Review - GeoScienceWorld

7 downloads 0 Views 2MB Size Report
Review. Formation of Archaean continental lithosphere and its diamonds: the root of the problem. D. G. PEARSON & N. WITTIG. Northern Centre for Isotopic and ...
Journal of the Geological Society, London, Vol. 165, 2008, pp. 895–914. Printed in Great Britain.

Review Formation of Archaean continental lithosphere and its diamonds: the root of the problem D. G . P E A R S O N & N. W I T T I G Northern Centre for Isotopic and Elemental Tracing, Department of Earth Sciences, Durham University, South Road, Durham DH1 4QE, UK (e-mail: [email protected]) Abstract: Cratonic lithospheric mantle plays an integral role in defining the physical behaviour of ancient continents and their mineral potential. Bulk compositional data show that modern-day melting residues from a variety of tectonic settings can be as depleted in Al and Ca as cratonic peridotites. Cratonic peridotites are strongly affected by secondary introduction of pyroxenes and garnet such that the extent and depth of melting cannot be reliably determined. Olivine compositions are probably the most reliable tracer of the original melting process and indicate that typical cratonic peridotites have experienced 40% or more melt extraction. Homogeneous levels of depletion indicated by olivine compositions, combined with mildly incompatible trace element evidence, indicate that melting took place at shallow depths, dominantly in the spinel stability field. Consideration of melt production models shows that shallow (,3 GPa) anhydrous melting is not capable of producing residues dominated by large degrees of melt extraction. Instead, a critical role for water is indicated, implicating the formation of cratonic peridotites within Archaean subduction zones. This melting occurred in the Neoarchaean in some cratonic blocks, initially forming dunitic residues that are still evident in the xenolith inventory of some cratons. Release and migration up-section of siliceous melt produced during orthopyroxene breakdown metasomatizes the proto-lithospheric via re-enrichment in orthopyroxene crystallizing from this hydrous Si-rich melt, forming the variably orthopyroxene-rich refractory harzburgites typical of most cratonic roots. Melting in Archaean subduction zones is followed by subduction stacking to form the cratonic root. Gravitational forces may then be responsible for the loss of imbricated mafic crust during periods of transient thermal and physical disturbances prior to final long-term tectonic stability. Most diamonds form in the base of these cratonic roots during pulses of thermal or tectonic activity, initially during root construction and subsequently associated with large-scale regional lithospheric events that may be correlated to pulses in global mantle dynamic evolution.

age of formation in relation to their associated crust, and we examine the timing of diamond formation in relation to the evolution of the lithosphere.

The stabilization of the first large robust continental land masses seems inexorably linked to the protective presence of thick, strong lithospheric roots. In addition to playing a critical role in defining the longevity of continents, through their unwillingness to be rifted apart, these roots beneath the continents play host to the Earth’s reserves of diamonds (Stachel & Harris 2008). The density, thermal structure and rheology of continental lithosphere also exercises a profound influence on continental dynamics, relative sea-level change and climate change. Hence, understanding the formation and evolution of continental roots is central to many issues is Earth Sciences. In this review we will focus primarily on the effects of melt depletion in cratonic lithospheric mantle, together with its consequences and implications for the formation and evolution of cratonic roots. For several decades the depth and environment of melting of cratonic lithosphere was a subject of fierce debate, with a multitude of models put forward ranging from deep plume melting (Boyd 1989; Griffin et al. 1999) to shallow melting and subduction (e.g. Schulze 1986; Canil & Wei 1992). More recently the consensus is moving towards the original melt extraction taking place at shallow mantle depths prior to subduction stacking (Canil 2004; Lee 2006; Simon et al. 2008). Below we outline recent and new constraints on the melting environments that produced the first continental roots, we address their

Cratons: records of the early history of the Earth The history of the early evolution of the Earth is contained within and beneath regions of continental crust in excess of 2.5 Ga old, known as cratons (Fig. 1). The oldest rocks within these cratons are 4 Ga old (Bowring et al. 1990), although zircons have been recovered that are older than 4.4 Ga, covering over 96% of Earth’s history (Fig. 2). Although no cratonic lithosphere has yet been found that approaches the age of the oldest zircons, the recent discovery of diamonds included in Hadean zircons (Menneken et al. 2007) has raised the possibility that deep continental material may have existed in parts of the Earth as early as 4.2 Ga ago. The oldest terrestrial record of melting within the mantle comes from very unradiogenic Os isotope ratios recorded in Re-poor, Os-rich osmiridium grains (Fig. 2). Some detrital grains within the Witwatersrand basin yield Os model ages of 4.1 Ga (Malitch & Merkle 2004) and are hence within 90% of the age of the Earth. We will return to the significance of these ancient events below. 895

896

D. G . P E A R S O N & N. W I T T I G

Fig. 1. Global distribution of the cratons (regions of crust .2.5 Ga old) modified after Bleeker (2003). Regional outcrops of Archaean crustal rocks are indicated in grey (those beneath Greenland are extrapolated under the ice cap) and other definable fragments of composite cratons in brown (from Bleeker 2003). The approximate outline of units that are relatively well defined as whole, composite cratons is shown by black dotted lines. Red dashed lines show the estimated extent of cratonic regions amalgamated from Archaean blocks during the Proterozoic. Blue dotted lines extended across oceanic areas show links between cratonic fragments that are thought to have once formed single cratonic blocks. NAC, North Atlantic Craton.

The nature of continental roots Xenoliths: a biased view? Several recent reviews have documented in detail the chemical and physical character of the rocks that make up continental roots (Menzies 1990; Pearson et al. 2003; Carlson et al. 2005; Lee 2006) and the reader is referred to those works for details. Here we will focus on a limited number of features relevant to the issues being addressed. Mineral concentrate data from kimberlite pipes indicate that the dominant rock type forming the lithosphere beneath cratonic crust is peridotite, with eclogite and pyroxenitic assemblages making up less than 2% of the volume (Schulze 1989). Although this estimate is consistent with seismological evidence it is important to realize the nature of the sample upon which our view is based. Almost all our samples of cratonic lithospheric mantle originate as xenoliths carried to the Earth’s surface by volcanic rocks, predominantly kimberlites (e.g. Pearson et al. 2003). The elemental chemistry and mineralogy of these samples are often strongly influenced by activity that is directly or indirectly related to the host volcanic rock and its precursory activity. Hence the likelihood is that the xenolith sample, in most instances, samples the wall rock of magma conduits that have been extensively modified by the host melt

prior to, during and after eruption. Hence the sample provided by xenoliths, or their disaggregated counterparts (mineral concentrates) is not necessarily representative of the lithosphere as a whole. Increasing evidence indicates that a significant proportion of the clinopyroxene and some garnet included in cratonic peridotites are of secondary origin, related to metasomatic events, either introducing the phases, in some cases relatively recently (Shimizu et al. 1997; van Achterbergh et al. 2001; Gre´goire et al. 2002; Simon et al. 2003), or significantly modifying their chemistry (e.g. Burgess & Harte 2004). Hence using the chemistry of these phases in concentrate form to characterize the properties of the cratonic lithospheric mantle and the processes that have affected it in entirety (e.g. Griffin et al. 1999) may provide a highly biased view. Nevertheless, the xenolith sample is in most cases the only sample of cratonic lithospheric mantle available for a given craton. The samples can still provide essential information regarding craton origin but, as we will document below, it is important to try to discern which chemical and mineralogical feature might result from the original depletion event (or events) and which are products of subsequent re-enrichment. Lastly, the compiled averages for compositional data presented in this paper focus on those cratons where there are enough samples to make meaningful statistical analyses. This means that peridotites from some cratons (e.g. North China) are

A R C H A E A N C O N T I N E N TA L L I T H O S P H E R E A N D D I A M O N D S

Fig. 2. Age ranges of the oldest terrestrial minerals and rocks compared with the average and oldest ages reported for Archaean lithospheric mantle. Diamond ages are from Menneken et al. (2007), oldest mantle melting event is from osmiridium data of Malitch & Merkle (2004) and Pearson et al. (2007). Open symbols for cratonic lithospheric mantle and oldest crust (Slave) show the mode in the statistical distributions, which is significantly less than the oldest ages.

897

Fig. 3. Averages and standard deviations (error bars) for olivine Mg numbers in low-temperature peridotites from various cratons and other, more recent samples from other tectonic settings. Shaded ovals on the North Atlantic craton (NAC) and Slave craton lines give the statistical modes for the most depleted peridotites (i.e. harzburgites and dunites). Shaded vertical oblong shows the apparent range for olivines from depleted cratonic peridotite. Literature data compiled by D. G. Pearson are available on request. Forearc data are for peridotites from Leg 125 (Ishii et al. 1992); OIB data are from Simon et al. (2008); abyssal average is taken from Niu et al. (2004); Iwanaidake peridotite data are presented separately for harzburgites and dunites (Kubo 2002). Diamond inclusion data are from Stachel & Harris (2008).

not evaluated and discussed because of the as yet relatively small but growing database for these areas (e.g. Gao et al. 2002; Menzies et al. 2007).

Melt depletion and chemical buoyancy Cratons have, by definition, remained convectively stable for periods of time in excess of 2.5 Ga. This fact, together with the lack of observable gravity excess over these areas of anomalously low heat flow (Pollack & Chapman 1977), led Jordan (1975) to construct the isopycnic hypothesis. The isopycnic hypothesis states that the negative buoyancy imparted by the low temperature of the cratonic lithospheric mantle relative to convecting mantle is offset by the positive buoyancy generated by depletion of the cratonic lithospheric mantle in melt-forming elements. Peridotite that has experienced extensive melt depletion has a significantly lower intrinsic density than more fertile, undepleted mantle (Boyd & McAllister 1976; Lee 2003). One of the characteristic features of cratonic mantle is the dominant signature of melt depletion, evident in a number of parameters such as their very high average Mg number of olivine (Fig. 3), elevated bulk MgO contents and low bulk Al and Ca contents (Fig. 4). The important elements from a density perspective are Fe and Al, the latter because of its requirement to make the dense mineral garnet. The significant depletion of these elements with respect to fertile mantle (Fig. 5) is responsible for cratonic lithospheric mantle densities being significantly lower than that of the ambient convecting mantle. The standard temperature and pressure (STP) density difference between fertile peridotite with a bulk Mg number of 88 and typical cratonic lithospheric mantle with a bulk Mg number of 92 is c. 1.5% (3390 kg m3 v.

3340 kg m3 ), rising to .2.5% for very depleted residues with Mg numbers approaching 94. According to the isopycnic hypothesis, cratons are effectively in isostatic equilibrium as a result of the thermal and chemical buoyancy factors cancelling each other out at any given depth within the lithosphere. Although the evidence for this has been frequently discussed since the inception of isopycnic theory, the consensus is that conditions are reasonably close to those required for neutral buoyancy (e.g. Lee 2003), with scatter and uncertainty resulting from the likelihood that the xenoliths are derived from magma conduit wall rocks and are not fully representative of the density of the bulk of the cratonic lithospheric mantle. It has been generally assumed that the longevity of cratonic lithosphere is a direct result of its anomalous degree of melt depletion compared with other peridotites of younger age (e.g. Griffin et al. 1999). Whereas this is true if cratonic peridotites are compared with abyssal or orogenic peridotites, Boyd (1989, 1996), Simon et al. (2008) and the extensive compilation in Figure 4 show clearly that peridotites from modern-day forearcs, ophiolites and even some ocean island basalts (OIBs) are more depleted in CaO and Al2 O3 than most cratonic peridotites. Hence, for these elements, there is no clear temporal evolution in the degree of depletion of melting residues. If differences in olivine Mg number are interpreted as resulting from differences in the levels of melt depletion in peridotites (see below) then a clear difference does exist between cratonic samples and most peridotites from other tectonic settings (Boyd & Mertzman 1987; Pearson et al. 2003; Bernstein et al. 2006a; Lee 2006; Fig. 3).

898

D. G . P E A R S O N & N. W I T T I G

Fig. 4. Average and standard deviation of bulk-rock Al2 O3 and CaO for low-temperature cratonic peridotites compared with present-day peridotites from other tectonic settings (forearcs, OIBs, abyssal), orogenic peridotites, ophiolites and ‘off-craton’ peridotite xenoliths predominantly from continental rifts. NAC, North Atlantic craton (East and West Greenland); Churchill (broadly equivalent to Rae in Fig. 1), peridotites from Somerset Island, Churchill Province, Northern Canada. Cratonic datasets are from Boyd & Mertzman (1987), Boyd et al. (1997), Bernstein et al. (1998), Lee & Rudnick (1999), Schmidberger & Francis (1999), Kopylova & Russell (2000), Bizzarro & Stevenson (2003), Irvine et al. (2003), Pearson et al. (2004), Wittig et al. (2008b), and Boyd (unpubl. data). Orogenic peridotites are from a dataset compiled by Bodinier & Goddard (2003). Off-craton data are from a compilation by Canil (2004). Forearc data are from Ishii et al. (1992), abyssal data are from Niu et al. (2004); OIB data are taken from the Canaries dataset of Neumann et al. (1995) and are meant to be representative of the more depleted OIB peridotite endmember rather than OIB peridotites as a whole (for review, see Simon et al. 2008). Ophiolite data are from Jaques & Chappell (1980) and Marchesi et al. (2006).

None the less, few studies other than that of Boyd (1996) have recognized that the modern-day Earth is capable of producing residues, in ophiolites or forearcs, that are as depleted in major elements as any cratonic peridotite, or more depleted (Fig. 3). These ultra-depleted residues will have densities as low as or lower than cratonic peridotites and this may explain why their fate is to be obducted not subducted; this implication has interesting implications for models of craton root formation in the Archaean Earth and also for forming new stable continental keels today (Jackson et al. 2008).

