Secular evolution of the continental crust - Wiley Online Library

Geochemistry Geophysics Geosystems

3

Article

G

Volume 9, Number 12 5 December 2008 Q12010, doi:10.1029/2008GC002262

AN ELECTRONIC JOURNAL OF THE EARTH SCIENCES Published by AGU and the Geochemical Society

ISSN: 1525-2027

Click Here for

Full Article

Secular evolution of the continental crust: Implications for crust evolution models Hugh Rollinson Department of Geographical, Earth and Environmental Sciences, University of Derby, Kedleston Road, Derby DE22 1GB, UK ([email protected])

[1] The present-day flux from mantle to crust is basaltic and yet the average composition of the continental crust is andesitic. This is the crust composition paradox. A new solution to this paradox is proposed whereby the secular evolution in the composition of the continental crust reflects a changing flux from mantle to crust over time. Thus it is proposed that the present-day composition of the continental crust is a time-integrated average. Crustal growth curves show that 48–54% of the continental crust was formed by the end of the Archaean. A mass balance model based upon a tonalite-trondhjemite-granodiorite compositional model for the Archaean continental crust shows that the post-Archaean mantle to crust flux was predominantly basaltic and likely a mix of arc-plume basalts. Trace element modeling, however, reveals that additional processes also contributed to the average crust composition. Balancing Y, Ho, and Yb concentrations requires a garnetiferous mafic granulite lower Archaean crust, which in turn drives the post-Archaean flux toward a high mg # andesite. This suggests that there was a slab melt contribution to the continents, in addition to basalt. An excess of fluid mobile elements in the continental crust can be explained either by the addition of a slab melt or small fraction melts. A deficiency in Sr requires that the post-Archaean crustal composition has been modified by erosion. Both Archaean and post-Archaean continental crust contain contributions from basalt and a slab melt. In the Archaean crust the slab melt contribution is dominant. In the post-Archaean crust the basaltic contribution is dominant. Components: 8326 words, 5 figures, 1 table. Keywords: continental crust; Archaean; tonalite-trondhjemite; arc magmas. Index Terms: 1020 Geochemistry: Composition of the continental crust; 1009 Geochemistry: Geochemical modeling (3610, 8410); 1031 Geochemistry: Subduction zone processes (3060, 3613, 8170, 8413). Received 8 February 2008; Revised 3 October 2008; Accepted 27 October 2008; Published 5 December 2008. Rollinson, H. (2008), Secular evolution of the continental crust: Implications for crust evolution models, Geochem. Geophys. Geosyst., 9, Q12010, doi:10.1029/2008GC002262.

1. Introduction [2] One of the outstanding problems in determining the origin of the continental crust is the discrepancy between the average composition of the continental crust today, which is andesitic [Taylor and McLennan, 1985; Rudnick and Gao,

Copyright 2008 by the American Geophysical Union

2003], and the composition of the present-day mass flux from mantle to crust across the Moho, which is widely regarded as basaltic [Davidson and Arculus, 2006]. This problem has become known as the ‘‘continental crust composition paradox.’’ [3] Previous solutions to the crust composition paradox have suggested that the composition of 1 of 14


3

G

rollinson: secular evolution of the continental crust 10.1029/2008GC002262

the continental crust has been modified through chemical weathering [Albarede, 1998] or by the addition of small fraction mantle melts [O’Nions and McKenzie, 1988]. However, these mechanisms fail as the principal means by which crustal compositions have been modified. The weathering and erosion model would require the continental crust to be peraluminous, which it is not [Rudnick, 1995], and that its fractionated rare earth element (REE) pattern be produced from that of basalt by weathering, which is improbable [Davidson and Arculus, 2006]. The small melt fraction mechanism [O’Nions and McKenzie, 1988] fails because these melts are also basaltic and so cannot shift the bulk composition toward an andesitic composition. [4] An alternative approach is that the continental crust was formed from basaltic magmas but that it was modified by the removal of mafic and ultramafic cumulates from the lower crust back into the mantle during or shortly after continent genesis. The effect of this process is to drive basaltic compositions toward an andesitic bulk composition by the removal of dense garnet pyroxenites [Hawkesworth and Kemp, 2006a], formed perhaps by the metamorphism of arc lower-crustal cumulates during the continental arc stage of crustal accretion [Lee et al., 2007]. Documented examples of lower crustal delamination include the eastern block of the North China Craton [Gao et al., 2004], the Sierra Nevada batholith in the United States [Boyd et al., 2004; Lee et al., 2007], and the Talkeetna Arc, Alaska [Kelemen et al., 2003a]. Plank [2005] estimated from Th/La mass balance that as much as 25 to 60% of the basaltic parent to the continental crust was delaminated during crust generation and Lee et al. [2007] estimated that between 60 and 75% of the original basalt was delaminated in the formation of the crust of the Sierran and Peninsular Ranges batholiths in the western United States. [5] A different type of solution to the crust composition paradox, proposed by Kelemen et al. [2003a], is that the flux from mantle to crust is andesitic and so there is no paradox. High mg # andesites (54 – 65 wt % SiO2, mg # >0.45, a lithological grouping which includes adakites) formed in the Aleutian arc have compositions which are close to average crust [Kelemen et al., 2003a]. Many authors think that these rocks formed from a slab melt which reacted with the overlying mantle wedge [e.g., Kelemen et al., 2003a], although this view is not universally held [Macpherson et al., 2006]. This model also has

implication for how Archaean felsic crust might have formed [Martin et al., 2005]. [6] This paper proposes a new type of solution to the continental crust paradox. Here a new model is proposed to explain the andesitic composition of the continental crust by invoking a change in the supra-Moho flux through time, from dominantly silicic in the Archaean to basalt-andesite in the post-Archaean period. Thus the present bulk composition of the continental crust is a time-integrated average of a process of secular evolution which has been taking place over Earth history. The firstorder observation which drives this hypothesis is the bimodal nature of Archaean continental crust [Barker and Arth, 1976] in which, today, the dominant component is the highly siliceous magmatic tonalite-trondhjemite-granodiorite (TTG) suite. The model follows the work of previous authors [e.g., Hawkesworth and Kemp, 2006a] in that crustal growth must be seen as the balance between a series of fluxes both into and out of the crustal reservoir [Rollinson, 2007]. However, here an ‘‘end-member’’ model is adopted such that crustal growth over time has been dominated by fluxes into the continental crust and that fluxes out have been comparatively minor. In detail the model explores the secular evolution hypothesis through mass balance calculations which, to a first approximation, are consistent with the model. However, some trace elements do not fit the ‘‘secular change model’’ exactly, implying, as proposed by previous workers that secondary modifying processes also play a role.

