Multistage Evolution of Dolerites in the Karoo Large Igneous …folk.uio.no/hensven/Neumann_JP2011_Karoo.pdf · 2011-04-13 · Multistage Evolution of Dolerites in the Karoo Large

Multistage Evolution of Dolerites in theKaroo Large Igneous Province,Central South Africa

ELSE-RAGNHILD NEUMANN1*, HENRIK SVENSEN1,CHRISTOPHE Y. GALERNE1,2 AND SVERRE PLANKE1,3

1PGPçPHYSICS OF GEOLOGICAL PROCESSES, UNIVERSITY OF OSLO, PO BOX 1048 BLINDERN, 0316 OSLO, NORWAY2GEODYNAMIK, STEINMANN-INSTITUT, UNIVERSITA« T BONN, NUSSALLEE 8, 53115 BONN, GERMANY3VBPRçVOLCANIC BASIN PETROLEUM RESEARCH, OSLO INNOVATION CENTER, 0349 OSLO, NORWAY

RECEIVED JANUARY 15, 2010; ACCEPTED FEBRUARY 25, 2011

EarlyJurassic sheet-like intrusions (sills and dykes) are abundant

in the Karoo Basin in South Africa, and were emplaced as a part of

the Karoo Large Igneous Province. Here we discuss the evolutionary

history of dolerite sills and dykes in different parts of the basin on

the basis of new major and trace element analyses of dolerite samples

collected from drill-cores (five sites spanning 1700 m of basin stra-

tigraphy) and previously published data on sills and dykes in the

Golden Valley Sill Complex (GVSC). In addition, we present Sr^

Nd isotope data for selected samples.The dolerites are subalkaline

tholeiitic basalts and basaltic andesites characterized by enriched

trace element patterns, variable degrees of depletion in Nb^Ta rela-

tive to light rare earth elements, negative to positive Pb anomalies,

and mild to moderate enrichment in initial Sr^Nd isotopic ratios.

The aim of this study is to unravel the evolutionary history of the

melts that gave rise to the dolerites. We propose that the primary

melts were derived from sub-lithospheric mid-ocean ridge basalt (or

ocean island basalt) source mantle and had acquired a weak subduc-

tion signature (relative depletion in Nb^Ta, mildly enriched

Sr^Nd isotopic ratios) through interaction with metasomatized

lithospheric mantle. In the deep crust the magmas underwent assimi-

lation and fractional crystallization (AFC) processes involving up

to 10% assimilation of granulites with strong arc-type geochemical

signatures. The AFC processes may alternatively have taken place

in the uppermost mantle. Distinct geochemical characteristics

among the GVSC and drill-core units reflect different amounts of

AFC. During and/or after intrusion into the sedimentary rocks in

the Karoo Basin the magmas underwent a second stage of fractional

crystallization (50^60%) and local contamination by their

sedimentary wall-rocks. High U concentrations and U/Th ratios

in some dolerites in the southwestern part of the Karoo Basin

were probably caused by fluids released from shales rich in organic

material (e.g. Ecca Group shales) during devolatilization and con-

tact metamorphism. Contamination in a GVSC unit may reflect

interaction withTa^ThÛ-rich minerals of the type found in strati-

form uranium ore bodies in the Karoo Basin, or fluids that have

interacted with such rocks. Considering that continental flood basalts

are emplaced through continental crust and sedimentary basins, it

is likely that other LIPs have similar evolutionary histories to that

proposed for the Karoo Basin.

KEY WORDS: Gondwana; geochemistry; Nd and Sr isotopes; AFC

processes; Karoo

I NTRODUCTIONContinental flood basalts (CFB) occur world-wide andrepresent an important style of intra-plate magmatism inthe continents. CFBs are characterized by the emplace-ment of large volumes of magma (of the order of severalmillion km3) within time periods of only a few millionyears or less (e.g. Cox, 1988; White & McKenzie, 1989;Duncan et al., 1997; Jourdan et al., 2007). Extrusion of lavasis accompanied by extensive intrusion of sills and dykesinto contemporary sedimentary basins (e.g. Symondset al., 1998; Chevallier & Woodford, 1999; Planke et al.,

*Corresponding author. E-mail: [email protected]

� The Author 2011. Published by Oxford University Press. Allrights reserved. For Permissions, please e-mail: [email protected]

JOURNALOFPETROLOGY VOLUME 0 NUMBER 0 PAGES1^26 2011 doi:10.1093/petrology/egr011

Journal of Petrology Advance Access published April 11, 2011

at University of O

slo Library on April 12, 2011

petrology.oxfordjournals.orgD

ownloaded from

http://petrology.oxfordjournals.org/

2005; Polteau et al., 2008). Many CFBs are believed to beassociated with the opening of new oceans (e.g. White &McKenzie, 1989; Coffin & Eldholm, 1990; Klausen, 2009).Among the geochemical features common to many CFBsare negative Nb^Ta anomalies and enriched Nd^Sr iso-tope ratios (e.g. Gallagher & Hawkesworth, 1992; Puffer,2001).The origin of these features has been discussed exten-sively over the last decades; proposed models include (1)partial melting of heterogeneous sub-continental litho-spheric mantle (SCLM; e.g. Gallagher & Hawkesworth,1992; Jourdan et al., 2007, 2009); (2) contamination duringpassage through the SCLM of magmas derived fromsub-lithospheric plume or mid-ocean ridge basalt(MORB) sources, or mixing between sub-lithospheric andlithospheric mantle melts (e.g. Marsh & Eales, 1984;Arndt & Christensen, 1992; Ellam et al., 1992; Riley et al.,2005; Jourdan et al., 2007; Heinonen & Luttinen, 2008,2010; Heinonen et al., 2010); (3) assimilation of sediments(Elburg & Goldberg, 2000); (4) polybaric fractional crys-tallization (Marsh & Eales, 1984); (5) contamination bysedimentary country rocks combined with fractional crys-tallization in addition to deep processes (e.g. Riley et al.,2005, 2006; Jourdan et al., 2007).All of these models have been proposed for the Karoo

Large Igneous Province (LIP) in South Africa (Fig. 1; e.g.Duncan et al., 1984; Marsh & Eales, 1984; Ellam & Cox,1989, 1991; Ellam et al., 1992; Harmer et al., 1998; Elburg &Goldberg, 2000; Ellam, 2006; Riley et al., 2006; Jourdan

et al., 2007, 2009). The Karoo CFB is thus eminently suitedfor further studies of the evolutionary history of floodbasalts from their mantle source to solidification in theupper crust.Galerne et al. (2008) recently presented major and trace

element data on 327 samples from sills and dykes exposedwithin a restricted area in the Karoo Basin, the GoldenValley Sill Complex (GVSC; Figs 1 and 2). The aimof that paper was to test geochemical similarities or dif-ferences between sills and dykes in the Golden ValleySill Complex to throw light on their emplacement mechan-ism. Forward Stepwise Discriminant Function Analysis(FS-DFA) showed that all samples from the same unit(sill or dyke) have the same geochemical signature, where-as different units generally have different signatures.These results imply that most sills and dykes in the GVSCformed from different magma batches, although one batchgave rise to two large sills and one small sill. The emplace-ment mechanism is thus a combination of the followingmodels: (1) the sills are fed by chemically distinct batchesof magma through separate feeders; (2) single sill com-plexes are generated from a single magma batch, suggest-ing a mechanism of sill feeding sill, thereby forming aninterconnected network.The aim of this study is to use the large dataset on the

Golden Valley Sill Complex to throw light on the evolu-tionary processes that led to the different geochemicalcharacteristics of the various GVSC units. In addition, we

Fig. 1. Simplified geological map of the Karoo Basin showing the locations of the studied boreholes.The blue dashed lines mark the boundariesbetween basement terranes or domains. Based on maps by DeWit et al. (1992), Cole (1998) and Schmitz (2002), and references therein.

JOURNAL OF PETROLOGY VOLUME 0 NUMBER 0 MAY 2011

2

at University of O



ownloaded from


have collected a new set of samples of sills from drill-coresat various stratigraphic levels in the Karoo Basin, as wellas Sr^Nd isotope data on representative samples fromboth groups. We believe that the results throw additionallight on magmatic differentiation processes in crust andmantle in the Karoo LIP, and on magmatic processes inCFBs in general. Furthermore, we show that the detailedgeochemical data collected on the GVSC, with numeroussamples from each of a limited number of magmatic units,provide unique information that cannot be obtained with-out such extensive sampling campaigns. The results alsohave implications for our understanding of the formationof ore complexes associated with LIPs [e.g. Norilsk-typeNi^PGE (platinum group element) complexes such as theEnziswa formation; Lightfoot et al., 1984; Marsh et al.,2003] and the processes that may lead to LIP-relatedclimatic changes.

GEOLOGICAL SETT INGThe Early Jurassic Karoo magmatism occurred over theentire southern African subcontinent and extended intoadjacent areas in Antarctica (e.g. Eales et al., 1984). Thetotal volume of magma emplaced amounts to several mil-lion km3, making this event one of the largest CFBs in theworld (e.g.White & McKenzie, 1989; Storey & Kyle, 1997;White, 1997). 40Ar/39Ar age determinations suggest that

emplacement of the Karoo CFB lasted from 184 to 176Ma(e.g. Riley & Knight, 2001; Riley et al. 2005; Jourdan et al.2007, 2008). However, U/Pb zircon ages suggest that insome parts of the Karoo Basin, including the GVSC, vol-canism and sill emplacement may have occurred as ashort pulse at about 182·5Ma (Encarnacio¤ n et al., 1996;Svensen et al., 2007). The 1·8 km thick Drakensberg lava se-quence in Lesotho (Fig. 1) is the remnant of a much widerlava cover (e.g. Duncan et al., 1997). The total area affectedby the Karoo magmatism is demonstrated by numeroussill intrusions and dykes (some are hundreds of kilometreslong; e.g. Chevallier & Woodford, 1999) which covernearly two-thirds of southern Africa (Fig. 1). Locally thesills form up to 70% of the Karoo Basin stratigraphy(Roswell & De Swardt, 1976).The Karoo Basin sills have intruded essentially un-

deformed sedimentary rocks of the Karoo Supergroup,which were deposited from about 310Ma until the initi-ation of volcanism (Visser et al., 1997; Catuneanu et al.,1998). The majority of the sills are strata bound and con-centrated in the Ecca and Beaufort groups (Fig. 1), whichessentially consist of shales, siltstones, mudstones and sand-stones (e.g. Cole, 1992; Veevers et al., 1994; Johnson et al.,1997). Sill intrusions commonly follow lithological bound-aries. There are virtually no sills in the basement com-plexes west and east of the Karoo Basin; it is thereforepossible that the most distal sills (and dykes) in the KarooBasin were fed laterally from feeder channels in the centralparts of the basin (J. Marsh, personal communication,2009). Sill emplacement resulted in widespread contactmetamorphism and the formation of hundreds of verticalpipe-like structures that released thermogenic carbongases into the atmosphere (Jamtveit et al., 2004; Svensenet al., 2004, 2006, 2007, 2008; Aarnes et al., 2010).The Karoo basaltic lavas have been divided into low-Ti

and high-Ti basalts (e.g. Cox et al., 1967). Both groups arecharacterized by negative Nb^Ta anomalies and enrichedSr^Nd^Pb isotopic compositions (e.g. Hawkesworth et al.,1984; Ellam & Cox, 1991; Marsh et al., 1997; Elburg &Goldberg, 2000), but also show significant geochemicalvariations (e.g. Pemberton, 1978; Marsh et al., 1997;Jourdan et al., 2007; Galerne et al., 2008). The GoldenValley Sill Complex and the drill-core sills discussed inthis study belong to the low-Ti group.The GVSC comprises four major saucer-shaped sills,

two large dykes and some minor sills and dykes (Fig. 2)intruded into the upper part of a sequence of sandstonesand shales of the Beaufort Group. The drill-core sampleswere collected in various parts of the Karoo Basin (Fig. 1).One drill-core (LA1/68, here called LA1), located NW ofthe Lesotho lava area, comprises sills emplaced in theupper Beaufort and Ecca groups. The QU1/65 borehole(referred to as QU1) in the western part of the KarooBasin intersects sills emplaced into the Dwyka, Ecca, and

Fig. 2. Simplified map showing the main sills and dykes in theGolden Valley Sill Complex (after Galerne et al., 2008). Differentcolors distinguish the geochemical signatures established by Galerneet al. (2008) by forward stepwise-discriminant function analysis; sillsnot included in this study are shown in grey. GVS, GoldenValley Sill;GS, Glen Sill; HS, Harmony Sill; MS, Morning Sun Sill; L1, L1 sill;GVD, Golden Valley Dyke. The position of the Golden Valley SillComplex is indicated in Fig. 1. The c. 100 km long, up to 30m wideCradock dyke (Marsh & Mndaweni, 1998) is located 70 km west ofthe GoldenValley Sill Complex.

