www.elsevier.com/locate/tecto

Tectonophysics 370 (2003) 49–76

Microstructures, petrofabrics and seismic properties of ultra

high-pressure eclogites from Sulu region, China: implications

for rheology of subducted continental crust and origin

of mantle reflections

Shaocheng Jia,*, Kazuko Saruwataria,1, David Mainpriceb, Richard Wirthc,Zhiqin Xud, Bin Xiae

aDepartement des Genies Civil, Geologique et des Mines, Ecole Polytechnique de Montreal, Succursale Centre-Ville,

Montreal, Quebec, Canada H3C 3A7bLaboratoire de Tectonophysique, Universite de Montpellier II, 34095 Montpellier, Cedex 05, France

cGFZ-Potsdam, Telegrafenberg, D-14473, Potsdam, GermanydChinese Continental Scientific Drilling Program (CCSD), Institute of Geology, Chinese Academy of Geological Sciences, Beijing, PR China

eGuangzhou Institute of Geochemistry, Chinese Academy of Sciences, Wushan, Guangzhou 510640, PR China

Accepted 31 March 2003

Abstract

Ultra high-pressure (UHP) eclogites from Sulu region (China) represent mafic components of the continental crust,

which were first subducted to mantle depths greater than 100 km and then exhumed to the earth’s surface. Detailed

investigation of microstructures, chemical compositions, petrofabrics and seismic properties of the UHP eclogites can

provide important information on the operating deformation mechanisms and rheology of subducted continental crust and

on the origin of seismic reflections within the upper mantle. We present here results from field, optical and TEM

observations, electron back-scattered diffraction (EBSD) measurements and numerical computations of the seismic

properties of UHP eclogites collected from fresh surface outcrops at the drill site (Maobei, Donghai County, Jiangsu

Province) of the Chinese Continental Scientific Drilling Program (CCSD). Two types of eclogites have been distinguished:

Type-1 (coarse-grained) eclogites deformed by recovery-accommodated dislocation creep at the peak metamorphic

conditions, and Type-2 (fine-grained) eclogites which are composed of reworked Type-1 materials during recrystallization-

accommodated dislocation creep in shear zones which were active during the exhumation of the UHP metamorphic rocks.

Both garnet and omphacite in these eclogites deformed plastically and the flow strength contrast between these two

constituent minerals is apparently much less than an order of magnitude under the UHP metamorphic conditions. Plasticity

of eclogites under UHP conditions can effectively facilitate channeled flow along the interplate shear zone. The

0040-1951/03/$ - see front matter D 2003 Elsevier B.V. All rights reserved.

doi:10.1016/S0040-1951(03)00177-X

* Corresponding author. Fax: +1-514-3403970.

E-mail address: sji@polymtl.ca (S. Ji).1 Present address: Department of Earth Science, Faculty of Science, University of Tokyo, Hongo 7-3-1 Bunkyo-ku Tokyo, Japan.

S. Ji et al. / Tectonophysics 370 (2003) 49–7650

preservation of the relict crustal materials within the continental lithosphere may produce regionally extensive, strong,

seismic reflections in the upper mantle. This may explain the origin of mantle reflections observed in many areas of the

world.

D 2003 Elsevier B.V. All rights reserved.

Keywords: Eclogite; Microstructure; Petrofabrics; Rheology; Subduction; Garnet; Omphacite; Seismic reflection

1. Introduction Wang and Liou, 1991) and diamond (Xu et al., 1992)

Whereas the upper crust is accessible to geological

sampling and mapping, the deeper portions of the crust

and the upper mantle are relatively inaccessible. Much

of our understanding about the composition, structure

and evolution of the deep crust and upper mantle

beneath continents has been derived from various

seismic refraction and reflection measurements. Inter-

pretation of these seismic data is mainly based on our

knowledge of the elastic properties of rocks and rock-

forming minerals, the latter obtained largely from

laboratory measurements of ultrasonic velocities and

density (e.g., Christensen, 1965; Kern and Richter,

1981; Kern, 1990; Ji and Salisbury, 1993; Christensen

and Mooney, 1995; Burlini and Tancredi, 1998; see Ji

et al., 2002, for summary). However, it is not easy to

simulate the pressure and temperature environments of

the earth’s mantle in the laboratory. Thus, theoretical

calculations of seismic properties based on the elastic

theory (Crosson and Lin, 1971; Ji and Mainprice,

1988; Mainprice, 1990) become another way to esti-

mate the anisotropic seismic properties of any kind of

possible rocks when all the physical properties such as

elastic constants, density and crystallographic pre-

ferred orientation (CPO) are known for each constitu-

ent mineral (e.g., Mainprice and Silver, 1993; Ji et al.,

1994; Mainprice et al., 2000; Mauler et al., 2000;

Saruwatari et al., 2001). The CPO measurement of

ultrahigh pressure (UHP) eclogites from former conti-

nental subduction zones has recently been made pos-

sible by technological developments of SEM-based

measurements of minerals of any symmetry, such as

cubic garnet (Prior et al., 2000) and omphacite (Mauler

et al., 2000; Bascou et al., 2002).

The Qinling–Dabie–Sulu ultrahigh pressure (UHP)

metamorphic belt, which is about 1000 km long and

120 km wide, is the largest among seven well-recog-

nized UHP belts in the world and has been investigated

extensively since the discovery of coesite (Xu, 1987;

from the region. The UHP rocks were a part of

lithospheric slab subducted at the intracontinental col-

lision zone between the Sino-Korean and Yangtze

blocks in the Late Triassic or Early Jurassic (e.g., Cong,

1996; Liou, 1999). Previous studies (e.g., Hirajima et

al., 1992; Yang et al., 1993; Zhang et al., 1994) suggest

that the UHP rocks from Dabie and Sulu region was

subducted to a depth up to about 150 km before

being rapidly exhumed to the earth’s surface. Pet-

rologic, mineralogical and geochemical aspects of

these UHP rocks have already been thoroughly

investigated during last decade (e.g., Zhang et al.,

1994; Cong, 1996; Liou, 1999). However, deforma-

tion microstructures of these UHP rocks have re-

ceived little attention although such investigations

may provide important information on deformation

mechanisms, rheological and petrophysical proper-

ties of the continental lithosphere first subducted

into and then exhumed from the upper mantle. In

this paper, we report the results of our investiga-

tions on microstructures, crystallographic preferred

orientation (CPO) and calculated seismic properties

of the UHP-eclogites from the Sulu UHP metamor-

phic terrane. These results are then applied to constrain

deformation mechanism and rheological behavior of

subducted slab and origin of seismic reflections in the

upper mantle.

2. Geological setting

The Sulu UHP metamorphic terrane, which has

been displaced about 530 km by the sinistral, strike–

slip Tanlu fault, is an eastern extension of the

Triassic Qinling–Dabie collision zone (e.g., Wang

and Liou, 1991; Wang et al., 1992). On the north,

the UHP metamorphic terrane is bounded by the

Yantai–Qingdao–Wulian fault with the Proterozoic–

Archean basement (Fig. 1), which is overlain by

Fig. 1. Simplified geological map of the Sulu ultrahigh-pressure (UHP) metamorphic terrane, eastern China. The Sulu region is situated in the

northern Jiangsu and eastern Shandong province. Star symbol represents the drill site (Maobei, Donghai County, Jiangsu Province) of the

Chinese Continental Scientific Drilling Program (CCSD) where eclogite samples were collected. Modified from Zhang et al. (1994).

S. Ji et al. / Tectonophysics 370 (2003) 49–76 51

Paleozoic cratonal cover and the Mesozoic volcano-

clastic rocks (Zhang et al., 1994). On its southern

limit, the high-pressure (HP) terrane includes blues-

chists bounded by the Haizhou–Siyang fault. The

UHP terrane consists of eclogite, gneiss, marble,

quartzite and meta-ultramafic rocks. The eclogite

occurs as sheared lenses, pods and layers ranging

from tens of centimeters to hundreds of meters in

size within foliated ultramafic blocks, metapelites

(garnet-quartz-jadeite gneisses), kyanite quartzites

and marbles. These eclogitic blocks are obviously

tectonic boudins. The age of UHP metamorphism

and the time of collision between the Sino-Korean

and Yangtze blocks is about 210–220 Ma (see Cong,

1996 for summary), as derived from independent

studies including Sm–Nd isochrons (Li et al., 1989,

1992), U–Pb zircon ages (Ames et al., 1993; Li et

al., 1993a,b), 40Ar/Ar39 dating (Hacker and Wang,

1995) and Rb–Sr dating (Cong, 1996). The protolith

of the UHP rocks are considered to be Archean

( = 1.66 Ga, Jahn et al., 1996) or Proterozoic

(f 800 Ma, Li et al., 1993a,b; Jahn et al., 1996).

The present-day crustal structures under the

Dabie–Sulu region are divided into four-layer struc-

tures (i.e., upper, middle, upper lower-crust and

lowermost crust) with an average total thickness of

34 km that is much thinner than the crust (45–70 km)

of western China (Li and Mooney, 1998; Kern et al.,

Fig. 2. Microstructures of the UHP eclogites from Maobei, Donghai, China. (a) Field photograph of compositional banding with alternating

garnet-and omphacite-rich layers. (b) Optical microphotograph of a typical garnetite layer. XZ sections. (c) Optical microphotograph of a coarse-

grained eclogite. Both garnet (Grt) and omphacite (Omp) are flattened and elongated. XZ sections.

S. Ji et al. / Tectonophysics 370 (2003) 49–7652

S. Ji et al. / Tectonophysics 370 (2003) 49–76 53

1999). Both the crustal thinning and exhumation of

the UHP rocks are attributed to an intensive tectonic

extension immediately after the collision between the

Sino-Korean and Yangtze cratons.

P-wave refraction and reflection profiles across the

Sulu UHP terrane from Tangchen to Lianshui (Yang et

al., 1999; Yang and Yu, 2001) have a series of

northwest-dipping reflectors at 0–500, 900–1100

and 3000–4000 m depth. These highly reflective

layers have an average P-wave velocity varying from

6.8 to 7.3 km/s, probably corresponding to a series of

sheared slices of UHP eclogite- and/or garnet lherzo-

lite-bearing felsic gneisses as exposed at the surface.

A high portion of eclogite/lherzolite (30–40 vol.%) is

needed to explain such high P-wave velocities (Kern

et al., 2002).

The studied samples of eclogites were collected

from fresh surface outcrops in stone pits at the drill site

(Maobei, Donghai County, Jiangsu Province) of the

Chinese Continental Scientific Drilling Program

(CCSD). This drill site is located in the southern part

of the Sulu UHP metamorphic terrane, about 30 km

east of the Tanlu fault and approximately 70 km west

of the Yellow sea (Fig. 1). The CCSD drilling started in

July 2001, has reached the depth of 3000 m, and

planned to reach a depth of 5000 m in the year of

2004, penetrating through all the observed seismic

reflectors and high Vp layers within the upper crust

of the region. Petrophysical properties of the studied

coesite-bearing eclogites may provide critical con-

straints on geological interpretations of the seismic

refraction and reflection data obtained from the CCSD,

and further shed light upon the deformation mecha-

nisms of the subducted slab and exhumation of UHP

metamorphic rocks.

Fig. 3. SEM orientation contrast (OC) image (a) and backscattered

electron image (b) of a typical coarse-grained eclogite. XZ sections.

