.................................................................The possible subduction ofcontinental material todepths greater than 200 kmKai Ye, Bolin Cong & Danian Ye

Laboratory of Lithosphere Tectonic Evolution, Institute of Geology and Geophysics,Chinese Academy of Sciences, Beijing 100029, PO Box 9825, China.................................. ......................... ......................... ......................... ......................... ........

Determining the depth to which continental lithosphere can besubducted into the mantle at convergent plate boundaries is ofimportance for understanding the long-term growth of super-continents as well as the dynamic processes that shape suchmargins. Recent discoveries of coesite and diamond in regionalultrahigh-pressure (UHP) metamorphic rocks has demonstratedthat continental material can be subducted to depths of at least120 km (ref. 1), and subduction to depths of 150–300 km has beeninferred from garnet peridotites in orogenic UHP belts based onseveral indirect observations2–5. But continental subduction tosuch depths is difficult to trace directly in natural UHP meta-morphic crustal rocks by conventional mineralogical and petro-logical methods because of extensive late-stage recrystallizationand the lack of a suitable pressure indicator. It has been predictedfrom experimental work, however, that solid-state dissolution ofpyroxene should occur in garnet at depths greater than 150 km(refs 6–8). Here we report the observation of high concentrationsof clinopyroxene, rutile and apatite exsolutions in garnet withineclogites from Yangkou in the Sulu UHP metamorphic belt,China. We interpret these data as resulting from the high-pressureformation of pyroxene solid solutions in subducted continentalmaterial. Appropriate conditions for the Na2O concentrations andoctahedral silicon observed in these samples are met at depthsgreater than 200 km.

Supercontinents are generally thought to grow through subduc-tion and collision of smaller continental blocks—as is happening innorth India and southwest China, where the Indian continentalplate is subducting below the Eurasian continent, and causing theuplift of the Tibetan plateau. Subduction of the South Chinacontinent under the North China continent in the Triassic9 periodconsolidated the old Eurasia continent, and resulted in the forma-tion of various coesite- and diamond-bearing UHP metamorphicrocks in the oregon in eastern China (for a review, see ref. 10). TheSulu UHP metamorphic belt is the most deeply exhumed part of theDabieshan-Sulu oregon. This provides the possibility of tracing thesubduction depths of continental materials from fresh samples inthe Sulu UHP belt.

Unusual eclogites were found as lenses in a garnet-peridotite layerintercalated with layers of UHP metamorphosed crustal igneouscomposite11 at Yangkou. Partial transformation from low-pressureigneous textures (plagioclase stable) to UHP textures (coesitestable) have been reported in some meta-gabbroic layers of thecomposite12. Unlike the common eclogites with near equigranulartexture in the world-wide UHP belts, these eclogites contain anolder generation of garnet (grt-I) which occurs as larger (up to2 cm), purple, plastically deformed porphyroblast in the matrix oflater medium-grained (0.05–0.5 mm) garnet (grt-II), omphacite(cpx-II), rutile (rut-II) and apatite (ap-II) (Fig. 1a). Grt-I exhibitsabundant fine-grained euhedral rods or needles of clinopyroxene(cpx-I), rutile (rut-I) and apatite (ap-I) (Fig. 1a). The clinopyroxenerods are spindle-shaped in the long axis section (Fig. 1b), andhexagonal prismatic in the basal section (Fig. 1c). Apatite rods occuras hexagonal prism or rhombohedra. Like the hexagonal prismaticclinopyroxene, each group of apatite crystal planes is parallel to onegroup of rutile rods. In cases, several short apatite rods are dotted

along one long rutile rod (Fig. 1d). Backscattered electron images(see Supplementary Information) were used to estimate theamounts of the exsolutions. The rods increase in concentrationwith distance away from a rod-free edge zone (50–400 mm wide) toa maximum of 4–7 vol.% for clinopyroxene, 1–2 vol.% for rutileand 0.5–1 vol.% for apatite in the core.

