+ All Categories
Home > Documents > The Generation and Evolution of the Continental Crust

The Generation and Evolution of the Continental Crust

Date post: 27-Apr-2015
Category:
Upload: bobmackenzie
View: 167 times
Download: 0 times
Share this document with a friend
20
Journal of the Geological Society , London, Vol. 167, 2010, pp. 229–248. doi: 10.1144/0016-76492009-072. 229 Review The generation and evolution of the continental crust C. J. HAWKESWORTH 1,2 *, B. DHUIME 1 , A. B. PIETRANIK 3 , P. A. CAWOOD 4 , A. I. S. KEMP 5 & C. D. STOREY 1,6 1 Department of Earth Sciences, University of Bristol, Wills Memorial Building, Queens Road, Bristol BS8 1RJ, UK 2 Present address: School of Geography and Geosciences, University of St. Andrews, North Street, St. Andrews KY16 9AL, UK 3 Institute of Geological Sciences, University of Wroclaw, 50-205 Wroclaw, Poland 4 School of Earth and Environment, University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia 5 School of Earth and Environmental Sciences, James Cook University, Townsville, QLD 4811, Australia 6 School of Earth and Environmental Sciences, University of Portsmouth, Portsmouth PO1 3QL, UK *Corresponding author (e-mail: [email protected]) Abstract: The continental crust is the archive of the geological history of the Earth. Only 7% of the crust is older than 2.5 Ga, and yet significantly more crust was generated before 2.5 Ga than subsequently. Zircons offer robust records of the magmatic and crust-forming events preserved in the continental crust. They yield marked peaks of ages of crystallization and of crust formation. The latter might reflect periods of high rates of crust generation, and as such be due to magmatism associated with deep-seated mantle plumes. Alternatively the peaks are artefacts of preservation, they mark the times of supercontinent formation, and magmas generated in some tectonic settings may be preferentially preserved. There is increasing evidence that depletion of the upper mantle was in response to early planetary differentiation events. Arguments in favour of large volumes of continental crust before the end of the Archaean, and the thickness of felsic and mafic crust, therefore rely on thermal models for the progressively cooling Earth. They are consistent with recent estimates that the rates of crust generation and destruction along modern subduction zones are strikingly similar. The implication is that the present volume of continental crust was established 2–3 Ga ago. The continental crust constitutes some 40% of the surface area of the Earth, and yet it constitutes almost 70% of the total volume of the Earth’s crust. It is andesitic in composition, 25– 70 km thick, and less dense than the thinner (,10 km) oceanic crust of largely mafic composition. The differentiation of the crust of the Earth into these contrasting chemical–mechanical components in part reflects the horizontal movement of the lithosphere (crust and upper mantle) through plate tectonics. The contrasting density structure of continental and oceanic crust results in a pronounced bimodal elevation, a buoyant continental crust and an oceanic crust that is, except for very young rocks, gravitationally unstable (e.g. Cloos 1993) and sinks back into the asthenospheric mantle at subduction zones, resulting in no oceanic crust being older than 200 Ma. The continental crust is the geological archive of Earth history, it is different from analogues on nearby planets, and it influences global climate by being a sink for CO 2 (e.g. Garrels & Perry 1974; Zhang & Zindler 1993; Lowe & Tice 2004). It is andesitic in composition, and such magmas are not commonly in equili- brium with the upper mantle (e.g. Rudnick 1995; Walter 2003). Most models for the generation of new continental crust there- fore involve the generation of basalt and subsequent differentia- tion, by fractional crystallization and/or remelting, to higher silica compositions (Kuno 1968; Ellam & Hawkesworth 1988; Arndt & Goldstein 1989; Kay & Kay 1991; Rudnick 1995; Arculus 1999; Kemp & Hawkesworth 2003; Zandt et al. 2004; Plank 2005; Hawkesworth & Kemp 2006a,b) and return of residue or cumulate to the mantle. Differentiation of the continental crust primarily involves igneous processes, and an idealized crustal section consists of a lower part dominated by residue or cumulate and/or new mafic crust and an upper part composed mainly of rocks of granitic to granodioritic composi- tion. The residence times of elements in the upper crust appear to be much longer than those in the lower crust (Hawkesworth & Kemp 2006b). The longstanding questions are when the continental crust was generated, and how the processes involved in the generation of the continental crust have changed with time. This review is written at a time of considerable conceptual upheaval when old approaches are being questioned and new analytical techniques have recently come into play. It has long been argued that the upper mantle was depleted by the extraction of the continental crust, and so the two reservoirs are complementary (Jacobsen & Wasserburg 1979; O’Nions et al. 1980; Alle `gre et al. 1983). The implication was that the record of continental crust generation could then be broadly investigated from the evolution of the depleted upper mantle. If that depleted mantle reservoir is well mixed, its radiogenic isotope ratios should offer a more robust indication of the volumes of crust extracted than the remaining vestiges of old continental crust. Yet recent evidence suggests that the upper mantle was initially depleted by processes much older than the preserved continental crust (Carlson & Boyet
Transcript
Page 1: The Generation and Evolution of the Continental Crust

Journal of the Geological Society, London, Vol. 167, 2010, pp. 229–248. doi: 10.1144/0016-76492009-072.

229

Review

The generation and evolution of the continental crust

C. J. HAWKESWORTH 1,2* , B. DHUIME 1, A. B. PIETRANIK 3, P. A. CAWOOD4, A. I . S . KEMP5

& C. D. STOREY 1,6

1Department of Earth Sciences, University of Bristol, Wills Memorial Building, Queens Road, Bristol BS8 1RJ, UK2Present address: School of Geography and Geosciences, University of St. Andrews, North Street,

St. Andrews KY16 9AL, UK3Institute of Geological Sciences, University of Wrocław, 50-205 Wrocław, Poland

4School of Earth and Environment, University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia5School of Earth and Environmental Sciences, James Cook University, Townsville, QLD 4811, Australia

6School of Earth and Environmental Sciences, University of Portsmouth, Portsmouth PO1 3QL, UK

*Corresponding author (e-mail: [email protected])

Abstract: The continental crust is the archive of the geological history of the Earth. Only 7% of the crust is

older than 2.5 Ga, and yet significantly more crust was generated before 2.5 Ga than subsequently. Zircons

offer robust records of the magmatic and crust-forming events preserved in the continental crust. They yield

marked peaks of ages of crystallization and of crust formation. The latter might reflect periods of high rates of

crust generation, and as such be due to magmatism associated with deep-seated mantle plumes. Alternatively

the peaks are artefacts of preservation, they mark the times of supercontinent formation, and magmas

generated in some tectonic settings may be preferentially preserved. There is increasing evidence that

depletion of the upper mantle was in response to early planetary differentiation events. Arguments in favour of

large volumes of continental crust before the end of the Archaean, and the thickness of felsic and mafic crust,

therefore rely on thermal models for the progressively cooling Earth. They are consistent with recent estimates

that the rates of crust generation and destruction along modern subduction zones are strikingly similar. The

implication is that the present volume of continental crust was established 2–3 Ga ago.

The continental crust constitutes some 40% of the surface area

of the Earth, and yet it constitutes almost 70% of the total

volume of the Earth’s crust. It is andesitic in composition, 25–

70 km thick, and less dense than the thinner (,10 km) oceanic

crust of largely mafic composition. The differentiation of the

crust of the Earth into these contrasting chemical–mechanical

components in part reflects the horizontal movement of the

lithosphere (crust and upper mantle) through plate tectonics. The

contrasting density structure of continental and oceanic crust

results in a pronounced bimodal elevation, a buoyant continental

crust and an oceanic crust that is, except for very young rocks,

gravitationally unstable (e.g. Cloos 1993) and sinks back into the

asthenospheric mantle at subduction zones, resulting in no

oceanic crust being older than 200 Ma.