The composition of archetypal cratonic mantle lithosphere: what are the primary phases? To extract genetic information from cratonic lithospheric mantle peridotites a clear understanding of the effects of mantle metasomatic processes on the bulk composition and mineralogy of the rocks is necessary. Only recently have petrological and geochemical studies been able to resolve most of the likely metasomatic effects, in particular the impact on mineralogy. In this regard, the addition of minerals such as clinopyroxene, orthopyroxene and garnet needs to be addressed. Possible addi-

Fig. 5. Mg/Si v. Al/Si systematics for low-temperature cratonic peridotites compared with present-day peridotites from other tectonic settings (forearcs, OIBs (Canary Island only), abyssal), orogenic peridotites, ophiolites and ‘off-craton’ peridotite xenoliths predominantly from continental rifts. Data sources are as in Figure 4. Dark grey arrows for orthopyroxene (Opx) clinopyroxene (Cpx) and garnet (Gt) addition start from average North Atlantic craton (NAC) and are for typical low-T Kaapvaal peridotite mineral compositions (Boyd & Mertzman 1987). Light green large arrow labelled ‘high-T Opx’ is the vector for addition of high-temperature aluminous orthopyroxene estimated by Saltzer et al. (2001). Dashed black line is the ‘geochemical fractionation’ line of Jagoutz et al. (1979). The displacement to low Mg/Si in Kaapvaal peridotites and Phanerozoic ophiolites should be noted.

tion of any of these phases will potentially severely disturb both major and moderately incompatible trace element systematics in a suite of peridotites. This creates conflicting signatures as to the depth and environment of melt extraction. Investigating these issues is most reliably done by following the approach of Boyd & Mertzman (1987), who emphasized the importance of using large (preferably over 500 g) samples for the preparation of analytical powders, to minimize scatter from sample inhomogeneities. Alternatively, where point counting is used it is essential to follow procedures for obtaining statistically significant data to estimate reliable modal abundances (MacKenzie & Canil 1999). Early studies of modal mineralogy showed that a significant proportion of Kaapvaal peridotites contained more orthopyroxene than could be accounted for by partial melt extraction alone (Boyd 1989; Canil 1992). This was confirmed by further experimental petrology (Walter 1998; Herzberg 1999). Kesson & Ringwood (1989) first advocated a secondary origin for the excess orthopyroxene in Kaapvaal peridotites, via the addition of SiO2 rich fluids in an Archaean subduction zone. Recent major element, trace element and isotopic evidence (Kelemen et al. 1998; Lee 2006; Simon et al. 2008) indicates a secondary origin for orthopyroxene in Kaapvaal peridotites. The addition of orthopyroxene to Kaapvaal peridotites imparts their well-known Si-rich characteristic, which is also a feature, albeit to a lesser extent, of some Siberian and Tanzanian peridotites (Fig. 5). The addition of orthopyroxene to a more olivine-rich protolith is now widely regarded as the origin for the Si-rich nature of the Kaapvaal peridotites (Kelemen et al. 1992, 1998; Simon et al.

A R C H A E A N C O N T I N E N TA L L I T H O S P H E R E A N D D I A M O N D S

2008). A combination of the measured orthopyroxene, clinopyroxene and garnet compositions can explain the systematic shift of cratonic peridotites away from the magmatic fractionation line in Mg/Si–Al/Si space (Fig. 5). These are low-T phases whose compositions reflect conditions considerably below the peridotite solidus. An excellent fit to the Mg/Si–Al/Si cratonic trend is produced by addition of pre-exsolution high-temperature aluminous pyroxene of the type suggested by Saltzer et al. (2001; Fig. 5). The addition of orthopyroxene to depleted harzburgite can also explain the anomalously low FeO of Kaapvaal peridotites (Walter 2003; Lee 2006; Simon et al. 2008), as addition of substantial orthopyroxene dilutes bulk-rock FeO contents (Fig. 6). Ophiolites that have experienced extensive serpentinization have anomalously low Mg/Si, comparable with Kaapvaal peridotites (Fig. 5). This is caused by Mg loss during alteration. The distinctively higher FeO contents of these highly altered oceanic rocks compared with Kaapvaal peridotites (Fig. 7) make it unlikely that the silica enrichment or low Ca contents of some Kaapvaal peridotites (Schulze 1986) could originate by serpentinization of their protoliths. Kaapvaal low-temperature peridotites are so orthopyroxene rich that their average olivine contents fall far outside average values for other cratons (Fig. 8). In contrast, cratonic peridotites from the North Atlantic craton (Greenland) are poor in orthopyroxene, with modal compositions that are dominated by olivine, yet with very similar olivine Mg numbers (Fig. 8; Bernstein et al. 2006a). In particular, peridotite xenoliths from East Greenland (Bernstein et al. 1998) and Ubekendt Ejland, West Greenland (Bernstein et al. 2006b) are very olivine rich and closely approximate both the olivine Mg number and olivine content of highly depleted dunites from the Iwanaidaike peridotite massif, Japan (Figs 8 and 9). When other samples from West Greenland are considered (Bizzarro & Stevenson 2003; Wittig et al. 2008a) the average olivine content is still very high (.89 wt%) and the

Fig. 6. Bulk-rock FeO v. MgO for cratonic peridotites. The five curves forming a cluster are the experimentally derived batch melting trends of Herzberg (2004) for melting between 1 and 7 GPa. Also plotted are the fields for olivine compositions, Kaapvaal orthopyroxene, garnet, clinopyroxene plus amphibole compositions, and high-temperature experimental orthopyroxenes. Data sources are as for Figures 3 and 4. Only data from low-temperature peridotites are plotted in all cratonic examples.

899

Fig. 7. Bulk-rock FeO v. Mg/Si for Kaapvaal peridotites (data sources as in Fig. 4) and serpentinized ophiolitic peridotites from Cuba (Marchesi et al. 2006). The similar range in Mg/Si but elevated FeO of the ophiolitic peridotites should be noted. Only data from low-temperature peridotites are plotted for Kaapvaal samples.

Fig. 8. Average olivine Mg number v. weight per cent olivine for peridotites from different cratons and compared with peridotites from various present-day tectonic settings (forearcs, OIBs (Canary Islands only), abyssal), Iwanaidake massif (ophiolites containing dunites and harzburgites), Hokkaido, Japan (Kubo 2002). Field for orthopyroxenepoor ‘dunites’ from East Greenland (Bernstein et al. 1998) and Ubekendt Ejland, West Greenland (Bernstein et al. 2006b) is marked as a dashed box. Other West Greenland peridotites are more orthopyroxene rich (Bizzarro et al. 2002; Wittig et al. 2008b) and influence the less olivinerich North Atlantic craton average. Other data sources are as for Figures 3 and 4. Only data from low-temperature peridotites are plotted in all cratonic examples.

North Atlantic craton remains the most orthopyroxene-poor endmember of the cratonic suites, plotting close to the ‘oceanic trend’ of Boyd (1989; Fig. 8). The Tanzania (Lee et al. 1999) and Slave (Kopylova & Russell 2000) cratons are progressively

900

D. G . P E A R S O N & N. W I T T I G

more olivine poor whereas the Kaapvaal craton is the extreme orthopyroxene-rich endmember with only 64 wt% olivine on average (Fig. 8). Interestingly, the harzburgites of the Iwanaidaike ophiolite massif are the closest average compositions to North Atlantic craton peridotites in terms of modal olivine and olivine Mg number, with the depleted dunites from Iwanaidaike being very similar in composition to the Greenland dunites from the North Atlantic craton (Kubo 2002; Figs 8 and 9). This similarity between some cratonic peridotites and those of the Iwanaidaike massif, a product of Mesozoic partial melting, in a suprasubduction zone environment (Kubo 2002), has significant implications for the origins of cratonic peridotites in general and this will be discussed in a later section. The close compositional similarity of average cratonic peridotites (excluding Kaapvaal samples) to recent melt residues suggests that these relatively orthopyroxene-poor peridotites, highly depleted in Al and Ca, with high olivine Mg numbers, are likely to be the more archetypal components of cratonic lithospheric roots. In contrast, the Kaapvaal peridotites represent an extreme endmember. None the less, the scatter in olivine contents even in the North Atlantic craton peridotites (Fig. 9) may be a product of some orthopyroxene addition. Other processes that act to further obscure the chemical signature of the original depletion event (or events) carried by cratonic peridotites are the addition of modal clinopyroxene and garnet. There is also now clear evidence for a secondary origin for much of the clinopyroxene within cratonic peridotites. Addition of clinopyroxene as a metasomatic phase to cratonic peridotites has been suggested on the basis of the varied, generally enriched trace element patterns of the pyroxenes (e.g. Shimizu et al. 1997). Pearson et al. (2002) showed that the abundance of clinopyroxene in Kaapvaal cratonic residues is highly variable and significantly exceeds that expected for residual rocks with such high Mg numbers, even when the effects

of sub-solidus exsolution are accounted for. This indicates that most depleted cratonic peridotites contain excess clinopyroxene. This observation is extended to all cratonic peridotites in Figure 10, and is in contrast to the very low (2% and less) clinopyroxene abundances evident in similarly depleted harzburgites and dunites from the Iwanaidake peridotite massif. A clear indication that the excess clinopyroxene may have been introduced relatively recently, and cannot have all exsolved from a higher temperature orthopyroxene (e.g. Cox et al. 1987; Saltzer et al. 2001), comes from the lack of isotopic equilibrium between the clinopyroxene and coexisting garnet (Simon et al. 2003) together with Sr and Nd isotope compositions that are close to those of present-day Bulk Earth in many instances (Pearson & Nowell 2002). Most ancient isotopic signatures in these cratonic peridotites reflect enrichment rather than depletion (e.g. Menzies & Murthy 1980; Pearson & Nowell 2002). Although Cr-pyrope garnet is a mineral thought by many to characterize cratonic peridotites, its presence and composition depend on both the depth of the lithosphere at the time of xenolith sampling and the degree of depletion and subsequent metasomatism. Smith et al. (2008) have recently described a suite of very depleted chromite-bearing dunites and harzburgites from the Murowe mine, Zimbabwe, that appear to originate from depths within the diamond stability field. In these peridotites garnet is a scarce phase, presumably because of the high-pressure stability of high-Cr chromite and the lack of subsequent metasomatic addition that may have lowered the bulk-rock Cr number to the point where garnet was again stable. Experimental petrology (Ringwood 1977; Canil & Wei 1992; Stachel et al. 1998) has also shown that high-Cr, low-Ca garnets with high knorringite component cannot be produced as residual phases on the liquidus of cratonic peridotite and hence must be metamorphic products of the reaction between orthopyroxene and chromite. The origin of more typical Ca-saturated Cr-pyropes is less

Fig. 9. Olivine Mg number v. modal olivine for peridotites from the North Atlantic craton (Greenland; see Fig. 8 for data sources) compared with the harzburgite–dunite sequence from the Iwanaidake peridotite (Kubo 2002). Squares show average values for abyssal (A; Niu et al. 2004) and Kaapvaal (K) low-temperature peridotites (from Fig. 8). Only data from low-temperature peridotites are plotted in all cratonic examples.

Fig. 10. Olivine Mg number v. modal clinopyroxene content for cratonic peridotites (data sources as for Figs 3 and 4) compared with the harzburgite–dunite sequence from the Iwanaidake peridotite (Kubo 2002). The similar range in olivine Mg numbers for the two suites but the greatly contrasting variation in modal clinopyroxene should be noted. Only data from low-temperature peridotites are plotted in all cases.

A R C H A E A N C O N T I N E N TA L L I T H O S P H E R E A N D D I A M O N D S

clear, although in many samples such garnets have a complex trace element signature and very few have the highly radiogenic Nd and Hf or very unradiogenic Sr isotope systematics that would be consistent with an origin as ancient residual phases (e.g. Pearson & Nowell 2002). Although it may be argued that more recent exsolution of clinopyroxene and garnet from a highT aluminous orthopyroxene may reduce the likelihood of finding ancient residual isotopic compositions in these phases, the nature of Sm–Nd and Lu–Hf partitioning in orthopyroxene means that any exsolved garnet and clinopyroxene would inherit a depleted isotopic composition from their parent phase and following exsolution would rapidly evolve to compositions significantly more depleted than the Bulk Earth mean values observed (Pearson & Nowell 2002). In some peridotites textural and chemical evidence have combined to invoke a metasomatic origin for garnet (Bell et al. 2005). The last phase to add to the list of minerals that, in some circumstances, may not be of primary origin in cratonic peridotites is olivine itself. Olivines of low Mg number (86–88) have been identified in deformed dunitic peridotites from the Kimberley area of the Kaapvaal craton (Boyd et al. 1983) and also in some Greenland peridotites (Wittig et al. 2008a). These rocks appear to have experienced significant olivine addition and are hence strongly modified from their original solidus compositions. Whereas the Kimberley dunites are rare examples of this phenomenon, olivine addition may also account for the more olivine-rich nature of Kaapvaal high-T peridotites (Kelemen et al. 1992). Overall it is clear that, despite their simple mineralogy, cratonic peridotites are rocks of extreme complexity, in which every major mineral phase, to greater or lesser extents, may have been modified or introduced by melt–fluid interaction. This multi-stage history obscures much of their early history and it is thus no surprise that their origin has been so difficult to constrain.

