Journal of Petrology Advance Access originally published online on May 29, 2007
Journal of Petrology 2007 48(7):1411-1441; doi:10.1093/petrology/egm025
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Geochemistry, Geochronology and Isotopic Evolution of the Chewore–Rufunsa Terrane, Southern Irumide Belt: a Mesoproterozoic Continental Margin Arc
1Institute for Research on Earth Evolution, Japan Agency for Marine–Earth Science and Technology, 2-15 Natsushima-Cho, Yokosuka, Kanagawa-Ken, 237-0061, Japan
2Tectonics Special Research Centre, School of Earth and Geographical Sciences, University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia
3School of Mines, Geology Department, University of Zambia, Po Box 32379, Lusaka, Zambia
4Laboratory for Planetary Sciences, Tokyo Institute of Technology, 2-1-12 O-Okayama, Meguro, Tokyo, 152-8551, Japan
5National Institute of Polar Research, 9-10 Kaga 1-Chome, Itabashi-Ku, Tokyo 173-8515, Japan
RECEIVED AUGUST 28, 2006; ACCEPTED APRIL 5, 2007
 |
ABSTRACT |
|---|
The southern Irumide Belt (SIB) is an ENE–WSW-trending, late Mesoproterozoic orogenic belt located between the Congo–Tanzania–Bangweulu (CTB) and Kalahari cratons in central southern Africa. It is separated from the late Mesoproterozoic Irumide Belt (IB) to the north by Permo-Triassic graben, raising the possibility that the younger rifts reactivated a suture between the two belts that has been rendered cryptic as a result of younger Karoo cover. Both belts are dominated by calc-alkaline gneisses, but in addition the SIB contains abundant metavolcanic and metasedimentary rocks. In this study we present detailed geochemical, isotopic and geochronological data for volcanic and plutonic lithologies from the southernmost part of the SIB, the Chewore–Rufunsa Terrane. This terrane comprises a wide variety of supracrustal to mid-crustal rocks that have major- and trace-element compositions similar to magmas formed in present-day subduction zones. Chondrite-normalized rare earth element (REE) profiles and whole-rock Sm–Nd isotope compositions indicate that the parental supra-subduction melts interacted with, and were contaminated by sialic continental crust, implying a continental-margin-arc setting. Secondary ionization mass spectrometry dating of magmatic zircon has yielded crystallization ages between c. 1095 and 1040 Ma, similar to elsewhere in the SIB. U–Pb dating and in situ Lu–Hf isotopic analyses of abundant xenocrystic zircon extracted from the late Mesoproterozoic granitoids indicate that the contaminant continental basement was principally Palaeoproterozoic in age and had a juvenile isotopic signature at the time of its formation. These data are in contrast to those for the IB, which is characterized by younger, c. 1020 Ma, calc-alkaline gneisses that formed by the direct recycling of Archaean crust without significant addition of any juvenile material. We suggest that the SIB developed by the subduction of oceanic crust under the margin of an unnamed continental mass until ocean closure at c. 1040 Ma. Subsequent collision between the SIB and the CTB margin led to the cessation of magmatism in the SIB and the initiation of compression and crustal melting in the IB.
KEY WORDS: geochemistry; Mesoproterozoic; SHRIMP zircon U–Pb dating; Sm–Nd isotopes; Southern Irumide Belt
|
INTRODUCTION |
|---|
Central southern Africa is a complex region of Archaean cratons and enveloping orogenic belts that range in age from Palaeoproterozoic to Cambrian (Fig. 1a). Until recently, it was presumed that these cratons were mostly assembled in the mid- to late Mesoproterozoic during a sub-Saharan-wide ‘Kibaran Orogeny’ (Hanson, 2003
). However, analysis of the geochronological and sparse geochemical data from these various Mesoproterozoic belts has demonstrated that magmatism and compressional tectonics occurred at distinctly different times in different belts, and that most of the belts presumably owe their character to distinct and unrelated geological processes (De Waele et al., 2003; Johnson et al., 2005). The majority of the central African Mesoproterozoic belts are dominated by felsic calc-alkaline gneisses, but it has yet to be determined whether the protoliths of these gneisses formed by juvenile or crustal recycling processes. For instance, the calc-alkaline rocks may have formed in a continental-margin arc, or by accretion of juvenile or mature oceanic arcs to the continent margin, or by the complete tectonic recycling of older calc-alkaline basement gneisses in a hot, wide, Himalayan-style collisional orogen. In this study we present a detailed integrated geochemical, isotopic and geochronological investigation of part of one Mesoproterozoic Belt, the Southern Irumide Belt, which is situated between the Greater Congo Craton (the combined Congo–Tanzania–Bangweulu Craton or CTB) and the Kalahari Craton (Fig. 1a). We provide evidence to indicate that this belt formed as part of a continental-margin arc on the margin of an unnamed continent that subsequently collided with the CTB craton during the late Mesoproterozoic.