NEUMANN et al. EVOLUTION OF KAROODOLERITES

3

at University of O



ownloaded from


Beaufort groups. Three boreholes, G39856, G39974 andG39980 (referred to as G56, G74 and G80, respectively),in the westernmost part of the Karoo Basin cut sillsemplaced into the Ecca Group (Fig. 3). The borehole sillshave cumulative thicknesses varying from550m to severalhundred meters.The basement beneath the Karoo lavas and sills com-

prises the Kaapvaal Craton in the northern, and the

Proterozoic Namaqua^Natal (PNN) mobile belt in thewestern and southern parts of the Karoo Basin (Fig. 1; e.g.DeWit et al., 1992; Schmitz, 2002; and references therein).The PNN is bounded to the south and west by the CapeFoldbelt.The exposed boundary between these terranes in-dicates that borehole LA1 is located on Archaean base-ment, whereas the other locations lie within the PNNmobile belt.

Fig. 3. Borehole logs of the studied drill-cores showing sill thicknesses, positions of analyzed samples, and variations in host-rock lithology.Theapproximate intrusion level of the GoldenValley Sill Complex is also shown. The three leftmost logs are based on those of Svensen et al. (2007).


4

at University of O



ownloaded from


METHODS AND ANALYT ICALPROCEDURESSamplingSamples from the GVSC were collected during field workin 2004 and 2005. The borehole samples were collected in2004 at the Core Library of the Council for Geosciencesin Pretoria, South Africa. The cores were logged for lith-ology during sampling; those presented in Fig. 3 representcombinations of this logging and available logs fromthe Council for Geosciences. The boreholes were drilledin the 1960s by Soekor (Suid Afrikaanse Olie ExplorasieKorporasie) for stratigraphic purposes (QU1, LA1), and inthe 1990s by the Council for Geosciences for groundwater research (G56, G74 and G80). The boreholes wehave used were fully cored and represent a unique possibil-ity to study sill geochemistry throughout the basin. In add-ition to sill samples, we also collected a number ofsamples from the intruded sedimentary sequences for refer-ence. Moreover, a sample of a pseudo-coal (solid bitumen:borehole PC2; Fig. 1) was included in this study as refer-ence for the composition of bitumen in the Karoo Basin.

GeochemistryThe samples were analyzed for major and trace elementsby inductively coupled plasma atomic emission spectrom-etry (ICP-AES) and inductively coupled plasma massspectrometry (ICP-MS) at the University of London,Royal Holloway. Blocks of fresh rock, at least 3 cm in alldirections, were cut and crushed. Powders for geochemicalanalyses were prepared from 50 g of crushed rock using atungsten tema-mill; 0·2 g of powdered sample was weighedinto a graphite crucible and 1·0 g of LiBO2 added. Thepowders were carefully mixed and fused at 9008C for20min, then dissolved in 200ml of cold 5% nitric acid.Ga was added to the flux to act as an internal standard.This solution was then analysed for Si, Al and Zr byICP-AES using a Perkin Elmer Optima 3300R. The in-strument was calibrated with natural and synthetic stand-ards. The solution was also used to analyse for Cs, Nb,Rb, Ta, Th, Tl, U, Y, and rare earth elements (REE) byICP-MS on a Perkin Elmer Elan 5000 calibrated with nat-ural and synthetic standards. Another batch of 0·2 g ofpowdered sample was dissolved in 6ml of HFand HClO4

(2:1 mixture), evaporated to dryness, cooled and dissolvedin 20ml of 10% HNO3. This solution was analysed byICP-AES for Fe, Mg, Ca, Na, K, Ti, P, Mn, Ba, Co, Cr,Cu, Li, Ni, Pb, Sc, Sr,V, and Zn.Representative samples were also analyzed for Sr and

Nd isotopes at the Geoanalytical Facility, University ofBergen. As Pb appears to have been disturbed by late-stagecontamination processes (see below) we chose not to ana-lyze for Pb isotopes. The chemical processing was carriedout in a clean-room environment with HEPA-filtered airsupply and positive pressure, using reagents purified in

two-bottle Teflon stills. Samples were dissolved in a mix-ture of HFand HNO3. Sr was separated by specific extrac-tion chromatography using the method described by Pinet al. (1994). Sr was loaded on a double Re-filament andanalysed in static mode using a Finnigan 262 mass spec-trometer. Sr isotopic ratios were corrected for mass frac-tionation using an 88Sr/86Sr value of 8·375209. Repeatedmeasurements of the NBS 987 Sr standard yielded an aver-age of 0·710223�0·000007 (2s; n¼ 9; accepted value is0·710240). The REE were separated by specific extractionchromatography using the method described by Pinet al. (1994), and Nd was subsequently separated using amodified version of the method described by Richardet al. (1976). The separated Nd was loaded on a doubleRe-filament and analysed in dynamic mode. TheNd-isotopic ratios were corrected for mass fractionationusing a 146Nd/144Nd ratio of 0·7219. Repeated measure-ments of the La Jolla standard yielded an average143Nd/144Nd ratio of 0·511842�0·000006 (2s) (n¼ 8).

PETROGRAPHYThe petrographic description of dolerites in the GVSC pre-sented below is a summary of that given by Galerne et al.(2008, 2010) with some additions. The sills generally showa shift from fine-grained textures at their margins (upperand lower 5m) to medium- or coarse-grained in their cen-tral regions. Most dolerites consist of about 50 vol. %plagioclase (pl), 5% olivine (ol), 40% pyroxene and55%Fe^Ti oxides. Minor amounts (50·5%) of apatite, pyriteand very rare fluorite and titanite are present in some ofthe most evolved rocks. Olivine (euhedral to subhedral)occurs partly as single grains (0·4^2mm in diameter) andpartly in clusters. Plagioclase is generally zoned and formselongated laths (typically 1^2mm long, but may reach6mm) partly enclosed by pyroxene and Fe^Ti oxides.Plagioclase also forms aggregates of zoned, large, subhe-dral grains (2^4mm). Such aggregates are found both inthe chilled margins and in central parts of the sills. Themain pyroxene species is augite (cpx), but orthopyroxene(opx) and pigeonite (pig) have also been observed. Thepyroxenes occur mainly as oikocrysts (up to 3mm in diam-eter) enclosing plagioclase and olivine; some oikocrysts en-close euhedral to subhedral cores. Most samples alsocontain magnetite (sp) and ilmenite and small amounts ofsulfide. Minor alteration has caused local growth of id-dingsite, carbonate and serpentine at the expense of oliv-ine, and biotite and chlorite at the expense of pyroxene.Biotite and chlorite seem to occur exclusively as secondaryphases. The textures of the dolerites indicate the followingin situ crystallization sequence for the main phases: olivine,plagioclase, pyroxenes, Fe^Ti oxides.The drill-core dolerites have similar textures to those

described for the GVSC. Also in the drill-cores most sam-ples show evidence of minor alteration with growth of


5

at University of O



ownloaded from


biotite, chlorite and Fe^Ti oxides at the expense of primarysilicates. A few samples from boreholes G74 and LA1(Figs 1 and 3; Electronic Appendix 1) are stronglyaltered with biotite, chlorite/chamosite, smectite, quartzand zeolite (possibly laumonite) as the main alterationproducts.

WHOLE -ROCK CHEMISTRYMajor and trace element data on samples from the GVSCwere reported by Galerne et al. (2008). Major and traceelement data on the drill-core samples are given in

Electronic Appendix 1. Many of the drill-core sills are rela-tively thin (a few meters to a few tens of meters) and arerepresented by only one or a few samples (Fig. 3). Sillswithin the same drill-core are therefore shown with thesame symbol in the figures; altered samples are shown byopen symbols.Most of the dolerites classify as basalts to basaltic andes-

ites with tholeiitic, subalkaline affinity; a few evolved sam-ples classify as dacites [using the classification systems ofKuno (1968) and Le Bas et al. (1986)]. With the exceptionof two evolved samples with 2·1 and 2·2wt % TiO2, thedolerites contain 0·4^2·0wt % TiO2 (Fig. 4) and thus

Fig. 4. Whole-rock major element variations (wt %) among the main sills and dykes in (a) the GoldenValley Sill Complex (GVSC; data fromGalerne et al., 2008) and (b) the drill-core sills. Arrows indicate trends defined by the Glen Sill and Harmony Sill samples. GVS, GoldenValley Sill; GS, Glen Sill; HS, Harmony Sill; MS, Morning Sun Sill; GV dyke, GoldenValley Dyke; Cr dyke, Cradock Dyke. Locations of theGVSC units are shown in Fig. 2, drill-core locations in Fig. 1, and drill-core logs in Fig. 3. Glen Sill samples that, on the basis of trace elements,are believed to have undergone upper crustal contamination (*cont, contaminated; open circles) fall within the field of other Glen Sill samples.Altered drill-core samples show addition of SiO2 and K2O, and removal of CaO and Na2O. (See text for further information.)


6

at University of O



ownloaded from


belong to the low-Ti series. Total ranges in MgO and SiO2

are 2·1^16·8 and 48^65wt %, respectively, but most of thesamples lie within the range 5^9wt % MgO and 48^53wt % SiO2. The GVSC samples define clear trends ofincreasing concentrations of TiO2, Fe2O3, K2O, P2O5,and decreasing Al2O3, CaO and Cr with decreasingMgO (Figs 4 and 5). Most units also show trends of strong-ly decreasing Ni with decreasing MgO (Fig. 5). TheHarmony Sill forms separate trends in most diagrams.Also, the Cradock Dyke appears to define a low CaO^Nitrend.Among the drill-cores the LA1 and G80 sills have rela-

tively uniform compositions whereas the others show con-siderable variations (Electronic Appendix 1; Fig. 4b); thewidest range is found in QU1 (e.g. 2·7^8·1wt % MgO,0·8^2·7wt % TiO2).The majority of the drill-core samplesfall within the GVSC ranges. The largest scatter is foundamong the altered samples from holes G56 and QU1, andamong samples with MgO 410wt %, which also show

very high Ni and Cr contents (Fig. 5). The MgO-richrocks differ petrographically from the other rocks byhaving a high proportion of euhedral to subhedral olivinesurrounded by near-equal porportions of clinopyroxeneand plagioclase, plus some magnetite. These samples aregenerally located at the base of sills (G74-1-3, G56-2-3,QU1-1-7; Electronic Appendix 1; Fig. 6) and are probablycumulates. Very high Pb, K2O and U contents, suggestingcontamination, are found in sample QU1-1-8 near theroof of a sill, but also in sample QU1-1-13 in the middle ofa sill (c. 190 and 240m, respectively, in Fig. 6). The lower-most QU1 sill shows an upwards development towardsmore differentiated rocks. Otherwise there are no system-atic variations.The majority of the dolerites have uniform, essentially

parallel PM-normalized [where PM is Primitive Mantleas defined by McDonough & Sun (1995)] trace elementpatterns showing mild enrichment in the most incompat-ible elements relative to heavy REE (HREE; e.g.LaN¼ 8^33; YbN¼ 3·9^8·3; the subscript N indicates nor-malization to PM; Fig. 7a and b). These samples arereferred to as Group 1. Group 1 patterns also show weakto strong negative Nb^Ta anomalies, weak negativeanomalies for P and Ti, and enrichment in K relative toLa. Pb shows mildly negative to strongly positive anoma-lies. Sr shifts from mildly positive anomalies in the leastevolved samples to negative anomalies in the most evolvedones which, together with depletion in Eu relative to neigh-boring REE in the most enriched samples is compatiblewith the removal of plagioclase. The majority of thedrill-core samples have trace element patterns that are es-sentially identical to those of the GVSC Group 1 doleritesand thus belong to Group 1 (Fig. 7b). The main differenceis that a few REE-rich drill-core dolerites lack the markedenrichment in K relative to La shown by all GVSCGroup 1 samples. This may be an alteration effect.Both in the GVSC and in the drill-cores a few samples

have trace element patterns that deviate from those ofGroup 1 (Fig. 7c and d); these are referred to as Group 2.In the GVSC Group 2 samples are restricted to a singleprofile across the Glen Sill (Fig. 2). These samples havemajor element compositions similar to Group 1 (Fig. 4),but are strongly enriched in Th, U, Nb and Ta, and havehigh Ta/Nb ratios (Fig. 7c). In the drill-cores Group 2 isrepresented by the moderately to strongly altered samplesin drill-cores G56 and QU1 (G56-3-1, G56-3-2, G56-3-4,QU1-1-5, QU1-1-7, QU1-1-8 and QU1-1-13; ElectronicAppendix 1; Fig. 3). These rocks tend towards higher SiO2

and K2O, and lowerTiO2 and CaO than Group 1 sampleswith similar MgO contents (Fig. 4b), and show muchstronger enrichment in most of the strongly incompatibleelements (Fig. 7d).In GVSC Group 1 dolerites from the same unit,

pairs of incompatible elements define trends through

Fig. 4. Continued.


