The provenance of sub-cratonic mantle beneath the Limpopo Mobile Belt (South Africa)

Lithos 170–171 (2013) 90–104

Contents lists available at SciVerse ScienceDirect

Lithos journal homepage: www.elsevier.com/locate/lithos

The provenance of sub-cratonic mantle beneath the Limpopo Mobile Belt (South Africa) Quinten H.A. van der Meer a, b,⁎, Martijn Klaver a, Tod E. Waight b, Gareth R. Davies a a b

Department of Earth Science, Faculty of Earth and Life Science, VU University, Amsterdam, The Netherlands Department of Geosciences and Natural Resource Management, Øster Voldgade 10, 1350 Copenhagen K Denmark

a r t i c l e

i n f o

Article history: Received 20 August 2012 Accepted 25 February 2013 Available online 13 March 2013 Keywords: Mantle lithosphere characterisation Craton–craton collision Geothermobarometry Thermochemical lithosphere evolution Diamond potential

a b s t r a c t Petrological, whole rock major element and mineral chemical analysis of mantle xenoliths from the Venetia kimberlite pipes (533 Ma) in South Africa reveals an apparently stratified cratonic mantle beneath the Central Zone of the Limpopo Mobile Belt (LMB) that separates the Kaapvaal and Zimbabwe Cratons. Combined pressure–temperature (P–T) data and petrographic observations indicate that the mantle consists of an upper layer of Low-T coarse-equant garnet + spinel lherzolite (~50 to ~130 km depth). This layer is underlain by a region of mixed garnet harzburgites and garnet lherzolites that are variably deformed (~130 to ~ 235 km depth). An equilibrated geotherm did not exist at the time of kimberlite eruption (533 Ma) and a localised heating event involving the introduction of asthenospheric material to the High-T lithosphere below 130 km is inferred. Low-T garnet–spinel lherzolites are highly melt depleted (40% on average). In contrast, the High-T lithosphere (mostly at diamond stable conditions) consists of a mixed zone of variably sheared and melt depleted (30% on average) garnet harzburgite and mildly melt depleted (20% on average) garnet lherzolite. The chemistry of the High-T xenoliths contrasts with that of minerals included in diamond originating from the same depth. Inclusions suggest diamond crystallisation in a more melt depleted lithosphere than represented by either Low- or High-T xenoliths. High-T xenoliths are proposed to represent formerly melt depleted lithosphere, refertilised by asthenosphere-derived melts during the diapiric rise of a proto-kimberlitic melt pocket. This process is coupled to the positive temperature perturbation observed in the High-T xenoliths and may represent a common process in the lower lithosphere related to localised but intense tectono-magmatic events immediately preceding kimberlite eruption. The presence of clinopyroxene, garnet and abundant orthopyroxene in the Low-T lherzolite implies a history of melt depletion followed by metasomatic addition of Si–Al–Ca, forming high-temperature orthopyroxene from which clinopyroxene and garnet exsolved. Si enrichment is a characteristic feature of the majority of the Kaapvaal Craton to the south of the LMB but not of the Zimbabwe Craton to the north, implying a Kaapvaal origin. The provenance of the High-T lithosphere beneath the LMB is less well constrained as it is intensely modified by kimberlitic magmatism and diamond inclusion chemistry does not show significant systematic variation across the cratons. The presence of rare, mildly silica enriched high-temperature harzburgites suggests that a Kaapvaal origin for the entire lithosphere beneath the LMB is most likely. © 2013 Elsevier B.V. All rights reserved.

1. Introduction The Venetia kimberlite cluster in northernmost South Africa was emplaced at ~533 Ma (Allsopp et al., 1995) and is one of South Africa's most diamondiferous kimberlites with a grade of >1.2 ct/ton (Field et al., 2008). In total, 15 bodies were intruded over an area of ~3 km2 (Tait and Brown, 2008) into the central zone of the Limpopo Mobile Belt (LMB), which forms the ~2.65 Ga (Barton and van Reenen, 1992; Gerdes and Zeh, 2009; Zeh et al., 2007) collision zone between the ⁎ Corresponding author at: Department of Geosciences and Natural Resource Management, Øster Voldgade 10, 1350 Copenhagen K, Denmark. Tel.: +45 35324357. E-mail address: [email protected] (Q.H.A. van der Meer). 0024-4937/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.lithos.2013.02.019

Kaapvaal and Zimbabwe cratons in southern Africa. The LMB can be subdivided into three distinct zones (Fig. 1). The Northern and Southern Marginal Zones are commonly interpreted to be high grade metamorphic equivalents of the Zimbabwe and Kaapvaal cratons, respectively (e.g., Du Toit et al., 1983; Van Reenen et al., 1992). The Central Zone, hosting the Venetia kimberlite cluster, acted as the overriding plate during the Kaapvaal–Zimbabwe collision (Durrheim et al., 1992) but is itself of a debatable origin (see review in Rigby et al., 2008) and an origin as part of either the Kaapvaal or Zimbabwe Craton has been proposed. Alternatively, the Central Zone could be an allochtonous block that formed at 3.28 Ga (Zeh et al., 2007) and was subsequently incorporated during the Kaapvaal–Zimbabwe collision. The LMB underwent subsequent high grade metamorphism and tectonism at 2.65 Ga, the

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

91

Fig. 1. Map of the Kalahari Craton and subdivision between the separate terrains after Griffin et al. (2003) and Richardson et al. (2009), including the Limpopo Mobile Belt and its subdivision. Selected kimberlites shown for reference after Field et al. (2008): K = Kimberly cluster, P = Premier/Cullinan kimberlite, M = Murowa/Sese kimberlites, V = Venetia kimberlite cluster. Outcrop of the 2.05 Ga Bushveld LIP is shown in black. Dashed lines indicate national boundaries. SA, South Africa; NAM, Namibia; LS, Lesotho; SW, Swaziland; BW, Botswana; ZIM, Zimbabwe.

proposed age of collision and craton stabilisation (e.g., de Wit et al., 1992; Zeh et al., 2007). A second metamorphic episode at 2.0 Ga is related to deformation and mafic underplating which has been proposed to be either associated with the emplacement of the Bushveld Large Igneous Province (LIP) (Millonig et al., 2010; Richardson and Shirey, 2008; see Fig. 1) or due to tectonism as a consequence of the 2.0 Ga Magondi Orogeny on the north-western margin of the Zimbabwe Craton (McCourt and Armstrong, 1998). Despite a tectonic setting within a mobile belt, the LMB central zone is underlain by thick cratonic lithosphere (James et al., 2001; Kopylova et al., 1997). Previous studies of mantle xenoliths (Barton and Gerya, 2003; Hin et al., 2009; Stiefenhofer et al., 1999), diamond inclusions (e.g., Viljoen, 2002; Viljoen et al., 1999) and garnet concentrates (Griffin et al., 2003) from the Venetia kimberlites have shown that the subcontinental lithospheric mantle (SCLM) is depleted in magmaphile elements. This melt depletion is thought to have occurred in the Archaean as suggested by xenolith whole rock rhenium depletion ages of >2.5 Ga (Carlson et al., 1999). Moreover, the petrography of mantle xenoliths at Venetia suggests that the underlying SCLM lacks the extensive phlogopite metasomatism common elsewhere in South Africa (Hin et al., 2009; Stiefenhofer et al., 1999). The occurrence of eclogite in the LMB central zone cratonic mantle is rare. Deines et al. (2001) report that less than 10% of inclusion-bearing diamonds are of eclogitic origin and only 10% of the mineral inclusions in diamonds analysed by Viljoen et al. (1999) were of pyroxenite or eclogite composition. Following extensive studies of xenolith populations by staff and students from the VU University (~4 man months), only one mantle-derived eclogite xenolith has been identified to date. This contrasts with other kimberlites within the Kaapvaal Craton where the abundance of eclogitic xenoliths can locally be as high as 15% (Schulze, 1989) but lower eclogite abundances seem normal for the Zimbabwe cratonic lithosphere (Bulanova et al., 2012; Smith et al., 2009). Eclogitic sulphide inclusions in diamond have been dated as being coeval with the emplacement of the Bushveld LIP at

~2.05 Ga (Richardson and Shirey, 2008). The significance of this age is difficult to assess and may indicate diamond growth with only minor modification of the lithosphere (Cartigny et al., 2009; Thomassot et al., 2009) as Venetian eclogitic silicate diamond inclusions are interpreted to originate from subduction processes, not a LIP (Aulbach et al., 2002). A combined Sr–Nd isotope study of four peridotitic silicate diamond inclusions from Venetia does not produce coherent age relationships. The limited data set is interpreted to indicate that diamonds have a maximum age of 2.3 Ga (Richardson et al., 2009). These workers also report an isotopic study of garnet macrocrysts from the Venetian kimberlite and argue for modification of Archaean mantle by Bushveld type magmas at circa 2.0 Ga. The Cambrian emplacement age of the Venetia kimberlite predates Phanerozoic regional magmatic events in Southern Africa such as the Karoo flood basalts (Marsh et al., 1997) and the Mesozoic group 1 and 2 kimberlites (Smith et al., 1985). Hence, mantle xenoliths from Venetia have the potential to retain petrographic and geochemical information about the structure and composition of the SCLM, unaffected by the Mesozoic magmatism that influences a large proportion of xenoliths beneath the Kaapvaal Craton (e.g., Simon et al., 2003, 2007). Such processes are thought to modify the lithosphere sampled by kimberlites and thus post-Archaean thermo-chemical processes are a major component recorded in the mantle xenoliths (Kobussen et al., 2008, 2009). When using mantle xenoliths to describe Archaean processes it is therefore important to identify these later events that altered the mantle and to assess how they affected mantle chemistry and thermal conditions. As noted above, although the Venetian kimberlite predates major kimberlite activity in the region, collision at 2.65 Ga and the emplacement of the Bushveld LIP have been proposed to have had some effect on the Limpopo SCLM (Richardson and Shirey, 2008; Richardson et al., 2009). This article aims to provide a better understanding of the origin of the cratonic mantle beneath the LMB central zone and its development

92

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

up to the time of kimberlite eruption. Additionally we assess if the thick SCLM beneath the LMB is contiguous with either or both of the adjacent cratons. An alternative possibility is that the Limpopo SCLM could have formed elsewhere as a micro-continent or as part of another craton that was subsequently tectonically emplaced between the Kaapvaal and Zimbabwean Cratons. This question is addressed by characterising the major petrographic and chemical characteristics of the Limpopo Mobile Belt SCLM and making a detailed comparison with the SCLM of the adjacent Kaapvaal and Zimbabwe cratons. Whole rock and mineral major element compositions of 36 newly collected Venetian mantle xenoliths are presented along with pressure–temperature estimates to provide a comprehensive description of the local SCLM. 2. Methods Xenolith samples were processed to yield polished 200 μm thick sections, whole rock powders and mineral separates at the Geological Technical Laboratory of the VU University, Amsterdam. Weathering and alteration rims were first removed using a diamond-blade saw. Care was taken to remove metallic streaks on the samples originating from the saw blade as these could form a source of contamination for siderophile elements. Samples large enough to yield a representative whole rock powder (>100 g) were crushed using a jaw crusher and agate mills. Fused glass beads and pressed powder pellets were prepared from the whole rock powder and measured for major and minor element concentrations respectively using a Philips Pananalytical MagiX Pro XRF spectrometer at the Department of Earth Science, VU University Amsterdam. Interference-corrected spectra intensities were converted to oxide-concentrations using calibration curves consisting of natural standards closely approximating a peridotite matrix. Loss on ignition was determined but is presumed to derive entirely from hydration of olivine and orthopyroxene due to weathering. Hence, anhydrous whole rock compositions recalculated to 100% are reported, with iron expressed as total ferrous iron (FeO*). Polished thick sections of the samples were examined under a petrographic microscope and using backscattered electron images. Representative, non-altered minerals were analysed for major and minor elements using a Jeol JXA 8800M electron microprobe (EMP) at the Department of Earth Science, VU University Amsterdam. The EMP was operated at an acceleration voltage of 15 kV and an electron beam intensity of 25 nA. A focussed spot was used for all phases and acquisition time was 25 s on the peak followed by 12.5 s background on both sides of the peak for major elements. For minor elements both acquisition times were increased by 50%. All analyses were calibrated using the ZAF-method against natural and synthetic mineral standards. Modal mineral abundances were calculated by least squares fitting of EMP mineral data to whole rock compositions using a Monte Carlo simulation (Table 1). Emphasis was placed on constraining elements least likely to be distorted by kimberlite infiltration and weathering. For samples without a whole rock analysis, or whenever the simulation yielded improbable results, point counting was used to estimate mineral modes (Table 1). Errors for point counting include the assumption that a thin section may not be representative of larger rock volumes and are set at 20%, and around 10% for the Monte Carlo simulation. 3. Results 3.1. Petrography The studied samples were collected from the coarse oversize dump and drill core sections at the Venetia diamond mine; all samples except one pyroxenite originate from the main pipe, K1 (Tait et al., 2006). A careful selection of the samples was made and all visibly different types of peridotite (garnet-bearing and garnet-free, fine-grained and coarse-grained etc.) were included to ensure a representative population.

