Journal of Petrology | Pages |
© 1998 Oxford University Press |
Metamorphic belts world-wide often feature thrust-bound windows of older, previously metamorphosed basement. In such slabs, basement rocks are variably overprinted by the deformations accompanying the orogenesis of the enclosing metamorphic belt. In most metamorphic belts reworking of basement is typically pervasive with earlier mineral parageneses rarely preserved; examples include the Hellenides, European Alps, Ubendian Belt (Ring, 1993) and the Kaoko Belt (Durr & Dingeldey, 1996). However, basement slabs that preserve earlier mineral parageneses and have not been pervasively reworked by the enclosing metamorphic belt are recognized, typically as anhydrous thrust-bound slabs such as in the Grenville Province (Rivers et al., 1989), Ungava Orogen (St-Onge & Ijewliw, 1996) and the Chewore Inliers (this study). In such terranes, new mineral parageneses are largely restricted to corona textures (St-Onge & Ijewliw, 1996) or within domains that involved recrystallization, such as later shear zones and foliation seams. St-Onge & Ijewliw, (1996) recognized that only some samples, those closest to the margin of the tectonic window, are chemically re-equilibrated, resetting multi-equilibria P-T calculations. This study of the Chewore Inliers attempts to quantify the degree of chemical re-equilibration of the older mineral parageneses within a basement terrane enclosed within a younger metamorphic belt. Re-equilibration during reworking without pervasive recrystallization is tested by comparing equilibrium thermodynamics P-T calculations from throughout the basement terrane with metamorphic conditions in the enclosing metamorphic belt. The Chewore Inliers of Zimbabwe provide a useful area for testing the effects of differential reworking of previously metamorphosed terranes by younger tectonometamorphic events. The inliers are situated centrally within the Zambezi Belt; this is part of the Pan-African Orogenic System, which cuts across and reworks the older Magondi and Irumide Belts and Archaean gneisses at the northern margin of the Zimbabwe Craton (Fig. 1). Furthermore, recent geochronology recognizes Mesoproterozoic protoliths (Johnson et al., 1996; Goscombe et al., 1998) and tectonometamorphic events at approximately 945 ± 20 Ma, 830 ± 28 Ma and 538 ± 15 Ma throughout most of the Zambezi Belt (Dirks et al., 1997; Vinyu et al., 1997; Goscombe et al., 1998).
The Chewore Inliers were geologically unexplored, except for visits by
Wiles, (1956) and
Munyanyiwa, (1993), until they were mapped by the Zimbabwe Geological Survey during 1990-1994. Structural analysis and geochronology of the inliers have been presented by
Goscombe et al., (1994, , 1998). These studies recognize a high-grade metamorphic cycle at 943 ± 34 Ma (M1) and reworking in an amphibolite facies metamorphic cycle (M2) during the Pan-African Orogeny at 524 ± 16 Ma. A thrust-bound terrane, the Granulite Terrane, was delineated and shown to preserve the M1 parageneses with only minor reworking during M2. Thus the setting of the Granulite Terrane is ideal for testing the effect of a second metamorphic event on older mineral parageneses that were not substantially recrystallized during reworking. The present study interprets the metamorphic evolution of all the terranes recognized in this critical area and attempts to quantify the chemical re-equilibration and resetting of P-T calculations from the Granulite Terrane by the younger Pan-African Orogeny. The regional setting of the Chewore Inliers is presented in Fig. 1, and correlation with the regional geology has been presented by
Goscombe et al., (1998). The Inliers are central within the east-west-trending Zambezi Belt at the northern margin of the Archaean Zimbabwe Craton. The Zambezi Belt constitutes the central portion of the Neoproterozoic-early Palaeozoic Pan-African Orogenic System that stretches across southern Africa, between the Damara Belt-Lufilian Arc and the Mozambique Belt. The Zambezi Belt is largely composed of reworked pre-Pan-African basement gneisses of Archaean to Mesoproterozoic age. The Zambezi Supracrustals, interpreted to be of 820-880 Ma age (Wilson et al., 1993; Hanson et al., 1994), constitute only a minor component of the Zambezi Belt to the east of the Mwembeshi shear zone and appear to be restricted to south Zambia (Fig. 1). The age of the reworked basement gneisses is poorly known throughout most of the Zambezi Belt, but present geochronology suggests largely Mesoproterozoic ages. Concordant orthogneisses within older supracrustals from the Granulite and Zambezi Terranes of the Chewore Inliers have been dated by SHRIMP at 1071 ± 8 and 1083 ± 8 Ma (Goscombe et al., 1998). A plagiogranite dyke within what is interpreted as sheeted meta-dolerite dykes in the Ophiolite Terrane has been SHRIMP dated at 1393 ± 22 Ma (Johnson et al., 1996). A high-grade tectonometamorphic event including granite emplacement, of approximately 945 ± 20 Ma age (age determinations spanning 920-960 Ma), is recognized in the Chewore Inliers and also at widely scattered localities in south Malawi (Andreoli et al., 1981; Haslam et al., 1983; Cahen et al., 1984), in the Mozambique Belt (Jourde & Wolff, 1974; Costa et al., 1983; Cahen et al., 1984) and in the Zambezi Belt in east Zambia (Barr et al., 1978). These ages are similar to the 960-1100 Ma age of accretion of voluminous calc-alkaline magmas in the Mozambique province of the Mozambique Belt (Haslam et al., 1983; Pinna et al., 1993). The southernmost margin of the Zambezi Belt, at the northern margin of the Zimbabwe Craton, is a collage of metamorphic gneisses of Archaean age (Loney, 1969; Barton et al., 1991; Hahn et al., 1992; Dirks et al., 1997; Vinyu et al., 1997) that have been multiply reworked within the Zambezi Belt. A high-grade tectonometamorphic event of ~945 ± 20 Ma age (age determinations spanning 920-960 Ma), is recognized in the Chewore Inliers and also at widely scattered localities elsewhere in the Zambezi Belt in east Zambia (Barr et al., 1978), south Malawi (Andreoli, 1981; Haslam et al., 1983; Cahen et al., 1984) and in the Mozambique Belt (Jourde & Wolff, 1974; Costa et al., 1983; Cahen et al., 1984). The first metamorphic event recognized in the Chewore Inliers is high-T-low-P metamorphism of the Granulite Terrane (M1). Metamorphic zircon overgrowths in the Granulite Terrane are high in U and have low Th/U ratios, and have been dated by SHRIMP at 943 ± 34 Ma and interpreted as the minimum age of M1 metamorphism (Goscombe et al., 1998). Syn-tectonic granite emplacement and associated metamorphism at ~830 ± 28 Ma (age determinations spanning 780-880 Ma) occurred throughout most of the Zambezi Belt and has been interpreted as a major orogenic