Sampling

Samples were collected over a fairly restricted area within the OCD to ensure that the rocks had all been subjected to similar metamorphic histories. Samples were selected to be representative of the major lithostratigraphic units (Fig. 2). A description of the samples and more detailed localities are provided in the Appendix.

Figure 2. Simplified geological map of the Okiep Copper District and the location of samples analysed in the present study.

Analytical procedure

Samples of 1-2 kg were subjected to routine heavy mineral separation, which involved crushing and subsequent Wifley Table, heavy liquid and Frantz magnetic separation. Zircons were hand-picked, mounted in epoxy, and polished. Individual zircon grains were subjected to U-Pb isotopic analysis using the Sensitive High-Resolution Ion Microprobe (SHRIMP II) developed at the Research School of Earth Sciences at The Australian National University, Canberra.

Detailed SHRIMP analytical procedures have been reported by Compston et al., (1984) and Williams & Claesson, (1987). Briefly, the technique focuses a primary beam of negative oxygen ions in vacuo onto the zircon surface, from which a small area (25-30 µm diameter) of sputtered positive secondary ions is extracted. Secondary ions, which include Zr, Th, U and Pb from the zircon, are passed through a curvilinear flight path in a strong magnetic field and then counted at a mass resolution of 6500 on a single collector using cyclic magnetic stepping. Isotopic ratios and inter-element fractionation are monitored by continuous reference to a standard Sri Lankan zircon (SL13), fragments of which are mounted with each sample. Progressive changes in the Pb/U ionic ratio during sputtering were corrected using an empirical quadratic relationship between Pb⁺/U⁺ and UO⁺/U⁺ determined for the standard zircon (Claoue-Long et al., 1995). A radiogenic ²⁰⁶Pb/²³⁸U ratio of 0·0928 for the standard zircon, corresponding to an age of 572 Ma, is obtained through standard isotope dilution analysis. Initial Pb isotope compositions of the analysed zircons are assumed to be similar to that of model-derived average crustal Pb of similar age according to Cumming & Richards, (1975).

The analytical precision of Pb isotope ratios is controlled by machine counting statistics whereas the precision of Pb/U ratios is affected by uncertainties in the standard calibration. Ages were calculated using the recommended Steiger & Jäger, (1977) decay constants. Weighted mean ²⁰⁷Pb/²⁰⁶Pb ages were obtained for zircons exhibiting an obvious clustering and indistinguishable ²⁰⁷Pb/²⁰⁶Pb ages at a 95% confidence level (i.e. t[sgr], where t is the Fisher's t). Individual analyses in the data tables and concordia diagrams are presented at 1[sgr]. Errors associated with scatter within the cluster are obtained by standard statistical techniques. Dispersion within the cluster is attributed to modern Pb loss. Analyses that do not fall within the clusters of weighted mean ages probably reflect uncertainties caused by ancient Pb loss or overlap of the ion beam onto domains of the zircon crystal with different isotopic ratios.

Zircon morphology, U-Th contents and recognition of magmatic vs metamorphic growth

The interpretation of age data in this study depends on reliably distinguishing between xenocrystic, magmatic and metamorphic zircon. Most zircon types can be classified by morphology and by the assumption that zircon rims are younger than the cores they overgrow. Magmatic and metamorphic events are distinguished with the accumulation of data and the development of systematic patterns (see Fig. 9, below) that accord with geological events in the region. In the present study, zircon types were also recognized from their U and Th contents (Fig. 4, below).

Zircons in rocks of the OCD are composite and heterogeneous. They reflect the complex association between preservation or incorporation of xenocrystic grains, magmatic growth and metamorphic overgrowth. Zircons can be subdivided morphologically into (1) composite grains comprising a core and, generally, a single rim overgrowth, and (2) homogeneous, mostly structureless grains (Fig. 3). Cores represent either xenocrysts or magmatic grains; the latter are commonly recognized in the orthogneisses of the OCD by delicate magmatic zonation that is texturally distinct from the more homogeneous metamorphic rims. Structureless grains may be either magmatic or metamorphic. The distinction is particularly unclear in igneous rocks emplaced during peak metamorphism. Metamorphic overgrowth of new zircon on older cores is most likely either during the mineral reactions occurring at the metamorphic peak, or during high-temperature fluid infiltration. In these rocks the likely sources of fluid are syn-metamorphic magmas or crystallizing migmatite leucosomes (Waters, 1988). In any case the overgrowths are linked to the climax of a regional metamorphic event and give an indication of its age. Overgrowths on zircons from the Concordia and Rietberg granites, however, record distinctly younger ages than the major regional period of metamorphism and it is, therefore, necessary to interpret each case in its geological context. It is notable that few magmatic zircon overgrowths appear to nucleate around xenocrystic cores.