901

from metasomatic activity for considerable time periods (Bernstein et al. 2006a). Modern high-quality phase equilibria studies allow the change in mineral chemistry of residual phases to be correlated to the extent of melting experienced (Fig. 11; e.g. Baker & Stolper 1994; Walter 1998). This relationship was exploited by Bernstein et al. (2006a) to estimate the amount of melt extraction experienced by typical cratonic peridotites to be c. 40%, based on their very consistent, high olivine Mg numbers, which average 92.8 from their database. Our own compilation of the literature data (Fig. 3) reveals some differences from the averages recorded by Bernstein et al. (2006a), either because of increased available data (in the case of the North Atlantic craton and the Kaapvaal craton) or perhaps because we have chosen to include a wider variety of rock types, such as for the Slave craton, where the statistical distribution of olivine Mg numbers is clearly bimodal (Fig. 12). For Slave cratonic peridotites the lower olivine Mg numbers belong to lherzolites whereas olivines with the highest Mg numbers belong to the ‘ultra-depleted layer’ thought to consist predominantly of harzburgite. This contrasts with the mixed lherzolite–harzburgite suite from the Kaapvaal craton in which many lherzolites have olivine Mg numbers that are just as high as those of typical harzburgites. None the less, the low Mg number tail in the distribution of Kaapvaal olivines is dominated by lherzolites (Fig. 12). Despite some interesting inter-craton variations in the statistical distribution of olivine Mg numbers, we concur with Bernstein et al. (2006a) that although modal mineralogy varies widely, there is remarkable consistency in the average olivine Mg numbers for cratonic peridotites (Figs 3 and 12), the modes of which vary from 92.5 to 92.8 for every craton. Even for cratons

Extent of melting As outlined above, cratonic peridotites contain extensive evidence for the addition of orthopyroxene, clinopyroxene and most probably some garnet. Because modal mineralogy varies strongly depending on equilibration depth, it is not possible to use modal abundance data alone to assess the likely extent of melting. Although bulk chemistry approaches such as the significant depletion of CaO and Al2 O3 clearly indicate extensive melt extraction, the abundances of these elements are also sensitive to the addition of clinopyroxene and garnet in particular. However, at any melting depth, mantle peridotites with very high olivine contents (.90%) probably represent residues from very extensive melt extraction (40% or more), as long as such rocks contain highly magnesian olivines. In fact, the chemistry of olivine may be the most robust indicator of the extent of melting in a peridotite. Despite the evidence of later olivine introduction into some cratonic peridotites noted above, in the majority of cratonic peridotites it seems likely that olivine compositions might be relatively faithful recorders of the melting history of the rock. The Mg number of olivine in particular is relatively insensitive to the introduction of high Mg number phases such as orthopyroxene (e.g. Kelemen et al. 1992, 1998). Another indication of the robustness of olivine compositions to later compositional changes is the observation that the average Mg number of olivines included in diamonds (Fig. 3) is very similar (92.8) to that of the averages for peridotites from numerous other cratons (92.5) even though the olivine in diamond has been armoured

Fig. 11. Mg number of olivine v. per cent melting (F) in residual peridotites from experiments at varying pressures and starting compositions. Error bars give estimated experimental uncertainties on the melt fraction. Source of experiments: Baker & Stolper 1994; Walter 1998. Arrows mark the changes in melting reactions denoted by clinopyroxene-out and orthopyroxene-out. It should be noted that this plot is equivalent to that of figure 3 of Bernstein et al. (2006a) except for the exclusion of some experimental data and the inclusion of uncertainties on F.

902

D. G . P E A R S O N & N. W I T T I G

Fig. 13. Summary of possible effects of metasomatism on olivine Mg number v. modal olivine relations for cratonic peridotites (s). Arrows for different stages relate to: (1) melt extraction, probably accompanied by melt infiltration (Niu et al. 2004); (2) melt extraction where orthopyroxene melts incongruently leaving an olivine-rich residue; (3) orthopyroxene addition at constant olivine Mg number (horizontal trend; major arrow), or orthopyroxene  clinopyroxene and garnet addition at varying olivine Mg number (diagonal trends; minor arrows); (4, 5) melt addition creating higher or lower olivine Mg numbers; (6) melt addition creating increased olivine Mg number plus increased modal orthopyroxene  clinopyroxene and garnet.

Fig. 12. Probability density plots of olivine Mg numbers for peridotites from the Kaapvaal craton, the Slave craton and the North Atlantic craton (West and East Greenland). Only data from low-temperature peridotites are plotted in all cases. A constant uncertainty of 0.2 is used for all olivine Mg numbers. Number of analyses plotted is given as n in curve labels. Shaded oblongs at the top of the diagram show the averages of OIBs (Canary Islands only), abyssal peridotites and the Papuan ophiolite belt. Data for the Iwanaidake ophiolite includes dunites and harzburgites. Curve labelled ‘all cratons’ is a compilation of all the cratons listed in Figure 3. Forearc data are from Leg 125 peridotites (Ishii et al. 1992). Other data sources are as in Figure 3.

where only small databases are available for cratonic peridotite xenoliths, such as the North China craton (Gao et al. 2002), the average Mg number of olivine is 92.6 (Bernstein et al. 2006a). Both the very olivine-rich East Greenland peridotites and the olivine-poor Kaapvaal peridotites have very similar olivine Mg numbers and so orthopyroxene addition does not appear to have been the most significant factor dictating olivine chemistry (Figs 8 and 9). The orthopyroxene addition vector is therefore subhorizontal on plots of olivine Mg number v. modal olivine (Fig. 13) whereas other processes (e.g. olivine, clinopyroxene or melt addition) may increase or decrease olivine Mg numbers and hence create scatter. Bernstein et al. (2006a) suggested that an olivine Mg number of 92.6 marks the orthopyroxene ‘out’ stage in the melting of peridotite (Stage 2 of Fig. 13). Although the constancy of the olivine compositional modes for cratonic peridotites may support this view, there are a small number of samples with both higher and lower olivine Mg numbers. These

variations may be caused by melt–rock interaction, with the effect on olivine and modal compositions depending upon the melt–rock fraction (Bodinier & Godard 2003; Fig. 13). Overall, the main trend of constant olivine Mg number at widely varying modal olivine composition is consistent with the idea that cratonic peridotites melted to the stage of orthopyroxene ‘out’ (Stage 2; Fig. 13) and then experienced later, varying degrees of orthopyroxene addition (Bernstein et al. 2006a). This multi-stage model implies that most cratonic peridotite residues (at least those with olivine Mg numbers close to 92.6) were dunites at some stage. This model is supported by the statistical distribution of olivine compositions in the Iwanaidake peridotite massif. Olivines from this mixed dunite–harzburgite massif have a bimodal distribution (Fig. 12) in which the main mode at Mg number 92.5 belongs to dunites (Fig. 3) and is identical to cratonic peridotites. Bernstein et al. (2006a) suggested that the clear relationship between olivine Mg number and extent of melting observed in experiments indicates that cratonic peridotites are the residues of 40% partial melt extraction. Our plotting of the experimental database suggests that there is some uncertainty in this estimate, depending on the depth and hence pressure at which the melting took place (Fig. 11). Some difficulty arises because of the different starting compositions used by experimentalists. Using experiments performed at different pressures and with the same starting composition (e.g. Walter 1998) it is evident that higher pressure melt extraction more rapidly depletes Fe in residues leading to higher olivine Mg numbers at a given degree of melting (compare regressions for 3 and 7 GPa, Fig. 11). This means that for an average olivine Mg number of 92.6 the degree of melt extraction is estimated to be between 37 and 47%, depending on the pressure of melt extraction. Despite this uncertainty, the indicated extent of melt extraction is clearly

A R C H A E A N C O N T I N E N TA L L I T H O S P H E R E A N D D I A M O N D S

exceptional and rarely achieved in the modern-day Earth except in forearc environments where previously depleted residues are remelted by water fluxing and decompression (Parman & Grove 2004; Fig. 3; see ‘Environment of melting’). Even for the significant tail of Kaapvaal peridotites with olivine Mg numbers of around 92, melting extents are between 27 and 36% (i.e. still very high) if these compositions are primary and not lowered by metasomatism. The power of using olivine compositions to assess levels of melt depletion in peridotites is apparent when compared with bulk analytical data. Average CaO and Al2 O3 contents of cratonic peridotites are certainly low, being significantly less than those for most abyssal peridotites. However, some xenoliths erupted by OIBs such as at Gran Canaria (Neumann et al. 1995) also have very low CaO and Al2 O3 contents that are more depleted than in most cratons (Boyd 1996; Simon et al. 2008; Fig. 4). In contrast, the olivine compositions of these OIB peridotites are significantly less magnesian (91.4  0.6) than those of average cratonic peridotites (Fig. 12), supporting the view that the bulk Ca and Al contents of cratonic peridotites have been enhanced by later metasomatism. We therefore believe that olivine compositions, used with caution, offer the most reliable estimate of the degree of melt depletion in peridotites. Having established that the statistical distribution of olivine compositions for cratonic peridotites is generally tightly unimodal (with the exception of the Slave craton), it is important to establish the melting environment and the depth of melting that may lead to such a distribution, because this will have an important bearing on the tectonic environment in which cratonic roots are formed. Both theoretical and experimental approaches have been used to quantify the melting behaviour of peridotite (e.g. McKenzie & Bickle 1988; Iwamori et al. 1995; Herzberg 2004). The results from the two approaches are broadly in agreement. Polybaric melting beginning at high pressure (e.g. between 5 and 7 GPa) at a high potential temperature of 1650 8C produces a large fraction of melting over a relatively small

Fig. 14. Fraction of melting v. depth or pressure for the polybaric melting of peridotite. Shaded regions bounded by dashed blue (fractional) and continuous green (batch) lines are for theoretical melting paths calculated by Iwamori et al. (1995). Continuous black lines are melting trajectories plotted from the experimental data summarized by Herzberg (2004). Potential temperatures of melting trajectories are indicated (in 8C). Shaded vertical oblong denotes the region of the garnet–spinel transition.

903

pressure drop (Fig. 14). In contrast, melting at lower potential temperatures, which initiates melting at lower pressures (4 GPa or less), is more linear in nature. Hence, of the two scenarios (deep, high-pressure melting or shallow, low-pressure melting), high-pressure melting may be more likely to yield residues where the majority have experienced large fractions of melt depletion

Fig. 15. Whole-rock MgO (wt%) v. Al2 O3 (wt%) for cratonic peridotites. Continuous black lines are the melting trajectories for various pressures (2, 3, 5, 7, 10 GPa) taken from Herzberg (2004). Melt extraction equiline of 30% is shown (larger extents of melting are to the right of the line). Field for ‘opx’ depicts the range in orthopyroxene compositions observed in low-T Kaapvaal peridotites (Boyd & Mertzman 1987; Simon et al. 2008). Field for ‘high T opx’ is for liquidus unexsolved orthopyroxenes from data sources summarized by Saltzer et al. (2001) and Simon et al. (2008). Whole-rock analytical data sources are as for Figure 4.

Fig. 16. Whole-rock Lu (ppm) v. Yb (ppm) elemental systematics in cratonic peridotite suites. Curves are shown for polybaric fractional melting beginning at 2, 3 and 7 GPa and isobaric melting at 7 GPa following melting models described by Wittig et al. (2008a). Highpressure (7 GPa) isobaric or polybaric melting would produce residues lying to the high Lu and Yb side of the Primitive Upper Mantle (PRIMA) reference point. No cratonic peridotites are displaced in this direction. Data sources are indicated in the diagram. Data sources are: Wittig et al. (2008a), Simon et al. (2007), Schmidberger & Francis (1999) and Friend et al. (2002).

904

D. G . P E A R S O N & N. W I T T I G

and therefore have similar olivine Mg numbers. Only melting at potential temperatures in excess of 1600 8C in either low- or high-pressure regimes can achieve melt fractions in excess of 40%, the likely average experienced by cratonic peridotites (Fig. 11). Even in these circumstances, if melting is occurring in a ridge or plume environment, only a small fraction of the residues, located in the axial part of the melting column, will experience these very high levels of melt depletion. To further constrain which melting scenario and hence tectonic environment can best explain the statistical distribution of olivine compositions we will now try to better understand the likely depth of melting experienced by cratonic peridotites.

of orthopyroxene, clinopyroxene and/or garnet (Fig. 15). Some of the spread of the East Greenland data towards the high-pressure melting trends in Al2 O3 –MgO space could be the result of minor olivine addition that is a likely result of melt–rock interaction (e.g. Kelemen et al. 1998). In contrast, a shift towards lowerpressure melting trends may be caused by extensive serpentinization of peridotites, inducing Mg loss and a relative increase in the concentration of Al (e.g. Niu 2004). From these considerations it is clear that major element approaches to constraining the pressure of melt depletion experienced by cratonic peridotites are not likely to be accurate because major elements are relatively insensitive to melting depth but very sensitive to modal metasomatism.

Depth of melting Trace element considerations Major element considerations Although it has long been established that the cratonic lithospheric mantle is highly depleted in melt-forming elements relative to fertile mantle (Nixon & Boyd 1973; Maaloe & Aoki 1977; Boyd 1989), the depth at which this melting took place has been a far more contentious issue. This question is of paramount importance in establishing an accurate picture of craton formation and there have been a number of attempts to constrain the issue over the last 20 years. The initial approaches were largely based around bulk-rock major elements (Schulze 1986; Boyd 1989; Walter 1998, 1999, 2003) whereas more recently the usefulness of moderately incompatible trace elements has been realized (Kelemen et al. 1998; Canil 2004; Lee 2006; Wittig et al. 2008b). Even when plotting only bulk compositions that have been reliably estimated from large samples, the effects of metasomatism on major element bulk compositions cause some significant confusion in extracting depth of melting information. For instance, when cratonic peridotites are compared with the major element compositions of melt residues predicted from experiments, the data scatter across melting trends for both high- and low-pressure melting (Figs 6 and 15). A case has been made for using FeO as a melting depth proxy in some non-cratonic peridotites that have experienced minimal metasomatism (Ionov & Hofmann 2007) but for cratonic rocks this is not practical. The low FeO contents of Kaapvaal peridotites in particular have been cited as evidence of high-pressure melt extraction because of the sensitivity of FeO to melting pressure (Pearson et al. 1995a; Walter 1998; Aulbach et al. 2007). Once a secondary origin for much of the orthopyroxene is accepted, a convincing case can be made for the dilution of FeO in these rocks by orthopyroxene addition (Walter 2003; Lee 2006; Simon et al. 2008; Figs 6 and 15), and this removes one of the main arguments for high-pressure melting. The compositions of orthopyroxene-poor Greenland peridotites, argued above to be more representative of archetypal cratonic lithospheric mantle compositions (Bernstein et al. 2006), are displaced clearly towards the low-pressure melting trend in terms of FeO–MgO relations (Fig. 6), in agreement with the generally low pressures of melting indicated for other major elements such as Al2 O3 v. MgO (Fig. 15). However, there is still considerable scatter in the major element bulk compositions of this and other cratonic peridotite suites. This scatter in cratonic peridotite datasets is partly due to latestage secondary Fe enrichment (e.g. Boyd et al. 1997) and may also be influenced by the addition of other phases such as clinopyroxene. Considerable scatter across the Al2 O3 –MgO diagram is largely because of the effects of post-melting addition