|
|
|
|
|
|
REGIONAL GEOLOGY |
|---|
The Southern Irumide Belt (SIB) (Fig. 1b) is a lithologically varied and locally multiply deformed belt of metasedimentary, metavolcanic and metaplutonic rocks that are distinct from the monotonous granitoid gneisses that form much of the adjacent Irumide Belt (sensu stricto) (De Waele, 2005
; Johnson et al., 2005, 2006; De Waele et al., 2006b). The SIB is exposed in southern Zambia, northern Zimbabwe, northern Malawi and western Mozambique (Fig. 1a–c) but the contact with the Irumide Belt is obscured by a Permo-Triassic ‘Karoo’ graben, raising the possibility that these younger rifts reactivate and mask a cryptic suture between the two (Johnson et al., 2006). Based on lithostratigraphy, structure and metamorphic characteristics, Mapani et al. (2001, 2004) provisionally subdivided the SIB in Zambia into several lithologically distinct terranes bounded by ductile shear zones. This architecture was used as the basis for an extensive U–Pb sensitive high-resolution ion microprobe (SHRIMP) zircon geochronological study of the region (Johnson et al., 2006) that resulted in some modifications, as shown in Fig. 1b. The Chewore–Rufunsa Terrane (Fig. 1b), the focus of the present study, comprises late Mesoproterozoic calc-alkaline gneisses and metavolcanic rocks. Major- and trace-element and rare earth element (REE) geochemical analyses from the Chewore Inliers of northern Zimbabwe indicate that the lithologies formed via supra-subduction-zone magmatic processes (Oliver et al., 1998; Johnson & Oliver, 2000, 2004); however, there are no published isotopic data to support this interpretation. The structurally overlying Luangwa–Nyimba Terrane (Fig. 1b) to the east is dominated by calcareous, psammitic and pelitic metasedimentary rocks, and the Petauke–Sinda Terrane to the east of that (Fig. 1b) consists of a suite of late Mesoproterozoic calc-alkaline plutonic rocks and late Cambrian post-tectonic granites and syenites (Johnson et al., 2006). The easternmost terrane in Zambia, the Chipata Terrane, is composed of variably deformed high-temperature mafic and felsic granulite including meta-igneous lithologies (charnockite and charno-enderbite) and various sillimanite-bearing metasedimentary lithologies of Palaeoproterozoic and Mesoproterozoic age (Johnson et al., 2006). In the poorly studied regions of northern Malawi and western Mozambique, the nature of the SIB is obscure and the locations of the terrane boundaries are unknown; however, various Rb–Sr and Sm–Nd whole-rock–mineral isochrons, zircon Pb-evaporation ages and unpublished U–Pb SHRIMP zircon ages indicate the presence of similar Mesoproterozoic protoliths (Kröner et al., 1997; Evans et al., 1999; Mäkitie et al., 2006; Mänttäri et al., 2006). Metamorphism throughout the SIB in Zambia is generally high-temperature (>850°C) and low-pressure (<4·5 kbar) in nature and has been dated between 1·07 and 1·04 Ga (Goscombe et al., 1998, 2000; Schenk & Appel, 2001, 2002; Cox et al., 2002; Johnson & Oliver, 2004; Johnson et al., 2006); however, parts of the SIB were reworked under high- to very high-pressure conditions (
20 kbar) during the Neoproterozoic–Cambrian Pan-African Zambezi Orogeny (c. 550–515 Ma) (John et al., 2004; Johnson et al., 2004; Johnson et al., 2005). This multiple reworking has completely obliterated any pre Pan-African structures and metamorphic assemblages, making the interpretation of Mesoproterozoic events especially cryptic.
In contrast, the Irumide Belt (sensu stricto; IB) comprises a basement of granitoid gneiss emplaced between 2·05 and 1·93 Ga overlain by a supracrustal sequence, the Muva Supergroup, at c. 1·87–1·86 Ga (De Waele & Fitzsimons, 2004). Both basement gneiss and Muva Supergroup were locally intruded by small granitoid bodies at 1·65–1·55 Ga and later by voluminous K-feldspar megacrystic granitoid batholiths between 1·05 and 0·95 Ga (De Waele, 2005; De Waele et al., 2006b). These late Mesoproterozoic calc-alkaline granitoids were derived by recycling of basement gneisses without addition of any juvenile material (De Waele et al., 2006a). Medium-pressure, high-temperature metamorphism (8 kbar at 850°C) and contractional deformation accompanied the main magmatic event at c. 1·02 Ga (Daly, 1986; De Waele, 2005) producing strong NE–SW structural trends.
|
GEOLOGY OF THE CHEWORE–RUFUNSA TERRANE |
|---|
 TOPABSTRACTINTRODUCTIONREGIONAL GEOLOGY GEOLOGY OF THE CHEWORE-RUFUNSA... ANALYTICAL TECHNIQUESMAJOR, TRACE AND REE...ZIRCON U-PB SHRIMP GEOCHRONOLOGYWHOLE-ROCK AND ZIRCON ISOTOPESDISCUSSIONREFERENCES |
|---|
The Chewore–Rufunsa (C–R) Terrane is exposed along the deeply incised Zambian Zambezi Valley fault escarpment and within isolated and remote basement horsts (known as the Chewore Inliers) that sit within the Mid Zambezi Valley of northern Zimbabwe (Fig. 1c). Mapping by the Zambian and Zimbabwean (formally Rhodesian) Geological Surveys (Goscombe et al., 1994
, 1998, 2000; Barr, 1998) has revealed the presence of a wide variety of mafic to felsic gneisses, K-feldspar augen gneisses, meta-pelites and abundant quartzites (Fig. 1c). The metamorphic grade increases dramatically from greenschist facies in the northern part of the terrane around Rufunsa (Barr, 1998) to high- to moderate-pressure amphibolite facies and local granulite facies in the south (Goscombe et al., 1998, 2000; Johnson & Oliver, 2000, 2004). A striking feature of the terrane is the presence of four mafic to felsic complexes, composed of plutonic and volcano-sedimentary rocks, here named the Chakwenga, Chongwe, Chewore (the Ophiolite Terrane of Oliver et al. 1998) and Ikondo complexes (Fig. 1c).