The continental crust is the geological archive of Earth history,

it is different from analogues on nearby planets, and it influences

global climate by being a sink for CO2 (e.g. Garrels & Perry

1974; Zhang & Zindler 1993; Lowe & Tice 2004). It is andesitic

in composition, and such magmas are not commonly in equili-

brium with the upper mantle (e.g. Rudnick 1995; Walter 2003).

Most models for the generation of new continental crust there-

fore involve the generation of basalt and subsequent differentia-

tion, by fractional crystallization and/or remelting, to higher

silica compositions (Kuno 1968; Ellam & Hawkesworth 1988;

Arndt & Goldstein 1989; Kay & Kay 1991; Rudnick 1995;

Arculus 1999; Kemp & Hawkesworth 2003; Zandt et al. 2004;

Plank 2005; Hawkesworth & Kemp 2006a,b) and return of

residue or cumulate to the mantle. Differentiation of the

continental crust primarily involves igneous processes, and an

idealized crustal section consists of a lower part dominated by

residue or cumulate and/or new mafic crust and an upper part

composed mainly of rocks of granitic to granodioritic composi-

tion. The residence times of elements in the upper crust appear

to be much longer than those in the lower crust (Hawkesworth &

Kemp 2006b).

The longstanding questions are when the continental crust was

generated, and how the processes involved in the generation of

the continental crust have changed with time. This review is

written at a time of considerable conceptual upheaval when old

approaches are being questioned and new analytical techniques

have recently come into play. It has long been argued that the

upper mantle was depleted by the extraction of the continental

crust, and so the two reservoirs are complementary (Jacobsen &

Wasserburg 1979; O’Nions et al. 1980; Allegre et al. 1983). The

implication was that the record of continental crust generation

could then be broadly investigated from the evolution of the

depleted upper mantle. If that depleted mantle reservoir is well

mixed, its radiogenic isotope ratios should offer a more robust

indication of the volumes of crust extracted than the remaining

vestiges of old continental crust. Yet recent evidence suggests

that the upper mantle was initially depleted by processes much

older than the preserved continental crust (Carlson & Boyet

Page 2: The Generation and Evolution of the Continental Crust

2008; Tolstikhin & Kramers 2008). Similarly, it has been widely

assumed that the geological record provides a representative

record of the evolution of the continental crust, and yet now

there is increasing evidence that it may be biased by the tectonic

settings in which the rocks were generated (Hawkesworth et al.

2009).

In this review we explore current thinking on when and how

new continental crust was generated, and the extent to which

there were peaks of crust generation and variations in the rates at

which new crust was generated. We consider the constraints on

the composition of early continental crust in the Hadean (older

than 4.0 Ga), and links between granite magmatism, crustal

growth and high-grade metamorphic events. It is timely also to

review the approaches to these longstanding issues.

Old problems and new ways forward

The oldest known rocks on Earth are the 4 Ga Acasta Gneisses

from the Slave Province in NW Canada (Bowring & Williams

1999), and the oldest minerals are detrital zircons that are up to

4.4 Ga old from the Jack Hills area in Western Australia (Wilde

et al. 2001). The Jack Hills zircons, and their diverse inclusions

of minerals such as quartz, feldspar, muscovite and monazite,

suggest that geochemically evolved rocks, and hence by implica-

tion continental crust, were present from at least 4.4 Ga.

However, the volume of continental crust cannot be assessed

from the small number of isolated records that are preserved.

Figure 1 summarizes some of the data and interpretations that

have been a focus for many discussions of the generation and

evolution of the continental crust in the last 20 years. A number

of models sought to describe the increase in the volume of

continental crust through time. Many of the models are based on

radiogenic isotopes; for example, the Nd isotope ratios of shales

(McCulloch & Wasserburg 1978; O’Nions et al. 1983; Allegre &

Rousseau 1984). Strictly speaking, they therefore describe the

increase in the volume of continental crust that survived for long

enough for the isotope ratios to change in response to radioactive

decay, which is of the order of several hundred million years. As

pointed out by Armstrong (1981), crust that is generated one day

and destroyed soon thereafter will not be visible to records based

on the radiogenic isotope ratios of crustal rocks. Formally,

therefore, the curves, which are often described as showing

increases in the volume of continental crust through time, in

practice describe the volume of (relatively) stable continental

crust. In general these curves are smooth (Fig. 1), and they

contrast with the spiky distribution of the ages of igneous rocks

derived from the mantle that therefore represent the generation

of new continental crust (McCulloch & Bennett 1994; Condie

1998, 2005).

The details of the igneous record of new crustal growth remain

difficult to establish, and this plot is included because, albeit in

slightly different versions, it has been a focus of discussion for

the last 20 years. The distinctive feature of the igneous record is

that there are two pronounced age peaks at 2.7 and 1.9 Ga, with

some compilations suggesting a third at 1.2 Ga, and then none in

the last 1 Ga. One interpretation is that the age peaks reflect

periods of exceptional rates of crustal growth, that such peaks

are difficult to ascribe to processes linked to plate tectonics, and

they may therefore be evidence for periods of crustal growth

linked to the emplacement of superplumes (e.g. Condie 2004).

The lack of conspicuous age peaks in the last billion years might

in turn reflect the period in which the generation of new

continental crust was largely dominated by plate-tectonic pro-

cesses. The underlying question is the extent to which these age

distributions are a true record of the generation and evolution of

the continental crust, or whether they are artefacts of preserva-

tion in the geological record (e.g. Gurnis & Davies 1985, 1986;

Hawkesworth et al. 2009).

One of the marked developments in recent years is our ability

to measure isotope and trace elements in situ by ion microprobe

and laser ablation inductively coupled plasma mass spectromery

(LA-ICP-MS). It is now much easier to characterize what is

being analysed, and there have been exciting new developments

in topics as diverse as the generation of granites, the records

preserved in detrital and inherited zircons, and the dating of

high-grade metamorphic processes. Recent studies of crustal

evolution have relied on zircons, which typically yield high-

precision U–Pb ages and so have been the cornerstone of most

attempts to develop an accurate geological time scale. Zircons

can be analysed in situ for U–Pb, Hf and O isotopes and trace

elements. They are resilient under most crustal conditions and

they survive metamorphism and prolonged sedimentary recy-

cling. Thus they arguably provide more representative records,

particularly from older geological terrains, than the rock record.

However, most zircons crystallize from medium- to high-silica

magmas and so they more readily offer insights into the

generation and evolution of more evolved compositions in the

crust than into the processes of crust generation and the

evolution of the depleted upper mantle.

Terms

The literature on the generation and evolution of the continental

crust and the mantle contains a number of terms that are widely,

but perhaps not consistently used. It is therefore helpful to start

with definitions of some of these terms as they can mean

different things to different people, which may in turn lead to

fundamentally different implications. Following Rudnick & Gao

Fig. 1. A histogram of the volume distribution of juvenile continental

crust based on a compilation of U–Pb zircon ages integrated with Nd

isotope ratios and lithological associations. This is compared with models

of continental growth with 100% representing the present-day cumulative

volume of crust (adapted from Condie 2005; after Cawood et al. 2009).

Early models, such as that by Hurley & Rand (1969), were based on the

geographical distribution of Rb–Sr and K–Ar isotope ages, and these are

likely to have been reset by younger orogenic events. Some models

suggested rapid early growth of continental crust, and a slower rate of

growth or even a decrease in continental volume through time (Fyfe

1978; Reymer & Schubert 1984; Armstrong 1991). Other models require

a more linear growth (unlabelled) or rapid growth during the late

Archaean followed by steady-state growth driven by island arc

magmatism (Taylor & McLennan 1985).