Figure 3. Microphotographs showing the composite morphology of zircons from various rock types in the Okiep Copper District; plane transmitted light (PT) and cathodoluminescence imagery (CL). Ages shown for individual spot analyses are ²⁰⁷Pb/²⁰⁶Pb ages given in Tables 2- 5. (a) Zircons with distinctive cores and overgrowths from the Modderfontein gneiss (PT). (b) Zircon with distinctive core and overgrowth from anorthosite of the Koperberg Suite (PT). (c, d) Zoned zircons from the Concordia granite. Zircon overgrowths in this sample, as well as the Rietberg granite, yielded young ages of ~850 Ma. (PT and CL.) (e, f) Zircons from the two-pyroxene granulite revealing no apparent zonation in plane transmitted light; CL imagery shows a distinctive core and a complex pattern of overgrowths. (g, h) Zircons from the diorite of the Koperberg Suite. CL imagery shows the presence of structureless grains or distinctive low-U cores with pronounced overgrowths.

In the OCD, magmatic and metamorphic zircon domains exhibit diagnostic U and Th contents. Most metamorphic overgrowths show higher U contents and lower Th/U ratios than magmatic or xenocrystic zircons in the same rock (Fig. 4). This may result from selective partitioning of U (but not Th) into the fluid that accompanied granulite-facies metamorphism and growth of new zircon. Although experimental work at high temperatures is limited, Nguyen Trung, (1985) has shown that at T = 400°C and 5 <= pH <= 9, the UO₂(OH)₂° complex could stabilize uranium in an aqueous solution. The metamorphic fluid would have to have been sufficiently oxidizing to convert tetravalent U into its labile hexavalent form, without similarly affecting Th.

Figure 4. Plots of U vs Th for the various zircon grains and overgrowths analysed in the present study, for (a) the Little Namaqualand Suite and (b) the Spektakel Suite.

In a U vs Th plot for the Little Namaqualand Suite, the distinction between metamorphic rims vs magmatic cores and structureless grains is obvious (Fig. 4a). Although the Th contents of all grains range from 50 to 250 ppm, the U contents of metamorphic rims are markedly higher. The Rietberg granite, however, contains a population of clear, structureless zircon grains (a type interpreted as magmatic in the Little Namaqualand Suite), which has U contents characteristic of metamorphic growth that distinguish it from the magmatic population of this sample (Fig. 4b).

Zircon U-Pb isotope data

Gladkop Suite (NAM6 and NAM7)

The granitic Brandewynsbank orthogneiss (NAM6) contains two types of zircon; large, brown composite grains with well-defined cores and rims, along with small, clear, structureless grains. The latter are regarded as a magmatic population that yields a weighted mean ²⁰⁷Pb/²⁰⁶Pb age of 1822 ± 36 Ma (Fig. 5a). This is indistinguishable from the whole-rock Rb-Sr age of 1824 ± 70 Ma (Barton, 1983) for the Brandewynsbank orthogneiss suite and is likely to be the emplacement age of the igneous precursor to the suite. Zircon cores are mainly magmatic but are discordant and underwent Pb loss, probably during metamorphism of the suite. One xenocrystic core has a ²⁰⁷Pb/²⁰⁶Pb age of 2018 ± 8 Ma (Table 2). The zircon rims define a tight cluster of data with a weighted mean ²⁰⁷Pb/²⁰⁶Pb age of 1032 ± 18 Ma (Fig. 5a). This is interpreted as the age of metamorphism of the suite.

Table 2. U-Th-Pb SHRIMP data for the Gladkop Suite

Figure 5. ²⁰⁶Pb/²³⁸U vs ²⁰⁷Pb/²³⁵U concordia plots for zircons from (a) Brandewynsbank gneiss and (b) Noenoemaasberg gneiss of the Gladkop Suite.

Zircons from the leucocratic Noenoemaasberg orthogneiss (NAM7), which is intimately interlayered with NAM6, are characterized by very high uranium contents and discordant U-Pb isotopic ratios (Fig. 5b). Little meaningful age information is available from this sample although the available data are consistent with it having been subjected to a similar history to the Brandewynsbank gneiss.

Little Namaqualand Suite (NAM1 and NAM2) and two-pyroxene granulite (NAM3)

The Modderfontein orthogneiss (sample NAM1), which intrudes the Nababeep orthogneiss near the Springbok dome, yields well-constrained U-Pb zircon age data. Zircons are typically composite with well-defined cores and rims, as well as sparse clear, structureless grains. Cores and clear, structureless grains define a weighted mean ²⁰⁷Pb/²⁰⁶Pb age of 1199 ± 12 Ma, whereas the rims define an equally well-constrained age of 1032 ± 12 Ma (Fig. 6a). The former date is statistically indistinguishable from whole-rock Rb-Sr dates (1179 ± 28 Ma, Barton, 1983; 1223 ± 48 Ma, Clifford et al., 1995) and probably represents the emplacement age of the magmatic protolith. The latter date is identical to that of zircon rims from the Gladkop Suite and is interpreted as the age of a metamorphic overprint that affected both units.