For this study, 33 peridotitic and three pyroxenitic xenoliths were selected for analysis from ~1000 examined samples of which roughly 95% were of peridotitic paragenesis. The remaining 5% consisted of pyroxenite. Sample sizes range from 5 to 40 cm but size was a subordinate discriminator to mineralogy and degree of alteration during sample selection. The majority of the examined samples are affected to variable degrees by post emplacement low temperature alteration that locally can result in the total replacement of mantle mineralogy. A minimum requirement for the samples selected for this study was the preservation of ±mm sized pristine mineral cores. Peridotite xenoliths with obvious signs of extensive kimberlite veining and metasomatism such as phlogopite, amphibole, high concentrations of clinopyroxene, purported to be late in the temporal evolution of the xenoliths (Erlank et al., 1987; Simon et al., 2003) are a rarity among Venetian xenoliths (Stiefenhofer et al., 1999). These samples were, however, avoided in this study when identified. Of the selected samples, two contain trace amounts of phlogopite (see Table 1) while all samples are subject to minor kimberlite infiltration along cracks and grain boundaries. Based on the petrography of the 36 samples, five distinct groups were recognised: a) Low-temperature (T) lherzolites, generally undeformed coarse garnet (+spinel) bearing lherzolites; b) Low-T spinel peridotites, garnet-free coarse spinel peridotites; c) harzburgites, clinopyroxene-free coarse to porphyroclastic garnet harzburgites, classified on the criteria of having no visible clinopyroxene in hand specimen and no primary clinopyroxene in thin section (i.e., cpx not in reaction rims or veins), except for very minor exsolution lamellae in orthopyroxene; d) High-T lherzolites, spinel-free garnet lherzolites with a mosaic porphyroclastic structure; and e) pyroxenites. This subdivision is comparable to previous subdivisions of the Venetian mantle xenolith suite in Stiefenhofer et al. (1999) and Hin et al. (2009).

3.1.1. Low-T garnet (spinel) lherzolites The first group consists of 16 garnet (+ spinel) lherzolites that are characterised by a coarse equant texture (terminology after Harte, 1977) with up to 1 cm orthopyroxene and olivine crystals. Although olivine and orthopyroxene generally have equilibrated straight crystal boundaries, some curving boundaries and inclusions of olivine in orthopyroxene are also observed in the majority of the samples. Garnet, spinel and clinopyroxene have an irregular lobate form and can occur in disrupted aggregates associated with or adjacent to orthopyroxene grains (Fig. 3). The poorly equilibrated boundaries between garnet, clinopyroxene and spinel aggregates and the coarse olivine and orthopyroxene structure are indicative of a common exsolution origin of cpx-gt from orthopyroxene (Cox et al., 1987; Saltzer et al., 2001). In some samples, exsolution lamellae of clinopyroxene in orthopyroxene are also visible. Modal olivine content in the samples ranges between 55% and 75% and makes up the bulk of the samples together with orthopyroxene. Clinopyroxene, garnet and spinel concentrations range from trace amounts to several % (Table 1). Only one studied sample (ATC 752) has over 5% clinopyroxene and is thus the only classical lherzolite according to the scheme of Streckeisen (1976). One sample (ATC 722) has a coarse laminated structure forming a shear fabric. The individual crystals in this sample have equilibrated boundaries and appear strain free. The shear fabric has been overprinted by static recrystallisation, and deformation must have significantly pre-dated kimberlite entrainment of the sample (in the order of at least several Myr, Evans et al., 2001). One spinel-free garnet lherzolite (ATC 753) has a coarse granuloblastic texture with irregularly shaped garnets of up to 5 mm. Olivine and orthopyroxene (2–5 mm) have grain boundaries that are smooth and curving indicating that they are not fully equilibrated. Undulose extinction is observed in olivine and orthopyroxene in some of the samples indicating minor low-strain deformation.

Table 1 Whole rock major element compositions of all the analysed samples, calculated modal mineral abundances by weight and calculated pressure temperature and depth of mineral equilibration. Major element oxides were measured by XRF in glass beads made from whole rock powder with the exception of NiO and Cr2O3 which were calculated from Ni and Cr concentrations measured in pressed powder pellets. All Fe measured as Fe2O3 and recalculated to FeO. Depths were calculated from pressure estimates using an average pressure increase of 0.03 GPa/km in the crust and 0.035 GPa/km for the mantle (Winter, 2002). Note; samples AT 1315 and AT 1337 were not treated in further discussion and figures because of (probable) disequilibrium in major element concentrations, see 3.3. Samples AT 1317 and AT 1342 were left out of the table for clarity and lack of WR analyses for these samples.

SiO2 TiO2 Al2O3 Cr2O3 FeO MnO MgO NiO CaO Na2O K2O P2O5 BaO Total LOI% Mg# Garnet Spinel Olivine opx cpx Phlogopite T [°C] P [GPa] Depth [km]

Spinel–peridotites

AT 1324

AT 1348

AT 1350

ATC 722

ATC 724

ATC 725

ATC 726

ATC 730

ATC 731

ATC 732

ATC 740

ATC 742

ATC 745

ATC 747

ATC 752

ATC 753

AT 1333

ATC 720

48.49 0.04 1.55 0.40 5.60 0.10 43.71 0.26 0.78 0.05 0.12 0.01 0.01 101.12 5.06 93.29 0.01 0.01 0.56 0.41 0.01 – 858 3.17 98

46.85 0.01 1.58 0.40 5.85 0.09 44.87 0.28 0.50 0.06 0.15 0.01 0.00 100.66 6.35 93.22 – 0.03 0.58 0.36 0.03 – 774 3.01 93

45.82 0.02 1.00 0.31 6.14 0.09 47.07 0.31 0.52 0.04 0.11 0.02 0.00 101.43 8.98 93.23 – 0.02 0.69 0.26 0.03 – 817 2.92 91

45.87 0.01 1.47 0.35 5.99 0.10 45.39 0.29 0.79 0.06 0.14 0.04 0.01 100.50 6.55 93.11 0.03 0.01 0.69 0.26 0.01 – 827 2.53 80

51.07 0.01 1.69 0.40 6.20 0.19 37.84 0.31 1.05 0.25 0.10 0.01 0.00 99.12 7.31 91.58 0.01 0.01 0.61 0.35 0.02 – 784 2.13 68

46.72 0.01 1.45 0.38 6.01 0.10 44.69 0.28 1.02 0.11 0.18 0.04 0.01 100.98 1.85 92.99 0.00 0.00 0.65 0.28 0.04 0.03 757 1.92 62

46.29 0.00 1.56 0.33 6.18 0.10 45.92 0.31 0.52 0.02 0.02 0.00 0.00 101.25 6.88 92.98 0.05 0.00 0.68 0.26 0.00 – 836 2.91 90

45.71 0.04 1.59 0.37 6.41 0.10 45.65 0.30 0.79 0.05 0.13 0.01 0.00 101.16 3.01 92.70 0.04 0.00 0.72 0.22 0.02 – 802 2.70 84

45.96 0.02 1.27 0.35 5.96 0.10 45.08 0.31 1.06 0.19 0.33 0.07 0.02 100.71 3.18 93.09 0.00 0.01 0.66 0.31 0.02 – 799 2.44 77

46.86 0.03 1.67 0.37 6.13 0.11 44.17 0.28 1.06 0.08 0.14 0.02 0.01 100.92 5.90 92.78 0.01 0.01 0.62 0.32 0.02 0.02 679 1.78 58

46.10 0.00 1.79 0.37 6.14 0.10 45.01 0.29 0.75 0.03 0.05 0.00 0.00 100.64 1.92 92.89 0.06 0.00 0.64 0.28 0.01 – 776 2.59 81

46.70 0.11 1.51 0.36 6.72 0.11 44.33 0.28 1.08 0.12 0.24 0.03 0.01 101.58 8.91 92.16 0.05 0.00 0.65 0.28 0.02 – 831 3.21 99

48.29 0.07 4.70 0.82 6.99 0.24 34.71 0.32 2.90 0.40 0.08 0.02 0.02 99.56 10.79 89.86 0.06 0.00 0.65 0.28 0.02

46.91 0.04 2.28 0.49 6.24 0.14 43.29 0.31 1.00 0.09 0.19 0.02 0.01 101.00 7.76 92.52 0.07 0.01 0.58 0.33 0.01 – 794 2.61 82

46.61 0.01 2.44 0.41 6.78 0.12 41.19 0.30 1.83 0.16 0.06 0.00 0.00 99.92 3.05 91.55 0.07 0.00 0.59 0.29 0.05 – 748 2.12 68

46.41 0.01 3.10 0.32 6.89 0.11 42.08 0.29 1.53 0.12 0.11 0.02 0.00 100.98 8.70 91.59 0.07 0.00 0.56 0.34 0.03 – 822 2.64 83

45.79 0.03 1.42 0.38 6.53 0.10 45.43 0.29 0.66 0.05 0.08 0.01 0.01 100.77 6.95 92.54 – 0.01 0.53 0.43 0.03 – – – –

48.14 0.01 1.52 0.35 6.26 0.12 45.05 0.31 0.47 0.08 0.03 0.01 0.02 102.37 9.37 92.77 – 0.01 0.58 0.39 0.01 – – – –

796 2.40 78

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

Low-T garnet (spinel) lherzolites

(continued on next page)

93

94

Table 1 (continued) Garnet harzburgites

Pyroxenites

K1 kimb.

AT 1304

AT 1305

AT 1311

AT 1334

AT 1338

AT 1345

AT 1303

AT 1314

AT 1315

AT 1337

AT 1355

AT 1361

AT 1198

AT 1477

ATC 727

AT-kim

46.59 0.06 1.36 0.34 6.68 0.10 44.22 0.32 0.91 0.12 0.07 0.01 0.01 100.78 9.56 92.24 0.05 – 0.66 0.28 0.00 – 1510 6.75 200

47.50 0.01 1.17 0.36 6.66 0.09 45.63 0.31 0.53 0.07 0.03 0.01 0.01 102.39 6.27 92.48 0.05 – 0.66 0.29 0.00 – 1288 5.37 161

44.48 0.05 1.11 0.34 6.98 0.09 47.10 0.33 0.34 0.02 0.03 0.01 0.01 100.89 10.97 92.38 0.04 – 0.79 0.17 0.00 – 1423 6.13 182

51.97 0.05 1.32 0.32 6.20 0.10 40.07 0.37 0.87 0.20 0.04 0.00 0.00 101.53 11.43 92.06 0.04 – 0.36 0.58 0.02 – – – –

45.51 0.04 1.40 0.37 6.64 0.10 45.70 0.29 0.86 0.10 0.13 0.04 0.01 101.19 7.45 92.51 0.05 – 0.72 0.22 0.01 – 1417 6.25 186

43.93 0.06 1.34 0.33 6.79 0.10 47.96 0.30 0.48 0.08 0.22 0.03 0.01 101.64 9.10 92.69 0.04 – 0.85 0.11 0.00 – 1120 5.09 153

– – – – – – – – – – – – – – – – – – – – – – 1452 5.44 163

– – – – – – – – – – – – – – – – – – – – – – 1339 5.79 173

– – – – – – – – – – – – – – – – 0.05 – 0.73 0.21 0.01 – 1162 4.33 131

44.05 0.07 1.31 0.25 8.49 0.12 44.99 0.29 2.20 0.15 0.15 0.01 0.01 102.09 8.21 90.43 0.06 – 0.83 0.11 0.01 – 1309 5.03 151

44.87 0.04 0.87 0.40 7.94 0.10 46.26 0.31 0.47 0.09 0.07 0.02 0.00 101.44 8.60 91.29 0.04 – 0.81 0.15 0.00 – 1449 5.33 159

42.95 0.05 1.09 0.22 7.46 0.10 47.58 0.30 0.64 0.05 0.04 0.01 0.00 100.51 7.67 91.98 0.04 – 0.86 0.08 0.01 – 1321 5.14 154

44.20 0.04 1.15 0.29 7.60 0.10 46.81 0.33 0.82 0.11 0.09 0.02 0.00 101.58 8.53 91.65 0.04 – 0.79 0.15 0.02 – 1324 5.13 154

47.68 0.46 8.17 0.31 8.23 0.20 17.60 0.03 16.02 0.81 0.05 0.03 0.00 99.59 3.15 79.21 0.20 – – – 0.65 – 506 0.52 22

– – – – – – – – – – – – – – – – 0.18 – – 0.50 0.32 – 863 3.43 105

49.10 0.07 1.68 1.85 7.50 0.15 38.11 0.20 1.35 0.14 0.08 0.01 0.00 100.23 2.15 90.06 – 0.02 0.36 0.57 0.05 – – – –

42.08 1.04 4.44 0.21 8.05 0.18 30.60 0.16 9.89 0.51 2.90 0.38 0.11 100.53 10.06 87.15 – – – – – – – – –

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

SiO2 TiO2 Al2O3 Cr2O3 FeO MnO MgO NiO CaO Na2O K2O P2O5 BaO Total LOI% Mg# Garnet Spinel Olivine opx cpx Phlogopite T [°C] P [GPa] Depth [km]

High-T garnet lherzolites

AT 1302

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

95

Fig. 2. a–c Selection of scanned thin sections. a) Coarse equant garnet spinel lherzolite (ATC 740) with abundant pale yellow orthopyroxene and colourless olivine with dark serpentine filled cracks. Aggregates of purple garnet with bright green clinopyroxene and irregular patches of black spinel are also present. b) Coarse equant garnet harzburgite (AT1304) with pale yellow orthopyroxene, partly serpentinised olivine and purple subcalcic garnet. c) Mosaic porphyroclastic garnet lherzolite (AT 1361) with rounded garnet, orthopyroxene and clinopyroxene porphyroclasts in a fine grained serpentinised olivine matrix, each image is approximately 25 by 40 mm.

