Journal of Petrology

This Article Abstract FREE Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (17) Request Permissions Google Scholar Articles by BRANDT, S. Articles by OKRUSCH, M. Search for Related Content GeoRef GeoRef Citation Social Bookmarking          What's this?

Journal of Petrology | Volume 44 | Number 6 | Pages 1121-1144 | 2003
© Oxford University Press 2003

Ultrahigh-Temperature Metamorphism and Multistage Evolution of Garnet–Orthopyroxene Granulites from the Proterozoic Epupa Complex, NW Namibia

SÖNKE BRANDT^*, REINER KLEMD and MARTIN OKRUSCH

MINERALOGISCHES INSTITUT, UNIVERSITÄT WÜRZBURG, AM HUBLAND, 97074 WÜRZBURG, GERMANY

Telephone: ++49-(0)931-8885406. Fax: ++49-(0)931-8884620. E-mail: soenke.brandt{at}mail.uni-wuerzburg.de

RECEIVED FEBRUARY 1, 2002; ACCEPTED DECEMBER 17, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION REGIONAL SETTING LITHOLOGY OF THE EPUPA... PETROGRAPHY OF THE SEMIPELITIC... MINERAL CHEMISTRY REACTION HISTORY P-T CONDITIONS DISCUSSION AND CONCLUSION REFERENCES

 
Migmatitic semipelitic granulites of the Proterozoic Epupa Complex, NW Namibia, underwent ultrahigh-temperature metamorphism as is indicated by the high alumina contents of orthopyroxene (8–11 wt % Al₂O₃) coexisting with garnet. Peak P–T conditions of 970°C and 9·5 kbar are calculated from conventional garnet–orthopyroxene geothermobarometry. Conspicuous reaction textures document a multistage retrograde uplift–cooling path: post-peak decompression initially under still ultrahigh temperatures (940°C and 8 kbar) is recorded by coronas of aluminous orthopyroxene + plagioclase around garnet. Continued decompression (6 kbar and 800°C) is evident from subsequently formed symplectites of cordierite + lower-alumina orthopyroxene and cordierite + lower-alumina orthopyroxene + spinel, both replacing garnet. Subsequent regrowth of garnet and biotite, mainly formed at the expense of the symplectitic phases, presumably reflects back-reactions with crystallizing melts during near-isobaric cooling to upper amphibolite-facies conditions (660°C and 5 kbar). Rims of low-alumina orthopyroxene around retrograde biotite point to renewed decompression subsequent to cooling. The deduced clockwise retrograde P–T path reflects the thinning and later cooling of former thickened lower crust. Because of the limited structural and geochronological data it remains uncertain whether initial ultrahigh-temperature metamorphism was induced by a collision event or by crustal extension.

KEY WORDS: Namibia; geothermobarometry; granulite; P–T path; ultrahigh-temperature metamorphism

	INTRODUCTION

TOP ABSTRACT INTRODUCTION REGIONAL SETTING LITHOLOGY OF THE EPUPA... PETROGRAPHY OF THE SEMIPELITIC... MINERAL CHEMISTRY REACTION HISTORY P-T CONDITIONS DISCUSSION AND CONCLUSION REFERENCES

 
In recent years an increasing number of granulite-facies terranes have been recognized in which ultrahigh-temperature (UHT) crustal metamorphism with temperatures in the range of 900–1050°C at moderate pressures of 7–13 kbar is recorded [reviewed by Harley (1989, 1998a)]. Based on well-constrained theoretical and experimental data [reviewed by Hensen & Harley (1990), Bertrand et al. (1991) and Harley (1998a)] mineral assemblages in rocks of highly aluminous and magnesian bulk composition (MgAl-rich granulites) are the key indicators of such extreme metamorphic conditions (e.g. Harley, 1998a). In contrast, conventional geothermobarometry, which is mainly based on Fe–Mg exchange reactions, generally yields erroneous low temperatures as the mineral compositions are commonly affected and modified by post-peak diffusional cation exchange (e.g. Frost & Chacko, 1989; Harley, 1989, 1998a; Fitzsimons & Harley, 1994).

This study details the petrology, mineral chemistry and reaction history of garnet + orthopyroxene-bearing meta-semipelitic gneisses from a newly recognized Proterozoic UHT terrane in the geologically poorly studied Epupa Complex of NW Namibia (Fig. 1). As a result of their large grain size (up to 5 cm in diameter) and favourable reaction kinetics, garnets in these granulites partly preserve their peak-metamorphic mineral composition, and therefore UHT conditions are retrieved from conventional geothermobarometry. Well-preserved reaction textures are used to deduce a retrograde P–T path, which is characterized by post-peak decompression followed by cooling.

View larger version (80K): [in this window] [in a new window]   Fig. 1. (a) Location of the Epupa Complex within the Pan-African Damara orogen of Namibia [modified from Miller (1983)]. The box shows the area of (b). (b) Simplified geological map of the studied part of the Epupa Complex [modified from Köstlin (1967) and Menge (1998)] showing the occurrence of UHT granulites-facies rocks (Epembe Unit) and upper amphibolite-facies rocks (Orue Unit) and the locations of the microprobe samples.

 
Our study represents the first detailed petrological investigation of the high-grade rocks of the Epupa Complex, which is interpreted to represent the southwestern margin of the Archaean to Palaeoproterozoic Congo craton (Tegtmeier & Kröner, 1985). Therefore, the P–T path recorded in the semipelitic granulites will place important constraints on the tectonothermal evolution of the Congo craton during Proterozoic times.

	REGIONAL SETTING

TOP ABSTRACT INTRODUCTION REGIONAL SETTING LITHOLOGY OF THE EPUPA... PETROGRAPHY OF THE SEMIPELITIC... MINERAL CHEMISTRY REACTION HISTORY P-T CONDITIONS DISCUSSION AND CONCLUSION REFERENCES

 
The Epupa Complex (EC) forms a pre-Pan-African basement complex that is situated at the eastern margin of the Pan-African Kaoko belt of NW Namibia (e.g. South African Committee for Stratigraphy, 1980; Fig. 1a) and extends into southern Angola where it is termed the ‘Schist, quartzite and amphibolite Complex’ (de Carvalho & Alves, 1990). The western part of the EC was affected by the Pan-African event and was partly incorporated in the Pan-African mobile belt (e.g. Dingeldey et al., 1994). In contrast, the northeastern portion of the EC, which forms the studied area, was a stable part of the Congo craton at this time. In this region the rocks of the EC are unconformably overlain by undeformed Neoproterozoic sediments of the Damara Supergroup (Fig. 1a and b).

The EC comprises a sequence of high-grade metamorphic para- and orthogneisses of varying ages. A U–Pb zircon age of 1795 +33/-29 Ma for a syntectonic granitic gneiss from Ruacana (Tegtmeier & Kröner, 1985), situated at the easternmost margin of the EC (Fig. 1a), correlates well with Palaeoproterozoic ages between 1700 and 2000 Ma from other basement complexes of Namibia and southern Angola (see Tegtmeier & Kröner, 1985; Seth et al., 1998; Franz et al., 1999), such as the Kamanjab Inlier (Fig. 1a). These ages indicate that the whole region was affected by a major crust-forming magmatic and metamorphic event (Tegtmeier & Kröner, 1985), which corresponds to the Eburnian Orogeny known from western, central and eastern Africa (e.g. Cahen et al., 1984). Mainly on the basis of these ages, Tegtmeier & Kröner (1985) suggested that the region was part of the Congo craton during pre-Pan-African times and was partly reworked during the Pan-African event. In clear contrast, Mesoproterozoic U–Pb zircon ages of 1480–1390 Ma are reported for granitic gneisses and amphibolites of the studied part of the EC (Allsopp & Burger; in Menge, 1998). These Mesoproterozoic ages are consistent with Pb–Pb garnet ages of 1490–1447 Ma obtained for the granulites of our studied area, which are interpreted as the peak-metamorphic event (Seth et al., 2001, submitted). U–Pb zircon studies on the same granulite samples yielded Palaeoproterozoic ages (1810–1635 Ma) for zircon cores and Mesoproterozoic ages (1520–1510 Ma) for zircon rims (Seth et al. in preparation). The Mesoproterozoic ages are interpreted to date prograde zircon growth during high-grade metamorphism that affected Palaeoproterozoic protoliths (Seth et al., in preparation). The core ages suggest that the protoliths of the studied granulites were part of the Congo craton during the Palaeoproterozoic Eburnian Orogeny. In clear contrast, Mesoproterozoic ages of 1500 Ma have as yet not been recognized from any other high-grade metamorphic belts of southern Africa, and thus point to an isolated and unique metamorphic event in the investigated field area.

In the studied area the basement rocks are intruded by anorthositic rocks of the Kunene Intrusive Complex (KIC; Fig. 1b), which represents one of the largest massif-type anorthosite bodies in the world (Ashwal & Twist, 1994; Drüppel et al., 2001). The anorthosites contain lenses of the surrounding basement and show no evidence for a tectonometamorphic overprint (Drüppel et al., 2001), which clearly indicates that their emplacement post-dates the high-grade metamorphism observed in the rocks of the EC. Two generations of anorthosites are recognized: older ‘white anorthosites’ are intruded by younger ‘dark anorthosites’ (Drüppel et al., 2001). U–Pb zircon ages of 1385 ± 25 Ma for the younger ‘dark anorthosites’ (Drüppel et al., 2000b) are consistent with U–Pb zircon ages of 1370 ± 4 Ma for a cogenetic mangeritic vein from the Angolan part of the KIC (Mayer et al., 2000). A contact thermal effect induced by the emplacement of the younger ‘dark anorthosites’ is restricted to a narrow reaction zone up to several metres in width and is recorded by undeformed Grt–Crd–Sil rocks [mineral abbreviations after Kretz (1983)]. Radiogenic ages for the older ‘white anorthosites’ have not been reported so far, but the ages for the ‘dark anorthosites’ provide a minimum age for their emplacement. The anorthosites, in turn, are intruded by undeformed granitic rocks termed ‘Red Granites’ with Rb–Sr whole-rock ages between 1400 and 1300 Ma (de Carvalho et al., 1987). Similar granites have been recognized in the studied area at the southern margin of the KIC, where they form a broadly ENE–WSW-trending intrusion (Fig. 1b). Subsequently, syenitic and mafic intrusions were emplaced into the basement rocks and the anorthosites, mainly along broadly east–west-trending and more rarely along NE–SW-trending faults (Fig. 1b). U–Pb zircon ages of 1216 and 1213 Ma were recently obtained for the emplacement of the syenites and nepheline syenites (Littmann et al., 2000). To the south and west, the rocks of the EC are unconformably overlain by non-metamorphosed Neoproterozoic clastic sediments of the Damara Supergroup (Fig. 1b), indicating that the studied part of the EC was not affected by the Pan-African tectonothermal event.