period called the Zambezi Orogeny (Barton et al., 1991), and interpreted to be extensional (Dirks et al., 1997). Pervasive granitoid emplacement has not been recognized in the Chewore Inliers, nor is there geochronological evidence of an 830 Ma event in this region. Syntectonic granite sheets are most pervasive at the southern margin of the Zambezi Belt (Loney, 1969; Barton et al., 1991; Dirks et al., 1997). Syn-tectonic granites of this age are also recognized in south Zambia, as the Lusaka Granite and Ngoma orthogneiss (Barr et al., 1978; Hanson et al., 1988), in south Malawi (Andreoli, 1981; Haslam et al., 1983) and in the Mozambique province of the Mozambique Belt (Pinna et al., 1993). The Zambezi Belt experienced SW-directed tectonic transport and crustal shortening (Barton et al., 1991; Goscombe et al., 1994) concomitant with east-west shortening in the Mozambique Belt (Pinna et al., 1993). Collisional orogenesis within the Zambezi Belt has previously been interpreted to have occurred at 830 Ma (Hanson et al., 1988; Barton et al., 1991; Goscombe et al., 1994). However, recent zircon geochronology (Vinyu et al., 1997; Goscombe et al., 1998) and earlier geochronological studies (Barr et al., 1978; Hahn et al., 1992) constrain the peak of metamorphism to be ~538 ± 15 Ma (age determinations ranging from 510 to 560 Ma age) (Goscombe et al., 1998). Thin metamorphic zircon overgrowths in the Granulite Terrane have low U and extremely low Th/U, and have been SHRIMP dated at 524 ± 16 Ma (Goscombe et al., 1998). These overgrowths are interpreted to have grown at the peak of metamorphism during reworking of the Granulite Terrane in the M2 metamorphic cycle. Concordant single zircon ages and lower intercept zircon ages of 518-543 Ma from NE Zimbabwe are also correlated with the peak of metamorphism during collisional orogenesis (Vinyu et al., 1997). Consequently, collisional orogenesis and amphibolite facies metamorphism of the Zambezi Belt occurred during the early Palaeozoic Pan-African Orogeny (Dirks et al., 1997; Vinyu et al., 1997; Goscombe et al., 1998). Collisional orogenesis at this time is recorded from throughout the Pan-African Orogenic System, from the Mozambique Belt (U. Ring, personal communication, 1997) to the Damara Belt (Miller, 1983). Pan-African orogenic events were initiated by tectonometamorphic events during the period 600-700 Ma; this is particularly evident in the Lufilian Arc, Kaoko Belt, Central Damara Orogen and the Mozambique Belt in Tanzania, but not in the Zambezi Belt. Pan-African orogenesis culminated in a distinct phase at ~510-560 Ma as discussed, with widespread reworking by overthrusting, crustal over-thickening, associated amphibolite facies metamorphism and consequent post-tectonic pegmatites and granites at 440-490 Ma (Kroner, 1980; Barton et al., 1991, , 1993; Munyanyiwa & Blenkinsop, 1993; Vinyu et al., 1997; Goscombe et al., 1998). Uplift subsequent to the Pan-African events gave rise to denudation to deep crustal levels. Mesozoic crustal extension (150-285 Ma) gave rise to grabens and half-grabens such as the Luangwa Rift and Zambezi Rift, within which the Chewore Inliers are situated (Fig. 1). The Chewore Inliers are composed of four lithologically distinct terranes: the Granulite (GT), Quartzite (QT) and Zambezi Terranes (ZT) (Goscombe et al., 1994, , 1998) and the Ophiolite Terrane (OT) (Johnson & Oliver, 1997) (Fig. 2). Each terrane is of uniform metamorphic grade and distinct from the other terranes, and the ZT is further subdivided into the north ZT of upper-amphibolite facies grade and south ZT of middle-amphibolite facies grade (Figs 2 and 3). The QT is dominated almost entirely by quartzite and psammo-pelites, and the ZT by a wide variety of layered and migmatized quartzo-feldspathic gneiss (QFG) and concordant granitic orthogneiss with units of metapelites, quartzites, mafics and calc-silicate gneisses. The high-grade GT is composed of anhydrous sillimanite- or orthopyroxene-bearing garnet paragneiss and concordant orthogneisses with minor metapelites and mafic granulite, and is devoid of calc-silicates and quartzite. Concordant granitic orthogneisses in both the GT (1083 ± 8 Ma) and ZT (1071 ± 8 Ma) were emplaced before all recognized tectonic events, and are thought to have been emplaced soon after formation of the supracrustal sequence (Goscombe et al., 1998). The OT is composed mostly of mafic and ultramafic gneisses that have recently been interpreted as an ophiolite sequence and may represent a crustal suture (Johnson et al., 1996; Johnson & Oliver, 1997). The southeasternmost portion of this terrane is high-P white schists of talc-kyanite ± staurolite, rarely also with yoderite (Johnson & Oliver, 1997).
Two clearly distinct tectonothermal cycles have been recognized in the Chewore Inliers, M1 and M2 (Table 1). The first tectonothermal cycle (M1) is preserved in the GT, and is responsible for the character of this terrane. The M1 metamorphic cycle is dated at 943 ± 34 Ma and involved isoclinal folding during south to north transport, coeval with high-T-low-P metamorphism, possibly in an extensional regime. In contrast, the other terranes were almost totally reworked during the Pan-African metamorphic cycle (M2) and expressions of the M1 metamorphic cycle are only preserved as inclusion parageneses and overgrown foliations in the ZT and QT. The M2 metamorphic cycle was collisional, with early intense fabric development and subsequent recrystallization accompanying the peak of amphibolite facies metamorphism (524 ± 16 Ma) and a clockwise P-T path. Table 1. Summary of tectonothermal events experienced during evolution of the Chewore Inliers.
Structures associated with the M1 metamorphic cycle are only preserved within the GT; this is an artefact of all other terranes being pervasively reworked during M2. The M1 metamorphic cycle is entirely responsible for the character of the GT because expressions of reworking by M2 are almost non-existent (Goscombe et al., 1994). Compositional and gneissic layering (SG1) and the sub-parallel tectonic fabric (SG2) trend east-west and dip steeply north, characterizing this high-grade terrane that is continuous for 30 km to the northwest into Zambia (Barr, 1970; Ramsay & Ridgway, 1977). The M1 metamorphic cycle involved high-T-low-P metamorphism accompanying the first two isoclinal fold events, DG2 and DG3, which plunge shallowly west (Fig. 4a). A steep, NNE-plunging stretching lineation (LG2) and tectonic foliation formed during DG2 and were later statically annealed to coarse granoblastic textures at the peak of M1 metamorphism. Sigma-type asymmetrical augen, associated with LG2, and also asymmetrical DG3 folds, both indicate northward-directed transport during DG2-DG3 (Goscombe et al., 1994).