Overall, the rods occur in groups of parallel rods; each group isorientated in one of four directions. The effect is to show anequilateral triangular form in cross-section (Fig. 1a). Universal-stage measurements show that the long-axis orientations of the rodsare consistent with four cubic 〈111〉 directions. In a periodiccylindrical packing model13, these are dominant glide planes.Their concentration, topography and orientation suggest that therods of clinopyroxene, rutile and apatite in the porphyroblasticgarnet are generated by exsolution, rather than by epitaxial replace-ment or accidental inclusion. It is reasonable to assume that thecubic oxygen sublattice of garnet would have the same relation tothe pseudohexagonal oxygen sublattice of clinopyroxene. Thiswould imply a topotaxy between garnet and hexagonal prismaticcpx in Fig. 1c as: (100)cpx // (110)grt; (010)cpx // (1̄10)grt; (1̄11)cpx //(221)grt.

Mineral compositions were determined by means of a CamebaxSX51 microprobe analyser. Longer counting times were pro-grammed for Ti (20 s), Cr (20 s), P (60 s) and Na (60 s) in garnetthan for other common elements (10 s). Representative analyses arelisted in Table 1. The two stages of garnet and clinopyroxene differwidely in their composition. The porphyroblastic garnet shows aclear zonation. The exsolution-rich core contains higher Mg and Feconcentrations, and a lower Ca concentration. The exsolution-freerim has a composition similar to that of the matrix garnet. Itcontains higher Ca, and lower Fe and Mg (Table 1). Particularlyhigh concentrations of P2O5 (0.06–0.17 wt%) and Na2O (0.05–0.06 wt%) were detected in the core of the porphyroblastic garnet.However, the Ca-rich rim and the matrix garnet contain almost noP2O5 (, 0.03 wt%) or Na2O (, 0.02 wt%) (Table 1). The exsolvedcpx-I contains lower jadeite (Xjd = 0.17–0.25) than the matrixomphacite (Xjd = 0.29–0.35). (Here Xjd is the volume per cent ofjadeite component.) The average composition of the exsolved cpx-I,

letters to nature

734 NATURE | VOL 407 | 12 OCTOBER 2000 | www.nature.com

grt-Icpx

rut

ap

rut

cpx

rut

a b

c dgrt-II+cpx-II+rut-II+ap-II

Figure 1 Transmitted light photomicrographs of garnet (grt) with exsolution rods ofclinopyroxene (cpx), rutile (rut) and apatite (ap) along crystallographically controlledplanes. a, Porphyroblastic garnet (grt-I) in the matrix of garnet (grt-II), omphacite (cpx-II),rutile (rut-II) and apatite (ap-II), the exsolution rods oriented in an equilateral triangularform. Scale bar, 1 mm. b, Spindle-shaped long-axis section of clinopyroxene rods inporphyroblastic garnet. Scale bar, 0.1 mm. c, Hexagonal prismatic basal sections ofclinopyroxene rods in porphyroblastic garnet. Note that each group of pyroxene crystalplanes is parallel to one group of needle-like rutile rods. Scale bar, 0.06 mm. d, Threeshort apatite rods dot one long rutile rod. Scale bar, 0.1 mm.

© 2000 Macmillan Magazines Ltd

rut-I and ap-I were restored to the core composition of grt-Ifollowing the estimated volume proportions of grt91.5cpx7rut1ap0.5.The calculated original garnet (no. 6 in Table 1) is depleted in Al(with respect to the chemical formula) and contains 0.04 more Sithan the ideal value of 3 for 12 oxygens, necessitating the existenceof 0.06 (SiTotal + PIV − 3.00) 6-fold Si in octahedral coordination(Table 1). Here the roman numerals indicate the degree of coordi-nation of the atoms. It also exhibits high Na2O (0.34 wt%), TiO2

(1.07 wt%) and P2O5 (0.28 wt%) values. A slightly lower averageNa2O content (0.28 wt%) was confirmed for the core of theporphyroblastic garnet with the highest amount of clinopyroxeneexsolution by broad-beam electron microprobe analysis.