7

at University of O



ownloaded from


the origin, whereas samples from different units mayhave contrasting ratios (Fig. 8). Group 2 samples falloff the Group 1 trends. Drill-core Group 1 samplesfall within, or close to, the trends defined by GVSCGroup 1 (Fig. 9), whereas Group 2 samples are mildlyto strongly enriched in Rb, Ba, Th, La and Pb relativeto Group 1 samples with similar Y contents. Pb differs

from other incompatible elements by showing signifi-cant variations that are unrelated to concentrations inother incompatible elements (e.g. Y) and wide rangesin concentration within each unit. Also, Sr shows somescatter unrelated to Y, and a weak increase withincreasing loss on ignition (LOI) among the GVSCsamples (not shown).

Fig. 5. Variations in Ni and Cr (ppm) relative to MgO (wt %) among dolerites from the GoldenValley Sill Complex (data from Galerne et al.,2008) and drill-cores in the Karoo Basin (Electronic Appendix 1). The field termed ‘basaltsþ dykes’ shows the range covered by low-Ti dykesin eastern Lesotho (data from Riley et al., 2006) and low-Ti basalts and dykes in Botswana^Zimbabwe (data from Jourdan et al., 2007).Decreasing Ni and Cr with decreasing MgO are the typical consequences of fractionation of olivine and clinopyroxene.With a few exceptionsamong the drill-cores, the dolerites of this study have significantly lower Ni and Cr contents than the most mafic ‘basaltsþ dykes’, implyingthat they are more evolved. Abbreviations as in Fig. 4.


8

at University of O



ownloaded from


SR^ND ISOTOPESThe dolerites also show a significant range in Nd^Sr iso-topic composition [initial ratios 87Sr/86Sr183: 0·7047^0·7080and eNd183: �0·72 to �4·3, calculated assuming an age of�183Ma determined by Svensen et al. (2007) for a sill inG74; Electronic Appendixes 1 and 2; Fig. 10]. Such Sr^Ndisotopic compositions strongly suggest contaminationby crustal rocks (e.g. Carlson et al., 1981; DePaolo, 1981;DePaolo et al., 1982).Within each GVSC unit (sill or dyke) Group 1 samples

show uniform initial Sr^Nd isotope ratios, whereas thereare significant differences between units. Exceptions are theGoldenValley Sill and the adjacent Glen Sill (Fig.10), whichhave similar ratios. In the Glen Sill Group 1 and 2 samplesshow identical initial Sr^Nd isotope ratios.The isotope datathus mimic the geochemical variations among the GVSCunits that were obtained through statistical analysis of majorand trace elements (Galerne et al., 2008). With respect to87Sr/86Sr183êNd183 the GVSC units define a trend fromabout (0·705, �1) to (0·708, �4·5) with the L1 sill, G74 andG80atthedepletedend,andtheHarmonySillat themosten-riched end (Fig. 10). Most Group 1 drill-core dolerites fallwithin the GVSC field, but samples from three sills indrill-core LA1 have slightly higher eNd183 at any given87Sr/86Sr183. The Group 2 drill-core samples with the mostdeviating trace element compositions (Fig.4b) scatter to loweNd183 and high

87Sr/86Sr183 (Fig.10).

COMPAR ISON WITH OTHERKAROO-TYPE MAGMATICROCKSKaroo-type lavas and dykes cover a wide range in traceelement and Sr^Nd isotope compositions. However, mostlow-Ti basaltic rocks (lavas and dolerites) in the KarooBasin (Fig. 11; data from Riley et al., 2006; Jourdan et al.,2007; J. Marsh, unpublished data) fall within a narrow87Sr/86Sr183êNd183 range from about (0·7045, þ1) to(0·7065, �4), which overlaps the GVSC drill-core field.Karoo-type lavas and dykes in South Africa andAntarctica have been divided into the chemical typesCT1^CT4 (Luttinen et al., 1998; Luttinen & Furnes, 2000),Groups 1^4 (Riley et al., 2005) and D-FP and E-FP(depleted and enriched ferropicrites, respectively;Heinonen & Luttinen, 2008). The key discriminant param-eters are listed in Table 1. The Group 1 dolerites of thisstudy show similarities to many of the chemical groupslisted inTable 1, but no overall match. Compared with theother groups, the most significant feature of the Group 1dolerites is their high average (Pb/Ce)N ratio (Table 1), al-though Group 2 dykes show a similar range in Pb anoma-lies. However, the dolerites differ from Group 2 by theirnegative Nb^Ta and positive K anomalies (Fig. 7, Table 1).With respect to the Drakensberg lavas the dolerites show

clear similarities to the Lesotho Formation and the south-ern part of the Barkley East Formation (south of 308S).The latter is, like the GVSC and drill-cores other thanLA1 (Fig. 1), located in the Proteozoic Namaqua^NatalMobile Belt. Many of the lavas in the northern part ofthe Barkley East Formation, located on the ArchaeanKaapvaal Craton, have significantly more enriched Sr^Nd isotopic signatures and significantly higher Ba/Nbratios (Table 1).

DISCUSS IONThe observed variations in major and trace element com-positions and in initial Sr^Nd isotope ratios (Figs 4^10)among the GVSC and drill-core dolerites show systematicsthat make it possible to reconstruct (at least parts of) theevolutionary history of the melts that gave rise to thesedolerites. We will show that the GVSC units andthe drill-core sills have a complex post-magma generationhistory that involves fractional crystallization and contam-ination at different depths. These processes are discussedbelow.

Fractional crystallizationSignificant variations in major and trace element concen-trations (e.g. 3·6^8·3wt % MgO), similar ratios betweenpairs of incompatible trace elements (Figs 7 and 8) andsimilar Sr^Nd isotopic ratios (Fig. 10a), as shown by eachGroup 1 GVSC unit, are typical of fractional

Fig. 6. Variations in selected chemical parameters in the sillintrusions from boreholes LA1 and QU1. The geochemical data areplotted against cumulative sill thickness in the boreholes; that is,by excluding the sedimentary rocks from the logs in Fig. 3. Theindividual sill intrusions are highlighted as alternating grey andwhite bands.


9

at University of O



ownloaded from


crystallization.We tried to mimic the trends formed by thevarious sample groups using the MELTS model ofGhiorso & Sack (1995) and Smith & Asimow (2005). Theresults for the groups showing the largest compositionalranges (Group 1 samples from the Harmony Sill anddrill-core QU1) are shown in Fig. 12. Although QU1 repre-sents several sills, the two groups show similar trends. Thesystematics of the MELTS curves shown in Fig. 12 are rep-resentative of all the modelling results, irrespective of start-ing composition. The MELTS curves show a series ofevolutionary stages: QU1: olþ l, cpxþ l, pl� cpx�olþ l,pl�ol�opx�pigþ l, plþopxþ spþ l; Harmony Sill:olþ l, olþplþ l, pl� ol� cpxþ l, plþ cpxþ spþ l. Thefirst stages, which are semi-parallel to the dolerite trends,are compatible with the crystallization sequence indicatedby the dolerite textures and trace element variations(Figs 5, 7 and 8). The last stage, involving removal ofspinel/magnetite, is not seen in the dolerite trends,

indicating that spinel/magnetite was never a fractionatingphase. The MELTS results imply that the stage of stronglydecreasing Al2O3 and CaO, and increasing Fe2O3 in thedolerite trends requires the combined removal of plþ cpx(� ol). This is compatible with decreasing Ni and Cr withdecreasing MgO (Fig. 5), and increasingly negative Sranomalies with increasing REE contents (Fig. 7) in thedolerites.We were, however, unable to find a set of physicochem-

ical conditions and starting compositions (based on theanalyzed samples) that exactly reproduced the doleritetrends. The main difference is that the MELTS model pre-dicts olivine as the only crystallizing phase to significantlylower MgO contents than we observe in the doleritetrends.This problem showed up in the MELTS results irre-spective of starting composition. Lower oxygen fugacity,addition of H2O (Fig. 12) and higher pressures (notshown) enhance this problem. Under more oxidizing

Fig. 7. Trace element concentrations normalized to Primordial Mantle (PM; McDonough & Sun, 1995) for uncontaminated (Group 1) andcontaminated samples (Group 2) in the Golden Valley Sill Complex (a and c, respectively) and drill-cores (b and d). (See text for furtherexplanation.)


10

at University of O



ownloaded from


conditions spinel enters the liquidus at higher MgO con-centrations; this is not compatible with the dolerite trends.Another important difference between the observedand modelled trends lies in the Na2O^MgO relationship(Fig. 12).The dolerites show increasing Na2O with decreas-ing MgO, whereas in the MELTS model removal ofplagioclase causes Na2O to decrease. A possible explan-ation of the discrepancy between the dolerite trends andthe MELTS results may be that flow of melt through alarge cooling sill involves more complex processes thansimple fractional crystallization. This is reflected in differ-ent compositional profiles. Proposed explanations includeprolonged continuous magma influx (Gorring & Naslund,1995), convective flux of refractory components within thecrystal^liquid mush in the boundary layer during in situ

differentiation or compositional convection (e.g. Tait &Jaupart, 1996), gravitational settling (e.g. Frenkel et al.,1989), Soret fractionation (e.g. Latypov, 2003; Latypovet al., 2007) and post-emplacement melt flow (Aarneset al., 2008; Galerne et al., 2010). These complex processesare expressed in a variety of compositional profiles in theGVSC (Galerne et al., 2010). The discrepancy betweenthe data and the model may also be due to problems inthe MELTS model. Putirka (2005) reported that in a com-parison with experimental melts that all had plagioclaseon the liquidus, the MELTS or pMELTS model failed toplace plagioclase on the liquidus for more than half thetest melts.Despite some differences between the MELTS and the

dolerite trends the results obtained from MELTS provideimportant information.The results imply that the compos-itional variations among the Group 1 dolerites involve ex-tensive fractional crystallization at low pressure (in situ),relatively oxidizing conditions (c. FMQþ1, where FMQis the fayalite^magnetite^quartz buffer) and dry condi-tions (or negligible water contents). Furthermore, underthese conditions, the MELTS results suggest that a MgOreduction from �7·5 to �3·5% requires about 50^60%crystallization (F¼ 0·4�0·5; Fig. 12). An estimate basedon the range of thorium contents in the Harmony Sill,1·5^3·4 ppm (Electronic Appendix 1), gives a similarresult, 40^45% crystallization (F¼ 0·55�0·60) assuminga bulk mineral/melt distribution coefficient for Th in therange 0·01^0·1 [based on data from Francalanci (1989)and Green et al. (2000)].However, the parent melts of the dolerites must also

have experienced some differentiation processes beforeintruding the upper crust (pre-intrusion processes). This isreflected in the evolved compositions of all the dolerites ofthis study. Their Ni and Cr contents (Fig. 5) are, with twoexceptions among the high-MgO^Ni^Cr drill-core sam-ples, significantly lower than those found in the mostmafic low-Ti basalts and dykes in other parts of the KarooBasin (e.g. Riley et al., 2006; Jourdan et al., 2007). The

Fig. 8. Variations in incompatible trace element abundances (ppm)between units in the Golden Valley Sill Complex. GVS, GoldenValley Sill; GS, Glen Sill; HS, Harmony Sill; MS, Morning Sun Sill;GV dyke, GoldenValley Dyke; Cr dyke, Cradock Dyke. Thin dashedlines indicate trends of fractional crystallization; thick dotted greylines marked ‘cont’ indicate the effects of contamination. (See text forfurther discussion.)