The partitioning of mildly incompatible trace elements such as V, Sc and Yb is sensitive to the presence or absence of garnet in peridotite melting residues and hence can be used as a powerful tool to constrain the depth of melting. Using this approach, Canil (2004) has provided convincing evidence that most peridotite residues reported in the literature, both cratonic and postArchaean, originate via low-pressure melting (generally ,3 GPa). This conclusion has also been made on the basis of Yb partitioning in cratonic peridotites (Kelemen et al. 1998; Lee 2006; Simon et al. 2008). Heavy REE such as Yb are perhaps more preferable than Sc and V for this use because they are less susceptible to the possible introduction of clinoproxene, or interaction with the host kimberlite, which is generally Yb-poor (Wittig et al. 2008b). The approach has been extended using coupled Yb–Lu elemental systematics that show clearly, for cratonic and off-craton peridotites, that they must have experienced most of their melting under garnet-absent conditions, most probably at pressure less than 3 GPa (Fig. 16; Wittig et al. 2008a). Hence, there is a clear consensus from workers using mildly incompatible elements that most of the residues of melting present within the lithospheric mantle as a whole, and particularly cratonic mantle, form by melting at low pressures, mostly less than 3 GPa. Whereas estimates of melting pressure constrained by mildly incompatible elements such as Yb and Lu are relatively insensitive to modal or host rock metasomatism, because high-pressure (e.g. 7 GPa) melt residues evolve in the opposite direction to low-pressure (e.g. 3 GPa) melt residues (Fig. 16) the extents of melt extraction predicted using these elements are affected. None the less, many cratonic peridotites retain evidence of large extents of melt extraction, in excess of 25% or more (Wittig et al. 2008a). In summary, the strongest evidence for the depth of melting experienced by cratonic peridotites comes from a range of mildly incompatible elements, with the consensus being that melt extraction took place at pressures of 3 GPa or less (Stachel et al. 1998; Canil 2004; Lee 2006; Simon et al. 2008; Wittig et al. 2007, 2008a). This firm constraint can now be used to evaluate the likely tectonic environment in which cratonic peridotites were generated.

Environment of melting Any successful genetic model for cratonic peridotites must successfully explain their high and yet relatively uniform levels of melt depletion, produced in a shallow melting environment. Consideration of the melting relations of peridotite (Fig. 14) indicates that single-stage melting at low pressures (3 GPa and less) is unlikely to generate abundant residues that have experi-

A R C H A E A N C O N T I N E N TA L L I T H O S P H E R E A N D D I A M O N D S

enced c. 40% or more melt extraction, even at the high potential temperatures that may have existed in the Archaean mantle. This apparent contradiction can be resolved by examination of residues produced in more recent, better understood melting environments. Evaluation of both bulk-rock chemistry (Fig. 4) and olivine compositions (Figs 3 and 12) shows that the only modern-day melting environments capable of producing melt residues of comparable depletion to cratonic peridotites are in arc settings, notably the Izu–Bonin forearc (Ishii et al. 1992), the Iwanaidake peridotite massif (Kubo 2002) and ophiolites such as the highly depleted Papuan and Cuban ophiolite massifs (Jaques & Chappell 1980; Marchesi et al. 2006). This is also confirmed from examination of mildly incompatible element abundances (Wittig et al. 2008a). These residues are all produced at low pressures, with most of the melting probably occurring in the spinel stability field, at ,3 GPa, in a mantle environment where hydrous fluids are present. The similarity between residues produced in some modern-day subduction zones and cratonic peridotites has been previously noted by Parman et al. (2004), who proposed that the production of Barberton komatiites in an Archaean subduction zone led to the generation of residual Kaapvaal peridotites. A multi-stage melting model, with the final stage of melt extraction occurring in a subduction environment, has also been suggested by Simon et al. (2008) and Wittig et al. (2008a) to explain the modal, trace element and isotopic characteristics of Kaapvaal and Greenland (North Atlantic craton), respectively. This ‘wet melting’ model is also lent support in terms of explaining the statistical distribution of olivine compositions in cratonic peridotites that have pronounced modes at very depleted compositions. The low-pressure (,3 GPa) melt extraction indicated by trace element systematics means that initial melting is likely to have taken place at a spreading centre, either at a mid-ocean ridge or in an arc-related setting or in a shallow forearc environment. Single-stage polybaric melting at a ridge is unlikely to produce abundant residues dominated by highly depleted compositions, even at Archaean mantle potential temperatures (Fig. 14). The involvement of hydrous fluids, lowering the solidus, means that melting can extend to greater melt fractions at a given temperature (e.g. Hirschmann et al. 1999; Parman & Grove 2004), leading to greater levels of melt extraction and greater volumes of highly residual peridotite. In fact, theoretical modelling of the effects of melt depletion in peridotites using MELTS (Hirschmann et al. 1999) indicates that after clinopyroxene exhaustion, isobaric melt productivity decreases dramatically, by a factor of four, before rising again close to the orthopyroxene-out reaction. This drop in melt productivity results from a decrease in the temperature derivative of the bulk solid composition occurring as the melting reaction changes (Hirschmann et al. 1999). Hence, in anhydrous melting environments the clinopyroxene-out reaction is a natural impediment to the production of residues of very large degrees of melt extraction. The addition of water acts to lower the temperature required to reach a given extent of melt depletion. For instance, addition of 0.2% H2 O under isothermal conditions at 1350 8C and 1 GPa produces an increase of 15% melting compared with anhydrous conditions and the effect is greater at higher temperatures (Hirschmann et al. 1999). Given that the effect of melting past clinopyroxene exhaustion under hydrous conditions produces the same drop in melt productivity, it seems that the likely higher potential temperature of the Archaean mantle will assist, along with water, in consistently producing residues from very large extents (c. 40%) of melting. As melt productivity rises again near the orthopyroxene-out

905

boundary (under hydrous or anhydrous conditions) and then declines rapidly again thereafter, the orthopyroxene-out reaction would appear to produce a natural barrier, even to melting under hydrous conditions, and this may explain why few olivine Mg numbers exceed the value indicated by experiments as corresponding to this reaction (Fig. 11). Hydrous melting is thus capable of producing the strongly peaked statistical distributions in olivine compositions seen in cratonic peridotites. This is confirmed by examination of olivine compositions from hydrous melting environments such as the Izu–Bonin forearc and Iwanaidake massif peridotites and many ophiolites (Figs 3 and 12). A model involving two stages of melt extraction, with the final stage occurring in an arc, will also encourage melting in garnetfree conditions. The stability of spinel in mantle peridotite is enhanced at depleted, high Cr/Al bulk compositions (e.g. Ringwood 1977). Hydrous melting of depleted mantle in a subduction zone environment means that melting is much more likely to be taking place with spinel on the solidus at greater depths than for normal fertile mantle. This may explain the apparent complete absence of a strong garnet signature in any of the cratonic or modern subduction zone peridotites (Fig. 16). The combined weight of evidence from bulk compositions, olivine compositions and trace element systematics points clearly towards a shallow final stage of melting in a subduction zone environment. Whether melting occurred in two stages, first at a ridge and then in a subduction zone (e.g. Simon et al. 2008), or whether it took place predominantly in a subduction zone, with the simultaneous production of komatiites (Parman et al. 2004) cannot easily be resolved. However, hydrous melting in presentday subduction zones often involves mantle that has experienced previous melt depletion events (Parman et al. 2004), some of which took place billions of years previously (Parkinson et al. 1998; Pearson et al. 2007). As noted above, melting of previously depleted peridotite also favours melting taking place in the spinel stability field. The consequence of invoking a melting model that involves the last major period of melt extraction taking place at relatively shallow depths (,100 km) in a subduction zone environment is that this then requires tectonic stacking to thicken to lithospheric root to depths of 150 km or more (e.g. Helmstaedt & Schulze 1989). A result of subduction accretion of residues beneath existing crustal blocks is that this process decouples residue compositions from overlying melt compositions and also decouples the residue age from that of the overlying crust. Both these consequences are the opposite of the relations expected from the plume model of craton formation.

Timing of melt depletion and diamond formation in cratonic peridotites Lithosphere formation events Early Re–Os isotope studies of cratonic peridotites found that they retained signatures of Archaean melt depletion and hence appeared to be broadly as old as the cratonic crust lying above them (Walker et al. 1989; Carlson & Irving 1994; Pearson et al. 1995a, b). Since then, Archaean melt depletion ages have been found in every suite of cratonic peridotites analysed and a much larger database has been generated (for reviews, see Pearson et al. 2003; Carlson et al. 2005). One of the newest localities from which cratonic peridotites have become available is the North Atlantic craton, where Archaean melt depletion ages are evident in xenolith suites from both East and West Greenland (Hanghøj et al. 2001; Bernstein et al. 2006b; Wittig et al. 2007; Fig. 17). The early Re–Os studies of cratonic peridotites suggested that

906

D. G . P E A R S O N & N. W I T T I G

the main mass of cratonic lithospheric mantle was Mesoarchaean in age, similar to much of the early crust building phase (e.g. Pearson et al. 1995a). However, the large database now available, especially for the Kaapvaal craton (Carlson et al. 2005), allows a more detailed examination of the age information that can be retrieved from these rocks and reveals that there is no direct age relationship between crust and lithospheric mantle. In addition to whole-rock Re–Os analyses of peridotites, another useful development has been the use of sulphide Re–Os dating (Pearson et al. 1998; Pearson, N. J., et al. 2002) and the two methods offer much complementary information (Fig. 17). Although wholerock Re–Os analyses of peridotites are certainly integrations of numerous potential events that may have affected the evolution of a cratonic peridotite, and sulphide analyses have been suggested to be superior in terms of retrieving age information (Griffin et al. 2004), it is important to recognize the problems inherent in both approaches. There is abundant evidence for the recent introduction of significant Re (Walker et al. 1989; Pearson et al. 1995a; Carlson et al. 2005) and some Os (Chesley et al. 1999) into whole-rock peridotites and this is why model ages that are calculated assuming a Re content of zero (the timing of Re depletion: TRD age) are often preferred (Walker et al. 1989). These factors mean that the TRD age for whole-rock peridotites is likely to be an underestimate of the actual melting age. The sulphide dating approach also has significant pitfalls, unless the sulphide is trapped in a diamond at the time of its formation. The low solidi of sulphides means that they are one of the first phases to enter the melt during peridotite melting. This means that at the high melt fractions extracted from cratonic peridotites, common sulphides belonging to the monosulphide solid solution (mss) series will no longer be residual phases (Pearson et al. 2004; Lorand & Gre´goire 2006). Hence, any such sulphides, or Ni-rich pentlandites, are metasomatic in nature (Luguet et al. 2003; Lorand & Gre´goire 2006) and so by definition will carry a mixed signal themselves. Furthermore, peridotites with easily located high-T sulphides are relatively scarce among cratonic peridotites, especially when weathered. For instance, an intensive search for high-T base metal sulphides in a suite of peridotites from the Argyle diamond mine by Luguet et al. (2006) was unsuccessful and whole-rock analysis was the only possible approach. The low melting temperature of Cu–Ni–Fe sulphides means that they are easily disturbed and this produces a very large spread in resulting model ages. For instance, in a study of 168 sulphides from the Kaapvaal craton, Griffin et al. (2004) found that 33 sulphides gave ages older than the age of the Earth and 28 sulphides yielded future ages, using the Re–Os TMA model ages (model age using Re/Os ratio of sample) approach, with ages ranging from 79 to +130 Ga. Part

Fig. 17. Probability density plots of Re–Os model ages for whole-rock peridotites and sulphides for the Kaapvaal, Slave and North Atlantic cratons. Whole-rock data sources are as follows. Kaapvaal craton: Pearson et al. (1995a), Carlson et al. (1999), Menzies et al. (1999), Irvine et al. (2001), Carlson & Moore (2004) and Simon et al. (2008); Slave craton: Irvine et al. (2003), Aulbach et al. (2004) and Westerlund et al. (2006); North Atlantic craton: Hanghøj et al. (2001), Bernstein et al. (2006), Webb (2008) and Wittig et al. (2008b). All TRD ages are corrected to the eruption age of the kimberlite. Model age equations for TRD and TMA have been presented by Walker et al. (1989) and Pearson (1999) and are calculated to reference values for chondritic mantle of 187 Os=188 Os ¼ 0:1283 and 187 Re=188 Os ¼ 0:422. Number of samples plotted for each population is given by n.