Experiments have predicted the structural transformation ofpyroxene to garnet and the solid-state dissolution of pyroxeneinto garnet at very-high-pressure conditions (5–17 GPa)6–8. Con-version of pyroxene to the garnet structure requires some of thefour-fold tetrahedral Si in pyroxene to be changed to six-foldoctahedral coordination in converted majoritic garnet, becausethe ratio of Si to O in pyroxene is 4:12, rather than the garnetvalue of 3:12. From a crystal chemistry point of view, substitution ofNaVIII + SiVI → CaVIII + AlVI in the garnet structure can describethese transformations14. Direct access to xenoliths sampled bydiamond kimberlite pipes from the stable mantle lithosphere atdepths greater than 300 km have revealed pyroxene dissolution inhomogeneous garnet included in diamond15, and decompressionresulted in exsolution of clinopyroxene in garnet16, which is iden-tical to the occurrence of clinopyroxene in garnet reported here.Experiments17,18 have shown that garnet can host high Ti, Na and Pcontents under very-high-pressure conditions. Phosphorus hasbeen suggested18 to enter the garnet structure through the coupledsubstitution of PIV + NaVIII → SiIV+ CaVIII. Ringwood and co-workers6,17 proposed that (CaNa2)Ti2Si3O12 and (Ca2Na)(AlTi)-Si3O12 are probable Ti-bearing garnet components, based on com-positions of garnets synthesized under high-pressure conditions,and they hence suggested that Ti enter garnet through substitutionof NaVIII + TiVI → CaVIII + AlVI. Such Na-, P- and Ti-bearing garnetshad been reported in ultra-deep xenoliths in diamond kimberlitepipes14,15 and UHP metamorphic crustal rocks19.

Because of the lack of a suitable pressure indicator for the

bimineralic eclogitic assemblage, the pressure conditions underwhich the garnet was once homogeneously pyroxene-substitutedmust be inferred from experimental data on basaltic systems undersubduction conditions. As pressure increases, the solubility ofpyroxene in eclogitic garnet increases gradually7,20, and the Na, Tiand P contents in garnet also have a positive pressure dependence18.The calculated Na2O content (0.34 wt%) and octahedral siliconvalues (0.06) of the original Yangkou garnet are higher than those ofthe garnets synthesized21 at 7 GPa and 1,000 8C using startingmaterial of a natural eclogite from Dabieshan, and also higherthan those of garnet synthesized20 at 7 GPa and 1,400 8C in a MORBsystem: but the values of 0.34 wt% and 0.06 are consistent withgarnet synthesized7 at 7.6 GPa and 1,200 8C in an olivine tholeiitesystem. In these synthesis experiments, the temperatures of 1,400 8Cand 1,200 8C are higher than in the subduction regime, and theNa2O content in garnet is known20 to be higher at higher tempera-tures. Thus, it is reasonable to conclude that the original garnet inthe Yangkou eclogites was formed at pressures greater than 7 GPa.This puts the origin of the studied Yangkou eclogites at depthsgreater than 200 km.

The tectonic settings of the protolith of Yangkou eclogite and thehost garnet-peridotite are not clear. However, lower-pressuremineral inclusions have been reported in garnet of similar garnet-peridotite from nearby localities in the same belt22. This impliespossible crustal association of the studied eclogite before subduc-tion. An alternative interpretation is that the peridotite was under-plated at the base of the continental basement then underwentultrahigh-pressure metamorphism22. But regardless of the exacttectonic settings, our results have implications for the dynamicsof continental subduction and exhumations of the subducted rocks.

Density calculations based on experimental data confirm thatsubducted oceanic slab is substantially denser than the surroundingpyrolitic mantle at depths of 100–650 km (ref. 7). Oceanic slab hasbeen observed to be subducted to depths exceeding 600 km in themantle23. Subduction of the less-dense continental crust to greatmantle depths is traditionally thought to be difficult, because theoperation of buoyancy forces would drive it back to the crustal level.However, subducted continental crust, and terrigenous and pelagicsediments, can be important in the geochemical evolution of the

letters to nature

NATURE | VOL 407 | 12 OCTOBER 2000 | www.nature.com 735

Table 1 Representative compositions of clinopyroxene (cpx) and garnet (grt)

1 2 3 4 5 6 7 8 9...................................................................................................................................................................................................................................................................................................................................................................