3. Microstructures

3.1. Field, optical and SEM observations

In the field, the UHP eclogites reveal a strong

planar fabric defined by a compositional banding with

alternating garnet- and omphacite-rich layers (Fig. 2a).

These garnet-rich layers, containing more than 85–90

vol.% garnet, can be referred to as garnetite. The

omphacite-rich compositional layers are dominated

by omphacite (>80 vol.%) and can be referred to as

pyroxenite. These layers are 5–80 cm thick. On a

given outcrop, garnetite layers are usually thicker than

pyroxenite layers (Fig. 2a). The stretching lineation is

defined by the elongation of garnet and particularly

omphacite. In thin sections, both garnet and ompha-

cite are flattened and elongated in the eclogites (Fig.

2b and c). The compositional layering (Fig. 2a) is

believed to originate from mass-transfer-related meta-

morphic differentiation.

In thin sections cut parallel to the XZ plane, where

X is parallel to the stretching lineation and Z is nor-

mal to the flattened foliation, garnet grains generally

show a shape of elongate lenses with two smoothly

curved boundaries subparallel to the foliation (Fig. 2b

and c). Intersections of these two boundaries form

two-outward-pointing cusps extending into the folia-

tion plane. Some garnet grains show a well-devel-

oped pinch-and-swell structure. These microstructural

Fig. 4. Grain size distributions of garnet and omphacite in Type-1 (sample 96-MB3C) and Type-2 (sample 96-MB4) eclogites.

S. Ji et al. / Tectonophysics 370 (2003) 49–7654

characteristics provide unambiguous evidence for

garnet plasticity. In the XZ sections, omphacite shows

a mean aspect ratio (2.5–5.0) larger than garnet

(2.0–3.5). In the XY and YZ sections, omphacite also

shows a mean aspect ratio larger than garnet. The

larger aspect ratio of omphacite could be due to a fact

that garnet grains were equidimensional while om-

phacite crystals had a columnar shape prior to de-

Fig. 5. Flinn diagram showing the garnet shape fabrics of Type-1

(solid circle) and Type-2 (open circle) eclogites. X, Y and Z are the

mean crystal lengths in the directions parallel to the stretching

lineation, perpendicular to the stretching lineation and in the

foliation plane and normal to the foliation, respectively.

Fig. 6. SEM orientation contrast (OC) images showing porphyro-

clastic texture of garnetite layers in Type-2 eclogite (sample 96-

MB20D).

S. Ji et al. / Tectonophysics 370 (2003) 49–76 55

formation. It is necessary to note that no strong

deflection of the foliation and layering defined by

flattened omphacite grains has been observed around

garnet crystals (Fig. 2b and c). This fact suggests that

the rheological contrast between garnet and clinopyr-

oxene is much less under the UHP conditions than

those in amphibolite- and granulite-facies mylonites

from the deep crust.

Fig. 7. TEM microphotographs of dislocation substructures in deformed gar

regularly spaced, linear array of straight dislocations. Bright field. Sample 9

polygonal array of dislocations of two (d) or three (c) discrete orientations.

Thin sections which were first polished mechani-

cally and chemically, then coated with carbon and

were examined in a Philips XL30 SEM using a

forescatter detector system to collect orientation con-

trast (OC) images. On these OC images (Fig. 3), grain

boundaries and fractures can be clearly distinguished.

Internal structure such as cellular subdomains of

different crystallographic orientations (Prior et al.,

net grains. (a and b) Dislocation walls defined by well-organized and

6-MB3 (Type-1 eclogite). (c and d) Dislocation networks defined by

Bright field (c) and dark field (d). Sample 96-MB1 (Type-2 eclogite).

S. Ji et al. / Tectonophysics 370 (2003) 49–7656

2000) were occasionally detected in some garnet

grains (Fig. 6b).

Based on grain size, two characteristic types of

UHP eclogites can be identified. The first, referred to

as ‘‘Type-1 eclogites’’, is characterized by coarse-

grained garnet and omphacite. Garnet ranges from

0.15 to 1.4 mm with an average grain size of 0.52

mm, and omphacite (0.32–3.0 mm) has a mean grain

size of 0.93 mm (Fig. 4). Measurements made on thin

sections parallel to XY planes show that garnet and

omphacite have average aspect ratios of 2.06 and

2.58, respectively. On YZ sections, omphacite has a

slightly higher average aspect ratio (1.68) than garnet

(1.46). The garnet grains in Type-1 eclogites have thus

prolate shapes with a Flinn coefficient K = 1.4 (Fig. 5).

The second type of eclogites, referred to as ‘‘Type-2

eclogites’’, is characterized by relatively fine-grained

garnet and omphacite (Fig. 4). Garnet grain size

ranges from 0.03 to 1.4 mm with an average grain

size of 0.28 mm, and omphacite (0.03–1.5 mm) has a

mean grain size of 0.41 mm. Garnet and omphacite on

the XY sections have average aspect ratios of 1.4 and

1.78, respectively. On YZ sections, the average aspect

Fig. 8. TEM (bright field) microphotographs of dislocation substructures

(sample 96-MB3, Type-1 eclogite). (b and c) Dislocation junctions (samp

ratio is 1.56 for omphacite and 1.46 for garnet. The

garnet grains are characterized by quasi-plane-strain

shape with K = 1 in Flinn diagram (Fig. 5). Thus,

Type-1 and Type-2 eclogites can be considered as

LHS and LS tectonites (Helmstaedt et al., 1972),

respectively. The LHS tectonites deformed in the

field of apparent constriction while the LS tectonites

deformed in the field of plane strain and probably

indicates a simple shear deformation (Philippot and

van Roermund, 1992; Godard and van Roermund,

1995; Abalos, 1997; Zulauf, 1997).

Both types of eclogites consist of the same

assemblage of garnet (57F 6 vol.%), omphacite

(f 39F 10 vol.%), rutile (f 1–3%) and phengite

(f 0.5%), and minor amounts (less than 3 vol.%) of

secondary minerals such as amphibole and epidote.

Rutile occurs either as interstitial grains between

garnet and omphacite or as inclusions in these min-

erals. Coesite relicts and their quartz pseudomorphs

occur as inclusions in garnet and omphacite in Type-1

eclogites. Radial extensional fractures from these

inclusions are found in both garnet and omphacite.

Some omphacite grains are rimmed by a very fine-

in deformed garnet grains. (a) Free dislocations within a subgrain

le 96-MB1, Type-2 eclogite).

S. Ji et al. / Tectonophysics 370 (2003) 49–76 57

grained symplectite of plagioclase and Ca-pyroxene

or omphacite with a lower jadeite content. Amphibole

and epidote are the retrograde products which are

restricted to some fractures. Both types of eclogites

are pervasively cracked by extensional fractures

which are predominantly perpendicular to the stretch-

ing lineation (Figs. 2 and 3). These lineation-normal

fractures are very common in granulite facies mylon-

ites and UHP metamorphic rocks and thought to be

formed during the latest stages of the exhumation (Ji

et al., 1997).

When both types of eclogites coexist in the same

blocks, the coarse-grained Type-1 eclogites are

bounded by the fine-grained Type-2 eclogite layers.

The latter represent localized shear zones which over-

printed on the early coarse-grained Type-1 eclogite.

The shear zones are approximately parallel to the

regional penetrative mylonitic foliation in the strongly

deformed wall rocks such as garnet-bearing ultramafic

rocks, UHP paragneiss, quartzite and marble. Fore-

scatter orientation contrast imaging showed that the

Table 1

Bulk compositiona of representative eclogites from Maobei, Donghai Cou

Sample Lithology SiO2 TiO2 Al2O3 FeO+

Fe2O3

Mn

96-MB1B Type-2 eclogite 38.34 3.77 18.74 23.06 0.3

96-MB2 Type-1 eclogite 43.51 0.06 22.01 9.02 0.1

96-MB3C Type-1 eclogite 47.46 1.45 17.97 13.54 0.2

96-MB4 Type-2 eclogite 48.66 0.55 17.88 6.65 0.1

96-MB20D Type-2 eclogite 44.48 0.21 17.74 10.38 0.1

01-MB20 Type-1 eclogite 43.05 0.08 21.23 9.89 0.1

01-MB23 Type-1 eclogite 44.75 0.23 21.92 9.69 0.1

01-MB25 Type-1 eclogite 44.86 0.42 20.33 9.82 0.1

01-MB26 Type-1 eclogite 42.91 0.37 22.11 11.29 0.1

01-MB27 Type-1 eclogite 46.65 0.49 18.53 8.19 0.1

01-MB31 Type-2 eclogite 45.98 1.32 16.64 13.41 0.2

01-MB35 Type-2 eclogite 46.01 1.73 15.91 13.79 0.2

MB98-02 Banded eclogite 43.57

MB98-03 Omp-rich eclogite 46.17

MB98-04 Type-2 eclogite 43.99

MB98-08 Retrograde eclogite 44.34

MB98-19 Garnetite 43.32

D12 Type-1 eclogite 48.65 0.85 19.06 9.03 0.1

a Major oxides are expressed as wt.%.

garnetite layers from Type-2 eclogites developed

porphyroclastic texture with large porphyroclasts en-

closed in a fine-grained matrix (Fig. 6a). The latter

developed a foam structure where grain boundaries

are smoothly curved and aligned 120j triple junctions

(Fig. 6b). No coesite relicts were observed in either

garnet or omphacite from Type-2 eclogites, and quartz

occurs as elongated polycrystalline aggregates be-

tween fine-grained garnet/garnet, garnet/omphacite

or omphacite/omphacite grains. It is thus likely that

Type-2 eclogites resulted from dynamic recrystalliza-

tion of Type-1 eclogites along shear zones which were

active during the exhumation of the UHP metamor-

phic rocks.

3.2. TEM observations

A Philips CM200 transmission electron micro-

scope (TEM, GFZ-Potsdam) operating at 200 kV

was used to characterize the dislocation microstruc-

tures in the eclogites. Dislocation microstructures are

nty, Jiangsu Province

O MgO CaO Na2O K2O P2O5 Total Source

4.2 11.22 1.15 0.02 0.02 100.82 This study

3 12.86 11.57 1.21 0.05 0.03 100.45 This study

6.34 11.03 2.71 0.04 0.15 100.89 This study

2 8.1 15.61 1.48 0.05 0.09 99.19 This study

5 17.41 8.47 0.85 0.12 0.22 100.03 This study

5 14.61 10.11 0.85 0.02 0.03 100.02 This study

5 9.39 11.82 1.95 0.12 0.02 100.04 This study

4 10.87 12.14 2.01 0.07 0.02 100.68 This study

7 11.02 11.56 1.13 0.1 0.03 100.69 This study

1 10.73 12.6 2.66 0.16 0.02 100.14 This study

8 10.26 9.55 3.12 0.01 0.2 100.77 This study

7 10.2 9.42 3.13 0.02 0.08 100.56 This study

Kern et al.,

2002

Kern et al.,

2002

Kern et al.,

2002

Kern et al.,

2002

Kern et al.,

2002

4 8 12.57 0.88 0.19 0.25 99.62 Zhang et al.,

1994

S. Ji et al. / Tectonophysics 370 (2003) 49–7658

similar in garnets from both types of eclogites.