	LITHOLOGY OF THE EPUPA COMPLEX

TOP ABSTRACT INTRODUCTION REGIONAL SETTING LITHOLOGY OF THE EPUPA... PETROGRAPHY OF THE SEMIPELITIC... MINERAL CHEMISTRY REACTION HISTORY P-T CONDITIONS DISCUSSION AND CONCLUSION REFERENCES

 
In the investigated area at the southern margin of the KIC, the EC can be subdivided into two distinct metamorphic units displaying significant differences in metamorphic grade (Fig. 1b):

1. the EC mainly consists of upper amphibolite-facies ortho- and paragneisses, which are grouped together as the ‘Orue Unit’;
   
2. ultrahigh-temperature granulites have been recognized in a broadly east–west-trending terrane of limited areal extent (50 km x 10 km) named the ‘Epembe Unit’.
   

These units are separated by broadly east–west- and NE–SW-trending, subvertical ductile shear zones, along which the mafic granulites of the Epembe Unit are retrogressed to fine-grained, mylonitic amphibolites. To the west, the Epembe Unit is covered by soil; therefore, a western continuation is possible. Both basement units are transected by cataclastic WNW–ESE faults that post-date the formation of the amphibolite-facies shear zones and even displace the young syenite intrusions.

The upper amphibolite-facies Orue Unit is dominated by migmatitic granitic orthogneisses (Fig. 1b). Their plutonic precursors intruded a volcano-sedimentary succession that is exposed in the eastern part of the Orue Unit. This sequence predominantly consists of migmatitic meta-semipelites with intercalations of migmatitic metapelites, meta-arkoses, meta-quartzites and rare calc-silicate felses. The metasediments were intruded by mafic dykes, which were metamorphosed to garnet-bearing amphibolites. Granulite-facies relics have not been observed in the rocks of the Orue Unit; consequently a derivation as a result of retrogression of former granulites of the Epembe Unit is unlikely. The occurrence of abundant granitic leucosomes in the paragneisses and the peak-metamorphic muscovite-free Grt–Sil–Kfs–Bt–Qtz assemblages of the metapelites indicate crossing of the reaction Ms + Qtz Kfs + Sil + L, which points to temperatures >725°C [at 6–7 kbar and a(H₂O) = 1]. In clear contrast to the rocks of the Epembe Unit (see below), the mafic and felsic rocks of the Orue Unit never contain orthopyroxene; therefore these rocks are ascribed to the upper amphibolite facies. Preliminary results of geothermobarometric calculations cluster at temperatures of 680–760°C (garnet–biotite Fe–Mg geothermometry) at pressures of 6–7 kbar [garnet–aluminosilicate–quartz–plagioclase (GASP) geobarometry] throughout the Orue Unit.

The Epembe Unit comprises a variety of granulite-facies rock types. Its eastern and northwestern part is dominated by massive to weakly banded mafic two-pyroxene granulites with rare garnet (Opx + Cpx + Pl ± Grt); felsic orthogneisses (Opx ± Grt + Pl + Qtz) of enderbitic composition are locally interlayered whereas paragneisses are absent (Fig. 1b). In contrast, the central part of the Epembe Unit is dominated by orthopyroxene-bearing orthogneisses of charnockitic to enderbitic composition (Opx ± Grt + Pl ± Kfs + Qtz). Concordantly interlayered mafic orthogneisses are similar to those of the eastern and northwestern part and are essentially quartz-free two-pyroxene granulites with sporadic garnet (Opx + Cpx + Pl ± Grt). Intercalated paragneisses are dominated by relatively Fe-rich (bulk-rock Mg-number 0·39–0·47) quartz-bearing or quartz-free Grt–Sil–Kfs metapelites that locally preserve spinel–quartz associations. The more magnesian and less aluminous garnet + orthopyroxene-bearing meta-semipelites (bulk-rock Mg-number 0·52–0·56) discussed in this study appear as concordant layers of up to 20 m in width and show gradual contacts towards the metapelites. The meta-semipelites contain rare lenses of sapphirine-bearing MgAl-rich granulites that preserve peak-metamorphic UHT assemblages of Opx + Sil ± Qtz ± Grt ± Spr (Brandt et al., 2001, in preparation). All these lithotypes show a planar foliation or banding that dips subvertically to the north. They are intruded by mafic dykes that cross-cut the regional foliation at a low angle and have been transformed to two-pyroxene granulites or garnet-bearing enderbites.

	PETROGRAPHY OF THE SEMIPELITIC GRANULITES

TOP ABSTRACT INTRODUCTION REGIONAL SETTING LITHOLOGY OF THE EPUPA... PETROGRAPHY OF THE SEMIPELITIC... MINERAL CHEMISTRY REACTION HISTORY P-T CONDITIONS DISCUSSION AND CONCLUSION REFERENCES

 
The coarse-grained semipelitic granulites preserve peak-metamorphic assemblages of garnet + orthopyroxene + quartz + plagioclase ± K-feldspar ± biotite with minor amounts of ilmenite, rutile, zircon and rare apatite. Mineral inclusions are attributed to the prograde evolution, whereas abundant and well-preserved reaction textures yield evidence for a multistage retrograde evolution.

Stage 1: prograde evolution; Stage 2: peak-metamorphic assemblages
The meta-semipelites show a characteristic migmatitic texture. The original compositional banding is modified by leucocratic layers and streaks, which are composed of garnet, orthopyroxene, plagioclase, quartz and rare K-feldspar. They are interpreted as leucosomes derived from partial melting during the prograde evolution. The leucosomes are commonly concordant to the compositional banding; in rare cases they cross-cut the banding. The melanocratic layers and lenses comprise a restitic mineralogy that is dominated by orthopyroxene and garnet. Peak-metamorphic garnet, orthopyroxene, rutile and, if preserved, biotite are aligned and define a weak foliation (S₂) concordant to the regional foliation and banding, indicating mineral growth before, or synchronous with, ductile deformation (D₂). The highest-grade mineral assemblages are commonly dominated by distinct megacrystic garnet porphyroblasts (Grt2) up to 5 cm in diameter that are set in a granoblastic quartz–feldspar matrix. The porphyroblasts are always surrounded by broad corona textures (see below). Grt2 coexists with porphyroblastic orthopyroxene (Opx2; up to 6 mm in diameter) as is indicated by locally preserved adjacent straight grain boundaries. Plagioclase (Pl2) and perthitic K-feldspar of the matrix appear as isolated and anhedral grains that show local recrystallization to fine-grained neoblasts at grain boundaries. Early biotite (Bt2) is rare in the quartz–feldspar matrix and is isolated from garnet and orthopyroxene; therefore a retrograde formation is unlikely. Bt2 is aligned parallel to the regional S₂ foliation and is therefore interpreted as a relic of the prograde evolution. The dominant Ti phase of the matrix is rutile, but minor ilmenite (Ilm2) may occur.

The peak-metamorphic minerals preserve rare mineral inclusions that are ascribed to the prograde metamorphic assemblages. The inclusions show no significant preferred orientation; thus, evidence for an early foliation pre-dating the external S₂ foliation is not preserved in the meta-semipelites. Porphyroblastic Grt2 contains rare inclusions of corroded biotite (Bt1) and irregular trails of fibrolitic and prismatic sillimanite (Sil1), which is absent from the matrix. Further inclusions in Grt2 are rutile, quartz, ilmenite (Ilm1) and plagioclase (Pl1). Porphyroblastic Opx2 encloses biotite (Bt1), plagioclase (Pl1) and quartz. In places, inclusion-free garnet (Grt1) is enclosed as sub- to euhedral grains (<0·25 mm) in Opx2 cores (Fig. 2), suggesting virtually contemporaneous growth of both phases. The inclusions indicate that early Bt–Sil–Pl–Qtz assemblages were replaced by the coarse-grained peak-metamorphic Grt–Opx assemblages.

View larger version (179K): [in this window] [in a new window]   Fig. 2. Photomicrograph illustrating prograde evolution of the meta-semipelitic granulites. Subhedral inclusion of Grt1 with partly preserved straight grain boundaries against hosting porphyroblastic Opx2. Grt1 is partly replaced by Crd–Opx symplectites. Zoning profile of the Grt1 inclusion is given in Fig. 6a.

 

View larger version (29K): [in this window] [in a new window]   Fig. 6. Zoning profiles of garnet. Each profile extends from rim to rim through the core of the garnet grains. (a) Grt1 inclusion in porphyroblastic Opx2 (see Fig. 2). (b) Coarse-grained (5 mm) porphyroblastic Grt2 with Grt4 regrowth rims (see Fig. 4d) and medium-grained (1·5 mm) porphyroblastic Grt2. (c) Euhedral Grt4 (see Fig. 4c). [Note larger scale of (a) and (c).]

 
Stage 3: corona and symplectite formation
Peak-metamorphic Grt2 is extensively replaced by conspicuous symplectite and corona textures (Fig. 3a–e). All replacement textures are essentially undeformed, indicating that their formation post-dates the development of the S₂ foliation. Complex corona textures are developed between Grt2 and matrix quartz (Fig. 3a). An outer relatively coarse-grained collar consists of orthopyroxene (Opx3a) and granoblastic plagioclase (Pl3a). The inner part of the corona in contact to Grt2 is composed of a double-layer symplectitic intergrowth of orthopyroxene (Opx3b) and cordierite (Fig. 3a and b). These Crd–Opx symplectites probably formed subsequently to the outer Opx–Pl collar, as they are localized on individual garnet grains whereas the Opx–Pl coronas envelop several garnet grains (Fig. 3a). The inner part of the Crd–Opx symplectite is formed by a dactylitic Crd–Opx intergrowth that is dominated by cordierite (Fig. 3b). The outer part is predominantly or completely composed of orthopyroxene (Fig. 3b). The Crd–Opx symplectites may be dispersed by fine-grained plagioclase (Pl3b) and by late ilmenite (Ilm3), and even occur in cracks present in the outer parts of Grt2. In contrast to the scarce Opx–Pl collars, the Crd–Opx symplectites are present in all studied semipelitic granulites partially or completely replacing Grt2. Cracks in Grt2 cores are filled by a very fine-grained symplectitic intergrowth consisting of orthopyroxene (Opx3c), cordierite and green spinel (Fig. 3c). Symplectitic orthopyroxene and spinel commonly appear as lamellar intergrowths. Similar Opx–Spl symplectites are also formed between porphyroblastic Opx2 and Grt2 (Fig. 3d). In addition, symplectites composed of cordierite and green spinel that are surrounded by a monomineralic cordierite corona may occur in Grt2 (Fig. 3e). Preserved Sil1 inclusions in Grt2 suggest formation of Crd–Spl symplectites through local replacement of former sillimanite inclusions. Between porphyroblastic Opx2 and matrix quartz a monomineralic corona of fine-grained granoblastic cordierite is developed, which locally displays straight grain boundaries against Opx2 (Fig. 3f).