No intense penetrative fabric was developed in the GT during the M2 metamorphic cycle, reworking being restricted almost entirely to spaced, fine biotite-sillimanite seams that overprint the SG2 fabric. Other late-stage episodes in the GT include open, north-south-trending warps with a axial planar fracture-cleavage, coronal reaction textures, mantle sub-grains and strained M1 minerals. The timing of open warps is not known but is tentatively correlated with the north-south warps in the QT and ZT that formed late in the M2 metamorphic cycle (Table 1). A garnet-chloritoid-biotite-hornblende-bearing granitic orthogneiss body intruded the GT subsequent to DG2-DG3 deformation. This granite exhibits a weak foliation parallel to SZ4 in the ZT and is interpreted to have been emplaced and deformed during the M2 metamorphic cycle (Goscombe et al., 1994). Included phases, such as chloritoid, spinel, kyanite and sillimanite, which are absent from matrix mineral parageneses of the sample, are considered relics of earlier mineral parageneses (Vernon, 1978). These early parageneses either formed during the prograde metamorphism associated with the matrix assemblage or in an earlier unrelated tectonothermal event. In the ZT early foliations of aligned sillimanite ± biotite ± ilmenite inclusions rarely occur in the cores of porphyroblastic garnet and define an overgrown pre-SZ2 foliation (see Fig. 6e, below). This early sillimanite, and rarely also spinel, is inconsistent with the prograde portion of the M2 clockwise P-T path which culminated in decompression through the peak of metamorphism from the kyanite-staurolite stability field into the sillimanite stability field (see below). Consequently, these early sillimanite ± spinel parageneses in the ZT are thought to be relic phases formed during the earlier high-T-low-P, M1 metamorphic cycle and are unrelated to M2. Pre-SZ2 foliations are also preserved in lenticular low-strain domains enveloped by the SZ2 foliation in ZT quartzo-feldspathic gneisses (Fig. 4b). The Quartzite, Zambezi and Ophiolite Terranes all experienced the same deformational history and are pervasively reworked during the M2 metamorphic cycle (Table 1; Goscombe et al., 1994). An intense SZ2-LZ2 fabric (Fig. 4a) is everywhere developed in these terranes and though parallel to, overprints an earlier gneissic and migmatitic layering (SZ1) (Table 1, Fig. 5a). The earlier SZ1 migmatitic and gneissic layering formed either during prograde metamorphism in the M2 metamorphic cycle or in a pre-M2 metamorphic cycle, during either M2 or an unrecognized tectonic event (Table 3, below) (Goscombe et al., 1998). SZ2 is defined by a sub-mylonitic to mylonitic foliation of quartz-feldspar aggregate ribbons (Fig. 4d) and aligned micas that has been annealed to medium- to coarse-grained granoblastic textures (see Fig. 7a and b, below). Elongate minerals such as hornblende and primary sillimanite and kyanite laths are contained within SZ2 and in textural equilibrium with the granoblastic matrix, but typically are only vaguely aligned with the LZ1 mineral-aggregate lineation (Fig. 4d). Consequently, the SZ2 fabric is interpreted to have been statically annealed at the peak of metamorphism, giving rise to the medium- to coarse-grained granoblastic and porphyroblastic matrix (Goscombe et al., 1994). Sigma-type augen and shearbands (Fig. 4c) indicate both SW- and NE-directed transport along the SW-plunging LZ2 stretching lineation. Unfolding of DZ3 effects indicates transport during DZ2 was NE over SW (Goscombe et al., 1994). DZ3 deformation gave rise to widespread ductile fold repetition on all scales, by tight to isoclinal SW-plunging folds, with a weak to moderate axial planar foliation (SZ3) (Fig. 4e) that is typically fine grained and parallel to SZ2 (Figs 6d and 7b). DZ3 fold axes are sub-parallel to LZ2 (Fig. 4f) and these folds are interpreted to have formed in a non-coaxial shear environment by transport along the same NE-SW trending transport vector as LZ2 (Goscombe et al., 1998). Thus DZ2 and DZ3 constitute a period of intense progressive deformation resulting in crustal shortening and over-thickening by fold repetition, immediately before the peak of M2 metamorphism. The last pervasive folding event (DZ4) gave rise to tight to open, steeply inclined folds with weak axial planar foliations (Fig. 5a and c). DZ4 and DZ5 crenulation cleavages both involved NE over SW directed tectonic transport. Thus all ductile deformations from DZ2 to DZ5 involved SW-directed transport and constitute a single orogenic cycle (Table 1) which is correlated with the Pan-African Orogeny of identical tectonic transport from throughout the Zambezi Belt (Fig. 1) (Goscombe et al., 1998).
Deformation-induced recrystallization and metamorphic reaction textures that overprint the matrix mineral parageneses are minor but widely developed in the QT and ZT. In metapelites, these include thin SZ3 seams of fine-grained biotite-sillimanite ± garnet developed sub-parallel to SZ2 (Figs 6d and 7b). At a high angle to SZ2 are SZ4 biotite-sillimanite seams and SZ5 crenulation cleavages with aligned micas, fine-grained sillimanite or hornblende. DZ3-DZ5 foliations formed during and immediately after the peak of M2 metamorphism at similar temperatures but lower pressures than peak (matrix) conditions. Similar late-stage sillimanite-biotite foliations in the GT (SG4) are interpreted to have formed during reworking of the GT in the M2 metamorphic cycle. The southern margin of the GT is a 100 m wide zone with numerous steep, north-dipping mylonites that formed during oblique transport of the GT to the SW and onto the QT (DG5) (Goscombe et al., 1994). The contact between the QT and ZT is a sharp discontinuity that, by comparison with the south margin of the GT, is also interpreted as a steep thrust contact (DZ6). The GT, QT and ZT were brought into their present juxtaposition subsequent to all ductile fold events, and SW-directed transport of the GT suggests that juxtaposition occurred in the latter stages of the M2 metamorphic cycle (Table 1) (Goscombe et al., 1994). Overthrusting is in part responsible for reorientation of the GT and QT into sub-vertical orientations. The OT is fault bound (Johnson et al., 1996), but the contact appears to be folded by DZ3 folds (Fig. 1) (Goscombe et al., 1994). Thus the OT may have been brought into juxtaposition with the ZT before, or early in the M2 metamorphic cycle. NW-SE-trending muscovite ± garnet pegmatites in the ZT post-date all ductile deformations and are unstrained (Table 1). The Chewore Ultramafic Complex and undeformed vertical dolerite dykes were emplaced subsequent to DG5-DZ6 juxtaposition of terranes (Goscombe et al., 1994) and may be related to Mesozoic rifting. Metapelite and psammo-pelite rocks in the GT contain stromatic partial melt segregations and have a well-annealed granoblastic matrix of largely unaligned phases. The granoblastic matrix comprises the peak metamorphic assemblage of garnet-sillimanite-biotite-K-feldspar-quartz-ilmenite ± rutile ± plagioclase ± hercynitic spinel (Table 2). A few samples contain kyanite grains close to peak metamorphic sillimanite; the timing of the kyanite is not confidently known (see later discussion). Hercynitic spinel occurs rarely as grains within the peak metamorphic matrix and most commonly as inclusions along with sillimanite, ilmenite and biotite within garnet and ilmenite (Fig. 6a). Early prograde foliations of well-aligned fine sillimanite ± biotite inclusions in garnet are common (Table 3). Table 2. Petrology of representative aluminous rocks from the Chewore Inliers.
Table 3. Summary of mineral parageneses and metamorphic reaction textures in the Chewore Inliers.
Garnet porphyroblasts are syntectonic with the SG2 fabric and in textural equilibrium with the matrix assemblage. Garnet also occurs as idioblastic coronas enclosing hercynite-ilmenite grains and both primary sillimanite and fine late-stage sillimanite-biotite seams, as well as constituting garnet-ilmenite ± sillimanite overgrowths on garnet porphyroblasts. Garnet porphyroblasts are corroded by aggregates of sillimanite-biotite ± garnet ± ilmenite and by partial coronas of sillimanite or biotite. Fine-grained secondary sillimanite ± biotite grows on the margins of primary sillimanite laths, K-feldspar, ilmenite and biotite (Table 3). The matrix assemblage is overprinted by fine-grained foliated seams of sillimanite and biotite, which are generally sub-parallel to the gneissic layering and the annealed coarse-grained sillimanite-biotite fabric (SG2). Other late-stage fabrics recognized are fine isolated sillimanite needles or muscovite laths growing across the earlier SG2 fabric. Late-stage deformation is further evidenced by undulose extinction in some quartz and feldspar grains. QFG and pelitic QFG in the GT are near-anhydrous gneisses with well-annealed granoblastic textures. These gneisses have similar assemblages, reaction textures and late-stage foliations to the metapelites except that all samples are dominated by plagioclase, K-feldspar and less commonly perthite, and a few samples are devoid of sillimanite but contain peak metamorphic orthopyroxene or hornblende. Orthopyroxene and hornblende margins are partially replaced by blue-green hornblende and garnet coronas. Hornblende coronas also enclose ilmenite and biotite grains and rarely there is also an outer concentric corona of garnet. Ilmenite is enclosed by coarse sillimanite coronas. Nearly all mafic gneisses in the GT are concordant coarse-grained mafic granulites with unaligned granoblastic assemblages of orthopyroxene-clinopyroxene-plagioclase-quartz-ilmenite ± hornblende ± biotite (Table 4, Figs 3 and 6f). Concordant mafic amphibolites with hornblende-quartz-plagioclase ± biotite ± garnet ± clinopyroxene ± sphene ± ilmenite assemblages also occur but are less common. Rutile and ilmenite appear to be in textural equilibrium. Garnet is rarely a peak metamorphic phase but is a very common corona phase, often enclosing thin inner plagioclase or actinolite coronas on matrix clinopyroxene and orthopyroxene (Fig. 7f). These textures indicate post-M1 retrogressive actinolite growth followed by a later heating event (M2) producing the garnet coronas. Blue-green hornblende ± ilmenite form partial coronas on pyroxenes, biotite and ilmenite. Primary hornblende is typically brown-green and uncommonly enclosed by aggregates of ilmenite or clinopyroxene ± ilmenite ± hornblende. Table 4. Petrology of representative mafic and calc-silicate rocks from the Chewore Inliers.