Garnets developed extensively regular dislocation

arrays and dislocation networks. The dislocation

arrays (Fig. 7a and b) can be interpreted as tilt

subgrain boundaries while the dislocation networks

(Fig. 7c and d) imply an activation of more than two

active slip systems. Those well-organized dislocation

microstructures indicate that diffusion-assisted recov-

ery mechanism such as dislocation climb and cross-

Fig. 9. Compositional profiles of pyrope (Prp), grossular (Grs), almandine (

lenticular garnets. (a) Sample 96-MB3C (Type-1 eclogite). (b) Sample 96-M

imageries.

slip efficiently occurred. Within the subgrains sur-

rounded by dislocation walls, free dislocations were

imaged with a density in the range from 1011 to 1012

m� 2 (Fig. 8a). Dislocation junctions, which are

complex intersections (tangles) of several disloca-

tions, were also observed in both types of eclogites

(Fig. 8b and c). The presence of dislocation junctions

indicates that dislocation interactions and conse-

quently the operation of multiple slip systems took

Alm) and spessartine (Sps) contents along two orthogonal profiles of

B4 (Type-2 eclogite). Sketch maps based on backscattered electron

S. Ji et al. / Tectonophysics 370 (2003) 49–76 59

place in the deformed garnet (Voegele et al., 1998;

Wang and Ji, 1999). The above observations agree

with previous TEM investigations on other naturally

deformed garnets (Ando et al., 1993; Ji and Mar-

tignole, 1994; Doukhan et al., 1994; Voegele et al.,

1998; Prior et al., 2000), all suggesting that the

garnet can be deformed plastically as long as tem-

perature, pressure, differential stress and strain rate

are appropriate.

4. Chemical compositions and the metamorphic

temperature

Major elements of 27 UHP eclogites collected from

the region were determined by the X-ray fluorescence

(XRF) in McGill University. The results of 12 repre-

sentative samples are listed in Table 1. Analytical

precision is better than F 1%. The UHP eclogites

Fig. 10. Compositional profiles (b and c) of pyrope, grossular, almandine

garnet porphyroclast (a) from Type-2 eclogite.

have significantly lower contents of K2O, Na2O and

SiO2 compared with typical basalts, indicating that

these relatively mobile components have immigrated

into fluids during the eclogitization. The above chem-

ical composition also reflects a fact that the UHP

eclogites, which have subjected to little retrograde

metamorphism, are mainly bimineralic aggregates of

garnet and omphacite and contains < 1–2% coesite

and/or quartz.

Chemical compositions of garnet and omphacite

were measured using a JEOL 8700 Super-probe,

with an accelerating voltage of 15 kV, a beam

current of 20 nA and a beam diameter of 5 Am.

Each estimation of elements was performed with 20 s

counting time on peaks and 10 s on background. The

precision in the chemical compositions is about

F 1%. Garnet composition varies from sample to

sample; such differences are related to differences in

bulk composition of protoliths. In one garnet-rich

and spessartine contents along profiles AB and CD of an elongated

Fig. 11. Plots of the KD values [Fe2 +/Mg)gt/(Fe2 +/Mg)px] of eclogite

samples versus the grossular content [XCa =Ca/(Ca + Fe +Mg+

Mn)] of garnet. Also shown are the 3.0 GPa isotherms of Ellis and

Green (1979, continuous lines) and Powell (1985, broken lines).

S. Ji et al. / Tectonophysics 370 (2003) 49–7660

layer and in one textural situation in the sample scale,

the garnet grains generally have nearly constant

compositions, varying only by F 3%. For example,

microprobe analyses of 209 garnet grains from sam-

ple 96-MB3C, which is a typical Type-1 eclogite,

yield the ranges of pyrope (38.7–44.5%), grossular

(30.7–36.0%), almandine (22.6–25.6%) and spes-

sartine (0.3–0.7%) with an average composition

Alm24.0F 0.5Prp41.7F 1.1Grs33.7F 1.1Spe0.5F 0.1.

Four typical lenticular garnet grains with length/

width ratios larger than 2 were selected from Type-1

eclogite for compositional X-ray maps of Mg, Ca and

Fe using the electron microprobe. These maps (not

illustrated here) show that each lenticular crystal is

rather homogeneous in compositions and do not

show any clear compositional zoning. Zoning pro-

files from rim to rim through the core were also

carried out along two mutually perpendicular direc-

tions which coincide with the major and minor axes

of several elongated garnet grains in the XZ sections.

These profiles (Fig. 9a) show only slight composi-

tional variations with respect to the uncertainties of

measurements, providing no evidence for the exis-

tence of compositional zoning in garnet grains from

Type-1 eclogites. The absence of compositional zon-

ing in the garnet grains from Type-1 eclogites can be

attributed to homogenization by diffusion at peak

metamorphic conditions (T>800 jC and P>3 GPa,

Zhang et al., 1994) and to the absence of effective

retrogression from the eclogite to granulite or am-

phibolite facies.

In Type-2 eclogites, most of garnet porphyroclasts

do not show significant compositional zoning (Fig.

9b) with some rare exceptions (Fig. 10). When the

compositional zoning occurs, it is characterized by a

relatively low pyrope component at both the center

and the rims while the variation of grossular compo-

nent is reverse (Fig. 10). Higher pyrope component

and lower grossular component occur between the

center and the rims. The almandine and spessartine

components remain generally constant. The zoning

pattern can be interpreted as growth zoning during the

exhumation of the UHP metamorphic terrane. Rapid

exhumation and resultant strain localization might

result in grain boundary migration of garnet and

formation of the compositional zones with distinct

pyrope and grossular components. The radial fractur-

ing around coesite inclusions in Type-1 eclogites and

recrystallization of Type-2 eclogites were probably

contemporaneous with this tectonic uplift. However,

omphacite grains are nearly homogeneous with jadeite

components of Xjd = 0.25–0.60 and show no specific

compositional distinction between these two types of

eclogites.

Equilibrium temperatures were estimated from the

empirical relationship between Ca-content in garnet

[XCa =Ca/(Ca + Fe +Mg+Mn)] and Fe–Mg partition-

ing (KD) between coexisting homogeneous garnet and

omphacite. KD is defined as the ratio of Fe2 +/Mg for

garnet to that for omphacite. Since the existence of

coesite pseudomorphs implies that the garnet and

omphacite assemblage was stable in the coesite sta-

bility field, the peak temperature can be estimated at a

pressure of 3 GPa using the garnet-clinopyroxene

geothermometers of Ellis and Green (1979) and

Powell (1985). These geothermometers generally give

reasonable temperature estimates with garnets of low

to medium grossular components (XGrs < 0.5, Koziol

and Newton, 1989). As shown in Fig. 11, the results

suggest that both types of the eclogites from Maobei

drilling site were equilibrated at the similar tempera-

ture range of 850F 50 jC, consistent with previous

Fig. 12. Garnet compositional variations in an old porphyroclast (profiles AB and CD) and the adjacent matrix of recrystallized new grains

(profiles EF and GH). Type-2 eclogite (sample 96-MB20D).

S. Ji et al. / Tectonophysics 370 (2003) 49–76 61

Fig. 13. Histogram showing density distributions of Type-1 and

Type-2 eclogites.

S. Ji et al. / Tectonophysics 370 (2003) 49–7662

estimates for eclogites from other localities of the Sulu

regions (Zhang et al., 1994).

Compositional variations of garnet relict porphyr-

oclasts and polygonal recrystallized grains were

investigated in sample 96-MB20D, which is a typ-

ical Type-2 eclogite. The results (Fig. 12) are

presented in terms of pyrope, grossular, almandine

and spessartine components. The line drawing (Fig.

12a), which was based on an electron backscattered

image, shows a typical site chosen for microprobe

measurements. The porphyroclast has a serrated

boundary, which is interpreted as the result of grain

boundary migration. The recrystallized matrix sur-

rounding the porphyroclast is characterized by a

mosaic of polygonal fine grains. The pyrope, gros-

sular, almandine and spessartine components display

no variations within the range of uncertainties of

microprobe measurements along two orthogonal

profiles AB and CD in the garnet porphyroclast

(Fig. 12b and c). Along profiles EF and GH, which

started from the rims of the porphyroclast and then

extended progressively to the recrystallized matrix, a

marked continuous decrease in pyrope component

and an increase in both grossular and almandine

components are detected while spessartine content

remains constant (Fig. 12d and e). The observed

tendency of compositional variations has two impli-

cations: (1) the dynamic recrystallization of garnet

took place during uplift of the UHP metamorphic

terrane with continuous decreasing in T and P; and

(2) the dynamic recrystallization of garnet in the

UHP eclogites, which probably occurred through

grain boundary migration, was driven by not only

dislocation strain energy but also by chemical-free

energy.

5. Density data

Densities of the eclogites were measured using

Archimedes methods at ambient conditions. The

densities obtained vary from 3.46 to 4.10 g/cm3

(Fig. 13). The Type-2 eclogites have an average

density of 3.76 g/cm3, which is higher than that

(3.54 g/cm3) of Type-1 eclogites. These mean den-

sities are consistent with the bulk densities of eclo-

gites from Dabie UHP belt (Kern et al., 1999), but

higher than those compiled by Rudnick and Fountain

(1995) (3.43 g/cm3) and Christensen and Mooney

(1995) (3.48 g/cm3).

6. Crystallographic preferred orientations

CPOs were measured on garnet, omphacite and

rutile using electron backscattered diffraction (EBSD),

JEOL JSM 5600 SEM, at Laboratoire de Tectonophy-

sique, Universite de Montpellier II. The CPO dia-

grams were projected in equal area and lower

hemisphere respect to the X, Y and Z directions of

the tectonic framework (Fig. 14).

6.1. Garnet CPO

Each EBSD measurement with more than two

hundreds of garnet grains shows that garnets from

both types of eclogites have weak strengths and

complex CPO patterns with numerous maxima of

h100i, h111i and h110i (Fig. 14). Bulk patterns of

those CPOs cannot be interpreted simply by assum-

ing the activation of any single slip system. The

complex patterns may result from multiple slip of

many systems such as those of 1/2 h111i {110},

{112} or {123}, h100i {010} and h100i {011}.

Activation of these slip systems has been documented

by previous deformation experiments of garnet single

crystals (Garem et al., 1982; Rabier et al., 1976;

Karato et al., 1995; Voegele et al., 1998; Wang and

Ji, 1999). We are not surprised by observing such

weak and complex CPOs for garnets because for a

Fig. 14. Stereographic projections (lower hemisphere equal area) of crystallographic preferred orientations of garnet, omphacite and rutile,

determined by EBSD system. X, Y and Z are defined in the text.

S. Ji et al. / Tectonophysics 370 (2003) 49–76 63

S. Ji et al. / Tectonophysics 370 (2003) 49–7664

given finite strain produced by dislocation slip,

crystals with more active independent slip systems

should develop more complex and weaker CPOs than

those with only single or two active slip systems.

During slip, an unconstrained crystal with five or

more independent slip systems changes its shape but

its lattice may subject to little rotation with respect to

the instantaneous stretching axes of bulk flow. Unlike

triclinic minerals such as plagioclase (Ji and Main-

price, 1988), the bcc garnet has high symmetry and

12 potential slip systems, there is a wide choice of

slip systems. The slip plane does not have to undergo

much rotation before the resolved shear stress

becomes high on another 1/2 h111i {110} slip

system, for example. Even though the slip occurs

predominantly on the {110} planes, it is important to

realize that three {110}-type planes intersect in a

[111] direction and screw dislocations with a 1/2

Fig. 15. Misorientation angle distributions (a) and pole figures of the miso

h111i Burgers vector may move at random on the

{111} planes with high resolved shear stress. Thus,

the weakness of the garnet LPOs cannot provide any

unequivocal evidence for diffusion creep or against

dislocation creep as a deformation mechanism of the

garnet grains in the UHP eclogites (Ji and Mar-

tignole, 1996).