A R C H A E A N C O N T I N E N TA L L I T H O S P H E R E A N D D I A M O N D S

of this spread may be due to open-system behaviour and part is due to significant problems in the application of Re corrections to small Os beams during laser ablation analyses (Nowell et al. 2008). This large fraction of aberrant ages led Griffin et al. (2004) to screen sulphides such that only ages from sulphides with Re/Os ,0.08 were deemed reliable. Although this narrows the spread in sulphide ages, assessing their accuracy is difficult. There are two factors that have caused confusion about the relative proportions of ancient lithospheric mantle ages provided by sulphides. One is the fact that, for the Kaapvaal craton, despite the large number of sulphides analysed, these data come from only 24 rocks (Griffin et al. 2004) compared with the 228 samples analysed for whole rocks. The second factor is the misleading practice by some workers (e.g. Griffin et al. 2004) of plotting on probability density diagrams age data where single ages have very large differences in uncertainty. A large variation in the apparent uncertainty of a Re–Os model age is common in sulphide datasets but is due largely to the increased precision of high-Os, low-Re/Os sulphides. Such data points create a false impression of the relative proportion of lithosphere of a given age if they are plotted alongside data of considerably worse precision (e.g. Griffin et al. 2004). In view of the fact that the dominant uncertainty on any Re–Os model age is our lack of precise knowledge of the Os isotope evolution of the Earth’s mantle, it is preferable to plot all such ages with a uniform conservative uncertainty (Fig. 17). A final factor that has distorted views on the relative merits of sulphide v. whole-rock Re–Os ages is the comparison of sulphide ages and whole-rock ages in rocks with abundant sulphides. Peridotites that have obviously experienced such metasomatism would be avoided by workers applying whole-rock analytical approaches. Another issue relating to the interpretation of Re–Os ages in cratonic peridotites is the growing evidence for considerable Os isotopic variation in the convecting mantle (see summary by Pearson et al. 2007). Re depletion ages of .1 Ga are now routinely found in samples recently extracted from the convecting mantle such as abyssal peridotites (e.g. Harvey et al. 2006) and Pt-group alloy grains (e.g. Pearson et al. 2007), and some samples show evidence of melting events that occurred . 2Ga ago. This creates severe difficulties in estimating the significance of small numbers of samples that have Re depletion model ages of 1–2 Ga occurring with other samples of younger age, within a particular lithospheric peridotite suite (e.g. Hanssler & Shimizu 1998). In such sample suites, where all the Os isotopic diversity is within that shown by abyssal peridotites, or where only one or two samples are less radiogenic than the ‘convecting mantle’ range, it is difficult to have confidence that any specific lithospheric formation event is being dated without independent constraints. However, isotopic diversity in the convecting mantle will naturally be less in the early stages of mantle evolution. Recent samples of convecting mantle with evidence of .2.5 Ga melting ages are extremely rare (1% of all data: Pearson et al. 2007). These two points together with the predominance of Archaean ages from all cratonic peridotite suites (Fig. 17) make it very likely that the abundant Archaean Re–Os model ages for these samples reflect melt extraction during lithospheric formation. In summary, it is likely that whole-rock TRD ages provide minimum estimates for the age of melting in cratonic peridotites whereas sulphide and whole-rock TMA ages are likely to overestimate depletion ages. Despite the fact that it is difficult to interpret either whole-rock or sulphide model ages with high levels of certainty, some striking observations can be made for cratons where both types of data have been obtained in some

907

abundance (i.e. the Kaapvaal and Slave cratons). The statistical distributions of whole-rock TMA and sulphide TMA model ages for the Kaapvaal craton show remarkable similarities (Fig. 17), with prominent but rather broad modes at 2.9–3.0 Ga. The distribution of whole-rock TMA ages shows a greater proportion of ages in the Eoarchaean to Hadean range but this probably reflects Re addition or problems with Re interference corrections (see Nowell et al. 2008) causing overestimation of TMA ages. In contrast, Kaapvaal peridotite TRD ages show a tighter, more pronounced mode at 2.7 Ga, with relatively few ages that are Mesoarchaean or older (Fig. 18). The clustering of whole-rock TRD model ages at around 2.7 Ga is unlikely to result from variable metasomatic addition of sulphides as suggested by some workers (e.g. Griffin et al. 2004), as such processes would lead to a more diffuse spread in ages. Although there is clearly a significant spread in ages, the fairly good agreement between the positions of the prominent modes for all the model age approaches could be taken to indicate that the major melt extraction event for the majority of Kaapvaal peridotites occurred between 2.7 and 3.0 Ga. This timing is younger than the age for the Kaapvaal lithospheric mantle commonly inferred from the 3.3 Ga diamond inclusion Sm/Nd model ages obtained by Richardson et al. (1984). It is coincident with collisional suturing of the Witwatersrand and Kimberley blocks of the Kaapvaal craton at c. 2.9 Ga, probably along the present-day Colesberg magnetic lineament (Schmitz et al. 2004). The Neoarchaean has been proposed by Pearson et al. (2002) to be the time of the major formation and stabilization of the lithospheric mantle beneath the Kaapvaal craton, and the recently defined palaeo-subduction zone between the two main blocks of the Kaapvaal craton provides a mechanism to make Kaapvaal peridotites in an arc environment at this time (Schmitz et al. 2004; Simon et al. 2008). This association is strengthened by the petrological arguments documented above. Further support for the timing of major subduction at 2.9 Ga is provided by the 2.89  0.06 Ga Re–Os isochron for eclogitic sulphides included in diamonds (Richardson et al. 2001) and 2.7–3.0 Ga Re–Os model ages for whole-rock eclogites from this region (Shirey et al. 2001). Evidence for the involvement of older lithospheric material, perhaps transposed from the older Witwatersrand block, comes from pre 3 Ga Re–Os model ages for eclogites and peridotites from the Newlands kimberlite (Menzies et al. 1999, 2003). This may explain the 3.3 Ga Sm/Nd model ages for peridotitic diamonds from the Kimberley block. The Witwatersrand block of the Kaapvaal craton contains crustal material formed between 3.7 and 3.3 Ga that seems to have formed a coherent block by 3.2 Ga (Schmitz et al. 2004). The older mini-peak in the mode for sulphide TMA ages at 3.1 Ga (Fig. 17) has been interpreted to represent older lithospheric mantle generation in the Witwatersrand block (Griffin et al. 2004). Although whole-rock TMA ages are consistent with this view of a significant fraction of Mesoarchaean lithosphere (Fig. 17), the TRD whole-rock ages are much more tightly clustered in the Neoarchaean. Further work is required to investigate this issue, although it does not seem unreasonable that some substantial fraction of lithospheric mantle from the Witwatersrand block formed in the Mesoarchaean, as has been suggested for lithosphere formation beneath the Siberia craton (Pearson et al. 1995b; Pearson, N. J., et al. 2002) and the eastern part of the North Atlantic craton (Hanghøj et al. 2001). The mode in age distributions for both whole-rock TRD and sulphide TMA ages are tight and in good agreement at c. 2.7– 2.8 Ga for Slave peridotites (Fig. 18). Davis et al. (2003) and Heaman & Pearson (2008) have noted that this peak in peridotite

908

D. G . P E A R S O N & N. W I T T I G

Re depletion ages corresponds to the timing of final stabilization of the Slave craton, as marked by intense 2.7 Ga granite plutonism. The suggestion of the presence of older diamondiferous lithosphere beneath the Slave craton at 3.5 Ga from diamond inclusion dating (Westerlund et al. 2006) and the observation of intense granitic plutonism at 2.7 Ga, implying very high thermal gradients at the base of the crust, are difficult to reconcile (Davis et al. 2003). To explain these observations, Percival & Pysklywe (2007) have suggested that the entire cratonic lithospheric mantle becomes inverted during eclogite delamination at c. 2.7 Ga. It is also possible that the 3.5 Ga sulphide isochron of Westerlund et al. (2006) is a mixing age produced by resetting of older lithospheric material during diamond crystallization at 2.7 Ga, as indicated by the ubiquitous 2.7 Ga TMA model ages of these sulphides. These problems and observations have led Davis et al. (2003) and Heaman & Pearson (2008) to favour a Neoarchaean formation and stabilization age for much of the Slave cratonic lithosphere. A Neoarchaean stabilization age is also indicated by wholerock TRD ages from the North Atlantic craton (Fig. 17; Wittig et al. 2008b). It seems a striking coincidence that peridotites from the three cratons with the most abundant Re–Os isotope data all have their modes in depletion ages between 2.7 and 3.0 Ga, suggesting that the Neoarchaean may have been a time when the majority of the now extant cratonic lithospheric mantle was produced and stabilized, mostly probably in Archaean subduction zones. There are certainly more ancient Mesoand Eoarchaean model ages evident for peridotites from the above three cratons and from Siberia (Pearson et al. 1995b). These may represent earlier lithospheric mantle formation episodes; for instance, for parts of the Witwatersrand block of the Kaapvaal craton or the northern and eastern portions of the North Atlantic craton exposed on Greenland. Some of these more ancient ages also may represent earlier melting events either at ridges, or in other environments that did not result in the accretion of these residues to the lithosphere. There are analogues to this situation in the modern Earth, where increasing evidence is being found for ancient depleted components existing in convecting mantle (Parkinson et al. 1998; Harvey et al. 2006; Pearson et al. 2007).

Lithosphere modification events and diamond formation As the great majority of diamonds are formed in the lithospheric mantle (Stachel & Harris 2008) the formation and subsequent evolution of the lithosphere are clearly critical factors in the formation of diamonds. One of the best defined diamond formation ages for the Kaapvaal craton is the 2.89  0.06 Ga eclogitic sulphide isochron for inclusions within Kimberley diamonds (Richardson et al. 2001). This event is most obviously correlated with the time of proposed suturing between the Witwatersrand and Kimberley blocks of the Kaapvaal craton (Schmitz et al. 2004) and the Archaean diamond crystallization event is synchronous with the likely formation of much of the lithospheric mantle in the Kimberley block, in a subduction zone environment. There are a number of additional post-Archaean peaks in the statistical distribution of sulphide and whole-rock Re–Os models ages for cratonic peridotites (Fig. 17). The most prominent is the peak between 0.9 and 1.2 Ga observed in sulphide TMA and whole-rock TRD ages for the Kaapvaal craton (Fig. 17). This peak correlates with diamond-forming events in the Kaapvaal lithosphere at 0.99 Ga (Orapa; Richardson et al. 1990) and 1.1 Ga (Pearson et al. 1998) and corresponds to the timing of the Namaqua high-temperature thermal event

(Tankard et al. 1982). As formation of the eclogitic hosts to these Proterozoic diamonds took place in the Archaean (Menzies et al. 1999; Shirey et al. 2001) they are likely to be of metasomatic origin, not associated with any discrete subduction event. Their association with high-temperature metamorphism in the deep crust indicates rising thermal gradients as a result of either lithospheric rifting or plume impingement. The thermal imprint in the lower crust of these Proterozoic thermal events is preserved by zircon ages in 0.964–1.11 Ga granulite-facies xenoliths (Schmitz & Bowring 2000). The broad coincidence of all these ages, allowing for the larger inaccuracies of the mantle xenolith Re–Os model ages, is strongly suggestive of a prolonged and significant lithospheric-scale thermal event that seems to have involved the crystallization of diamonds in some parts of the cratonic root. Given the relatively lowl temperatures of diamond formation indicated by trapped inclusions (1150– 1240 8C, Stachel & Harris 2008), this growth is likely to take place on the up-temperature side of such a thermal event, with carbon-bearing fluids being driven off ahead of a rising plume. The c. 2 Ga diamond-forming event at Premier (Richardson et al. 1993; Fig. 18) correlates with the major lithospheric thermal disturbance manifest as the Bushveld intrusion and its thermal imprint. There are a number of Kaapvaal peridotites of this age (Fig. 17) and new data acquired by Morel et al. (2006) indicate significant new peridotitic lithosphere added at 2 Ga beneath the central Witwatersrand block. This new data provide further support for suggestions by Richardson et al. (1993) and Shirey et al. (2002) for a link between pulses of diamond growth and major lithospheric thermo-tectonic events. There are exceptions to this general rule in that there seems to be no indication of a mantle or crustal event to accompany the c. 1.5 Ga ages found for diamonds from Finsch and Jwaneng (Richardson et al. 1990, 2004). None the less, the correlation between major lithospheric thermo-tectonic events manifest in both crust and mantle and the formation of diamonds is strongly indicative of a link and indicates a pulsed nature of diamond formation associated with such events. It seems more than coincidental that significant secondary age peaks in the lithosphere can be identified in the lithospheric mantle of the North Atlantic craton at similar ages to the diamond–lithosphere ages seen in the Kaapvaal craton (Fig. 17) and we will await diamond dating to see how these ages may relate to the age of diamonds from this region. In a broader context, a global correlation between mantle melting events and crustal ages has recently been recognized through Os isotope studies of platinum group metal alloy grains (Pearson et al. 2007), with c. 1.1 Ga and 2.2 Ga pulses of enhanced activity evident. The correlations identified here predict that, ultimately, there may be very strong temporal links between crust formation, continental lithospheric mantle generation, the depletion of the convecting mantle in general and diamond formation that are intrinsically related to the episodic nature of mantle convection and melting.

A model for the formation of cratonic roots In the discussion above we have shown that the bulk compositions and olivine chemistry of cratonic peridotites are similar to those of Phanerozoic to recent peridotites that have been melted in shallow environments, in the presence of hydrous fluids (i.e. in arc settings). The generation of cratonic peridotites in subduction zone environments has been gaining support from different angles for the past two decades (Ringwood 1977; Kesson & Ringwood 1989; Rudnick et al. 1994; Kelemen et al. 1998; Canil 2004; Parman et al. 2004; Bell et al. 2005; Lee 2006; Simon et

A R C H A E A N C O N T I N E N TA L L I T H O S P H E R E A N D D I A M O N D S

al. 2008) and this seems to be the best overall scenario to explain the shallow depth of melting, the very high and uniform levels of depletion, and the variable later introduction of orthopyroxene. Here we present a general petrogenetic (Fig. 19) and tectonic model (Fig. 20) for the formation of cratonic peridotites that integrates and builds on these earlier studies. We will use a plot of Mg number of olivine v. modal olivine through the sequence of events because it is useful to illustrate the different processes involved and we make analogy to the well-studied residual rocks of the Iwanaidake peridotite massif to graphically illustrate the possible reactions between melt and residue (Fig. 19). We start from the premise that most refractory cratonic peridotites require two stages of melting to generate their highly depleted, relatively uniform compositions. The first stage of melting may have occurred at an Archaean mid-ocean ridge where melting took place at 3 GPa or below to produce a fertile harzburgite; Stage 1 of Figure 19. This stage of melting produced the curved trend of increasing olivine Mg number v. increasing modal olivine, referred to as ‘the Oceanic Trend’ by Boyd (1989). If this melting took place in the geometrically complex environment of an ocean ridge then it is likely that some olivine addition also contributed to the trend defined by Stage 1 as a result of the crystallization of olivine-saturated melt at higher levels in the lithosphere (e.g. Niu 1994). Stage 1 may have been the immediate precursor to the enhanced melting represented by Stage 2 if all the melt extraction occurred during a single stage, at the wet solidus, in a subduction zone environment. Stage 2 is extended, water-fluxed melting to on average 40% (Bernstein et al. 2006a; Fig. 11) or more melt removal, close to, or past, the point of orthopyroxene exhaustion to produce orthopyroxene-poor dunitic residues. This is the composition of archetypal cratonic peridotite suggested by Kesson & Ringwood (1989), Bernstein et al. (2006) and Wittig et al. (2008a). Although it is possible to produce such refractory compositions without the aid of water because of the high potential temperatures of the Archaean mantle (Fig. 11), such melting is unlikely to produce the uniformly highly depleted residues observed (Figs 9 and 12) if it is constrained to pressures ,3 GPa, and hence hydrous melting is most appropriate. The best examples of these highly refractory Stage 2 residues are the olivine-rich endmembers of cratonic peridotites from the East and West Greenland parts of the North Atlantic craton (Bernstein et al. 1998, 2006; Wittig et al. 2008a), which contain no garnet but very high-Cr spinel. In some instances these chromite dunites may transform at the base of cool cratonic lithosphere into highCr garnet dunites (Ringwood 1977; Schulze 1986; Kesson & Ringwood 1989; Stachel et al. 1998). Stage 3 is the addition of orthopyroxene to refractory dunitic residues (Fig. 19). Kelemen et al. (1992), Bell et al. (2005) and Simon et al. (2008) proposed that orthopyroxene formed from high-T Si-rich fluids in the subduction zone environment, whereas Rudnick et al. (1994) favoured siliceous melts from basaltic slab melting. However, a more local source of Si-rich melts is available that is capable of generating orthopyroxenerich peridotites. The melting reaction of enstatite is peritectic in nature, forming olivine and a Si-rich melt. The extended melting that creates refractory dunitic residues will form a melt capable of reacting with olivine in parts of the nearby already depleted mantle that has risen (because of its lower density) and cooled slightly, to produce more Mg-rich orthopyroxene that will be close to equilibrium with its host and hence unlikely to alter the Mg number of olivine in the resulting rock. The close to solidus temperatures of this reaction environment will be conducive to the formation of high-temperature aluminous orthopyroxene,