GT gneisses are recrystallized to new fine-grained, aligned granoblastic assemblages within DG5 mylonite zones. Mylonitized mafics formed clinopyroxene-hornblende-garnet-biotite-plagioclase-quartz-ilmenite assemblages. Mafic granulites adjacent to these mylonites have numerous garnet coronas on primary hornblende and ilmenite, and coronas of blue-green hornblende on primary pyroxenes. Mylonitized QFGs have quartz-albite ± K-feldspar-garnet-biotite-ilmenite ± sillimanite assemblages. Metapelite rocks in the QT do not differ markedly from those in the north ZT, described in detail below. In the QT, quartz-rich metapelites that are largely devoid of feldspars and partial melt segregations are more common than typical metapelites. Peak metamorphic assemblages comprise a well-annealed granoblastic framework of porphyroblastic garnet, biotite, kyanite, sillimanite, quartz, K-feldspar, ilmenite and less commonly rutile, albite and muscovite (Fig. 7c and d). The kyanite, sillimanite and micas are coarse grained and vaguely aligned and in coarse seams (SZ2). Typically, kyanite and sillimanite coexist and are in textural equilibrium, with few samples containing only one Al-silicate phase (Fig. 3). Garnet is typically syntectonic with SZ2, rich in inclusions and in equilibrium with the granoblastic matrix. Early garnet porphyroblasts were also pulled apart during DZ2-DZ3 and the resulting garnet sub-grains are in textural equilibrium with the matrix assemblage. Ilmenite and rutile often occur in textural equilibrium though rutile coronas on ilmenite do occur. Relic coarse-grained prograde muscovite is overgrown by peak metamorphic kyanite laths. In both the QT and north ZT, garnet often contains coarse-grained inclusions of prograde kyanite (Fig. 6d), biotite, ilmenite and feldspars. Less commonly, aligned fine sillimanite needles and biotite inclusions preserve a pre-SZ2 foliation (Fig. 6e). The margins of primary kyanite and sillimanite are overgrown by fine sillimanite (Fig. 7c), and secondary kyanite has not been recorded. Secondary sillimanite also occurs in a multitude of coronal reaction textures, most commonly within a corona aggregate of secondary garnet, ilmenite and biotite overgrowing primary garnet margins. Secondary fine sillimanite also grows on the margins of primary K-feldspar, plagioclase, biotite, garnet and muscovite, and is scattered throughout the rock. Primary garnet is rarely enclosed by coronas of plagioclase ± ilmenite ± biotite (Fig. 7e), and ilmenite is commonly enclosed by garnet coronas. Late-stage retrogressive reactions include the partial replacement of kyanite and biotite by chlorite, and replacement of garnet by muscovite or biotite (Table 3). The coarse-grained matrix assemblage is overprinted by aligned fine sillimanite needles and less commonly muscovite and thin cross-cutting seams of fine-grained sillimanite and biotite (SZ4 and SZ5). These post-matrix assemblage foliations do not contain kyanite and equilibrated in the sillimanite stability field. Conformable mafic gneisses are rare and have an aligned granoblastic matrix of green hornblende with plagioclase, quartz, ilmenite and sphene, and less commonly minor garnet, biotite or clinopyroxene (Table 4, Fig. 3). Garnet is enclosed by plagioclase coronas; garnet-ilmenite coronas enclose both primary hornblende and secondary blue-green hornblende. Metapelites are composed of a granoblastic matrix of biotite, quartz, sillimanite or kyanite, porphyroblastic garnet, minor amounts of ilmenite, rutile, K-feldspar and less commonly plagioclase and muscovite (Table 2, Fig. 7c). In contrast to the QT, only a few samples contain coexisting kyanite and sillimanite in textural equilibrium. Biotite and Al-silicates define coarse-grained, annealed SZ2 seams in which these phases define a vague foliation but are not strongly lineated (Fig. 7a-c). These seams formed in response to ductile shear during prograde metamorphism (DZ2) but have been annealed at peak metamorphic conditions subsequent to DZ2. Most metapelites in the north ZT contain leucosomes (Fig. 5a) in equilibrium with the peak metamorphic assemblage. Garnet in metapelites rarely contains inclusions of aligned sillimanite, ilmenite and biotite defining a pre-SZ2 foliation (Fig. 6e), and in one sample also hercynite (Table 2). Secondary sillimanite occurs in all metapelite samples as part of corona reaction textures, and also within SZ3-SZ5 foliations that overprint the annealed SZ2 fabric (Figs 6d and 7b). Fine-grained sillimanite overgrowths form on the margins of primary sillimanite and kyanite. Sillimanite also forms monomineralic coronas on garnet, ilmenite and biotite, and fine sillimanite ± biotite forms at plagioclase and K-feldspar margins (Table 3, Fig. 7d). Garnet grew throughout prograde metamorphism giving syntectonic garnet porphyroblasts. Garnet continued to grow after the formation of the matrix assemblage, giving rise to garnet-sillimanite-biotite ± ilmenite overgrowths on garnet porphyroblasts. Garnet ± ilmenite coronas also form on primary sillimanite and as overgrowths enclosing fine-grained SZ3 and SZ4 sillimanite ± biotite seams. These sillimanite-biotite seams and enclosing garnet together constitute mineral parageneses very similar to the matrix assemblage but without kyanite. Consequently, DZ3-DZ4 deformations occurred during continued heating accompanying decompression from the kyanite to the sillimanite stability field. Mafics are typically massive, have a coarse-grained aligned granoblastic matrix of hornblende-plagioclase-quartz, often with large garnet porphyroblasts with plagioclase-rich leucosomes as moats that are often aligned with SZ4, indicating decompression (Green & Ringwood, 1967) during DZ4 (Fig. 5b). Minor phases include K-feldspar, biotite, clinopyroxene, scapolite, ilmenite, rutile and sphene. Inclusions of most phases represented in the matrix assemblage also occur in garnet and hornblende porphyroblasts (Fig. 6b). Mafic assemblages in the north ZT differ from those in the south ZT by the near-absence of epidote or clinozoisite and common presence of garnet and clinopyroxene (Table 4, Fig. 3). Peak metamorphic garnet in the mafics is pre- to syn-tectonic and rarely enclosed by coronas of plagioclase and symplectites of clinopyroxene or hornblende with plagioclase and ilmenite. Plagioclase rarely has scapolite coronas or muscovite growth at grain margins. Some hornblende has partial garnet coronas and biotite or chlorite growth on grain margins. Ilmenite and rutile are commonly enclosed by sphene coronas. Late-stage foliations are evident both as isolated but aligned biotite platelets that cut across matrix hornblende and as SZ5 crenulation cleavages with axial planar blue-green hornblende. Epidote occurs as a retrogressive phase partially replacing garnet, hornblende and clinopyroxene. There is a wide variety of calc-silicate gneisses in the north and south ZT, but these are rare in the QT and absent from the GT. Most are layered leucocratic gneisses with a granoblastic matrix of quartz-plagioclase-epidote ± clinozoisite ± poikiloblastic hornblende, commonly with minor sphene and ilmenite. The north ZT and QT also contain