Of note are the data of crystallographic misorien-

tation between two adjacent grains, which may pro-

vide some information for the prevailing deformation

mechanism (Mainprice et al., 1993; Jiang et al.,

2000). The distributions of minimum misorientation

(some times called disorientation) angles and rotation

axes between adjacent garnet grains are presented in

Fig. 15a and b, respectively. There is no clear

distinction between Type-1 and Type-2 eclogites. In

both types of eclogites, the rotation axes are almost

randomly distributed, with no special correlation with

rientation axes (b) of garnet grains in Type-1 and Type-2 eclogites.

S. Ji et al. / Tectonophysics 370 (2003) 49–76 65

the tectonic framework (X–Y–Z axes). The histo-

grams shown in Fig. 15a display a maximum fre-

quency at a disorientation angle of about 45j. For arandom distribution of cubic crystals such as garnet,

the distribution of disorientations has a maximum

peak at exactly 45j (Mackenzie, 1958). Hence, it is

further argument to say that these garnets have a very

weak CPO.

6.2. Omphacite

Both types of eclogites developed strong CPOs

(Fig. 14) with similar preferred orientations of [001],

which is subparallel to the stretching lineation (X).

Both [100] and [010] direction forms girdles perpen-

dicular to the lineation. However, the maximum con-

centrations of the [010] are different for Type-1 and

Type-2 eclogites. The [010] maxima are close to the

Z-direction for Type-1 eclogite while they are located

middle way between the Y and Z directions in the

plane normal to the lineation for Type-2 eclogite. The

omphacite CPO pattern of Type-1 eclogite is very

similar to those of so-called L-type eclogites subjected

to a constriction or coaxial extensional strain (e.g.,

Godard and van Roermund, 1995; Abalos, 1997;

Mauler et al., 2000). The omphacite CPO pattern of

Type-2 eclogite is similar to those observed by Abalos

(1997). Both patterns of the omphacite CPOs have

been reported for eclogites from Western Norway

(Boundy et al., 1992), Western Alps (Philippot and

van Roermund, 1992; Piepenbreier and Stockhert,

2001), Swedish Caledonids and Western France (God-

ard and van Roermund, 1995), NW Spain (Abalos,

1997) and for eclogites xenoliths from East Corolado

Plateau (Kumazawa et al., 1971).

The omphacite CPOs of eclogites (Fig. 14) can be

readily explained according to the recent viscoplastic

self-consistent (VPSC) numerical simulations of Bas-

cou et al. (2002). The simulations were made using

the slip systems observed by TEM in naturally de-

formed omphacite. Slip has been observed on 1/2

h110i {110}, [001] {110} and [001] (100) by van

Roermund and Boland (1981), Buatier et al. (1991), Ji

et al. (1993) and Godard and van Roermund (1995).

Type-1 eclogite has strong point maximum of [001]

with a density of 11.3 m.u.d. subparallel to the

lineation, a girdle of [010] in the YZ plane with a

maximum of 7.6 m.u.d. normal to the foliation (Z) and

more dispersed distribution of [100] in the YZ plane

with a maximum of 3.9 m.u.d. between Y and Z. Both

simple shear and pure shear simulations have these

general characteristics. Deformation of Type-1 eclo-

gite was probably in the constrictional field given the

prolate shape data for omphacite. Type-2 has stronger

point maximum of [001] with a density of 12.9 m.u.d.

subparallel to the lineation, a weaker girdle of [010] in

the YZ plane with a maximum of 6.0 m.u.d. between Y

and Z and stronger concentation of [100] in the YZ

plane with a maximum of 4.7 m.u.d. near Z. The

transtension simulation (Bascou et al., 2002) has these

general characteristics. The transtension used in the

simulation has a kinematic vorticity number of

Wk = 0.9, that is about 90% simple shear with a small

component of tension parallel to the lineation. The

shape data on omphacite for Type-2 is nearly plane

strain which is entirely compatible with a transtension

deformation.

6.3. Rutile

Rutile CPOs are shown in Fig. 14. [001] directions

of rutile in both types of eclogites are concentrated

near the lineation. As rutile is of tetragonal symmetry

where [100] and [010] axes are crystallographically

equal, those axes show similar patterns. [100] and

[010] in Type-1 eclogite show clear girdles perpen-

dicular to the lineation, whereas both these axes in

Type-2 eclogite show relatively dispersing fabric

patterns. Stoichiometric rutile has two principal glide

systems, h101i {101} is active above 600 jC and

[001]{110} above 900 jC (Blanchin et al., 1990).

However, [001] {110} has a critical resolved shear

stress that is twice as high as h101i {101}, hence is

difficult to activate at laboratory strain rates (Blanchin

and Faisant, 1979). As h101i {101} provides only

four slip systems, rutile is extremely difficult to be

fully plastic below 900 jC. Hence, it is likely that the

observed rutile CPO is due either to anisotropic

crystal growth or to the passive rotation of elongate

crystals in the ductile matrix.

7. Seismic wave properties

We calculated the seismic velocities (Vp is re-

ferred to P-wave velocity, and Vs1 and Vs2 are,

S. Ji et al. / Tectonophysics 370 (2003) 49–7666

respectively, the velocities of the first and second

arrived S-waves), shear-wave anisotropy [A(Vs)] and

polarization direction of Vs1 on the basis of the

CPO, density, volume fraction and elastic stiffness

coefficients of each constituent mineral. The elastic

stiffness coefficients of garnet, omphacite and rutile

are derived from Chai et al. (1997), Bhagat et al.

(1992) and Bass (1995), respectively. The seismic

velocities and anisotropies (Fig. 16) have been cal-

culated for each type of eclogite using the Voigt–

Reuss–Hill average. Both types of eclogites yield

comparable Vp anisotropies (1.4–1.5%) with similar

minimum (8.67–8.70 km/s) Vp and maximum

(8.79–8.84 km/s) Vp values. The maximum Vp

occurs at the direction subparallel to the lineation

while the minimum Vp occurs at a high angle to the

lineation. Kern et al. (2002) carried out laboratory

measurements on three fresh UHP eclogite samples

from the same locality as our samples and obtained

mean Vp values at 600 MPa ranging from 8.48 to

8.64 km/s and Vp anisotropies from 1.0% to 2.0%

(Table 2). Although the calculated and measured Vp

anisotropy values are almost the same, the calculated

zero-pressure Vp values of crack-free and retrogres-

sion-free eclogite aggregates are slightly higher than

those values measured at 600 MPa (Table 2). This

fact indicates that the calculated velocity actually

corresponds to a velocity measured at a pressure

above 600 MPa for eclogites. It is not surprising

because the critical pressure over which all pores and

cracks existent within hard rocks such as mantle

xenoliths and eclogites are closed at room tempera-

ture should be significantly higher than 600 MPa

(e.g., Christensen, 1965; Manghnani et al., 1974;

Kern et al., 1999; Wang and Ji, 2001).

Both types of eclogites have almost identical cal-

culated mean Vs1 (4.98–5.00 km/s) and Vs2 (4.96–

4.97 km/s) values, yielding weak Vs anisotropies

( < 1.5%). The low values of A(Vs) are mainly caused

by the volumical predominance of cubic garnet which

Fig. 16. Seismic properties of typical Type-1 and Type-2 eclogites

from the Sulu UHP metamorphic terrane. All the properties were

calculated for the zero-porosity and ambient conditions. Lower

hemisphere projections. Contours for P-wave velocity (Vp), fast S-

wave velocity (Vs1) and slow S-wave velocity (Vs2) are in km/s, S-

wave velocity anisotropy [A(Vs)] is in %. Vs polarization is

indicated by the trace of the fast S-wave vibration plane.

is elastically quasi-isotropic. The calculated values

(Fig. 16) are slightly higher than those values mea-

sured by Kern et al. (2002) at 600 MPa. They yielded

Table 2

Modal compositions, bulk densities and P- and S-wave velocities (at 600 MPa) of UHP eclogites from Maobei, Donghai, China

Sample Description Modal Composition Density Vp(km/s) A(Vp),

%

Vs(km/s) A(Vs),

%

Reference

MB98-02 Banded eclogite Grt57, Omp42, Rt0.6,

Phn0.2, Opq0.3

3.59 8.57 1.0 4.90 1.1 1

MB98-04 Fine-grained eclogite Grt39, Omp58, Hb1.5,

Phn1.0, Rt0.1, Ep.0.1,

Opq0.3

3.52 8.48 1.1 4.80 2.9 1

MB98-19 Garnet-rich eclogite Grt77, Omp14, Ms5,

Phn3, Qtz1

3.62 8.64 2.0 4.93 1.4 1

96-MB3 Coarse-grained eclogite Grt57.5, Omp39.2,

Rt1.5, Qtz1.8

3.54 8.73 1.4 4.97 1.4 2

96-MB4 Fine-grained eclogite Grt62.4, Omp33.7,

Rt2.9, Hb0.8, Opq0.2

3.76 8.77 1.5 4.99 1.3 2

1: experimental data from Kern et al. (2002); 2: calculated data from this study.

Abbreviations: Grt, garnet; Omp, omphacite; Rt, rutile; Phn, phengite; Hb, hornblende; Qtz, quartz; Ms, muscovite; Ep, epidote; Opq, opaques.

S. Ji et al. / Tectonophysics 370 (2003) 49–76 67

the eclogites mean Vs values ranging from 4.80 to 4.93

km/s at this pressure (Table 2). Vs1 fast polarization

directions, as shown in Fig. 16, are similar for Type-1

and Type-2 eclogites except at the Y-direction. The fast

polarization directions are subparallel to the X-direc-

tion for rays parallel to the XY-plane in Type-2 eclogite,

whereas those are oblique to the X-direction for rays

parallel to the Y-direction in Type-1 eclogite.

Fig. 17. Tentative P–T paths for Type-1 and Type-2 eclogites.

Type-1 eclogite deformed under the peak metamorphic conditions

within the wide subduction shear zone at the greater depth while

Type-2 eclogite resulted from reworking of Type-1 eclogite during

the exhumation along some localized narrow shear zones.

8. Discussion

8.1. Transition between constriction and plane strain

In the coarse-grained Type-1 eclogites, both garnet

and omphacite are of prolate shape (LHS), indicating

an apparent constrictional strain (e.g., Godard and van

Roermund, 1995; Zulauf, 1997). In the fine-grained

Type-2 eclogites, however, shapes of either garnet or

omphacite display a plane-strain (LS) which was most

likely related to a non-coaxial shear (e.g., Philippot

and van Roermund, 1992; Abalos, 1997) or trans-

tension as suggested here. It is interesting to discuss

the possible origin of these two types of eclogites.

Field observations show that the deformation fab-

rics of coarse-grained Type-1 eclogite and fine-grained

Type-2 eclogite were not coeval: the early coaxial

fabrics (LHS) were overprinted by the late simple

shear (LS) along the localized shear zones. This feature

implies that the variable microstructures of the UHP

eclogites were not due to heterogeneous and strongly

partitioned deformation—a mechanism proposed by

Henry et al. (1993) for the UHP rocks from the Dora–

Maira massif, Western Alps, Italy. Thus, the distinct

deformation fabrics of the Sulu UHP eclogites should

be formed at different periods of tectonic history and

most likely at different depths (Fig. 17).