909

which seems the best fit for explaining the bulk compositional trends of many cratonic peridotites (Figs 5 and 6). A further attraction of this model, where the orthopyroxene-forming melt is derived relatively locally, is that melt/rock ratios could be high, allowing extensive percolation and deposition of significant volumes of orthopyroxene. Derivation of orthopyroxene solely from much more distal sources, as exogenous water-rich fluids (Bell et al. 2005; Simon et al. 2008) or silica-rich melts (Rudnick et al. 1994; Kelemen et al. 1998) has the difficulty of needing to extensively pervade large columns of lithospheric mantle. The mechanism of orthopyroxene enrichment we propose is essentially ‘autometasomatic’ in that the Si-rich melt is locally derived and close to equilibrium with its cooler host. This mechanism can create the subhorizontal trend on the olivine Mg number v. modal olivine plot that is characteristic of cratonic peridotite suites (Boyd 1989; Menzies 1990; Boyd 1996; Figs 2, 8, 9 and 13). A single cycle of these melting and metasomatic events will produce a mixture of dunite and orthopyroxene-rich refractory harzburgites. Multiple cycles of this process, where cooler early formed residues are reacting with siliceous fluids being generated in new melting environments below, will result in a lithospheric section dominated by refractory harzburgites with widely varying orthopyroxene contents at relatively constant olivine Mg number (Figs 13 and 19). Melting of newly arrived but previously depleted mantle, as buoyant residues ascend to form proto-lithosphere, maintains melting in the spinel facies. Addition of olivine to fertile harzburgite that has not been fluxed by water, or olivine formation in refractory harzburgite that has not progressed to the orthopyroxene ‘out’ stage, will create additional spread away from the Stage 3 vector (Fig. 19) but all these processes are envisioned to take place in the Archaean melting environment. Subsequent addition of minor garnet and clinopyroxene, predominantly associated with magmatic activity in the Proterozoic to Phanerozoic, may create more diverse spread in modal olivine and olivine compositions (Fig. 19). The tectonic manifestation of the above model is depicted in Figure 20. We envisage that Stages 1–4 of ridge melting and subduction followed by physical maturity occurred in the Neoarchaean, between 2.9 and 2.7 Ga in the case of the Kimberley block of the Kaapvaal craton, and possibly the Slave and the SW part of the North Atlantic craton. It can be seen that one obvious consequence of a model that invokes tectonic stacking of oceanic lithosphere is the likely incorporation of oceanic crust of abundant eclogitic slabs in the newly stabilized ‘continental’ lithosphere. Even the crudest estimate of the abundance of this eclogitic component is of the order of 10% by volume. Estimates of the abundance of eclogite now residing in the cratonic lithosphere are considerably less than this, at around 1% by volume (Schulze 1989). This difference may be because large dense slabs of eclogite are not gravitationally stable in lowdensity cratonic peridotite. Percival & Pysklywe (2007) showed via dynamical modelling that a large horizontal slab of eclogite located at the base of the cratonic crust could sink rapidly through the underlying peridotite lithosphere causing significant thermal disturbance. Although this model leads to some appealing results in terms of transient thermal spikes at the base of the crust to produce Neoarchaean granites above older but thermally disturbed peridotite, it is a greatly simplified scenario. A subduction stacking model for the lithosphere is likely to have multiple slabs of eclogite inclined at some angle from the horizontal. This situation is also not one of long-term gravitational stability and it is likely that the lithosphere will tend to ‘cleanse itself’ of much of the dense eclogite, leaving behind a

910

D. G . P E A R S O N & N. W I T T I G

Fig. 18. Age correlations between crust, lithospheric mantle and diamond formation events for the Kaapvaal craton. Duration or age uncertainty of events is indicated by width of box. For diamond ages, line denotes age and box width denotes uncertainty. Colour coding for crust and mantle events is to delineate those evident in the Witwatersrand block (orange), the Kimberley block (green) and the marginal cratonic Namaqua metamorphic event (pink). Diamond ages (grey) show ages for silicate and sulphide inclusion dating. Crustal geochronology is from Tankard et al. (1982), Schmitz & Bowring (2000) and Schmitz et al. (2004). Mantle ages are from Figure 17 and references listed in the caption. Diamond ages are from Richardson et al. (1984, 1990, 1993, 2004), Pearson et al. (1998) and Pearson & Harris (2006).

Fig. 19. Petrogenetic model for the formation of cratonic peridotites via multi-stage melting and metasomatism. The model depicts the formation of ultradepleted dunitic residues starting at an oceanic ridge (Stage 1) via hydrous melting in a subduction zone environment (Stage 2). The resulting olivine-rich refractory lithologies from Stage 2 are probably the starting compositions for most cratonic lithosphere (Kesson & Ringwood 1989; Bernstein et al. 2006a; Figs 8 and 9), which then evolve to varying degrees of orthopyroxene enrichment as a result of metasomatism by silica-rich melts (Stage 3) driven off from later forming dunites during hydrous melting. Melting is within the spinel facies, aided by the melting of previously depleted mantle from the ridge. Multiple cycles of such water-fluxed melting then create ultra-depleted harzburgites with very variable orthopyroxene (and hence olivine) contents at relatively constant and high olivine Mg numbers. Later, infiltration of small-degree Si-poor melts causes the generation of lherzolites via clinopyroxene and some garnet crystallization. The legend gives the identity of key minerals in the upper illustrations during the evolution from fertile harzburgite to dunite to refractory orthopyroxene-rich harzburgite to refractory lherzolite. Time sequence is from left to right. Graphs show how complexity evolves with each stage in terms of olivine Mg number v. modal olivine; the arrows with numbers are explained in Figure 13.

A R C H A E A N C O N T I N E N TA L L I T H O S P H E R E A N D D I A M O N D S

911

Fig. 20. Tectonic schematic illustration showing the generation and evolution of the early stages of the formation of a craton focusing on the processes taking place in the lithospheric mantle root. Initial melting to create relatively fertile harzburgite is shown taking place at an Archaean oceanic ridge in Stage 1. The location of the two-stage melting and metasomatic model illustrated in Figure 19 is shown in the mantle wedge above an Archaean subduction zone (Stage 2), and shows how newly created ultra-depleted lithospheric mantle might grow and cool. Stage 3 indicates a proto-cratonic root thickened by tectonic stacking that contains a significant volume of eclogitic crust (black). This eclogitic crust is dense and hence gravitationally unstable, sinking through the peridotite to leave a more mature craton with only minor, localized eclogite lenses (Stage 4). The transition from Stage 3 to Stage 4 probably involves significant physical and thermal instability in the mantle and lower crustal sections of the evolving craton. For the Kaapvaal craton we envisage that Stages 1–4 took place in the Neoarchaean between 2.9 and 2.7 Ga for the Kimberley block.

more thermally and physically mature craton with a root of lowdensity peridotite dominated lithospheric mantle with minor eclogitic pods nested within it (Stage 4 of Fig. 20). It is likely that during the loss of the eclogite component there is significant physical and thermal instability in the mantle and lower crustal sections of the evolving craton. Following Percival & Pysklywe (2007) we suggest that this activity may create one or more thermal spikes sufficient to cause crustal melting and Neoarchaean crustal magmatism. The physical and thermal instability of a craton during this period may be able to explain the apparently very active early history of the Kaapvaal craton in which a major rifting event that produced the Ventersdorp lavas was occurring at 2.7 Ga, shortly after the supposed stabilization of the craton (de Wit et al. 1992) following east–west suturing (Schmitz et al. 2004). Ostensibly, the model presented in Figures 19 and 20 suffers from a similar problem to most other models of cratonic peridotites; that is, lack of evidence of suitable volumes of large fraction mafic melts associated with the ultra-depleted residues that form the lithosphere. However, although this is a seemingly critical problem in any plume-driven hypothesis for making

cratonic mantle because there is no evidence of anomalous crustal thicknesses beneath well-studied cratons such as the Kaapvaal (James et al. 2001; Nguuri et al. 2001), the subduction stacking model has the consequence of physically transporting and decoupling residues from melts, especially if any resulting cogenetic melts that do become incorporated into the lithospheric mantle become later ‘refined out’ by gravitational processes (Fig. 20). This model also allows for significant decoupling of the ages of lithospheric mantle and overlying crust. We thank T. Stachel, J.-L. Bodinier, Y. Niu, E.-R. Neumann, N. Simon and D. Francis for supplying additional bulk-rock and mineral chemistry data used in this review. S. Parman provided a critical sounding board for the ideas presented here, and discussions with C. Garrido helped to sharpen some of the arguments.

References Aulbach, S., Griffin, W.L., Pearson, N.J., O’Reilly, S.Y., Kivi, K. & Doyle, B. 2004. Mantle formation and evolution, Slave craton: constraints from HSE abundances and Re–Os isotope systematics of sulfide inclusions in mantle

912

D. G . P E A R S O N & N. W I T T I G

xenocrysts. Chemical Geology, 208, 61–88. Aulbach, S., Griffin, W.L., Pearson, N.J., O’Reilly, S.Y. & Doyle, B.J. 2007. Lithosphere formation in the Central Slave Craton (Canada): plume accretion or lithosphere accretion. Contributions to Mineralogy and Petrology, 154, 409–428. Baker, M.B. & Stolper, E.M. 1994. Determining the composition of highpressure mantle melts using diamond aggregates. Geochimica et Cosmochimica Acta, 58, 2811–2827. Bell, D.R., Gre´goire, M., Grove, T.L., Chatterjee, N., Carlson, R.W. & Buseck, P.R. 2005. Silica and volatile-element metasomatism of the Archean mantle: a xenolith-scale example from the Kaapvaal craton. Contributions to Mineralogy and Petrology, 150, 251–267. Bernstein, S., Kelemen, P.B. & Brooks, C.K. 1998. Depleted spinel harzburgite xenoliths in Tertiary dykes from East Greenland: Restites from high melting degrees. Earth and Planetary Science Letters, 154, 221–235. Bernstein, S., Kelemen, P.B. & Hanghøj, K. 2006a. Consistent olivine Mg# in cratonic mantle reflects Archean mantle melting to the exhaustion of orthopyroxene. Geology, 35, 459–462. Bernstein, S., Hanghøj, K., Kelemen, P.B. & Brooks, C.K. 2006b. Ultradepleted, shallow cratonic mantle beneath West Greenland: dunitic xenoliths from Ubekendt Ejland. Contributions to Mineralogy and Petrology, 152, 335– 347. Bizzarro, M. & Stevenson, R.K. 2003. Major element composition of the lithospheric mantle under the North Atlantic craton: Evidence from peridotite xenoliths of the Sarfartoq area, southwest Greenland. Contributions to Mineralogy and Petrology, 146, 223–240. Bleeker, W. 2003. The late Archean record: a puzzle in ca. 35 pieces. Lithos, 71, 99–134. Bodinier, J.-L. & Godard, M. 2003. Orogenic, ophiolitic and abyssal peridotites. In: Holland, H.D. & Turekian, K.K. (eds) Treatise on Geochemistry. Elsevier, Amsterdam, 103–170. Bowring, S.A., Housh, T.B. & Isachsen, C.E. 1990. The Acasta gneisses; remnant of Earth’s early crust. In: Newsom, H.E. & Jones, J.H. (eds) The Origin of the Earth. Oxford University Press, New York, 319–343. Boyd, F.R. 1989. Compositional distinction between oceanic and cratonic lithosphere. Earth and Planetary Science Letters, 96, 15–26. Boyd, F.R. 1996. Origins of peridotite xenoliths: Major element considerations. In: Ranalli, G., Lucchi, F.R., Ricci, C.A. & Trommsdorff, T. (eds) High Pressure and High Temperature Research on Lithosphere and Mantle Materials: Proceedings of the International School Earth and Planetary Sciences. University of Siena, 89–106. Boyd, F.R. & McAllister, R.H. 1976. Densities of fertile and sterile garnet peridotites. Geophysical Research Letters, 3, 509–512. Boyd, F.R. & Mertzman, S.A. 1987. Composition and structure of the Kaapvaal lithosphere, Southern Africa. In: Mysen, B.O. (ed.) Magmatic Processes: Physicochemical Principles. Geochemical Society Special Publications, 1, 3–12. Boyd, F.R., Jones, R.A. & Nixon, P.H. 1983. Mantle metasomatism: the Kimberley dunites. Carnegie Institution of Washington Yearbook, 82, 330– 336. Boyd, F.R., Pokhilenko, N.P., Pearson, D.G., Mertzman, S.A., Sobolev, N.V. & Finger, L.W. 1997. Composition of the Siberian cratonic mantle: evidence from Udachnaya peridotite xenoliths. Contributions to Mineralogy and Petrology, 128, 228–246. Burgess, S.R. & Harte, B. 2004. Tracing lithosphere evolution through the analysis of heterogeneous G9–G10 garnets in peridotite xenoliths, II: REE chemistry. Journal of Petrology, 45, 609–634. Canil, D. 1991. Experimental evidence for the exsolution of cratonic peridotite from high-temperature harzburgite. Earth and Planetary Science Letters, 106, 64–72. Canil, D. 1992. Orthopyroxene stability along the peridotite solidus and the origin of cratonic lithosphere beneath southern Africa. Earth and Planetary Science Letters, 111, 83–95. Canil, D. 2004. Mildly incompatible elements in peridotites and the origins of mantle lithosphere. Lithos, 77, 375–393. Carlson, R.W. & Irving, A.J. 1994. Depletion and enrichment history of subcontinental lithospheric mantle: An Os, Sr, Nd and Pb isotopic study of ultramafic xenoliths from the northwestern Wyoming Craton. Earth and Planetary Science Letters, 126, 457–472. Carlson, R.W. & Moore, R.O. 2004. Age of the Eastern Kaapvaal mantle: Re– Os isotope data for peridotite xenoliths from the Monastery mine. South African Journal of Geology, 107, 81–90. Carlson, R.W., Pearson, D.G., Boyd, F.R., Shirey, S.B., Irvine, G., Menzies, A.H. & Gurney, J.J. 1999. Re–Os systematics of lithospheric peridotites: Implications for lithosphere formation and preservation. In: Gurney, J.J., Gurney, J.L., Pascoe, M.D. & Richardson, S.H. (eds) Proceedings of the VIIth International Kimberlite Conference. Red Roof Designs, Cape Town, 99–108.