The Chewore Complex
Detailed investigations of the Chewore Inliers (Goscombe et al., 1998, 2000; Oliver et al., 1998; Johnson & Oliver, 2000, 2004) have provided a wealth of geochemical and geochronological data. The Chewore Inliers have been divided into four fault-bounded lithotectonic slices [termed terranes by Goscombe et al. (1994) and Oliver et al. (1998)], the Granulite, Quartzite, Zambezi and Ophiolite (Chewore Complex of this study) terranes. The Chewore Complex, one of the four regional mafic to felsic complexes in the C–R Terrane, comprises a variable suite of rocks ranging from ultramafic to felsic in composition, which exhibit the geochemical signatures of a marginal basin ophiolite [the Chewore Ophiolite of Oliver et al. (1998) and Johnson & Oliver (2000)] and a low-K tholeiitic island arc [the Kaourera Arc of Johnson & Oliver (2004)]. Low-strain zones in both the ophiolite and arc suites reveal primary volcanic features such as vesicular textures and pillows, indicating that parts of this complex are of supracrustal origin (Johnson & Oliver, 2000, 2004). SHRIMP dating of igneous zircon from a plagiogranite dyke within the Chewore Ophiolite and a metadacite in the Kaourera Arc indicate ocean crust formation at c. 1393 Ma (Oliver et al., 1998) and island arc formation at c. 1082 Ma (Johnson & Oliver, 2004), respectively. Felsic to intermediate orthogneisses that make up the Zambezi Terrane and intrude metasediments of the Granulite Terrane also display arc-like geochemical affinities (Johnson & Oliver, 2004) and crystallization ages similar to the arc rocks at 1071 Ma and 1083 Ma (Goscombe et al., 2000), leading to the interpretation that the different Chewore Inlier terranes represent different levels within a single arc complex (Johnson & Oliver, 2004).
The Chakwenga Complex
Lithologies of the Chakwenga Complex are best exposed along the Chakwenga River and its tributaries within the Lower Zambezi National Park (Figs 1c and 2). All lithologies have been metamorphosed to at least lower- to mid-amphibolite facies during the Neoproterozoic to Cambrian Zambezi orogeny and carry a strong south-dipping planar and SE-plunging linear fabric (Fig. 2), defined by the alignment of biotite and hornblende and/or quartz–feldspar aggregates. The complex can be divided into two main units, fine-grained mafic to intermediate mylonitic gneiss and medium- to coarse-grained hornblende gneiss and metagabbro. The northern part of the complex (north of UTM 8290; Fig. 2) is dominated by fine-grained amphibolitic mylonitic gneiss composed of quartz, plagioclase (± K-feldspar), biotite and hornblende (Figs 2 and 3a) with minor hornblende gneiss in the extreme NE of the mapping area (Fig. 2). Sparse felsic rocks contain fine-grained quartz and feldspar with minor biotite and in low-strain zones are commonly observed to contain relict tuffaceous textures (e.g. GR 0780867–8292595–sample 38). Garnet-bearing muscovite schist and quartzite layers containing intense internal isoclinal to tight shear folds are common.
To the south of UTM 8290 (Fig. 2), medium- to coarse-grained hornblende gneiss and metagabbro dominate (Figs 2 and 3b). The hornblende gneisses are composed of flattened and aligned feldspar, hornblende and biotite with little to no free quartz, and the metagabbros are composed of elongate, aligned hornblende and feldspar. In low-strain zones some of the metagabbros display relict sub-cumulate textures. At GR 0779614–8288973, a mylonitized fine- to medium-grained hornblende gneiss (sample 45a) contains thin (usually less than 10 cm wide) boudinaged lenses and layers (0·5–5 m in length) of coarse-grained, mildly strained metagabbro (sample 45b; Fig. 3b). These lenses and layers make up 30% of the outcrop and have sharp, finer-grained (but not strained) margins with the surrounding gneiss, suggesting that they may represent relict dykes. The presence of thrust faults is interpreted on the basis of decimetre-wide ductile shear zones that are associated with distinct changes in lithology.
The Chongwe Complex
The Chongwe Complex is best exposed along the Chongwe and Chowe (also locally known as Chiawa) rivers that deeply dissect the Zambezi fault escarpment in Zambia (Figs 1 and 4). Lithologies in both river sections have attained, at least, upper amphibolite-facies conditions during the late Neoproterozoic to Cambrian (John et al., 2004) and garnet is an abundant phase in rocks of variable compositions.
The Chongwe River section (Figs 4 and 5a) is dominated by garnet-bearing and occasionally biotite-rich amphibolites that are interbanded with metre- to decimetre-thick leucocratic, garnet–feldspar–quartz–biotite gneiss (Fig. 3c). Amphibolite predominates over the felsic gneiss throughout the section but in localized zones up to 100 m in width, the felsic gneiss can form up to 75% of the rock volume. All lithologies carry a strong planar tectonic fabric and the contacts between the units are sharp with no apparent gradational boundaries.