Figure 6. ²⁰⁶Pb/²³⁸U vs ²⁰⁷Pb/²³⁵U concordia plots for zircons from (a) Modderfontein gneiss, (b) Nababeep gneiss and (c) two-pyroxene granulite of the Little Namaqualand Suite.

Sample NAM2 from the Nababeep gneiss was collected close to its intrusive contact with the Modderfontein orthogneiss (Fig. 2). Zircons are very similar in appearance to those from the Modderfontein gneiss. Isotopically, however, the grains exhibit a more complex age pattern (Fig. 6b). A discordant zircon core (grain 6) reveals an inherited xenocrystic component with a minimum ²⁰⁷Pb/²⁰⁶Pb age of 1482 ± 18 Ma. Most of the remaining data points plot as a cluster of variably discordant analyses, with the nine most concordant analyses giving a weighted mean ²⁰⁷Pb/²⁰⁶Pb age of 1212 ± 11 Ma. This group includes clear, structureless zircon grains, as well as more complex crystals with structureless cores that grade into compositionally zoned margins. These zircons are of magmatic origin (see U and Th contents in Table 3 and Fig. 4a) and probably reflect the emplacement age of the magmatic protolith.

Table 3. U-Th-Pb SHRIMP data for the Little Namaqualand Suite

Zircon overgrowths in sample NAM2 are variably discordant and exhibit a range in ²⁰⁷Pb/²⁰⁶Pb ages (Table 3). Four of these overgrowths probably grew between 1030 and 1060 Ma in response to the same metamorphic event as recorded by the Modderfontein gneiss and older Gladkop Suite. Other overgrowths appear to have grown at ~1165 Ma (analyses 4.1 and 7.1, Table 3). These are interpreted as the products of polymetamorphic growth, in response to either local (contact?) metamorphism of the NAM2 sample (Nababeep gneiss) by intrusion of the nearby Modderfontein gneiss at 1200 Ma, or the regional metamorphism between 1060 and 1030 Ma. Age patterns of all the zircons in the Nababeep gneiss are further complicated by multi-stage Pb loss.

The two-pyroxene granulite is an enigmatic unit of mafic to intermediate composition that occurs as layers and elongate lenses over a large area of the NMC. It has been interpreted as xenoliths of the Lammerhoek Subgroup, or as a metamorphosed mafic sill. Zircons from the two-pyroxene granulite are typically subhedral with no well-defined core-rim texture. Although most of the zircons are apparently homogeneous in appearance, they define two age populations, which generally cannot be related to texturally identifiable isotopic domains. The older zircon domains have a weighted mean ²⁰⁷Pb/²⁰⁶Pb age of 1168 ± 9 Ma, whereas the younger domain has an age of 1063 ± 16 Ma (Fig. 6c). The former date resembles a multiple zircon U-Pb concordia age of 1160 ± 50 Ma (Clifford et al., 1981) that has been interpreted as the age of emplacement of the protolith. This is consistent with an interpretation of the two-pyroxene granulite as a sill in the ~1220 Ma Nababeep gneiss. The 1063 Ma age reflects a metamorphic overprint that produced the poorly structured zircon growths.

Spektakel Suite (NAM4 and NAM8)

Zircons in the Concordia granite (NAM4) are generally small, cracked and metamict as a result of high U and Th contents (Table 4). Most grains are composite and have well-defined cores and rims. One core (grain 1.1, Fig. 7a) is xenocrystic and has a ²⁰⁷Pb/²⁰⁶Pb age of 1161 ± 15 Ma. The other cores commonly exhibit a primary growth banding (Fig. 3) and appear to be a magmatic population, which is consistent with their U and Th contents (Fig. 4b). Zircon cores yield a weighted mean ²⁰⁷Pb/²⁰⁶Pb age of 1064 ± 31 Ma, interpreted as the emplacement age of the granite. This is somewhat younger than a whole-rock Rb-Sr errorchron age of 1105 ± 24 Ma (Clifford et al., 1995). Although zircon rims are discordant because of metamictization and recent lead loss, they provide a relatively imprecise age of 861 ± 45 Ma. This age is similar to K-Ar and Ar-Ar whole-rock ages of 800-850 Ma (Clifford et al., 1995), a range attributed to cooling plateaux in these isotope systems. This explanation does not, however, accord with growth of zircon rims at this time. The latter may reflect the initiation of a new orogenic episode at ~850 Ma (see below).

Table 4. U-Th-Pb SHRIMP data for the Spektakel Suite

Figure 7. ²⁰⁶Pb/²³⁸U vs ²⁰⁷Pb/²³⁵U concordia plots for zircons from (a) Concordia granite and (b) Rietberg granite of the Spektakel Suite.