2. Evidence for the Secular Evolution of the Continental Crust [7] Given that the Earth’s continental crust was extracted from the mantle, albeit indirectly, it is reasonable to suppose that a decline in mantle potential temperature over geological time [Richter, 1988; Campbell and Griffiths, 1992] will have influenced aspects of the evolution of the continental crust. This change is documented in the geological record, in particular in the upper continental crust. The principal evidence comes from granitoid geochemistry, for there is evidence from the REE and Nd-isotope geochemistry of granitoids that their compositions have changed over geological time [Martin, 1986; DePaolo, 1988]. A similar secular change has been noted in the Rb/Sr ratios and Th/U ratios of granitoids [Ellam and Hawkesworth, 1988; Kemp and Hawkesworth, 2003]. There are also differences in average heat 2 of 14


3

G


flow data from Archaean, Proterozoic, and Palaeozoic crust [Jaupart and Mareschal, 2003], apparently reflecting increased concentrations of crustal heat producing elements over geological time. [8] In addition a secular change in upper crustal compositions is recorded by clastic sedimentary rocks [Taylor and McLennan, 1985]. There are recognizable differences in REE patterns, incompatible element concentrations and ratios, and Pbisotope ratios between Archaean and post-Archaean clastic sediments [Condie, 1993; McLennan and Hemming, 1992]. However, the significance of these measurements is uncertain for they probably record the combined effect of compositional changes in the protolith coupled with secular change in weathering and diagenetic processes related to the evolution of the Earth’s atmosphere.

[13] 2. There has been minimal loss of mass from the continental crust over geological time. In other words this model sits in direct contrast to the delamination model of crustal growth as proposed by Hawkesworth and Kemp [2006a] and Kelemen et al. [2003a]. [14] 3. The mass of continental crust formed by the end of the Archaean can be estimated from the crustal growth curve. [15] 4. There was a change in the mode of crust formation at the end of the Archaean at about 2.5 Ga [Martin, 1986; Kemp and Hawkesworth, 2003]. [16] 5. The bulk composition of the continental crust today is known. In this paper the composition calculated by Rudnick and Gao [2003] is used.

[9] More contentious is whether or not seismic velocity measurements record a secular change in lower continental crust compositions. There is evidence to indicate that the present-day lower continental crust is basaltic [Rudnick and Fountain, 1995], although Behn and Kelemen [2003] have proposed a wider compositional range. A more felsic composition is inferred for the Archaean lower continental crust [Rudnick and Presper, 1990, section 3.2]. The matter is complex because of the varied nature of Archaean lower crustal xenoliths, both felsic- and mafic-dominated suites are recorded, and the observation that Archaean granulite terrains tend to become more mafic with depth [Rudnick and Gao, 2003].

[17] 6. Post-Archaean crustal growth includes contributions from both arc and plume magmas, although the dominant contribution is from arc magmas [Rudnick, 1995; Plank and Langmuir, 1998; Barth et al., 2000; Hawkesworth and Kemp, 2006a].

[10] There is also evidence that there was a secular evolution of crust compositions during the Archaean [Martin and Moyen, 2002]. This observation has been taken into account in the modeling below.

[19] The model presented here requires an estimate of the mass of continental crust formed by the end of the Archaean. This volume was calculated from the crustal growth curves of Nagler and Kramers [1998]. These are based on the Nd-isotopic depletion of the upper mantle and indicate that between 48% and 54% of the continental crust formed by 2.5 Ga (Figure 1). Other plausible growth curves are those of Collerson and Kamber [1999] based upon the depletion of the mantle in Th, U, and Nb, and Condie [2000], based upon a compilation of U-Pb zircon crystallization ages. However, Hawkesworth and Kemp [2006b] have shown from combined oxygen-Hf isotope measurements in zircon that the crystallization age of some zircons is not the same as the time of the derivation of their parent granitoid from the mantle, which may have been much earlier. This means that crustal growth curves based on zircon ages, such as that of Condie [2000], must be considered as minimum estimates of the crustal

3. A Mass Balance Model for the Secular Evolution of the Continental Crust [11] Central to this paper are mass balance calculations to test the secular evolution hypothesis proposed here. The principal goal of these calculations is to estimate the composition of the post-Archaean flux into the continental crust. The calculations are based upon the following assumptions: [12] 1. The continental crust has grown progressively over time in a nonlinear manner [Nagler and Kramers, 1998; Collerson and Kamber, 1999; Condie, 2000].

[18] 7. The composition of the Archaean continental crust is dominated by the compositions of magmas of the tonalite-trondhjemite-granodiorite suite [Barker and Arth, 1976; Martin, 1986]. Two variants of this model are presented below.

3.1. Mass of Continental Crust Formed by the End of the Archaean

3 of 14


3

G


view is rejected here for it has become clear since the work of Taylor and McLennan [1986] that although in some cratons greenstone belts outcrop over a relatively large surface area, they are relatively shallow structures [De Wit and Ashwal, 1997; Stettler et al., 1995; Ranganai et al., 2008] and so must make up only a small component of the total volume of the Archaean crust. Here we argue that the major part of the Archaean middle continental crust and in some cases the Archaean lower continental crust [Rollinson and Tarney, 2005] is TTG in composition.

Figure 1. Crustal growth curves after Nagler and Kramers [1998], Collerson and Kamber [1999], and Condie [2000]. The growth curves of Nagler and Kramers [1998] are used to estimate that the mass of continental crust formed by the end of the Archaean at 2.5 Ga is between 48 and 54% the mass of the presentday continental crust. Curves A and B represent two different assumptions about the amount of crust formed in the Hadean.

mass. Crustal growth models in which a large volume of continental crust formed very early in Earth history [Armstrong, 1981] are rejected on the grounds that they are not supported by timeintegrated isotopic and geochemical data for the Earth’s mantle [Kramers, 2007; Shirey et al., 2008]. Hence, the crustal growth curve adopted here is a cumulative growth curve for juvenile crust over time and is independent of any crustal recycling that may have also taken place. It indicates progressive, nonlinear crustal growth.

[21] Two models of Archaean crust are tested here. Model 1 is fundamentally TTG in composition but includes a greenstone belt component. The TTG component has two features. First, it reflects the possible secular evolution of TTGs that took place within the Archaean [Martin and Moyen, 2002], and second it is differentiated into a potassic granite upper crust and a depleted felsic granulite lower crust of TTG composition. There is no mafic lower crust in this model. Model 2 is identical to model 1 except that the lower crust is a mafic granulite (Figure 2). The bulk compositions of both models plot in the granodiorite field of the Middlemost [1994] total alkalis versus silica rock

3.2. Average Composition of Archaean Continental Crust [20] An assumption, central to the model proposed here, is that Archaean continental crust, as evidenced by Archaean cratons worldwide, is dominated by the felsic rocks of the TTG suite. This assumption is fundamentally different from that used by earlier workers on the composition of the continental crust. For example, Taylor and McLennan [1986] argued that the mafic rocks of Archaean greenstone belts were the dominant Archaean rock type; they concluded from a study of the distribution of heat-producing elements in Archaean sediments that the bulk Archaean crust was a mixture of a 67% mafic greenstone belt component and a 33% felsic TTG component. This

Figure 2. Archaean crustal models. Model 1 Archaean crust assumes that the crust was differentiated into 10% potassic granite upper crust, 60% TTG middle crust (a weighted average of Archaean crust with ages >3.5 Ga, 3.5 – 3.0 Ga, and 3.0– 2.5 Ga), and 30% felsic granulite facies TTG lower crust. Five percent of the crustal volume is greenstone belt comprising 90% tholeiite and 10% komatiite. Model 2 is the same as model 1 except that the lower crust is 70% mafic granulite. Average compositions are given in Table 1. 4 of 14


3

G


classification diagram. Both models assume that a typical Archaean greenstone belt comprises 90% Archaean tholeiite and 10% high-alumina komatiite [De Wit and Ashwal, 1997] and that greenstone belts make up only 5% of Archaean crust. In this study the major element compositions of Archaean tholeiites and komatiites were taken from averages from Rollinson [1999], Polat and Hofmann [2003] (tholeiites), and Parman et al. [2003] (komatiites).

felsic granulite. This is to reflect our uncertainty in the composition of the lower Archaean crust and to represent a different end-member. The mafic granulite is an average of the basic granulites described by Weaver and Tarney [1980] from the Lewisian of northwest Scotland; trace element compositions are modeled on their sample 20N.