C. J. HAWKESWORTH ET AL .230

Page 3: The Generation and Evolution of the Continental Crust

(2003), the continental crust is taken to extend vertically from

the Earth’s surface to the Mohorovicic discontinuity and laterally

to the break in slope in the continental shelf. It is the layer of

granitic, sedimentary and metamorphic rocks that form the

continents and the areas of shallow seabed close to their shores,

the continental shelves. It is linked to the upper mantle and

together these form the lithosphere, the rigid, highly viscous lid

of the Earth that is divisible into a series of plates. The

lithospheric plates range in thickness from less than 100 km for

oceanic lithosphere, and probably as low as 10 km close to mid-

ocean ridges, to up to 200–300 km under old Archaean nuclei.

The plates move both with respect to each other along plate

boundaries, and with respect to the underlying mantle astheno-

sphere from which they are separated by the low-velocity zone.

We are here concerned with the continental crust as a reservoir

of rocks of relatively differentiated compositions at the surface

of the Earth, and the volume that may be present at different

times. We are interested in generic models for how the average

‘andesitic’ composition of the continental crust may have been

generated, and less with the details of how it may have been

generated in different ways at different localities. New continen-

tal crust is generated from the mantle, but its average composi-

tion would not be in equilibrium with typical mantle

compositions. Thus it is likely to have been generated in more

than one stage as, for example, in the generation, crystallization

and remelting of basalt.

The term crust generation is therefore used to denote the

formation of new continental crust in a generic sense; that is,

the emplacement of new magma directly from the mantle. The

growth of continental crust is then formally the increase in its

volume through time. This necessarily takes account both of the

volumes of new crust generated and the amounts destroyed by

erosion and returned to the mantle. In practice, growth of

continental crust is difficult to tie down, because radiogenic

isotopes constrain only the volume of crust that has been stable

for long enough for significant differences in isotope ratios to

be developed from radioactive decay. However, even short-lived

crust may leave a legacy in the complementary depletion of the

upper mantle. The assembly of continental crust from different

segments that were generated elsewhere and juxtaposed tectoni-

cally increases the volume of continental crust in the region

being considered, but not the volume of continental crust

overall, in the sense that the assembled fragments were already

present elsewhere. The proportion of continental crust is

inversely proportional to oceanic and transitional crust on a

constant radius Earth. Thus as the volume of continental crust

has grown (assuming it has) then the proportion of oceanic

crust has decreased.

In this review crustal recycling is taken to mean the recycling

of continental crust back through the mantle and into the crust

again. Crustal reworking is the remobilization of pre-existing

crust by partial melting and/or erosion and sedimentation, but all

at sites within the continental crust.

The timing of events: crystallization and crustformation ages

Accurate and precise ages underpin our understanding of the

generation and the evolution of the continental crust. In practice,

two types of ages are involved. The first is the geological age of

the material being analysed, which can be the age of deposition

of a sediment or the crystallization of a single mineral or an

igneous rock. Zircons are widely used because they yield high-

precision U–Pb ages, and these ages may be determined by

dissolving the whole crystal, and by in situ isotope measurements

using secondary ionization mass spectrometry (SIMS) or laser

ablation ICP-MS. Zircons tend to yield relatively reliable isotope

results, and they are arguably easier to interpret than whole-rock

data, which may be more readily disturbed by metamorphic and

low-temperature alteration events. The advantage of in situ

measurements is that the material being analysed can be well

characterized by microbeam imaging, in contrast to whole-rock

analyses (Fig. 2). Moreover, small portions, 10–50 �m across,

can be dated and so the history of more complicated grains can

be unravelled.

The second types of ages are those that constrain when the

crustal source of igneous or sedimentary rock was itself derived

from the mantle. These are critical to this discussion because

they seek to constrain when new crust was generated, they are

termed model ages, and they are widely used for Hf and Nd

isotopes. The key assumption of these model ages is that the

ratio of the parent to the daughter isotope (Lu/Hf and Sm/Nd)

changes in the processes involved in the generation of new crust,

and then it is not fractionated further by processes of remelting,

erosion and sedimentation within the continental crust. Thus the

parent/daughter ratio of the crustal source material can be

estimated, and used to calculate the time when that portion of

new crust was initially extracted from the mantle. These ages are

known as model Nd and Hf ages, and they are best illustrated on

isotope evolution diagrams of isotope ratios against time (Fig.

3). In many cases the Nd and Hf isotope ratios of sediments will

be hybrid values averaging the contributions of material from

different source rocks. They are less likely to reflect discrete

crust-forming events. Thus it is helpful to distinguish model age

that are derived from sedimentary sources from those derived

from igneous sources, and this can be done using O isotopes.

The 18O/16O ratio, expressed as �18O, is readily changed only

by low-temperature and surficial processes, and so the �18O of

mantle-derived magmas (5.3 � 0.3‰; Valley 2003) are different

from those of rocks that have experienced a sedimentary cycle or

low-temperature hydrothermal alteration on the sea floor, which

Fig. 2. (a) Cenozoic granulite-facies gneiss from the Hidaka

Metamorphic belt, northern Japan. The coin is about 2 cm in diamter.

(b) A cathodoluminesence image from a zircon recording crystallization

events at 54 and 19 Ma (Kemp et al. 2007b). Leucosomes have

developed with a consistent orientation, typically containing garnet.

Granulite generation involved the reworking of older protoliths through

enhanced heat flux associated with subduction zone retreat.

THE CONTINENTAL CRUST 231

Page 4: The Generation and Evolution of the Continental Crust

have elevated �18O (c. 7–25‰; Eiler 2001, and references

therein). Magmas that contain a contribution from sedimentary

rocks should therefore have elevated �18O values, and these in

turn will be preserved in zircons that crystallized from such

magmas. Thus, high �18O values in zircon are a ‘fingerprint’ for

a supracrustal component in the generation of felsic igneous

rocks, with the implication that such zircons are likely to yield

hybrid model ages. Empirical studies have established that

oxygen diffusion in zircon is sufficiently sluggish that the

original igneous �18O value remains intact, even through pro-

tracted metamorphism and crustal fusion (King et al. 1998; Peck

et al. 2003). O isotopes can be measured in situ in zircon with

excellent precision (,0.5‰) by large radius ion microprobes that

have multi-collector capability (Valley 2003), and so it is

possible to determine U–Pb, Hf and O isotope and trace element

data on the same grain of zircon (or part there of), all to high

precision, and using in situ techniques.

Variations in Sm/Nd and Lu/Hf ratios and thecomposition of initial continental crust

Key points of discussion are the ages of crust generation, as

inferred from model ages, and the composition of that initial

crust. The calculation of model ages presumes that the parent/

daughter ratios are constant in the dominant crustal lithologies.

For Nd isotopes, it is generally accepted that the Sm/Nd ratios of

different crustal rocks are similar (but see below), and so model

Nd ages are routinely calculated using the measured Sm/Nd ratio

of the sample analysed. In contrast, zircons are widely analysed

for Hf isotopes because they have extremely low Lu/Hf ratios,

and so their present-day Hf isotope ratios are similar to those

when they crystallized. It follows that the Lu/Hf ratio of zircon

is very different from that of the likely source rocks of their host

magma in the continental crust, and so Hf isotopes in zircon

model ages are typically calculated using the average Lu/Hf ratio

of the continental crust (e.g. Griffin et al. 2002). A characteristic

of the Hf isotope–time plot is that the slopes of straight lines

depend on their parent–daughter Lu/Hf ratios. It follows that

straight isotope evolution lines defined by analyses of zircons of

different ages, but from magmas from similar source rocks, have

slopes that reflect the Lu/Hf ratios of the crustal source rocks. In

detail the Lu/Hf ratios of igneous rocks decrease with increasing

silica and trace element indices of differentiation such as Rb/Sr

(see Fig. 4, and the associated references); so if the slope of the

Hf isotope evolution line for the crustal source rocks can be

determined it should be possible to evaluate whether those

crustal source rocks are mafic or granitic in composition. It is

helpful, therefore, to evaluate the range of Lu/Hf and Sm/Nd

ratios in common rock types.