Zircons from the Rietberg granite (NAM8) are similar to, but of even poorer quality than, those from the Concordia granite. Cores and rims exhibit similar age patterns to those seen in zircons from the Concordia granite. One xenocrystic core (grain 12.1, Fig. 7b) has a ²⁰⁷Pb/²⁰⁶Pb age of about 1229 ± 34 Ma. The other cores are probably magmatic; they yield a weighted mean ²⁰⁷Pb/²⁰⁶Pb age of 1058 ± 30 Ma. Although this age is statistically indistinguishable from that of the Concordia granite, it is consistent with field observations indicating that the Rietberg intrusion cuts the Concordia granite. Zircon rims yield an age of 841 ± 23 Ma, similar within error to those from the Concordia granite. The Rietberg granite also contains a few small, clear, structureless zircons with variable discordant ages from 800 to 950 Ma. These grains have higher U contents than the magmatic cores (Fig. 4b) and probably grew during the 850 Ma event.

Koperberg Suite (NAM9, NAM10 and NAM11)

The units of the Koperberg Suite sampled for the present study are part of a sequence that cuts the Concordia granite (Fig. 2; Van Zweiten et al., 1996) and should, therefore, be younger than 1064 Ma. Anorthosite sample NAM9 is a large xenolith within diorite and hypersthenite of the Koperberg Suite. Most zircons from NAM9 are composite grains with well-defined cores and rims, along with a few clear structureless grains (Fig. 3). One of the zircon cores is xenocrystic with a ²⁰⁷Pb/²⁰⁶Pb age of 1732 ± 17 Ma (Table 5). Most of the other cores, together with a single analysis of a structureless grain, yield a weighted mean ²⁰⁷Pb/²⁰⁶Pb age of 1202 ± 25 Ma (Fig. 8a). Analyses of zircon overgrowths are generally discordant (correlating with high U concentrations) and some have substantial common Pb contents. An imprecise age of 1037 ± 86 Ma can be calculated from three nearly concordant analyses (3.3, 5.2 and 9.2), and is interpreted as the timing of the regional metamorphic overprint. These data agree with the U-Pb zircon data of Clifford et al., (1995) for andesine anorthosite that cuts the Nababeep gneiss.

Table 5. U-Th-Pb SHRIMP data for the Koperberg Suite

Figure 8. ²⁰⁶Pb/²³⁸U vs ²⁰⁷Pb/²³⁵U concordia plots for zircons from (a) anorthosite (b) diorite and (c) hypersthenite of the Koperberg Suite. Data are plotted according to the zircon type (core, overgrowths, structureless), as for the other concordia plots.

Zircons in the Koperberg Suite diorite (NAM10), which intrudes the anorthosite, are generally structureless, brown, sub- to euhedral grains of excellent quality and clarity. A smaller population of composite grains of similar habit, but with well-defined cores and rims, is also present. Isotopic ratios from all these zircon types are indistinguishable (i.e. structureless grains 1061 ± 20 Ma, cores 1053 ± 20 Ma and overgrowths 1056 ± 11 Ma). Collectively, the NAM10 zircons yield a weighted mean ²⁰⁷Pb/²⁰⁶Pb age of 1057 ± 8 Ma ( Fig. 8b). In contrast to the data for zircon overgrowths in other samples (NAM1, NAM6 and NAM11), the overgrowths from NAM10 do not show the distinctive low Th/U ratios of metamorphic rims and appear to be related to the same magmatic-thermal event as the cores and unstructured grains. The diorite was probably emplaced at 1057 ± 8 Ma, coeval with the onset of regional metamorphism which peaked at ~1030 Ma. Early Pb loss (as a result of regional metamorphism) caused some scatter in the ²⁰⁷Pb/²⁰⁶Pb ratios, and possibly blurred the age distinction between the two events. Alternatively, the two events may be statistical artefacts. In this case, magmatism and metamorphism might have been coeval between 1060 and 1030 Ma.

Zircons from the Koperberg Suite hypersthenite (NAM11), which intrudes both the diorite and anorthosite, are identical in habit to those in the diorite, although there are fewer composite zircon grains in the hypersthenite (Fig. 3). Three zircon cores are xenocrystic and yield ²⁰⁷Pb/²⁰⁶Pb ages of 1135 ± 24, 1272 ± 49 and 1561 ± 67 Ma. The remaining analyses of structureless grains and overgrowths have ²⁰⁷Pb/²⁰⁶Pb ages of 1028 ± 11 Ma and 1049 ± 15 Ma, but are statistically indistinguishable. A combined weighted mean ²⁰⁷Pb/²⁰⁶Pb age of 1037 ± 8 Ma (Fig. 8c) for the hypersthenite is consistent with the fact that it intrudes the NAM10 diorite. The overgrowths, however, have the low Th/U signature interpreted in other OCD samples as the signature of metamorphic zircon growth. The age of the hypersthenite zircon overgrowths is also recorded by other rock units, in which it reflects the peak regional granulite-facies overprint. Structureless zircons from the hypersthenite sample have Th/U ratios that suggest magmatic growth. The hypersthenite is, therefore, regarded as having been emplaced at 1037 ± 8 Ma, coeval with the mid-crustal thermal perturbation responsible for the peak of regional prograde granulite-facies metamorphism in the region.