11

at University of O



ownloaded from


Fig. 9. Variations in incompatible trace element abundances (ppm) vsY between drill-cores. Dashed lines indicating fractional crystallizationare from Fig. 8. Dotted vertical arrows labelled ‘cont’ indicate the effects of contamination. (See text for further discussion.)


12

at University of O



ownloaded from


highest Fo content in olivine is 82, and the highestmg-number in pyroxenes is similar (Galerne et al, 2008,2010; E.-R. Neumann, unpublished data). Data on ultra-mafic Karoo-type rocks and mantle xenoliths in the

Karoo Basin imply that the primary melts should have sig-nificantly more Mg-rich compositions. Olivine cores in pi-critic dykes in Vestfjella (Dronning Maud Land,Antarctica) have Fo contents up to 92 (Heininen &Luttinen, 2010). Fo contents reported for olivine in mantlexenoliths from various South African localities fall in therange 88^94 (e.g. Boyd et al., 1999; Konzett et al., 2000;Gre¤ goire, et al., 2003, 2005; Simon et al., 2007). The lowestFo values are observed in MARID-type xenoliths inwhich metasomatism involving the addition of iron ap-pears to be associated with kimberlite magmatism atc. 80^140Ma (e.g. Konzett et al., 1998, 2000), thuspost-dating the Karoo event at c. 183 Ma. Fe-rich dunitexenoliths (Fo87^89) in the Kimberley kimberlite have beeninterpreted by Rehfeldt et al. (2007) to be recrystallized cu-mulates related to fractional crystallization of low-Ti typeKaroo flood basalt magmas. These data confirm that theprimary low-Ti Karoo magmas must have been signifi-cantly more mafic than the dolerites of this study.Pre-intrusion processes are discussed below.

In situ contaminationAs shown above, the altered and contaminated Group 2samples fall off the trends defined by incompatible elem-ents in Group 1 samples, and they tend towards higher87Sr/86Sr183 (Figs 7^10). These differences are interpretedas the results of local crustal contamination that has af-fected a few samples in different ways, implying contrast-ing contaminants.Group 2 samples in drill-cores G56 and QU1 in the west-

ern part of the Karoo Basin (Fig. 1) are found in sills thathave intruded Ecca shales (Fig. 3). Ecca shales and horn-felses are markedly enriched inTh, U, K, light rare earthelement (LREE) and Pb relative to heavy REE (HREE),have high U/Th ratios and show strongly negative Ba,Nb^Ta, P and Ti anomalies (Electronic Appendix 3;Figs 13a and 14). Also, pseudo-coal from borehole PC2(Fig. 1) is highly enriched with a high U/Th ratio but it dif-fers from other sedimentary rocks and from Group 2drill-core samples by having a strong positive Sr anomaly(Fig. 13b). Compared with Group 1 the Group 2 drill-coredolerites are more enriched in strongly incompatible elem-ents (Fig. 13b).Sedimentary rock samples collected in drill-core G74

at different distances from the sill contact (Figs 1 and 3)show dissimilar variations in the concentrations of traceelements (Electronic Appendix 3; Fig. 13a). This diversitysuggests a complex redistribution of elements betweenrestite minerals, new metamorphic phases and fluidsduring contact metamorphism and devolatilization of thewall-rock black shales. It is beyond the aim of this studyto go into detail about these processes. The explanationfavored at present is that during contact metamorphismand devolatilization a number of elements, including U,Th and LREE, are preferentially partitioned into the

Fig. 10. Variations in initial Sr^Nd isotope composition, assumingan age of 183 Ma determined for sills in the Karoo Basin (Svensenet al., 2007; Svensen & Corfu, personal communication, 2009), among(a) GVSC units and (b) drill-cores (note the different scales).Abbreviations as in Fig. 8. The crustal contamination trend is esti-mated on the basis of the EC-RAFC model of Spera & Bohrson(2004) using dyke SA.20.1 from southern KwaZulu^Natal, SouthAfrica (Riley et al., 2006) as the model initial melt and a combinationof the Proterozoic mafic granulite xenoliths HCA28 and HCA35from the northern Lesotho^central Cape province (Huang et al.,1995) as the contaminant. EC-RAFC parameters are given inTable 2. (See text for discussion.) Tick marks on the model curve are5% increments of crustal contamination.


13

at University of O



ownloaded from


fluid. Very high concentrations closest to the sill contactstrongly suggest that enriched fluids not only ascended to-wards shallower levels, but also moved towards (and into)the sill.The latter effect may be caused by the developmentof a negative pressure gradient at the sill margins owingto volume reduction during cooling and crystallization(Aarnes et al., 2008; Galerne et al., 2010). The chemical dif-ferences between Group 2 dolerites in the drill-cores andGroup 1 dolerites are compatible with contamination byenriched fluids. Influx of fluids into the sills must have con-tinued after crystallization of the dolerites and causedalteration.In the GVSC contamination is seen only in a single pro-

file across the Glen Sill (Group 2). Contamination in thisprofile has caused different degrees of enrichment in Th,U, Ta, and less extensively in Nb, relative to HREE ascompared with the Glen Sill Group 1 samples (Figs 7 and8); Sr^Nd isotopic ratios and major elements appear un-changed (Figs 4 and 10). The trace element patterns ofthese Group 2 samples differ significantly from those ofthe sedimentary wall-rocks (Figs 3 and 13c), comprisingsandstones and siltstones of the Beaufort Group (e.g. Cole,1992; Veevers et al., 1994; Johnson et al., 1997) that have

uncommonly high Ta/Nb ratios up to 0·23. Lavas anddykes from the Botswana^Zimbabwe province also tend to-wards highTa/Nb ratios (0·060^0·35; Jourdan et al., 2007).Lavas and dykes in the Karoo Basin and Vestfjella(Antarctica) in general, including Group 1 GVSC anddrill-core dolerites, fall within the range Ta/Nb¼ 0·05^0·07, similar to MORB, ocean island basalt (OIB) andtheir mantle sources (0·055^0·056; e.g. data fromHofmann, 1988; Sun & McDonough, 1989; McDonough& Sun, 1995). Sedimentary rocks in the Karoo Basin haveTa/Nb ratios in the range 0·070^0·16 (ElectronicAppendix 3); averages for the upper, middle and lowercrust are 0·075, 0·060, and 0·12, respectively (Rudnick &Gao, 2003). The highTa/Nb ratios in the Group 2 doleritesimply that the contaminant is not an ordinary rock type.One possibility is that it comprises fluids that have inter-acted with U^Ta^Th-rich minerals. The Beaufort Groupincludes stratiform uranium ore bodies believed to haveformed during the Cape orogeny (e.g. le Roux, 1993, andreferences therein). In addition to U, these bodies show en-richment in Pb, As, Co and Hg and very high U/Thratios (e.g. le Roux, 1993); unfortunately, data on Ta andNb are not available. No ore body is reported from the

Fig. 11. Initial Sr^Nd isotope compositions at 183 Ma for the GVSC and drill-cores of this study (dark grey field marked GVSC), comparedwith data for low-Ti type basalts, dykes and sills from other parts of the Karoo Basin. BAS1 and BAS2, Lesotho lavas (J. Marsh, personal com-munication, 2008), Botswana^Zimbabwe lavas (Jourdan et al., 2007); KwaZulu^Natal dykes (Riley et al., 2006); Effingham intrusive rocks (J.Marsh, personal communication, 2004); Rooi Rand dykes (RRd; high- and low-Ti; J. Marsh, personal communication, 2010). The modelcurves are calculated using the EC-RAFC model of Spera & Bohrson (2004) using the starting melt, assimilant and EC-RAFC parameterslisted inTable 2. (See text for detailed discussion.)


14

at University of O



ownloaded from


vicinity of the GVSC, but that does not exclude the possiblepresence in the area of layers rich in U^Ta^Th-rich min-erals. Interaction with such minerals or with fluidsformed through contact metamorphism might lead tolocal strong U^Ta^Nb-enrichment. Although importantinformation is lacking, we believe that this is the best ex-planation for the contamination in the Glen Sill.In addition to the contamination shown by the Group 2

dolerites, the wide ranges in Pb within each unit shownby Group 1 rocks strongly suggest addition throughlocal contamination. Local contamination has also causeda weak increase in Sr with increasing LOI. We foundno correlation between sill thickness and degree ofcontamination.

Pre-intrusion differentiation processesThe evolved character of all the Group 1 GVSC anddrill-core samples and the distinct trace element and Sr^

Nd isotope signatures exhibited by the different units(Figs 7^10 and 15) suggest that pre-intrusion processescomprise a combination of fractional crystallization andassimilation^contamination (AFC). We have used theEC-RAFC model of Spera & Bohrson (2004) to test thispossibility.Primitive-melt compositions for use in the EC-RAFC

modeling were sought among those Karoo-type lavas anddolerites that show the most depleted Sr^Nd isotope andtrace element compositions and the least evidence of crust-al contamination, based on data from Luttinen et al.(1998), Luttinen & Furnes (2000), Riley et al. (2005, 2006),Jourdan et al. (2007, 2008), Heinonen & Luttinen (2008,2010), Heinonen et al. (2010) and unpublished data (J. S.Marsh, personal communication, 2009). The primitivemelt must be at least as depleted as the least enrichedGVSC dolerite (87Sr/86Sr183�0·7049; eNd183� ^0·72;Nd� 9·5 ppm; Figs 10 and 16). The Sr^Nd isotope

Table 1: Comparison between selected chemical parameters in GVSC and drill-core Group 1 dolerites and basalts in the

Drakensberg Group, Lesotho, South Africa (Fig. 1) and chemical types CT1-CT4, Groups 1^4, and depleted and enriched

ferrobasalts (D-FP and E-FP, respectively) inVestfjella, Dronning Maud Land, Antarctica

Ba/Nb (Nb/La)N La/Yb (Pb/Ce)N87Sr/86Sr183 eNd183

range av. range av. range av. range av. range range

Low-Ti

GVSC1 14–67 27 0·41–1·0 0·72 2·4–6·0 4·4 0·3–4·9 2·7 0·7049–0·7080 �0·7 to �4·3

Drill-cores 10–51 29 0·48–0·86 0·66 3·6–6·7 4·4 0·4–5·2 3·2 0·7053–0·7086 �0·9 to �3·8

Lesotho Fm2 28–36 0·47–0·63 0·7048–0·7069 0·14 to �7·0

Barkley East (S)2 17–43 0·56–1·0 0·7049–0·7075 �1·0 to �3·6

Barkley East (N)2 31–83 0·24–0·65 0·7056–0·7165 �1·9 to �14·4

CT13 7·1–254 95 0·37–0·56 0·44 4·1–9·1 6·2 1·0–2·1 1·4 0·7070–0·7096 �2·5 to �11·1

CT2 sill3 22 0·62 2·9 0·9 0·7033 þ7·6

CT33 25–152 73 0·48–0·85 0·56 2·6–4·5 3·8 0·8–1·6 1·2 0·7037–0·7053 þ2·0 to �1·3

CT44 8·6–9·4 9 0·95–1·1 0·98 12–13 12 1·1–1·2 1·1 0·70434 þ0·7

Group 15 21–43 36 0·48–0·57 0·52 6·6–7·3 7·0 1·2–1·5 1·4 0·7034–0·7085 �5·8 to �6·4

Group 25 7·8–17 10 0·88–0·95 0·92 3·9–4·2 4·1 0·6–4·2 1·3 0·7034–0·7046 þ1·7 – þ0·7

High-Ti

CT2 lavas3 26–51 35 0·39–0·62 0·42 4·5–8·0 5·7 1·0–1·4 1·2 0·7058–0·7082 �0·2 to �7·5

Group 35 4·5–41 10 0·40–1·1 0·97 2·5–6·4 3·5 0·4–2·0 0·7 0·7035–0·7062 þ9·0 – þ5·0

Group 45 14–44 17 0·45–0·83 0·67 11–29 17 0·5–0·7 0·5 0·7048–0·7059 þ2·4 to �4·6

E-FP6 7·2–58 20 0·39–1·1 0·82 1·8–9·1 6·0 0·7033–0·7070 þ7·6 to �10·7

Variable Ti

D-FP6 3·2–49 13 0·62–0·84 0·73 3·9–8·7 5·4 0·7030–0·7036 þ8·3 – þ4·8

1Galerne et al. (2008).2Range in averages of single lava flows; data from J. Marsh (personal communication, 2008).3Data from Luttinen & Furnes (2000).4Data from Luttinen et al. (1998).5Riley et al., (2005).6Data from Heinonen & Luttinen (2008) and Heinonen et al. (2010).