View larger version (191K): [in this window] [in a new window]   Fig. 3. Photomicrographs illustrating reaction textures resulting from the breakdown of peak-metamorphic Grt2 and Opx2 (stage 3). (a) Composite corona separating Grt2 from Qtz. The outer Opx–Pl collar surrounds several Grt2 fragments, indicating initial garnet breakdown according to the reaction Grt + Qtz Opx + Pl. The inner Crd–Opx symplectite is localized on individual Grt2 fragments, indicating subsequent garnet breakdown according to the reaction Grt + Qtz Crd + Opx. (b) Detail of the Crd–Opx symplectite resorbing Grt2. The finger-like Crd–Opx symplectite dominated by cordierite is rimmed by an Opx–Crd intergrowth that is dominated by orthopyroxene. (c) Crd–Opx–Spl symplectite replacing Grt2 along cracks in garnet cores, indicating the reaction Grt Crd + Opx + Spl. (d) Lamellar Opx–Spl symplectite between Grt2 and Opx2. (e) Crd–Spl symplectite rimmed by a corona of cordierite replacing former Sil1 inclusions in Grt2, indicating the reaction Grt + Sil + Qtz Crd followed by the reaction Grt + Sil Crd + Spl. (f ) Monomineralic corona of cordierite separating Opx2 from Qtz, indicating the reaction high-Al Opx + Qtz low-Al Opx + Crd. Cordierite is partly replaced by late Bt4.

 
Stage 4: regrowth of biotite and garnet
The Crd–Opx symplectites developed in stage 3 are in turn surrounded and replaced by broad rims of platy biotite (Bt4; Fig. 4a). In addition, Bt4 appears as monomineralic rims or as Bt–Qtz symplectites around porphyroblastic and coronitic orthopyroxene (Fig. 4b). Locally, Bt4 is intergrown with plagioclase. Although Bt4 occurs throughout the restitic domains, it is most extensively formed along the margins of the leucosomes, suggesting that biotite regrowth is related to interaction of the crystallizing melt with the minerals of the restitic layers. In more ferroan samples, the Crd–Opx symplectites are partially replaced by a second generation of garnet (Grt4), which forms fine-grained euhedral crystals (<0·3 mm) within the Crd–Opx pseudomorphs after Grt2 (Fig. 4c). Regrown Grt4 preserves inclusions of symplectitic orthopyroxene (Fig. 4c) and/or cordierite that clearly indicate regrowth of garnet at the expense of the symplectitic phases. Grt4 furthermore appears as narrow regrown rims on relict porphyroblastic Grt2, locally forming straight grain boundaries against Crd–Opx symplectites or enclosing the symplectitic phases, which preserve their original orientation (Fig. 4d).

View larger version (185K): [in this window] [in a new window]   Fig. 4. Photomicrographs illustrating reaction textures resulting from the regrowth of garnet and biotite (stage 4). (a) Rim of platy Bt4 around Crd–Opx symplectite pseudomorphing former porphyroblastic Grt2. Bt4, in turn, is partly rimmed by granoblastic Opx5. (b) Bt4–Qtz symplectite replacing porphyroblastic Opx2. (c) Euhedral Grt4 present in Crd–Opx symplectites after former Grt2. Grt4 preserves inclusions of symplectitic Opx3b that indicate late-stage garnet growth according to the reaction Crd + Opx Grt + Qtz. Zoning profile of Grt4 is given in Fig. 6c. (d) Grt4 regrowth rim on porphyroblastic Grt2. Grt4 encloses symplectitic Opx3b with the same orientation as Opx3b in the external Crd–Opx symplectite.

 
Stage 5: formation of late orthopyroxene and cordierite
Retrograde Bt4, in turn, is locally overgrown by fine-grained granoblastic orthopyroxene (Opx5), which forms monomineralic corona textures around Bt4 (Figs 4a and 5a) or occurs as very fine-grained intergrowth with cordierite (Fig. 5b). The textures indicate that the development of Opx5 post-dates the reappearance of biotite.

View larger version (93K): [in this window] [in a new window]   Fig. 5. Photomicrographs illustrating reaction textures resulting from the formation of late orthopyroxene and cordierite (stage 5). (a) Bt4 rimmed by a corona of granoblastic Opx5. (b) Platy Bt4 partly replacing the Crd–Opx symplectite. Bt4, in turn, is resorbed and rimmed by an intergrowth of Opx5 and Crd.

 

	MINERAL CHEMISTRY

TOP ABSTRACT INTRODUCTION REGIONAL SETTING LITHOLOGY OF THE EPUPA... PETROGRAPHY OF THE SEMIPELITIC... MINERAL CHEMISTRY REACTION HISTORY P-T CONDITIONS DISCUSSION AND CONCLUSION REFERENCES

 
All analyses were performed with a CAMECA SX50 electron microprobe at the University of Würzburg. Operating conditions were 15 kV accelerating voltage and 15 nA beam current with 1 µm beam size for garnet, orthopyroxene, cordierite, spinel and plagioclase, and 5 µm for biotite. Representative analyses of garnet, orthopyroxene, cordierite, spinel, biotite and plagioclase are presented in Tables 1–6 and illustrated in Figs 6–9. Sample locations are given in Fig. 1b. Variably high Fe³⁺ values for garnet and orthopyroxene (calculated from ideal stoichiometry) probably reflect systematic underestimation of Si.

View this table: [in this window] [in a new window]   Table 1: Representative analyses of garnet

 

View this table: [in this window] [in a new window]   Table 2: Representative analyses of orthopyroxene

 

View this table: [in this window] [in a new window]   Table 3: Representative analyses of cordierite (total iron as Fe2+)

 

View this table: [in this window] [in a new window]   Table 4: Representative analyses of biotite (total iron as Fe2+)

 

View this table: [in this window] [in a new window]   Table 5: Representative analyses of plagioclase

 

View this table: [in this window] [in a new window]   Table 6: Representative analyses of spinel

 

View larger version (24K): [in this window] [in a new window]   Fig. 7. Compositional variations of XMg and Altot in orthopyroxene depending on its textural occurrence.

 

View larger version (23K): [in this window] [in a new window]   Fig. 8. Compositional variations in cordierite depending on its textural occurrence.

 

View larger version (33K): [in this window] [in a new window]   Fig. 9. Phase relationships in the meta-semipelitic granulites of the Epupa Complex. (a) A–S–FM projection from feldspars indicating bulk-rock compositions of the meta-semipelitic granulites (open symbols) of the Epembe Unit and illustrating mineral reactions implied by reaction textures discussed in the text. (b–d) A–F–M projections from quartz and feldspars for three representative samples of the meta-semipelitic granulites of the Epembe Unit. Arrows indicate rimward zoning in garnet and orthopyroxene. Initial assemblages are joined by continuous lines; Grt2(rim)–Crd–Opx–Qtz assemblages by dashed lines. For reasons of clarity, compositions of Crd–Opx–Spl symplectites and cordierite coronas around Opx2 have been excluded. Source of illustrated mineral chemical data is Tables 1–6.

 
Garnet
Garnet of the meta-semipelites is essentially a pyrope–almandine solid solution with low spessartine and grossular component (Table 1). Significant chemical variations, especially in X_Mg [= Mg/(Mg + Fe²⁺)], are observed for garnet of distinct generations (Fig. 6a–c). Prograde Grt1 inclusions (<0·25 mm) preserved in cores of porphyroblastic Opx2 reveal minor zoning (Fig. 6a) with decreasing pyrope content from core (X_Mg = 0·35; Prp₃₃Alm₆₂Grs₂Sps₂) to rim (X_Mg = 0·33; Prp₃₁Alm₆₄Grs₃Sps₂). The inclusions are probably affected by significant retrograde Fe–Mg exchange with the Opx2 host, as is indicated by their relatively low X_Mg when compared with matrix Grt2 core composition. The highest pyrope contents for matrix Grt2 were obtained in core plateaux (X_Mg = 0·44–0·46; Prp_42–44Alm_52–54Grs_3–4Sps_1–2), which are interpreted to result from intracrystalline diffusional homogenization during the highest-grade metamorphism (Fig. 6b). Significant rimward zoning to lower pyrope content in Grt2 (X_Mg = 0·29–0·39) presumably developed during the retrograde formation of the surrounding Crd–Opx symplectites, as Mg was preferentially partitioned into symplectitic orthopyroxene (X_Mg = 0·52–0·66) and cordierite (X_Mg = 0·70–0·86). The preservation of core plateaux depends on the grain size of Grt2, as they are preserved only in grains of >4 mm in diameter. Smaller grains are completely modified (Fig. 6b). The grossular component shows minor but continuous zoning to lower Ca content from core (Grs_3–4) to rim (Grs_1–2). Late euhedral Grt4, present in Crd–Opx symplectites (Fig. 4c), is significantly less magnesian (X_Mg = 0·23–0·24; Prp₂₂ Alm_71–72Grs₄Sps₃) than porphyroblastic Grt2 and, in contrast to the latter, almost unzoned (Fig. 6c). The grossular content is in the range of those of late Grt4 regrowth rims (X_Mg = 0·28–0·32; Prp_27–30Alm_65–68Grs_3–4Sps₂) on relic porphyroblastic Grt2 (Fig. 6b).