Events in the Okiep Copper District

The present study confirms the existence of an episodic history in the Palaeo- to Mesoproterozoic development of crust in Namaqualand; a summary of the sequence of events in the OCD, as recorded in the new SHRIMP U-Pb zircon ages is presented in Fig. 9. The 1822 ± 36 Ma orthogneisses of the Gladkop Suite, in the northern portion of the OCD, are the only major remnant of Kheisian crust in the area. These rocks are mainly of amphibolite facies (Fig. 1) and petrographic evidence for an earlier metamorphism is lacking. However, xenocrystic zircon cores from the Brandewynsbank orthogneiss, and detrital zircon grains in Khurisberg Group quartzites, yield ages of 2020, 1940 and 1920 Ma (Fig. 9), similar to published whole-rock isochron ages for the Orange River Sequence (Table 1). In addition, most detrital grains in the Khurisberg quartzite display overgrowths that record a well-defined concordia intercept age of 1850 Ma. The age of these rims is evidence of a metamorphic overprint (Fig. 9) in the Kheisian source rocks that supplied the detritus to the Khurisberg Group (L. J. Robb et al., unpublished data, 1995).

Figure 9. A summary of the sequence of events in the Okiep Copper District as recorded in the SHRIMP U-Pb zircon dates of the present study.

Although two xenocrystic zircon cores record ages around 1500 Ma (Fig. 9), little else is preserved in the U-Pb zircon dating record in the OCD until a major pulse of magmatism associated with the Kibaran Orogeny. This magmatism is represented by intrusion of the voluminous Nababeep (at 1212 ± 11 Ma) and Modderfontein (at 1199 ± 12 Ma) orthogneisses, as well as a mafic sill (at 1168 ± 9 Ma). The regionally significant D₂ deformation either accompanied or immediately post-dated this event, and possibly persisted until the Namaquan (Fig. 9). The D₂ deformation may comprise two discrete events, D_2a and D_2b, as suggested by Raith & Harley (1998). The abundant xenocrystic zircon cores within the younger Spektakel and Koperberg Suites, which record ages of between 1140 and 1270 Ma (Fig. 9), also reflect the widespread development of Kibaran crust in the OCD.

A major pulse of magmatism also accompanied the Namaqua Orogeny, which took place between 1065 and 1030 Ma. Voluminous, sheet-like bodies of granite (the Spektakel Suite) were intruded between 1064 and 1058 Ma, parallel to recumbent D₂ fabrics and lithologies. A compressional stress regime persisted with the onset of D₃, forming open folds and fabric rotation that resulted in the formation of steep structures. The steep structures were the preferred sites of emplacement of the Koperberg Suite intrusions between 1060 and 1040 Ma (Fig. 9).

Petrographic evidence indicates that granulite-facies metamorphism of the NMC post-dated or outlasted the D₂ deformation. Zircon overgrowths in rocks of the Gladkop and Little Namaqualand Suites, dated at 1060-1030 Ma, are widespread. Furthermore, both metamorphic and magmatic zircon domains in the Spektakel and Koperberg Suites are essentially indistinguishable in age and also fall in the range 1060-1030 Ma. Thus, the metamorphic peak in the OCD was contemporaneous with Namaquan magmatism. In contrast, Raith & Harley, (1998) have suggested that maximum granulite-facies conditions were reached during D_2a (between 1200 and 1060 Ma, before the intrusion of the Spektakel Suite) and that a period of cooling intervened before a second peak in the OCD at 1030 Ma. The present study does not support this interpretation, as no metamorphic zircon growth was recorded before 1060 Ma, nor is there evidence for magmatic input of heat in the OCD in the interval 1170-1060 Ma.

Other metamorphic events are also seen in the zircon age data. In the Nababeep gneiss, several zircon overgrowths appear to define a Pb-loss discordia chord between ~1200 and 1000 Ma (Fig. 6a), suggesting that they formed during contact metamorphism that attended intrusion of Modderfontein orthogneiss, and then lost Pb during the regional granulite-facies overprint. In the Spektakel Suite, most zircon overgrowths are dated at 860-840 Ma. These appear to define a metamorphic overprint of less severe magnitude or more restricted extent than the regional granulite-facies event. These ages might reflect the onset of an early Pan-African, amphibolite-grade, overprint which is well expressed in the coastal belt to the west of the OCD.