[22] In model 1 the felsic (TTG) component of Archaean crust is assumed to be differentiated and comprises an upper potassic granite layer (10%), a middle crust comprising average TTG as discussed below (60%), and a lower crust made up of Archaean granulite facies TTG (30%). Average potassic granite compositions are calculated from data in the work of Sylvester [1994] and Kleinhanns et al. [2003] and from the author’s unpublished data from Sierra Leone. Average middle crust compositions were calculated from average TTG data for crust older than 3.5 Ga, crust between 3.5 and 3.0 Ga, and crust 3.0 to 2.5 Ga in age. A weighted average was calculated assuming 15% crustal growth by 3.5 Ga, 28% by 3.0 Ga, and 50% by 2.5 Ga. Major element compositions for early, middle, and late Archaean TTGs were taken from Martin et al. [2005]. Trace element data were taken from the following sources: Condie [2005], Martin et al. [2005], Kamber et al. [2002], Feng and Kerrich [1992], and the author’s unpublished data from Sierra Leone (Table 1). [23] In model 1 Archaean lower crust is assumed to be predominantly felsic and so differs from that of Rudnick and Fountain [1995] whose Archaean crustal composition was based on mafic xenolith data for the lower crust. Support for a felsic Archaean lower crust comes from the deep crustal rocks found in Archaean granulite belts. These are variable in composition and include both felsic and mafic varieties, but with felsic lithologies dominant [Niu and James, 2002], a view which is consistent with xenolith studies [Rudnick and Gao, 2003], recent geochemical modeling [Rollinson and Tarney, 2005; Hawkesworth and Kemp, 2006b], and the Th/La mass balance arguments of Plank [2005]. Here average Archaean felsic lower crust was calculated from data for the Lewisian granulites [Moorbath et al., 1969; Pride and Mucke, 1980; Rollinson and Windley, 1980; Weaver and Tarney, 1981], with Cs data from Rudnick and Taylor [1987].

[25] The mass balance calculation developed here uses the compositions of the Archaean continental crust computed in models 1 and 2 above and the mass of continental crust formed by the end of the Archaean (48– 54%). Since the composition of average continental crust is known [Rudnick and Gao, 2003], it is possible to calculate the composition of the post-Archaean net flux from mantle to continental crust.

[24] Model 2 is identical to model 1 except that the Archaean lower crust is 70% mafic granulite 30%

3.3. Results of the Mass Balance Modeling

[26] The major element results of the mass balance calculation are plotted on total alkalis versus silica and magnesium number versus silica diagrams (Figure 3). For an Archaean TTG crust (model 1) the post-Archaean contribution to the continental crust is between basalt and basaltic andesite in composition, with SiO 2 in the range 52.1 to 53.9 wt %, MgO 7.3 to 8.1 wt%, K2O 1.76 wt%, Na2O 1.4 to 1.7 wt %, and a mg # between 58 and 58.3 (Table 1, Figure 3). For Archaean crust with a mafic lower crust the post-Archaean contribution is a basaltic andesite with SiO2 in the range 55.7 to 56.7 wt %, MgO 6.5 to 7.0 wt%, K2O 1.8 wt%, Na2O 2.0–2.2 wt %, and a mg # between 57.8 and 58.2 (Table 1, Figure 3). These compositions plot within the range of the high mg # andesites of Kelemen et al. [2003a, Figure 3b]. [27] The calculated trace element compositions of the post Archaean flux into the continental crust calculated according to mass balance models 1 and 2 are presented on mantle normalized trace element diagrams (Figure 4) and trace element ratio diagrams (Figure 5) and plotted relative to a mix of mafic plume and arc magmas, reflecting the likely dual contribution to crustal growth [Rudnick, 1995; Plank and Langmuir, 1998; Barth et al., 2000; Kemp and Hawkesworth, 2003]. The calculated concentrations for models 1 and 2 show trace element patterns progressively enriched in the more incompatible elements and with a positive Pb anomaly and give a close match to a mafic composition representing a 20:80 plume-arc mix for the elements Nb, Ta, La, Ce, Pr, Nd, Sm, Zr, Hf, Eu, and Ti. In addition Archaean crust with a mafic 5 of 14

mg# Cs Rb Ba Th U K Nb Ta La Ce Pb Pr Sr Nd Sm Zr Hf Eu Ti Y Ho Yb

60.60 0.70 15.90 6.70 0.10 4.70 6.40 3.10 1.80 0.10 100.10 55.56 2.00 49.00 456.00 5.60 1.30 14942 8.00 0.70 20.00 43.00 11.00 4.90 320.00 20.00 3.90 132.00 3.70 1.10 4197 19.00 0.77 1.90

70.29 0.39 15.44 2.93 0.04 1.01 3.06 4.65 2.06 0.13 100.00 38.05 2.58 75.69 419.43 3.14 0.78 18944 6.05 0.39 26.10 48.24 15.80 5.57 327.69 21.12 3.76 167.75 4.38 0.98 2296 11.10 0.52 0.92

TTG >3.5

70.57 0.36 15.55 2.77 0.06 1.08 3.00 4.70 1.76 0.14 100.00 41.10 4.30 63.87 336.00 4.40 1.86 14102 7.10 0.60 25.71 45.44 8.79 4.81 603.44 16.50 2.80 188.44 4.42 0.80 2324 8.04 0.29 0.79

TTG < 3.5 > 3.0

69.29 0.39 15.73 2.95 0.05 1.38 3.27 4.76 2.03 0.15 100.00 45.44 4.30 66.52 744.80 6.70 1.13 15855 5.74 0.84 28.38 58.88 20.10 6.32 503.26 21.13 3.28 145.28 4.70 0.90 2173 9.16 0.33 0.55

TTG < 3.0

69.92 0.38 15.60 2.90 0.05 1.19 3.14 4.71 1.97 0.14 100.00 42.29 3.79 68.58 540.90 5.04 1.22 16326 6.19 0.64 27.00 52.19 15.87 5.70 476.64 19.92 3.30 163.24 4.53 0.90 2249 9.45 0.38 0.72

30% > 3.5 26% 3.5-3.0 44% < 3.0

Weighted TTG Crust

64.98 0.54 16.31 4.45 0.07 2.39 5.50 4.63 0.97 0.16 100.00 48.87 0.27 6.64 694.70 0.28 0.08 7944 4.23 0.44 29.56 57.17 9.04 6.80 434.60 26.77 3.11 151.00 6.93 1.15 3227 9.33 0.31 0.64

Archaean Lower Crust

49.68 0.88 17.43 10.29 0.18 7.13 11.37 2.35 0.62 0.07 100.00 55.25 0.27 3.00 286.00 0.94 0.35 5161 7.00 0.80 13.00 34.00 4.00 4.80 200.00 21.90 5.50 99.00 2.86 1.71 5285 42.00 1.57 4.61