15

at University of O



ownloaded from


Fig. 12. Compositional variations among Group 1 dolerites from the Harmony Sill and drill-core QU1 (recalculated to a total of 100%) com-pared with trends of fractional crystallization estimated by the MELTS model (Ghiorso & Sack,1995; Smith & Asimow, 2005). Blue continuousline and numbers: 1 kbar, dry, fO2 buffer curve FMQþ1, sample QU1-1-9 as starting composition; blue dashed line: same, but using the FMQbuffer curve; red continuous line: 0·5 kbar, dry, fO2 buffer curve FMQþ1, sample K05-196 as starting composition; red dashed line: same, butassuming 0·3% H2O in the starting melt; vertical lines indicate changes in the phase assemblage. Because all Fe is assumed to be trivalent inthe starting composition, the MELTS compositions, with both FeO and Fe2O3, are shifted to slightly higher concentrations in SiO2, Al2O3,etc. The MELTS curves do not reproduce the dolerite trends in detail, but they clearly indicate that the stage of strongly decreasing Al2O3

and CaO, and increasing Fe2O3 with decreasing MgO requires the removal of both plagioclase and clinopyroxene (� olivine). Numbers indi-cate proportion of melt remaining (F). Small irregularities along the MELTS curves reflect variations in the liquidus assemblages (e.g. pl,plþol and plþ cpx in the pl�ol� cpx field). ol, olivine; pl, plagioclase; cpx, clinopyroxene; opx, orthopyroxene; pig, pigeonite; sp, spinel.(See text for discussion.)


16

at University of O



ownloaded from


relationships of representative candidates are shown inFig. 17. Karoo basalts or dolerites that meet these restric-tions are relatively rare. In general crustal rocks in theKaroo are markedly depleted in Nb and Ta relative toother strongly incompatible elements (Eglington et al.,1989; Huang et al., 1995; Cornell et al., 1996; Eglington &

Fig. 13. Concentrations of incompatible trace elements normalizedto Primordial Mantle (PM; McDonough & Sun, 1995) for (a) repre-sentative sedimentary rocks in the Karoo Basin (ElectronicAppendix 3), (b) sedimentary rocks (grey shaded field) and alteredor contaminated drill-core samples (Group 2), and (c) sedimentaryrocks (grey shaded field) and contaminated (Group 2) dolerites inthe Glen Sill. (See text for discussion.)

Fig. 14. U/ThÛ relationships among the drill-core dolerites com-pared with black shales (U/Th¼ 0·67, U¼ 23·0 ppm) and pseudo-coal (PC2) (U/Th¼ 2·2, U¼ 29·4 ppm): (a) overview; (b) details.Concentrations in black shale hornfels at different distances from thecontact (red stars) were measured in drill-core G74 (Fig. 3;Appendix 3). The U/ThÛ relationships in altered samples(Group 2) in drill-cores G74 and QU1 are compatible with contamin-ation by fluids released from black shales and/or pseudo-coal.Somewhat elevated Ucontents also suggest that some of the QU1sam-ples assigned to Group1may have undergone minor contamination.


17

at University of O



ownloaded from


Armstrong, 2003; Kampunzu et al., 2003; Lana et al., 2004).Sr^Nd isotope ratios are available for Archaean andProterozoic mafic granulite xenoliths (Huang et al., 1995).These rocks cover a significant range in eNd183 (þ14 to ^19) and relatively low 87Sr/86Sr183 (0·7035^0·7110; Fig. 17),which is typical of the lower to middle crust. Sedimentaryrocks retrieved from drill-cores (Eglington & Armstrong,2003) are enriched in terms of eNd183 (�4·4 to �21·2) and

87Sr/86Sr183 (0·708^0·738). The assimilant used in the mod-elling must be at least as enriched as the most enricheddolerite (87Sr/86Sr183�0·7086; eNd183��4·4).EC-RAFC parameters are listed inTable 2. The low Nd

concentrations in the dolerites (Fig. 15) pose important re-strictions on the starting conditions used in the modelling.If the solidus temperature (Ts) is higher than the initialtemperature of the assimilant (Tao) the isotope ratios inthe melt will stay constant until the assimilant reaches Tsand starts melting; at the same time cooling and fractionalcrystallization of the melt will increase its content of in-compatible elements (dotted lines in Fig. 16). To mimic theisotope^trace element trend of the GVSC Group 1 units

Fig. 15. Relationships between initial Sr^Nd isotopic compositionsand size of Nb^Ta anomalies (expressed by the ratio NbN/LaN)among Group 1 GVSC and drill-core dolerites. The dolerites show aclear decrease in NbN/LaN with decreasing 87Sr/86Sr183 and decreas-ing eNd183, indicating that these parameters are closely related. GVS,Golden Valley Sill; GS, Glen Sill; HS, Harmony Sill; MS, MorningSun Sill; L1, L1 sill; GV dyke, GoldenValley Dyke; Cr dyke, CradockDyke. The model curve is derived using the EC-RAFC model ofSpera & Bohrson (2004), with mafic dyke SA.20.1, KwaZulu^Natal,South Africa (Riley et al., 2006) as the starting melt and a combin-ation of Proterozoic granulite xenoliths HCA35 and HCA28 (Huanget al., 1995) as the assimilant. EC-RAFC parameters are listed inTable 2. (See text for discussion.)

Fig. 16. Relationships between initial Sr^Nd isotopic compositionsand Nd concentrations compared to EC-RAFC model curves (con-tinuous lines) using starting materials as for Fig. 15, and the param-eters given inTable 2. Dotted lines are EC-RAFC model curves usinga starting melt with Ndmelt¼ 3 ppm, Ts¼ 7108C, and other param-eters as inTable 2. Abbreviations as in Fig. 15.


18

at University of O



ownloaded from


Ts^Tao must be very small and/or the Ndmelt very low.Themodels shown in Figs 10, 11, 15 and 16 are based onNdmelt¼ 9·5 ppm and Tao¼Ts (Table 2). For Tao¼Ts theinitial assimilant temperature must be several hundreddegree centigrades; we chose 7008C, the exact value is notimportant. Such a highTao means that the EC-RAFC pro-cesses must have taken place at elevated pressures, in thelower crust or upper mantle. Riley et al. (2006) proposedthat enrichment in the KwaZulu^Natal dykes was causedby contamination by Ecca Group hornfels (ElectronicAppendix 3). However, in addition to the high temperatureand pressure required, EC-RAFC runs show that the highNd content (43 ppm) in the Ecca hornfels makes it an im-possible candidate for an assimilant in the dolerites of thisstudy.The most primitive Karoo-type rocks have been

found in Vestfjella (Antarctica; Luttinen & Furnes, 2005;Heinonen et al., 2010). A high-Ti, high-Nd sample, Z.1816.2,was used to test the possibility that the GVSC dolerites ori-ginated from an isotopically depleted mantle source(Table 2; Figs 10 and 11). However, independent of theassimilant, bulk partition coefficient or temperature set-tings used in the modelling, this sample failed to reproducethe dolerite trend. Similarly, Riley et al. (2006) discardedthe Z.1816.2 sample as a model primitive melt composition

for AFC processes in dykes in the KwaZulu^Natal area.Tests with different primitive-meltâssimilant combin-ations showed that the best fit was obtained using maficdyke SA.20.1 (KwaZulu^Natal; Riley et al., 2006) as themodel primitive-melt composition, and a combination ofthe Proterozoic mafic granulites HCA28 (trace elements)and HCA35 (Sr^Nd isotopes) as assimilant model com-position (HCA28/35; data from Huang et al., 1995).Although the modelling results (Figs 10, 11, 15 and 16)

do not fit the GVSC trend perfectly, they show that it ispossible to develop the trace element^Nd^Sr isotope char-acteristics of the different GVSC and drill-core units byEC-RAFC processes at elevated pressures, involving�10% assimilation of crustal rocks. The results imply thatthe initial melt had mildly enriched Sr^Nd isotope ratiosand was, most probably, somewhat more depleted in Nb^Ta relative to other strongly incompatible elements thansample SA.20.1 (Fig. 15).We cannot exclude the possibility that AFC processes

took place in the upper SCLM. Interaction with the litho-spheric mantle has been proposed by several workers (e.g.Marsh & Eales, 1984; Arndt & Christensen, 1992; Ellamet al., 1992; Gallagher & Hawkesworth, 1992; Luttinen &Furnes, 2000; Riley et al., 2005, 2006; Jourdan et al., 2007,2009; Heinonen et al., 2010). However, the relatively flatHREE patterns ([Dy/Yb]N¼1·0^1·2; Fig. 7) of the doler-ites preclude interaction with garnet-bearing materials.The weak subduction-like signature indicated for theprimitive melt(s) may either have been inherited from thesub-lithospheric mantle source of the primary magmas, orbeen acquired during ascent through the SCLM.These al-ternatives are discussed below.

Mantle^crust plumbing systemThe questions regarding the composition of the primitivemelt(s) and location of AFC processes require a review ofthe composition of the mantle beneath the PNN and adja-cent part of the Kaapvaal Craton, and the SCLM.There has been extensive discussion about the source of

the primary Karoo magmas, involving partial meltingof heterogeneous SCLM (e.g. Gallagher & Hawkesworth,1992; Jourdan et al., 2007, 2009), or derivation fromsub-lithospheric mantle plume or MORB-source mantle(e.g. Marsh & Eales, 1984; Arndt & Christensen, 1992;Ellam et al., 1992; Riley et al., 2005; Jourdan et al., 2007;Heinonen & Luttinen, 2008, 2010; Heinonen et al., 2010).Strongly depleted Nd^Sr isotope compositions andMORB- or OIB-type trace element patterns, indicating a‘normal’ sub-lithospheric mantle source have been mainlyreported for high-Ti and variable-Ti Karoo groups(Group 3, E-FP and D-FP; Table 1). However, there arealso a few examples of depleted Sr^Nd isotope compos-itions among low-Ti groups (CT2 sill and Rooi Randdykes; Table 1). In a recent study, Heinonen et al. (2010) sug-gested that heterogeneous sub-lithospheric, long-term

Fig. 17. Sr^Nd isotope compositions of primitive Karoo-type floodbasalts and crustal rocks in the Karoo Basin, recalculated to an ageof 183 Ma. Karoo-type mafic rocks (Karoo Basin and Vestfjella,Antarctica): data from Luttinen et al. (1998), Luttinen & Furnes(2000), Riley et al., (2005, 2006) and Heinonen et al. (2010). Archaeanand Proterozoic mafic granulite xenoliths from northern Lesothoand central Cape Province, South Africa (Huang et al., 1995). Uppercrust: granite, granodiorite, gneiss, hornfels retrieved from 2500^6000m depths in drill-cores in the Karoo Basin (Eglington &Armstrong, 2003). (See text for discussion.) Sample numbers indicaterock compositions used in the EC-RAFC modelling.