3.1.2. Low-T spinel peridotites Four samples of garnet-free spinel peridotite have been studied. They are texturally and mineralogically similar to the garnet (spinel) lherzolites with the exception of the absence of garnet and, in some samples, clinopyroxene (see modal abundances in Table 1). Spinel is

present as an opaque or deep red Cr-rich phase and forms anhedral grains, generally of up to 0.5 mm in size. Clinopyroxene is rare or absent in the spinel peridotites, but when present it is usually anhedral and occurs in the interstices of olivine and opx as shown in Fig. 3b and d.

Fig. 3. a, b, c, d. Backscatter electron images (scale on figures) illustrating textures in garnet (spinel) lherzolite. a: Fine aggregate of garnet (gt), clinopyroxene (cpx) and spinel (sp) between coarse larger olivine and orthopyroxene (opx) grains in Low-T lherzolite ATC 724. Opx and olivine have equilibrated grain boundaries while clinopyroxene, spinel and garnet form more irregular grains and “intrude” olivine and opx grains. Serpentinisation (black in BSE) with bright secondary Fe oxide mineralisation is visible in cracks in olivine. b: Wedge-shaped laths of cpx (white) exsolved from orthopyroxene in sample ATC 752. Interstitial cpx is visible in the top right corner. c: Euhedral orthopyroxene crystal encapsuled within garnet porphyroclast in High-T lherzolite AT1361. Fine grained kelyphite has developed along some of the cracks in the garnet and on the garnet–orthopyroxene grain boundaries. d: Irregularly shaped anhedral garnet and clinopyroxene intergrown with orthopyroxene and olivine in Low-T lherzolite ATC 740.

96

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

Table 2 Mineral major element compositions of representative samples for each peridotite group. Complete dataset presented as electronic supplement. Low-T lherzolite: ATC 740

SiO2 TiO2 Al2O3 Cr2O3 FeO MnO MgO NiO CaO Na2O Total Mg# Cr#

Spinel–peridotite: AT 1317

Garnet

Spinel

Olivine

opx

cpx

Spinel

Olivine

opx

cpx

42.48 0.00 22.38 2.66 7.01 0.38 20.79 0.00 5.26 0.01 100.99 84.09 7.37

0.03 0.01 18.24 55.15 13.66 0.61 14.23 0.09 0.00 0.02 102.04 64.99 66.98

41.61 0.00 0.00 0.01 7.10 0.06 52.01 0.43 0.01 0.01 101.25 92.88 93.83

58.06 0.00 1.12 0.40 4.36 0.10 35.92 0.09 0.32 0.06 100.43 93.62 19.19

54.84 0.01 3.01 1.85 1.19 0.07 16.06 0.04 21.14 1.91 100.11 96.01 29.17

0.03 0.01 26.26 45.19 13.64 0.20 14.93 0.10 0.01 0.01 100.37 66.12 53.58

41.21 0.00 0.00 0.00 6.77 0.09 50.94 0.42 0.01 0.00 99.45 93.06 100.00

57.14 0.00 1.85 0.46 4.32 0.11 35.01 0.09 0.46 0.04 99.48 93.52 14.21

54.07 0.00 3.27 1.16 1.32 0.05 16.04 0.04 22.06 1.39 99.40 95.60 19.18

High-T lherzolite: AT 1361

SiO2 TiO2 Al2O3 Cr2O3 FeO MnO MgO NiO CaO Na2O Total Mg# Cr#

Garnet harzburgite: AT 1304

Garnet

Olivine

opx

cpx

Garnet

Olivine

opx

42.26 0.21 20.31 4.30 6.32 0.37 21.68 0.02 4.98 0.04 100.49 85.95 12.44

40.63 0.00 0.04 0.00 8.24 0.11 50.30 0.38 0.08 0.04 99.83 91.58 7.12

57.42 0.08 1.31 0.35 4.98 0.13 33.82 0.12 1.31 0.30 99.83 92.37 15.17

55.32 0.12 2.30 0.95 3.29 0.13 19.70 0.07 16.21 1.60 99.69 91.43 21.61

42.64 0.03 20.01 5.37 5.78 0.24 22.01 0.03 3.92 0.01 100.05 87.16 15.25

41.20 0.00 0.03 0.06 7.24 0.10 50.13 0.40 0.05 0.01 99.21 92.51 55.58

57.77 0.01 0.95 0.35 4.40 0.11 34.62 0.14 0.70 0.09 99.13 93.34 19.68

3.1.3. Garnet harzburgites The seven garnet harzburgitic xenoliths differ from the lherzolitegroups by the absence of primary clinopyroxene, although minor clinopyroxene exsolution lamellae and rims are found in b5% of the orthopyroxene grains. Their texture varies between equant coarse and (mosaic-) porphyroclastic and the degree of recrystallisation of olivine is variable between and within samples. Up to 10 mm large olivine crystals can be present in the least recrystallised samples. Round garnet grains are present in all samples and range from 4 to 10 mm. In contrast to the Low-T garnet (spinel) lherzolites, spinel is completely absent from the harzburgite samples apart from finegrained kelyphitic rims around garnet. In addition, olivine is on average more abundant than in the Low-T garnet (spinel) lherzolites and constitutes 66–84% of the samples. Hence, the harzburgitic xenoliths more closely resemble the High-T garnet lherzolites discussed below than the Low-T lithologies.

3.1.5. Pyroxenites Three pyroxenite samples with variable mineralogies have been examined. One sample (AT 1198) is a fine grained (~1 mm grains) mosaic garnet clinopyroxenite with accessory apatite, calcite and rutile. Garnet makes up ~20% of the rock, with ~65% clinopyroxene and ~15% of altered minerals, probably after opx in which case the sample should be classified as a websterite. The second sample is a coarse equant garnet websterite that consists of ~18% garnet with 50% orthopyroxene and 32% clinopyroxene. The phases are similar in size, about 5 mm across. The last sample is a spinel and clinopyroxene-bearing olivine orthopyroxenite. This coarse grained sample has up to 1 cm sized orthopyroxene (56%) and olivine (36%) with interstitial clinopyroxene (~5%) and Cr rich spinel (~2%) that have an irregular shape. This lithology has textural similarities with the spinel peridotite group but higher spinel, clinopyroxene and orthopyroxene contents. 3.2. Major elements

3.1.4. High-T garnet lherzolites Six samples were assigned to the High-T garnet lherzolite group. All samples have a mosaic porphyroclastic fabric with bimodal grain size distribution between fine grained olivine and porphyroclasts of clinopyroxene, orthopyroxene and garnet (Fig. 2c), indicative of deformation. The samples have a higher abundance of olivine than the samples from the Low-T sample groups ranging between ~75% and 85%. Most of the fine grained olivine has suffered from intense alteration to serpentine, while some larger, >1 mm olivine grains preserve pristine cores. Orthopyroxene makes up between 8% and 20% of the samples and forms 1–5 mm large porphyroclasts often intergrown with clinopyroxene (trace to 2% of sample). Orthopyroxene is generally deformed, preserving kink bands and signs of dynamic recrystallisation. All samples have large (up to 1 cm) round or rounded garnet porphyroclasts making up 4–5% of the samples. Garnet grains have kelyphite rims that vary in thickness from 100 μm to total consumption of the mineral.

Whole rock major element oxide concentrations of representative samples along with their calculated modal mineral abundances are presented in Table 1. Table 2 lists representative mineral major element concentrations. Mineral cores and rims were analysed but no significant (>3 times SD or above typical detection limits) major element variations between core and rim were found apart from where rims were altered. No variation between primary minerals within a single sample was found regardless of the texture of the mineral. For example, clinopyroxene was homogeneous when occurring as exsolution lamellae in orthopyroxene, as (>0.5 mm) independent minerals or in a fine grained (b 100 μm sized minerals) aggregate with garnet, spinel and orthopyroxene (see also Fig. 2a and b). Some altered rims appear to have interacted with kimberlite, for example resulting in the introduction of Ti. Major element classification (after Dawson and Stephens, 1975; Gurney, 1984) of garnet (Fig. 4) indicates that only 2 out of 7 harzburgites possess Ca undersaturated garnet (G10), while 2 others

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

are Ca-saturated (G9) in contrast with the observed clinopyroxene free mineralogy. The remaining harzburgitic samples have garnet compositions on the border between Ca saturated and undersaturated compositions. Although websterite AT1198 differs from all others in texture and mineralogy, garnet Mg# and Ca# (Ca/(Mg + Ca)) and in particular the low sodium-content indicate a likely mantle origin of the sample (c.f. Schulze, 2003).

3.3. Pressure temperature calculation Pressures and temperatures of chemical equilibration between mineral cores were calculated using the barometer PNG85 of Nickel and Green (1985) (based on garnet-orthopyroxene Al–Cr exchange) combined with the TNG09 garnet–orthopyroxene Fe–Mg exchange thermometer (Nimis and Grütter, 2010) and are listed in Table 1 and plotted in Fig. 5. Calculated errors on the used thermo-barometry couple are ±34 °C (TNG09) and ±0.4 GPa (PNG85) (Nimis and Grütter, 2010). The choice of barometer and thermometer is based on the review of thermometry in Nimis and Grütter (2010) and their effectiveness on natural samples (e.g., Mather et al., 2011). The versatility of the thermometer allows comparison of temperatures between lherzolites and clinopyroxene-free harzburgites using the same phase relations. In addition, all samples with garnet and orthopyroxene major element concentrations reported in earlier xenolith studies (Hin, 2008; Stiefenhofer et al., 1999), (n = 19) have been recalculated and are included in the data set, presented as open symbols in Fig. 4. Following the approach of Nimis and Grütter (2010), temperatures for cpx-bearing lithologies were calculated with different thermometers to assess the extent of major element equilibrium between garnet and pyroxenes. Samples found to be in disequilibrium (i.e. >70 °C difference between TNG09 and enstatite in clinopyroxene thermometer TNT (Nimis and Taylor, 2000)) are not included in the discussion as they may not be representative of mantle conditions prior to kimberlite eruption. Two lherzolites (AT 1315 and AT 1337) analysed in the present study and five lherzolites reported in the literature were discarded on these grounds. Based on the P–T results and in accordance with petrography, garnet (spinel) lherzolites can be classified as (b1000 °C) Low-T lherzolites and sheared garnet lherzolites as (>1000 °C) High-T lherzolites (e.g., Boyd and Mertzman, 1987), separated by a gap in data between 4 and 4.5 GPa (~120 to ~135 km). Garnet-free 20 Diamond inclusions (Viljoen et al., 1999)

18 16 14

Cr2O3

12 ~72%

10

G10

8

G9

6 4 2 0 0

1

2

3

4

5

6

7

8

9

10

CaO Low-T Lherzolite

Pyroxenite

High-T Lherzolite

Garnet Harzburgite

Fig. 4. Garnet classification of studied and literature samples from Venetia based on CaO and Cr2O3 after Grütter et al. (2004). Filled symbols: garnet in xenolith, this study. Open symbols: garnet in xenolith from Stiefenhofer et al. (1999) and Hin et al. (2009). Field representing garnet included in diamond after Viljoen et al. (1999).