Mineral cpx cpx grt grt grt grt grt grt grtGroup E M CP RP M O SY SY SYSiO2 55.36 55.56 39.39 38.93 39.21 40.32 41.43 41.49 40.94TiO2 0.06 0.01 0.11 0.06 0.06 1.07 0.15 0.03 1.49Al2O3 6.14 7.90 21.37 21.01 20.87 20.18 22.46 22.77 22.56Cr2O3 0.01 0.06 0.04 0.00 0.05 0.06 – 0.19 –FeO 3.85 4.03 21.75 18.60 18.92 20.19 13.61 13.05 12.29MgO 12.50 10.82 9.57 6.49 5.98 9.41 12.28 16.20 9.96MnO 0.05 0.05 0.49 0.43 0.45 0.47 – 0.26 –CaO 17.51 16.14 7.12 14.45 14.90 8.36 9.67 5.61 12.46Na2O 4.31 5.54 0.06 0.01 0.02 0.34 0.38 0.06 0.27K2O 0.00 0.01 0.00 0.00 0.00 0.00 – 0.06 –P2O5 0.01 0.00 0.09 0.01 0.00 0.28 – – –Total 99.78 100.12 99.99 99.99 100.46 100.68 99.98 99.72 99.97...................................................................................................................................................................................................................................................................................................................................................................

Cation proportions based on 6 (for cpx) and 12 (for grt) oxygens

Si 1.995 1.992 3.001 2.991 3.005 3.039 3.046 3.019 3.019Ti 0.002 0.000 0.006 0.003 0.003 0.061 0.008 0.002 0.083Al 0.261 0.334 1.919 1.902 1.885 1.793 1.946 1.953 1.961Cr 0.000 0.002 0.002 0.000 0.003 0.004 – 0.011 –Fe 0.116 0.121 1.386 1.195 1.213 1.273 0.837 0.794 0.758Mg 0.671 0.578 1.087 0.743 0.683 1.057 1.346 1.757 1.095Mn 0.001 0.002 0.032 0.028 0.029 0.030 – 0.016 –Ca 0.676 0.620 0.581 1.189 1.224 0.675 0.762 0.437 0.984Na 0.301 0.385 0.009 0.001 0.003 0.050 0.054 0.008 0.039K 0.000 0.000 0.000 0.000 0.000 0.000 – 0.006 –P 0.000 0.000 0.006 0.001 0.000 0.018 – – –...................................................................................................................................................................................................................................................................................................................................................................Values given are for analyses of minerals in Yangkou eclogites (nos 1-5); no. 6 represents calculated original garnet; nos 7, 8 and 9 are garnets synthesized in basaltic systems at 7.6 GPa and 1,200 8C byIrifune et al.7, 7 GPa and 1,000 8C by Okamoto and Maruyama21, and 7 GPa and 1,400 8C by Ono and Yasuda20, respectively. Abbreviations; cpx: clinopyroxene; grt: garnet; E: exsolution; M: matrix, CP:core of porphyroblastic garnet, RP: rim of porphyroblastic garnet, O: original garnet, SY: synthesized garnet.

© 2000 Macmillan Magazines Ltd

source of many alkali intraplate hotspot magmas—such sources areat extreme depths in the mantle24–26. Recent discoveries of coesiteand diamond in eclogites, metamorphic sedimentary rocks1 andeven granitic gneisses27 in continental subduction orogenic beltsreveal that large, coherent volumes of continental materials could besubducted to depths of more than 120 km. The present studyfurther extends continental subduction to depths of more than200 km. Experiments and density calculations8 show that thedensities of the continental crust are substantially smaller thanthat of the surrounding pyrolitic mantle (3.4 g cm−3) at pressureslower than 6 GPa (3.0 g cm−3), slightly less-dense in the interval of6.5–8.5 GPa (3.2–3.3 g cm−3), and much denser above 9–10 GPauntil 24 GPa (3.7–3.9 g cm−3)8. Once the subducted continentalmaterials reach a depth of about 200 km, their densities approachthat of the surrounding mantle, and exceed the density of thesurrounding mantle below 300 km. Thus, the buoyancy forces areeffectively cancelled and entrainment by the sinking slab becomesthe predominant influence. In the Sulu case, a subducted oceanicslab is thought to have hauled the following continental materials todepths greater than 200 km. Only one mechanism is reasonablefor exhumation of such deeply subducted rocks: that they are raftedby rocks beneath them, propelled by buoyancy forces2. Thus, adetachment was required along the subducting slab at some depthsbetween 200 and 300 km, so that the buoyancy forces could drivethe deeply subducted continental materials back to the crustallevel.