19

at University of O



ownloaded from


depleted, peridotitic mantle with enriched components isthe main source for the primary Karoo melts.An important observation is that although the Group 1

dolerites were sampled partly on the Archaean KaapvaalCraton (LA1 drill-core; Fig. 1) and partly on differentparts of the PNN mobile belt (all others), they show verysimilar trace element and Sr^Nd isotope characteristics(Figs 7^11). There is some controversy with regard to theevolution and composition of the mantle lithosphere be-neath these two crustal domains. Hopp et al. (2008) re-ported 40Ar/39Ar ages of 1·0^1·2 Ga for phlogopites inkimberlite-hosted mantle xenoliths in the southernKaapvaal Craton. These ages coincide with the Kibaranorogenic cycle that formed the PNN, and Hopp et al.(2008) proposed that during this period melts and fluidscaused significant metasomatism of the PNN mantle litho-sphere, and that the effects reached far into the KaapvaalCraton. However, based on PGE and ReÔs isotope data

on mantle xenoliths from off-craton kimberlites, Janneyet al. (2010) concluded that the PNN mantle lithospherebecame strongly depleted through moderate degrees ofpartial melting during the earliest Proterozoic, and add-itional moderate to high degrees of partial melting at1·3^1·0Ga, and was undisturbed from about 1 Ga untilthe time of the Karoo event. They found no mantlewedge signature in the PGE patterns of the PNN perido-tites. Unfortunately, Janney et al. (2010) did not presentREE, high field strength element (HFSE) or Sr ^Nd iso-tope data that might be directly compared with theGVSC and drill-core data.There is consensus of opinion that the SCLM beneath

the Kaapvaal Craton has been subjected to repeated en-richment processes (e.g. Simon et al., 2003, 2007; Hoppet al., 2008), but the resulting geochemical signatures areunclear. Hopp et al. (2008) proposed subduction-relatedmetasomatism. Simon et al. (2007) reconstructed the

Table 2: EC-RAFC parameters

Tlm liquidus of magma 13008C

Tmo initial temperature of magma 13008C

Tla liquidus of assimilant 11008C

Tao initial temperature of assimilant 7008C

Ts solidus (melt and assimilant) 7008C

Cpm specific heat of magma 1495 J kg�1 K�1

Cpa specific heat of assimilant 1400 J kg�1 K�1

Hcry heat of crystallization 395000 J kg�1

Hfus heat of fusion 354000 J kg�1

Sr (ppm) Nd (ppm) Nb (ppm) La (ppm) 87Sr/86Sr eNd

Magmas

Z.18.16.21 218 25 0·703518 8·75

A1292 150 9·6 0·70385 4·12

SA.20.13 256 9·5 5·9 12·8 0·704607 �0·83

SA.20.1-mod 256 9·5 12·8 0·704607 �0·83

Assimilants

HCA35/284 481 4·9 2·0 7·0 0·711026 �13·9

QU1/65 (2993m)5 658 19·9 0·708423 �21·2

Bulk D0 magma 0·2 0·15 0·15 0·15

Enthalpy magma 0 0 0 0

Bulk D0 assimilant 0·2 0·15 0·15 0·15

Enthalpy assimilant 0 0 0 0

1Dolerite, Dronning Maud Land, Antarctica (Riley et al., 2005).2Dyke, Rooi Rand, South Africa (J. Marsh, personal communication, 2010).3Mafic dyke, KwaZulu–Natal, South Africa (Riley et al., 2006).4Proterozoic granulite xenolith (Huang et al., 1995).5Granodiorite, borehole (Eglington & Armstrong, 2003).


20

at University of O



ownloaded from


pre-kimberlite compositions of kimberlite-hosted xenolithsin the Kaapvaal Craton and found that before the kimber-lite metasomatism typical mantle compositions hadconcave-upward trace element patterns with weakly nega-tive to positive Nb^Ta anomalies. According to Schmitz& Bowring (2004) the effects of the Late ProterozoicNamaquan orogeny is manifested in the formation oflower crustal granulites, both in the PNN mobile belt andalong the southern margin of the Kaapvaal Craton.Other parts of the cratonic crust in southern Africa showstrong arc-like trace element signatures.In reality, any combination of the compositional charac-

teristics proposed for the Kaapvaal and PNN mantle do-mains may be possible. The chemical similarity of sillsemplaced in cratonic (LA1; Fig. 1) and PNN terranes(all others) may thus be explained by ascent throughSCLM domains with similar geochemical signatures. Analternative explanation, which allows for the possibilitythat the two mantle domains had different geochemicalsignatures, is that the primary magmas were transportedover long distances (e.g. Encarnacio¤ n et al., 1996; Elliotet al.,1999), so that the magmas giving rise to all the sills as-cended through the same SCLM domain, which had anarc-like signature. A third possibility is that all themagmas ascended through depleted PNN lithosphericmantle without interaction with the mantle wall-rocks.Luttinen & Furnes (2000) found evidence among Karoo-type flood basalts in Antarctica that melts extruded onProterozoic crust were not contaminated by lithosphericmantle material, whereas melts extruded on cratonic crustwere contaminated by cratonic mantle material. If thePNN mantle was depleted and the melts ascended withoutinteracting with the wall-rocks, the sublithospheric sourcemust have had a mild arc-like signature.

Assuming different combinations of the evolutionaryprocesses outlined above and of compositions in lithospher-ic and sublithospheric mantle domains, four main scen-arios are possible. These are schematically outlined inTable 3.We consider Scenario 1most likely.

SUMMARY MODELAssuming Scenario 1 inTable 3 the evolutionary history ofthe dolerites in the sills and dykes in the GVSC and indrill-cores in various parts of the Karoo Basin, SouthAfrica, is summarized below, and shown schematically inFig. 18.

(1) The primary magmas originated in sub-lithosphericmantle with MORB (or OIB) source composition.

(2) We accept the results of Hopp et al. (2008) thatthe lithospheric mantle beneath the ProterozoicNamaqua^Natal mobile belt and the adjacentKaapvaal Craton had acquired an arc-like signature(enriched Sr^Nd isotopic compositions and negativeNb^Ta anomalies) owing to metasomatism duringthe Kibaran orogenic cycle at 1·2^1·0 Ga.

(3) The magmas that gave rise to the dolerites of thisstudy interacted with lithospheric mantle wall-rocksduring ascent to the near-surface and thus acquired aweak arc-like signature (Stage 1).

(4) The magmas intruded the deep crust, where theyunderwent different degrees of contamination bycrustal rocks with a strong arc-type geochemical sig-nature, accompanied by fractional crystallization(AFC processes; Stage 2). Assimilation of �10%lower crustal rocks led to different degrees of enrich-ment in strongly incompatible elements and Sr^Ndisotopic ratios, and relative deletion in Nb and Ta.

Table 3: Schematic outline of possible scenarios for the evolution of dolerites in the Golden Valley Sill Complex

and drill-cores

Scenario 1 Scenario 2 Scenario 3 Scenario 4

Upper crust FC FC FC FC

Local contamination Local contamination Local contamination Local contamination

Lower crust W-r: strong arc sign. ???? W-r: strong arc sign. W-r: strong arc sign.

AFC AFC AFC

Upper SCLM W-r: mild arc sign. W-r: strong arc sign. W-r: depleted W-r: mild arc sign.

Melt interaction AFC No melt interaction

Lower SCLM W-r: mild arc sign. W-r: mild arc sign. W-r: depleted Partial melting

Melt interaction Melt interaction No melt interaction

Sublithospheric MORB type (or OIB) MORB type (or OIB) Mild arc signature

source Partial melting Partial melting Partial melting

W-r, wall-rocks; arc sign., arc signature.


21

at University of O



ownloaded from


High-grade metamorphic rocks with strong arc-typesignatures are found in the PNN as well as in theKaapvaal lower crust.

(5) Different degrees of contamination gave rise todissimilar geochemical characteristics in the meltbatches that formed the various GVSC and drill-coreunits.

(6) Contaminated and somewhat evolved melts ascendedto the upper crust where they were intruded to formsills and dykes in the layered sedimentary sequence.

(7) During and/or after emplacement into the upper crustthe melts were subjected to a new stage of fractionalcrystallization (up to 60%: Stage 3).

(8) Locally the magmas also underwent a new type ofcontamination (Stage 3), possibly caused by fluidsreleased from the sedimentary wall-rocks during con-tact metamorphism and devolatilization. Differenttypes of wall-rocks gave rise to fluids with contrastingchemical characteristics.

(9) In drill-cores G56 and QU1 that intruded black shalesrich in organic material within the Ecca Group, somesamples were contaminated by fluids released from

the shales, resulting in marked addition of Si, K, Cs,Rb, Ba, Th, Pb and LREE, increased U/Th ratiosand more enriched Sr^Nd isotopic ratios.

(10) In the GVSC contamination in one of the sills (theGlen Sill) led to strong enrichment in Th, U and Ta,and less extensively in Nb, whereas Sr^Nd isotopicratios and major elements appear unchanged. Thiscontamination may have been caused by interactionwith ore minerals of the type found in stratiform ur-anium ore bodies in the Karoo Basin, or with fluidsthat had interacted with such minerals.

ACKNOWLEDGEMENTSWe are indebted toJulian S. Marsh for letting us use his un-published Sr^Nd isotope data on magmatic rocks fromthe Karoo Basin, for helpful discussions based on hisvast knowledge of the geology and petrology of the Karoomagmatism, and for feedback on an earlier version of thispaper. Many thanks are also due to N. S. C. Simon for en-lightening discussions, and to P. E. Janney, who gave usaccess to a paper on the evolution of the lithosphericmantle in the Proterozoic Namaqua^Natal mobile belt

Fig. 18. Schematic presentation of the ascent through the lithosphere of the melts that gave rise to the dolerites in the GVSC and drill-cores,and the processes believed to have taken place at different depths and in different areas.The figure is not to scale and the possibility of significantlateral magma transport (e.g. Encarnacio¤ n et al., 1996; Elliot et al., 1999) is not considered. Color variations between the intrusions in the deepcrust indicate different degrees of contamination (imposing an arc signature) and fractional crystallization (Stage 2); vertical color variationsin the shallow sills indicate a new stage of fractional crystallization locally combined with contamination that reflects the local sedimentarycountry-rocks (Stage 3). Information about variations in the thickness of the crust and lithospheric mantle from the southern KaapvaalCraton, across the Namaqua^Natal mobile belt into the Cape Foldbelt is taken from Nguuri et al. (2001), James et al. (2003) and Li & Burke(2006). Alternative scenarios are presented inTable 3. (See text for further information and discussion.) E, Ecca Group; B, Beaufort Group; S,Stormberg Group.


22

at University of O



ownloaded from


before publication.We would also like to thank Doug Colefor kindly providing the PC2 pseudo-coal sample, andLuc Chevallier for discussions about the Karoo drill-cores.The paper was improved significantly through construct-ive criticism by the reviewers Fred Jourdan, Arto Luttinenand Teal R. Riley.

FUNDINGThe project was made possible through financial supportfrom the Norwegian Research Council (NFR) for the pro-ject 159824/V30 ‘Emplacement mechanisms and magmaflow in sheet intrusions in sedimentary basins’, includinga doctoral fellowship to C.Y.G. The work has also beensupported by a Centre of Excellence grant from theNorwegian Research Council to Physics of GeologicalProcesses.

SUPPLEMENTARY DATASupplementary data for this paper are available at Journalof Petrology online.

REFERENCESAarnes, I., Podladtchikov, Y. Y. & Neumann, E.-R. (2008).

Post-emplacement flow induced by thermal stresses: Implicationsfor differentiation in sills. Earth and Planetary Science Letters 276,152^166.

Aarnes, I., Svensen, H., Connolly, J. A. D. & Podladchikov, Y. Y.(2010). How contact metamorphism can trigger global climatechanges: Modeling gas generation around igneous sills in sediment-ary basins. Geochimica et Cosmochimica Acta 74, 7179^7195.

Arndt, N. T. & Christensen, U. (1992). The role of lithospheric mantlein continental flood volcanism: thermal and geochemicalconstraints. Journal of Geophysical Research 97, 10967^10981.

Boyd, F. R., Pearson, D. G. & Mertzman, S. A. (1999). Spinel-faciesperidotites from the Kaapvaal root. In: Gurney, J. J., Gurney, J.L., Pascoe, M. D. & Richardson, S. H. (eds) Proceedings of the 7th

International Kimberlite Conference, Cape Town, 1998. Cape Town: RedRoof Design, pp. 40^48.

Carlson, R.W., Lugmair, G.W. & McDougall, J. D. (1981). ColombiaRiver volcanism: The question of mantle heterogeneity or crustalcontamination. Geochimica et Cosmochimica Acta 45, 2483^2499.

Catuneanu, O., Hancox, P. J. & Rubidge, B. S. (1998). Reciprocalflexural behaviour and contrasting stratigraphies: a new basin de-velopment model for the Karoo retroarc foreland system, SouthAfrica. Basin Research 10, 417^439.

Chevallier, L. & Woodford, A. (1999). Morpho-tectonics and mechan-ism of emplacement of the dolerite rings and sills of the westernKaroo, South Africa. South AfricanJournal of Geology 102, 43^54.