97

peridotites only allow for temperature calculation which was performed for pre-set pressures between 1.5 and 5 GPa. The intersection of the P–T range for each garnet free sample with the P–T array defined by the garnet-bearing samples is adopted as their pressure and temperature of equilibration (Fig. 5). 4. Discussion In order to gain a fuller understanding of the formation and evolution of the SCLM beneath the Limpopo Mobile Belt Central Zone, a detailed comparison is made between Venetian xenoliths and literature data from the adjacent Kaapvaal and Zimbabwe Cratons. Initially, the thermal state of the mantle beneath the LMB is assessed. Second, the major element chemistry of the Low-T and High-T domains of the mantle is compared to the Kaapvaal and Zimbabwe Cratons. Additionally, a comparison is made between Venetian xenoliths and published diamond inclusion data (Viljoen et al., 1999) to assess the chemical evolution of the High-T (diamond stability) portion of the mantle between diamond formation at > 2.0 Ga (Richardson et al., 2009) and kimberlite eruption at 533 Ma. 4.1. P–T calculations 4.1.1. Mantle stratification? The calculated pressures and temperatures of all samples in which the different phases are in major element equilibrium (following Nimis and Grütter, 2010) are reported in Fig. 5. All garnet harzburgites are included in the figure but it is acknowledged that an assessment of the extent of chemical equilibrium within these clinopyroxene-free peridotites is impossible. Based on regional tomographic studies a crustal thickness of 50 km is assumed (Nguuri et al., 2001). Garnet (spinel) lherzolites and two garnet pyroxenites define a linear array between 1.6 GPa and 4 GPa on a P–T diagram with a narrow spread in data points (b ±25 °C; Fig. 5) around the array. These data correspond to a layer between 50 km and 130 km, starting from just below the depth of the Moho (or crust–mantle boundary). Temperatures for garnet-free spinel peridotites were calculated at different preset pressures (thin lines on Fig. 5) and these data indicate an origin from a similar depth as the Low-T samples, in accordance with their similar petrographic characteristics. The absence of garnet in these samples is hard to explain as they do not contain significantly higher Cr contents (cf. Klemme, 2004) and may therefore be an artefact of inhomogeneity of the mantle on a larger than xenolith (>10 cm) scale. Samples derived from depths corresponding to 130–235 km, mostly within the diamond stability field, are garnet harzburgite and High-T lherzolite. Surprisingly, harzburgites or lherzolites do not show any systematic relations in terms of depth of origin, and based on these P–T data they appear to be intermixed in the lower lithosphere. These High-T lithologies are derived from depths that extend to the base of the lithosphere as inferred from seismic tomography (Fouch et al., 2004). Nearly half of all calculated High-T samples (10 out of 21) cluster between 5 and 6 GPa and 1300 and 1450 °C, perhaps suggesting that this part of the diamond stability field was preferentially sampled by the kimberlite. Compared to the Low-T samples, the High-T peridotites have calculated temperatures with a far greater variability (up to ±250 °C) and scatter markedly around the empirical, xenolith based Kalahari geotherm of Rudnick and Nyblade (1999) (Fig. 5). This greater temperature variability at depth is highlighted by both garnet harzburgite and High-T lherzolites with temperatures up to 200 °C higher than the Kalahari geotherm. The control on major element equilibrium in the two harzburgite samples is admittedly poor and some of the outliers may reflect disequilibrium between garnet and orthopyroxene. The reported High-T lherzolite samples, however, are in major element equilibrium and appear to provide unambiguous evidence of a positive temperature perturbation compared to the Kalahari geotherm. This finding is associated with

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

T [ºC] 0

0

500

1000

1

1500

2000

Moho

2

P [GPa]

3 4 5 6 7 8 Low-T Lherzolite

Pyroxenite

High-T Lherzolite

Garnet Harzburgite

Fig. 5. Calculated pressures and temperatures for garnet- and orthopyroxene-bearing samples (filled symbols: this study, open symbols: Stiefenhofer et al., 1999; Hin, 2008) Included are the diamond–graphite stability equilibrium boundary (Kennedy and Kennedy, 1976), approximate Moho pressure for a 50 km thick crust and range of mantle adiabats adopted from Pearson et al. (2003) (TP 1300–1400 °C). All temperatures and pressures are calculated using the PNG85 (Nickel and Green, 1985) barometer coupled with the TNG09 (Nimis and Grütter, 2010) thermometry couple. Temperatures for garnet-free samples were determined at a range of preset pressures using a modification of the Ca-in-opx thermometer (Nimis and Grütter, 2010). The three garnet-free lherzolites (thin black lines) and one garnet-free orthopyroxenite (thin dashed line) intersect the range of calculated P–T of the Low-T garnet (spinel) lherzolites. Data compared to the Kalahari geotherm of Rudnick and Nyblade (1999) (thick dashed line).

the sheared texture of the rocks, indicating deformation, which could be a consequence of material and heat transport occurring prior to upward transport in the kimberlite. Modelling of the annealing time of the b1 mm olivine fabric at >1300 °C restricts the shear fabric formation to ≪1 year prior to kimberlite eruption (Karato, 1989). Therefore the deformation is considered to have been ongoing up to the time of kimberlite eruption. 4.1.2. Transient geotherm The Low-T samples define a linear P–T array that crosses the empirical Kalahari geotherm of Rudnick and Nyblade (1999) (Fig. 5) with the shallowest derived samples (1.5–2 GPa) having a systematically higher temperature while the deeper (3–4 GPa) samples have lower temperatures, by up to 150 °C. Porphyroclastic lherzolites have a P–T range similar to garnet harzburgites and thus the exclusion of the latter lithologies, as proposed by some previous workers (e.g., Mather et al., 2011), does not change the inferred palaeogeotherm, indicating thermal equilibration between samples. This observation indicates similar thermal equilibration for the garnet harzburgites and High-T lherzolites. The Venetia data therefore display a significant misfit with the Kalahari geotherm. An alternative palaeogeotherm can be fitted to the data from the LMB central zone (Fig. 6, ‘Transient geotherm’). This palaeogeotherm is compatible with both a thicker heat producing crust than is typical of the Kaapvaal Craton, as found beneath the LMB (Nguuri et al., 2001), and the steep increase in temperature observed between 5 GPa and 6 GPa. There is a distinct change of slope in the P–T data below 4 GPa, at the transition from the Low-T samples to the High-T samples. This apparent kink below 4 GPa cannot represent an equilibrated conductive geothermal gradient and is interpreted to be a transient palaeogeotherm specific to the time of kimberlite eruption. The feature, however, does not appear to be an artefact of late heating during transport in the host kimberlite because the analysed samples record no core to rim variation in mineral chemistry. Furthermore, application of different thermometers based on

different ion exchange reactions to the Low-T samples produces consistent temperature results. The transient palaeogeotherm therefore appears to be the likely effect of heat input elevating the temperature of the lower lithosphere by more than 200 °C prior to kimberlite eruption. The deformed nature of the sheared High-T lherzolites and the apparent random distribution of lherzolite and harzburgite at these depths may be explained by metasomatism by asthenosphere derived Ca–Al–Fe rich melt (e.g. Mercier, 1979; Moore and Lock, 2001) and subsequent ductile mobilisation of the metasomatised peridotite. It is noteworthy that the deepest samples with an excess in temperature correlate best with the empirical Kalahari geotherm and therefore, an excess temperature in the lower lithosphere appears to be a common feature for mantle sampled by kimberlite beneath southern Africa (e.g., Bell et al., 2003; Rudnick and Nyblade, 1999). An inflection of palaeogeothermal gradients has mainly been observed in xenoliths in the abundant Phanerozoic kimberlites from within the Kaapvaal Craton. Hence, the thermal inflection has been interpreted to represent heating of the lower SCLM on a regional scale and has been attributed to relatively young magmatic events such as the ~ 180 Ma Karoo Large Igneous Province (Marsh et al., 1997) and the opening of the Atlantic and the ~ 130 Ma (Peate, 1997) Etendeka/Parana province (e.g., Carlson et al., 2005; Kobussen et al., 2008). Such an interpretation is successful in explaining the observed temperature inflection at depth and satisfies the observation that there is no surface heat flow anomaly because the migration of heat through the lithosphere would take several 100s of Ma to reach the surface (Nyblade, 1999). An identical temperature profile in the 533 Ma Venetia kimberlite emplaced much earlier than the Karoo and Etendeka/Parana flood basalts, however, is incompatible with this interpretation as it must predate these events. Similar temperature profiles are also reported from kimberlites with different ages (e.g., the ~1 Ga Premier kimberlite; Viljoen et al., 2009). At the time of eruption of these older kimberlites there are no recorded recent regional magmatic events such as the Karoo flood basalts on the Kaapvaal and Zimbabwe cratons. Therefore the observed temperature excess may be better explained by localised heating preceding kimberlite eruption, this interpretation is further explored in Section 4.1.4.

T [ºC] 0

0

500

1000

1500

1

2000

Moho

2 3

P [GPa]

98

DI

4 5 6

Diamond window

7 8 Fig. 6. Diagram showing P–T conditions of the SCLM beneath Venetia. The Transient palaeogeotherm is fitted to match the observed P–T calculations while the steady state palaeogeotherm is an extrapolation from the Low-T samples. Symbols (shaded) as in Fig. 4, garnet free peridotite lines removed for clarity. P–T calculations from four garnet– orthopyroxene pairs included in Venetian diamonds are shown as filled black circles (Viljoen et al., 1999). Shaded field indicates P–T determinations from diamond inclusions (DI) from the southern Zimbabwe Craton (Smith et al., 2009). Diamond window indicates the diamondiferous mantle column for the steady-state geotherm scenario. Arrows indicate heating from a steady-state palaeogeotherm to a Transient palaeogeotherm preceding kimberlite eruption. Grey line indicates the 40 mWm−2 conductive geotherm (Pollack and Chapman, 1977) that yields a similar lithosphere thickness but overlaps only partly with calculated pressures and temperatures.

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

4.1.3. Steady state geotherm and diamond potential The intersection between the palaeogeotherm and the mantle adiabat represents the lowermost possible extent of cratonic lithosphere, defined as a thermal boundary layer (e.g. Eaton et al., 2009). Both the Kalahari geotherm and the Transient geotherm beneath Venetia imply that there is only a small (b 25 km) window within the diamond stability field. This geometry would result in a diamond-poor or -absent mantle column. In contrast, the Venetia kimberlite is one of the most diamondiferous in South Africa with a grade of >1.2 ct/ton (Field et al., 2008). Geothermobarometric calculations from silicate inclusions in diamond can be subject to error because of a lack of knowledge of the entire mineral assemblage. However, the pressure range defined by peridotitic Venetian diamond inclusions shows a far greater range than the thickness of the diamond stable lithosphere implied from the High-T xenolith data (minimum implied pressure range based on peridotite inclusions is 5.9–6.9 GPa; Viljoen et al. (1999); Fig. 6) equivalent to at least ~35 km. It therefore appears that neither the Kalahari palaeogeotherm nor the observed Transient palaeogeotherm describes P–T conditions for diamond formation beneath Venetia and potentially for many other kimberlites. The Transient and Kalahari palaeogeotherms are interpreted as the product of heating subsequent to diamond formation at >2 Ga (Richardson et al., 2009). The Low-T samples seem unaffected by this process. This would imply that the Low-T samples represent an older relict palaeogeothermal gradient. Based only on equilibrated Low-T lherzolites, this old, steady-state palaeogeothermal gradient may be linearly approximated (R2 = 0.98) as: PðGPaÞ ¼ 0:0091  Tð ∘CÞ−4:55: Including the Low-T lherzolites that are suspected to be in disequilibrium does not significantly change this temperature envelope but would increase the R 2. When the array is projected to greater depths it intersects with a conservative mantle adiabat (Pearson et al., 2003) at ~ 225 km (7.7 GPa) and 1345 °C (Fig. 6), which we interpret as an indication of a previous steady state lithospheric thickness. This thickness exactly matches estimates of the present day lithospheric thickness under the LMB and Zimbabwe Craton made from tomography (Fouch et al., 2004). This steady-state palaeogeotherm describes a mantle column with an extensive section of the lithosphere within diamond stability of about 90 km (~ 4.5–7.7 GPa, ‘Diamond window’ in Fig. 6) compared to the b25 km from either the Kalahari geotherm (Fig. 5) or the Transient geotherm (Fig. 6). P–T estimates for Venetian peridotitic diamond inclusions (Viljoen et al., 1999) and a more complete dataset of P–T calculations of diamond mineral inclusions (Smith et al., 2009) for the similar aged (538 Ma, Smith et al., 2004) Murowa and Sese kimberlites ~ 200 km to the northeast of Venetia on the Zimbabwe Craton are also plotted in Fig. 6. The reported P–T equilibration conditions of these inclusions lie on an extension of the palaeogeotherm defined by the Low-T lherzolites and record the P–T conditions during diamond formation. This suggests that diamond formation occurred under steady-state geotherm conditions prior to more recent heating of the lower lithosphere. The steady state geothermal gradient indicates a cool cratonic surface heat flow of ~ 40 mWm −2 (Pollack and Chapman, 1977) but it must be noted that such simple conductive heat flow models are unlikely to represent realistic geothermal gradients (Mather et al., 2011), which is also evident from the misfit of conductive heat flow models with the calculated pressures and temperatures. 4.1.4. Transient heating induced by proto-kimberlitic activity The inferred short term change in thermal regime suggests a general relationship between kimberlite magmatism and a deep thermal anomaly arguing for localised heat input that does not heat the lithosphere on a regional scale. Locally xenoliths equilibrated to >200 °C higher than ambient temperatures over the ~85 km thick column represented by High-T lithosphere xenoliths. A rise in temperature of ~350 °C (Fig. 6)