Much larger-scale extremely deeply subducted crustal materialsmay be present in the Sulu UHP belt. Such materials would beindicated by the presence of appropriate mineralogical or micro-textural evidence—such as TiO2 with the a-PbO2 structure (stableat 5–7 GPa), K-wadeite (6.5–7 GPa), or high-pressure C2/C clin-oenstatite (6–7 GPa). We note that the presence of high-pressureC2/C clinoenstatite was previously predicted28 for the Alpe Aramiperidotite, and that this was subsequently confirmed3. M

Received 13 June; accepted 24 August 2000.

1. Chopin, C. & Sobolev, N. V. in Ultrahigh Pressure Metamorphism (eds Coleman, R. G. & Wang, X.)

96–131 (Cambridge Univ. Press, New York, 1995).

2. Dobrzhinetskaya, L., Green, H. W. & Wang, S. Alpe Arami: a peridotite massif from depths of more

than 300 kilometers. Science 271, 1841–1845 (1996).

3. Bozhilov, K. N., Green, H. W. & Dobrzhinetskaya, L. Clinoenstatite in Alpe Arami peridotite:

additional evidence of very high pressure. Science 284, 128–132 (1999).

4. Van Roermund, H. L. M. & Drury, M. R. An ultra-deep (.200 km) orogenic peridotite body in

Western Norway. Eos 79, F971 (1998).

5. Yang, J. J., Godard, G., Kienast, J. R., Lu, Y. & Sun, J. Ultrahigh-pressure (60 kbar) magnesite-bearing

garnet peridotites from northeastern Jiangsu, China. J. Geol. 101, 541–554 (1993).

6. Ringwood, A. E. & Major, A. Synthesis of majorite and other high pressure garnets and perovskites.

Earth Planet. Sci. Lett. 12, 411–418 (1971).

7. Irifune, T., Sekine, T., Ringwood, A. E. & Hibberson, W. O. The eclogite-garnet transformation at high

pressure and some geographical implications. Earth Planet. Sci. Lett. 77, 245–256 (1986).

8. Irifune, T., Ringwood, A. E. & Hibberson, W. O. Subduction of continental crust and terrigenous and

pelagic sediments: an experimental study. Earth Planet. Sci. Lett. 126, 351–386 (1994).

9. Ames, L., Tilton, G. R. & Zhou, G. Z. Timing of collision of the Sino-Korean and Yangtze cratons:

U-Pb zircon dating of coesite-bearing eclogites. Geology 21, 339–342 (1993).

10. Cong, B. L. Ultrahigh-Pressure Metamorphic Rocks in the Dabieshan-Sulu Region of China (Science

Press/Kluwer Academic, Beijing, 1996).

11. Wallis, S. R. et al. Occurrence and field relationships of ultrahigh-pressure metagranitoid and coesite

eclogite in the Su-Lu terrane, eastern China. J. Geol. Soc. Lond. 154, 45–54 (1997).

12. Zhang, R. Y. & Liou, J. G. Partial transformation of gabbro to coesite-bearing eclogites from Yangkou,

the Sulu terrane, eastern China. J. Metamorph. Geol. 15, 183–202 (1997).

13. Anderson, S. & O’Keeffe, M. Body-centered cubic cylinder packing and the garnet structure. Nature

267, 605–606 (1977).

14. Sobolev, N. V., Lavrent’ev, G. & Yu, G. Isomorphic sodium admixture in garnets formed at high

pressure. Contrib. Mineral. Petrol. 31, 1–12 (1971).

15. Moore, R. O. & Gurney, J. J. Pyroxene solid solution in garnets included in diamond. Nature 318, 553–

555 (1985).

16. Haggerty, S. E. & Sautter, V. Ultradeep (greater than 300 kilometers), ultramafic upper mantle

xenoliths. Science 248, 993–996 (1990).

17. Ringwood, A. E. & Lovering, J. F. Significance of pyroxene-ilmenite intergrowths among kimberlite

xenoliths. Earth Planet. Sci. Lett. 7, 371–375 (1970).

18. Tompsom, R. N. Is upper-mantle phosphorus contained in sodic garnet? Earth Planet. Sci. Lett. 26,

417–424 (1975).

19. Schertl, H. P., Schreyer, W. & Chopin, C. The pyrope-coesite rocks and their country rocks at Parigi,

Dora Maira Massif, Western Alps: detailed petrography, mineral chemistry and P-T path. Contrib.