Coffin, M. F. & Eldholm, O. (eds) (1990). Large Igneous Provinces:

JOI/USSAC Workshop Report. University of Texas at Austin Institute for

GeophysicalTechnical Report 114, , 79 p.Cole, D. I. (1992). Evolution and development of the Karoo Basin.

In: deWit Maarten, J. & Ransome, I. G. D. (eds) InversionTectonicsof the Cape Fold Belt, Karoo and Cretaceous Basins of Southern Africa.Rotterdam, Netherlands: A.A. Balkema, pp. 87^99.

Cole, D. I. (1998). Palaeogeography and palaeocurrent distribution ofthe Beaufort Group in the Karoo Basin, south Africa, during theLate Permian. Journal of African Earth Sciences 27, Special Issue,Gondwana 10: Event Stratigraphy of Gondwana, 46^47.

Cornell, D. H., Thomas, R. J., Bowring, S. A., Armstrong, R. A. &Grantham, G. H. (1996). Protolith interpretation in metamorphicterranes: a back-arc environment with Besshi-type base metal po-tential for the Quha Formation, Natal Province, South Africa.Precambrian Research 77, 243^271.

Cox, K. G. (1988). The Karoo Province. In: Macdougall, J. D. (ed.)Continental Flood Basalts. Boston, MA: Kluwer Academic,pp. 239^271.

Cox, K. G., Macdonald, R. & Hornung, G. (1967). Geochemical andpetrologic provinces in the Karoo basalts of Southern Africa.American Mineralogist 52, 1451^1474.

DePaolo, D. J. (1981). A neodymium and strontium isotopic study ofthe Mesozoic calc-alkaline granitic batholiths of the SierraNevada and Penninsular Ranges, California. Journal of GeophysicalResearch 86, 10470^10488.

DePaolo, D. J., Manton, W. I., Grew, E. S. & Halpern, M. (1982).Sm^Nd, Rb^Sr and U^Th^Pb systematics of granulite faciesrocks from Fyfe Hills, Enderby Land, Antarctica. Nature 298,614^618.

De Wit, M. J., Roering, C., Hart, R. J., Armstrong, R. A., deRonde, C. E. J., Green, R. W. E., Tredoux, M., Peberdy, E. &Hart, R. A. (1992). Formation of an Archaean continent. Nature357, 553^562.

Duncan, A. R., Erlank, A. J. & Marsh, J. S. (1984). Regional geo-chemistry of the Karoo igneous province. Special Publication of the

Geological Society of South Africa 13, 355^388.Duncan, R. A., Hooper, P. R., Rehacek, J., Marsh, J. S. & Duncan, A.

R. (1997). The timing and duration of the Karoo igneous event,southern Gondwana. Journal of Geophysical Research 102, 18127^18138.

Eales, H. V., Marsh, J. S. & Cox, K. G. (1984). The Karoo IgneousProvince: an introduction. Special Publication of the Geological Society

of South Africa 13, 1^26.Eglington, B. M. & Armstrong, R. A. (2003). Geochronological and

isotopic constraints on the Mesoproterozoic Namaqua^Natal Belt:evidence from deep corehole intersections in South Africa.Precambrian Research 125, 179^189.

Eglington, B. M., Harmer, R. E. & Kerr, A. (1989). Isotope andgeochemical constraints on Proterozoic crustal evolution inSouth-East Africa. Precambrian Research 45, 159^174.

Elburg, M. & Goldberg, A. (2000). Age and geochemistry of Karoodolerite dykes from northeast Botswana. Journal of African Earth

Sciences 31, 539^554.Ellam, R. M. (2006). New constraints on the petrogenesis of the

Nuanetsi picrite basalts from Pb and Hf isotope data. Earth and

Planetary Science Letters 245, 153^161.Ellam, R. M. & Cox, K. G. (1989). A Proterozoic lithospheric source

for Karoo magmatism: Evidence from the Nuanetsi picrites. Earthand Planetary Science Letters 92, 207^218.

Ellam, R. M. & Cox, K. G. (1991). An interpretation of Karoo picriticbasalts in terms of interaction between asthenospheric magmasand the mantle lithosphere. Earth and Planetary Science Letters 105,330^342.

Ellam, R. M., Carlson, R. W. & Shirey, S. B. (1992). Evidence fromReÔs isotopes for plume^lithosphere mixing in Karoo floodbasalt genesis. Nature 359, 718^721.

Elliot, D. H., Fleming, T. H., Kyle, P. R. & Foland, K. A. (1999).Long-distance transport of magmas in the Jurassic Ferrar LargeIgneous Province, Antarctica. Earth and Planetary Science Letters 167,89^104.


23

at University of O



ownloaded from

http://petrology.oxfordjournals.org/cgi/content/full/egr011/DC1


Encarnacio¤ n, J., Fleming, T. H., Elliot, D. H. & Eales, H. V. (1996).Synchronous emplacement of Ferrar and Karoo dolerites and theearly breakup of Gondwana. Geology 24, 535^538.

Francalanci, L. (1989).Trace element partition coefficients for mineralsin shoshonite and calc-alkaline rocks from Stromboli Islands.NeuesJahrbuch fu« r Mineralogie, Abhandlungen 160, 229^247.

Frenkel, M. Y., Yaroshevsky, A. A., Ariskin, A. A., Barmina, G. S.,Koptev-Dvornikov, E.V. & Kireev, B. S. (1989). Convective cumu-late model simulating the formation process of strified intrusions.In: Bonin, B., Didier, J., Le Fort, P., Propach, G., Puga, E. &Vistelius, A. B. (eds) Magma^Crust Interactions and Evolution. Athens:Theophrastus, pp. 3^88.

Galerne, C. Y., Neumann, E.-R. & Planke, S. (2008). Emplacementmechanisms of sill complexes: Information from the geochemicalarchitecture of the Golden Valley Sill Complex, South Africa.Journal of Volcanology and Geothermal Research 177, 424^440.

Galerne, C. Y., Neumann, E.-R., Aarnes, I. & Planke, S. (2010).Magmatic Differentiation processes in saucer-shaped sills: evidencefrom the Golden Valley Sill in the Karoo Basin, South Africa.Geosphere 6(2), 1^26.

Gallagher, K. & Hawkerworth, C. (1992). Dehydration melting andthe generation of continental flood basalts. Nature 358, 57^59.

Ghiorso, M. S. & Sack, R. O. (1995). Extrapolation of liquid^solidequilibria in magmatic systems at elevated temperatures and pres-sures. Contributions to Mineralogy and Petrology 119, 197^212.

Gorring, M. L. & Naslund, H. R. (1995). Geochemical reversalswithin the lower 100m of the Palisades Sill, New Jersey.Contributions to Mineralogy and Petrology 119, 263^276.

Green, T. H., Blundy, J. D., Adam, J. & Yaxley, G. M. (2000). SIMSdeterminatioon of trace element partition coefficients betweengarnet, clinopyroxene and hydrous basaltic liquids at 1^7·5GPaand 1080^12008C. Lithos 53, 165^187.

Gre¤ goire, M., Bell, D. R. & le Roex, A. P. (2003). Garnet lherzolitesfrom the Kaapvaal craton (South Africa): trace element evidencefor a metasomatic history. Journal of Petrology 44, 629^657.

Gre¤ goire, M., Tinguely, C., Bell, D. R. & le Roex, A. P. (2005). Spinellherzolite xenoliths from the Premier kimberlite (Kaapvaalcraton, South Africa): Nature and evolution of the shallow uppermantle beneath the Bushweld complex. Lithos 84, 185^205.

Harmer, R. E., Lee, C. A. & Eglington, B. M. (1998). A deep mantlesource for carbonatite magmatism: evidence from the nephelinitesand carbonatites of the Buhera district, SE Zimbabwe. Earth and

Planetary Science Letters 158, 131^142.Hawkesworth, C. J., Marsh, J. S., Duncan, A. R., Erlank, A. J. &

Norry, M. J. (1984). The role of continental lithosphere in thegeneration of the Karoo volcanic rocks: evidence from combinedNd- and Sr-isotope studies. Special Publication of the Geological Societyof South Africa 13, 341^354.

Heinonen, J. S. & Luttinen, A. V. (2008). Jurassic dikes of Vestfjella,western Dronning Maud Land, Antarctica: Geochemical tracingof ferropicrite sources. Lithos 105, 347^364.

Heinonen, J. S. & Luttinen, A. V. (2010). Mineral chemical evidencefor extremely magnesian subalkaline melts from the Antarcticextension of the Karoo large igneous province. Mineralogy and

Petrology 99, 201^217.Heinonen, J. S., Carlson, R. W. & Luttinen, A. V. (2010). Isotopic

(Sr, Nd, Pb, and Os) composition of highly magnesian dikes ofVestfjella, western Dronning Maud Land, Antarctica: A key tothe origins of the Karoo large igneous province. Chemical Geology277, 227^244.

Hofmann, A. W. (1988). Chemical differentiation in the Earth: therelationship between mantle, continental crust and oceanic crust.Earth and Planetary Science Letters 57, 421^436.

Hopp, J., Trieloff, M., Brey, G. P., Woodland, A. B., Simon, N. S. C.,Wijbrans, J. R., Siebel, W. & Reitter, E. (2008). 40Ar/39Ar-ages ofphlogopite in mantle xenoliths from South African kimberlites:Evidence for metasomatic mantle impregnation during theKibaran orogenic cycles. Lithos 106, 351^364.

Huang, Y.-M., van Calsteren, P. & Hawkesworth, C. J. (1995). Theevolution of the lithosphere in southern Africa: A perspective onthe basic of granulite xenoliths from kimberlites in South Africa.Geochimica et Cosmochimica Acta 59, 4905^4920.

James, D. E., Niu, F. & Rokosky, J. (2003). Crustal structure of theKaapvaal craton and its significance for early crustal evolution.Lithos 71, 413^429.

Jamtveit, B., Svensen, H., Podladchikov, Y. Y. & Planke, S. (2004).Hydrothermal vent complexes associated with sill intrusions insedimentary basins. In: Breitkreutz, C. & Petford, N. (eds)Physical Geology of High-Level Magmatic Systems. Geological Society,

London, Special Publications 234, 233^241.Janney, P. E., Shirey, S. B., Carlson, R.W., Pearson, D. G., Bell, D. R.,

le Roex, A. P., Ishikawa, A., Nixon, P. H. & Boyd, F. R. (2010).Age, composition and thermal characteristics of South Africanoff-craton mantle lithosphere: Evidence for a multi-stage history.Journal of Petrology 51, 1849^1890.

Johnson, M. R., van Vuuren, C. J., Visser, J. N. J., Cole, D. I.,Wickens, H. deV., Christie, A. D. M., Roberts, D. L. & Brandl, G.(1997). The foreland Karoo Basin, South Africa. In: Selley, R. C.(ed.) African Basins. Sedimentary Basins of the World 3. Amsterdam:Elsevier, pp. 269^317.

Jourdan, F., Bertrand, H., Scha« rer, U., Blichert-Toft, J., Fe¤ raud, G. &Kampunzu, A. B. (2007). Major and trace elements and Sr, Nd,Hf, and Pb isotope compositions of the Karoo Large IgneousProvince, Botswana^Zimbabwe: lithosphere vs mantle plumecontribution. Journal of Petrology 48, 1043^1077.

Jourdan, F., Fe¤ raud, G., Bertrand, H.,Watkeys, M. K. & Renne, P. R.(2008). The 40Ar/39Ar ages of the sill complex of the Karoo largeigneous province: Implications for the Pliensbachian^Toarcianclimate change. Geochemistry, Geophysics, Geosystems 9(6), Q06009,doi:10.1029/2008GC001994.

Jourdan, F., Bertrand, H., Fe¤ raud, G., Le Gall, B. & Watkeys, M. K.(2009). Lithospheric mantle evolution monitored by overlappinglarge igneous provinces: case study in southern Africa. Lithos 107,257^268.

Kampunzu, A. B.,Tombale, A. R., Zhai, M., Bagai, Z., Majaule,T. &Modisi, M. P. (2003). Major and trace element geochemistry ofplutonic rocks from Francistown, NE Botswana: evidence for aNeoarchaean continental active margin in the Zimbabwe craton.Lithos 71, 431^460.

Klausen, M. B. (2009). The Lemombo monocline and associatedfeeder dyke swarm: Diagnostic of a successful and highly volcanicrifted margin.Tectonophysics 468, 42^62.