99

is recorded between equilibrated samples at 4 and 4.5 GPa, signifying a very large temperature difference over only 17 km depth. To explain this observation a change in mode of heat transport is implied. A likely explanation for this observation may be found in the coincidence with the ductility threshold of olivine (~1000 °C, Passchier and Trouw, 2005) changing the mode of heat transport from advection (deformation) and conduction to conduction only in the shallower part of the lithosphere. This conclusion is in accordance with the petrography of the Venetian samples that record an undeformed Low-T lithosphere above 130 km depth (~1000 °C in steady state geotherm: Fig. 6) overlying a variably deformed High-T lithosphere that represents hot material introduced prior to kimberlite activity. The elevation of temperatures at depths greater than 130 km suggests that this event resulted in a >200 °C rise in temperature in the lower ~100 km of the SCLM. A lack of knowledge on the breadth over which the thermal anomaly was produced hinders modelling the time of the formation of the thermal anomaly. The time scale for thermal diffusion over distances characterising the thermally conductive lithosphere implies relaxation times of hundreds of Ma for major mantle heating events (e.g., Artemieva and Mooney, 2001; Bell et al., 2003; Nyblade and Sleep, 2003). Significantly, today there is no regional negative seismic anomaly beneath Venetia (Chevrot and Zao, 2007; Fouch et al., 2004; Li and Burke, 2006; Priestley et al., 2006), strongly suggesting that any thermal anomaly formed prior to kimberlite eruption at Venetia must have been of a local scale and is no longer present. Thermal diffusion calculations imply that the width of the 100 km thick anomaly cannot be greater than ~10 km or it would have survived for 500 Ma and hence be resolvable in regional seismic studies today (e.g., Artemieva and Mooney, 2001). The formation of such a narrow thermal upwelling within the lower lithosphere may be explained by similar mechanisms to (proto-) kimberlite melt genesis and migration through the mantle involving discrete melt pockets separated by peridotite wallrock. The peridotites would ductilly deform but inefficiently mix with rising melt pockets due to the large density contrast (cf. Grégoire et al., 2006). The presented Venetian xenolith P–T data suggest that the input of melt, accompanied by metasomatism and deformation continues into the lower part of the lithosphere up to the depth where olivine ductility is inhibited (~1000 °C at ~130 km depth or ~4 GPa, see Fig. 6). Tomographic modelling of the Kaapvaal Craton, LMB and southern Zimbabwe Craton filtered for the expected lithospheric temperature variations allows for isolation of a composition dependant signal (Artemieva, 2009; Griffin et al., 2009). Such modelling indicates that at depths of >150 km, kimberlite occurrences generally correlate with compositionally slower lithosphere than the typical surrounding SCLM. Based on the xenolith population, sheared lherzolites (relatively Fe-rich) appear to make up a considerable proportion of the deeper lithosphere beneath Venetia, which could (in part) explain these lower seismic velocities (e.g., Jordan, 1978; Lee, 2003) as already proposed by Griffin et al. (2009). 4.2. Magmaphile element extraction and kimberlite contamination Melt depletion of Low-T lherzolites and harzburgites at Venetia occurred in the Archaean (Carlson et al., 1999). Whole rock major element compositions and olivine (Ol) Mg# reflect a high degree of melting for Low-T xenoliths (Ol) Mg# 92 to 93.5; an average of 92.8 reflects melting to 25–50% with an average of 40% following the approach of Bernstein et al. (2007). Similarly High-T harzburgites reflect melting to 20–50% with an average of 30% (Ol Mg# 91.5–93.5 average 92.3) but have a mineralogy reflecting addition of magmaphile elements (Ca–Fe–Al) as commonly observed in both the Kaapvaal and Zimbabwe Cratons (e.g., Boyd, 1989; Pearson and Wittig, 2008; Smith et al., 2009). Olivine Mg# in sheared High-T lherzolite is lower (91–92, average 91.5 reflecting melting of 15–30% with an average of 20%) comparable to compositions found in High-T lherzolite

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

94

60

Average DI

93

Zimbabwe Kaapvaal Low-T

40

92

25

91

15

Kaapvaal High-T

90

89 100

90

80

70

% Melt Extraction

xenoliths throughout the Kaapvaal Craton (e.g. Boyd and Mertzman, 1987). These degrees of melt-depletion can be interpreted as minima because melt-metasomatism, Fe-rich relative to the bulk xenoliths, would lower Mg# and thus the inferred melt depletion. Elevated whole rock concentrations of Na2O, P2O5 and K2O (mobile and highly incompatible elements) and highly variable concentrations of TiO2 in Low-T lherzolite indicate b 1–5% contamination of the samples by the host kimberlite. These geochemical signatures are not reflected in the compositions of cores and rims of pristine minerals, therefore individual minerals retain a chemical signature representative of the mantle prior to kimberlite entrainment. Below, the geochemical characteristics of regional Low-T mantle are examined to establish the extent of melt depletion they have undergone and if there are any characteristics that can be used to distinguish the SCLM beneath the Kaapvaal and Zimbabwe Cratons.

Mg# in olivine

100

10

60

50

40

Modal % olivine 4.3. Post melt-depletion enrichment and craton characterisation 4.3.1. Distinction of Kaapvaal and Zimbabwe Low-T mantle In their review, Pearson and Wittig (2008) argued that the similarity in extent of melt depletion of all Archaean cratons does not allow discrimination of different cratonic areas based on Mg#. This conclusion appears valid for the Kaapvaal and Zimbabwean SCLM with olivine Mg# indicating that the majority of Low-T peridotites have undergone between 25 and 50% melt extraction (Bernstein et al., 2007; Fig. 7). Low-T peridotites from Venetia have olivine Mg#s that are indistinguishable from Kaapvaal and Zimbabwean SCLM (Fig. 7). Despite comparable olivine compositions, however, the majority of SCLM xenolith samples from beneath Kaapvaal and Zimbabwean Cratons have distinct modal compositions (Fig. 7). Boyd (1989) was the first to highlight that the Kaapvaal SCLM has anomalously low olivine abundances compared to the Mg# in olivine that is reflected as relatively Si-rich peridotites with orthopyroxene (opx) contents higher than inferred from melting models. Numerous models have been proposed to explain the orthopyroxene enrichment (e.g., see review in Wasch et al., 2009). Currently the most widely accepted explanation is for metasomatic Si addition after initial extensive melt extraction, which ultimately leads to reactions that produced orthopyroxene (e.g., Kelemen et al., 1998; Pearson and Wittig, 2008). The Low-T samples from Venetia appear to be typical of Low-T xenoliths retrieved from the Kaapvaal Craton (Boyd, 1989; Boyd and Mertzman, 1987; Simon et al., 2003) in that they are characterised by relatively low olivine modes 50–85% and high orthopyroxene (up to 40%; Fig. 7). Extensive Si addition is not observed in Low-T xenoliths from the Zimbabwe Craton in which orthopyroxene modes never exceed 10% (Smith et al., 2009). Zimbabwean xenoliths are characterised by olivine contents that are generally above 80% (Fig. 7). Therefore the Low-T SCLM and some of the High-T harzburgites are Kaapvaal-like and distinct from the Zimbabwe SCLM. 4.3.2. Si-metasomatism and mineral exsolution at Low-T The Mg#s in olivine in Low-T lherzolites correspond to melt extraction experiments that would leave melt depleted harzburgite or dunite residua (e.g., Bernstein et al., 2007; Walter, 1998). The resultant Cr-spinel dunites effectively have a major element chemistry consisting of SiO2, MgO, Cr2O3 and NiO, the most compatible elements during melt extraction (Walter, 2004). Therefore, the presence of Ca–Al bearing minerals such as clinopyroxene and garnet but also orthopyroxene in the LMB xenoliths, requires extensive metasomatic addition of Si, Ca and Al after melt depletion. The widespread presence of clinopyroxene in Kaapvaal SCLM, the earliest mineral to be exhausted in melting of such peridotitic samples, has been found to originate from metasomatism related to regional kimberlite activity (e.g., Grégoire et al., 2003; Griffin et al., 2003; Simon et al., 2003, 2007). The major element concentrations in clinopyroxene in the Low-T samples, however, are not in accordance with this interpretation. Clinopyroxene compositions

Low-T lherzolite

High-T Lherzolite

Harzburgite

Fig. 7. Calculated modal abundance of olivine plotted against the Mg# in olivine. The dashed black curve indicates the ‘Oceanic Melting Trend’ and the grey field defined for Kaapvaal Craton Low-T peridotite xenoliths after Boyd (1989), open field defines ‘High-T Southern Kaapvaal’ after Boyd and Mertzman (1987). Vertical dashed line indicates average Mg# of Venetian olivine diamond inclusions from Viljoen et al. (1999). The field defining the Zimbabwe Craton is based on the most common Low-T xenolith compositions from the Murowa kimberlite, xenoliths shown as the small black dots (n = 20, Smith et al., 2009; C. Smith Pers. Comm.). The right hand scale is the approximate degree of melt extraction and is estimated from melt experiment compilation after Bernstein et al. (2007).

have low concentrations of TiO2 (b0.05 wt.% with one exception of 0.09 wt.%) and Na2O (b2 wt.% with one exception of 2.1 wt.%) as well as high Mg# that are systematically higher than orthopyroxene. These compositions indicate Fe–Mg equilibration with orthopyroxene rather than with coexisting olivine or with an incompatible element enriched melt (Dawson, 2004). Furthermore, textural relationships with orthopyroxene (see Fig. 3) suggest an origin for clinopyroxene as an exsolved phase from a former High-T orthopyroxene (Canil, 1991; Cox et al., 1987; Saltzer et al., 2001). Based on textural observations and the low overall TiO2 content and the formation of fine-grained aggregates of coexisting garnet + clinopyroxene ± orthopyroxene ± spinel, a similar exsolution origin for garnet and (part of the) spinel is likely (see Fig. 3). The mineralogy of the Low-T lherzolites can therefore be explained by a single metasomatic event involving the introduction of Si with minor Al and Ca. The net effect of the metasomatism has been to enrich the depleted lithospheric mantle in a High-T equivalent of orthopyroxene (e.g. Pearson and Wittig, 2008). After cooling and/or decompression of the peridotites, clinopyroxene, garnet and spinel would exsolve and recrystallise. As there is no apparent major element and mineralogical difference between the b130 km lithospheric mantle beneath the LMB and that of the Kaapvaal Craton other than mineralogies associated with late kimberlite-related metasomatism, a similar mechanism is envisaged to have formed all these silica-enriched lithologies. We conclude that the SCLM beneath both the upper section of the LMB and the Kaapvaal Craton have many features in common and were formed by the same process and likely as part of a single craton. A large proportion of the Kaapvaal SCLM has suffered from subsequent local metasomatic overprinting related to Mesozoic tectono-magmatic events (Erlank et al., 1987; Kobussen et al., 2008, 2009; Simon et al., 2003, 2007). Notably the detailed petrographic studies reported here suggest that the Venetian SCLM has undergone less Mesozoic metasomatic overprinting than recorded by the Kaapvaal SCLM, consistent with the older emplacement age of the Venetia kimberlite cluster. 4.3.3. The High-T lithosphere At Venetia the High-T lithosphere forms a domain of mixed harzburgites and lherzolites and these xenoliths record a higher geothermal gradient than the Low-T samples. The majority (~ 80%) of the High-T xenoliths equilibrated at diamond stability conditions