clinopyroxene ± poikiloblastic garnet within the matrix assemblage, whereas the south ZT calc-silicates may also contain biotite and scapolite. Clinopyroxene is often partially replaced by blue-green hornblende or enclosed by garnet ± ilmenite coronas. Aluminous schists are restricted to the south ZT and comprise 2-20 m thick laterally continuous concordant units. Aluminous schists have a coarse-grained matrix of aligned phlogopite and quartz with porphyroblasts of kyanite, garnet and staurolite all in textural equilibrium. Typical peak metamorphic assemblages include quartz-phlogopite-staurolite ± garnet-ilmenite ± plagioclase ± muscovite and quartz-phlogopite-muscovite-ilmenite-kyanite ± garnet ± staurolite (Table 2). Feldspars are rare, being mostly plagioclase inclusions. Fe-Ti oxides are exclusively ilmenite and rutile, often in textural equilibrium and as inclusions in preference to being matrix phases. Inclusions within garnet include chloritoid, plagioclase, quartz, biotite, ilmenite and tourmaline (Fig. 6c). Chloritoid is absent from the peak metamorphic assemblages and is considered representative of the prograde mineral parageneses. Sillimanite is absent from the south ZT (Fig. 3). Garnet in aluminous schists is mostly syn-tectonic to SZ2 with numerous inclusions; garnet is also pulled apart in deformation events that are accompanied by prograde mineral growth. Staurolite is mostly elongate, poikiloblastic and syntectonic (SZ2), and folded into sigmoidal shapes during DZ3. Kyanite is similarly syntectonic and often poikiloblastic and overgrown by either fine sillimanite or muscovite and chlorite. Garnet and staurolite porphyroblast margins are in part idioblastic and often inclusion free, indicating continued growth subsequent to deformation. Muscovite is aligned parallel to and mostly in equilibrium with phlogopite, but fine muscovite occurs as retrogressive coronas on kyanite and staurolite, and rarely overgrows phlogopite. Prograde muscovite is also evident as coarse muscovite laths overgrown by kyanite porphyroblasts. Chlorite is a late-stage phase on the margins of all other phases and also as foliations that cut across matrix phlogopite. There is a wide variety of fine- to medium-grained mafic gneisses throughout both the south Zambezi Terrane and dominating the Ophiolite Terrane (Table 4). Diagnostic matrix assemblages are hornblende-quartz-plagioclase-epidote/clinozoisite ± biotite ± sphene and, in contrast to the north ZT, only very rarely with garnet or clinopyroxene (Fig. 3). Garnet has idioblastic rims and poikiloblastic cores with quartz-biotite-epidote inclusions and ilmenite-rutile inclusions in the rims. Core inclusion assemblages are aligned and equivalent to matrix mineral parageneses. Garnet more commonly occurs as a corona phase enclosing ilmenite, rutile and plagioclase. Hornblende and plagioclase typically constitute the medium-grained polygonal granoblastic matrix, but also occur as poikiloblasts. Conformable, laterally continuous meta-ultramafic units and lenses of up to 40 m width are common in the OT and rare in the north ZT. Most ultramafic samples are totally retrogressed to serpentinite or tremolite-chlorite-ilmenite, talc-tremolite-ilmenite or chlorite-ilmenite-rutile-quartz schists. White schists from the extreme southeast exposures of the OT (Fig. 3) have quartz-talc-muscovite-kyanite-staurolite ± tourmaline and garnet-talc assemblages (Table 2). Johnson & Oliver, (1997) have also described quartz-free talc-kyanite-yoderite assemblages and inferred conditions of formation of 590-795°C and 13-25 kbar by comparison with experimental work. The white schists preserve evidence of much higher pressures than are preserved by the majority of rocks in the OT. These high-P schists form a unit only at the southeast margin of the OT and therefore may be a discrete fault-bound block of high-P metamorphic rocks (Fig. 3). Analyses were performed on the Cameca SX50 electron microprobe at the University of Tasmania, with an operating voltage of 15 kV and 20 nA for all phases except micas (10 nA) and feldspar (15 nA), and a beam area of 15 µm2 for most phases and 60 µm2 for micas and feldspars. The range in mineral chemistry by rock type from all terranes is summarized in Table 5 and representative mineral analyses are presented in Table 6. Mineral end-member activities were calculated largely after
Powell & Holland, (1985) as summarized in Appendix C. Table 5. Summary of mineral compositional ranges in all rock types and terranes.
Table 6. Representative core mineral analyses.
Average P-T conditions of equilibration of core and rim assemblages were determined using equilibrium thermodynamics by the method of
Powell & Holland, (1985, , 1988). All calculations were performed using the computer program THERMOCALC v2.0b (Powell & Holland, 1988), and results are presented in Appendix A and Figs 8 and 9. All results satisfy the [chi]2 test with f values <1·45 in all but one calculation (sample C323, average T). Errors average ±110°C and ±1·63 kbar in the GT, and ±96°C and ±1·44 kbar in the other terranes (Appendix A). Where there is no solution for average P-T loci the intersection of solutions for average P and average T calculations has been used to define the P-T loci of equilibration conditions (Figs 8 and 9). In a few samples, P-T loci were defined by the intersection of equilibrium thermodynamics results with geothermometry and geobarometry results (see below). The best constrained P-T locus of the maximum temperature conditions, of each sample analysed, is summarized in Table 7 and Figs 8 and 9. Table 7. Best estimates of peak metamorphic conditions preserved in Chewore Inlier samples.
Average P-T values from cores cluster as tight groups that define the peak metamorphic conditions in each terrane (Fig. 10, Table 7). Peak GT samples cluster around a mean of 738 ± 45°C and 4·4 ± 0·9 kbar (Table 7, Fig. 8). These peak temperatures are 60°C lower than the two-pyroxene mafic granulite experimental stability field based on the incoming of orthopyroxene at temperatures >800°C (Spear, 1981) (Fig. 8). Thus the temperatures derived by mineral calculations are closure temperatures representing cessation of cation exchange during cooling (Harley, 1992). Calculated pressures are only slightly higher than the stability field of hercynitic spinel in aluminous rocks (Fig. 8). Spinel in the GT was stabilized at these slightly higher pressures by 1·36-5·21 wt % ZnO (Bohlen & Dollase, 1983; Vielzeuf, 1983). Rims and corona assemblages equilibrated at lower temperatures (20-60°C lower) than cores, and at similar to slightly lower pressures (Fig. 8).
The remainder of the spread of average P-T results from the GT constitutes two groups, both of lower T and higher P. These sample groups are interpreted to have been either totally recrystallized and/or re-equilibrated during the M2 metamorphic cycle (Fig. 8). The two mylonite samples from the southern margin of the GT cluster around mean conditions of 590 ± 60°C and 7·7 ± 0·2 kbar (Table 7). These samples were recrystallized during DG5 in the latter stages of the M2 metamorphic cycle, at conditions similar to those experienced in the ZT (Fig. 11).