Mafic Lower Crust

72.58 0.29 14.27 1.84 0.03 0.42 1.43 4.54 4.52 0.09 100.00 29.07 3.66 170.32 174.71 26.47 4.37 37569 12.24 1.42 68.66 116.33 26.30 16.26 272.15 51.99 8.05 235.33 5.44 1.21 1730 22.70 0.76 1.96

Archaean Granite

68.70 0.42 15.68 3.26 0.05 1.47 3.68 4.67 1.92 0.14 100.00 44.63 2.72 60.17 550.42 5.75 1.19 15936 6.20 0.66 31.93 60.10 14.86 7.09 443.58 25.18 3.72 166.78 5.34 1.00 2491 10.74 0.40 0.82

10% grt 60% av TTG 30% lower crust

Differentiated Crust

65.49 0.49 15.91 4.48 0.08 2.47 4.91 4.19 1.85 0.12 100.00 49.53 2.72 59.41 464.60 5.89 1.25 15351 6.79 0.73 28.46 55.23 13.81 6.67 394.31 24.16 4.22 155.86 4.49 1.12 2923 17.60 0.66 1.65

10% grt 60% av TTG 30% lower crust 70% mafic

Differentiated Crust

51.50 0.97 14.33 11.68 0.20 9.01 10.16 1.85 0.19 0.11 100.00 57.89 1.44 30.02 46.13 0.48 0.09 6747 1.93 0.16 3.06 8.88 2.44 1.39 49.19 7.15 2.41 56.67 1.50 0.81 5094 17.86 0.65 1.72

90% thol 10% kom

Greenstone Belt Component

G

3

SiO2 TiO2 Al2O3 FeO(tot) MnO MgO CaO Na2O K2O P2O5

Average CCrust

Archaean Crust - Input Data (Major Element Normalized to 100%)

Table 13 (Sample). Compositions Used in Mass Balance Calculations [The full Table 1 is available in the HTML version of this article at http://www.g-cubed.org]



6 of 14


3

G


Figure 3. The results of the mass balance calculations shown on (a) a total-alkalis silica diagram and (b) a mg # (Mg+Fe(II)/Mg ratio) versus silica diagram showing that the calculated post-Archaean flux from mantle to continental crust is basaltic or basaltic andesite, consistent with what is observed today. The symbols represent model 1 and model 2 Archaean crust compositions, ‘‘Av CC’’ is the average composition of the continental crust [Rudnick and Gao, 2003], and the numbered points are calculated post-Archaean fluxes for model 1 and 2 Archaean crust. The compositional fields in Figure 3b are from Kelemen et al. [2003b] and Leat et al. [2002].

lower crust shows good agreement for the element Y, Ho, and Yb.

has important consequences for models of crustal evolution.

[28] To a first approximation these results provide strong support for the view that the presentday continental crust is a mixture, in almost equal proportions, of Archaean TTG magmas and basaltic magmas, from a mix of plume and arc sources. The TTG magmas represent the principal flux into the continental crust up until 2500 Ma, and the basaltic magmas represent the principal post-Archaean mantle to continental crust flux. This fundamental difference in flux into the continental crust over geological time

[29] It is evident, however, from Figure 4 that although a secular evolution model for the bulk composition of the continental crust works for a large number of major and trace elements, it does not fully explain all the trace element data. In particular there are two groups of trace elements which do not fit the model. The most significant are the subduction-fluid elements, Cs, Rb, Ba, Th, U, K, and Pb whose concentrations are higher that those predicted by a plume-arc basaltic mix. This implies that there is an additional flux of these elements into the continental crust in addition to 7 of 14


3

G


Figure 4. The results of the mass balance calculations plotted on mantle normalized [Sun and McDonough, 1989] trace element diagrams. Calculated compositions for model 1 Archaean crust are shown in grey and for model 2 Archaean crust are shown in black. The calculated compositions are compared with the composition of average arc (ARC) and average ocean island basalts (OIB) (Table 1) and plume-arc mixing curves, shown at 10% intervals. Those elements which do not fit the arc-plume mixing curves are identified and discussed in the text. Average OIB is from Sun and McDonough [1989]; average arc is calculated from data in the work of Turner et al. [1997], McCulloch and Gamble [1991], Hawkesworth and Kemp [2006a, 2006b], and Davidson and Arculus [2006].

Figure 5. Trace element ratio plots for model 1 and model 2 Archaean crust and the calculated post-Archaean flux, relative to a mix of OIB and ARC magmas. (a) Nb/La versus Sr/Nd ratio diagram shows that the Sr/Nd ratio is strongly divergent from that predicted for the 20% OIB mixing model; (b) Nb/La versus Zr/Hf ratio plot shows a closer correspondence between the predicted (20% OIB mixing) model and model 2 crust of this study. ‘‘H & K New continental crust’’ is the model composition for new crust used by Hawkesworth and Kemp [2006a], plotted here for comparison. ‘‘Av CC’’ is the bulk crust composition of Rudnick and Gao [2003]. 8 of 14


3

G


the ones assumed in the model. A second element is Sr, perhaps accompanied by Na and Ca, whose flux is lower than that predicted by a plume-arc basaltic mix, implying that the bulk continental crust has lost Sr.

4. Discussion 4.1. A Sr Deficit in the Continental Crust [30] The possibility that the continental crust has ‘‘lost’’ Sr was also noted by Hawkesworth and Kemp [2006b], who showed that the Sr/Nd ratio of the continental crust is lower than in geochemically similar arc magmas and that of new material being added to the crust (Figure 5a). A possible explanation for the missing Sr is that it was removed from the continental crust by chemical weathering and erosion. There is some support for this hypothesis from the Sr-isotope evolution curve for seawater [Shields and Veizer, 2002], which closely parallels the Nagler and Kramers [1998] crustal evolution curve from 3000 to 800 Ma, indicating that over much of geological time before metazoan life emerged, the dominant flux of Sr into the oceans has been from the continents. This flux therefore reflects an increased radiogenic Sr contribution from old crust. [31] In a semiquantitative test of this hypothesis the mass of Sr ‘‘lost’’ from the continental crust was calculated on the assumption that the continental crust is a 50:50 mix of Archaean TTG crust and basalt (an 80:20 arc-plume mix). Using the presentday crustal mass, it is possible to calculate that there should be 6.7 1018 kg of Sr in the continental crust [from Rudnick and Gao, 2003]. The amount predicted in the model however is 8.4–8.9 1018 kg of Sr indicating that 1.7–2.2 1018 kg of Sr is missing. The present-day concentration of Sr in the oceans is 8.1 ppm [Turekian, 1968] and so the mass of Sr in the ocean today is 1.1 1016 kg. Thus the missing mass of Sr equates to between 154 and 198 ‘‘ocean equivalents.’’ Given the ocean residence time for Sr is 5.1 Ma (http://www.mbari.org/chemsensor/summary.html), the missing mass of Sr could be removed from the continents to oceanic sediments by weathering in less than 1.0 Ga. [32] The alternative possibility that Sr was removed from the continental crust by the delamination of lower crustal mafic rocks has been proposed by Kelemen et al. [2003a] and Hawkesworth and Kemp [2006a]. Using element concentrations in lower crustal garnet granulites from the Kohistan

arc (Sr = 270 ppm) [Garrido et al., 2006] and in the amount of residual material returned to the mantle in the model of Hawkesworth and Kemp [2006a] (Sr = 354 ppm) it is possible to calculate the mass of missing rock required to explain the missing mass of Sr from the crust. Expressed as a fraction of the whole continental crust this equates to 39% and 30%, respectively. The garnet granulites from the Sierra Nevada, described by Lee et al. [2007], have much lower Sr and would require the amount of delaminated crust to exceed the present crustal mass.