Carlson, R.W., Pearson, D.G. & James, D.E. 2005. Physical, chemical, and geochronological characteristics of continental mantle. Reviews of Geophysics, 43, paper number 2004RG000156. Chesley, J.T., Rudnick, R.L. & Lee, C.-T. 1999. Re–Os systematics of mantle xenoliths from the East African Rift: Age, structure, and history of the Tanzanian craton. Geochimica et Cosmochimica Acta, 63, 1203–1217. Cox, K.G., Smith, M.R. & Beswetherick, S. 1987. Textural studies of garnet lherzolites: evidence of exsolution from high-temperature harzburgites. In: Nixon, P.H. (ed.) Mantle Xenoliths. Wiley, New York, 537–550. Davis, W.J., Jones, A.G., Bleeker, W. & Grutter, H. 2003. Lithosphere development in the Slave craton: a linked crustal and mantle perspective. Lithos, 71, 575–589. de Wit, M.J., Roering, C. & Hart, R.J. et al. 1992. Formation of an Archaean continent. Nature, 357, 553–562. Friend, C.L., Bennett, V.C. & Nutman, A.P. 2002. Abyssal peridotites .3,8000 Ma from southern West Greenland: field relationships, petrography, geochronology, whole-rock and mineral chemistry of dunite and harzburgite inclusions in the Itsaq Gneiss Complex. Contributions to Mineralogy and Petrology, 143, 71–92. Gao, S., Rudnick, R.L., Carlson, R.W., McDonough, W.F. & Liu, Y.-S. 2002. Re–Os evidence for replacement of ancient mantle lithosphere beneath the North China craton. Earth and Planetary Sciences, 198, 307–322. Gre´goire, M., Bell, D.R. & Le Roex, A.P.L. 2002. Garnet lherzolites from the Kaapvaal Craton (South Africa): trace element evidence for a metasomatic history. Journal of Petrology, 44, 629–657. Griffin, W.L., O’Reilly, S.Y. & Ryan, C.G. 1999. The composition and origin of sub-continental lithospheric mantle. In: Fei, Y., Bertka, C.M. & Mysen, B.O. (eds) Mantle Petrology: Field Observations and High Pressure Experimentation: A Tribute to Francis R. (Joe) Boyd. Geochemical Society, Special Publications, 6, 13–45. Griffin, W.L., Graham, S., O’Reilly, S.Y. & Pearson, N.J. 2004. Lithosphere evolution beneath the Kaapvaal Craton: Re–Os systematics of sulfides in mantle-derived peridotites. Chemical Geology, 208, 195–215. Hanghøj, K., Kelemen, P.B., Bernstein, S., Blustztajn, J. & Frei, R. 2001. Osmium isotopes in the Wiedemann Fjord mantle xenoliths: A unique record of cratonic mantle formation by melt depletion in the Archean. Geochemistry, Geophysics, Geosystems, 2, paper number 2000GC000085. Harvey, J., Gannoun, A., Burton, K., Rogers, N.W., Alard, O. & Parkinson, I.J. 2006. Ancient melt extraction from the oceanic upper mantle revealed by Re–Os isotopes in abyssal peridotites from the Mid-Atlantic Ridge. Earth and Planetary Sciences Letters, 244, 606–621. Hassler, D.R. & Shimizu, N. 1998. Osmium isotopic evidence for ancient subcontinental lithospheric mantle beneath the Kerguelen Islands, southern Indian Ocean. Science, 280, 418–420. Heaman, L.M. & Pearson, D.G. 2008. Nature and evolution of the Slave subcontinental lithospheric mantle. In: Lithoprobe – Parameters, Processes and the Evolution of a Continent. Lithoprobe Synthesis Volume II. NRC Earth Science Series Monographs, NRC Research Press, in press. Helmstaedt, H.H. & Schulze, D.J. 1989. Southern African kimberlites and their mantle sample—implication for Archean tectonics and lithosphere evolution. In: Ross, J. (ed.) Kimberlites and Related Rocks, Vol. 1. Geological Society of Australia Special Publication, 14, 358–368. Herzberg, C. 1999. Phase equilibrium constraints on the formation of cratonic mantle. In: Fei, Y., Bertka, C.M. & Mysen, B.O. (eds) Mantle Petrology: Field Observations and High Pressure Experimentation. Geochemical Society, Special Publications, 6, 241–257. Herzberg, C. 2004. Geodynamic information in peridotite petrology. Journal of Petrology, 45, 2507–2530. Hirschmann, M.M., Asimow, P.D., Ghiorso, M.S. & Stolper, E.M. 1999. Calculation of peridotite partial melting from thermodynamic models of minerals and melts III. Controls on isobaric melt production and the effect of water on melt production. Journal of Petrology, 40, 831–850. Ionov, D.A. & Hofmann, A.W. 2007. Depth of formation of subcontinental offcraton peridotites. Earth and Planetary Science Letters, 261, 65–83. Irvine, G.J., Pearson, D.G. & Carlson, R.W. 2001. Lithospheric mantle evolution in the Kaapvaal craton: A Re–Os isotope study of peridotite xenoliths from Lesotho kimberlites. Geophysical Research Letters, 28, 2505– 2508. Irvine, G.J., Pearson, D.G., Kjarsgaard, B.A., Carlson, R.W., Kopylova, M.G. & Dreibus, G.E. 2003. A Re–Os isotope and PGE study of kimberlitederived peridotite xenoliths from Somerset Island and a comparison to the Slave and Kaapvaal craton. Lithos, 71, 461–488. Ishii, T., Robinson, P.T., Maekawa, H. & Fiske, R. 1992. Petrological studies of peridotites from diapiric serpentinite seamounts in the Izu–Ogasawara forearc. In: Fryer, B.P., Pearce, J.A. & Stokking, L.B. (eds) Proceedings of the Ocean Drilling Program, Scientific Results. Ocean Drilling Program, College Station, TX, 125, 445–485. Iwamori, H., McKenzie, D. & Takahashi, E. 1995. Melt generation by isentropic

A R C H A E A N C O N T I N E N TA L L I T H O S P H E R E A N D D I A M O N D S mantle upwelling. Earth and Planetary Science Letters, 134, 253–266. Jackson, J., McKenzie, D., Priestley, K. & Emmerson, B. 2008. New views on the structure and rheology of the lithosphere. Journal of the Geological Society, London, 165, 453–466. Jagoutz, E., Palme, H. & Baddenhausen, H. et al. 1979. The abundances of major, minor and trace elements in the Earth’s mantle as derived from primitive ultramafic nodules. In: Merrill, R.B. (ed.) Proceeding of the 10th Lunar and Planetary Science Conference. Geochimica et Cosmochimica Acta Supplement, 2031–2050. James, D.E., Fauch, M.J., Van deCar, J.C. & van der Lee, S. 2001. Tectospheric structure beneath southern Africa; The Kaapvaal project; formation and evolution of cratons. Geophysical Research Letters, 28, 2485–2488. Jaques, A.L. & Chappell, 1980. Petrology and trace element geochemistry of the Papuan ultramafic belt. Contributions to Mineralogy and Petrology, 75, 55–70. Jordan, T.H. 1975. The continental tectosphere. Reviews of Geophysics and Space Physics, 13, 1–13. Kelemen, P.B., Dick, H.J.B. & Quick, J.E. 1992. Formation of harzburgite by pervasive melt/rock reaction in the upper mantle. Nature, 358, 635–641. Kelemen, P.B., Hart, S. & Bernstein, S. 1998. Silica enrichment in the continental upper mantle via melt/rock reaction. Earth and Planetary Science Letters, 164, 387–406. Kesson, S.E. & Ringwood, A.E. 1989. Slab–mantle interactions 2: the formation of diamonds. Chemical Geology, 78, 97–118. Kinzler, R.J. & Grove, T.L. 1999. Origin of depleted harzburgite by deep fractional melt extraction and shallow olivine cumulate infusion. In: Gurney, J.J., Gurney, J.L., Pascoe, M.D. & Richardson, S.H. (eds) Proceedings of 7th International Kimberlite Conference. Red Roof Design, Cape Town, 437– 443. Kopylova, M. & Russell, K. 2000. Chemical stratification of cratonic lithosphere: Constraints from the northern Slave craton. Earth and Planetary Sciences, 181, 71–87. Kubo, K. 2002. Dunite formation processes in highly depleted peridotite: Case study of the Iwanaidake Peridotite, Hokkaido, Japan. Journal of Petrology, 43, 423–448. Lee, C.-T. 2003. Compositional variation of density and seismic velocities in natural peridotites at STP conditions: implications for seismic imaging of compositional heterogeneities in the upper mantle. Journal of Geophysical Research, 108, 2441–2451. Lee, C.-T. 2006. Geochemical/petrological constraints on the origin of cratonic mantle. In: Benn, K., Mareschal, J.-C. & Condie, K.C. (eds) Archean Geodynamics and Environments. American Geophysical Union Monograph, 00, 89–114. Lee, C.T. & Rudnick, R.L. 1999. Compositionally stratified cratonic lithosphere: Petrology and geochemistry of peridotite xenoliths from the Labait volcano, Tanzania. In: Gurney, J.J., Gurney, J.L., Pascoe, M.D. & Richardson, S.H. (eds) Proceedings of the VIIth International Kimberlite Conference. Red Roof Designs, Cape Town, 503–521. Lorand, J.-P. & Gre´goire, M. 2006. Petrogenesis of base metal sulphide assemblages of some peridotites from the Kaapvaal craton (South Africa). Contributions to Mineralogy and Petrology, 151, 521–538. Luguet, A.L., Lorand, J.-P. & Seyler, S. 2003. Sulfide petrology and highly siderophile element geochemistry of abyssal peridotites: a coupled study of samples from the Kane Fracture Zone (458W, 23820N, MARK area, Atlantic Ocean). Geochimica et Cosmochimica Acta, 67, 1553–1570. Luguet, A.L., Pearson, D.G., Jaques, A.L., Bulanova, G.P., Smith, C.B., Roffey, S. & Rayner, M.J. 2006. Deep structure of the lithospheric mantle under a mobile zone: the Re–Os isotope approach. Geophysical Research Abstracts, 8, 04320. Maaloe, S. & Aoki, K. 1977. The major element composition of the upper mantle estimated from the composition of lherzolites. Contributions to Mineralogy and Petrology, 63, 161–173. MacKenzie, J.M. & Canil, D. 1999. Composition and thermal evolution of cratonic mantle beneath the central Slave province, NWT, Canada. Contributions to Mineralogy and Petrology, 134, 313–324. Malitch, K.N. & Merkle, R.K.W. 2004. Ru, Os, Ir, Pt and Pt Fe alloys from the Evander gold field, Witwatersrand basin, South Africa: Detrital origin inferred from compositional and osmium isotope data. Canadian Mineralogist, 42, 631–650. Marchesi, C., Garrido, C.J., Godard, M., Proenza, J.A., Gervilla, F. & Blanco-Moreno, J. 2006. Petrogenesis of highly depleted peridotites and gabbroic rocks from the Mayari–Baracoa ophiolite belt (eastern Cuba). Contributions to Mineralogy and Petrology, 151, 171–736. McKenzie, D. & Bickle, M.J. 1988. The volume and composition of melt generated by extension of the lithosphere. Journal of Petrology, 29, 625–679. Menneken, M., Nemchin, A.A., Geisler, T., Pidgeon, R.T. & Wilde, S.A. 2007. Hadean diamonds in zircon from Jack Hills, Western Australia. Nature, 448, 917–920.