Six samples from the GT, representing all major rock-types, have been re-equilibrated at mean conditions of 631 ± 41°C and 5·6 ± 1·0 kbar (Table 7). These samples still preserve M1 mineral parageneses without pervasive later recrystallization and fabric development. The calculated average P-T results for these samples are of identical temperature and 2·3 kbar lower pressure than that experienced at the peak of M2 in the ZT and QT (Fig. 11). These samples equilibrated at significantly lower temperatures than the stability field of the matrix mineral assemblage (M1) of the sample. Consequently, the mineral chemistry of the primary phases constituting the M1 granulite facies assemblages is interpreted to have been re-equilibrated at P-T conditions consistent with M2 metamorphism. The P-T loci calculated from mineral cores form a tight distribution, whereas rim results give a scatter to these data, with rims equilibrating at both higher and lower T and P than the cores (Fig. 8). The average P-T results from M2 mineral parageneses from all the other terranes fall into three distinct groups corresponding to the south ZT and OT together, north ZT and QT sample groups (Fig. 9). Peak metamorphic conditions for each of these groups form tight clusters with mean temperatures of 591 ± 14°C in the OT and south ZT, 630 ± 27°C in the north ZT and 717 ± 19°C in the QT, and 7·9-8·6 kbar in all terranes (Table 7). These estimates of peak metamorphic conditions are entirely compatible with phase stability constraints for the matrix assemblages (Tables 2 and 4, Fig. 9). Kyanite-staurolite assemblages in aluminous schists and epidote-bearing, garnet-free amphibolites in the south ZT occur at conditions identical to those calculated by equilibrium thermodynamics (Fig. 9). Similarly, garnet amphibolites and coexisting sillimanite and kyanite (Fig. 9) are diagnostic of the north ZT, and kyanite-sillimanite-bearing migmatitic metapelites of the QT (Harte & Hudson, 1980) are compatible with the average P-T results derived by equilibrium thermodynamics from these terranes (Fig. 9). Staurolite is absent from the QT and the results of all P-T calculations from this terrane fall outside the staurolite stability field. Average P-T results from rims are of significantly lower pressures (1-4 kbar lower) and slightly lower temperatures (30-65°C lower) than those from cores (Fig. 9). Published geothermometers and geobarometers have been used to constrain, in part, some low-variance samples with insufficient mineral end-members to calculate average P-T loci by equilibrium thermodynamics (Appendix B). These include, in particular, the late-stage igneous bodies; for example, the syn-M2 granitic orthogneiss in the GT and post-tectonic pegmatite are constrained in this way. The syn-M2 granitic orthogneiss was annealed after emplacement, giving rise to garnet-hornblende-feldspar-chloritoid assemblages. Plagioclase-hornblende and two-feldspar geothermometry (Powell & Powell, 1977; Spear, 1981) give tightly constrained results from rim pairs, averaging 580°C (Table 7, Fig. 8). Garnet-hornblende-plagioclase (Kohn & Spear, 1990) and Al in hornblende geobarometry (Hammarstrom & Zen, 1986; Hollister et al., 1987; Schmidt, 1992) also give a tight cluster of results that are mutually consistent (Appendix B) and average 8·5 kbar (Table 7). These conditions are identical to those experienced in the south ZT at the peak of M2 metamorphism. Consequently, this granitic orthogneiss body equilibrated at P-T conditions consistent with emplacement during M2. The Si in phengite geobarometer of
Massonne & Schreyer, (1987) gives a very rough guide to the pressures of equilibration of the post-tectonic garnet-muscovite pegmatite to be 2·7 kbar. Temperature estimates of 440°C from this pegmatite by the garnet-muscovite geothermometer of
Green & Hellman, (1982) are considered reasonable results, as this geothermometer also gives results for a south ZT aluminous schist that is consistent with the phase stability field of the sample (Appendix B, Fig. 9). The garnet-spinel geothermobarometer in the MAS-Zn system (Nichols et al., 1992) and the garnet-plagioclase-sillimanite geobarometer of
Perchuk et al., (1985) give results, for GT samples, that are entirely consistent with average P-T results by equilibrium thermodynamics (Fig. 8). The garnet-biotite geothermometer of
Dachs, (1990), using the non-ideal activity model for garnet after
Hoinkes, (1986), was used to fix P-T loci by intersection with equilibrium thermodynamics, average T and average P results that have low [Delta]P/[Delta]T slopes. This garnet-biotite geothermometer was used for some low-variance QT samples, and gives results that are identical to the temperature results from other QT samples by equilibrium thermodynamics (Fig. 9). The P-T loci of two ZT samples (B54 and C72) are, in part, constrained by the garnet-hornblende (Perchuk et al., 1985) and garnet-muscovite (Green & Hellman, 1982) geothermometers and garnet-ilmenite-rutile-sillimanite geobarometer (Bohlen et al., 1983) (Appendix B). The P-T loci of these samples are identical to equilibrium thermodynamics P-T loci from other ZT samples and are consistent with the phase stability fields of their matrix assemblage (Fig. 9). Diagnostic, peak M1 assemblages, such as spinel-garnet-sillimanite migmatized metapelite, two-pyroxene mafics and orthopyroxene-garnet QFG, indicate crystallization at temperatures >800°C and pressures <4·5 kbar (Fig. 11). Equilibrium thermodynamics results indicate that matrix assemblages equilibrated at 740°C and 4·4 kbar. These lower temperature results suggest that the peak metamorphic minerals continued to re-equilibrate during cooling immediately after the peak, with chemical exchange ceasing at the conditions derived by equilibrium thermodynamics. Thus peak M1 conditions are approximately >800°C and 4·4 kbar. The near-isobaric spread in T results by equilibrium thermodynamics and near-isobaric vector between core and rim P-T loci (Fig. 8) suggest that the peak of M1 metamorphism was possibly terminated by isobaric cooling. Isobaric cooling subsequent to an anti-clockwise P-T path is suggested by phase stability constraints. For example, hercynitic spinel and coexisting sillimanite are often early phases enclosed by peak metamorphic garnet (Tables 2 and 3), indicating increasing pressure accompanying the peak of metamorphism (Fig. 8) (Harley, 1992). Furthermore, inclusion spinel has lower ZnO contents than matrix spinel, suggesting lower-pressure conditions before the peak of metamorphism. Garnet coronas enclosing pyroxenes in mafic gneisses and peak metamorphic sillimanite in metapelites (Tables 2, 3 and 4), constrain cooling to be isobaric or up-P across reactions of shallow positive [Delta]P/[Delta]T (Fig. 8) (Harley, 1992). Six samples from the GT equilibrated at conditions centred on 631°C and 5·6 kbar. These samples preserve peak M1 matrix assemblages, with only minor, or absent, corona reaction textures and SG4 foliations. These samples have re-equilibrated at 120-170°C lower than conditions under which the matrix assemblage crystallized. Re-equilibration occurred at pressures intermediate between the peak of M1 and M2 and at temperatures identical to the peak of M2. It is suggested that these samples have been re-equilibrated during the later M2 metamorphic cycle. The higher pressures are consistent with further burial of the GT during crustal over-thickening during the collisional Pan-African Orogeny (M2). The anhydrous M1 mineral parageneses were not conducive to recrystallization during the M2 metamorphic cycle and so were preserved. The paucity of mineral reaction textures, sub-grains and penetrative reworking fabrics suggest that re-equilibration of these samples did not involve significant recrystallization and formation of new mineral parageneses. Chemical re-equilibration without recrystallization was equally effective in all rock-types, with QFG, mafic gneiss and metapelites samples all having been re-equilibrated (Table 7). The timing of kyanite growth in the GT is not confidently known. In all cases, kyanite does not pre-date peak metamorphic sillimanite and because M1 mineral parageneses are far removed from the kyanite stability field (Fig. 8), kyanite is interpreted to have formed subsequent to M1. Kyanite only occurs at the southern margin of the GT and so is interpreted to have grown during reworking of the GT in the M2 metamorphic cycle (Fig. 11). Further expressions of the M2 metamorphic cycle include: garnet ± sillimanite ± ilmenite overgrowths on garnet porphyroblasts; garnet