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

(see Figs. 5 and 8). Hence a comparison can be made between mineral inclusions in diamonds that represent High-T lithosphere at the time of diamond formation > 2 Ga (Richardson and Shirey, 2008; Richardson et al., 2009) and High-T xenoliths representative of the mantle at the time of kimberlite eruption. From these observations the lithosphere during the main diamond forming period may be simplistically explained by a single stage of ~ 50% melt extraction leaving a garnet-poor harzburgite. This explanation, however, does not account for the compositions of the deformed High-T xenoliths. The Mg#s of olivines from the High-T xenoliths are lower than those of olivine inclusions in diamond. This is a common feature of sheared High-T lherzolites found in kimberlite (e.g. Boyd and Mertzman, 1987; Griffin et al., 1989, 1996; Nixon and Boyd, 1973; Smith and Boyd, 1987, 1989). Fig. 8 expands on this observation by comparing olivine Mg#s to depth of equilibration of the xenolith samples. Superimposed on Fig. 8 is a histogram of olivine diamond inclusion Mg# from Venetia (Viljoen et al., 1999). The histogram is scaled so it lies below a depth of 150 km, the shallowest depth of diamond stability beneath Venetia. Olivines from garnet harzburgites and High-T lherzolites from depths within the diamond stability field generally have lower Mg#s than olivine in diamond inclusions. Moreover, the most melt depleted xenoliths (highest olivine Mg#) occur at depths above the diamond stability field and are thus not directly related to the minerals included within diamonds. If the diamond inclusions are assumed to be representative of lithosphere in which the diamond formed, then the data presented here indicate that the lithosphere has undergone enrichment in Fe and presumably other magmaphile elements since the formation of the majority of the diamonds. Variable enrichment of the lithosphere in Ca and Al at depths > 130 km is evident by the presence of predominantly calciumsaturated garnet and clinopyroxene in the mantle xenoliths derived from depths at which diamond is stable. Only two harzburgites in

Mg# in olivine 90 0

91

92

93

94

95

100

Graphite

60

Diamond

50

150

40 30

200 Distribution in diamond inclusions (n=85)

20 10

% Diamond Inclusions

Depth [km]

50

0

250 Low-T Lherzolite

Garnet Harzburgite

High-T Lherzolite

Fig. 8. Calculated depth against Mg# in olivine from Venetian xenoliths. Barometer and thermometer as in Fig. 5, open symbols from Stiefenhofer et al. (1999) and Hin et al. (2009), diamond stability depth from intersection between diamond–graphite boundary and Kalahari geotherm as in Fig. 5 and therefore a minimum estimate of the depth relative to steady state geotherm in Fig. 6. Also plotted is a smoothed histogram of olivine Mg# from Venetian diamond inclusions after Viljoen et al. (1999), with the percentage of olivine in bins with a width of 1 Mg# indicated by vertical lines.

101

the current study contain clearly Ca-undersaturated G10 garnet (terminology after Dawson and Stephens, 1975; see also Fig. 4). This finding contrasts with the mainly Ca-undersaturated G10, garnets and chromite in mineral inclusions in diamonds from Venetia (Fig. 4; Viljoen et al., 1999) and the Zimbabwe Craton (Bulanova et al., 2012; Smith et al., 2009). It is likely that Ca-saturated garnet was formed through re-fertilisation of the lower lithosphere by asthenospheric melts after diamond formation (e.g., Griffin et al., 1999). Garnet harzburgites have magmaphile element concentrations intermediate between those of the diamond inclusions and High-T lherzolites and may represent cratonic lithosphere that has been metasomatised to a lesser extent than the lherzolites. Distinctive chemical properties of the High-T lithosphere are thus obscured in the xenolith record (see also Gaul et al., 2000). 4.3.4. Characteristics of High-T lithosphere enrichment The High-T lithologies are characterised by a fine-grained olivine fabric, indicating that deformation of the samples occurred up until the moment of kimberlite eruption or these structures would be (partially) overprinted by static recrystallisation. Occurrence of deformation in the most refertilised lithologies may be explained if trace amounts of water accompanied metasomatism, weakening olivine rheology (e.g., Mei and Kohlstedt, 2000). Petrographic observations are useful in helping to constrain the relationship between deformation and metasomatism. A euhedral orthopyroxene crystal encapsuled by a rounded garnet porphyroclast in High-T lherzolite AT1361 has identical major element chemistry to orthopyroxene porphyroclasts within the olivine matrix (Fig. 3c). This observation establishes that at least the latest stage of deformation was isochemical and postdates any metasomatic chemical alteration. High-T lherzolite and harzburgite samples studied here lack major element zoning in garnet and other constituent minerals. Major element zoning in garnet is often described as a typical feature of High-T lithologies caused by interaction with the kimberlite shortly prior to eruption (e.g. Griffin et al., 1989; Griffin et al., 1996; Griffin et al., 1999; Smith & Boyd, 1987, 1989; Smith et al., 1993). The absence of major element zoning in the Venetian samples demonstrates that late kimberlite interaction is not a viable mechanism to explain refertilisation by magmaphile elements. Enrichment must have occurred previous to kimberlite eruption, allowing the rock to reach full chemical equilibrium before transport to the surface. Given diffusion speeds for Ca, Mg, Fe and Mn in garnet (Carlson, 2006) of >10− 16 2 −1 m s at 1200 °C (the temperature of the coolest High-T lherzolites), the presence of unzoned garnets up to 1 cm in size constrains the time of metasomatic introduction of an Fe–Ca–Al component to the High-T lherzolites to more than several kyr prior to kimberlite eruption. The minimal age of diamond formation of >2.0 Ga (Richardson et al., 2009) implies that metasomatism of the High-T lithosphere occurred between 2.0 Ga and kimberlite eruption at 533 Ma. We propose that the metasomatism of deep lithologies is coupled to their positive temperature perturbation (Fig. 6) and deformation of the lower lithosphere. The variable metasomatism by asthenospheric melts is suggested to have occurred in a contact zone (e.g. Mercier, 1979) with a diaper of rising melt (Grégoire et al., 2006). 4.4. Origin of the Limpopo SCLM Based on a comparison of garnet compositions in heavy mineral concentrates (effectively minerals originating as xenocrysts from xenoliths) from the Venetia and Zimbabwean kimberlites, Smith et al. (2009) postulated that the LMB represents a thin-skinned deformed crust underlain by SCLM similar to that of the southern Zimbabwe Craton. This conclusion was based on the high abundance of G10 garnet in the concentrate from both these regions and lower G10 abundances in heavy mineral concentrate derived from the northern Kaapvaal Craton. The current study of peridotite xenoliths establishes that the majority of G10 garnet at Venetia must originate from disaggregated

102

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

High-T harzburgites, as G10 garnet compositions were not found in the Low-T peridotites or, as expected, in the High-T deformed lherzolites (Fig. 4). They represent the High-T lithosphere and have compositions that contrast strongly with the Si-rich Low-T lithosphere (and rare Si-rich High-T harzburgites) that is indistinguishable from typical Kaapvaal lithosphere (Fig. 7). These interpretations are not necessarily contradictory. The LMB acted as the overriding plate during the Kaapvaal–Zimbabwe collision (Durrheim et al., 1992), which could have resulted in a Kaapvaal related Low-T lithosphere underlain by a Zimbabwe related High-T lithosphere. The significance of the garnet concentrate data, however, is questionable. As argued in Sections 4.3.3 and 4.3.4, kimberlite borne xenoliths from the High-T lithosphere commonly reflect variable but pervasive refertilisation related to (proto) kimberlitic activity. The same history is expected for garnet concentrates. Variable major element re-enrichment of a depleted lithology would be dependent on the local intensity of preceding kimberlite activity rather than reflecting compositions characteristic of an entire craton. To avoid the potential effects of recent re-enrichment, minerals included in diamond are used for comparison. Diamonds act as time capsules and effectively protect inclusions from younger melting and metasomatic events. Available age constraints on diamond formation in the region suggest most formed during the Archaean and early Proterozoic (Gurney et al., 2010). Major element chemistry of garnet included in diamonds from Venetia (Viljoen et al., 1999), Kaapvaal (Stachel and Harris, 2008) and Zimbabwe (Bulanova et al., 2012; Smith et al., 2009) establishes that mineral inclusions are comparable, although published data from the Zimbabwe Craton is admittedly currently limited. Comparable inclusion compositions in the three regions do not allow distinction between the deepest parts of the cratons. This observation further implies that the difference in garnet concentrates between the LMB–Zimbabwe region and the Kaapvaal Craton is a post diamond formation characteristic, i.e. post Mid Proterozoic. We conclude that the Low-T lithosphere beneath Venetia can be positively identified to be of a Kaapvaal Craton affinity and the Si-enriched High-T harzburgites (Fig. 7) imply that the entire lithosphere is most likely of Kaapvaal origin. 5. Conclusions The Venetia kimberlite contains mantle xenoliths derived from between 50 km and 235 km depth, spanning the entire thickness of sub-continental lithospheric mantle beneath the central zone of the Limpopo Mobile Belt. The LMB central zone has an apparently stratified SCLM with an upper 50–130 km layer formed by up to 50% melt depletion and subsequent Si metasomatism. A metasomatic event where olivine reacted to a High-T orthopyroxene equivalent followed by cpx–garnet-(spinel) exsolution can explain the observed mineralogy and appears the major metasomatic event affecting the shallower SCLM beneath the LMB other than minor kimberlite-related metasomatism. This Low-T lithosphere is indistinguishable from that of the Kaapvaal Craton. The deeper SCLM from 130 to 235 km lacks pervasive silica enrichment and is characterised by highly melt depleted (~ 50%) diamond inclusions with a high abundance of sub-calcic garnet (Viljoen et al., 1999) similar to the SCLM of both the Kaapvaal (Stachel and Harris, 2008) and the Zimbabwe Craton (Smith et al., 2009). While the Low-T lithosphere can be positively identified to be of a Kaapvaal Craton affinity, data on the High-T lithosphere are less conclusive. The presence of rare, mildly silica enriched High-T harzburgites suggests a Kaapvaal origin for the entire lithosphere beneath the LMB is most likely. Consequently, the LMB SCLM formed as part of the Kaapvaal Craton or accreted to the Kaapvaal Craton prior to collision with the Zimbabwe Craton at 2.65 Ga. High-T xenoliths derived from 130 to 235 km, record a lithosphere enriched in Fe–Ca–Al relative to diamond inclusion data. These sheared harzburgites and lherzolites have equilibrated to anomalously high