S. Ji et al. / Tectonophysics 370 (2003) 49–7668

Systematic analyses of seismic focal mechanisms

led geophysicists to recognize three tectonic strain

regimes along subducting slabs (Isacks and Molnar,

1969, 1971): simple shear at shallow depths ( < 100

km), down-dip extension at intermediate depths

(100–400 km) and down-dip compression at deep

upper mantle depths (400–660 km). The boundaries

between these regimes correspond approximately to

those of phase transformations. At depths greater than

400 km, olivine of the subducting slab is transformed

to spinel while clinopyroxene progressively dissolves

into the garnet structure and finally eclogite complete-

ly transforms to garnetite. Both spinel and garnetite

are denser than olivine and eclogite, resultant negative

buoyancy causes an additional pulling force on the

subducting slab at 100–400 km depth. As the sub-

ducting slab is less dense than perovskite-dominant

lower mantle below the 660 km discontinuity, the

resultant buoyancy will resist against penetration of

the slab into the lower mantle. As a result, the

subducting slab is subject to down-dip compression

at 400–660 km depth. The non-coaxial strain regime

is prevailing along the boundary thrust shear zone

between the subducting slab and its overlying litho-

sphere at depth of the slab bend ( < 100 km). Although

eclogitization of crustal rocks such as amphibolite,

anorthosite and granulite-facies gneiss can start from a

depth of f 40 km, the transformation may not be

fully completed until a depth of f 80 km if the

materials are relatively dry (e.g., Austrheim, 1987;

Andersen et al., 1991).

Type-1 eclogites recorded an apparent constric-

tional strain which could be driven by a slab-pull

mechanism at the depth deeper than the slab bend

(f 100 km), whereas Type-2 eclogites recorded a

plane, non-coaxial strain, predominantly simple shear

which could take place a depths shallower than 100

km deep. This also reasonably explains why both

garnet and omphacite are coarser in Type-1 eclogites

than in Type-2 eclogites and why inclusions of coesite

pseudomorphs surrounded by radial fractures are

observed in Type-1 eclogites rather than in the

recrystallized matrix of Type-2 eclogites. If the above

conclusion is correct, Type-2 eclogites might be

reworked and recrystallized Type-1 eclogites from

shear zones resulted from strain localization during

the uplift of the UHP metamorphic rocks, possibly in

extension as indicated by the transtension interpreta-

tion of omphacite CPOs. It is thus reasonable to

assume that Type-1 eclogite was equilibrated above

the critical pressure for the quartz/coesite transforma-

tion while Type-2 eclogite was equilibrated near or

below this pressure. As the deformation temperatures

estimated from geothermometric mineral assemblages

for both types of eclogites are almost the same (Fig.

11) within the certainties of the geothermometers, the

initial uplift of the UHP metamorphic terrane ap-

peared to be nearly isothermal (Okay, 1993; Zhang

et al., 1994; Wang and Cong, 1999). The existence of

some zoning garnet grains from Type-2 eclogites may

indicate depressurization of these rocks. Furthermore,

the transition from plane, non-coaxial strain (L–S) to

constrictional strain (LHS) within the subducting

slab is believed to occur approximately at the depth

of the transformation of quartz to coesite (f 100 km).

This fabric transition depth, which may depend on

tectonic setting and geothermal gradient, varies from

region to region and may be as shallow as about

40 km depth (e.g., in central European Variscides:

Zulauf, 1997; south Norwegian Caledonides: Ander-

sen et al., 1991).

Previous TEM investigations of the bcc garnets

(e.g., Voegele et al., 1998) revealed that the disloca-

tion slip occurs on several planes, {110}, {111},

{112} and {123} but predominantly in the close-

packed h111i direction. The h111i dislocations are

extensively dissociated into two partials 1/2 h111i,between which a stacking fault is produced. An

extended screw dislocation cannot cross slip unless

the partial dislocations recombine into a perfect dis-

location. The greater the width of the stacking fault,

the difficult it is to produce constrictions in the

stacking fault. An increase in hydrostatic pressure will

tend to reduce partial dislocation separation and in-

crease the possibility of cross-slip (Poirier and Ver-

gobbi, 1978; Montardi and Mainprice, 1987). Because

recovery and recrystallization are two antagonist

(competing) processes, a decrease in hydrostatic pres-

sure due to tectonic exhumation will reduce the ability

of dislocations to cross-slip so that the dislocation

density rises and recrystallization occurs. This may

explain why recovery-accommodated creep is preva-

lent in garnet from Type-1 eclogites which deformed

likely at higher P (= 3.2 GPa), while recrystallization

occurs in garnet from Type-2 eclogite which deformed

likely at lower (P= 2.8 GPa, Fig. 17).

hysics 370 (2003) 49–76 69

8.2. Rheological contrast between garnet and om-

phacite

The studied UHP eclogites are typically two-phase

rocks that are mainly composed of garnet and ompha-

cite. Microstructural observations suggested that the

eclogites can be plastically deformed and are not as

rigid as thought previously. Moreover, plastic flow of

eclogites under UHP conditions may effectively fa-

cilitate channeled flow within the shear zone between

two lithospheric plates. The enhanced counterflow

within a wedge-shaped flow channel can force the

UHP metamorphic rocks to be rapidly exhumed. The

rapid exhumation in turn precludes effective heat

transfer from the surrounding mantle to the wedge,

rendering a cool thermal structure within the sub-

ducted crustal materials (5–10 jC/km, Fig. 17). This

makes the subducted crust possible to preserve the

mineralogic assemblages and deformation microstruc-

tures formed at UHP metamorphic conditions.

Our microstructural observations suggest that the

rheological contrast between garnet and omphacite

under the UHP conditions is lower than that com-

monly observed between garnet and clinopyroxene in

typical deep crustal rocks such as amphibolite or

granulite facies metamorphic rocks (e.g., Ji et al.,

1993). Garnet is most likely several times rather than

several orders stronger than omphacite in UHP eclo-

gites where both garnet and omphacite plastically

deform.

Using a Griggs apparatus, Jin et al. (2001) de-

formed hot-pressed quartz eclogite (50% garnet, 40%

omphacite and 10% quartz), garnet aggregate and

omphacite aggregate. They found that ‘‘the flow

properties of eclogite are very similar to those of

peridotite’’. However, garnet grains in their deformed

samples are extensively fractured rather than flattened,

indicating little plastic deformation of garnet. Eclo-

gites from the Sulu UHP metamorphic terrane show

remarkable evidence of plastic deformation: well-

developed foliation, pronounced stretching lineation

and deformation-metamorphism-induced, alternating

garnet-and omphacite-rich compositional layers (Fig.

2). Both garnet and omphacite grains are flattened and

elongated in the eclogites (Fig. 3). TEM observations

of deformed garnets from the UHP eclogites (Figs. 7

and 8) display well-organized dislocations and regu-

larly spaced subgrain boundaries. EBSD measure-

S. Ji et al. / Tectonop

ments demonstrated that garnet developed complex

CPOs. Furthermore, recrystallization-induced por-

phyroclastic texture is observed in Type-2 eclogites

(Fig. 6). The above observations indicate that garnet

grains in Type-1 and Type-2 eclogites deformed

viably by recovery- and recrystallization-accommo-

dated dislocation creep, respectively.

It is well known that the mechanical strength of a

material strongly depends on operating deformation

mechanisms (e.g., Stockhert and Renner, 1998; Wang

and Ji, 2000). For example, quartz is stronger than

feldspar when quartz deforms by dislocation creep

while feldspar by microfracturing and cataclastic flow,

whereas quartz is weaker than feldspar when both

quartz and feldspar deform by dislocation creep

(Dell’Angelo and Tullis, 1996). Jin et al. (2001)

experiments failed to duplicate the typical microstruc-

tures observed optically and with the TEM in natu-

rally deformed UHP eclogites. Deformation mecha-

nisms in their deformed samples are not the same type

as those operated in natural UHP eclogites from

former subducting slabs. Thus, the mechanical

strength data of the eclogite deformed in the non-

steady-state semibrittle regime cannot be reliably

extrapolated to the UHP eclogites deformed plastical-

ly at much lower natural strain-rate and temperature.

The UHP eclogites from the Sulu region contain

< 50% SiO2 (Table 1) and < 1-2% coesite or quartz

pseudomorphs as inclusions in strong minerals such

as garnet. However, the eclogite specimens studied by

Jin et al. (2001) contain of 10 wt.% intergranular

quartz. The quartz, which did not enter in the stability

field of coesite under their experimental conditions, is

the weakest phase in the eclogite samples and takes

the much greater part of the strain with reference to its

volume fraction. The addition of 10 wt.% quartz into

eclogites as intergranular ‘‘lubricant’’ results in con-

siderable weakening. Therefore, the ‘‘flow’’ strength

of quartz eclogite, reported by Jin et al. (2001), cannot

represent that of bimineralic (garnet and omphacite)

eclogites in the subducting slab.

8.3. Geophysical implications

Seismic reflectors have been observed in the upper

mantle beneath many areas of the world, for example,

the Northwest Territories of Canada (Cook et al.,

1999), the Abitibi–Opatica of the Canadian Superior

S. Ji et al. / Tectonophysics 370 (2003) 49–7670

Province (Calvert et al., 1995) and the north coast of

Scotland (Warner and McGeary, 1987; Warner et al.,

1996). These mantle reflectors are generally gently

dipping and extend from the Moho to at least 110 km

depth. The mantle reflectors display the following

characteristics: (1) Reflection coefficients (Rc) higher

than around 0.1 are needed to cause such reflections

within the upper mantle overlaid by the 40-km-thick

crust (Warner and McGeary, 1987). (2) Seismic

reflectors represent the sharp boundaries (Morgan et

al., 1994; Bostock, 1997). (3) Because seismic re-

flection methods have a vertical and horizontal reso-

lution of about 75 and 3000 m, respectively, at depths

of 40 km (assuming average crustal velocities of 6

km/s and a dominant frequency of 20 Hz), each

mantle reflector must be considered as a zone com-

posed of layered rocks with high impedance contrast.

Such a zone should be regionally extensive, have a

thickness of several hundreds meters to several kilo-

meters and have a sharp, well-defined, continuous

upper surface.

Several models, as discussed below, have been

proposed to interpret the origin of mantle reflections.

(1) Ductile shear zone. The deformed peridotites in a

mantle shear zone have developed strong CPOs of

olivine and pyroxenes and are thus anisotropic,

while the wall rocks are undeformed, have no

CPO and therefore isotropic. The boundary

between deformed and undeformed rocks can be

seismically reflective if the resulting reflection

coefficient is large enough (i.e., Rc >0.5–0.10).

Previous investigations (e.g., Kern et al., 1996; Ji

et al., 1994; Ben Ismaıl and Mainprice, 1998)

show that most of peridotites have a CPO-induced

P-wave anisotropy < 7% with an average value

of f 4%. A simple calculation, assuming that

both deformed and undeformed peridotites have

the same density, can readily demonstrate that the

reflection coefficient related to such an anisotropy

is generally < 0.03. Therefore, the boundary

between sheared (anisotropic) peridotites and

their undeformed (isotropic) protoliths cannot be

a candidate for the observed mantle reflections.