The Chowe River section (Fig. 5b) exposes similar mylonitic lithologies to those in the Chongwe River (Fig. 3h) but additionally contains strongly metasomatized rocks that have been metamorphosed to whiteschist (high-pressure talc–kyanite assemblages) during the Neoproterozoic Zambezi Orogeny (John et al., 2004). All lithologies in this river section have been intensely folded into millimeter and decimeter-scale sheath-folds (Fig. 3d). In the lower-strain hinge zones of these folds, many primary igneous textures and features have been preserved including discrete tuffaceous layers and lenses (e.g. GR 0755541–8268374–sample C7; Fig. 3e). Importantly, at GR 0755536–8268404 a strongly foliated quartz–feldspar–biotite–garnet leucogneiss (sample C9a) hosts 1–20 cm thick, foliation-parallel garnet-amphibolite layers (sample C9b). In the lower-strain regions the amphibolite is observed to cross-cut the dominant felsic gneiss foliation and crude layering (Fig. 3f), demonstrating that some of the amphibolites were once dykes. Rarely, these dykes retain texturally preserved chilled margins. The Chowe River section also contains up to 100 m thick lenses of mylonitized K-feldspar augen gneiss (Fig. 3g). At GR 0755513–8268419 (sample C10) tabular K-feldspar augen 10 cm by 5 cm dominate a matrix composed of quartz and feldspar with minor biotite and muscovite. These augen are intensely deformed and in places the gneiss contains centimeter-scale layers and bands dominated by very fine-grained ultramylonitized K-feldspar augen. The margins of the augen gneiss bodies are intensely mylonitized, indicating that these units represent imbricate thrust lenses.
Because the whiteschist lithologies have undergone such intense metasomatic alteration (Johnson & Oliver 2002; John et al., 2004) it is almost impossible to determine their parental, pre-metasomatic geochemical composition and so they are not considered further here.
|
ANALYTICAL TECHNIQUES |
|---|
Whole-rock major- and trace-element analyses
All samples were prepared and analysed by conventional X-ray fluorescence spectrometry (XRF) using the procedure and instrumentation described in Tani et al. (2005
). Fresh samples were cut into slabs with a diamond-blade saw and the altered surfaces removed. The slab faces were polished and rinsed in an ultrasonic bath to remove any iron introduced by the diamond blade saw, before being coarsely crushed and washed in distilled water and acetone. The dried samples were crushed in a mortar and then powdered using an oscillatory alumina mill. Major- and trace-elements were analysed on glass beads and pressed powder pellets, respectively, using a RIGAKU SIMULTIX12 for major elements and a RIGAKU RIX3000 for trace elements. Instrumental procedures, standards, analytical conditions and detection limits have been outlined in detail by Tani et al. (2005). The major, trace and REE data were plotted using GDCkit v2.2.1 (Janousek et al., 2006). Numerical data are presented in Table 1.
Sensitive high-resolution ion-microprobe U–Pb zircon dating
Fresh rocks samples were crushed and heavy mineral separates obtained through conventional panning and high-density liquid separation in an ultra-clean environment. Zircon grains were picked under a binocular microscope and mounted alongside the CZ3 and/or FC1 standard zircon in an epoxy cast, which was polished to expose the grains mid-section. The mount was first imaged using an optical microscope, then coated with a thin layer of carbon and imaged on a JEOL 6400 electron microscope fitted with a cathodoluminescence (CL) detector. Operating conditions for CL imaging were 15 keV accelerating voltage, 5 nA current and a working distance of 16–25 mm. The mount was then repolished and thoroughly cleaned to minimize contamination (Pb), and then coated with a thin layer of Au to provide conductivity during the SHRIMP II analyses. The mount was loaded into the sample lock 24 h prior to analysis and pumped to high vacuum to allow outgassing, thereby minimizing hydride interference during the analysis. The analyses were conducted in two separate sessions, the first (Session A in Table 2) at the John de Laeter Centre for Mass Spectrometry (Curtin University of Technology) and the second (Session B in Table 2) at the National Institute for Polar Research (NIPR) in Tokyo. Operating procedures for the SHRIMP followed that described by Nelson (1996). In session A (Curtin University SHRIMP II), the primary beam intensity was 2 nA with a slightly elliptical spot size of 30 µm, whereas at NIPR (SHRIMP II) the primary beam intensity was 4·4 nA with an elliptical spot size of 30 µm. Analyses of unknown zircon were interspersed with analyses on the standard zircon (CZ3 at Curtin University or FC1 at NIPR) at a ratio of 3:1 to allow calibration of the 206Pb/238U ratio. In the case of standard FC1 (which has variable U concentrations), two analyses of zircon standard SL13 (with a uniform U concentration of 238 ppm) were conducted to calculate the U concentration for the unknown analyses. Corrected ratios were calculated using SQUID software (Ludwig, 2001b), and calculation of pooled ages and plotting were done using ISOPLOT (Ludwig, 2001a). All data in Table 2 are reported at the 1
confidence level. Age data for single zircons are reported in the text at the 1 confidence level whereas the pooled ages (concordia ages, weighted mean 207Pb/206Pb or 206Pb/238U and lower or upper intercept ages) are reported at the 95% confidence level.
Whole-rock Sm–Nd isotopic analyses
Sr and Nd isotope ratios were determined by thermal ionization mass spectrometry (TIMS) at the Department of Geology and Geophysics at the University of Adelaide on a Finnigan MAT 262 system in static mode. All ground samples were leached in 3N HCl for 30 min at c. 100°C. The supernatant liquid was pipetted off, the sample washed in deionized water and the water pipetted off. The residue was then analysed for its isotopic composition. The long-term average for the in-house Nd standard (J&M specpure Nd2O3) is 0·511603 ± 9 (1 of total population, n = 105). The LaJolla standard gave 0·511828 ± 11 (n = 9) and BCR-1 was 0·512593 ± 16 (n = 12). Typical blanks are in the order of 100–200 pg for Nd. The average for the NBS987 Sr standard is 0·710258 ± 18 (n = 56). Typical Sr blanks are better than 1·5 ng, which is negligible compared with a typical sample size of 10–100 mg of Sr. Initial ratios and model ages were calculated using the present-day Chondritic Uniform Reservoir (CHUR) of Goldstein et al. (1984) (0·512638).