Recent dating in other Mesoproterozoic terranes, such as Natal, also suggests that crustal growth might have occurred in two discrete episodes, at 1220-1170 Ma and 1060-1030 Ma. The voluminous rapakivi-textured granitoids of the Oribi Gorge Suite, which underlie much of the Natal Metamorphic Complex, have been dated at 1068-1029 Ma (Thomas et al., 1993), whereas other orogenic granitoids that pre-date these appear to have an age of ~1200 Ma. In the Kirwanveggen, Sverdrupfjella and Heimefrontfjella regions of Mesoproterozoic Antarctica, however, a major event of zircon growth-overgrowth took place at 1150-1100 Ma (Arndt et al., 1991; Harris et al., 1995), with little or no record of the discrete Kibaran and Namaquan events described above.

Crustal growth

Continental crust in pre-Kibaran times consisted of the amalgamated Archaean Kaapvaal-Zimbabwe Craton, which had accreted onto its western and southwestern flanks a complex 2000-1800 Ma Kheisian terrane, the origin of which is still not fully understood (Hartnady et al., 1985; Thomas et al., 1994). In western Namaqualand, north of the OCD, a pristine, relatively unmetamorphosed, remnant of this crust (the Richtersveld terrane) is preserved as the Vioolsdrif Suite and Orange River Sequence, which comprise a calc-alkaline composite batholith and associated extrusives perhaps akin to a magmatic arc adjacent to a convergent continental margin (Reid & Barton, 1983). The southern margin of the Richtersveld terrane is bounded by the southward verging Groothoek thrust, on the footwall side of which the largely amphibolite-grade Gladkop Suite orthogneisses are located. The region south of the Groothoek thrust is made up largely of potassic, high-SiO₂ orthogneisses of distinctly peraluminous character, some of which were emplaced at 1822 ± 36 Ma (Fig. 5). Although the available initial ⁸⁷Sr/⁸⁶Sr ratio (R₀) is relatively low (i.e. 0·705 ± 2; Barton, 1983) the suite is interpreted as an S-type granite that was transformed to gneiss early in its crustal history (Reid & Barton, 1983). Parts of the Kheisian crust were metamorphosed at 1850 Ma, as suggested by the age of zircon rims on detrital grains in Khurisberg Group quartzites (Fig. 9). The pre-Kibaran crust in the OCD, therefore, consisted of both Richtersveld and Gladkop material, the only remnants of which are tectonically interleaved slices, or digested components reflected in the xenocrystic zircon population. A volcano-sedimentary supracrustal package, the Aggeneys Sequence, was also deposited between 1600 and 1300 Ma (Table 1), possibly in response to rifting of the existing continent and deposition in extensional, intracratonic basins (Moore et al., 1990; Thomas et al., 1994).

Kibaran crustal formation in the OCD began at ~1210 Ma with the intrusion of the Nababeep orthogneiss (and its correlatives) over a wide area, followed by intrusion of the Modderfontein orthogneiss over a more restricted area at 1200 Ma. The Little Namaqualand Suite displays a high R₀ of 0·725 ± 3 (Barton, 1983) and represents partial melt products of older Proterozoic crust. The age and origin of the protolith is uncertain because Kheisian aged zircon xenocrysts are scarce in the samples analysed (Fig. 9). McCarthy, (1976) suggested that these orthogneisses are the residues from various increments of melt extraction. However, Reid & Barton, (1983) viewed these rocks as an accumulated melt fraction derived from progressive partial melting of a crustal source below the present exposure level of the OCD (because the NMC records limited evidence of in situ anatexis and migmatization; Waters, 1988), with some in situ crystal fractionation. Kibaran anatexis might have accompanied the crustal thickening and convergence associated with the D₂ deformation, in response to the collision between Kheisian crust and the Kaapvaal-Zimbabwe Craton (Thomas et al., 1994).

Crustal formation during the Namaqua Orogeny is represented by Spektakel Suite granites and the Koperberg Suite dated at 1060-1030 Ma. The Concordia granite is a potassic, marginally peraluminous, high-SiO₂ intrusion with a markedly negative, crustal, [epsilon]_Nd signature (-7; Clifford et al., 1995), but relatively primitive R₀ (0·709 ± 3). Raith, (1995) regarded the suite as S-type, but did not preclude contributions from A- or I-type magmatic precursors. The markedly differentiated trace element chemistry of the suite suggests that in situ feldspar fractionation took place within the intrusions (McCarthy, 1976). The Concordia granite was emplaced in late D₂ times, but before D₃, because many steep structures are present in it.