temperatures and record a non steady state geotherm at the time of kimberlite eruption. In contrast, mineral inclusions in diamond have retained cooler equilibration conditions, conforming to steady-state geotherm conditions in the SCLM above 130 km. The lack of mineral zonation in xenoliths from below 130 km indicates that, although the disequilibrium geotherm does not represent an ancient stable feature, it is not simply caused by abrupt recent (immediately preceding emplacement) changes of chemical environment such as kimberlite magmatism. The temperature perturbation represents local metasomatism by asthenosphere-derived melt and subsequent deformation, presumably during diapiric rise of a proto-kimberlitic melt pocket (Grégoire et al., 2006). This metasomatic event was followed by kimberlite eruption and seems to be a common feature for many kimberlite-borne xenoliths. Regional tectono-magmatic events such as the emplacement of LIP and continental rifting are not responsible for this observed heating and metasomatism of the High-T lithosphere. Although no thermal perturbation of the deep lithosphere beneath Venetia is visible today, the rise of melt pockets in the High-T lithosphere may provide a cause for the low seismic velocities found beneath regions with abundant kimberlite clusters (Artemieva, 2009; Griffin et al., 2009). Based on the complex tectono-magmatic evolution of the LMB region, the best constraints on ancient craton forming processes will be preserved within Low-T peridotites and diamond inclusions. Both the Low-T peridotites and minerals included in diamond have remained largely unaffected by the thermal and chemical modification immediately preceding kimberlite eruption. Furthermore, we concur with previous workers that the earliest generation of kimberlites within a region is more likely to retain ancient geochemical signatures unaffected by kimberlite-related magmatism (Kobussen et al., 2008, 2009; Simon et al., 2007). Acknowledgements We would like to thank Sonja Aulbach and an anonymous reviewer for their constructive reviews that helped to significantly improve the manuscript and Andrew Kerr for efficient editorial handling. Paul Smit (i.m.), Wynanda Koot, Bouk Laçet, Roel van Elsas, Elodie Tronche and Wim Lustenhouwer are thanked for their assistance with laboratory procedures. DeBeers and Venetian diamond mine staff, Mark Tait, Johann Stiefenhofer, Paolo Nimis, Cees Passchier, Brett Davidheiser, Anouk Borst, Remco Hin, Ineke Wijbrans and Roula Dambrink are thanked for their assistance in the field and for fruitful discussions. Johann Stiefenhofer and Chris Smith are thanked for supplying unpublished data. Partial funding for the field work was provided by grants from the Dr. Schürmann Foundation and the VU University Amsterdam. Appendix A. Supplementary data Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.lithos.2013.02.019. References Allsopp, H.L., Smith, C.B., Seggie, A.G., Skinner, E.M.W., Colgan, E.A., 1995. The emplacement age and geochemical character of the Venetia kimberlite bodies, Limpopo Belt, northern Transvaal. South African Journal of Geology 98, 239–244. Artemieva, I.M., 2009. The continental lithosphere: reconciling thermal, seismic, and petrologic data. Lithos 109, 23–46. Artemieva, I.M., Mooney, W.D., 2001. Thermal thickness and evolution of Precambrian lithosphere — a global study. Journal of Geophysical Research, Planets 106, 16387–16414. Aulbach, S., Stachel, T., Viljoen, K.S., Brey, G.P., Harris, J.P., 2002. Eclogitic and Websteritic diamond sources beneath the Limpopo Belt — is slab melting the link? Contributions to Mineralogy and Petrology 143, 56–70. Barton, J.M., Gerya, T.V., 2003. Mylonitization and decomposition of garnet: evidence for rapid deformation and entrainment of mantle garnet-harzburgite by kimberlite magma, K1 Pipe, Venetia Mine, South Africa. South African Journal of Geology 106, 231–242. Barton, J.M., van Reenen, D.D., 1992. When was the Limpopo Orogeny? Precambrian Research 55, 7–16.

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104 Bell, D.R., Schmitz, M.D., Janney, P.E., 2003. Mesozoic thermal evolution of the southern African mantle lithosphere. Lithos 71, 273–287. Bernstein, S., Kelemen, P.B., Hanghøj, K., 2007. Consistent olivine Mg# in cratonic mantle reflects Archean mantle melting to the exhaustion of orthopyroxene. Geology 35, 459–462. Boyd, F.R., 1989. Compositional distinction between oceanic and cratonic lithosphere. Earth and Planetary Science Letters 96, 15–26. Boyd, F.R., Mertzman, S.A., 1987. Composition and structure of the Kaapvaal lithosphere, Southern Africa. In: Mysen, B.O. (Ed.), Magmatic Processes: Physicochemical Principles: The Geochemical Society, Special Publication, 1, pp. 13–24. Bulanova, G.P., Marks, A., Smith, C.B., Kohn, S.C., Walter, M.J., Gaillou, E., Shirey, S.B., Trautman, R., Griffin, B.J., 2012. Diamonds from Sese and Murowa Kimberlites (Zimbabwe) — evidence or extreme peridotitic lithosphere depletion and Timetasomatism. Proceedings 10th International Kimberlite Conference Abstract Reg. No. 10IKC-218. Canil, D., 1991. Experimental evidence for the exsolution of cratonic peridotite from high-temperature harzburgite. Earth and Planetary Science Letters 106, 64–72. Carlson, W.D., 2006. Rates of Fe, Mg, Mn, and Ca diffusion in garnet. American Mineralogist 91, 1–11. Carlson, R.W., Pearson, D.G., Boyd, F.R., Shirey, S.B., Irvine, G., Menzies, A.H., Gurney, J.J., 1999. Re-Os systematics of lithospheric peridotites: implications for lithosphere formation and preservation. Proceedings of the 8th International Kimberlite Conference, the J. Barry Hawthorne Volume. J. L, pp. 99–108. Carlson, R.W., Pearson, D.G., James, D.E., 2005. Physical, chemical, and chronological characteristics of continental mantle. Reviews of Geophysics 43, 1–24. Cartigny, P., Farquhar, J., Thomassot, E., Harris, J.W., Wing, B., Masterson, A., McKeegan, K., Stachel, T., 2009. A mantle origin for Paleoarchean peridotitic diamonds from the Panda kimberlite, Slave Craton: evidence from 13C-, 15N- and 33,34S-stable isotope systematic. Lithos 112 (Supplement 2), 852–864. Chevrot, S., Zao, L., 2007. Multiscale finite frequency Rayleigh wave tomography of the Kaapvaal craton. Geophysical Journal International 169, 201–215. Cox, K.G., Smith, M.R., Beswetherick, S., 1987. Textural studies of garnet lherzolites: evidence of exsolution origin from high-temperature harzburgites. In: Nixon, P.H. (Ed.), Mantle Xenoliths. John Wiley and Sons, pp. 537–550. Dawson, J.B., 2004. A fertile harzburgite–garnet lherzolite transition: possible inferences for the roles of strain and metasomatism in upper mantle peridotites. Lithos 77, 553–569. Dawson, J.B., Stephens, W.E., 1975. Statistical classification of garnets from kimberlite and associated xenoliths. Journal of Geology 589–607. de Wit, M.J., de Ronde, C.E., Tredoux, M., Roering, C., Hart, R.J., Armstrong, R.A., Green, R.W.E., Pederby, E., Hart, R.A., 1992. Formation of an Archaean continent. Nature 357, 553–562. Deines, P., Viljoen, F., Harris, J.W., 2001. Implications of the carbon isotope and mineral inclusion record for the formation of diamonds in the mantle underlying a mobile belt: Venetia, South Africa. Geochimica et Cosmochimica Acta 65, 813–838. Du Toit, M.C., van Reenen, D.D., Roering, C., 1983. Some aspects of the geology, structure and metamorphism of the Southern Marginal Zone of the Limpopo Metamorphic Complex. Special Publications Geological Society South Africa 8, 89–102. Durrheim, R.J., Barker, W.H., Green, R.W.E., 1992. Seismic studies in the Limpopo Belt. Precambrian Research 55, 187–200. Eaton, D.W., Darbyshire, F., Evans, R.L., Grütter, H., Jones, A.G., Yuan, X., 2009. The elusive lithosphere–asthenosphere boundary (LAB) beneath cratons. Lithos 109, 1–22. Erlank, A.J., Waters, F.G., Hawkesworth, C.J., Haggerty, S.E., Allsopp, H.L., Rickard, R.S., Menzies, M.A., 1987. Evidence for mantle metasomatism in peridotite nodules from the Bultfontein Floors, Kimberley, South Africa. In: Menzies, M.A., Hawkesworth, C.J. (Eds.), Mantle Metasomatism. Academic Press, London, pp. 221–311. Evans, B., Renner, J., Hirth, G., 2001. A few remarks on the kinetics of static grain growth in rocks. International journal of earth sciences 90, 88. Field, M., Stiefenhofer, J., Robey, J., Kurszlaukis, S., 2008. Kimberlite-hosted diamond deposits of southern Africa: a review. Ore Geology Reviews 34, 33–75. Fouch, M.J., James, D.E., VanDecar, J.C., van der Lee, S., Group, K.S., 2004. Mantle seismic structure beneath the Kaapvaal and Zimbabwe Cratons. South African Journal of Geology 107, 33–44. Gaul, O.F., Griffin, W.L., O’Reilly, S.Y., Pearson, N.J., 2000. Mapping olivine composition in the lithospheric mantle. Earth and Planetary Science Letters 182, 223–235. Gerdes, A., Zeh, A., 2009. Zircon formation versus zircon alteration — new insights from combined U–Pb and Lu–Hf in-situ LA-ICP-MS analyses, and consequences for the interpretation of Archean zircon from the Central Zone of the Limpopo Belt. Chemical Geology 261, 230–243. Grégoire, M., Bell, D.R., Le Roex, A.P., 2003. Garnet lherzolites from the Kaapvaal Craton (South Africa): trace element evidence for a metasomatic history. Journal of Petrology 44, 629–657. Grégoire, M., Rabinowicz, M., Janse, A.J.A., 2006. Mantle mush compaction: a key to understand the mechanisms of concentration of kimberlite melts and initiation of swarms of kimberlite dykes. Journal of Petrology 47, 631–646. Griffin, W.L., Smith, D., Boyd, F.R., Cousens, D.R., Ryan, C.G., Sie, S.H., Suter, G.F., 1989. Trace-element zoning in garnets from sheared mantle xenoliths. Geochimica et Cosmochimica Acta 53, 561–567. Griffin, W.L., Smith, D., Ryan, C.G., O'Reilly, S.Y., Win, T.T., 1996. Trace-element zoning in mantle minerals; metasomatism and thermal events in the upper mantle. The Canadian Mineralogist 34, 1179–1193. Griffin, W.L., Shee, S.R., Ryan, C.G., Win, T.T., Wyatt, B.A., 1999. Harzburgite to lherzolite and back again: metasomatic processes in ultramafic xenoliths from the Wesselton kimberlite, Kimberley, South Africa. Contributions to Mineralogy and Petrology 134, 232–250. Griffin, W.L., O'Reilly, S.Y., Natapov, L.M., Ryan, C.G., 2003. The evolution of lithospheric mantle beneath the Kalahari Craton and its margins. Lithos 71, 215–241.