4.2. An Excess of Subduction Fluid-Mobile Elements [33] The results of the mass balance calculations presented here indicate that there is an additional source of subduction-fluid-mobile elements in the post-Archaean contribution to the continental crust in excess of that provided by the arc-plume mix. Albarede [1998] recognized and discussed the problem of excess K in the continental crust and suggested that it was fluxed into the crust via nonmagmatic processes, perhaps through aqueous fluids. Here this observation is extended to the fluid mobile elements Rb, Ba, Th, U, K, and Pb (the Cs data are probably not sufficiently accurate). This enrichment is seen irrespective of the precise crustal compositions used and is a robust feature of this model. [34] This suite of elements is characteristic of subduction-related magmatism and is thought to originate in such a setting from the dewatering of the subducting slab. Typically, these fluids are removed from the slab to the mantle wedge and from there to the crust in a melt phase. Three mechanisms are proposed for the additional flux of subduction fluid-mobile elements. First, it is possible that a small volume of subduction fluids may be stored in hydrous phases such as amphibole and phlogopite in the subcontinental lithospheric mantle (SCLM). Thus these elements can be added to the lower and middle continental crust through the injection of small fraction mafic melts of the SCLM as proposed by O’Nions and McKenzie [1988]. In this study model potassic lamproites are used to characterize the melt flux. The mass of rock required to explain the element gain, expressed as a fraction of the mass of the continental crust, is calculated using data from Wilson [1989] and Chakrabarti et al. [2007]. Calculations show that the addition of 1–5% of mafic lamproite melt can explain the excess of Ba, U, K, Rb, Th, and Pb. 9 of 14


3

G


However, there is a problem with this mechanism for high fluid mobile element concentrations are known to be a very common feature of the upper continental crust [Rudnick and Gao, 2003], an observation which appears to be inconsistent with the ‘‘point source’’ distribution proposed through the addition of small, highly enriched mafic bodies. This apparent conflict can be resolved through the subsequent reworking of the lamproite additions to the crust during later crustal differentiation. [35] A second possible solution is that the elevated fluid-mobile element concentrations in the continental crust are the product of a differentiation process. This mechanism implies that the bulk continental crust is the geochemical complement to an extensive mafic residue, which has now been removed. This mechanism has been modeled using Nb/La versus Sr/Nd ratios by Hawkesworth and Kemp [2006a, 2006b] and supports crustal delamination as a major process. However, there are some uncertainties with the status of Sr in geochemical models of this type, as has been discussed above, which render this result equivocal. Further, any delamination process which involves garnet is improbable from the evidence of the elements Y, Ho, and Yb, as discussed below. [36] A third mechanism of enrichment of the continental crust in subduction fluid-mobile elements is through the addition to the crust of a slab-melt component. This will be discussed further below.

4.3. Balancing Y, Ho, and Yb [37] The preferred mass balance solution in this study is the model 2 solution for Archaean crust with a mafic lower crust. A model 1 solution, in which the entire Archaean continental crust (upper, middle, and lower crust) is made up of TTG magmas, apart from a small greenstone belt component, shows an enrichment in the elements Y, Ho, and Yb, in the post-Archaean flux. A better solution therefore is that there is a significant source of Y, Ho, and Yb in the Archaean continental crust. This is most likely in the phase garnet and most probably located in lower crustal garnet granulites. Thus a model Archaean TTG crust in which a significant fraction of the lower crust is a mafic garnet-granulite (model 2) provides a more satisfactory solution. [38] However, if the Archaean crust was more mafic, as model 2 implies, the bulk composition of the calculated post-Archaean flux (SiO2 = 55– 57 wt%, mg # = 58) places it in the compositional

range of high mg # andesites (Figure 3b). Kelemen et al. [2003a] have argued that this composition is derived in a very specific manner and is the product of eclogite melting in a subducting slab which is enriched in Mg through interaction with the mantle during ascent. [39] There are three implications of this observation. First, the post-Archaean flux into the continental crust must contain a slab melt component, in addition to the ‘‘standard’’ arc component and the OIB component. If this is correct this component of the post-Archaean flux might also contain the excess subduction fluid-mobile element contribution. Second, the difficulty of balancing the elements Y, Ho, and Yb in model 1 and the solution proposed in model 2 suggests that the removal of garnetiferous mafic rocks, rich in Y, Ho, and Yb, through the process of lower crustal delamination is unlikely to be a major process in the formation of continental crust since the Archaean. Third, the nature and origin of Archaean lower crustal, mafic, garnet granulites should be reinvestigated, since their contribution to the bulk Archaean crust appears to be significant.

4.4. Genesis of Continental Crust Over Geological Time [40] A central feature of this model is that Archaean continental crust was silica-rich, an observation that is based upon the abundance of TTG magmas in Archaean cratons. Experimental studies show that TTG magmas are produced by the partial melting of a hydrated basaltic parent at sub-Moho depths [Wyllie et al., 1997; Rapp, 1997]. Their juvenile isotopic character, as evidence by the coincidence of Nd-depleted mantle model ages and isochron ages [e.g., Whitehouse, 1989] and the coincidence of Nd-depleted mantle model ages and U-Pb zircon crystallization ages [e.g., Jelsma et al., 1996], indicates that their basaltic source was produced by mantle melting only a short time before TTG production. The tectonic setting of TTG generation has been much disputed. However, a recent reevaluation of the probable melting regime in which they were produced suggests that a subduction-related, slab-melting model has much to commend it [Moyen and Stevens, 2005]. Further, middle to late Archaean TTGs have very similar compositions to high-silica adakites, produced by slab melting in modern subduction environments [Martin et al., 2005], although alternative mechanisms for producing such melts within a subduction environment have also been proposed [Dreher 10 of 14