Figure 4 and Table 1 summarize the median values of the

Sm/Nd and Lu/Hf ratios in basalts, granites and sedimentary

rocks compiled from a database of c. 12 000 analyses from

GEOROC (http://georoc.mpch-mainz.gwdg.de/georoc/). Overall

there is a broad correlation between the two ratios but the range

in the median values for Lu/Hf is almost twice that for Sm/Nd.

Because the half-life of 176Lu is a third that of 147Sm, the

parent/daughter ratio, and hence the bulk composition of early

formed crust, is better approached using Hf rather than Nd

isotopes. It follows that there is a greater range for Lu/Hf than

for Sm/Nd for each rock type (Table 1). Island arc basalts and

basaltic andesites, and mid-ocean ridge basalt (MORB), have

higher Sm/Nd and Lu/Hf ratios than the rocks of the continental

crust, and both ratios decrease from estimates of the average

lower continental crust, through the values for the bulk crust to

the upper crust. The abundances of insoluble elements in the

upper crust are estimated from those in continental sediments,

and so they both plot together in Figure 4. Granitic rocks and

the bulk crust have similar Sm/Nd and Lu/Hf ratios, as do flood

basalts and the lower crust. The average lower crustal composi-

tion is similar to estimates of model new continental crust; that

is, the mantle-derived magma from which the more evolved

compositions of the crust then differentiated (Hawkesworth &

Kemp 2006a). However, the similarity in Sm/Nd and Lu/Hf

ratios between flood basalts and the lower crust may be simply

coincidental, and it should not be taken to imply that both were

necessarily generated in similar settings. Current models attrib-

ute the trace element ratios of the continental crust to the

combination of processes involved in the generation of new

crust from the mantle, and to the subsequent foundering of

lower crustal material (Ellam & Hawkesworth 1988; Arndt &

Goldstein 1989; Kay & Kay 1991; Rudnick 1995; Arculus

1999; Kemp & Hawkesworth 2003; Zandt et al. 2004; Plank

2005; Hawkesworth & Kemp 2006a,b).

In seeking to interpret the variations in Lu/Hf and Sm/Nd

ratios it is helpful to evaluate how they vary with indices of

magma differentiation. For crustal systems, Rb/Sr ratios are more

sensitive indices of differentiation than bulk SiO2 contents, and

so Figure 4 summarizes Rb/Sr�SiO2, Rb/Sr–Sm/Nd, and Rb/Sr–

Lu/Hf for various rock units. A positive correlation is observed

in the Rb/Sr–SiO2 diagram and the Sm/Nd and Lu/Hf ratios

show a negative correlation with Rb/Sr. What is striking is the

shift in Lu/Hf and Sm/Nd from the arc basalts or basaltic

andesites into the granitic rocks and sediments. In detail, there-

fore, Sm/Nd and Lu/Hf both decrease with increasing differentia-

tion in the continental crust, and this may introduce additional

uncertainties into the calculations of model ages, especially in

strongly differentiated samples.

Zircons as archives of the continental crust

Zircons typically crystallize from granitic magmas with .60%

SiO2, although they are found in lower silica magmas in lesser

abundance. They are therefore a feature of magmatic rocks in the

Fig. 3. Model Hf ages. A plot illustrating the changes in Hf isotope

ratios with time in the bulk Earth, the depleted mantle and in average

continental crust. The present-day Hf isotope ratio of a zircon is similar

to that at the time it crystallized, because it has a very low Lu/Hf ratio.

Its model Hf age is calculated as the age at which the evolution of the

source of the magma from which the zircon crystallized intersects the

curve for the depleted mantle. The (crustal) source of the magma is

traditionally assumed to have the 176Lu/177Hf ratio of average continental

crust.

C . J. HAWKESWORTH ET AL .232

Page 5: The Generation and Evolution of the Continental Crust

upper continental crust, they yield high-precision (U–Th)–Pb

crystallization ages, and they survive erosion and sedimentation

to be preserved in continent-derived sedimentary rocks. Zircon

has the unique combination of physiochemical resilience and

high concentrations of important trace elements that include two

radiogenic isotope systems of geochronological importance

(namely, U–Pb, Th–Pb) and another (Lu–Hf) that is widely used

as a tracer of crustal evolution. It follows that zircons offer an

excellent record of the evolution of differentiated compositions

in the continental crust, and it may be more difficult to use

zircons to chart the generation of new crust and hence the

evolution of the depleted mantle, for example. Continent-derived

sediments, and in particular shales, have been widely used to

obtain representative average compositions of the upper crust

(e.g. Taylor & McLennan 1991; Condie 1993). This works well

for insoluble elements, the REE, Y, Sc, Th and Nb, and also for

Ti, Zr and Hf, but the abundances of the second group are

controlled by the distribution of heavy minerals. Zircon is one

such mineral, and so the question becomes how representative of

their source terrain, and hence of the upper crust in a particular

area, are zircons preserved in continental sediments. Zircons, for

example, are concentrated in more mature sands, and yet average

compositions, and Nd model ages, are estimated from shales.

Figure 5 summarizes U–Pb crystallization and Hf model ages

on zircons from Australia in both magmatic rocks, and as detrital

grains in sediments. Just over 17 000 zircons have been dated,

and there are Hf model ages on over 2000 of them, and so they

provide a good basis for exploring the information that is

available from the study of zircons. The striking observation is

that there are marked peaks in the ages of crystallization,

although the peak at 2.7 Ga partly reflects the intensive study of

economically important rocks of this age in Western Australia,

and peaks of similar ages are present in both the magmatic and

detrital zircons. It appears therefore that the detrital zircons offer

representative records of the magmatic history seen in the

exposed igneous provenance.

The peaks of crystallization ages primarily represent peaks

of magmatic activity, although we shall return to the issue of

whether the record may in some way be biased by the nature

of the geological record. It is not possible to say from the

crystallization ages whether these magmatic episodes involved

the generation of significant volumes of new crust or not, but

in principle this may be investigated by the determination of

Hf isotope ratios, and hence of Hf model ages. In practice,

two approaches have been used. First, the analysis of Hf and

O isotopes in magmatic zircons (i.e. those that crystallized at

the same time as their host rock) is used to unpick the relative

contributions of mantle and crustal sources in their petrogen-

esis (Kemp et al. 2007a). Such detailed studies constrain the

amounts of new crust generated in the different peaks of

magmatic activity identified from the crystallization ages of

the zircons.

Hf model ages are also measured in detrital and inherited

zircons. However, at present there is little sense of the kinds of

magma from which the zircons crystallized, apart from the

observation that they tend to have high silica contents. The

difficulty is that the magmas may contain mixed contributions

from the mantle and the crust at the time they crystallized, and

some magmas also contain contributions from sediments, which

are in themselves typically derived from mixed sources. Thus,

Fig. 4. Median Lu/Hf–SiO2, Lu/Hf–Sm/Nd, Rb/Sr–SiO2, Rb/Sr–Sm/Nd, and Rb/Sr–Lu/Hf in basalts and basaltic andesites [SiO2 , 55%, mid-ocean

ridge basalts (MORB): n ¼ 331; oceanic island basalts (OIB): n ¼ 1520; island arcs: n ¼ 644; continental flood basalts (CFB): n ¼ 2514], granitoids (SiO2

> 55%; n ¼ 7371) and continental sediments (n ¼ 63) (GEOROC compilation, http://georoc.mpch-mainz.gwdg.de/georoc/). Rb/Sr is used as an index of

differentiation. Primitive mantle (PUM) and CI Chondrites after Sun & McDonough (1989), continental crust from Rudnick & Gao (2003), cratonic shales

from Condie (1993) and GLOSS (global subducting sediment) from Plank & Langmuir (1998). The horizontal and vertical bars represent the deviation of

the median from the mean and mode values reported in Table 1.