coronas enclosing actinolite coronas after M1 pyroxenes (Fig. 7f); and fine secondary sillimanite growth on M1 sillimanite, feldspars, Fe-Ti oxides, garnet and in fine late-stage (SG4) sillimanite ± biotite seams. These reaction textures suggest a post-M1 heating event (M2) crystallizing garnet and sillimanite and possibly at higher pressures than M1, passing through the kyanite field. Mylonite assemblages from the marginal shear zone of the GT were crystallized at approximately 590°C and 7·7 kbar. These mylonites formed during oblique overthrusting of the GT at the latter stages of deformation in the M2 metamorphic cycle. These mylonites record the maximum pressure conditions the GT experienced in response to further burial during crustal thickening in the M2 metamorphic cycle. Furthermore, the un-recrystallized but re-equilibrated GT samples discussed above preserve the peak temperature conditions experienced by the GT during M2 reworking. Pressures at this thermal maximum are lower than preserved in the mylonites, suggesting decompression through the thermal maximum (Fig. 11). Thus, together, the mylonites and re-equilibrated samples document two points on the clockwise P-T path that the GT experienced during reworking in the M2 metamorphic cycle (Fig. 11). Further burial and metamorphism with subsequent decompression through the thermal peak are typical of P-T paths accompanying crustal over-thickening (England & Thompson, 1984). Such a P-T path is compatible with the regional setting, delineating the GT as a thrust-bound terrane of basement incorporated within the Zambezi Belt (Fig. 1). In conclusion, Pan-African reworking resulted in further burial of the GT, as a thrust-bound slab, within over-thickened crust. Burial and heating during the M2 metamorphic cycle recrystallized very little of the earlier M1 parageneses in the GT, recrystallization being largely confined to the marginal shear zone and some late-stage, thin sillimanite-biotite foliation seams (SG4) and coronal garnet and sillimanite (Tables 2 and 3). Reworking during M2 resulted in chemical re-equilibration of M1 assemblages without recrystallization of mineral phases, re-equilibration being largely confined to within 2 km of the margin of the GT. The Quartzite and Zambezi Terranes both preserve evidence of having experienced at least two tectonothermal cycles. Aligned sillimanite inclusions within peak metamorphic garnet preserve a fabric formed during a metamorphic event before the pervasive tectonothermal events of the Pan-African Orogeny (M2). These early sillimanite ± spinel inclusions are incompatible with being prograde mineral parageneses (Fig. 11) formed during the moderate-P, clockwise P-T path experienced during M2 metamorphism (see below). The sillimanite ± spinel inclusion parageneses are compatible with the high-T-low-P metamorphic conditions experienced during M1 in the GT. The OT, ZT and QT were almost totally recrystallized during the M2 metamorphic cycle that accompanied NE over SW tectonic transport and crustal over-thickening in the Zambezi Belt during the Pan-African Orogeny at 524 ± 16 Ma. The peak metamorphic conditions experienced differ within each of these three terranes. Equilibrium thermodynamics and geothermobarometry indicate that average P-T conditions at the peak of metamorphism were ~7·9-8·6 kbar in all terranes and 590°C, 630°C and 717°C in the south ZT and OT, north ZT and QT, respectively. As previously discussed, these P-T estimates are entirely consistent with the stability of the diagnostic peak metamorphic assemblages of each terrane (Figs 3 and 11, Table 3). Vectors from core to rim P-T loci suggest near-isothermal decompression through the peak of metamorphism (Fig. 9); this is also well constrained by phase stability relationships. Decompression through the peak of metamorphism is evidenced by pervasive secondary sillimanite occurring as fine overgrowths on primary kyanite and sillimanite, sillimanite coronas enclosing a variety of peak metamorphic phases and in syn- to post-peak metamorphic foliations (SZ3-SZ5) of sillimanite-biotite ± ilmenite (Tables 2 and 3). The matrix assemblage equilibrated within the kyanite field in the south ZT and at the kyanite-sillimanite transition in the north ZT and QT. Thus the absence of secondary kyanite and common occurrence of sillimanite in reaction textures and SZ3-SZ5 foliations indicate decompression into the sillimanite field, possibly at constant to increasing temperature, immediately after crystallization of the matrix assemblages. A variety of garnet overgrowths and coronas are common in both aluminous and calcareous rock-types in the north ZT and QT, and garnet grew in equilibrium with SZ3-SZ4 foliations (Table 3). This secondary garnet growth accompanied decompression through the thermal peak of metamorphism. Plagioclase coronas on garnet porphyroblasts in mafic gneisses also indicate decompression through the peak (Green & Ringwood, 1967). Chloritoid inclusions within garnet porphyroblasts in kyanite-staurolite aluminous schists from the south ZT (Table 3) suggest that prograde metamorphism progressed from the chloritoid stability field into the staurolite stability field, at moderate pressures. The progression from early chloritoid parageneses to staurolite-kyanite matrix assemblages and further, to pervasive late-stage sillimanite growth in the other terranes, is indicative of a clockwise P-T path during the M2 metamorphic cycle. All samples preserve retrogressive hydration reaction textures that formed during cooling subsequent to the peak of M2 metamorphism. These are dominated by the formation of chlorite, muscovite and blue-green hornblende, and document cooling to temperatures <500°C, but do not constrain the pressures experienced during thermal relaxation. The only P constraint on the post-M2 cooling path is that the north ZT resided at ~2·7 kbar during emplacement of post-tectonic pegmatites at 480 ± 2 Ma (Goscombe et al., 1998). Mineral parageneses and equilibrium thermodynamics calculations indicate that the OT and south ZT equilibrated under the same P-T conditions (Tables 4 and 7, Appendix A), whereas the southeast margin of the OT is a distinct unit of talc-kyanite ± staurolite white schists. These white schists and yoderite-bearing white schists reported by
Johnson & Oliver, (1997) indicate equilibration at pressures much higher than the rest of the OT. Thus these white schists constitute a discrete fault-bounded block of high-P metamorphic rocks at the southeasternmost margin of the Chewore Inliers (Goscombe et al., 1998). The terranes that constitute the Chewore Inliers are lithologically distinct and have experienced variable reworking at different metamorphic grades. Only the OT appears to be of different age, and is interpreted to be metamorphosed oceanic crust of ~1393 Ma age. The GT and ZT have protoliths older than the 1071-1083 Ma orthogneisses and so may be contemporaneous with the OT. The present juxtaposition of the Chewore Inlier terranes occurred in two events. The OT was brought into contact with the ZT before DZ3 folding, with no further time constraints, and so could have occurred early in the Pan-African Orogeny (M2) or in an unrelated earlier tectonic event. High-P metamorphic rocks and eclogites occur at widely scattered localities defining a trace from the Mwembeshi shear zone in Zambia, along the centre of the Zambezi Belt to south Malawi (Andreoli, 1981; Dirks et al., 1997; Goscombe et al., 1998). The occurrence of what is interpreted to be oceanic crust in the OT, on this hypothetical trace, suggests that it may be a crustal suture (Johnson et al., 1996). Though geochronological data are limited, this hypothetical suture roughly coincides with the southern limit of localities that have experienced the 945 Ma tectonothermal event (M1) in the Zambezi and Mozambique Belts (Goscombe et al., 1998). Consequently, this hypothetical suture and juxtaposition of the OT with the terranes to the north occurred during or after the M1 tectonometamorphic events, as all subsequent tectonic events occur both north and south of this hypothetical suture (Goscombe et al., 1998). All the other Chewore Inlier terranes are bound by steep oblique thrusts and were brought into juxtaposition in the latter stages of the Pan-African Orogeny. The overthrusting of disparate terranes is a defining aspect of Neoproterozoic to early Palaeozoic tectonics in the Pan-African Orogenic System of south and east Africa. Crustal-scale thrusts and shear zones stacked and juxtaposed terranes of different character in the 510-560 Ma Pan-African