3

G


et al., 2005; Macpherson et al., 2006]. An alternative to the subduction model for TTG generation is that they were formed by the partial melting of mafic rocks in the lower crust [Smithies et al., 2005; Whalen et al., 2002]. However, this is thought to be unlikely to be the principal mechanism of crust generation in the Archaean, for trondhjemites produced by the partial melting of Phanerozoic mafic, lower-oceanic-crust, probably a realistic scenario for much of the Archaean, produce melts which are very different in trace element composition from typical Archaean TTGs [Rollinson, 2008]. [41] The results of this model have some important implications for the processes of crustal evolution both in the Archaean and in post-Archaean times. First, it is evident that Archaean continental crust has another, minor, component in addition to its TTG and greenstone belt components. This is the mafic component which makes up the Archaean lower continental crust. The details of these lithologies need to be reinvestigated. Their enrichment in Y, Ho, and Yb content requires either that the lower crust experienced melt extraction, with these elements sequestered and concentrated in residual garnet, or that the garnet in the lower crust is a cumulate from magma crystallization. Geochemical evidence indicates that these mafic rocks are unrelated to the TTG rocks of the lower crust [Weaver and Tarney, 1980] and implies the addition of a separate, mantle-derived component to the lower crust. A further consequence of this model is that the Archaean TTG component of the continental crust has a relatively low mg # (38–45), relative to primary mantle materials, and thus contributes to the low mg # of the continental crust as a whole. This implies that the TTG magmas did not equilibrate with the mantle wedge as they rose to the surface. Third, the post Archaean flux into the continents has long been assumed to be predominantly basaltic and is thought to be an arcplume mix [Kemp and Hawkesworth, 2003]. A finding of this study, however, is that this postArchaean flux may also contain a slab melt component. Thus the post-Archaean flux is a three component mix of arc basalt, plume basalt, and slab melt. [42] These observations imply that the fluxes into both Archaean and post-Archaean continental crust are similar. The two components are basalt and a slab melt. In the Archaean the dominant flux was from a slab melt and the minor flux was basalt, whereas in the post Archaean the proportions were

reversed. It is this variability which accounts for the secular change in the composition of the continental crust.

5. Conclusions [43] The mass balance calculation presented here shows that for a significant number of trace elements, the composition of the bulk continental crust equates to an approximate 50:50 mix of Archaean TTG crust and post-Archaean basaltic or basaltic andesite contribution. This observation supports the hypothesis that the mantle to continental crust flux has changed with time from being predominantly felsic (TTG) in the Archaean to mafic in the post-Archaean. Hence the andesitic bulk composition of the continental crust is a time-integrated average of fluxes from mantle to crust which have changed over time. In addition, [44] 1. Trace element data for Sr indicate that the bulk crust composition has been also modified by erosion. [45] 2. Trace element data for the subduction fluidmobile elements Rb, Ba, Th, U, and K, indicate that the bulk crust composition has an excess of these elements, implying an addition source. This may be through by the addition of enriched mafic melts to the lower and middle crust or through addition of a slab melt to the crust. [46] 3. Trace element data for Y, Ho, and Yb require a mafic Archaean lower crust which in turn implies a high mg # andesite flux into the crust. High mg # andesites tend to be the product of slab melting. [47] 4. Thus the post-Archaean flux into the continental crust is dominantly basaltic with a small slab melt component, whereas in the Archaean the dominant flux was a slab melt with a small basaltic component. [48] 5. This study supports the view of O’Nions and McKenzie [1988] and Hawkesworth and Kemp [2006b] that the residence times of elements within the continental crust are highly variable.

Acknowledgments [49] Richard Arculus, Matthias Barth, Ken Collerson, Chris Hawkesworth, Peter Kelemen, Tony Kemp, and Jan Kramers are all thanked for their helpful comments on the several versions of this manuscript.

11 of 14


3

G


References Albarede, F. (1998), The growth of the continental crust, Tectonophysics, 296, 1 – 14, doi:10.1016/S0040-1951 (98)00133-4. Armstrong, R. L. (1981), Radiogenic isotopes: The case for crustal recycling on a near steady state, no-continental-growth model, Philos. Trans. R. Soc. London, Ser. B, 301, 443 – 472, doi:10.1098/rsta.1981.0122. Barker, F., and J. G. Arth (1976), Generation of trondhjemitictonalitic liquids and Archaean bimodal trondhjemite-basalt suites, Geology, 4, 596 – 600, doi:10.1130/0091-7613 (1976)42.0.CO;2. Barth, M. G., W. F. McDonough, and R. L. Rudnick (2000), Tracking the budget of Nb and Ta in the continental crust, Chem. Geol., 165, 197 – 213, doi:10.1016/S0009-2541 (99)00173-4. Behn, M. D., and P. B. Kelemen (2003), Relationship between seismic P wave velocity and the composition of anhydrous igneous and meta-igneous rocks, Geochem. Geophys. Geosyst., 4(5), 1041, doi:10.1029/2002GC000393. Boyd, O. S., C. H. Jones, and A. F. Sheehan (2004), Foundering lithosphere imaged beneath the southern Sierra Nevada, California, USA, Science, 305, 660 – 662, doi:10.1126/ science.1099181. Campbell, I. H., and R. W. Griffiths (1992), The changing nature of mantle hotspots through time: Implications for the chemical evolution of the Mantle, J. Geol., 92, 497 – 523. Chakrabarti, R., A. R. Basu, and D. K. Paul (2007), Nd-Hf-SrPb isotopes and trace element chemistry of Proterozoic lamproites from southern India: Subducted komatiite in the source, Chem. Geol., 236, 291 – 302, doi:10.1016/j.chemgeo. 2006.10.006. Collerson, K. D., and B. Kamber (1999), Evolution of the continents and the atmosphere inferred from Th-U-Nb systematics of the depleted mantle, Science, 283, 1519 – 1522, doi:10.1126/science.283.5407.1519. Condie, K. C. (1993), Chemical composition and evolution of the upper continental crust: Contrasting results from surface samples and shales, Chem. Geol., 104, 1 – 37, doi:10.1016/ 0009-2541(93)90140-E. Condie, K. C. (2000), Episodic continental growth models: Afterthoughts and extensions, Tectonophysics, 322, 153 – 162, doi:10.1016/S0040-1951(00)00061-5. Condie, K. C. (2005), High field strength element ratios in Archaean basalts: A window into evolving sources of mantle plumes, Lithos, 79, 491 – 504, doi:10.1016/j.lithos.2004. 09.014. Davidson, J. P., and R. J. Arculus (2006), The significance of Phanerozoic arc magmatism in generating continental crust, in Evolution and Differentiation of the Continental Crust, edited by M. Brown and T. Rushmer, pp. 135 – 172, Cambridge Univ. Press, New York. DePaolo, D. J. (1988), Age dependence of the composition of the continental crust: Evidence from Nd isotopic variations in granitic rocks, Earth Planet. Sci. Lett., 90, 263 – 271, doi:10.1016/0012-821X(88)90130-6. De Wit, M. J., and L. D. Ashwal (1997), Greenstone Belts, 809 pp., Clarendon Press, Oxford, U. K. Dreher, S. T., C. G. Macpherson, D. G. Peason, and J. P. Davidson (2005), Re-Os isotope studies of Mindanao adakites: Implications for sources of metal and melts, Geology, 33, 957 – 960, doi:10.1130/G21755.1. Ellam, R. M., and C. J. Hawkesworth (1988), Is average continental crust generated at subduction zones?, Geology, 16,

