. The amphibole LPOs in the retrograded eclogites were mostly similar to omphacite LPOs. The low misorientation angle between amphibole in direct contact with omphacite suggests that the amphibole LPO is formed due to topotactic growth of amphibole during the retrogression process of eclogites. There was a lack of intracrystalline deformation features in amphibole, such as undulose extinction and subgrains, indicating that there might have been little deformation after the retrogression of eclogites.
The seismic properties of retrograded eclogites were more severely affected by amphibole than by other minerals due to the high fraction of amphibole (33-91.5%) in retrograded eclogites. During the retrogression process of eclogites, the seismic velocity slowed and the seismic anisotropy increased. R C , which was calculated using the V P of minerals, showed that the contact boundaries between relatively fresh eclogite and hydrated mantle rocks, such as serpentinized peridotites, showed high R C , indicating that these contact boundaries can be easily detected using the reflected seismic wave, but the contact boundaries between dunites or blueschists and eclogites showed low R C , independent of the degree of retrogression. This indicates that these contact boundaries are expected to be difficult to detect using the reflected seismic wave.