913

Menzies, A., Westerlund, K., Grutter, H., Gurney, J., Carlson, J., Fung, A. & Nowicki, T. 2004. Peridotitic mantle xenoliths from kimberlites on the Ekati Diamond Mine property, N.W.T., Canada: major element compositions and implications for the lithosphere beneath the central Slave craton. Lithos, 77, 395–412. Menzies, A.H., Shirey, S.B., Carlson, R.W. & Gurney, J.J. 1999. Re–Os systematics of Newlands peridotite xenoliths: implications for diamond and lithosphere formation. In: Gurney, J.J., Gurney, J.L., Pascoe, M.D. & Richardson, S.H. (eds) Proceedings of the 7th International Kimberlite Conference. Red Roof Design, Cape Town, 566–583. Menzies, A.H., Carlson, R.W., Shirey, S.B. & Gurney, J.J. 2003. Re–Os systematics of diamond-bearing eclogites from the Newlands kimberlite. Lithos, 71, 323–336. Menzies, M. & Murthy, R.V. 1980. Enriched mantle: Nd and Sr isotopes in diopsides from kimberlite nodules. Nature, 283, 634–636. Menzies, M., Xu, Y., Zhang, H. & Fan, W. 2007. Integration of geology, geophysics and geochemistry: A key to understanding the North China Craton. Lithos, 96, 1–21. Menzies, M.A. 1990. Archaean, Proterozoic and Phanerozoic lithospheres. In: Menzies, M.A. (ed.) Continental Mantle. Clarendon Press, Oxford, 67–86. Morel, M.L.A., Simon, N.S.C., Davies, G.R. & Pearson, D.G. 2006. Modification of cratonic lithosphere: Influence of tectono-magmatic events on Kaapvaal craton (South Africa). Geochimica et Cosmochimica Acta, 70, A478. Neumann, E.R., Wulff-Pederson, P.E., Johnsen, K., Andersen, T. & Krogh, E. 1995. Petrogenesis of spinel harzburgite and dunite suite xenoliths from Lanzarote, eastern Canary Islands; implications for the upper mantle. Lithos, 35, 83–107. Nguuri, T.K., Gore, J. & James, D.E. et al. 2001. Crustal structure beneath southern Africa and its implications for the formation and evolution of the Kaapvaal and Zimbabwe cratons. Geophysical Research Letters, 28, 2501– 2504. Niu, Y. 2004. Bulk-rock major and trace element compositions of abyssal peridotites: Implications for mantle melting, melt extraction and postmelting processes beneath mid-ocean ridges. Journal of Petrology, 45, 2423–2458. Nixon, P.H. & Boyd, F.R. 1973. Petrogenesis of the granular and sheared ultrabasic nodule suite in kimberlite. In: Nixon, P.H. (ed.) Lesotho Kimberlites. Cape and Transvaal, Cape Town, 48–56. Nowell, G.M., Pearson, D.G., Parman, S.W., Luguet, A.L. & Hanski, E. 2008. Precise and accurate 186 Os/188 Os and 187 Os/188 Os measurements by multicollector plasma ionisation mass spectrometry, part II: Laser ablation and its application to single-grain Pt–Os and Re–Os geochronology. Chemical Geology, 248, 394–426. Parkinson, I.J., Hawkesworth, C.J. & Cohen, A.S. 1998. Ancient mantle in a modern arc: Osmium isotopes in Izu–Bonin–Mariana Forearc peridotites. Science, 281, 2011–2013. Parman, S.W. & Grove, T.L. 2004. Harzburgite melting with and without H2 O: Experimental data and predictive modeling. Journal of Geophysical Research: Solid Earth, 109, paper number B02201. Parman, S.W., Grove, T.L., Dann, J.C. & de Wit, M.J. 2004. A subduction origin for komatiites and cratonic lithospheric mantle. South African Journal of Geology, 107, 107–118. Pearson, D.G. & Harris, J.W. 2006. Diamond geochronology—A record of continental lithosphere evolution. Geochimica et Cosmochimica Acta, 70, A478. Pearson, D.G. 1999. The age of continental roots. Lithos, 48, 171–194. Pearson, D.G. & Nowell, G.M. 2002. The continental lithospheric mantle: characteristics and significance as a mantle reservoir. Philosophical Transactions of the Royal Society of London, 360, 2383–2410. Pearson, D.G., Carlson, R.W., Shirey, S.B., Boyd, F.R. & Nixon, P.H. 1995a. Stabilisation of Archean lithospheric mantle: A Re–Os isotope study of peridotite xenoliths from the Kaapvaal Craton. Earth and Planetary Science Letters, 134, 341–357. Pearson, D.G., Shirey, S.B., Carlson, R.W., Boyd, F.R., Pokhilenko, N.P. & Shimizu, N. 1995b. Re–Os, Sm–Nd and Rb–Sr isotope evidence for thick Archaean lithospheric mantle beneath the Siberia craton modified by multistage metasomatism. Geochimica et Cosmochimica Acta, 59, 959–977. Pearson, D.G., Shirey, S.B., Harris, J.W. & Carlson, R.W. 1998. Sulfide inclusions in diamonds from the Koffiefontein kimberlite, S. Africa: Constraints on diamond ages and mantle Re–Os systematics. Earth and Planetary Science Letters, 160, 311–326. Pearson, D.G., Irvine, G.J., Carlson, R.W., Kopylova, M.G. & Ionov, D.A. 2002. The development of lithospheric mantle keels beneath the earliest continents: time constraints using PGE and Re–Os isotope systematics. In: Fowler, C.M.R., Ebinger, C.B. & Hawkesworth, C.J. (eds) The Early Earth: Physical, Chemical and Biological Development. Geological Society, London, Special Publications, 199, 65–90.

914

D. G . P E A R S O N & N. W I T T I G

Pearson, D.G., Canil, D. & Shirey, S.B. 2003. Mantle samples included in volcanic rocks: xenoliths and diamonds. In: Holland, H.D. & Turekian, K.K. (eds) Treatise on Geochemistry. Elsevier, Amsterdam, 171– 275. Pearson, N.J., Alard, O., Griffin, W.L., Jackson, S.E. & O’Reilly, S. 2002. In situ measurement of Re–Os isotopes in mantle sulfides by laser ablation multicollector inductively coupled plasma mass spectrometry: Analytical methods and preliminary results. Geochimica et Cosmochimica Acta, 66, 1037–1050. Pearson, D.G., Irvine, G.J., Ionov, D.A., Boyd, F.R. & Dreibus, G.E. 2004. Re-Os isotope systematics and Platinum Group Element fractionation during mantle melt extraction: A study of massif and xenolith peridotite suites. Chemical Geology, 208, 29–59. Pearson, D.G., Parman, S.W. & Nowell, G.M. 2007. A link between large mantle melting events and continent growth seen in Osmium isotopes. Nature, 449, 202–205. Percival, J.A. & Pysklywe, A. 2007. Are Archean lithospheric keels inverted? Earth and Planetary Science Letters, 254, 393–403. Pollack, H.N. & Chapman, D.S. 1977. On regional heat flow, geotherms and lithospheric thickness. Tectonophysics, 38, 279–296. Richardson, S.H., Gurney, J.J., Erlank, A.J. & Harris, J.W. 1984. Origin of diamonds in old enriched mantle. Nature, 310, 198–202. Richardson, S.H., Erlank, A.J., Harris, J.W. & Hart, S.R. 1990. Eclogitic diamonds of Proterozoic age from Cretaceous kimberlites. Nature, 346, 54– 56. Richardson, S.H., Harris, J.W. & Gurney, J.J. 1993. Three generations of diamonds from old continental mantle. Nature, 366, 256–258. Richardson, S.H., Shirey, S.B., Harris, J.W. & Carlson, R.W. 2001. Archean subduction recorded by Re–Os isotopes in eclogitic sulfide inclusions in Kimberley diamonds. Earth and Planetary Science Letters, 191, 257–266. Richardson, S.H., Shirey, S.B. & Harris, J.W. 2004. Episodic diamond genesis at Jwaneng, Botswana, and implications for Kaapvaal craton evolution. Lithos, 77, 143–154. Ringwood, A.E. 1977. Synthesis of the pyrope–knorringite solid-solution series. Earth and Planetary Science Letters, 36, 443–448. Rudnick, R.L., McDonough, W.F. & Orpin, A. 1994. Northern Tanzanian peridotite xenoliths: a comparison with Kaapvaal peridotites and inferences on metasomatic interactions. In: Meyer, H.O.A. & Leonardos, O.H. (eds) Proceedings of the 5th International Kimberlite Conference. CRPM, Brasilia, 336–353. Saltzer, R.L., Chatterjee, N. & Grove, T.L. 2001. The spatial distribution of garnets and pyroxenes in mantle peridotites: pressure–temperature history of peridotites from the Kaapvaal Craton. Journal of Petrology, 42, 2215–2229. Schmidberger, S.S. & Francis, D. 1999. Nature of the mantle roots beneath the North American craton: mantle xenolith evidence from Somerset Island kimberlites. Lithos, 48, 195–216. Schmitz, M.D. & Bowring, S.A. 2000. The significance of U–Pb zircon dates in lower crustal xenoliths from the southwestern margin of the Kaapvaal craton, southern Africa. Chemical Geology, 172, 59–76. Schmitz, M.D., Bowring, S.A., deWit, M.J. & Gartz, V. 2004. Subduction and terrane collision stabilize the western Kaapvaal craton tectosphere 2.9 billion years ago. Earth and Planetary Science Letters, 222, 363–376. Schulze, D.J. 1986. Calcium anomalies in the mantle and a subducted metaserpentinite origin for diamonds. Nature, 319, 483–485. Schulze, D.J. 1989. Constraints on the abundance of eclogite in the upper mantle. Journal of Geophysical Research, 94, 4205–4212. Shimizu, N., Pokhilenko, N.P., Boyd, F.R. & Pearson, D.G. 1997. Geochemical characteristics of mantle xenoliths from Udachnaya kimberlite pipe. Geologiya i Geofizika, 38, 194–205. Shirey, S.B., Carlson, R.W. & Richardson, S.H. et al. 2001. Archean emplacement of eclogitic components into the lithospheric mantle during

formation of the Kaapvaal craton. Geophysical Research Letters, 28, 2509– 2512. Shirey, S.B., Harris, J.W. & Richardson, S.H. et al. 2002. Diamond genesis, seismic structure, and evolution of the Kaapvaal–Zimbabwe craton. Science, 297, 1683–1686. Simon, N.S.C., Irvine, G.J., Davies, G.R., Pearson, D.G. & Carlson, R.W. 2003. The origin of garnet and clinopyroxene in ‘depletedh Kaapvaal peridotites. Lithos, 71, 289–322. Simon, N.S.C, Neumann, E.-R., Bonadiman, C., Coltorti, M., Delpech, G., Gregoire, M. & Widom, E. 2008. Ultra-refactory Domains in the Oceanic Mantle Lithosphere Sampled as Mantle Xenoliths at Ocean Islands. Journal of Petrology, 49, 1223–1251. Smith, C.B., Pearson, D.G., Bulanova, G.P., Beard, A.D. Carlson, R.W., Sims, K., Chimuka, L. & Muchemwa, E. 2008. Extremely depleted lithospheric mantle and diamonds beneath the southern Zimbabwe Craton. In: Extended Abstracts 9th International Kimberlite Conference, Frankfurt. Stachel, T. & Harris, J.W. 2008. The origin of cratonic diamonds—constraints from mineral inclusions. Ore Geology Reviews, doi: 10.1016/j.oregeorev. 2007.05.002. Stachel, T., Viljoen, K.S., Brey, G. & Harris, J.W. 1998. Metasomatic processes in lherzolitic and harzburgitic domains of diamondiferous lithospheric mantle. Earth and Planetary Science Letters, 159, 1–12. Tankard, A.J., Jackson, M.P.A., Eriksson, K.A., Hobday, D.K., Hunter, D.R. & Minter, W.E. 1982. Crustal Evolution of Southern Africa: 3.8 Billion Years of Earth History. Springer, New York. van Achterbergh, E., Griffin, W.L. & Stiefenhofer, J. 2001. Metasomatism in mantle xenoliths from the Letlhakane kimberlites: estimation of element fluxes. Contributions to Mineralogy and Petrology, 141, 397–414. Walker, R.J., Carlson, R.W., Shirey, S.B. & Boyd, F.R. 1989. Os, Sr, Nd, and Pb isotope systematics of southern African peridotite xenoliths: Implications for the chemical evolution of subcontinental mantle. Geochimica et Cosmochimica Acta, 53, 1583–1595. Walter, M.J. 1998. Melting of garnet peridotite and the origin of komatiite and depleted lithosphere. Journal of Petrology, 39, 29–60. Walter, M.J. 1999. Melting residue of fertile peridotite and the origin of cratonic lithosphere. In: Fei, Y., Bertka, C.M. & Mysen, B.O. (eds) Mantle Petrology: Field observations and high pressure experimentation: A tribute to Francis R. (Joe) Boyd. The Geochemical Society, Special Publications, 6, 225–239. Walter, M.J. 2003. Melt extraction and compositional variability in mantle lithosphere. In: Holland, H.D. & Turekian, K.K. (eds) Treatise on Geochemistry. Elsevier, Amsterdam, 363–394. Webb, M., 2008. Re–Os isotope constraints on the age of the lithospheric mantle beneath West Greenland. MSc thesis, University of Durham. Westerlund, K., Shirey, S.B., Richardson, S.H., Carlson, R.W., Gurney, J.J. & Harris, J.W. 2006. A subduction wedge origin for Paleoarchean peridotitic diamonds and harzburgites from the Panda kimberlite, Slave craton, evidence from Re–Os isotope systematics. Contributions to Mineralogy and Petrology, 152, 275–294. Wittig, N., Webb, M., Pearson, D.G., Nowell, G.M., Jensen, S.M. & Sand, K.K. 2007. Archetypal Archean lithosphere from West Greenland implicates shallow melting at .3.0 Ga. Geochimica et Cosmochimica Acta, 71, A1121. Wittig, N., Pearson, D.G. & Webb, M. et al. 2008a. Origin of cratonic lithospheric mantle roots: A geochemical study of peridotites from the North Atlantic Craton, West Greenland. Earth and Planetary Science Letters, in press. Wittig, N., Pearson, D.G., Webb, M., Ottley, C.J., Sand, K.K., Jensen, S.M. & Nowell, G.M. 2008b. Evolution of the mantle root beneath the North Atlantic Craton. In: Extended Abstracts 9th International Kimberlite Conference, Frankfurt.

Received 10 January 2008; revised typescript accepted 19 March 2008. Scientific editing by Rob Strachan