The Koperberg Suite includes andesine anorthosite, biotite diorite, leuconorite, norite and hypersthenite. The sequence of intrusion of the members of the suite exhibits a well-defined reversal of the normal differentiation trend, which suggests either progressive partial melting or the tapping of a differentiated magma chamber from the top downwards. The precursor to the Koperberg Suite had variable but high R₀ (up to 0·727), high initial U/Pb ratio and a markedly negative [epsilon]_Nd signature (-9; Clifford et al., 1995). Clifford et al. suggested that the Koperberg Suite was derived from partial melting of fertile lower-crustal material of intermediate composition. In contrast, Boer et al., (1994) and Van Zweiten et al., (1996) suggested that mafic members of the suite originated in the mantle and were contaminated by both wall rock and associated lower-crustal melts to form the more felsic components. This model is supported by detailed Sr isotopic studies of the Koperberg Suite, which show bimodality in the distribution of R₀ values (Brandriss & Cawthorn, 1996). Abundant xenocrystic zircon, particularly in high R₀ anorthosites (Clifford et al., 1995; Figs 8 and 9), also indicates the effects of significant crustal contamination.

Gibson et al., (1996) described the Namaquan orogenic event as a cycle of prograde metamorphism and compressive deformation (i.e. D₂ followed by D₃) culminating in a granulite-facies climax. Because it is improbable that a single deformation event could have persisted for >100 my, those workers regarded the deformation of the Concordia granite as late D₂. This model is inconsistent with the widespread preservation of metamorphism and deformation related to the Kibaran Orogeny in the OCD referred to above. The geochemical character of the Little Namaqualand Suite is, in fact, more compatible with a synorogenic setting than is that of the Spektakel Suite. Microstructural evidence (Waters, 1989, , 1990) indicates that D₂ deformation is intimately associated with the prograde metamorphism up to amphibolite grade, but that the granulite-facies assemblages are essentially post-D₂. The present study, therefore, proposes that the D₂ deformation and amphibolite-facies prograde metamorphism is an integral part of the Kibaran orogenic cycle at ~1200 Ma. The discrete Namaquan event at 1060-1030 Ma was a heating event that accompanied intrusion of the Spektakel and Koperberg magmatic suites, and was associated with limited compressional deformation (D₃).

P-T-t path

An anticlockwise P-T-t path for the evolution of the granulite-facies rocks of western Namaqualand (Waters, 1989) is well established. Granulite-facies metamorphism was previously assumed to coincide with the Rb-Sr errorchron age for the Little Namaqualand Suite [given as 1187 ± 22 Ma by Clifford et al., (1975), or 1223 ± 48 Ma by Clifford et al., (1995)] and to reflect the fact that these rocks were synorogenic, had been affected by the granulite-facies overprint and, accordingly, had their Rb-Sr isotopic system reset by the latter event. However, the Modderfontein orthogneiss, as well as other units in the region, clearly demonstrates zircon overgrowths at 1060-1030 Ma, indicating that peak metamorphism was considerably later than ~1200 Ma. Rather than reflecting a metamorphic resetting, the Rb-Sr data for the Little Namaqualand Suite (and indeed for other units in the region) probably provide an imprecise indication of the protolith age. Such behaviour is perhaps feasible in granulite-facies terranes where low aH₂O, and restricted or closed system element mobility, could account for the preservation of original Rb-Sr isotopic systems. The previously published age data for the NMC (Table 1) are very similar to the age constraints derived from this study.

The prograde P-T path for both lower and upper granulite-facies rocks of the NMC (Fig. 10) utilizes the thermobarometric constraints summarized by Waters, (1989) and the age data of this study. The overall shape of the Namaquan P-T path is also likely to be anticlockwise because the emplacement of sheet-like granitoids of the Spektakel Suite, followed by D₃ deformation, would increase pressure at a near-maximum temperature. The transition from amphibolite to lower granulite grades and the transition from the D₂ to D₃ stress regime accompanied a 100 my hiatus, between 1170 and 1060 Ma (Fig. 10). The granulite-facies prograde path was initiated, at ~1060 Ma, from heat supplied by under- or intra-plating of mantle-derived basic magma, by delamination of the lower lithosphere, or by both (Waters, 1990; Gibson et al., 1996).

Figure 10. Pressure-temperature-time path for the evolution of the Okiep Copper District during the Kibaran and Namaquan orogenies.

In contrast to a rapid onset of prograde metamorphism, the retrograde path appears to have been long lived. Mineral reactions suggest isobaric cooling until around 600°C. Subsequent decompression is evident because kyanite is not observed as a retrograde mineral. Whole-rock ⁴⁰Ar/³⁹Ar plateau ages of 790-845 Ma suggest that cooling and denudation of the Namaquan crust resulted from some 4-8 km of uplift per 100 my (Clifford et al., 1995). This process may, however, have been interrupted or complicated by early Pan-African orogenic processes to the northwest of the OCD (i.e. the Gariep Belt; Fig. 1), which produced zircon overgrowths at 850 Ma in rocks such as the H₂O-saturated Spektakel Suite (Raith, 1995). Later uplift brought the present exposures of the OCD to the surface by ~600 Ma, when the sediments of the Nama Group were unconformably deposited on Kibaran-Namaquan crust.