Orogeny in the southern Ubendian Belt (Lenoir et al., 1993; Ring, 1993), the Mozambique Belt (Pinna et al., 1993) and in the Zambezi Belt in NE Zimbabwe (Barton et al., 1991, , 1993; Vinyu et al., 1997) and the Chewore Inliers (Goscombe et al., 1998). The juxtaposition of these terranes was the culminating phase of pervasive ductile fold repetition, crustal over-thickening and associated amphibolite facies metamorphism with a clockwise P-T path. The Pan-African Orogeny was contractional and involved NE over SW transport in the Zambezi Belt, and is responsible for the tectonic grain of the Zambezi Belt. The degree of reworking of older basement gneisses by the Pan-African Orogeny is heterogeneous in the Chewore Inliers. The GT is only partially tectonically reworked, this being confined to the bounding shear zones and spaced, fine foliation seams. Thermal reworking is evident as coronal reaction textures and re-equilibration of multi-equilibria P-T calculations. Towards the core of this terrane both these effects diminish, leaving very little expression of reworking of this terrane during the Pan-African Orogeny. These observations suggest the main Pan-African orogenic-metamorphic front was at the southern margin of the Zambezi Belt and that the GT was transported SW into juxtaposition with the more pervasively reworked terranes in the latter stages of the Pan-African Orogeny. Juxtaposition of terranes that have been variably reworked during the orogenic period that defines the metamorphic belt is a feature of many metamorphic belts, such as the Grenville Province (Rivers et al., 1989), Ungava Orogen (St-Onge & Ijewliw, 1996), European Alps (Ricou & Siddans, 1986), Scandinavian Caledonides (Hossack & Cooper, 1986) and Namaqua Belt (Thomas, 1989; Grantham et al., 1994). Such a common feature is an artefact of some terranes acting as coherent anhydrous slabs, with strain being partitioned into the bounding shear zones and not pervasively reworking the terrane. Furthermore, the bounding shear zones allow considerable lateral displacements and thus incorporation of variably reworked terranes into the more intensely deformed portion of the orogenic-metamorphic belt can occur at different times in the orogenic cycle. This is a more dynamic and heterogeneous view of metamorphic belts, in contrast to homogeneous crustal thickening and reworking in a metamorphic belt by ductile fold repetition alone. As previously discussed, about half of the GT samples, those closest to the thrust margin (Fig. 10), have been re-equilibrated during the inter-thrusting of the GT within the Zambezi Belt during the M2 tectonothermal cycle. Re-equilibrated samples still preserve M1 mineral parageneses, but the M1 phases have been re-equilibrated during M2 metamorphism, giving rise to results from multi-equilibria P-T calculations that are similar to metamorphic conditions experienced by the other terranes during M2 metamorphism. There is no obvious recrystallization of M2 mineral parageneses, nor a higher proportion of metamorphic reaction textures or late-stage foliations in the re-equilibrated samples. Furthermore, all rock-types appear to be equally represented in the re-equilibrated sample set. Though the GT was incorporated within the Zambezi Belt during the Pan-African Orogeny (M2), the M1 mineral assemblages were preserved because of their anhydrous nature and the M2 thermal perturbation experienced within the GT was possibly short lived (see below). Only in the case of the mylonite samples, from the marginal shear zones, were new mineral parageneses crystallized. In these samples the calculated M2 results are almost identical to the peak of metamorphism during M2. Of particular interest is that many GT samples were apparently not re-equilibrated during M2. This is evidenced by the high-T-low-P cluster (738°C, 4·4 kbar) of equilibrium thermodynamics results, which is consistent with closure during isobaric cooling after formation of M1 assemblages. The samples that have not been re-equilibrated during M2 mostly occur >2 km distant from the southeastern margin of the GT and all re-equilibrated samples are within 2 km of this margin (Fig. 10). Consequently, the kinetic energy imparted into the inter-thrust GT, in response to the elevated crustal geotherms associated with the M2 metamorphic cycle, dropped off rapidly towards the core of the GT. This implies a significant local disturbance of the crustal isotherms of M2 metamorphism, in response to inter-thrusting of a relatively cooler basement terrane. Furthermore, the M2 thermal perturbation experienced within the GT must have been short lived or this local disturbance of the isotherms would have been superseded by the regional isotherms, giving rise to re-equilibration and possibly recrystallization also of the entire GT. This scenario is supported by the inter-thrusting and juxtaposition of the GT with the more intensely reworked terranes occurring in the latest deformational episode of the Pan-African Orogeny (DG5-DZ6, Table 1). No pervasive hydrothermal retrogression is recognized in the GT; therefore re-equilibration is considered to be due not to pervasive fluid flow but to thermal re-equilibration of a near-anhydrous terrane. Tectonic reworking of older high-grade terranes, by a later orogenic event involving crustal over-thickening, is recognized in other metamorphic belts (Goscombe, 1992; St-Onge & Ijewliw, 1996). In contrast to the GT in the Chewore Inliers, reworking typically involved significant degrees of recrystallization, giving rise to new mineral parageneses and formation of pervasive fabrics. The recognition of chemical re-equilibration of M1 mineral assemblages, without pervasive formation of new mineral parageneses, by the later M2 tectonometamorphic cycle, offers a model for apparent isobaric and up-P cooling paths in other metamorphic belts. Isobaric and up-P cooling paths have been well documented and confidently established by phase stability constraints in metamorphic belts world-wide (Clarke et al., 1987, , 1989; Harley, 1989). However, this study of the Chewore Inliers illustrates the potential for the incorrect interpretation of isobaric or up-P cooling paths if based on P-T calculations alone, without phase stability constraints. For example, a simplified interpretation of the GT, involving only one metamorphic cycle to explain the complete spread of calculated P-T results, without additional constraints such as metamorphic reaction textures and tectonometamorphic history of adjacent terranes, suggests an up-P cooling path (Fig. 8). This spread in average P-T results is in fact due to re-equilibration of some samples during a later tectonothermal event (M2). Consequently, other metamorphic belts with a spread in average P-T loci similar to that of the GT could potentially be misinterpreted as being due to an isobaric or up-P P-T path if re-equilibration by later tectonothermal events is not recognized. Edward and Fanta Gwera, Numeri Kaudur, Edimore Mupapudzi, Murrey and Ned are sincerely thanked for their great company and help with the fieldwork. Peter Fey, Fritz Both, Peter Zizhou and Mr Lunga are acknowledged for their mapping. The staff of the Zimbabwe Geological Survey and the directors, Dr J. Orpen and Mr S. M. N. Ncube are thanked for initiating and giving considerable support during this project. Tom Blenkinsop, Paul Dirks and Mike Vinyu are thanked for their helpful discussions. The efforts and comments of the reviewers, Kurt Bucher, Simon Harley, Michael Daly and Uwe Ring, are greatly appreciated. Dr Wieslaw Jablonski is thanked for his help with electron microprobe analyses at the Central Science Laboratories, University of Tasmania.INTRODUCTION
REGIONAL GEOLOGY AND GEOCHRONOLOGY
STRUCTURAL EVOLUTION OF THE CHEWORE INLIERS
M1 high-grade metamorphic cycle
M2 Pan-African metamorphic cycle
Late-stage events
PETROLOGY OF THE CHEWORE INLIERS
Granulite Terrane
Quartzite Terrane
North Zambezi Terrane
South Zambezi Terrane and Ophiolite Terrane
MINERAL CHEMISTRY
EQUILIBRIUM THERMODYNAMICS
GEOTHERMOBAROMETRY
TECTONOMETAMORPHIC EVOLUTION OF THE GRANULITE TERRANE
TECTONOMETAMORPHIC EVOLUTION OF THE QUARTZITE, ZAMBEZI AND OPHIOLITE TERRANES
DISCUSSION
Collage nature of the Zambezi Belt
Re-equilibration of the Granulite Terrane
Tectonic reworking and apparent `isobaric' cooling paths
ACKNOWLEDGEMENTS
REFERENCES
Appendix A: Chewore Inliers equilibrium thermodynamic results derived using Thermocalc v2.0b (Powell & Holland, 1988).
Appendix B: Chewore Inliers geothermobarometry results.
Appendix C: Calculation of end-member activities.