THE CONTINENTAL CRUST 233

Page 6: The Generation and Evolution of the Continental Crust

the Hf isotope record in the detrital and inherited zircons is

likely to be dominated by processes of mixing within the crust,

and so it is extremely difficult to establish the ages of the initial

crust-forming events, except in cases where O isotopes are also

available (Kemp et al. 2006; Pietranik et al. 2008). One

consequence is that, taken together, the Hf model ages of the

zircons from Australia show much less well-defined peaks than

do the crystallization ages (Fig. 5). There are peaks of Hf model

ages at c. 4.4 Ga, from the Jack Hills locality in Western

Australia, and at c. 3.2 Ga, but they are much less marked

subsequently. The implication is that these zircons provide a

record that is dominated by crustal reworking, as is also seen for

Nd isotopes in shales (Allegre & Rousseau 1984).

In practice, sedimentary and igneous rocks offer different

records and perspectives on the evolution of the continental crust

(see also Fig. 1). Igneous rocks are for the most part generated in

magmatic provinces that are restricted in space and time, and the

generation of new crust clearly involves magmatic processes.

Sediments, in contrast, are mixtures of the material that was

present in potential source regions at the time of deposition.

Thus their Nd and Hf model ages are necessarily hybrid ages that

in most cases do not reflect discrete crust-forming events; they

are difficult to interpret because the ages, and the relative

contributions of different source terrains, cannot be indepen-

dently determined from whole-rock samples (e.g. Allegre &

Rousseau 1984). Recently, two ways forward have been devel-

oped using zircons: (1) to use O isotopes to screen out zircons

that crystallized from magmas that may contain a contribution

from sedimentary source regions, and may therefore yield hybrid

model ages; (2) to use zircons to constrain the source regions,

and their relative contributions, represented in whole-rock sedi-

ments.

In subsequent sections we consider a series of case studies

showing new advances and/or perspectives in a number of key

areas pertaining to the continental growth and crust–mantle

differentiation. These include the differentiation of the infant

silicate Earth, the composition of nascent continents, rates of

crust generation and preservation in the geological record, and

the information available from the igneous and sedimentary

records.

Differentiation of the infant silicate Earth: formationof early depleted and enriched reservoirs

The depleted mantle is traditionally regarded as the complemen-

tary reservoir to the continental crust (Jacobsen & Wasserburg

1979; O’Nions et al. 1980; Allegre et al. 1983). Isotopically it is

depleted relative to the bulk Earth, it is sampled by MORB at

the present day, and that depletion has typically been attributed

to the extraction of the continental crust. In terms of radiogenic

Table 1. A summary of the SiO2 contents and selected parent–daughtertrace element ratios in selected geological units and rock types

SiO2 Rb/Sr Sm/Nd Lu/Hf

Continental crust1

Bulk 60.6 0.153 0.195 0.081Lower 53.4 0.032 0.255 0.132Upper 66.6 0.256 0.174 0.058

Cratonic shales2,3

Archaean 61.0 1.82 0.175 0.087Proterozoic 63.1 1.53 0.178 0.092Phanerozoic 63.6 1.20 0.178 0.102PAAS 62.8 0.800 0.175 0.086NASC 64.8 0.880 0.204 0.073

GLOSS4 58.6 0.175 0.214 0.102Mid-ocean ridge basalts (n ¼ 331)5

Mean 50.3 0.017 0.335 0.195Median 50.5 0.012 0.335 0.187Mode 51.0 0.003 0.327 0.178

Ocean island basalts (n ¼ 1520)5

Mean 47.6 0.042 0.233 0.085Median 48.0 0.037 0.223 0.060Mode 49.0 0.033 0.226 0.056

Island arcs (n ¼ 644)5

Mean 51.6 0.030 0.313 0.224Median 51.7 0.026 0.318 0.214Mode 51.0 0.013 0.335 0.159

Continental flood basalts (n ¼ 2514)5

Mean 49.8 0.061 0.256 0.112Median 50.1 0.041 0.254 0.105Mode 50.0 0.032 0.265 0.100

Granitoids (n ¼ 7371)5

Mean 65.6 1.07 0.235 0.112Median 64.4 0.072 0.224 0.086Mode 56.0 0.017 0.200 0.063

Archaean cratons sediments (n ¼ 63)5

Mean 65.6 0.635 0.174 0.052Median 64.9 0.160 0.173 0.056Mode 65.0 0.078 0.168 0.056

PAAS, Post-Archaean Australian Shale; NASC, North American Shale Composite;GLOSS, Global Subducting Sediment. 1Rudnick & Gao (2003); 2Condie (1993);3Nance & Taylor (1976); 4Plank & Langmuir (1998); 5GEOROC compilation(http://georoc.mpch-mainz.gwdg.de/georoc/)

Fig. 5. A summary of the U–Pb crystallization and Hf model ages on

zircons from Australia both in magmatic rocks and as detrital minerals.

This contrasts the peaks of crystallization ages for the zircons from

magmatic rocks and as detrital grains in sediments with the broader

distribution of the Hf model ages. The data sources are available upon

request.

C . J. HAWKESWORTH ET AL .234

Page 7: The Generation and Evolution of the Continental Crust

isotope ratios such arguments work well; the extraction of high

Rb/Sr crust, for example, leaves behind a low Rb/Sr residual

depleted mantle, and with time those reservoirs develop high and

low 87Sr/86Sr ratios respectively. In detail, the processes involved

are much debated. If the new material extracted from the mantle

as a precursor to the continental crust is basaltic, the residual

depleted mantle may be sufficiently infertile to generate more

basalt in subsequent melting events, unless some process of

refertilization is invoked. None the less, there has been consider-

able interest in determining the isotope evolution of the depleted

mantle, both as a reference frame for the calculation of model

ages (e.g. Fig. 3), and because it might constrain when the

continental crust was generated.

The initial isotope ratios from whole-rock analyses of rocks

derived from the mantle should in principle offer robust esti-

mates of isotope ratios of the mantle at those times. Figure 6

compares models for the evolution of the depleted mantle based

on initial whole-rock Hf isotope ratios with high �Hf values. The

continuous line was calculated using the 176Lu/177Hf and 176Hf/177Hf ratios of the depleted mantle at the present day (Griffin et

al. 2002) and this is the curve that is most widely used to

calculate the model ages of crustal rocks. In practice, the

observed maximum �Hf values do not simply increase with

decreasing age, as they should in mantle with a constant Lu/Hf,

and this has been attributed to enhanced crustal recycling in the

Archaean (Bennett 2003). The implication is that the depleted

mantle evolution would then be characterized by changes in its

Lu/Hf through time (short-dashed line, Fig. 6). Critically, how-

ever, both these interpretations based on whole-rock composi-

tions are different from those predicted from simple models in

which the depleted mantle and the continental crust are treated

as complementary reservoirs, and the crustal volume grows

through time (the long-dashed line in Fig. 6 is the estimated �Hf

evolution of the mantle adapted from �Nd in the mantle from

Nagler & Kramers (1998)). Particularly for Archaean samples, it

remains difficult to establish the extent to which the whole rocks

have been disturbed isotopically, and some have higher initial Nd

and Hf isotopes ratios than predicted by most models of early

crust extraction. Thus there has been increasing interest in

mineral archives that yield precise, typically U–Pb ages of

crystallization, and preserve robust records of other isotope

systems. Zircons have considerable potential, although because

they tend to crystallize from relatively evolved magmas, they

may offer greater insights into the evolution of the continental

crust than the evolution of the depleted mantle.