Orthopyroxene
Orthopyroxene is present in distinct textural domains and shows systematic compositional variations especially concerning the Al content (Table 2; Fig. 7a and b). The highest X_Mg and Al contents are preserved in cores of porphyroblastic Opx2 (Al^tot = 0·52–0·42 cations p.f.u.; X_Mg = 0·65–0·70) present in Grt2-absent domains. Opx2 coexisting with Grt2 is slightly less magnesian than Opx2 in Grt2-absent domains but has similar high Al contents in the cores (Al^tot = 0·45–0·39 cations p.f.u.; X_Mg = 0·61–0·65). Such high Al content in orthopyroxene coexisting with garnet is a characteristic feature of UHT granulites (Harley, 1998a). In both textural domains Opx2 displays significant rimward zoning to less aluminous composition and lower X_Mg (Al^tot = 0·30–0·12 cations p.f.u.; X_Mg = 0·59–0·62), suggesting release of the Mg-Tschermak's component (MgAl₂SiO₆) towards the adjacent reaction domains (Fig. 7a and b). Opx3a in Opx–Pl corona textures around Grt2 is less magnesian (X_Mg = 0·54–0·56) than Opx2 (Fig. 7b) presumably because it has grown from the relatively Fe-rich Grt2 rims. Al in Opx3a is generally high (Al^tot = 0·38–0·26 cations p.f.u.) but lower than in Opx2 cores, which testifies to falling, but still very high temperatures during corona formation. Symplectitic orthopyroxene is less aluminous than coronitic orthopyroxene (Fig. 7a and b), which indicates that symplectite formation proceeded at lower temperatures, supporting the interpretation that the symplectites formed subsequent to the outer Opx–Pl corona. The high X_Mg of orthopyroxene in Crd–Opx–Spl symplectites (Opx3c; Al^tot = 0·24–0·15 cations p.f.u.; X_Mg = 0·63–0·70) probably reflects symplectite formation at the expense of the relatively Mg-rich Grt2 cores. Orthopyroxene in Crd–Opx symplectites (Opx3b) has similar Al contents but is significantly less magnesian (Al^tot = 0·25–0·12 cations p.f.u.; X_Mg = 0·52–0·66), as it has grown from the relatively Fe-rich Grt2 rims. The highest X_Mg values are obtained for Opx3b in direct contact with Grt2 rims, suggesting Fe–Mg re-equilibration during later cooling. Late-stage corona-forming Opx5 replacing retrograde Bt4 is characterized by the lowest Al content (Al^tot = 0·16–0·09 cations p.f.u.; X_Mg = 0·52–0·64). To summarize, orthopyroxene shows a systematic decrease in Al content that correlates with its textural and temporal occurrence [matrix Opx2 (cores) > matrix Opx2 (rims) coronitic Opx3a > symplectitic Opx3b and 3c > late Opx5]. Variations in the X_Mg are mainly controlled by the composition of coexisting garnet in the various reaction domains.

Cordierite
Cordierite always has the highest X_Mg of the coexisting ferromagnesian symplectitic or coronitic phases, with variable compositions (X_Mg = 0·70–0·89) depending on its textural occurrence (Table 3; Fig. 8). In places, cordierite contains significant amounts of Na₂O (up to 0·5 wt %). Low total sums of 97–99 wt % point to the presence of H₂O or CO₂ fluids during cordierite growth. Like orthopyroxene (see above), cordierite in Crd–Opx–Spl symplectites (X_Mg = 0·81–0·87) is more magnesian than cordierite in Crd–Opx symplectites (X_Mg = 0·70–0·86), as it has grown from the relatively Mg-rich Grt2 cores. Partly, cordierite in the Crd–Opx symplectites is strongly zoned to lower X_Mg with distance from Grt2 (Fig. 8) as a result of retrograde Fe–Mg re-equilibration with Grt2. The preservation of such zonation depends on the width of the Crd–Opx corona, as it occurs only in sample 614-1, which has relatively broad coronas of 0·5–1 mm in width. In the more narrow coronas of the other samples cordierite is homogenized. Cordierite (X_Mg = 0·80–0·81) in Crd–Spl symplectites replacing former Sil1 inclusions in Grt2 and corona-forming cordierite (X_Mg = 0·79–0·80) surrounding Opx2 are unzoned.

Biotite
Matrix Bt2 is less magnesian and has higher Ti (X_Mg = 0·63–0·72; Ti = 0·43–0·63 cations p.f.u.) than late Bt4 formed at the expense of the Crd–Opx symplectites (X_Mg = 0·65–0·74; Ti = 0·39–0·57 cations p.f.u.; Table 4), suggesting that Bt2 was not, or only weakly, affected by retrograde Fe–Mg exchange. Bt4 in Bt–Qtz symplectites or Bt–Pl–Qtz intergrowths replacing Opx2 is even more magnesian (X_Mg = 0·71–0·78; Ti = 0·27–0·60 cations p.f.u.). The high fluorine content (1·1–2·6 wt % F) of Bt4 shows no variation depending on the textural occurrence.

Plagioclase
Prograde plagioclase inclusions (Pl1) in porphyroblastic Grt2 preserve the highest Ca contents, which decrease slightly from core (An₅₆) to rim (An₅₃; Table 5). Peak-metamorphic matrix plagioclase (Pl2) is significantly less calcic (An_27–42), with a smaller compositional range in each sample that results from a weak zonation with rimward increase of the An content of 4–5 mol %. Late plagioclase in Opx–Pl coronas (Pl3a: An_28–31) and in Crd–Opx symplectites (Pl3b: An_40–44) is unzoned and has a similar composition to matrix Pl2 of the same samples (Table 5).

Spinel
Spinel is always the most Fe-rich symplectitic phase (Table 6). Analysed spinel is essentially a hercynite–spinel solid solution with low gahnite (<3·5 mol %) and chromite (<1 mol %). Magnetite calculated from ideal stoichiometry is <2·5 mol %. Spinel in Crd–Opx–Spl symplectites is more magnesian (X_Mg = 0·38–0·47) than in Crd–Spl symplectites (X_Mg = 0·31–0·35).

Ilmenite
Early ilmenite (Ilm1) enclosed in Grt2 or present in the quartz–feldspar matrix (Ilm2) has higher MnO contents (0·7–2·2 wt % MnO) than late ilmenite (Ilm3) that occurs in the Crd–Opx symplectites (0·3–0·4 wt % MnO).

	REACTION HISTORY

TOP ABSTRACT INTRODUCTION REGIONAL SETTING LITHOLOGY OF THE EPUPA... PETROGRAPHY OF THE SEMIPELITIC... MINERAL CHEMISTRY REACTION HISTORY P-T CONDITIONS DISCUSSION AND CONCLUSION REFERENCES

 
The conspicuous reaction textures preserved in the meta-semipelitic granulites of the Epembe Unit are attributed to a multistage metamorphic evolution as evident from the application of A–S–FM and A–F–M diagrams (Fig. 9a–d) and the FMAS grid for high-grade metapelites (Fig. 10).

View larger version (35K): [in this window] [in a new window]   Fig. 10. P–T path of the meta-semipelitic granulites of the Epupa Complex, as deduced from mineral assemblages and reaction textures combined with the results of geothermobarometric calculations explained further in the text and in Fig. 11. The petrogenetic grid for high-grade metapelites in the system FMAS is based on Hensen & Green (1973), Hensen (1987), Hensen & Harley (1990) and Bertrand et al. (1991). Also shown is the fluid-absent dehydration melting reaction Bt + Pl + Qtz Opx + Grt/Spl/Crd + Kfs + L (Vielzeuf & Montel, 1994). The numbers on the P–T path refer to reactions discussed in the text.

 

View larger version (28K): [in this window] [in a new window]   Fig. 11. P–T plots showing the results of geothermobarometric calculations for the peak-metamorphic conditions and the retrograde stages of the metamorphic evolution applied to three meta-semipelitic granulites of the Epupa Complex. Source of data is Table 7. CH, Carswell & Harley (1989); LG, Lee & Ganguly (1988); NP, Newton & Perkins (1982); M, Moecher et al. (1988); HP1, Alm + Qtz Fs + Fe-Crd, Holland & Powell (1998); HP2, Prp + Qtz En + Mg-Crd, Holland & Powell (1998); IM, Indares & Martignole (1985); KR, Kleemann & Reinhardt (1994).

 
Stage 1: prograde evolution; Stage 2: peak-metamorphic assemblages
Inclusions of biotite, plagioclase, quartz and sillimanite in porphyroblastic Grt2 suggest progress of the fluid-absent dehydration-melting reaction

(1)

during prograde heating through the stability field of sillimanite. The unorientated appearance of the inclusions indicates static conditions during heating. The lack of sillimanite inclusions in porphyroblastic Opx2 suggests that orthopyroxene formed after the complete consumption of matrix sillimanite, exhausted by reaction (1), and consequently after initial garnet growth. This interpretation is in agreement with the presence of garnet inclusions in Opx2 and explains the lack of coexisting Opx–Sil–Qtz assemblages. Inclusions of biotite, plagioclase and quartz in Opx2 point to growth of orthopyroxene, accompanied by continued garnet growth, according to the fluid-absent dehydration-melting reaction

(2)

(Vielzeuf & Montel, 1994). The deduced reaction sequence, with orthopyroxene formed subsequently to initial garnet growth, is consistent with the observation that reaction (1) occurs at lower temperatures than reaction (2) (Montel & Vielzeuf, 1997). Progress of reactions (1) and (2) resulted in the almost complete replacement of matrix biotite and furthermore explains the production of melt that is recorded by the presence of garnet- and/or orthopyroxene-bearing leucosomes. According to fluid-absent biotite-dehydration melting experiments for an aluminous and moderately magnesian (bulk rock Mg-number 0·49) metagreywacke (Vielzeuf & Montel, 1994), similar in composition to our studied semipelitic granulites (bulk rock Mg-number 0·52–0·56), reaction (2) takes place at temperatures between 890 and 990°C (at 10 kbar; stippled field in Fig. 10) and produces garnet and highly aluminous orthopyroxene (Montel & Vielzeuf, 1997) similar in composition to high-Al Opx2 of the studied granulites. The resulting peak-metamorphic Grt–Opx–Pl–Kfs–Qtz assemblages of the meta-semipelitic granulites correlate well with the bulk-rock compositions (Fig. 9a). The presence of garnet in our samples indicates relatively high pressures, because at pressures <5 kbar cordierite instead of garnet would be formed (Vielzeuf & Montel, 1994). The weakly rimward decreasing Ca content in Grt2 (Fig. 6b) corresponds to a rimward increasing Ca content in matrix Pl2 (Table 5) and indicates slightly decreasing pressures during growth of garnet in the present Grt–Opx–Pl–Qtz assemblage. The presence of matrix biotite in only one of the studied meta-semipelites (sample 614-1) suggests peak temperatures close to the upper thermal stability limit of biotite. The preservation of this early biotite is probably related to its high Ti content (up to 5·5 wt % TiO₂), which may extend the thermal stability of biotite (e.g. Stevens et al., 1997).

Stage 3a: corona formation
The peak-metamorphic mineral assemblages are modified by well-preserved reaction textures that are diagnostic of decompression under ultrahigh temperatures in the first stage.

Initial breakdown of Grt2 is recorded by the formation of Opx–Pl coronas between Grt2 and quartz (Fig. 3a) that document progress of the reaction

(3)

According to Harley (1989), this reaction is indicative of near-isothermal decompression (near-ITD). The high Al content of coronitic Opx3a (up to 9 wt % Al₂O₃) suggests that corona formation took place under still ultrahigh temperatures.

Cordierite coronas between porphyroblastic Opx2 and quartz (Fig. 3f ) may suggest progress of the divariant FMAS reaction Opx + Sil + Qtz Crd. However, as coexisting Opx–Sil–Qtz assemblages were not present (see above), cordierite formation is interpreted to result from the release of the Mg-Tschermak's component (MgTs) of orthopyroxene according to the model MAS reaction (Fig. 9a)

(4)

or

which is consistent with rimward zoning to lower Al and X_Mg in porphyroblastic Opx2 (Fig. 7).