103

Griffin, W.L., O'Reilly, S.Y., Afonso, J.C., Begg, G.C., 2009. The composition and evolution of lithospheric mantle: a re-evaluation and its tectonic implications. Journal of Petrology 50, 1185–1204. Grütter, H.S., Gurney, J.J., Menzies, A.H., Winter, F., 2004. An updated classification scheme for mantle-derived garnet, for use by diamond explorers. Lithos 77, 841–857. Gurney, J.J., 1984. A correlation between garnets and diamonds. In: Glover, J.E., Harris, P.G. (Eds.), Kimberlite Occurrence and Origins: a Basis for Conceptual Models in Exploration. : Publication, 8. Geology Department and University Extension, University of Western Australia, pp. 143–166. Gurney, J.J., Helmstaedt, H.H., Richardson, S.H., Shirey, S.B., 2010. Diamonds through time. Economic Geology 105, 689–712. Harte, B., 1977. Rock nomenclature with particular relation to deformation and recrystallisation textures in olivine-bearing xenoliths. Journal of Geology 85, 279–288. Hin, R.C., 2008. Formation and modification of the Venetian mantle lithosphere. A major and trace element study of South-African mantle xenoliths, M.Sc. Thesis VU University Amsterdam. Hin, R.C., Nebel, O., Mason, P.R.D., van Westrenen, W., Davies, G.R., 2009. Formation and temporal evolution of the Kalahari sub-cratonic lithospheric mantle: constraints from Venetia xenoliths, South Africa. Lithos 112 (Supplement 2), 1069–1082. James, D.E., Fouch, M.J., vanDecar, J.C., Van der Lee, S., Group, K.S., 2001. Tectospheric structure beneath southern Africa. Geophysical Research Letters 28 (13), 2485. Jordan, T.H., 1978. Composition and development of the continental tectosphere. Nature 274, 544–548. Karato, S., 1989. Grain growth kinetics in olivine aggregates. Tectonophysics 168, 255–273. Kelemen, P.B., Hart, S.R., Bernstein, S., 1998. Silica enrichment in the continental upper mantle via melt/rock reaction. Earth and Planetary Science Letters 164 (1–2), 387–406. Kennedy, C.S., Kennedy, G.C., 1976. The equilibrium boundary between graphite and diamond. Journal of Geophysical Research 81, 2467–2470. Klemme, S., 2004. The influence of Cr on the garnet–spinel transition in the Earth's mantle: experiments in the system MgO–Cr2O3–SiO2 and thermodynamic modelling. Lithos 77, 639–646. Kobussen, A.F., Griffin, W.L., O'Reilly, S.Y., Shee, S.R., 2008. Ghosts of lithospheres past: imaging an evolving lithospheric mantle in southern Africa. Geology 36, 515–518. Kobussen, A.F., Griffin, W.L., O'Reilly, S.Y., 2009. Cretaceous thermo-chemical modification of the Kaapvaal cratonic lithosphere, South Africa. Lithos 112 (Supplement 2), 886–895. Kopylova, M.G., Gurney, J.J., Daniels, R.L.M., 1997. Mineral inclusions in diamonds from the River Ranch kimberlite, Zimbabwe. Contributions to Mineralogy and Petrology 129, 366–384. Lee, C.-T.A., 2003. Compositional variation of density and seismic velocities in natural peridotites at STP conditions: implications for seismic imaging of compositional heterogeneities in the upper mantle. Journal of Geophysical Research 108, 2441. Li, A., Burke, K., 2006. Upper mantle structure of southern Africa from Rayleigh wave tomography. Journal of Geophysical Research 111, B10303. Marsh, J.S., Hooper, P.R., Rehacek, J., Duncan, R.A., Duncan, A.R., 1997. Stratigraphy and age of Karoo basalts of Lesotho and implications for correlations within the Karoo Igneous Province. American Geophysical Union, Washington, DC (ETATS-UNIS, 443 pp.). Mather, K.A., Pearson, D.G., McKenzie, D., Kjarsgaard, B.A., Priestley, K., 2011. Constraints on the depth and thermal history of cratonic lithosphere from peridotite xenoliths, xenocrysts and seismology. Lithos 125, 729–742. McCourt, S., Armstrong, R.A., 1998. SHRIMP U–Pb zircon geochronology of granites from the central zone, Limpopo Belt, southern Africa; implications for the age of the Limpopo Orogeny. South African Journal of Geology 101, 329–338. Mei, S., Kohlstedt, D.L., 2000. Influence of water on plastic deformation of olivine aggregates 1. Diffusion creep regime. Journal of Geophysical Research 105, 21457–21469. Mercier, J.C., 1979. Peridotite xenoliths and the dynamics of kimberlite intrusions. In: Meyer, H.O.A., Boyd, F.R. (Eds.), The Mantle Sample: Inclusions in Kimberlites and Other Volcanics. AGU, Washington, DC, pp. 197–212. Millonig, L., Zeh, A., Gerdes, A., Klemd, R., Barton, J.M., 2010. Decompressional heating of the Mahalapye Complex (Limpopo Belt, Botswana): a response to Palaeoproterozoic magmatic underplating? Journal of Petrology 51, 703–729. Moore, A.E., Lock, N.P., 2001. The origin of mantle-derived megacrysts and sheared peridotites-evidence from kimberlites in the northern Lesotho-Orange Free State (South Africa) and Botswana pipe clusters. South African Journal of Geology 104, 23–38. Nguuri, T.K., Gore, J., James, D.E., Webb, S.J., Wright, C., Zengeni, T.G., Gwavava, O., Snoke, J.A., Kaapvaal Seismic Group, 2001. Crustal structure beneath southern Africa and its implications for the formation and evolution of the Kaapvaal and Zimbabwe cratons. Geophysical Research Letters 28, 2501–2504. Nickel, K.G., Green, D.H., 1985. Empirical geothermobarometry for garnet peridotites and implications for the nature of the lithosphere, kimberlites and diamonds. Earth and Planetary Science Letters 73, 158–170. Nimis, P., Grütter, H., 2010. Internally consistent geothermometers for garnet peridotites and pyroxenites. Contributions to Mineralogy and Petrology 159, 411–427. Nimis, P., Taylor, W.R., 2000. Single clinopyroxene thermobarometry for garnet peridotites. Part I. Calibration and testing of a Cr-in-Cpx barometer and an enstatite-in-Cpx thermometer. Contributions to Mineralogy and Petrology 139, 541–554. Nixon, R.H., Boyd, F.R., 1973. Petrogenesis of the granular and sheared ultrabasic nodule suite in kimberlite. In: Nixon, P.H. (Ed.), Lesotho Kimberlites. Cape and Transvaal Printers, South Africa, pp. 48–56. Nyblade, A.A., 1999. Heat flow and the structure of Precambrian lithosphere. In: Hilst, R.D.v.d, McDonough, W.F. (Eds.), Developments in Geotectonics. Elsevier, pp. 81–91. Nyblade, A.A., Sleep, N.H., 2003. Long lasting epeirogenic uplift from mantle plumes and the origin of the Southern African Plateau. Geochemistry, Geophysics, Geosystems 4, 1105. Passchier, C.W., Trouw, R.A.J., 2005. Microtectonics. Springer-Verlag, Berlin Heidelberg 59. Pearson, D.G., Wittig, N., 2008. Formation of Archaean continental lithosphere and its diamonds: the root of the problem. Journal of the Geological Society 165, 895–914.

104

Q.H.A. van der Meer et al. / Lithos 170–171 (2013) 90–104

Pearson, D.G., Holland, H.D., Turekian, K.K., 2003. Mantle samples included in volcanic rocks: xenoliths and diamonds. Treatise on Geochemistry 2, 171–275. Peate, D.W., 1997. The Parana–Etendeka Province. In: Mahoney, J., Coffin, B. (Eds.), Large Igneous Provinces: Continental, Oceanic and Planetary Flood Volcanism: AGU Geophysics Monograph, vol. 100, pp. 217–246. Pollack, H.N., Chapman, D.S., 1977. On the regional variation of heat flow, geotherms, and lithospheric thickness. Tectonophysics 38, 279–296. Priestley, K., McKenzie, D., Debayle, E., 2006. The state of the upper mantle beneath southern Africa. Tectonophysics 416, 101–112. Richardson, S.H., Shirey, S.B., 2008. Continental mantle signature of Bushveld magmas and coeval diamonds. Nature 453, 910–913. Richardson, S.H., Pöml, P.F., Shirey, S.B., Harris, J.W., 2009. Age and origin of peridotitic diamonds from Venetia, Limpopo Belt, Kaapvaal–Zimbabwe craton. Lithos 112 (Supplement 2), 785–792. Rigby, M., Mouri, H., Brandl, G., 2008. A review of the pressure–temperature–time evolution of the Limpopo Belt: constraints for a tectonic model. Journal of African Earth Sciences 50, 120–132. Rudnick, R.L., Nyblade, A.A., 1999. The thickness and heat production of Archaean lithosphere: constraints from xenolith thermobarometry and surface heat flow. In: Fei, Y., Bertka, C.M., Mysen, B.O. (Eds.), Mantle Petrology: Field Observations and High Pressure Experimentation: a Tribute to Francis, R. (Joe) Boyd. The Geochemical Society Special Publications, Houston, TX, pp. 3–12. Saltzer, R.L., Chatterjee, N., Grove, T.L., 2001. The spatial distribution of garnets and pyroxenes in mantle peridotites: pressure–temperature history of peridotites from the Kaapvaal Craton. Journal of Petrology 42, 2215–2229. Schulze, D.J., 1989. Constraints on the abundance of eclogite in the upper mantle. Journal of Geophysical Research 94, 4205–4212. Schulze, D.J., 2003. A classification scheme for mantle-derived garnets in kimberlite: a tool for investigating the mantle and exploring for diamonds. Lithos 71, 195–213. Simon, N.S.C., Irvine, G.J., Davies, G.R., Pearson, D.G., Carlson, R.W., 2003. The origin of garnet and clinopyroxene in “depleted” Kaapvaal peridotites. Lithos 71, 289–322. Simon, N.S.C., Carlson, R.W., Pearson, D.G., Davies, G.R., 2007. The origin and evolution of the Kaapvaal cratonic lithospheric mantle. Journal of Petrology 48, 589–625. Smith, C.B., Allsopp, H.L., Kramers, J.D., Hutchinson, G., Roddick, J.C., 1985. Emplacement ages of Jurassic–Cretaceous South African kimberlites by the Rb–Sr method on phlogopite and whole-rock samples. South African Journal of Geology 88, 249–266. Smith, D., Boyd, F.R., 1987. Compositional heterogeneities in a high temperature lherzolite nodule and implications for mantle processes. In: Nixon, P.H. (Ed.), Mantle Xenoliths. John Wiley and Sons, pp. 551–561. Smith, D., Boyd, F.R., 1989. Compositional heterogeneities in minerals of sheared lherzolite inclusions from African kimberlites. In: Ross, J. (Ed.), Proceedings 4th International Kimberlite conferenceGeol. Soc. Australia Special Publication, Blackwell Scientific Publications, Victoria, Australia, 14, pp. 709–724. Smith, D., Griffin, W.L., Ryan, C.G., 1993. Compositional evolution of high-temperature sheared lherzolite PHN 1611. Geochimica et Cosmochimica Acta 57, 605–613. Smith, C.B., Sims, K., Chimuka, L., Duffin, A., Beard, A.D., Townend, R., 2004. Kimberlite metasomatism at Murowa and Sese pipes, Zimbabwe. In: Mitchell, R.H., Grütter, H.S., Heaman, L.M., Scott Smith, B.H., Stachel, T. (Eds.), Proceedings of the 8th Inter-

national Kimberlite Conference, Volume 1, The C. Roger Clement Volume: Lithos, vol. 76, 1–4, pp. 219–232. Smith, C.B., Pearson, D.G., Bulanova, G.P., Beard, A.D., Carlson, R.W., Wittig, N., Sims, K., Chimuka, L., Muchemwa, E., 2009. Extremely depleted lithospheric mantle and diamonds beneath the southern Zimbabwe Craton. Lithos 112 (Supplement 2), 1120–1132. Stachel, T., Harris, J.W., 2008. The origin of cratonic diamonds-constraints from mineral inclusions. Ore Geology Reviews 34, 5–32. Stiefenhofer, J., Viljoen, K.S., Tainton, K.M., Dobbe, R., Hannweg, G.W., 1999. The petrology of a mantle xenolith suite from Venetia, South Africa. In: Gurney, J.J., Gurney, J.L., Pascoe, M.D., Richardson, S.H. (Eds.), P.H. Nixon Volume Proc. 7th Int. Kimb. Conf. Red Roof Design, Cape Town, pp. 836–845. Streckeisen, A., 1976. To each plutonic rock its proper name. Earth-Science Reviews 12, 1–33. Tait, M.A., Brown, R.J., 2008. Explosive fissure eruption of a large kimberlite pipe: Venetia K01 kimberlite pipe, Limpopo, RSA. Extended Abstract, 9th International Kimberlite Conference, Frankfurt am Main, Germany, August 2008, pp. 1–3. Tait, M.A., Brown, R.J., Mnyama, A., 2006. Internal architecture of the Venetia K1 kimberlite: a new geological model and implications for kimberlite emplacement processes, Venetia Mine, Limpopo RSA. 2006 Kimberlite Emplacement Workshop, Saskatoon, Canada (5 pp.). Thomassot, E., Cartigny, P., Harris, J.W., Lorand, J.P., Rollion-Bard, C., Chaussidon, M., 2009. Metasomatic diamond growth: a multi-isotope study (13C, 15N, 33S, 34S) of sulphide inclusions and their host diamonds from Jwaneng (Botswana). Earth and Planetary Science Letters 282, 79–90. Van Reenen, D.D., Roering, C., Ashwal, L.D., de Wit, M.J., 1992. Regional geological setting of the Limpopo Belt. Precambrian Research 55, 1–5. Viljoen, 2002. An infrared investigation of inclusion-bearing diamonds from the Venetia kimberlite, Northern Province, South Africa: implications for diamonds from craton-margin settings. Contributions to Mineralogy and Petrology 144, 98–108. Viljoen, K.S., Philips, D., Harris, J.W., Robinson, D.N., 1999. Mineral inclusions in diamonds from the Venetia Kimberlites, Northern Province, South Africa. Proceedings of the VII IKC, vol. 2, Red Roof Design, Cape Town, pp. 888–895. Viljoen, F., Dobbe, R., Smit, B., 2009. Geochemical processes in peridotite xenoliths from the Premier diamond mine, South Africa: evidence for the depletion and refertilisation of subcratonic lithosphere. Lithos 112 (Supplement 2), 1133–1142. Walter, M.J., 1998. Melting of garnet peridotite and the origin of komatiite and depleted lithosphere. Journal of Petrology 39, 29–60. Walter, M.J., 2004. Melt extraction and compositional variability in mantle lithosphere. In: Holland, H.D., Turekian, K.K., Carlson, R.W. (Eds.), Treatise on Geochemistry: the Mantle and the Core, 2. Elsevier Pergamon, Amsterdam, pp. 363–394. Wasch, L.J., van der Zwan, F.M., Nebel, O., Morel, M.L.A., Hellebrand, E.W.G., Pearson, D.G., Davies, G.R., 2009. An alternative model for silica enrichment in the Kaapvaal subcontinental lithospheric mantle. Geochimica et Cosmochimica Acta 73, 6894–6917. Winter, J.D., 2002. An Introduction to Igneous and Metamorphic Petrology. Prentice Hall, New Jersey 12. Zeh, A., Gerdes, A., Klemd, R., Barton, J.M., 2007. Archaean to Proterozoic crustal evolution in the central zone of the Limpopo Belt (South Africa–Botswana): constraints from combined U–Pb and Lu–Hf isotope analyses of zircon. Journal of Petrology 48, 1605–1639.