(2) Lithological interfaces between typical mantle

rocks. The presence of water-rich fluids may

cause the upper mantle partially melted. The melts

may crystallize into wehrlite, websterite and

pyroxenite, leaving a residue (harzburgite and

dunite) from partial melting. Although these

lithological units can be interlayered and suffi-

ciently thick (up to several hundred meters), the

boundaries between them will have low reflectiv-

ity for the following reasons: (a) Dunite and

pyroxenite have almost same density although

they are different in average P-wave velocity: 8.40

km/s for dunite and 7.94 km/s for pyroxenite (at a

confining pressure equivalent to 40 km depth,

Christensen and Mooney, 1995). The reflection

coefficient between these two rocks is thus

< 0.04. (b) The wehrlite, websterite and pyrox-

enite occur commonly as subvertical dykes or

veins (Wilshire and Kirby, 1989) which are

difficult to be imaged at mantle depths using

conventional reflection techniques. (c) Contacts

between the lithologic units may not be sharp

enough to be seismically reflective. Furthermore,

spinel lherzolite and garnet harzburgite have

almost the same density, same P-wave velocity

and same anisotropy. The reflection coefficient

between these two rocks is < 0.01. Thus, the

transition boundary from spinel to garnet in

peridotite should not be as reflective, as suggested

by O’Reilly and Griffin (1985).

(3) Transition zone between dislocation and diffusion

creep. Dislocation creep is considered to produce

CPO and thus anisotropy while neither diffusion

nor superplastic process can form CPO and

anisotropy (e.g., Karato and Wu, 1993). However,

the transition zone between the dislocation and

diffusion creep regimes does not cause the strong

reflectivity for the following reasons: (a) typical

maximum seismic anisotropies formed by dis-

location creep in the mantle rocks are f 3–5%

(Mainprice and Silver, 1993; Ji et al., 1994; Kern

et al., 1996; Ben Ismaıl and Mainprice, 1998;

Saruwatari et al., 2001). Even though the

assumption that no CPO is formed in the diffusion

regime is fully correct, the boundary between the

two regimes has a reflection coefficient < 0.02.

(b) Because the average grain size of olivine is

commonly larger than 0.5 mm, as demonstrated

by upper mantle xenoliths, dislocation creep

should be dominated over a thickness of at least

250–300 km (Ji et al., 1994). The transition zone

between dislocation and diffusion creep regimes

Fig. 18. Contents of garnet in HP (high pressure) and UHP

(ultrahigh pressure) eclogites. Data compiled from Birch (1960),

Boundy et al. (1992), Kern and Richter (1981), Kern et al. (1999,

2002), Mauler et al. (2000), Manghnani et al. (1974) and this study.

See Ji et al. (2002) for summary.

S. Ji et al. / Tectonophysics 370 (2003) 49–76 71

should not appear in the depth range of upper

mantle reflectors. (c) The transition zone between

dislocation and diffusion or superplastic creep

regimes may not be sharp enough to produce the

interface transition widths < 200 m required by

the seismic reflection data (Bostock, 1997).

(4) Metasomatized shear zones. Mantle shear zones

developed parallel to a subducting slab in the

mantle wedge above the subducting slab would

provide a favorable conduit for fluids to migrate

upwards. Under high temperature and high

pressure conditions, the fluids react with rocks,

forming seismically slower, less dense hydrated

minerals such as phlogopite and amphibole (e.g.,

Francis, 1976). If the concentration of these

hydrated minerals is higher than about 40% in

metasomatized shear zones, the boundary be-

tween the metasomatized shear zones and their

wall rock has impedance contrasts large enough to

be seismic reflective (Warner and McGeary, 1987;

Calvert et al., 1995). Then the metasomatized

shear zones should be low velocity zones.

However, studies of Morgan et al. (1994), Warner

et al. (1996) and Bostock (1997) suggest that the

mantle reflectors are most likely the faster and

denser layers embedded in the slower and less

denser rocks. Moreover, evidence from mantle

xenoliths (e.g., Francis, 1976) shows that the

concentration of hydrated minerals in metasom-

atized zones is typically only several percent and

much lower than the critical value needed for

causing mantle reflections. All the facts described

above exclude metasomatized shear zones from

the common candidates for the origin of mantle

reflectors.

(5) Boundary between eclogite and peridotite. Pre-

vious researchers (Morgan et al., 1994; Warner et

al., 1996; Calvert et al., 1995) believe that mantle

reflectors represent relict fragments of eclogitic

oceanic crust embedded within the subducted

lithosphere. However, this assumption has not

been fully supported by published petrophysical

data (Ji et al., 2002). According to Christensen

and Mooney (1995), laboratory measurements of

54 mafic eclogite samples at a confining pressure

equivalent to 40 km depth yield an average

density of 3.51 g/cm3 and an average P-wave

velocity of 8.19 km/s, while measurements of 51

peridotite samples under the same conditions gave

an average density of 3.33 and an average Vp of

8.40 km/s. These average values give a reflection

coefficient < 0.02 for the boundary between

eclogite and peridotite, making it unlikely a

candidate for the strong mantle reflectors. Thus,

Fountain and Christensen (1989), Gubbins et al.

(1994), Hynes and Snyder (1995) concluded that

eclogites transformed from mafic rocks are

indistinguishable in both densities and velocities

from peridotitic upper mantle.

S. Ji et al. / Tectonophysics 370 (2003) 49–7672

Eclogites are generally classified into high pres-

sure (HP) and ultrahigh pressure (UHP) ones that

are separated by the quartz-coesite equilibrium (e.g.,

Cong, 1996). The HP and UHP eclogites are also

called ‘‘cold eclogite’’ and ‘‘hot eclogite’’ (Chopin

et al., 1991), respectively. The UHP eclogites (Kern

et al., 1999; Kern et al., 2002; this study) have a

higher mean density and a higher mean Vp than the

HP eclogites (e.g., Fountain et al., 1994; Mauler et

al., 2000) because the first have higher contents of

garnet (Fig. 18) and lower contents of quartz,

amphibole, zoisite and mica than the latter. Thus,

mafic UHP eclogites are believed to be denser and

faster in Vp than typical peridotites while HP

eclogites have similar densities and velocities as the

peridotites.

Field and laboratory studies on lithology, defor-

mation structures and seismic properties of the

Dabie–Sulu UHP metamorphic terrane can provide

some important hints for understanding the origin of

mantle reflections. As stated above, the metamor-

phic terrane consists of interlayered pelitic and felsic

gneisses, quartzite, eclogite, garnet-peridotite and mar-

ble. All these rocks contain coesite as inclusions in

garnet, omphacite, jadeite and zoisite (Wang et al.,

1993; Cong, 1996; Zhang and Liou, 1996; Wang and

Cong, 1999). This suggests that all these rocks, which

could represent continental shelf limestone, quartzo-

feldspathic sandstone (grewacke) and mafic volano-

clastic sediments, were subducted into the upper

mantle at great depths (>90–100 km) and then tecton-

ically exhumed to the earth’s surface, during the

continental collision and post-collision extension. Gar-

net peridotite, on the other hand, may be from the

mantle wedge immediately overriding the subducting

slab.

The UHP eclogites are tectonic boudins since they

occur as pods, layers and blocks, ranging in size

from tens of centimeters to hundreds of meters,

within garnet-peridotite, metapelites (garnet-quartz-

jadeite gneisses), coesite-bearing quartzite and mar-

ble. All the rocks show well-developed foliation,

stretching lineation and compositional layering.

These foliations or compositional layers are generally

subparallel in ultramafic rocks, eclogite layers and

country rocks (gneisses, quartzite and marble), indi-

cating that in-situ contact relations between these

rocks were not considerably disturbed during tecton-

ic emplacement. When such a package of interlay-

ered crustal materials was once subducted and

remains in the upper mantle, a large reflection

coefficient (>0.10) should occur at interfaces be-

tween UHP eclogite or garnet peridotite and their

wall rocks such as quartzite, marble, pelitic and

felsic gneisses. Therefore, the regionally extensive

mantle reflectors probably indicate the preservation

of subducted crustal materials in the lithospheric

upper mantle.

9. Conclusions

The major conclusions of this study are listed

below.

(1) Eclogites from the Sulu region deformed plasti-

cally under the UHP metamorphic conditions and

are not rheologically rigid as thought previously.

Plasticity of eclogites under UHP conditions may

effectively facilitate channeled flow within the

interplate shear zone. The enhanced counterflow

within a wedge-shaped flow channel may force

the subducted crustal rocks to be rapidly ex-

humed.

(2) Observations of optical microstructures, petrofa-

brics and TEM structures suggest that the coarse-

grained (Type-1) and fine-grained (Type-2)

eclogites recorded two distinct plastic deforma-

tions: constrictional strain (LHS) in Type-1

eclogites while plane strain in Type-2 eclogites.

Omphacite, which could be only a few times

lower in flow strength than garnet, deformed

mainly by dislocation creep as demonstrated by

the occurrence of CPOs. Plastic deformation of

garnet in Type-1 eclogites was dominated by

recovery-accommodated dislocation creep as

testified by the pervasive presence of dislocation

walls and networks, whereas garnet in Type-2

eclogites was deformed by recrystallization-

accommodated dislocation creep as indicated by

the presence of porphyroclastic texture. As h111idislocations are commonly dissociated into two

partials 1/2 h111i in bcc garnet (e.g., Voegele et

al., 1998), the cross slip of h111i dislocations,

which is sensitive to the hydrostatic pressure,

may be the factor that controls the transition

S. Ji et al. / Tectonophysics 370 (2003) 49–76 73

from recovery-accommodated dislocation creep

at higher pressures (P>f 3.2 GPa) to recrystal-

lization-accommodated dislocation creep at lower

pressures (P <f 2.8 GPa). Thus, Type-1 eclo-

gites represent the subducted mafic rocks

deformed at the peak metamorphic conditions

while Type-2 eclogites resulted from recrystalli-

zation of Type-1 eclogites along some shear

zones which were active during the exhumation

of the UHP metamorphic rocks.

(3) It is widely accepted that the packages of inter-

layered eclogite, marble, quartzite, pelitic and

felsic gneisses from the Dabie–Sulu belt, which

were pieces of the continental crust, were

subducted to mantle depths greater than 80–100

km and then tectonically exhumed to the earth’s

surface (e.g., Cong, 1996; Liou, 1999; Wang and

Cong, 1999). Field and laboratory studies of these

UHP metamorphic rocks provided some impor-

tant hints for understanding the origin of mantle

reflections. As long as these interlayered UHP

rocks remain within the present upper mantle, the

interfaces between eclogite (or peridotite) and

their wall rocks such as quartzite, marble, pelitic

and felsic gneisses should be strong seismic

reflectors. The continuous boundaries between

garnet- and omphacite-rich layers or between

peridotite and garnet-rich eclogite layers can also

be seismically reflective. Thus, the regionally

extensive seismic reflectors from the upper

mantle may indicate the preservation of relict

crustal materials subducted within the continental

lithosphere.

Acknowledgements

S.C. Ji thanks the NSERC and LITHOPROBE of

Canada for research grants and the Alexander von

Homboldt Stiftung for a visiting fellowship. This is

LITHOPROBE contribution No. 1280. A. Lacombe

and C. Nevardo are thanked for drawing the figures

and for the polished thin sections for the EBSD

analysis, respectively. Crystal orientation measure-

ments were made with Laboratoire de Tectonophysi-

que EBSD/SEM system funded by grants from CNRS/

INSU and Universite Montpellier II (France).