Stage 3b: symplectite formation
Subsequent symplectite formation proceeded under lower, but still granulite-facies conditions as suggested by the lower Al content of symplectitic Opx3b and 3c when compared with coronitic Opx3a.

Crd–Opx–Spl symplectites that occur in cracks present in Grt2 cores (Fig. 3c) are interpreted to result from progress of the divariant FMAS reaction (Fig. 9a)

(5)

Reaction (5) requires the coexistence of Crd + Opx + Spl, which is stable on the low-pressure side of the univariant FMAS reaction Grt + Spr Opx + Crd + Spl (Fig. 10). Furthermore, reaction (5) must occur at temperatures above and pressures below the univariant FMAS reaction Opx + Sil Grt + Crd + Spl, as it requires the coexistence of Grt + Crd + Spl. Therefore, progress of reaction (5) documents decompression to P <8 kbar at high temperatures (Fig. 10).

Crd–Opx symplectites replacing Grt2 coexisting with matrix quartz (Fig. 3a and b) are interpreted to result from the divariant FMAS reaction (Fig. 9a)

(6)

which explains rimward zoning of Grt2 to an Mg-poorer composition (Fig. 9b–d).

Crd–Spl symplectites enclosed in Grt2, which are surrounded by monomineralic cordierite coronas separating the symplectite from the garnet host (Fig. 3e), are interpreted to result from the breakdown of former Sil1 inclusions in Grt2 according to the following divariant FMAS reactions (Fig. 9a):

(7)

(8)

Influx of SiO₂, probably along garnet cracks, may have resulted in the formation of the outer cordierite coronas at initially high SiO₂ activity. Consumption of the available SiO₂ and/or separation of sillimanite from the high-SiO₂ activity domains by the cordierite coronas resulted in localized low-SiO₂ activity domains and thus initiated the growth of the Crd–Spl symplectites on the former sillimanite. Crd–Spl assemblages are stable on the low-pressure side of the univariant FMAS reaction Grt + Sil + Spr Crd + Spl (Fig. 10) and therefore progress of reaction (8) documents decompression to P < 8 kbar.

Similar to reaction (3) of stage 3a, reactions (5)–(8) of stage 3b have a shallow dP/dT slope and are therefore consistent with continued near-ITD or heating. The latter process is ruled out by the lowered Al₂O₃ content of symplectitic orthopyroxene when compared with coronitic orthopyroxene.

Stage 4: regrowth of biotite and garnet
The formation of Bt4 at the expense of the Crd–Opx symplectites (Fig. 4a) and Bt4–Qtz symplectites surrounding porphyroblastic Opx2 (Fig. 4b) is in agreement with progress of the hydration reactions

(9)

(10)

The extensive formation of Bt4 at leucosome–melanosome contacts suggests that biotite regrowth is related to reactions of the crystallizing anatectic melt with the minerals of the restitic layers. This interpretation is supported by the presence of plagioclase that is locally intergrown with Bt4. Crystallization of the melt resulted in the release of H₂O fluids, which reacted with orthopyroxene and/or cordierite to produce biotite. This process could also explain the high F content of Bt4 (Table 4) that was probably induced by F-bearing fluids. Progress of reactions (9) and (10) is consistent with cooling to temperatures below 800°C (Vielzeuf & Montel, 1994; Fig. 10).

Texturally late Grt4, present in Crd–Opx symplectites (Fig. 4c) and as rims on relic porphyroblastic Grt2 (Fig. 4d), encloses the symplectitic phases, which points to garnet regrowth according to reaction (6) in a reverse direction

(11)

consistent with a cooling-dominated P–T path at this stage.

Stage 5: formation of late orthopyroxene and cordierite
Rims of low-Al Opx5 locally intergrown with cordierite around retrograde Bt4 (Fig. 5a and b) indicate late-stage, melt-absent biotite dehydration according to the KFMASH reactions

(12)

(13)

Orthopyroxene regrowth may be related to continued decompression under melt-absent and fluid-absent conditions, as formerly present H₂O fluids were exhausted by the biotite-forming reactions (9) and (10) or may result from renewed heating. The latter possibility seems rather unlikely because Opx5 has lower Al contents than symplectitic Opx3.

	P–T CONDITIONS

TOP ABSTRACT INTRODUCTION REGIONAL SETTING LITHOLOGY OF THE EPUPA... PETROGRAPHY OF THE SEMIPELITIC... MINERAL CHEMISTRY REACTION HISTORY P-T CONDITIONS DISCUSSION AND CONCLUSION REFERENCES

 
The first evidence for the P–T conditions attained during the metamorphic evolution of the meta-semipelitic granulites is given by the observed mineral assemblages, reaction textures and mineral compositions. The high Al content of peak-metamorphic orthopyroxene (8–11 wt % Al₂O₃) indicates UHT condition with temperatures >900°C (Harley, 1998a). UHT conditions are furthermore supported by the prograde biotite-dehydration reactions and the rare preservation of prograde biotite, which point to peak temperatures between 890 and 990°C at relatively high pressures (see above). The retrograde evolution is dominated by decompression to P <8 kbar as mainly evident from the formation of Crd–Opx–Spl and Crd–Spl symplectites replacing peak-metamorphic garnet.

The preservation of the peak-metamorphic mineral compositions and the abundant reaction textures provide an excellent possibility for the application of widely used geothermobarometers. Thus, a detailed documentation of the physical conditions attained during the peak-metamorphic event and during the different stages of the retrograde evolution is possible. P–T calculations were performed with mineral chemical data given in Tables 1–6 for three representative samples. The results are summarized in Table 7 and Fig. 11.

View this table: [in this window] [in a new window]   Table 7: Representative results of conventional geothermobarometry

 
P–T results calculated from conventional geothermobarometry are compared for the individual stages with P–T estimates obtained from the application of isopleth diagrams that were contoured for the composition of garnet and orthopyroxene in the assemblage Grt–Opx–Sil–Qtz at higher pressures and the assemblage Grt–Opx–Crd–Qtz at lower pressures. The isopleth diagrams of Hensen & Harley (1990) and of Harley (1998a), the latter based on the experiments of Carrington & Harley (1995), have been contoured for X_Al and X_Mg in orthopyroxene and X_Mg in coexisting garnet. The diagram of Aranovich & Berman (1996) has been contoured for X_Prp in garnet and for Al₂O₃ in orthopyroxene. In all diagrams X_Mg in garnet is an excellent pressure indicator as the X_Mg isopleths are very shallow in P–T space. In contrast, the Al content of orthopyroxene is an excellent temperature indicator as the Al-isopleths are very steep in P–T space. Thus, the systematic decrease in the Al content of orthopyroxene in the various textural domains of the studied meta-semipelitic granulites (Fig. 7) indicates that post-peak decompression was accompanied by a significant temperature drop.

Stage 2: peak-metamorphic conditions
Peak-metamorphic conditions were calculated using the core compositions of coexisting aluminous Opx2, Pl2 and Grt2. Garnet–orthopyroxene Fe–Mg exchange geothermometry (Lee & Ganguly, 1988; Carswell & Harley, 1989; Bhattacharya et al., 1991) yields ultrahigh temperatures between 930 and 1010°C at a reference pressure of 9·5 kbar. The values for sample 634 are minimum temperatures as Grt2 core plateaux are not preserved. The ultrahigh garnet–orthopyroxene Fe–Mg temperatures are supported by the results of garnet–orthopyroxene Al-geothermometry (Harley & Green, 1982; Aranovich & Berman, 1997), which range between 1000 and 1070°C. Furthermore, garnet–biotite Fe–Mg exchange geothermometry (Indares & Martignole, 1985; Bhattacharya et al., 1992; Kleemann & Reinhardt, 1994) performed with matrix Bt2 and Grt2 cores yields extreme temperatures of 870–980°C. The calculated ultrahigh temperatures are consistent with the very high Al contents of porphyroblastic Opx2 (Al₂O₃ = 8–11 wt %; Al^VI = 0·15–0·19 cations p.f.u.) coexisting with garnet. Application of the isopleth diagrams of Hensen & Harley (1990) and of Aranovich & Berman (1996) described above yields ultrahigh temperatures of 980–1050°C, whereas from the isopleth diagram of Harley (1998a) slightly lower temperatures of 920–950°C are obtained.

Corresponding pressures were calculated with the garnet–orthopyroxene–plagioclase–quartz geobarometer using the calibrations of Newton & Perkins (1982) and Eckert et al. (1991) for the Mg-endmember reaction and of Bohlen et al. (1983) and Moecher et al. (1988) for the Fe-endmember reaction. Calculated pressures are between 7·5 and 11·5 kbar at a reference temperature of 970°C, with the lower estimates obtained from the Mg-endmember reaction. Best-fit peak-metamorphic P–T conditions of 9·5 ± 2 kbar and 970 ± 40°C are in agreement with the lack of Opx–Sil–Qtz assemblages and the temperature range (890–990°C) obtained from the biotite-dehydration melting reactions (Fig. 10).

Stage 3a: corona formation
Initial breakdown of garnet is documented by the formation of the Opx–Pl coronas around Grt2 that indicate post-peak decompression. Grt–Opx Fe–Mg geothermometry (calibrations as above) performed with coronitic Opx3a coupled with Grt2 rim compositions yields ultrahigh temperatures of 910–970°C, close to the peak-metamorphic conditions and in agreement with the high Al content of coronitic Opx3a (6–9 wt % Al₂O₃). UHT conditions are consistent with Grt–Opx Al-geothermometry (calibrations as above), which yields temperatures of 1000°C. Corresponding Grt–Opx–Pl–Qtz pressures (calibrations as above) of 6·5–10·0 kbar are 1·5 kbar lower than peak-metamorphic pressures and indicate that initial garnet breakdown proceeded in response to decompression to 8 ± 2 kbar under still ultrahigh temperatures of 940 ± 30°C.

Stage 3b: symplectite formation
P–T conditions for subsequent symplectite development were estimated using Grt2 rim composition and the composition of symplectitic orthopyroxene, cordierite, spinel and plagioclase. Calculations with the Crd–Opx–Spl symplectites that replace Grt2 along cracks were performed with composition of adjacent Grt2. Temperatures calculated from Grt–Opx Fe–Mg and Al-geothermometry (calibrations as above) mainly cluster around 800°C for both the Crd–Opx–Spl and the Crd–Opx symplectites. Partially lower temperatures (660–830°C) were calculated from Grt–Crd Fe–Mg geothermometry (Bhattacharya et al., 1988; Dwivedi et al., 1998) and probably reflect intense retrograde Fe–Mg re-equilibration of cordierite with Grt2 that is evident from the cordierite zonation (Fig. 8). Consequently, even Grt–Opx Fe–Mg geothermometry provides minimum temperatures for the symplectite formation, as they were calculated with the modified Grt2 rim composition.