In situ zircon Lu–Hf isotopic analyses
In situ analyses of the Lu–Hf ratio of the inherited zircon were performed on a multi-collector (MC) laser ablation inductively coupled plasma mass spectrometry (LA-MC-ICPMS) system at the Department of Earth and Planetary Sciences, Tokyo Institute of Technology, using the procedures documented by Iizuka & Hirata (2005). Analyses were carried out with a beam diameter of 62 µm, 3–10 Hz repetition rates and 15 s ablation times. All analyses were collected in a single session. Mass discrimination effects were corrected by normalizing to 179Hf/177Hf = 0·7325 (Patchette et al., 1981) for Hf and Lu, and to 173Yb/171Yb = 1·12346 (Thirlwall and Anczkiewicz, 2004) for Yb using an exponential law. For the calculation of initial Hf isotope ratio, the decay constant for 176Lu proposed by Scherer et al. (2001) (1·865 x 10–11 year–1) was used.
|
MAJOR, TRACE AND REE GEOCHEMISTRY |
|---|
Bearing in mind that the rocks in this study have been metamorphosed to at least the mid-amphibolite facies, it is inappropriate to use fluid-mobile elements such as the alkali oxides CaO, Na2O and K2O and trace elements such as Rb, Sr, Ba for discussing tectonic environments or fractionation–assimilation pathways (Humphris & Thompson, 1978
; Brekke et al., 1988; Brouxel & Lapierre, 1988) and so we focus on the relatively immobile elements such as Zr, Y, Th and Nb, which are frequently used to distinguish the tectonomagmatic environment (e.g. Pearce & Norry, 1979; Meschede, 1986) of basaltic lithologies from low- to high-grade metamorphic terranes. The data for both the Chakwenga (open squares and stars) and Chongwe complexes (closed circles) (Table 1) are shown in a series of Harker variation (Fig. 6) and normalized trace-element and REE diagrams (Fig. 7), to discriminate the tectonic setting of these rocks.
|
|
|
Major- and trace-element data
The rocks in the Chakwenga and Chongwe complexes comprise a compositionally variable suite ranging from sub-alkaline basalt through to rhyodacite–rhyolite (Fig. 6a; Table 1). In general, the metabasalts of all groups have low Zr (<100 ppm), Y (<30 ppm), Th (<10 ppm) and Nb (<20 ppm) concentrations and high FeOt (>10 wt %), TiO2 (>1 wt %) and MgO (>5 wt %) contents (Table 1; Fig. 6b–h). With increasing silica content, FeOt, TiO2 and MgO progressively decrease and Zr, Y, Th and Nb progressively increase (Table 1; Fig. 6b–h). However, the Chakwenga hornblende gneisses and metagabbros have significantly higher Zr, Th and Nb contents at all SiO2 concentrations, and have less well-constrained relationships than the mafic–felsic volcanic suites of both the Chakwenga and Chongwe complexes. The compositions and trends of the Chakwenga and Chongwe metavolcanic suites are similar to those of the Chewore Complex (Johnson & Oliver, 2000
, 2004), where they were interpreted to represent a low-K tholeiitic suite that had undergone simple, closed-system fractionation.
In the normal mid-ocean ridge basalt (N-MORB)-normalized trace-element diagrams (Fig. 7a–c; normalization after Pearce, 1983), the metamafic lithologies from both the Chakwenga and Chongwe complexes display relatively flat profiles from Nb to Yb with element concentrations between 0·5 and 5 times that of N-MORB, and have relatively elevated (1·5–70 times) Th concentrations. All of the Chakwenga hornblende gneisses and meta-gabbros have much higher trace-element concentrations and plot with a positive sloping trace-element pattern. Although there appears to be a clear distinction between the Chakwenga hornblende gneisses or metagabbros and the Chakwenga metabasalts, the latter overlap with metabasalts from the Chongwe Complex. None of the samples display relative negative Nb anomalies, a feature usually considered characteristic of supra-subduction-zone magmas. Two Chakwenga hornblende gneiss samples from the north of the mapping area (37a and 37b; closed stars; Fig. 2) have much steeper trace-element profiles that crosscut the patterns for the other Chakwenga meta-basaltic samples. Four Chewore Complex metabasalts are shown in Fig. 7c for comparison (see also Table 1). All these rocks have a similar, flat trace-element profiles to the bulk of the Chakwenga–Chongwe metabasalts but have relative, negative Nb anomalies.
In Fig. 8, the compositions of the meta-basalts are plotted on two of the most popular basalt tectonic discrimination diagrams, the 2Nb–Zr/4–Y triangular plot of Pearce & Norry (1979) and the Zr/Y–Zr binary plot of Meschede (1986). The data exhibit significant scatter, reflecting the variable Y, Zr and Nb concentrations as illustrated in the Harker variation diagrams (Fig. 6e, g and h). Data from the Chewore Complex plot within the N-MORB–volcanic arc basalt field in both diagrams, attesting to their relatively low and uniform Nb and Zr contents. The metabasalts of the Chakwenga and Chongwe complexes scatter within the N-MORB, enriched MORB (E-MORB), volcanic arc basalt and within-plate tholeiite fields (Fig. 8). Samples 37a and 37b from the Chakwenga Complex plot consistently within the within-plate basalt field in both diagrams. These two metabasalts have similar Zr, Nb and Y concentrations to the other metabasalts (Table 1 and Fig. 6) but their element ratios, especially Zr/Y (Fig. 6i), are appreciably different, suggesting that they may have a different tectonic setting from the other Chakwenga and Chongwe metabasalts.