Mineral. Petrol. 108, 1–21 (1991).

20. Ono, S. & Yasuda, A. Compositional change of majoritic garnet in a MORB composition from 7 to 17

GPa and 1400 to 1600 oC. Physics Earth Planet. Inter. 96, 171–179 (1996).

21. Okamoto, K. & Maruyama, S. Multi-anvil re-equilibration experiments of a Dabie Shan ultrahigh-

pressure eclogite within the diamond-stability fields. Island Arc 7, 52–69 (1998).

22. Zhang, R. Y., Liu, J. G. & Cong, B. 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 (1995).

23. Widiyantoro, S. & Hist, R. Structure and evolution of lithospheric slab beneath the Sunda Arc,

Indonesia. Science 271, 1566–1570 (1996).

24. Amstrong, R. L. Radiogenic isotopes: the case for crustal recycling on a near-steady state no-

continental growth Earth. Phil. Trans. R. Soc. Lond. A 301, 443–472 (1981).

25. Dupre, B. & Allegre, C. Pb-Sr isotopic variations in Indian Ocean basalts and mixing phenomena.

Nature 303, 142–146 (1983).

26. Loubet, M., Sassi, R. & Di Donato, G. Mantle heterogeneities: a combined isotope and trace element

approach and evidence for recycled continental crust materials in some OIB sources. Earth Planet. Sci.

Lett. 89, 299–315 (1988).

27. Ye, K. et al. Large areal extent of ultrahigh-pressure (UHP) metamorphism in the Sulu ultrahigh-

pressure terrane of East China: new implications from coesite and omphacite inclusions in zircon of

granitic gneiss. Lithos 52, 157–164 (2000).

28. Green, H. W., Dobrzhinetskaya, L. & Bozhilov, K. N. Response to ‘‘Determining the origin of

ultrahigh-pressure lherzolites. By Hacker, B. R., Sharp, T., Zhang, R. Y., Liou, J. G. & Hervig, R. L.’’

Science 278, 704–707 (1997).

Supplementary Information is available on Nature’s World-Wide Web site(http://www.nature.com) or as paper copy from the London editorial office of Nature.

AcknowledgementsWe thank P. Xu for programming microprobe analyses, and H. W. Green for commentsthat significantly improved this Letter. This work was supported by the National NaturalScience Foundation of China

Correspondence and requests for materials should be addressed to K.Y.(e-mail: yekai@hotmail.com).

letters to nature

736 NATURE | VOL 407 | 12 OCTOBER 2000 | www.nature.com

.................................................................The population genetics of ecologicalspecialization in evolvingEscherichia coli populationsVaughn S. Cooper & Richard E. Lenski

Center for Microbial Ecology, Michigan State University, East Lansing,Michigan 48824, USA

................. ......................... ......................... ......................... ......................... .........................

When organisms adapt genetically to one environment, they maylose fitness in other environments1–4. Two distinct populationgenetic processes can produce ecological specialization—muta-tion accumulation and antagonistic pleiotropy5–8. In mutationaccumulation, mutations become fixed by genetic drift in genesthat are not maintained by selection; adaptation to one environ-ment and loss of adaptation to another are caused by differentmutations. Antagonistic pleiotropy arises from trade-offs, suchthat the same mutations that are beneficial in one environmentare detrimental in another. In general, it is difficult to distinguishbetween these processes5–8. We analysed the decay of unusedcatabolic functions in 12 lines of Escherichia coli propagated onglucose for 20,000 generations9,10. During that time, several linesevolved high mutation rates11. If mutation accumulation is impor-tant, their unused functions should decay more than the otherlines, but no significant difference was observed. Moreover, mostcatabolic losses occurred early in the experiment when beneficialmutations were being rapidly fixed, a pattern predicted byantagonistic pleiotropy. Thus, antagonistic pleiotropy appearsmore important than mutation accumulation for the decay ofunused catabolic functions in these populations.

We founded 12 populations from a strain of E. coli, and they alladapted to minimal medium supplemented with glucose9,10.Competitive fitness increased rapidly in the first few thousandgenerations. Fitness continued to improve, but the average rate of

© 2000 Macmillan Magazines Ltd

Supplementary Figure Caption:

Backscattered electron imagine showing high concentration of clinopyroxene (dark grey) rods in

garnet.