Figure 7 contrasts laser ablation zircon Hf isotope data and

solution Hf isotope data from digestion of zircon fractions from

the Amıtsoq gneisses of southern West Greenland (Vervoort et

al. 1996; Vervoort & Blichert-Toft 1999; Kemp et al. 2009a). It

is striking that the zircons analysed from solution tend to have

higher initial Hf isotope ratios, and therefore apparently offer

more insight into the Hf isotope ratios of the depleted mantle.

In contrast, positive �Hf values are not evident within the laser

ablation zircon data, apart from a single analysis of a meta-

morphic overgrowth. The laser ablation Hf isotope data define a

simple array that overlaps the less radiogenic end of the solution

Hf isotope field. The slope of the array is consistent with the

evolution of a crustal reservoir with 176Lu/177Hf c. 0.019, which

was remelted at c. 3.81, 3.71 and 3.65 Ga to yield zircons of

these ages, and such a 176Lu/177Hf ratio is consistent with mafic

to intermediate crust (see Fig. 4). The array provides permissive

evidence for extraction of the crustal source material from

either a chondritic reservoir at c. 3.85 Ga or model depleted

mantle at 4.0 Ga. There is no evidence from these laser ablation

data for the addition of juvenile material between 3.81 and 3.65

Ga, which would generate vertical mixing arrays extending to

higher �Hf .

Of concern is how to reconcile the laser ablation data with the

much larger solution Hf isotope datasets from the Amıtsoq

Gneisses, in which positive �Hf values are prevalent (Fig. 7).

Assuming that both datasets are representative, these data high-

light the difficulty of interpretation of solution �Hf values from

the Amıtsoq gneiss, given the complex zircon microstructures

and U–Pb age dispersions (Whitehouse et al. 1999). The most

Fig. 6. Evolution of �Hf through time in the

depleted mantle. The basalts, komatiites

and Archaean TTG whole-rock

compositions plotted are those with high

�Hf and a range of ages. Continuous black

line: depleted mantle evolution based on

present-day values of 176Lu/177Hf ¼ 0.0384

and 176Hf/177Hf ¼ 0.28325 (Griffin et al.

2002). Short-dashed line: evolution with

changing Lu/Hf through time based on

maximum �Hf in whole rocks (Bennett

2003). Long-dashed line: evolution adapted

from the model for the depleted mantle

based on �Nd from Nagler & Kramers

(1998), and the correlation between Nd and

Hf isotopes after Vervoort et al. (1999).

�Hf was calculated using present-day CHUR

values of 176Lu/177Hf ¼ 0.0336 and176Hf/177Hf ¼ 0.282785 (Bouvier et al.

2008).

THE CONTINENTAL CRUST 235

Page 8: The Generation and Evolution of the Continental Crust

appropriate age at which to calculate the �Hf values for the

Amıtsoq Gneiss zircons requires knowledge of the proportion of

Hf contributed from the different growth zones. This is not easy

to determine for solution analysis, as bulk zircon digestion

homogenizes Hf derived from domains of different age, and

potentially, of different 176Hf/177Hf. However, they can be

resolved by in situ U–Pb and Lu–Hf analysis of zircons from

the same samples from which the solution data were acquired.

The issues raised in the discussion of Figure 7 are important

in the study of the older, more complex Archaean terrains, and

in the analysis of complex detrital zircons, but they are less

significant for the analysis of simple magmatic zircons. Pietranik

et al. (2009) explored how the large volumes of Hf isotope data

on concordant zircons might now be used to constrain the

evolution of the depleted mantle. Figure 8 illustrates that only a

small number of zircons plot on the depleted mantle evolution

line of Griffin et al. (2002) and these are all ,2.7 Ga. Many

zircons also plot above the curve that reflects the sigmoidal

increase in the volume of continental crust through time (the

curve modified after Nagler & Kramers (1998) in Fig. 6). The

implications are as follows.

(1) The depleted mantle evolution line of Griffin et al. (2002)

may overestimate the depleted mantle Hf isotope values, espe-

cially over the first 2.0 Ga of the Earth history. The curve also

gives a high �Hf value of c. 1.3 at 4.56 Ga, using the CHUR

constraints of Bouvier et al. (2008).

(2) The initial mantle depletion took place before c. 3.8 Ga

ago, and increasing and significant depletion is recorded in

zircons before 3.0 Ga. Therefore, all the curves that treat

depleted mantle and continental crust as complementary reser-

voirs with the crust growing through time since a major growth

episode around 3.0 Ga (e.g. Taylor & McLennan 1995; Collerson

& Kamber 1999) may underestimate the depleted mantle values

before 3.0 Ga.

Pietranik et al. (2009) used intersections between trends for

the evolution of unmodified crustal sources (see the next section)

and peaks of ages of crust formation to obtain Hf isotope ratios

of the depleted mantle. The ages of new crust formation were

constrained by vertical trends on U–Pb and �Hf plots, which are

often characterized by stepped increases in the maximum values

of �Hf . This approach provides values of the mantle from which

most of the continental crust was derived and it overcomes the

difficulty that most of the zircons yields contaminated �Hf values.

Unexpectedly perhaps, the reconstructed evolution of the de-

pleted mantle is linear, it has present-day 176Lu/177Hf ¼ 0.0393

and �Hf of c. +15.9, and it is best approximated by the function

�Hf ¼ (�3.85 � 0.1)t + (15.9 � 0.2) (R2 ¼ 0.99), where t is the

depleted mantle age in billion years. It intersects chondrite

evolution (CHUR) at c. 4.0 Ga, consistent with point (2) above.

The linearity implies that the Hf isotope composition of the

depleted mantle was subsequently little affected by the genera-

tion of younger continental crust. Either the rates of crust

generation and recycling were broadly balanced (Armstrong

1981), or the crust was generated from a relatively large volume

of geochemically homogeneous, and previously depleted mantle

(Tolstikhin et al. 2006).

The low Lu/Hf material responsible for the early depletion of

the mantle could have been crust that stabilized at c. 4.0 Ga, or

an early enriched reservoir (EER) that resulted in an additional

depletion episode before the present continental crust was

formed (c. 4.4 Ga; Boyet & Carlson 2006; Tolstikhin et al. 2006;

Shirey et al. 2008). In the former, crust generation was

responsible for the degree of depletion observed, and so the

volume of mantle involved must have been relatively small. For

the latter, the volume of mantle could have been much larger (up

to whole mantle, Tolstikhin et al. 2006), but the EER must have

been sequestered away from well-mixed portions of mantle.

Figure 8 shows a good agreement between the mantle curve

based on zircon data and that representing a two-stage model of

mantle depletion (EER extraction followed by continental crust

Fig. 7. A plot of �Hf v. inferred

crystallization age for zircons from

Greenland gneisses, comparing data

obtained by solution analysis (Vervoort et

al. 1996; Vervoort & Blichert-Toft 1999)

with those measured by laser ablation using

both standard Hf isotope and concurrent

Pb–Hf isotope routines (�Hf values of the

latter are calculated using the laser ablation207Pb/206Pb age; error bars are indicated at

2 SE) (Kemp et al. 2009a). The short-

dashed line depicts the evolution of a

putative crustal reservoir (176Hf/177Hf =

0.019) derived from depleted mantle at 4.0

Ga (the depleted mantle evolution curve

was calculated using mean present-day

values of 176Lu/177Hf ¼ 0.0384 and 176Hf/177Hf ¼ 0.28325, after Griffin et al. 2002).