|
REE data
REE data are tabulated in Table 1 and shown as chondrite-normalized REE plots in Fig. 7d–f (after Nakamura, 1974
). The meta-basaltic lithologies from the Chakwenga and Chongwe complexes have gently sloping, parallel, light REE (LREE)-enriched patterns that become progressively more LREE-enriched with increasing REE content compared with chondrite (Fig. 7d and e). Again, the Chongwe metabasalts show a large compositional range and overlap with the Chakwenga hornblende gneisses and meta-gabbros. The less-evolved metabasalts (i.e. those with the lowest REE concentrations of 10–20 times chondrite) have average La/SmN and La/YbN ratios of 1·45 and 1·94 whereas the most enriched samples have REE concentrations 50–110 times that of chondrite with La/SmN and La/YbN ratios of 1·4 and 3·0. These enriched patterns are similar to those of the Chongwe felsic gneisses (stippled field in Fig. 7d; Table 1). Sample 37b has a much steeper REE profile that crosscuts the other Chakwenga metamafic rocks and has La/SmN and La/YbN ratios of 2·63 and 8·34, respectively. The three representative samples from the Chewore Complex have lower REE concentrations than the Chongwe and Chakwenga samples and flat REE patterns with La/SmN and La/YbN ratios of 1· 0 and 0·5 (Fig. 7f; Table 1). |
ZIRCON U–PB SHRIMP GEOCHRONOLOGY |
|---|
|
TOPABSTRACTINTRODUCTIONREGIONAL GEOLOGYGEOLOGY OF THE CHEWORE-RUFUNSA...ANALYTICAL TECHNIQUESMAJOR, TRACE AND REE...ZIRCON U-PB SHRIMP GEOCHRONOLOGYWHOLE-ROCK AND ZIRCON ISOTOPESDISCUSSIONREFERENCES |
|---|
Data for six Chongwe and one Chakwenga Complex sample are presented.
The Chongwe Complex
Sample C6 (banded mafic gneiss)
The abundant zircons extracted from this sample range in size from 100 to 300 µm, have length to width ratios from 2:1 to 3:1 and, in general, are euhedral with typical bipyramidal terminations. Cathodoluminescence (CL) imaging reveals that many grains contain oscillatory-zoned cores surrounded by unzoned to sector zoned rims up to 70 µm in thickness (Fig. 9a). U–Pb analyses were made on five core regions and six rims. Proportions of common 206Pb in total 206Pb (f 206) range from 0·00 to 0·19% for all analyses, except for the core region of zircon c6-1c, which has an elevated f 206 of 2·61% (Table 2). U and Th contents for the core analyses were in the range of 286–572 ppm and 149–335 ppm, respectively, leading to Th/U ratios of 0·5–0·6, which are typical for magmatic zircon (Rubatto & Gebauer, 2000). The U contents of the rims were much higher (719–1051 ppm) and Th contents much lower (61–93 ppm) than those of the magmatic cores, leading to Th/U ratios (0·06–0·12) consistent with their growth during a metamorphic event (Rubatto & Gebauer, 2000). Except for core analysis c6-1c, which has high f206 and exhibits significant reverse discordance (110%), the remaining core analyses are within 3% of concordia (97–102% concordant) and yield a concordia age of c. 1050 Ma. The four remaining core analyses regress to a mean 207Pb/206Pb age of 1051 ± 12 Ma [mean square weighted deviation (MSWD = 0·43)] (Fig. 10a), which we interpret to be the age of igneous crystallization of the gneiss. The six rim analyses plot on concordia and yield a concordia age of 573 ± 2 Ma (MSWD = 0·106), which we interpret to be the age of amphibolite-facies metamorphism and which is consistent with other Pan-African metamorphic ages in the region (Johnson et al., 2005).
|
|
|
Sample C9a (leucocratic garnet-bearing gneiss–metadacite)
Zircons from this sample are sub-spherical to elongate in shape with the longest axis being between 50 and 150 µm in length. Imaging under CL conditions reveals the presence of rounded oscillatory-zoned to sector-zoned cores, which are surrounded by CL bright rims up to 50–100 µm wide (Fig. 9b). We have analysed six zircons, including two core–rim pairs (c9-1 and c9-2), one rim (c9-3r) and three single-sector zircons (c9-4, c9-9 and c9-10). The proportion of f206 for all analyses is very low, between 0·00 and 0·21% (Table 2). U and Th contents vary in the range of 95–483 ppm and 50–93 ppm, respectively, for the cores and single-sector zircon, resulting in variable Th/U between 0·06 and 0·58. The rims display the lowest U and Th values (24–75 and 3–29 ppm, respectively), and have widely variable Th/U ratios (0·06–1·24; Table 2). The data range from strongly discordant to reversely discordant, with the three rim analyses displaying the highest discordance. The data define a poorly constrained discordia that intercepts the concordia line at 1046 ± 24 Ma and 166 ± 110 Ma (MSWD = 0·5) (Fig. 10b). Given the variable discordance and poor clustering of the data, the upper intercept age can only be taken as a very preliminary estimate of the crystallization age of zircon in the volcanic precursor rock of sample C9.