References

Abalos, B., 1997. Omphacite fabric variation in the Cabo Ortegal

eclogite (NW Spain): relationships with strain symmetry during

high-pressure deformation. J. Struct. Geol. 19, 621–637.

Ames, L., Tilton, G.R., Zhou, G., 1993. Timing of collision of the

Sino-Korean and Yangtze Cratons: U–Pb zircon dating of co-

esite-bearing eclogites. Geology 21, 339–342.

Andersen, T.B., Jamtveit, B., Dewey, J., Swensson, E., 1991. Sub-

duction of continental crust: major mechanisms during continent

collision and orogenic extensional collapse, a model based on

the south Norwegian Caledonides. Terra Nova 3, 303–310.

Ando, J.I., Fujino, K., Takeshita, T., 1993. Dislocation microstruc-

ture in naturally deformed silicate garnets. Phys. Earth Planet.

Inter. 80, 105–116.

Austrheim, H., 1987. Eclogitization of lower crustal granulites by

fluid migration through shear zones. Earth Planet. Sci. Lett. 81,

221–232.

Bascou, J., Tommasi, A., Mainprice, D., 2002. Plastic deformation

and development of clinopyroxene lattice preferred orientations

in eclogites. J. Struct. Geol. 24, 1357–1368.

Bass, J.D., 1995. Elasticity of minerals, glasses and melts. In: Ah-

rens, T.J. (Ed.), Mineral Physics and Crystallography: A Hand-

book of Physical Constants. AGU Reference Shelf, vol. 2.

American Geophysical Union, Washington, pp. 45–63.

Ben Ismaıl, W., Mainprice, D., 1998. An olivine fabric database: an

overview of upper mantle fabrics and seismic anisotropy. Tec-

tonophysics 296, 145–157.

Bhagat, S., Bass, J.D., Smyth, J., 1992. Single-crystal elastic prop-

erties of omphacite-C2/c by Brillouin spectroscopy. J. Geophys.

Res. 97, 6843–6848.

Birch, F., 1960. The velocity of compressional waves in rocks to 10

kilobar, Part 1. J. Geophys. Res. 65, 1083–1102.

Blanchin, M.G., Faisant, P., 1979. Compression a hautes temper-

atures et sous atmosphere oxydante de monocristaux de rutile

TiO2. Rev. Phys. Appl. 14, 619–627.

Blanchin, M.G., Bursill, L.A., Lafage, C., 1990. Deformation and

microstructure of rutile. Proc. R. Soc. Lond., A 249, 175–202.

Bostock, M.G., 1997. Anisotropic upper-mantle stratigraphy and

architecture of the Slave craton. Nature 390, 392–395.

Boundy, T.M., Fountain, D.M., Austrheim, H., 1992. Structural

development and petrofabrics of eclogite facies shear zones,

Bergen Arcs, western Norway: implications for deep crustal

deformational processes. J. Metamorph. Geol. 10, 127–146.

Buatier, M., van Roermund, H.L., Dury, M.R., Lardeaux, J.M.,

1991. Deformation and recrystallization mechanisms in natu-

rally deformed omphacites from the Sesia–Lanzo zone: geo-

physical consequences. Tectonophysics 195, 11–27.

Burlini, L., Tancredi, S., 1998. Experimental study of the seismic

properties of Isola d’Elba metamorphic basement (Tuscany,

Italy). Mem. Soc. Geol. Ital. 52, 101–110.

Calvert, A.J., Sawyer, E.W., Davis, W.J., Ludden, J.N., 1995. Ar-

chean subduction inferred from seismic images of a mantle

suture in the Superior Province. Nature 375, 670–674.

Chai, M., Brown, J.M., Slusky, L.J., 1997. The elastic constants of a

pyrope-grossular-almandine garnets to 20 GPa. Geophys. Res.

Lett. 24, 523–526.

S. Ji et al. / Tectonophysics 370 (2003) 49–7674

Chopin, C., Henry, C., Michard, A., 1991. Geology and petrology

of the coesite-bearing terrain, Dora Maira massif, western Alps.

Eur. J. Mineral. 3, 263–291.

Christensen, N.I., 1965. Compressional wave velocities in meta-

morphic rocks at pressures to 10 kilobars. J. Geophys. Res.

70, 6147–6164.

Christensen, N.I., Mooney, W.D., 1995. Seismic velocity structure

and composition of the continental crust: a global view. J. Geo-

phys. Res. 100, 9761–9788.

Cong, B. (Ed.), 1996. Ultrahigh-Pressure Metamorphic Rocks in

the Dabieshan–Sulu Region of China. Kluwer Academic Pub-

lishing, London. 224 pp.

Cook, F.A., van der Velden, A.J., Hall, K.W., 1999. Frozen sub-

duction in Canada’s Northwest Territories: lithoprobe deep

lithospheric reflection profiling of the western Canadian Shield.

Tectonics 18, 1–24.

Crosson, R.S., Lin, J.W., 1971. Voigt and Reuss prediction of ani-

sotropic elasticity of olivine. J. Geophys. Res. 76, 570–578.

Dell’Angelo, L.N., Tullis, J., 1996. Textural and mechanical evo-

lution with progressive strain in experimentally deformed aplite.

Tectonophysics 256, 57–82.

Doukhan, N., Sautter, V., Doukhan, J.C., 1994. Ultradeep, ultra-

mafic mantle xenoliths: transmission electron microscopy pre-

liminary results. Phys. Earth Planet. Inter. 82, 195–207.

Ellis, D.J., Green, D.H., 1979. An experimental study of the effect

of Ca upon garnet-clinopyroxene Fe–Mg exchange equilibrium.

Contrib. Mineral. Petrol. 71, 13–22.

Fountain, D.M., Christensen, N.I., 1989. Composition of the con-

tinental crust and upper mantle: a review. In: Pakiser, L.C.,

Mooney, W.D. (Eds.), Geophysical Framework of the Continen-

tal United States. The Geological Society of America, Denver,

CO, pp. 711–742.

Fountain, D.M., Boundy, T.M., Austrheim, H., Rey, P., 1994. Eclo-

gite-facies shear zones—deep crustal reflectors. Tectonophysics

232, 411–424.

Francis, D.M., 1976. The origin of amphibole in lherzolite xenoliths

from Nunivak Island, Alaska. J. Petrol. 17, 357–378.

Garem, H., Rabier, J., Veyssiere, P., 1982. Slip system in gadoli-

nium gallium garnet single crystals. J. Mater. Sci. 17, 878–884.

Godard, G., van Roermund, H.L.M., 1995. Deformation-in-

duced clinopyroxene fabrics from eclogites. J. Struct. Geol. 17,

1425–1443.

Gubbins, D., Banicoat, A., Cann, J., 1994. Seismological con-

straints on the gabbro-eclogite transition in subducted oceanic

crust. Earth Planet. Sci. Lett. 122, 89–101.

Hacker, B.R., Wang, Q., 1995. Ar/Ar geochronology of ultra-

high-pressure metamorphism in central China. Tectonics 14,

994–1006.

Helmstaedt, H., Anderson, O.L., Gavasci, A.T., 1972. Petrofa-

bric studies of eclogite, spinel-websterite and spinel-lherzolite

xenoliths from kimberlite-bearing breccia pipes in southeast-

ern Utah and northeastern Arizona. J. Geophys. Res. 77,

4350–4365.

Henry, C., Michard, A., Chopin, C., 1993. Geometry and structural

evolution of ultra-high pressure and high-pressure rocks from

the Dora-Maria massif, Western Alps, Italy. J. Struct. Geol. 15,

965–981.

Hirajima, T., Zhang, R., Li, J., Cong, B., 1992. Petrology of the

nyboite-bearing eclogite in the Donghai area, Jiansu Province,

eastern China. Mineral. Mag. 56, 37–46.

Hynes, A., Snyder, D., 1995. Deep-crustal mineral assemblages and

potential for crustal rocks below the Moho in the Scottish Ca-

ledonides. Geophys. J. Int. 123, 323–339.

Isacks, B., Molnar, P., 1969. Mantle earthquake mechanisms and

the sinking of the lithosphere. Nature 223, 1121–1124.

Isacks, B., Molnar, P., 1971. Distribution of stresses in the descend-

ing lithosphere from a global survey of focal mechanism solu-

tions of mantle earthquakes. Rev. Geophys. Space Phys. 9,

103–174.

Jahn, B.M., Cornichet, J., Cong, B., Yui, T.F., 1996. Ultrahigh-qNDeclogites from an ultrahigh-pressure metamorphic terrane of

China. Chem. Geol. 127, 61–79.

Ji, S.C., Mainprice, D., 1988. Natural deformation fabrics of pla-

gioclase: implications for slip systems and seismic anisotropy.

Tectonophysics 147, 145–163.

Ji, S.C., Martignole, J., 1994. Ductility of garnet as an indicator of

extremely high temperature deformation. J. Struct. Geol. 16,

985–996.

Ji, S.C., Martignole, J., 1996. Ductility of garnet as an indicator of

extremely high temperature deformation: reply. J. Struct. Geol.

18, 1375–1379.

Ji, S.C., Salisbury, M., 1993. Shear-wave velocities, anisotropy

and splitting of the high-grade mylonites. Tectonophysics 221,

453–473.

Ji, S.C., Salisbury, M., Hanmer, S., 1993. Petrofabric, P-wave ani-

sotropy and seismic reflectivity of high-grade tectonites. Tecto-

nophysics 222, 195–226.

Ji, S.C., Zhao, X., Francis, D., 1994. Calibration of shear-wave

splitting in subcontinental upper mantle beneath active orogenic

belts using ultramafic xenoliths from the Canadian Cordillera

and Alaska. Tectonophysics 239, 1–28.

Ji, S.C., Zhao, P., Saruwatari, K., 1997. Fracturing of garnet crystals

in anisotropic metamorphic rocks during uplift. J. Struct. Geol.

19, 603–620.

Ji, S.C., Wang, Q., Xia, B., 2002. Handbook of Seismic Properties

of Minerals, Rocks and Ores. Polytechnic International Press,

Montreal. 630 pp.

Jiang, Z., Prior, D.J., Wheeler, J., 2000. Albite crystallographic

preferred orientation and grain misorientation in a low-grade

mylonite: implication for granular flow. J. Struct. Geol. 22,

1663–1674.

Jin, Z.M., Zhang, J., Green II, H.W., Jin, S., 2001. Eclogite

rheology: implications for subducted lithosphere. Geology

29, 667–670.

Karato, S., Wu, P., 1993. Rheology of upper mantle: a synthesis.

Science 260, 771–778.

Karato, S., Wang, Z.C., Liu, B., Fujino, K., 1995. Plastic deforma-

tion of garnets: systematics and implication for the rheology of

the mantle transition zone. Earth Planet. Sci. Lett. 130, 13–30.

Kern, H., 1990. Laboratory seismic measurements: an aid in the

interpretation of seismic field data. Terra Nova 2, 617–628.

Kern, H., Richter, A., 1981. Temperature derivatives of compres-

sional and shear wave velocities in crustal and mantle rocks at

6 kbar confining pressure. J. Geophys. 49, 47–56.

S. Ji et al. / Tectonophysics 370 (2003) 49–76 75

Kern, H., Burlini, L., Ashchepkov, I.V., 1996. Fabric-related seis-

mic anisotropy in upper mantle xenoliths: evidence from

measurements and calculations. Phys. Earth Planet. Inter. 95,

195–209.