314 – 317, doi:10.1130/0091-7613(1988)0162.3.CO;2. Feng, R., and R. Kerrich (1992), Geochemical evolution of granitoids from the Archaean Abitibi Southern Volcanic Zone and the Pontiac subprovince, Superior Province Canada: Implications for tectonic history and source regions, Chem. Geol., 98, 23 – 70, doi:10.1016/0009-2541(92)90090-R. Gao, S., et al. (2004), Recycling lower continental crust in the North China Craton, Nature, 432, 892 – 897, doi:10.1038/ nature03162. Garrido, C. J., et al. (2006), Petrogenesis of mafic garnet granulite in the lower crust of the Kohistan paleo-arc complex (northern Pakistan): Implications for intra-crustal differentiation of island arcs and generation of continental crust, J. Petrol., 47, 1873 – 1914, doi:10.1093/petrology/egl030. Hawkesworth, C. J., and A. I. S. Kemp (2006a), The differentiation and rates of generation of the continental crust, Chem. Geol., 226, 134 – 143, doi:10.1016/j.chemgeo.2005.09.017. Hawkesworth, C. J., and A. I. S. Kemp (2006b), Evolution of the continental crust, Nature, 443, 811 – 817, doi:10.1038/ nature05191. Jaupart, C., and J.-C. Mareschal (2003), Constraints on crustal heat production from heat flow data, in Treatise on Geochemistry 3.02, edited by H. D. Holland and K. K. Turekian, pp. 65 – 84, Elsevier, New York. Jelsma, H. A., M. L. Vinyu, P. J. Valbracht, G. R. Davies, J. R. Wijbrans, and E. A. T. Verdurmen (1996), Constraints on Archaean crustal evolution of the Zimbabwe Craton: A U-Pb zircon, Sm-Nd and Pb-Pb whole rock isotope study, Contrib. Mineral. Petrol., 124, 55 – 70. Kamber, B. S., A. Ewart, K. D. Collerson, M. C. Bruce, and G. D. McDonald (2002), Fluid-mobile trace element constraints on the role of slab melting and implications for Archaean crustal growth models, Contrib. Mineral. Petrol., 144, 38 – 56. Kelemen, P. B., K. Hanghoj, and A. R. Greene (2003a), One view of the geochemistry of subduction-related magmatic arcs, with an emphasis on primitive andesite and lower crust, in Treatise on Geochemistry 3.18, edited by H. D. Holland and K. K. Turekian, pp. 593 – 659, Elsevier, New York. Kelemen, P. B., G. M. Yogodzinski, and D. W. Scholl (2003b), Along-strike variation in the Aleutian island arc: Genesis of high-Mg# andesite and implications for continental crust, in Inside the Subduction Factory, Geophys. Monogr. Ser., vol. 138, edited by J. Eiler, pp. 223 – 276, AGU, Washington, D. C. Kemp, A. I. S., and C. J. Hawkesworth (2003), Granitic perspectives on the generation and secular evolution of the continental crust, in Treatise on Geochemistry 3.11, edited by J. Eiler, pp. 349 – 410, Elsevier, New York. Kleinhanns, I. C., J. D. Kramers, and B. S. Kamber (2003), Importance of water for Archaean granitoids petrology: A comparative study of TTG and potassic granitoids from Barberton Mountain Land, South Africa, Contrib. Mineral. Petrol., 145, 377 – 389, doi:10.1007/s00410-003-0459-9. Kramers, J. D. (2007), Hierarchical Earth accretion and the Hadean Eon, J. Geol. Soc., 164, 3 – 17, doi:10.1144/001676492006-028. Leat, P., T. R. Riley, C. D. Wareham, I. L. Millar, S. P. Kelley, and B. C. Storey (2002), Tectonic setting of primitive magmas in volcanic arcs: An example from the Antarctic Peninsula, J. Geol. Soc., 159, 31 – 44, doi:10.1144/0016-764900132. Lee, C.-T. A., D. M. Morton, R. W. Kistler, and A. K. Baird (2007), Petrology and tectonics of Phanerozoic continent formation: From island arcs to accretion and continental 12 of 14


3

G


arc magmatism, Earth Planet. Sci. Lett., 263, 370 – 387, doi:10.1016/j.epsl.2007.09.025. Macpherson, C. G., S. T. Dreher, and M. F. Thirlwall (2006), Adakites without slab melting: High pressure differentiation of island arc magma, Mindanao, the Philippines, Earth Planet. Sci. Lett., 243, 581 – 593, doi:10.1016/j.epsl.2005.12.034. Martin, H. (1986), Effects of steeper Archaean geothermal gradient on geochemistry of subduction-zone magmas, Geology, 14, 753 – 756, doi:10.1130/0091-7613(1986) 142.0.CO;2. Martin, H., and J.-F. Moyen (2002), Secular changes in tonalite-trondhjemite-granodiorite composition as markers of the progressive cooling earth, Geology, 30, 319 – 322, doi:10.1130/0091-7613(2002)0302.0. CO;2. Martin, H., R. H. Smithies, R. Rapp, J.-F. Moyen, and D. Champion (2005), An overview of adakite, tonalitetrondhjemite-granodiorite (TTG) and sanukitoid relationships and some implications for crustal evolution, Lithos, 79, 1 – 24, doi:10.1016/j.lithos.2004.04.048. McCulloch, M. T., and J. A. Gamble (1991), Geochemical and geodynamical constraints on subduction zone magmatism, Earth Planet. Sci. Lett., 102, 358 – 374, doi:10.1016/0012821X(91)90029-H. McLennan, S. M., and S. Hemming (1992), Samarium/neodymium elemental and isotopic systematics in sedimentary rocks, Geochim. Cosmochim. Acta, 56, 887 – 898, doi:10.1016/0016-7037(92)90034-G. Middlemost, E. A. K. (1994), Naming materials in the magma/ igneous rocks system, Earth Sci. Rev., 37, 215 – 224, doi:10.1016/0012-8252(94)90029-9. Moorbath, S., H. Welke, and N. H. Gale (1969), The significance of lead isotope studies in ancient high-grade metamorphic basement complexes, as exemplified by the Lewisian rocks of northwest Scotland, Earth Planet. Sci. Lett., 6, 245 – 256, doi:10.1016/0012-821X(69)90164-2. Moyen, J.-F., and G. Stevens (2005), Experimental constraints on TTG petrogenesis: Implications for Archaean geodynamics, in Archaean Geodynamics and Environments, Geophys. Monogr. Ser., vol. 164, edited by K. Benn, J.-C. Mareschal, and K. Condie, pp. 149 – 175, AGU, Washington, D. C. Nagler, T. F., and J. D. Kramers (1998), Nd isotopic evolution of the upper mantle during the Precambrian: Models, data and the uncertainty of both, Precambrian Res., 91, 233 – 252, doi:10.1016/S0301-9268(98)00051-5. Niu, F., and D. E. James (2002), Fine structure of the lowermost crust beneath the Kaapvaal craton and its implications for crustal formation and evolution, Earth Planet. Sci. Lett., 200, 121 – 130, doi:10.1016/S0012-821X(02)00584-8. O’Nions, R. K., and D. P. McKenzie (1988), Melting and continent generation, Earth Planet. Sci. Lett., 90, 449 – 456, doi:10.1016/0012-821X(88)90142-2. Parman, S. W., N. Schimizu, T. L. Grove, and J. C. Dann (2003), Constraints on the pre-metamorphic trace element composition of Barberton komatiites from ion probe analyses of preserved clinopyroxenes, Contrib. Mineral. Petrol., 144, 383 – 396. Plank, T. (2005), Constraints from Th/La on sediment recycling at subduction zones and the evolution of the continents, J. Petrol., 46, 921 – 944, doi:10.1093/petrology/egi005. Plank, T., and C. H. Langmuir (1998), The chemical composition of subducting sediment and its consequences for the crust and mantle, Chem. Geol., 145, 325 – 394, doi:10.1016/S0009-2541(97)00150-2.