Sample C10 (K-feldspar augen gneiss)
The majority of zircons extracted from this rock are sub-rounded to elongate, between 100 and 150 µm in length, with length to width ratios of 2:1. Rare grains exhibit bipyramidal terminations typical of an igneous origin. CL imaging reveals that most grains contain cores and rims, both displaying oscillatory zoning (Fig. 9c). We have analysed eight grains, seven rims, two cores and a single grain that does not contain any rim. The proportion of f206 is low (between 0·00 and 0·47%) and there is no obvious systematic difference in U and Th content between cores and rims, all analyses having 289–1819 ppm U, 5–517 ppm Th and Th/U ratios between 0·01 and 1·32 (Table 2). The single grain (c10-1), and two cores (c10-5c, c10-6c) have Palaeoproterozoic 207Pb/206Pb ages of 1917 ± 5 Ma (c10-1, 99% concordant), 1938 ± 8 Ma (c10-6c, 96% concordant) and 1809 ± 6 Ma (c10-5c, 98% concordant) (Fig. 10c) and are interpreted to be inherited grains. It is interesting to note that these three grains have the highest Th/U ratios of 1·32, 1· 07 and 0·83, respectively, whereas the remaining core and rim analyses all have low to very low Th/U ratios of 0·003–0·01 and late Mesoproterozoic ages. Zircons with Th/U ratios <0·1 are usually considered to be of metamorphic origin (Rubatto & Gebauer, 2000); however, the oscillatory zoning displayed by both cores and rims strongly suggests that they are magmatic. Analysis c10-5r, which has the largest f 206 value of 0·47%, is reversely discordant (112%) and is not open to easy interpretation. Except for analysis c10-7r, which has a slightly older 207Pb/206Pb age of 1105 ± 9 Ma, the remaining core and rim analyses form a single age population with a weighted mean 207Pb/206Pb age of 1092 ± 9 Ma (MSWD = 0·9) and a concordia age of 1094 ± 2 Ma (n = 5, MSWD = 0·055) (Fig. 10c). The low MSWDs for both the pooled 207Pb/206Pb and concordia age indicate the coherence of the data and lend confidence to the interpretation that the igneous protolith crystallized at c. 1094 Ma.
Sample C61a (leucocratic garnet-bearing gneiss)
All zircons extracted from this sample are elongate (200–300 µm in length), euhedral grains with typical igneous bipyramidal terminations (Fig. 9d). Most grains contain 10–50 µm sized inclusions typical of zircon from volcanic rocks (Thomas et al., 2003). CL images indicate that all the grains are oscillatory-zoned and only very thin (<10 µm), bright (in CL) rims have been detected around the margins of some grains. U–Pb analyses of these rims are beyond the spatial resolution of the SHRIMP. We have analysed 12 separate grains and all have very low f206 values (0·00–0·13) and are relatively uniform in U and Th concentrations (118–307 ppm U and 64–218 ppm Th) with Th/U ratios between 0·43 and 0·90 (Table 2). Apart from three variably discordant analyses (c61a-3, c61a-7 and c61a-11) the remainder define a single, coherent age population with a weighted mean 207Pb/206Pb age of 1087 ± 11 Ma (MSWD = 0·26) and with the six most concordant grains giving a concordia age of 1088 ± 4 Ma (MSWD = 0·064) (Fig. 10d) which we interpret as the age of crystallization of the igneous rock. The three variably discordant grains define a Pb-loss trend from this crystallization age (c. 1088 Ma) toward a lower intercept at c. 545 Ma; that is, similar in age to the metamorphic grains in the other samples (Fig. 10d).
Sample C70 (leucocratic garnet-bearing gneiss)
Similar to sample C61a, all extracted zircons were clear, euhedral grains, 200–300 µm in length, with typical igneous bipyramidal terminations. All grains show oscillatory zoning under CL conditions and no core–rim relationships were observed (Fig. 9e). We analysed five grains all of which have low f206 values (0·00–0·04%) and uniform U and Th concentrations (114–324 ppm U and 97–206 ppm Th) with Th/U ratios of 0·42–0·88 (Table 2). All five grains are part of a single population that gave a mean weighted 207Pb/206Pb age of 1067 ± 13 Ma (MSWD = 0·35) and a concordia age of 1070 ± 3 Ma (MSWD = 0·028) (Fig. 10e), which we interpret to be the age of crystallization of the igneous protolith.
Chakwenga Complex
Sample Chak48a (meta-dacitic gneiss)
Zircons extracted from this sample were elongate (75–200 µm in length) with length to width ratios of 3:1 to 4:1 but generally showed rounded and abraded terminations (Fig. 9f) and contain numerous, small (<10 µm) inclusions. Imaging under CL showed that all grains were oscillatory-zoned and no core–rim relationships have been observed. We have analysed 15 grains and most have low f206 values between 0·00 and 1·29 and relatively low concentrations of U and Th (68–388 ppm U and 54–334 ppm Th) with Th/U ratios between 0·52–2·20 (Table 2). Ten grains define a single coherent age population with a mean weighted 207Pb/206Pb age of 1077 ± 30 Ma (MSWD = 0·43) and a concordia age of 1083 ± 18 Ma (MSWD = 0·72) (Fig. 10f and g) that we interpret to be the age of crystallization of the igneous rock. Five other grains give Palaeoproterozoic ages and we interpret these grains to be inherited. Grain Chak48-4c has a very discordant (78% concordant) 207Pb/206Pb age of 1718 ± 34 Ma, Chak48-6 has a 207Pb/206Pb age of 2178 ± 42 Ma (106% concordant), Chak48-7 has a 207Pb/206Pb of 1889 ± 26 Ma (101% concordant), Chak48-8 has a 207Pb/206Pb age of 1957 ± 38 Ma (94% concordant) and Chak48-11 has a 207Pb/206Pb age of 1938 ± 62 Ma (99% concordant) (Fig. 10f and g).