Kern, H., Gao, S., Jin, Z., Popp, T., Jin, S., 1999. Petrophysical

studies on rocks from the Dabie ultrahigh-pressure (UHP) me-

tamorphic belt, Central China: implication for the composi-

tion and delamination of the lower crust. Tectonophysics 301,

191–215.

Kern, H., Jin, Z., Gao, S., Popp, T., Xu, Z., 2002. In-situ physical

properties of ultrahigh-pressure metamorphic rocks from the

Sulu terrain, Eastern central China: implication for the seismic

structure at the Donghai (CCSD) drilling site. Tectonophysics

354, 315–330.

Koziol, A.M., Newton, R.C., 1989. Grossular activity-composition

relationships in ternary garnets determined by reversed dis-

placed-equilibrium experiments. Contrib. Mineral. Petrol. 103,

423–433.

Kumazawa, M., Helmstaedt, H., Masaki, K., 1971. Elastic proper-

ties of eclogite xenoliths from diatremes of the east Colo-

rado Plateau and their implication to the upper mantle structure.

J. Geophys. Res. 76, 1231–1247.

Li, S., Mooney, W.D., 1998. Crustal structure of China from deep

seismic sounding profiles. Tectonophysics 288, 105–113.

Li, S.G., Ge, N.J., Liu, D., Zhang, Z., Ye, X., Zheng, S., Peng, C.,

1989. The Sm–Nd isotopic age of C-type eclogite from the

Dabie group in northern Dabie mountains and its tectonic im-

plication. Chin. Sci. Bull. 34, 1623–1628.

Li, S.G., Liu, D., Chen, Y., Zhang, G., Zhang, Z., 1992. The Sm–

Nd isotopic age of coesite-bearing eclogite from the southern

Dabie Mountains. Chin. Sci. Bull. 37, 1638–1641.

Li, S.G., Chen, Y.Z., Ge, N.J., Liu, D.L., Zhang, Z.M., Zhang,

Q.D., Zhao, D.M., 1993a. U–Pb zircon ages of eclogite and

gneiss from Jiaonan group in Qingdao area. Chin. Sci. Bull. 38,

1773–1777.

Li, S.G., Xiao, Y., Liu, D., Chen, Y., Ge, N.J., Zhang, Z., Sun, S.S.,

Cong, B., Zhang, R., Hart, S.R., 1993b. Collision of the North

China and Yangtze blocks and formation of coesite-bearing

eclogites: timing and processes. Chem. Geol. 109, 89–111.

Liou, J.G., 1999. Petrotectonic summary of less intensively studied

UHP regions. Int. Rev. Geol. 41, 571–586.

Mackenzie, J.K., 1958. Second paper on statistics associated with

the random disorientation of cubes. Biometrika 45, 229–240.

Mainprice, D., 1990. A Fortran program to calculate seismic aniso-

tropy from the lattice preferred orientation of minerals. Com.

Geosci. 16, 385–393.

Mainprice, D., Silver, P.G., 1993. Interpretation of SKS-wave using

samples from the subcontinental lithosphere. Phys. Earth Planet.

Inter. 78, 257–280.

Mainprice, D., Lloyd, G.E., Casey, M., 1993. Individual orientation

measurements in quartz polycrystals: advantages and limitations

for texture and petrophysical property determinations. J. Struct.

Geol. 15, 1169–1187.

Mainprice, D., Barruol, G., Ben Ismaıl, W., 2000. The anisotropy of

the Earth’s mantle: from single crystal to polycrystal. In: Karato,

A.M., Forte, A.M., Liebermann, R.C., Masters, G., Stixrude, L.

(Eds.), Mineral Physics and Seismic Tomography: From Atomic

to Global. AGU Geophysical Monograph, vol. 117. American

Geophysical Union, Washington, DC, pp. 237–264.

Manghnani, M.H., Ramananantoandro, R., Clark, S.P., 1974. Com-

pressional and shear wave velocities in granulite facies rocks

and eclogites to 10 kilobars. J. Geophys. Res. 79, 5427–5445.

Mauler, A., Burlini, L., Kunze, K., Philippot, P., Burg, J.P., 2000.

P-wave anisotropy in eclogites and relationship to the Omphacite

crystallographic fabric. Phys. Chem. Earth (A) 25, 119–126.

Montardi, Y., Mainprice, D., 1987. A transmission electron micro-

scopic study of the natural plastic deformation of calcic plagio-

clases (An 68-70). Bull. Mineral. 110, 1–14.

Morgan, J.V., Hadwin, M., Warner, M.R., Barton, P.J., Morgan, R.,

Li, P., 1994. The polarity of deep seismic reflections from the

lithospheric mantle: evidence for a relict subduction zone. Tec-

tonophysics 232, 319–328.

Okay, A.I., 1993. Petrology of a diamond and coesite-bearing

metamorphic terrain: Dabie Shan, China. Eur. J. Mineral. 5,

659–675.

O’Reilly, S.Y., Griffin, W.L., 1985. A xenolith-derived geotherm

for southeastern Australia and its geophysical implications. Tec-

tonophysics 111, 41–63.

Philippot, P., van Roermund, H.L.M., 1992. Deformation processes

in eclogitic rocks: evidence for the rheological delamination of

the oceanic crust in deeper levels of subduction zones. J. Struct.

Geol. 14, 1059–1077.

Piepenbreier, D., Stockhert, B., 2001. Plastic flow of omphacite in

eclogites at temperatures below 500 jC—implications for in-

terplate coupling in subducting zones. Int. J. Earth Sci. 90,

197–210.

Poirier, J.P., Vergobbi, B., 1978. Splitting of dislocation in olivine,

cross-slip-controlled creep and mantle rheology. Phys. Earth.

Planet. Inter. 16, 370–378.

Powell, R., 1985. Regression diagnostics and robust regression in

geothermometer-geobarometer calibration: the garnet-clino-

pyroxene geothermometer revisited. J. Metamorph. Geol. 3,

231–243.

Prior, D.J., Wheeler, J., Brenker, F.E., Hart, B., Matthews, M.,

2000. Crystal plasticity of natural garnet: new microstructural

evidence. Geology 28, 1003–1006.

Rabier, J., Veyssiere, P., Grilhe, J., 1976. Possibility of stacking

faults and dissociation of dislocations in the Garnet structure.

Phys. Status Solidi 35, 259–268.

Rudnick, R.L., Fountain, D.M., 1995. Nature and composition of

the continental crust: a lower crustal perspective. Rev. Geophys.

33, 267–309.

Saruwatari, K., Ji, S.C., Long, C.X., Salisbury, M., 2001. Seismic

anisotropies of mantle xenoliths and constraints on the upper

mantle structures beneath the southern Canadian Cordillera.

Tectonophysics 339, 399–422.

Stockhert, B., Renner, J., 1998. Rheology of crustal rocks at

ultrahigh pressure. In: Hacker, B.R., Liou, J.G. (Eds.), When

Continents Collide: Geodynamics and Geochemistry of Ultra-

high-Pressure Rocks. Kluwer Academic Publishing, Amsterdam,

pp. 57–95.

van Roermund, H.L.M., Boland, J.N., 1981. The dislocation sub-

structures of naturally deformed omphacites. Tectonophysics 78,

403–418.

S. Ji et al. / Tectonophysics 370 (2003) 49–7676

Voegele, V., Cordier, P., Sautter, V., Sharp, T.G., Lardeaux, J.M.,

Marques, F.O., 1998. Plastic deformation of silicate garnets II.

Deformation microstructure in natural samples. Phys. Earth

Planet. Inter. 108, 319–338.

Wang, Q.C., Cong, B.L., 1999. Exhumation of UHP terranes: a case

study from Dabie mountains, Eastern China. Int. Geol. Rev. 41,

994–1004.

Wang, Z.C., Ji, S.C., 1999. Deformation of silicate garnets: brittle-

ductile transition and its geological implications. Can. J. Miner.

37, 525–541.

Wang, Z.C., Ji, S.C., 2000. Diffusion creep of fine-grained garne-

tite: implications for the flow strength of subducting slabs. Geo-

phys. Res. Lett. 27, 2333–2336.

Wang, Z.C., Ji, S.C., 2001. Elasticity of six polycrystalline silicate

garnets at pressure up to 3.0 GPa. Am. Mineral. 86, 1209–1218.

Wang, X.M., Liou, J.G., 1991. Regional ultrahigh-pressure coesite-

bearing eclogitic terrane in central China: evidence from country

rocks, gneiss, marble, and metapelite. Geology 19, 933–936.

Wang, X.M., Liou, J.G., Maruyama, S., 1992. Coesite-bearing eclo-

gites from the Dabie mountains, Central China: petrogenesis,

P–T paths, and implication for regional tectonics. J. Geol.

100, 231–250.

Wang, Q.C., Ishiwatari, A., Zhao, Z.Y., Hirajima, T., Hiramatsu, N.,

Enami, M., Zhai, M.G., Li, J.J., Cong, B.L., 1993. Coesite-

bearing granulite retrograded from eclogite in Weihai, eastern

China. Eur. J. Mineral. 5, 141–152.

Warner, M., Morgan, J., Barton, P., Morgan, P., Price, C., Jones, K.,

1996. Seismic reflection from the mantle represent relict sub-

duction zones within the continental lithosphere. Geology 24,

39–42.

Warner, M.R., McGeary, S., 1987. Seismic reflection coefficients

from mantle fault zones. Geophys. J. R. Astron. Soc. 89,

223–230.

Wilshire, H.G., Kirby, S.H., 1989. Dikes, joints and faults in the

upper mantle. Tectonophysics 161, 23–31.

Xu, Z.Q., 1987. Etudes tectonique et microtectonique de la chaine

Paleozoique et Triasique des Qinlings (Chine). Unpublished

PhD thesis, Universite de Montpellier II, France. 162 pp.

Xu, S.T., Okay, A.I., Ji, S., Sengor, A.M.C., Su, W., Liu, Y., Jiang,

L., 1992. Diamond from the Dabie Shan metamorphic rocks and

its implication for tectonic setting. Science 256, 80–82.

Yang, J.J., Godard, G., Kienast, J.R., Lu, Y.Z., Sun, J.X.,

1993. Ultrahigh-pressure (60 Kbar) magnesite-bearing garnet

peridotites from Northeastern Jiangsu, China. J. Geol. 101,

541–554.

Yang, W.C., Yu, C.Q., 2001. Kinetics and dynamics of development

of the Dabie–Sulu UHPM terranes based on geophysical evi-

dences. Chin. J. Geophys. 44, 346–359.

Yang, W.C., Cheng, Z.Y., Chen, G.J., Hu, Z.Y., Bai, J., 1999. Geo-

physical investigation in northern Sulu UHPM belt. Part I: Deep

Seismic reflection. Chin. J. Geophys. 42, 41–52.

Zhang, R.Y., Liou, J.G., 1996. Coesite inclusions in dolomite from

eclogite in the southern Dabie Mountains, China: the signifi-

cance of carbonate minerals in UHPM rocks. Am. Mineral. 81,

181–186.

Zhang, R.Y., Liou, J.G., Cong, B., 1994. Petrogenesis of garnet-

bearing ultramafic rocks and associated eclogites in the Su-LU

ultrahigh-P metamorphic terrane, eastern China. J. Metamorph.

Geol. 12, 169–186.

Zulauf, G., 1997. Constriction due to subduction: evidence for slab

pull in the Marianske Lazne complex (central European Varis-

cides). Terra Nova 9, 232–236.