Grt–Opx–Pl–Qtz pressures (calibrations as above) range between 5·0 and 8·0 kbar at a reference temperature of 800°C. The location of the Fe- and Mg-endmember reactions of the divariant reactions Grt + Qtz Opx + Crd and Grt + Sil Crd + Spl was calculated with the program THERMOCALC v.2.75 of Powell & Holland (1988) using the updated internally consistent dataset of Holland & Powell (1998) and the activity models of Berman (1990) for garnet, of Wood & Banno (1973) for orthopyroxene, and an ideal mixing model for anhydrous cordierite. Because of the presence of cordierite, the location of the reactions strongly depends on the water activity. The low total sums of cordierite (97–99 wt %) suggest the presence of H₂O or CO₂ fluids during symplectite formation, but because of the lack of suitable fluid inclusions the water activity could not be estimated. Therefore the results given in Table 7 and Fig. 11 were calculated for an intermediate value of a(H₂O) of 0·5. Thus calculated pressures of 4·5–7·5 kbar are consistent with the Grt–Opx–Pl–Qtz pressures (see above). Fluid-absent conditions would lower the pressures by the order of 1 kbar whereas calculations with pure H₂O fluids would increase the pressures by the order of 1 kbar.

The P–T results calculated from conventional geothermobarometry are consistent with P–T estimates obtained from the application of the isopleth diagrams of Hensen & Harley (1990) and Aranovich & Berman (1996) contoured for the Grt–Opx–Crd–Qtz assemblage that yields P–T conditions of 750–850°C at 5–6 kbar and of 770–920°C at 6–7 kbar, respectively. Best-fit P–T conditions of 800 ± 60°C at 6 ± 2 kbar for the formation of the symplectites indicate subsequent decompression accompanied by cooling from stage 3a.

Stage 4: regrowth of garnet and biotite
P–T conditions for the regrowth of garnet were calculated using core composition of euhedral regrown Grt4 and the composition of Grt4 regrowth rims on porphyroblastic Grt2 combined with the composition of adjacent symplectitic cordierite and orthopyroxene. Temperatures calculated from Grt–Opx Fe–Mg and Grt–Crd Fe–Mg geothermometry (calibrations as above) are between 630 and 700°C. Grt–Opx Al-geothermometry (calibrations as above) yields higher temperatures (720–830°C), suggesting that the Al₂O₃ content of orthopyroxene was not re-equilibrated during cooling. Another possible explanation for the lower Fe–Mg temperatures is provided by retrograde Fe–Mg exchange that affected the garnet–orthopyroxene pairs. As regrown Grt4 is almost unzoned (see Fig. 6c) the latter possibility seems rather unlikely and the Fe–Mg exchange temperatures are interpreted to be more realistic. Pressures of 3·0–6·0 kbar, calculated from the Grt–Opx–Crd–Qtz barometer (method as above), are consistent with estimates obtained from the isopleth diagrams of Hensen & Harley (1990) and Aranovich & Berman (1996) for the present Grt–Opx–Crd–Qtz assemblage. Based on the X_Mg of regrown Grt4, pressures of <5 kbar and 4·5–5·5 kbar, respectively, are estimated. Grt–Bt geothermometry (calibrations as above), using Grt2 rim compositions coupled with Bt4 present in the Crd–Opx symplectites, yields temperatures of 600–670°C that are broadly consistent with the results of Grt–Opx and Grt–Crd Fe–Mg geothermometry. Best-fit P–T conditions of 660 ± 40°C at 5 ± 2 kbar indicate that garnet and biotite regrowth proceeded in response to near-isobaric cooling of 140°C from stage 3b.

Stage 5: formation of late orthopyroxene and cordierite
Application of the isopleth diagrams of Hensen & Harley (1990) and Aranovich & Berman (1996) yields temperatures of <700°C for the growth of late low-Al Opx5 (2–3 wt % Al₂O₃). The temperatures are in the range of those calculated for stage 4 and support the interpretation that late orthopyroxene growth is related to continued decompression at low water activity rather than to renewed heating.

	DISCUSSION AND CONCLUSION

TOP ABSTRACT INTRODUCTION REGIONAL SETTING LITHOLOGY OF THE EPUPA... PETROGRAPHY OF THE SEMIPELITIC... MINERAL CHEMISTRY REACTION HISTORY P-T CONDITIONS DISCUSSION AND CONCLUSION REFERENCES

 
The P–T path
Based on textural observations, mineral chemistry and geothermobarometric calculations, a multistage P–T evolution is deduced for the garnet + orthopyroxene-bearing meta-semipelitic granulites of the Epupa Complex, which is illustrated in Fig. 10. The resulting P–T path is a combination of post-peak decompression followed by near-isobaric cooling and a second stage of decompression, resulting in a clockwise retrograde P–T evolution.

Prograde heating proceeded through the sillimanite stability field and resulted in the replacement of biotite and sillimanite through dehydration melting reactions by the coarse-grained peak-metamorphic garnet + Al-rich orthopyroxene assemblages. As a result of the preservation of the highest-grade mineral composition, conventional geothermobarometry yields UHT metamorphic conditions (970 ± 40°C; 9·5 ± 2 kbar) for the formation of the peak-metamorphic assemblages. Subsequent decompression led to the development of a sequence of conspicuous undeformed corona and symplectite textures. Early-stage near-ITD (stage 2 stage 3a) to 8 ± 2 kbar proceeded under still UHT conditions of 940 ± 30°C, as is evident from coronas of aluminous orthopyroxene + plagioclase growing at the expense of garnet and quartz. Further decompression (stage 3a stage 3b) was accompanied by significant cooling to 800 ± 60°C at 6 ± 2 kbar, mainly recorded by the development of delicate Crd–Opx symplectites formed at the expense of garnet and quartz. Uplift from lower-crustal depths (29 km) to mid-crustal levels (18 km) probably proceeded along the subvertical shear zones that surround the UHT granulites. Subsequent cooling (stage 3b stage 4) to upper amphibolite conditions (660 ± 40°C at 5 ± 2 kbar) is evident from the regrowth of biotite and garnet, formed at the expense of the symplectitic phases. The formation of late, low-alumina, orthopyroxene at the expense of retrograde biotite indicates a second stage of decompression at temperatures of <700°C (stage 4 stage 5). The continuity of the deduced reaction sequence suggests that the inferred retrograde P–T path from stage 2 to stage 5 is the result of a single Mesoproterozoic event, rather than of two or more tectonometamorphic events. This interpretation is consistent with the lack of Pan-African tectonism in the studied area. However, more detailed geochronological data, mainly on the retrograde phases, are needed to support this interpretation. The retrograde P–T path of the semipelitic granulites of the Epupa Complex is remarkably similar to those documented for sapphirine-bearing MgAl-rich UHT granulites from Mather Peninsula (Harley, 1998b) and Forefinger Point (Harley et al., 1990) of Antarctica and from the Palni Hills, southern India (Raith et al., 1997).

The preservation of UHT conditions
UHT metamorphic conditions, as documented for the meta-semipelitic granulites of the Epembe Unit, have been reported from an increasing number of granulite terranes worldwide in recent years [reviewed by Harley (1998a)]. The recognition of UHT conditions is mainly based on phase relationships in sapphirine-bearing MgAl-rich granulites. With rare exceptions, such as in Grt–Opx gneisses from Mather Peninsula, Antarctica (Harley, 1998b), UHT conditions are not recovered by conventional geothermometry, because of post-peak diffusional Fe–Mg exchange (e.g. Harley, 1998a; Frost & Chacko, 1989). Even though we have recognized sapphirine-bearing MgAl-rich granulites in the Epupa Complex (Brandt et al., 2001) we concentrate on garnet–orthopyroxene granulites in this study. These rocks have a more common semipelitic bulk-rock chemistry but are exceptional in their preservation of the highest-grade mineral compositions. As a consequence, UHT conditions are retrieved from conventional geothermobarometry without the necessity of back-calculation of mineral compositions to account for post-peak Fe–Mg exchange (Fitzsimons & Harley, 1994; Harley, 1998b). According to Harley (1998a) the retrieval of UHT conditions by Fe–Mg geothermometry is mainly favoured by (1) Mg-rich garnet compositions, (2) the lack of Fe–Mg phases adjacent to garnet, (3) large garnet grain size and (4) a rapid cooling rate. Because the peak-metamorphic garnet of the semipelitic granulites of the Epupa Complex is of intermediate (X_Mg 0·44–0·46) rather than magnesian composition and is surrounded by symplectitic Fe–Mg phases, the first two options are not very likely to explain the preservation of the highest-grade mineral compositions. Presumably the partially large grain size of analysed garnet (up to 5 mm) is the main factor to account for the preservation of the peak-metamorphic mineral compositions. Furthermore, a rapid cooling rate, which may be inferred from the rather unusual preservation of significant cordierite zoning (Fig. 8), could explain the limited retrograde diffusive Fe–Mg exchange between garnet and the surrounding Fe–Mg symplectite phases that resulted in the preservation of the core plateaux in the largest garnet grains.

Tectonic implications
The tectonic setting necessary for the formation of UHT granulites is still under discussion (e.g. Harley, 1998a), although it is generally accepted that UHT metamorphism requires an anomalous thermal input into the lower crust (e.g. Raith et al., 1997). Only in a few cases were possible heat sources constrained: UHT metamorphism in the Palni Hills, southern India, is interpreted to result from the emplacement of voluminous enderbitic intrusive rocks (Raith et al., 1997) whereas UHT metamorphism in the South Harris granulite belt of the Lewisian of Scotland is correlated with the emplacement of anorthosites and related magmatic rocks (Baba, 1999). In addition, Sengupta et al. (1999) documented UHT granulites from the Eastern Ghats Belt, southern India, which they interpreted to result from the accretion of voluminous basic melts.

The tectonic significance of the P–T path deduced for the semipelitic granulites of the Epupa Complex is ambiguous. Because of the lack of sufficient geochronological data and as evidence for the prograde evolution is poorly preserved, two possible geodynamic models are discussed in the following to account for UHT metamorphism and the clockwise retrograde evolution. The significant post-peak decompression of 4 kbar, consistent with an uplift of >10 km, requires a former overthickened crust. The two tectonic models mainly differ in the timing of this crustal thickening event, as described below.