Causes of granulite-facies metamorphism in the mid-crust

The evolution of granulite terranes along an anticlockwise, isobaric cooling path at relatively shallow depths is relatively uncommon (Harley, 1989). Isobaric cooling from high temperatures in the mid-crust implies that a significant thermal anomaly is present in continental material of normal, 25-40 km thickness. Because the Kibaran and Namaquan orogenies were compressive, and major magmatic pulses coincide with metamorphic zircon growth in the area, the granulite-facies terrane of the western NMC is probably related to magmatic accretion into and under pre-existing continental crust that was accompanied by moderate crustal thickening (Waters, 1989, , 1990). Other possible mechanisms, such as extension of normal thickness crust with or without magmatic underplating, or extension of previously overthickened crust (Harley, 1989), are inconsistent with the regional geological constraints.

Anticlockwise P-T-t paths in isobarically cooled granulites are recorded in rocks within or below the zone of magmatic accretion, but not above it. This is consistent with observations in the NMC, where the prograde path is preserved in supracrustal assemblages of suitable compositions that occur at the same crustal level as, or below, the zone occupied by the voluminous Spektakel Suite intrusions. Although the Spektakel and Koperberg suites are too small to produce the advected heat that would be required to metamorphose the NMC, they nevertheless do reflect a major increase in heat flux from depth. As suggested by Waters, (1990) and Gibson et al., (1996), the heat was supplied either by under- or intraplating of basic magma, or by delamination of the lower lithosphere. Seismic refraction studies of the NMC have shown that at depths greater than ~14 km, the region is underlain by a zone whose seismic velocities are consistent with amphibolitic or dioritic compositions. The presence of a mafic underplate is also supported by the petrogenetic (assimilation-fractional crystallization; AFC) characteristics of the Koperberg Suite and the evidence that the Spektakel Suite is a lower-crustalanatect.

What was the cause of the thermal pulse 100-150 my after the Kibaran convergence, magmatism and low-pressure amphibolite-facies metamorphism? Although plume activity is an attractive explanation, it does not account for synchronous magmatic activity elsewhere in the region. Delamination at the base of a thickened lithosphere is another possible solution, but the time gap is too great for the Namaquan event to be a direct consequence of Kibaran convergence. The plate tectonic setting of the region after the Kibaran collision is obscure, although assemblies of a Rodinian supercontinent have been attempted (Dalziel, 1991; Hoffman, 1991). Two other mechanisms for initiating a thermal event that is accompanied by moderate compression are: (1) renewal of subduction along the new outer edge of the continent, which lay presumably towards the south; or (2) stagnation following Kibaran collision and a change in the pattern of subduction leading to a build-up of heat in the subcontinental mantle, and possibly also delamination of existing mantle lithosphere.

Whatever its cause, this thermal event lasted only for 30 my and was followed, after a further interval, by break-up of the supercontinent along north-south trending zones. A thermal pulse associated with an 850 Ma rifting event may be recorded in the zircon overgrowths analysed from the Spektakel Suite. Alternatively, uplift and exhumation of crustal blocks may also have produced the ⁴⁰Ar/³⁹Ar plateau ages of 790-845 Ma found in the western part of the OCD (Clifford et al., 1995).

In a schematic north-south section constructed for the OCD in Namaquan times (Fig. 11), metamorphic zonation is related to tectonic juxtaposition of the relatively unmetamorphosed Richtersveld terrane with the OCD along the south-verging Groothoek shear, and the later development of a granulite-facies overprint in the thermally perturbed zone associated with Namaquan magmatic accretion. The lack of granulite-facies metamorphism during the Kibaran (1212-1175 Ma) orogeny may be related to the absence of a mafic underplate at that time. The thermal anomaly that produced the Little Namaqualand Suite was evidently insufficient for regional granulite-grade metamorphism and zircon growth. The 1060-1030 Ma Namaquan crust was probably not thickened by orogenic processes at this time because this would have resulted in noticeable decompression after the thermal climax and during isostatic readjustment. Instead, the crust may have thinned somewhat between the Kibaran and Namaquan orogenies (i.e. between D₂ and D₃) in response to thermal relaxation after the Little Namaqualand Suite was emplaced. Namaquan accretion could merely have reinstated the crust to its normal proportions.

Figure 11. Schematic model illustrating the mechanisms responsible for the development of mid-crustal granulite-facies rocks in the Okiep Copper District. A, amphibolite grade; LG, lower granulite grade; UG, upper granulite grade; KS, Koperberg Suite; SS, Spektakel Suite.

ACKNOWLEDGEMENTS

Simon Harley, Lewis Ashwal and Sorena Sorenson provided detailed editorial and scientific insights, which greatly helped to improve the manuscript. Sally Stowe and David Vowles of the ANU Electron Microscopy Unit are thanked for their assistance with the cathodoluminescence imagery. Gold Fields of SA Ltd and the Foundation for Research Development (FRD) in South Africa are acknowledged for their logistic and financial support of this project.