Konzett, J., Armstrong, R. A., Sweeney, R. J. & Compston,W. (1998).The timing of MARID metasomatism in the Kaapvaal mantle:An ion probe study of zircons from MARID xenoliths. Earth and

Planetary Science Letters 160, 133^145.Konzett, J., Armstrong, R. A. & Gu« nther, D. (2000). Modal meta-

somatism in the Kaapvaal craton lithosphere: constraints oftiming and genesis from U^Pb zircon dating of metasomatizedperidotites and MARID-type xenoliths. Contributions to Mineralogy

and Petrology 139, 704^719.Kuno, H. (1968). Differentiation of basalt magmas. In: Hess, H. H. &

Poldervaart, A. (eds) BasaltsçPoldervaart Treatise of Rocks of Basalt

Composition 2. NewYork: JohnWiley, pp. 623^688.Lana, C., Reimold, W. U., Gibson, R. L., Koeberl, C. &

Siegesmund, S. (2004). Nature of the Archean midcrust in the core


24

at University of O



ownloaded from


of the Vredefort Dome, Central Kaapvaal Craton, South Africa.Geochimica et Cosmochimica Acta 68, 623^642.

Latypov, R. M. (2003).The origin of basicûltrabasic sills with S-, D-,and I-shaped compositional profiles by in situ crystallization ofa single input of phenocryst-poor parental magma. Journal of

Petrology 44, 1619^1656.Latypov, R. M., Chistyakova, S. & Alapieti, T. (2007). Revisiting

problem of chilled margins associated with marginal reversals inmaficûltramafic intrusive bodies. Lithos 99, 178^206.

Le Bas, M. J., Le Maitre, R.W., Streckeisen, A. & Zanettin, B. (1986).A chemical classification of volcanic rocks based on the totalalkali^silica diagram. Journal of Petrology 27, 745^750.

Le Roux, J. P. (1993). Genesis of stratiform U^Mo deposits in theKaroo Basin of South Africa. Ore Geology Reviews 7, 485^509.

Li, A. & Burke, K. (2006). Upper mantle structure of southern Africafrom Rayleigh wave tomography. Journal of Geophysical Research 111,B10303, doi:10.1029/2006JB004321.

Lightfoot, P. C., Naldrett, A. J. & Hawkesworth, C. J. (1984).The geol-ogy and geochemistry of the Waterfall Gorge section of theInsizwa complex with particular reference to the origin of thenickel sulphide deposits. Economic Geology 79, 1857^1879.

Luttinen, A. V. & Furnes, H. (2000). Flood basalts of Vestfjella:Jurassic magmatism across an Archaean^Proterozoic lithosphericboundary in Dronning Maud Land, Antarctica. Journal of Petrology41, 1271^1305.

Luttinen, A. V., Ra« mo« , O. T. & Huhma, H. (1998). Neodymium andstrontium isotope and trace element composition of a MesozoicCFB suite from Dronning Maud Land, Antarctica: Implicationsfor lithosphere and asthenosphere contribution to Karoo magma-tism. Geochimica et Cosmochimica Acta 62, 2701^2714.

Marsh, J. S. & Eales, H. V. (1984). Chemistry and petrogenesis ofigneous rocks of the Karoo Central area, southern Africa. SpecialPublication of the Geological Society of South Africa 13, 27^68.

Marsh, J. S. & Mndaweni, M. J. (1998). Geochemical variations in along Karoo dyke, Eastern Cape. South AfricanJournal of Geology 101,119^122.

Marsh, J. S., Hooper, P. R., Rehacek, J., Duncan, R. A. & Duncan, A.R. (1997). Stratigraphy and age of the Karoo basalts of Lesothoand implications for correlations within the Karoo igneous prov-inces. In: Mahoney, J. J. & Coffin, M. F. (eds) Large Igneous

Provinces: Continental, Oceanic, and Planetary Flood Volcanism.

Geophysical Monograph, American Geophysical Union 100, 247^272.Marsh, J. S., Allen, P. & Fenner, N. (2003). The geochemical

structure of the Insizwa lobe of the Mount Ayliff Complex withimplications for the emplacement and evolution of the complexand its Ni-sulphide potential. South African Journal of Geology 106,409^428.

McDonough,W. F. & Sun, S.-S. (1995). The composition of the Earth.Chemical Geology 120, 223^253.

Nguuri, T. K., Gore, J., James, D. E., Webb, S. J., Wright, C.,Zengeni, T. G., Gwavava, O.Smoke, J. A. & Kaapvaal SeismicGroup (2001). Crustal structure beneath southern Africa and itsimplications for the formation and evolution of the Kaapvaal andZimbabwe cratons. Geophysical Research Letters 28, 2501^2504.

Pemberton, J. (1978). The geochemistry and petrology of Karoo basalts of

the Barkley East area, North Eastern Cape, MSc thesis, RhodesUniversity, Grahamstown,139 p.

Pin, C., Briot, D., Bassin, C. & Poitrasson, F. (1994). Concomitantseparation of strontium and samarium^neodymium for isotopeanalyses in silicate samples, based on specific extraction chroma-tography. Analytica Chimica Acta 298, 209^217.

Planke, S., Rasmussen, T., Rey, S. S. & Myklebust, R. (2005). Seismiccharacteristics and distribution of volcanic intrusions and

hydrothermal vent complexes in the V�ring and M�re basins.In: Dore¤ , A. G. & Vining, B. A. (eds) Petroleum Geology:

Northwestern Europe and Global PerspectivesçProceedings of the 6th

Petroleum Geology Conference. London: Geological Society,pp. 833^844.

Polteau, S., Mazzini, A., Galland, O., Planke, S. & Malthe-S�rensen, A. (2008). Saucer-shaped intrusions: occurrences, em-placement and implications. Earth and Planetary Science Letters 266,195^204.

Puffer, J. H. (2001). Contrasting high field strength element contents ofcontinental flood basalts from plume vs. reactivated-arc sources.Geology 29, 675^678.

Putirka, K. D. (2005). Igneous thermometers and barometers based onplagioclaseþ liquid equilibria: Tests of some existing and newcalibrations. American Mineralogist 90, 336^346.

Rehfeldt,T., Jacob, D. E., Carlson, R.W. & Foley, S. F. (2007). Fe-richdunite xenoliths from South African kimberlites: cumulates fromKaroo flood basalts. Journal of Petrology 48, 1387^1409.

Richard, P., Shimizu, N. & Alle' gre, C. J. (1976). 143Nd/l46Nd, a naturaltracer: an application to oceanic basalts. Earth and Planetary ScienceLetters 31, 269^278.

Riley, T. R. & Knight, K. B. (2001). Age of pre-break-up Gondwanamagmatism. Antarctic Science 13, 99^110.

Riley,T. R., Leat, P.T., Curtis, M. L., Millar, I. L. & Fazel, A. (2005).Early^Middle Jurassic dolerite dykes from western DronningMaud land (Antarctica): identifying mantle sources in the Karoolarge igneous province. Journal of Petrology 46, 1489^1524.

Riley, T. R., Curtis, M. L., Leat, P. T., Watkeys, M. K., Duncan, R.A., Millar, I. L. & Owens, W. H. (2006). Overlap of Karoo andFerrar magma types in KwaZulu^Natal, South Africa. Journal ofPetrology 47, 541^566.

Roswell, D. M. & De Swardt, A. M. J. (1976). Diagenesis in Cape andKaroo sediments, South Africa and its bearing on their hydrocar-bon potential.Transactions of the Geological Society of South Africa 79,81^153.

Rudnick, R. & Gao, S. (2003). Composition of the continental crust.In: Rudnick, R. I. (ed.) The Crust. Treatise on Geochemistry, Vol. 3.Oxford: Elsevier^Pergamon, pp. 1^64.

Schmitz, M. D. (2002). Geological and thermochronological evolution of the

lower crust od southern Africa, PhD thesis, Massachusetts Institute ofTechnology, Cambridge, MA, 269 p.

Schmitz, M. D. & Bowring, S. A. (2004). Lower crustal granuliteformation during Mesoproterozoic Namaqua^Natal collisionorogenesis, Southern Africa. South African Journal of Geology 107,261^284.

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 litho-spheric mantle. Journal of Petrology 48, 589^625.

Smith, P. M. & Asimow, P. D. (2005). Adiabat_1ph: A new publicfront-end to the MELTS, pMELTS, and pHMELTS models.Geochemistry, Geophysics, Geosystems 6, article no. Q02004, doi:10.1029/2004GC00816.

Spera, F. J. & Bohrson, W. A. (2004). Open-system magma cham-ber evolution: an energy-constrained geochemical model incorpor-ating the effects of concurrent eruption, recharge, variableassimilation, and fractional crystallization. Journal of Petrology 45,2459^2480.

Storey, B. C. & Kyle, P. R. (1997). An active mantle mechanism forGondwana breakup. In: Hatton, C. J. (ed.) Special Issue on the


25

at University of O



ownloaded from


Proceedings of the Plumes, Plates and Mineralization ’97 Symposium. South

AfricanJournal of Geology 100, 283^290.Sun, S.-S. & McDonough,W. F. (1989). Chemical and isotopic system-

atics of oceanic basalts; implications for mantle composition andprocesses. In: Saunders, A. D. & Norry, M. J. (eds) Magmatism in

the Ocean Basins. Geological Society, London, Special Publications 42,313^345.

Svensen, H., Planke, S., Malthe-S�renssen, A., Jamtveit, B.,Myklebust, R., Eidem, T. & Rey, S. S. (2004). Release of methanefrom a volcanic basin as a mechanism for initial Eocene globalwarming. Nature 429, 542^545.

Svensen, H., Jamtveit, B., Planke, S. & Chevallier, L. (2006).Structure and evolution of hydrothermal vent complexes in theKaroo Basin, South Africa. Journal of the Geological Society, London163, 671^682.

Svensen, H., Planke, S., Chevallier, L., Malthe-S�rensen, A.,Corfu, F. & Jamtveit, B. (2007). Hydrothermal venting of green-house gases triggering Early Jurassic global warming. Earth and

Planetary Science Letters 256, 554^566.Svensen, H., Bebout, G., Kronz, A., Li, L., Planke, S., Chevallier, L.

& Jamtveit, B. (2008). Nitrogen geochemistry as a tracer of fluidflow in a hydrothermal vent complex in the Karoo Basin, SouthAfrica. Geochimica et Cosmochimica Acta 72, 4929^4947.

Symonds, P. A., Planke, S., Frey, �. & Skogseid, J. (1998). Volcanicevolution of the Western Australian continental margin and its

implications for basin development. In: Purcell, P. G. &Purcell, R. R. (eds) The Sedimentary Basins of Western Australia 2.

Proceedings of the Exploration Society of Australia Symposium, Perth. West.Aust., Australia: Petroleum Exploration Society of AustraliaPerth, pp. 33^54.

Tait, S. R. & Jaupart, C. (1996). The production of chemically strati-fied and adcumulate plutonic igneous rocks. Mineralogical Magazine

60, 99^114.Veevers, J. J., Cole, D. I. & Cowan, E. J. (1994). Southern Africa:

Karoo Basin and Cape Fold Belt. In: Veevers, J. J. & Powell, C.McA. (eds) Permian^Triassic Pangean Basins and Foldbelts along the

Panthalassan Margin of Gondwanaland. Geological Society of America,

Memoirs 184, 223^279.Visser, J. N. J., Van Niekerk, B. N. & Van der Merwe, S. W. (1997).

Sediment transport of the Late Palaeozoic glacial Dwyka Groupin the southwestern Karoo Basin. South African Journal of Geology

100, 223^236.White, R. S. (1997). Mantle plume origin for the Karoo and

Ventersdorp flood basalts, South Africa. In: Hatton, C. J. (ed.)Special Issue on the Proceedings of the Plumes, Plates and

Mineralization ’97 Symposium. South African Journal of Geology 100,271^282.

White, R. & McKenzie, D. (1989). Magmatism at rift zones: the gen-eration of volcanic continental margins and flood basalts. Journalof Geophysical Research 94, 7685^7729.


26

at University of O



ownloaded from


Documents

Multistage Evolution of Dolerites in the Karoo Large Igneous …folk.uio.no/hensven/Neumann_JP2011_Karoo.pdf · 2011-04-13 · Multistage Evolution of Dolerites in the Karoo Large