The solution �Hf data from zircons of

Amıtsoq gneiss GGU-110999 recalculate to

�1.4 and �1.7 at the inferred magmatic

age of 3.65 Ga, as shown by the fine

arrowed line. The second arrowed line

connects the �Hf values of the metamorphic

zircon rim in GGU 125540 calculated at

3.65 Ga and 3.5 Ga. These highlight the

sensitivity of �Hf values calculated for

ancient zircons to crystallization age.

C. J. HAWKESWORTH ET AL .236

Page 9: The Generation and Evolution of the Continental Crust

extraction). In summary, the linear evolution of the depleted

mantle appears to require early extraction of an enriched

reservoir that leaves high Lu/Hf depleted mantle that is in turn

little changed by the subsequent extraction of the continental

crust. If correct, there is no straightforward link between the

isotope evolution of the depleted mantle and the rates of

generation of the continental crust.

Composition of the oldest crust

The composition of the early crust remains controversial. The

presence of Hadean zircons suggests that granitic magmas, and

thus felsic crust, were present at that time, but the key question

is the composition of the crust and mantle from which such

granitic magmas were derived. Lu/Hf ratios decrease with in-

creasing differentiation and they therefore provide a tool with

which to characterize variably differentiated crustal reservoirs

(Fig. 4). On a plot of 176Hf/177Hf v. crystallization age (Fig. 3) a

single crustal reservoir evolves along a straight line, and the

slope of that line corresponds to its Lu/Hf ratio, and hence to its

bulk composition.

The critical step is to recognize arrays representing single

sources that have not been affected by mixing with older or

younger crustal components, as was also required for evaluating

the Hf isotope ratios of the depleted mantle (Fig. 6). One

approach is to screen out zircons from magmas that contain

contributions from sedimentary source rocks, as they will tend to

yield hybrid Hf isotope ratios. Accepting only zircons with

mantle-like �18O (5.3 � 0.3‰) or close to mantle-like values

(usually zircons with �18O up to 6.5‰ are accepted to accom-

modate the analytical precision of ion microprobe analysis;

Cavosie et al. 2005) allows us to distinguish sources unmodified

by contamination with supracrustal material. Figure 9 shows that

zircons with mantle-like �18O form three distinctive peaks in

histograms of model ages for a range of Lu/Hf ratios: two peaks

occur in the Archaean (3.5 and 3.1 Ga) and one, broader peak

occurs at c. 1.5–2.0 Ga. The zircons plot as three arrays on plots

of U–Pb age v. initial �Hf with surprisingly consistent slopes

corresponding to 176Lu/177Hf ¼ 0.019–0.024 (i.e. Lu/Hf ¼0.135–0.17), typical of a basaltic source. The two Archaean

sources appear to coexist for over 0.5 Ga but then after c. 2.7 Ga

the signal of the older crust seems to diminish and the younger

Archaean source signal disappears after the 2.0 Ga source is

formed. Surprisingly, only three general sources have been

identified using zircons with mantle-like �18O, despite including

zircons from a number of areas (Australia, Slave craton,

Mississippi River and Ural) and there is a lack of a clear 2.7 Ga

old source despite extensive magmatism at that time. This may

in part reflect the present shortage of O isotope analyses in the

literature, and/or the selection of the curve for the evolution of

the depleted mantle.

Another way to identify evolution arrays of single crustal

sources is to select areas where one source was reworked for

over c. 0.5 Ga (the time needed to obtain a reasonable regression)

without the addition of new crust, and to determine the slope of

the evolution array formed by zircons with the highest �Hf .

Figure 10a shows the evolution array constrained for South

American zircons (Willner et al. 2008), and although this

approach remains difficult to verify statistically, the slope of the

array is again consistent with mafic sources for these magmas.

Fig. 8. A zircon-based model for the

evolution of �Hf in the depleted mantle. The

bold continuous line is a regression through

open circles representing depleted mantle

compositions based on the approach of

Pietranik et al. (2009). The short-dashed

line is the two-stage evolution of the

depleted mantle based on Tolstikhin et al.

(2006); the mantle was depleted

immediately after accretion (c. 4.6 Ga) by

extraction of an early enriched reservoir,

and the crust was extracted later following

the crustal growth rates of Taylor &

McLennan (1995); 90% of the mantle was

depleted. The fine grey line shows the

depleted mantle evolution based on the

present-day values for the depleted mantle

of 176Hf/177Hf ¼ 0.28325 and176Lu/177Hf ¼ 0.0384 (Griffin et al. 2002).

The long-dashed curve and the CHUR

parameters are as in Figure 6.

THE CONTINENTAL CRUST 237

Page 10: The Generation and Evolution of the Continental Crust

The dominance of implied mafic sources for ancient zircon

arrays is consistent with the expectation that new crust generated

from mantle peridotite is basaltic, rather than being similar to

average present-day andesitic crust.

Similar approaches can be applied to the oldest zircons known,

from Jack Hills in Western Australia. The solution Hf isotope data

of Amelin et al. (1999) define an array whose slope corresponds

to the evolution of a source with 176Lu/177Hf c. 0.022, which is

most typical of a mafic crustal composition (Amelin et al. 1999).

Other studies have identified zircons with lower apparent �Hf

values that might imply the remelting of felsic, tonalite–trondhje-

mite–granodiorite-like (TTG-like) sources with 176Lu/177Hf va-

lues ,0.01 (Harrison et al. 2005, 2008; Blichert-Toft & Albarede

2008). However, for samples younger than 4.0 Ga there is little

evidence that such source regions persisted, which is unexpected

given the dominance of TTG complexes in Archaean terrains. The

Archaean zircons with low �Hf appear to plot on ‘mafic’ arrays

with slopes corresponding to 176Lu/177Hf of 0.022 (Fig. 10b). The

evidence from zircons is that both mafic and felsic crust were

present in the Hadean and the early Archaean, and that the mafic

crust may have been predominant. This is consistent with heat

production arguments that felsic crust is unlikely to have been

thicker than c. 10 km, whereas mafic crust might have been c.

40 km thick (Kamber et al. 2005).

Fig. 9. (a) Evolution of �Hf through time in zircons with mantle-like

�18O. The black line is the depleted mantle evolution based on zircons,

as in Figure 8, and the grey line is the Griffin et al. (2002) evolution

curve. The zircons form three distinct crustal arrays with slopes

corresponding to 176Lu/177Hf ¼ 0.019–0.024. The zircons included in

(a) therefore also form peaks in the Hf model age histograms (b–d),

consistent with many zircons coming from a single source. The model

ages in (b–d) were calculated using the zircon-based mantle evolution

curve; using the Griffin et al. (2002) curve would result in model ages

higher by 100–200 Ma, but with a similar distribution of peaks. For the

model age calculations the selection of 176Lu/177Hf in the source

determines the slope of the arrays in (a). We included zircons that

formed peaks in model ages for a range of 176Lu/177Hf ratios typical for

mafic to intermediate magma sources, and including zircons with model

ages calculated using lower or higher 176Lu/177Hf had minimal effect on

the slopes of the crustal arrays. The zircon data are from Kemp et al.

(2006), Pietranik et al. (2008) and Wang et al. (2009).

Fig. 10. Crustal arrays defined in zircon age v. �Hf diagrams highlighting

zircons that crystallized from magmas extracted from the youngest

(a) and the oldest (b) crustal sources sampled by South American zircons

(Willner et al. 2008).