(1) Crustal thickening is related to a collision event between 1520 and 1450 Ma that resulted in the formation of the UHT granulites. Post-collisional collapse of the overthickened crust resulted in post-peak decompression followed by late cooling of the rapidly uplifted crust to the stable geotherm. As the time span between 1660 and 1400 Ma in southern Africa has generally been interpreted as a period of tectonic stability (e.g. Cahen et al., 1984; Olson, 2000), this model would imply the first recognition of a Mesoproterozoic collision event in this region restricted to the Epupa Complex. Following this, the Epupa Complex has to be reinterpreted as an exotic terrane accreted to the Congo craton after the 1500 Ma collision event. However, this scenario seems rather unlikely, as (a) the EC displays Palaeoproterozoic protolith ages (Seth et al., submitted) similar to the surrounding basement complexes and (b) younger accretion ages have not been constrained. The unmetamorphosed Neoproterozoic sediments in the studied area demonstrate that the region was not involved in the Pan-African orogeny. Furthermore, it has to be mentioned that typical collision structures, such as thrusting or large-scale folding, were not observed. Last but not least, feasible heat sources for the UHT metamorphism, such as Mesoproterozoic syn-collisional intrusive rocks, are not recognized in the studied area.

(2) Regarding the close spatial relationship between Mesoproterozoic anorthosites and UHT granulites in the Epupa Complex, a genetic connection between the two units may be considered (Drüppel et al., 2000a). This would imply that both the KIC and the EC are of similar age. Unfortunately, the age for the initial formation of the magma chamber that acted as source for the anorthosites is uncertain, so far. The emplacement age of 1380 Ma for the younger ‘dark anorthosites’ provides a minimum age, as the ‘dark anorthosites’ intrude older, but yet undated, ‘white anorthosites’ (Drüppel et al., 2001). Huge massif-type anorthosite bodies such as the KIC are commonly interpreted to be emplaced in extensional settings [see Ashwal (1993) for a review]. Following this, crustal thickening must be related to an older orogenic event such as the Eburnian Orogeny (2100–1660 Ma), which has indeed been documented for the different lithologies of the EC (Tegtmeier & Kröner, 1985; Seth et al., 1998, submitted; Franz et al., 1999). The thickened crust became unstable in the Mesoproterozoic and extensional tectonics commenced that may have resulted in the replacement of lithosphere by asthenosphere (e.g. Bird, 1979; Houseman & England, 1986; Sandiford & Powell, 1986). Magmatic accretion of mantle-derived basic melts at the base of the lower crust may have caused the formation of the UHT granulites in the lower crust. Subsequent uplift and near-isothermal decompression under still granulite-facies conditions could be related to persisting extension contemporaneous with the continued introduction of mantle-derived melts. Late cooling may have resulted from thermal relaxation of the rapidly uplifted crust to the former stable geotherm.

As a result of the limited geochronological and structural data and the isolated occurrence of the UHT granulites, it remains uncertain whether the deduced P–T evolution of the meta-semipelitic granulites is related to tectonic model (1) or model (2). Therefore, further petrological investigations will concentrate on the prograde evolution of the granulites, which may be the key for understanding of the tectonometamorphic evolution of the UHT granulites. Furthermore, additional geochronological and structural data are needed.

	ACKNOWLEDGEMENTS

 
We are grateful to K. Drüppel, A. Zeh and B. Seth for helpful discussions. We also thank U. Schüssler for his advice and help with the electron microprobe work, and P. Späthe for his excellent preparation of the thin sections. S. Harley and I. Fitzsimons are thanked for their constructive and helpful reviews. S. Harley is furthermore thanked for providing us with a correct version of the Aranovich & Berman (1997) Grt–Opx Al-thermometer formulation. This study is an outcome of the Postgraduate Research Programme ‘Interdisciplinary Geoscience Research in Africa’ at the University of Würzburg supported by the Deutsche Forschungsgemeinschaft (grant GRK 144/3).

	REFERENCES

TOP ABSTRACT INTRODUCTION REGIONAL SETTING LITHOLOGY OF THE EPUPA... PETROGRAPHY OF THE SEMIPELITIC... MINERAL CHEMISTRY REACTION HISTORY P-T CONDITIONS DISCUSSION AND CONCLUSION REFERENCES

 
Aranovich, L. Y. & Berman, R. G. (1996). Optimized standard state and solution properties of minerals. II. Comparisons, predictions, and applications. Contributions to Mineralogy and Petrology 126, 25–37.[CrossRef][Web of Science]

Aranovich, L. Y. & Berman, R. G. (1997). A new garnet–orthopyroxene thermometer based on reversed Al₂O₃ solubility in FeO–Al₂O₃–SiO₂ orthopyroxene. American Mineralogist 82, 345–353.[Abstract]

Ashwal, L. D. (1993). Anorthosites. Minerals and Rocks, 21. Berlin: Springer, 422 pp.

Ashwal, L. D. & Twist, D. (1994). The Kunene complex, Angola/Namibia: a composite massif-type anorthosite complex. Geological Magazine 131, 579–591.[Web of Science]

Baba, S. (1999). Sapphirine-bearing orthopyroxene–kyanite/sillimanite granulites from South Harris, NW Scotland: evidence for Proterozoic UHT metamorphism in the Lewisian. Contributions to Mineralogy and Petrology 136, 33–47.[CrossRef][Web of Science]

Berman, R. G. (1990). Mixing properties of Ca–Mg–Fe–Mn garnets. American Mineralogist 75, 328–344.[Abstract]

Bertrand, P., Ellis, D. J. & Green, D. H. (1991). The stability of sapphirine–quartz and hypersthene–sillimanite–quartz assemblages: an experimental investigation in the system FeO–MgO–Al₂O₃–SiO₂ under H₂O and CO₂ conditions. Contributions to Mineralogy and Petrology 108, 55–71.[CrossRef][Web of Science]

Bhattacharya, A., Mazumdar, A. C. & Sen, S. K. (1988). Fe–Mg mixing in cordierite: constraints from natural data and implications for cordierite–garnet thermometry in granulites. American Mineralogist 3, 338–344.

Bhattacharya, A., Krishnapur, K. R., Raith, M. & Sen, S. K. (1991). An improved set of a–X parameters for Fe–Mg–Ca garnets and refinements of the orthopyroxene–garnet thermometer and the orthopyroxene–garnet–plagioclase–quartz barometer. Journal of Petrology 32, 629–656.[Abstract/Free Full Text]

Bhattacharya, A., Mohanty, L., Maji, A., Sen, S. K. & Raith, M. (1992). Non-ideal mixing in the phlogopite–annite binary: constraints from experimental data on Mg–Fe partitioning and a reformulation of the biotite–garnet geothermometer. Contributions to Mineralogy and Petrology 111, 87–93.[CrossRef][Web of Science]

Bird, P. (1979). Continental delamination and the Colorado Plateau. Journal of Geophysical Research 104, 398–416.

Bohlen, S. R., Wall, V. J. & Boettcher, A. L. (1983). Geobarometry in granulites. In: Saxena, S. K. (ed.) Kinetics and Equilibrium in Mineral Reactions. Advances in Physical Geochemistry 3, 141–172.

Brandt, S., Will, T. M. & Klemd, R. (2001). Metamorphic evolution of sapphirine-bearing orthopyroxene–sillimanite gneisses and associated felsic granulites from the Proterozoic Epupa Complex, NW Namibia. EUG 2001, Abstract Volume, p. 571.

Cahen, L., Snelling, N. J., Delhal, J. & Yail, J. R. (1984). The Geochronology and Evolution of Africa. Oxford: Clarendon Press, 512 pp.

Carrington, D. P. & Harley, S. L. (1995). Partial melting and phase relations in high-grade metapelites: an experimental petrogenetic grid in the KFMASH system. Contributions to Mineralogy and Petrology 120, 270–291.[Web of Science]

Carswell, D. A. & Harley, S. L. (1989). Mineral thermometry and barometry. In: Carswell, D. A. (ed.) Eclogite Facies Rocks. Glasgow: Blackie, pp. 83–110.

de Carvalho, H. & Alves, P. (1990). Gabbro–anorthosite complex of SW-Angola/NW Namibia. Instituto de Investigacão Cientifica Tropical, Serie Ciencias da Terra, Lisboa, Communicacoes 2, 1–66.

de Carvalho, H., Crasto, J., Silva, Z. C. & Vialette, Y. (1987). The Kibarian cycle in Angola: a discussion. Geological Journal 22, 85–102.[Web of Science]

Dingeldey, D. P., Dürr, S. B., Charlesworth, E. G., Franz, L., Okrusch, M. & Stanistreet, I. G. (1994). A traverse through the northern coastal branch of the Damara Orogen west of Sesfontein, Namibia. Journal of African Earth Science 19, 315–329.

Drüppel, K., Brandt, S. & Okrusch, M. (2000a). Evolution of the anorthositic Kunene Intrusive Complex and the metamorphic rocks of the Epupa Complex, NW Namibia: evidence for cogenetic formation during Proterozoic crustal extension? GEOLUANDA 2000, Abstract Volume, pp. 47–48.

Drüppel, K., Littmann, S. & Okrusch, M. (2000b). Geo- und isotopenchemische Untersuchungen der Anorthosite des Kunene-Intrusiv-Komplexes (KIC) in NW Namibia. Beihefte zum European Journal of Mineralogy 12, 37.

Drüppel, K., von Seckendorff, V. & Okrusch, M. (2001). Subsolidus reaction textures in the anorthositic rocks of the southern part of the Kunene Intrusive Complex, NW Namibia. European Journal of Mineralogy 13, 289–309.[Abstract/Free Full Text]

Dwivedi, S. B., Mohan, A. & Lal, R. K. (1998). Recalibration of the Fe–Mg exchange reaction between garnet and cordierite as a thermometer. European Journal of Mineralogy 10, 281–289.[Abstract/Free Full Text]

Eckert, J. O., Newton, R. C. & Kleppa, O. J. (1991). The H of reaction and recalibration of garnet–pyroxene–plagioclase–quartz geobarometer in the CMAS system by solution calorimetry. American Mineralogist 76, 148–160.[Abstract]

Fitzsimons, I. C. W. & Harley, S. L. (1994). The influence of retrograde cation exchange on granulite P–T estimates and a convergence technique for the recovery of peak metamorphic conditions. Journal of Petrology 35, 543–576.[Abstract/Free Full Text]

Franz, L., Romer, R. L. & Dingeldey, D. P. (1999). Diachronous Pan-African granulite-facies metamorphism (650 Ma and 550 Ma) in the Kaoko Belt, NW Namibia. European Journal of Mineralogy 11, 167–180.[Abstract/Free Full Text]

Frost, B. R. & Chacko, T. (1989). The granulite uncertainty principle; limitations on thermobarometry in granulites. Journal of Geology 97, 435–450.[Web of Science]

Harley, S. L. (1989). The origins of granulites: a metamorphic perspective. Geological Magazine 126, 215–247.[Abstract]

Harley, S. L. (1998a). On the occurrence and characterization of ultrahigh-temperature crustal metamorphism. In: Treloar, P. J. & O'Brien, P. J. (eds) What Drives Metamorphism and Metamorphic Reactions? Geological Society, London, Special Publications 138, 81–107.