Polat, A., and A. Hofmann (2003), Alteration and geochemical patterns in the 3.7 – 3.8 Ga Isua greenstone belt, West Greenland, Precambrian Res., 126, 197 – 218, doi:10.1016/S03019268(03)00095-0. Pride, C., and G. K. Mucke (1980), Rare earth element geochemistry of the scourian complex, NW Scotland—Evidence for the granite granulite link, Contrib. Mineral. Petrol., 73, 403 – 412, doi:10.1007/BF00376632. Ranganai, R. T., K. A. Whaler, and C. J. Ebinger (2008), Gravity anomaly patterns in the south-central Zimbabwe Archaean craton and their geological interpretation, J. Afr. Earth Sci., 51, 257 – 276, doi:10.1016/j.jafrearsci.2008.01.011. Rapp, R. P. (1997), Heterogeneous source regions for Archaean granitoids: Experimental and geochemical evidence, in Greenstone Belts, edited by M. De Wit and L. D. Ashwal, pp. 256 – 266, Oxford Univ. Press, New York. Richter, F. M. (1988), A major change in the thermal state of the earth at the Archaean-Proterozoic boundary: Consequences for the nature and preservation of continental lithosphere, in Oceanic and Continental Lithospheres: Similarities and Differences, edited by M. A. Menzies and K. G. Cox, pp. 39 – 52, Oxford Univ. Press, Oxford, U. K. Rollinson, H. R. (1999), Petrology and geochemistry of metamorphosed komatiites and basalts from the Sula Mountains greenstone belt, Sierra Leone, Contrib. Mineral. Petrol., 134, 86 – 101, doi:10.1007/s004100050470. Rollinson, H. R. (2007), Early Earth Systems: A Geochemical Approach, 285 pp., Blackwell, Malden, Mass. Rollinson, H. R. (2008), Ophiolitic trondhjemites: An analogue for the formation of Hadean felsic ‘crust’, Terra Nova, 20, 364 – 369, doi:10.1111/j.1365-3121.2008.00829.x. Rollinson, H. R., and J. Tarney (2005), Adakites – the key to understanding LILE depletion in granulites, Lithos, 79, 61 – 81, doi:10.1016/j.lithos.2004.04.050. Rollinson, H. R., and B. F. Windley (1980), Geochemistry and origin of an Archaean granulite grade tonalite-trondhjemitegranite suite from Scourie, N.W. Scotland, Contrib. Mineral. Petrol., 72, 265 – 281, doi:10.1007/BF00376145. Rudnick, R. L. (1995), Making continental crust, Nature, 378, 571 – 578, doi:10.1038/378571a0. Rudnick, R. L., and S. Gao (2003), Composition of the continental crust, in Treatise on Geochemistry 3.01, edited by H. D. Holland and K. K. Turekian, pp. 1 – 63, Elsevier, New York. Rudnick, R. L., and D. M. Fountain (1995), Nature and composition of the continental crust: A lower crustal perspective, Rev. Geophys., 33, 267 – 309, doi:10.1029/95RG01302. Rudnick, R. L., and T. Presper (1990), Geochemistry of intermediate to high-pressure granulites, in Granulites and Crustal Evolution, edited by D. Vielzeufand and P. Vidal, pp. 523 – 550, Kluwer, New York. Rudnick, R. L., and S. R. Taylor (1987), The composition and petrogenesis of the lower crust: A xenolith study, J. Geophys. Res., 92, 13,981 – 14,005. Shields, G., and J. Veizer (2002), Precambrian marine carbonate isotope database: Version 1.1, Geochem. Geophys. Geosyst., 3(6), 1031, doi:10.1029/2001GC000266. Shirey, S. B., B. S. Kamber, M. J. Whitehouse, P. A. Mueller, and A. R. Basu (2008), A review of the isotopic and trace element evidence for mantle and crustal processes in the Hadean and Archaean: Implications for the onset of plate tectonic subduction, Geol. Soc. Am. Spec. Pap., 404, 1 – 29. Smithies, R. H., M. J. Van Kranendonk, and D. C. Champion (2005), It started with a plume – early Archaean basaltic protocontinental crust, Earth Planet. Sci. Lett., 238, 284 – 297, doi:10.1016/j.epsl.2005.07.023. 13 of 14


3

G


Stettler, E. H., J. H. de Beer, D. Eberle, J. Ludden, and M. Mareschal (1995), Geophysics and deep structures, in Greenstone Belts, edited by M. J. De Wit, and L. Ashwal, pp. 338 – 375, Oxford Univ. Press, Oxford, U. K. Sun, S.-S., and W. F. McDonough (1989), Chemical and isotopic systematics of oceanic basalts: Implications for mantle composition and processes, Geol. Soc. London Spec. Publ., 42, 313 – 345. Sylvester, P. J. (1994), Archaean granite plutons, in Archaean Crustal Evolution, edited by K. C. Condie, pp. 261 – 314, Elsevier, Amsterdam. Taylor, S. R., and S. M. McLennan (1985), The Continental Crust: Its Composition and Evolution, 312 pp., Blackwell, Oxford, U. K. Taylor, S. R., and S. M. McLennan (1986), The chemical composition of Archaean crust, in The Nature of the Lower Continental Crust, edited by J. B. Dawson et al., Geol. Soc. London Spec. Publ., 24, 173 – 178. Turekian, K. K. (1968), Oceans, Prentice-Hall, Upper Saddle River, N. J. Turner, S., C. J. Hawkesworth, N. Rogers, J. Bartlett, T. Worthington, J. Hergt, J. A. Pearce, and I. Smith (1997), 238 U-230Th disequilibria, magma petrogenesis and flux rates beneath the depleted Tonga-Kermadec island Arc, Geochim. Cosmochim. Acta, 61, 4855 – 4884, doi:10.1016/S00167037(97)00281-0.

Weaver, B. L., and J. Tarney (1980), Rare earth geochemistry of Lewisian granulite facie gneisses, northwest Scotland: Implications for the petrogenesis of the Archaean lower continental crust, Earth Planet. Sci. Lett., 51, 279 – 296, doi:10.1016/0012-821X(80)90211-3. Weaver, B. L., and J. Tarney (1981), Lewisian Gneiss geochemistry and Archaean crustal development models, Earth Planet. Sci. Lett., 55, 171 – 180, doi:10.1016/0012821X(81)90096-0. Whalen, J. B., J. A. Percival, V. J. McNicoll, and F. J. Longstaffe (2002), A mainly crustal origin for tonalitic granitoid rocks, Superior Province, Canada: Implications for late-Archaean tectonomagmatic processes, J. Petrol., 43, 1551 – 1570, doi:10.1093/petrology/43.8.1551. Whitehouse, M. J. (1989), Sm-Nd evidence for diachronous crustal accretion in the Lewisian complex of north-west Scotland, Tectonophysics, 161, 245 – 256, doi:10.1016/ 0040-1951(89)90157-1. Wilson, M. (1989), Igneous Petrogenesis: A Global Tectonic Approach, 466 pp., HarperCollins, New York. Wyllie, P. J., M. B. Wolf, and S. R. van der Laan (1997), Conditions for formation of tonalites and trondhjemites: Magmatic sources and products, in Greenstone Belts, edited by M. De Wit and L. D. Ashwal, pp. 256 – 266, Oxford Univ. Press, New York.

14 of 14