|
WHOLE-ROCK AND ZIRCON ISOTOPES |
|---|
Twenty-one samples of variable SiO2 composition and age from the Chakwenga, Chongwe and Chewore complexes were selected for whole-rock Sm–Nd and Rb–Sr isotopic analysis; additionally, inherited zircons from samples of Chongwe and Chakwenga leucocratic gneiss were analysed for their Lu–Hf isotope composition. Seven of the samples have well-constrained crystallization ages, but the precise ages of the remaining 14, mainly meta-basaltic samples, are unknown. Considering the relatively restricted time period (c. 1090 – 1040 Ma) for magmatism within the Chewore–Rufunsa Terrane documented here and for the SIB in general (Johnson et al., 2006
), it is appropriate to provide only maximum and minimum initial isotopic ratios for these samples calculated using the oldest (c. 1090 Ma) and youngest (c. 1040 Ma) determined magmatic ages. Five inherited zircon grains, all of Palaeoproterozoic age, were selected for in situ Lu–Hf analysis, to gain information on the isotopic composition of the basement into which these magmas were intruded and to constrain contamination–assimilation pathways.
The Chakwenga Complex
The measured whole-rock 147Sm/144Nd ratios for all samples lie between 0·119 and 0·180 (Table 3). The dated felsic sample Chak48a (1083 Ma) has an initial 143Nd/144Nd ratio of 0·51072, 
Nd(t) values of –10·14 and a TDM age of 2.34. The remaining three meta-felsic lithologies, which encompass the mafic to felsic volcanic rocks, hornblende gneiss and gabbro, have initial 143Nd/144Nd(1090–1040) values between 0·51077 and 0·511215 with Nd(1090–1040) values ranging between –0·62(1090) to –9·12(1090) and –0·82(1040) to –9·63(1040) (Table 3; Fig. 11a). The same samples give TDM model ages between 1·73 and 2·50 Ga. The initial whole-rock 87Sr/86Sr ratios for all samples lie between 0·691 and 0·726. The lower values indicate that the Rb/Sr ratios may have been altered during amphibolite-facies metamorphism; therefore, little emphasis can be placed on them. Four inherited zircon grains from sample Chak48a (6, 7, 8, 11) with 207Pb/206Pb ages of c. 1889–2178 have initial 176Hf/177Hf(t) ratios between 0·28148 and 0·28169 with Hf(i) values ranging between +2·9 and –3·7 (Table 4).
|
|
The Chongwe Complex
Two meta-basaltic samples (C6 and C64), two meta-andesitic samples (C7 and C61b) and three felsic lithologies (C9a, C61a and C62b) were analysed. The measured whole-rock 147Sm/144Nd ratios for all samples have a similar range to those from the Chakwenga Complex, between 0·115 and 0·179 (Table 3). The dated meta-mafic lithology, sample C6 (1051 Ma), has an initial 143Nd/144Nd ratio of 0·511188, an
Nd(t) value of –1·82 and a TDM model age of 1·90 Ga. The two felsic samples, C9a and C61a, with ages of 1040 Ma and 1088 Ma, have initial 143Nd/144Nd ratios of 0·510886 and 0·511064, Nd(t) values of –8·02 and –3·31, and TDM model ages of 2·27 Ga and 1·92 Ga, respectively. The remaining undated mafic to felsic samples have initial 143Nd/144Nd(1090–1040) values between 0·510874 and 0·51307, Nd(1090–1040) values ranging from 1·44(1090) and –7·03(1090) to 1·28(1040) and –7·50(1040), and give TDM model ages between 1·74 and 2·35 Ga. The whole-rock 87Sr/86Sr ratios for all samples lie between 0·711 and 0·762 and the initial 87Sr/86Sr ratios for samples C6, C9a and C61a are 0·714, 0·716 and 0·718, respectively. The remaining samples have 87Sr/86Sr(1090–1040) ratios between 0·697–0·712(1090) and 0·698–0·713(1040). One grain from sample C9a (C9a-4c) with an age of c. 2080 Ma has an initial 176Hf/177Hf(t) ratio of 0·28146 and an Hf(i) value of –0·9 (Table 4).
|
The Chewore Complex
Four meta-basaltic samples (165, 236, 286, 373) and one felsic sample (220a) from the Kaourera Arc, and one meta-basaltic sample (206) from the Chewore Ophiolite were analysed for their whole-rock isotopic compositions. The geochemical data for these samples were previously presented by Johnson & Oliver (2000
, 2004) and are also shown in Fig. 7c and f. The four Kaourera Arc metabasalts have whole-rock 147Sm/144Nd ratios between 0·1582 and 0·2240 with initial 143Nd/144Nd(1090–1040) ratios of 0·51108–0·51571 and Nd(1090–1040) values ranging between +5·2 to –3·22 and +5·38 to –1·36 (Table 3; Fig. 11a and b). Three of these samples have measured 143Nd/144Nd values close to, or higher than present-day value of CHUR (0·512638; Goldstein et al., 1984) and so give erroneous model ages (Table 3), but sample 165 with lower values provides a TDM age of 2·15 Ga. The felsic sample (220a) dated at c. 1082 Ma (Johnson & Oliver, 2004) has an initial 143Nd/144Nd ratio of 0·510553, an Nd(t) value of –13·47 (Table 3) and a measured 143Nd/144Nd ratio greater than present-day CHUR, precluding the calculation of a reliable model age. The metabasalt from the Chewore Ophiolite, which has an age of c. 1393 (Oliver et al., 1998), has an initial 143Nd/144Nd ratio of 0·510880, an Nd(t) v