Harley, S. L. (1998b). Ultrahigh-temperature granulites metamorphism (1050°C, 12 kbar) and decompression in garnet (Mg70)–orthopyroxene–sillimanite gneisses from the Rauer Group, East Antarctica. Journal of Metamorphic Geology 16, 541–562.[CrossRef][Web of Science]

Harley, S. L. & Green, D. H. (1982). Garnet–orthopyroxene barometry for granulites and peridotites. Nature 300, 697–701.[CrossRef]

Harley, S. L., Hensen, B. J. & Sheraton, J. W. (1990). Two-stage decompression in orthopyroxene–sillimanite granulites from Forefinger Point, Enderby Land, Antarctica: implications for the evolution of the Archean Napier Complex. Journal of Metamorphic Geology 8, 591–613.[Web of Science]

Hensen, B. J. (1987). P–T grids for silica-undersaturated granulites in the system MAS (n + 4) and FMAS (n + 3)—tools for the derivation of P–T paths of metamorphism. Journal of Metamorphic Geology 5, 255–271.[Web of Science]

Hensen, B. J. & Green, D. H. (1973). Experimental study of the stability of cordierite and garnet in pelitic compositions at high pressures and temperatures. III. Synthesis of experimental data and geological applications. Contributions to Mineralogy and Petrology 38, 151–166.[CrossRef][Web of Science]

Hensen, B. J. & Harley, S. L. (1990). Graphical analysis of P–T–X relations in granulite facies metapelites. In: Brown, M. & Ashworth, J. R (eds) High-grade Metamorphism and Crustal Anatexis. London: Allan & Unwin, pp. 19–55.

Holland, T. J. B. & Powell, R. (1998). An internally consistent data set for phases of petrological interest. Journal of Metamorphic Geology 16, 309–343.[CrossRef][Web of Science]

Houseman, G. & England, P. (1986). A dynamic model of lithospheric extension and sedimentary basin formation. Journal of Geophysical Research 91, 719–729.

Indares, A. & Martignole, J. (1985). Biotite–garnet geothermometry in the granulite facies: the influence of Ti and Al in biotite. American Mineralogist 70, 272–278.[Abstract]

Kleemann, U. & Reinhardt, J. (1994). Garnet–biotite thermometry revisited: the effect of Al^VI and Ti in biotite. European Journal of Mineralogy 6, 925–941.[Abstract/Free Full Text]

Köstlin, E. C. (1967). The geology of part of the Kunene Basic Complex, Kaokoveld, South West Africa. M.Sc. thesis, University of Cape Town.

Kretz, R. (1983). Symbols for rock-forming minerals. American Mineralogist 68, 277–279.[Abstract]

Lee, H. Y. & Ganguly, J. (1988). Equilibrium compositions of coexisting garnet and orthopyroxene: experimental determinations in the system FeO–MgO–Al₂O₃–SiO₂, and applications. Journal of Petrology 29, 93–113.[Abstract/Free Full Text]

Littmann, S., Romer, R. L. & Okrusch, M. (2000). Nephelinsyenite der Epembe-Swartbooisdrif-Alkali-Provinz (ESAP)/NW Namibia. Beihefte zum European Journal of Mineralogy 12, 115.

Mayer, A., Sinigoi, S., Miguel, L. G., Morais, E. & Petrini, R. (2000). Kibarian ages in the Kunene Intrusive Complex. GEOLUANDA 2000, Abstract Volume, p. 106.

Menge, G. F. W. (1998). The antiformal structure and general aspects of the Kunene Complex. Zeitschrift der Deutschen Geologischen Gesellschaft 149, 431–448.

Miller, R. Mc. G. (1983). The Pan-African Damara Orogen of South West Africa/Namibia. In: Miller, R. Mc. G. (ed.) Evolution of the Damara Orogen of South West Africa/Namibia. Geological Society of South Africa, Special Publication 11, 431–515.

Moecher, D. P., Essene, E. J. & Anovitz, L. M. (1988). Calculation and application of clinopyroxene–garnet–plagioclase–quartz geobarometers. Contributions to Mineralogy and Petrology 100, 92–106.[CrossRef][Web of Science]

Montel, J. M. & Vielzeuf, D. (1997). Partial melting of metagreywackes, Part II. Compositions of minerals and melts. Contributions to Mineralogy and Petrology 128, 176–196.[CrossRef][Web of Science]

Newton, R. C. & Perkins, D. (1982). Thermodynamic calibration of geobarometers based on the assemblages garnet–plagioclase–orthopyroxene (clinopyroxene)–quartz. American Mineralogist 67, 203–222.[Abstract]

Olson, S. F. (2000). The Proterozoic evolution of Africa. University of the Witwatersrand, Johannesburg, Economic Geology Research Unit, Information Circular 343, 1–61.

Powell, R. & Holland, T. J. B. (1988). An internally consistent dataset with uncertainties and correlations: 3. Applications to geobarometry, worked examples and a computer program. Journal of Metamorphic Geology 6, 173–204.[Web of Science]

Raith, M., Karmakar, S. & Brown, M. (1997). Ultra-high-temperature metamorphism and multistage decompressional evolution of sapphirine granulites from the Palni Hills Range, southern India. Journal of Metamorphic Geology 15, 379–399.[CrossRef][Web of Science]

Sandiford, M. & Powell, R. (1986). Deep crustal metamorphism during continental extension: modern and ancient examples. Earth and Planetary Science Letters 79, 151–158.[CrossRef][Web of Science]

Sengupta, P., Sen, J., Dasgupta, S., Raith, M., Bhui, U. K. & Ehl, J. (1999). Ultra-high-temperature metamorphism of metapelitic granulites from Kondapalle, Eastern Ghats Belt: implications for the Indo-Antarctic Correlation. Journal of Petrology 40, 1065–1087.[CrossRef][Web of Science]

Seth, B., Kröner, A., Mezger, K., Nemchin, A. A., Pidgeon, R. T. & Okrusch, M. (1998). Archean to Neoproterozoic magmatic events in the Kaoko Belt of NW Namibia and their geodynamic significance. Precambrian Research 92, 341–363.[CrossRef][Web of Science]

Seth, B., Brandt, S. & Kramers, J. D. (2001). First isotopic age determinations of mid-Proterozoic granulite facies metamorphism in the Epupa Complex. EUG 2001, Abstract Volume, p. 597.

South African Committee for Stratigraphy (1980). Stratigraphy of South Africa, Part I, Kent, L. E. (compiler). Geological Survey of South Africa, Handbook 8.

Stevens, G., Clemens, J. D. & Droop, G. T. R. (1997). Melt production during granulite-facies anatexis: experimental data from ‘primitive’ metasedimentary protoliths. Contributions to Mineralogy and Petrology 128, 352–370.[CrossRef][Web of Science]

Tegtmeier, A. & Kröner, A. (1985). U–Pb zircon ages for granitoid gneisses in northern Namibia and their significance for Proterozoic crustal evolution of southwestern Africa. Precambrian Research 28, 321–326.

Vielzeuf, D. & Montel, J. M. (1994). Partial melting of metagreywackes. Part I. Fluid-absent experiments and phase relationships. Contributions to Mineralogy and Petrology 117, 375–393.[CrossRef][Web of Science]

Wood, B. J. & Banno, S. (1973). Garnet–orthopyroxene and orthopyroxene–clinopyroxene relationships in simple and complex systems. Contributions to Mineralogy and Petrology 42, 109–124.[CrossRef][Web of Science]

CiteULike    Connotea    Del.icio.us    What's this?

This article has been cited by other articles:

				K. Sato, T. Miyamoto, and T. Kawasaki Fe2+-Mg partitioning experiments between orthopyroxene and spinel using ultrahigh-temperature granulite from the Napier Complex, East Antarctica Geological Society, London, Special Publications, January 1, 2008; 308(1): 431 - 447. [Abstract] [Full Text] [PDF]

				K. Druppel, T. Wagner, and A. J. Boyce EVOLUTION OF SULFIDE MINERALIZATION IN FERROCARBONATITE, SWARTBOOISDRIF, NORTHWESTERN NAMIBIA: CONSTRAINTS FROM MINERAL COMPOSITIONS AND SULFUR ISOTOPES Can Mineral, August 1, 2006; 44(4): 877 - 894. [Abstract] [Full Text] [PDF]

				K. Das, S. Bose, I. Ohnishi, and S. Dasgupta Garnet-spinel intergrowths in ultrahigh-temperature granulite, Eastern Ghats, India: Possible evidence of an early Tschermak-rich orthopyroxene during prograde metamorphism American Mineralogist, February 1, 2006; 91(2-3): 375 - 384. [Abstract] [Full Text] [PDF]

				K. DRUPPEL, J. HOEFS, and M. OKRUSCH Fenitizing Processes Induced by Ferrocarbonatite Magmatism at Swartbooisdrif, NW Namibia J. Petrology, February 1, 2005; 46(2): 377 - 406. [Abstract] [Full Text] [PDF]

				S. Kroner, J. Konopasek, A. Kroner, C.W. Passchier, U. Poller, M.T.D. Wingate, and K.H. Hofmann U-Pb and Pb-Pb zircon ages for metamorphic rocks in the Kaoko Belt of Northwestern Namibia: A Palaeo- to Mesoproterozoic basement reworked during the Pan-African orogeny South African Journal of Geology, September 1, 2004; 107(3): 455 - 476. [Abstract] [Full Text] [PDF]

This Article Abstract FREE Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (17) Request Permissions Google Scholar Articles by BRANDT, S. Articles by OKRUSCH, M. Search for Related Content GeoRef GeoRef Citation Social Bookmarking          What's this?

Online ISSN 1460-2415 - Print ISSN 0022-3530

Copyright © 2009 Oxford University Press

Oxford Journals Oxford University Press

• Site Map
• Privacy Policy
• Frequently Asked Questions

Other Oxford University Press sites: Oxford University Press Oxford Journals China Oxford Journals Japan American National Biography Booksellers' Information Service Children's Fiction and Poetry Children's Reference Corporate & Special Sales Dictionaries Dictionary of National Biography Digital Reference English Language Teaching Higher Education Textbooks Humanities International Education Unit Law Medicine Music Online Products Oxford English Dictionary Reference Rights and Permissions Science School Books Social Sciences Very Short Introductions World's Classics