C. J. HAWKESWORTH ET AL .238

Page 11: The Generation and Evolution of the Continental Crust

Peaks of ages, preservation and crust generation

Figure 1 illustrates the peaks of the ages of rocks that, on the

basis of their mantle-like initial isotope ratios, are thought to

represent new continental crust (McCulloch & Bennett 1994;

Condie 1998, 2005). As indicated above, there is considerable

debate over the extent to which these peaks are representative of

the evolution of the continental crust or are a consequence of

preferential preservation in the geological record. If they are

assumed to be representative, the presence of alternating peaks

and troughs of ages implies dramatic changes in the rates of

crustal production from periods in which unusually large

volumes of new crust were generated to periods of relatively

little new crustal production. It is difficult to envisage how the

global rate of crust generation would vary markedly if the

generation of new continental crust is in response to plate

tectonics. The apparent peaks of crust generation have therefore

been attributed to deep-seated thermal anomalies in the mantle,

and the emplacement of superplumes (Stein & Hofmann 1994;

Albarede 1998) or the triggering of mantle avalanches (Condie

1998; Nelson 1998). One difficulty is that intra-plate magmatic

rocks are not commonly recognized in Archaean granite–green-

stone terranes. Another is that experimental evidence for the

generation of the TTG indicates that they are likely to have been

generated from altered basalt (Foley et al. 2003; Brown &

Rushmer 2006; Clemens et al. 2006), and it is less clear how

hydrothermally altered basalt can be taken down to the site of

partial melting in an intra-plate setting.

Campbell & Allen (2008) summarized the crystallization ages

from c. 7 000 zircons, most of which were sampled from recent

sediments from over 40 rivers worldwide (Fig. 11). They too

yield marked peaks in crystallization ages and, again assuming

that they are representative of igneous events in the sedimentary

provenance, they would reflect periods of increased magmatic

activity in the evolution of the continental crust. Some of the

peaks of crystallization ages, most notably at 1.9 and 2.7 Ga, are

coincident with the periods in which relatively large volumes of

new crust were generated (as indicated in Fig. 1). However, this

is much less marked in the last 1.5 Ga, when the peaks of

crystallization ages appear to represent pulses of magmatic

activity that are not associated with the generation of significant

volumes of new crust. This is consistent with models in which

the volumes of new continental crust generated decrease expo-

nentially with time as the Earth cools (e.g. Walzer & Hendel

1997; Condie 2000; Grigne & Labrosse 2001), with less and less

new crust being generated in the younger magmatic events.

A number of studies have pointed out that the peaks of

crystallization ages occur at times when there were superconti-

nents on the Earth’s surface (Taylor & McLennan 1995; Condie

1998; and see Campbell & Allen 2008; Rino et al. 2008, for

summaries). It is less clear that these should have been periods

of unusual volumes of magmatism, but it may be that rocks

enveloped within supercontinents have more chance of being

preserved. Brown (2007) categorized high-grade orogenic belts

into high-, intermediate- and low-pressure high-temperature

belts. He noted that the high-pressure belts were restricted to the

last 600 Ma, and concluded that they reflect cold subduction as

observed at present along convergent margins. Intermediate- to

low-pressure high-temperature rocks are preserved dating back to

the late Archaean, and Kemp et al. (2007b) pointed out that their

ages are grouped in clusters similar to the peaks of crust

generation illustrated in Figure 1. The implication is that periods

of granulite-facies metamorphism are in some way linked to the

processes of crust generation as suggested by Kemp et al.

(2007b), and/or the peaks of the ages of crust generation and

granulite metamorphism are themselves a function of the

unevenness of the continental record.

One question is the extent to which the observed peaks of ages

are a robust record of the major magmatic events in a particular

area. This can be assessed by comparing the peaks of ages

sampled by zircons in sediments of different ages. Analyses are

available on over 11 000 detrital zircons from Australia, from

sediments that range in age from Recent to Archaean (Fig. 12).

The age distribution for all the zircons considered together is

marked by a sharp peak at c. 2.7 Ga, and smaller peaks at c.

1.6–1.8, c. 1.2 and 0.5 Ga. These four peaks are more clearly

seen in the zircons from sediments deposited in the last 50 Ma,

but then the zircons in sediments through much of the Phaner-

ozoic are dominated by the peaks at c. 1.2 and 0.5 Ga. The age

peaks at 1.6–1.8 Ga begin to appear in zircons from the

Ordovician and Cambrian sediments, and then the older peaks

dominate in the older sediments. It appears that whereas different

age peaks dominate in zircons from sediments of different ages,

all of the four main peaks are sampled by the youngest

sediments, and there is little evidence of age peaks that were

sampled only by older sediments and then not sampled subse-

quently. The implication is that the age peaks sampled in the

youngest sediments are a robust record of the major magmatic

events recorded in Australian geology. The question is then the

extent to which those age peaks are representative of the events

that shaped the geology of this continent.

Fig. 11. Plots of the age distribution of relative volumes of juvenile

continental crust (from Condie 2005, and as in Fig. 1), and of

crystallization ages for over 7000 detrital zircons (Campbell & Allen

2008). The peaks in the zircon crystallization ages are similar to the ages

of supercontinents. The crust generation rate curve illustrates a model in

which the volume of new crust generated decreases with decreasing age,

and the lightly shaded peaks schematically illustrate the relative volumes

of new crust that might consequently be associated with each peak of

zircon crystallization ages.

THE CONTINENTAL CRUST 239

Page 12: The Generation and Evolution of the Continental Crust

Fig. 12. A summary of the distribution of crystallization ages from detrital zircons from sediments of different ages in Australia. For the most part the

peaks of ages older than the host sediments do not change markedly in sediments of different ages, which implies that the age distributions obtained

provide a robust representation of the preserved geological record of Australia. The sketch maps illustrate the locations of the dominant units of different

ages.

C. J. HAWKESWORTH ET AL .240

Page 13: The Generation and Evolution of the Continental Crust

Hawkesworth et al. (2009) explored ways in which the peaks

of ages might be artefacts of preservation in the geological

record, and preferential preservation might be linked to the

development of supercontinents. The fossil record is biased by

the unevenness of the geographical and stratigraphical sampling

effort and inequality in the rock record available for sampling

(Smith 2007). It seems increasingly likely that a similar uneven-

ness biases the record of the generation and the evolution of the

continental crust.

Supercontinents are regarded as an inevitable consequence of

plate tectonics (e.g. Dalziel 1992); they are the outcome of

global tectonics and they come together in compressive phases

associated with subduction. The igneous record associated with

the development and break-up of supercontinents is therefore one

of subduction-related magmatism, collisional orogens and crustal

melting, and subsequently extensional magmatism (Fig. 13). At

issue is the volume of magma generated in each phase, and the

extent to which the magmatic record of each phase will be

preserved and specifically will be represented in the record of

detrital and inherited zircons.

The composition of the continental crust appears to be

dominated by geochemical signatures associated with compres-

sive plate margin magmatism (Taylor 1967; Rudnick 1995;

Hawkesworth & Kemp 2006a,b), yet rocks from this setting have

relatively poor preservation potential in the geological record.

The data compiled by Scholl & von Huene (2007, 2009)

highlight that the global rates of removal of continental and

island arc crust through subduction into the mantle are similar to

the rates at which crust is generated at modern magmatic arcs (c.

2.5 km3 a�1, Fig. 13). Clift & Vannucchi (2004) and Clift et al.

(2009) used different datasets to reach similar but slightly higher

values for the generation and recycling of continental material.

Mineral deposits that form predominantly in convergent margin

settings, such as epithermal and porphyry copper deposits and

orogenic gold deposits, are generally less than 100 Ma old, as a

result of rapid exhumation and erosion (Kesler & Wilkinson

2006; Bierlein et al. 2009), and this too is consistent with the

poor preservation potential of arc magmatic rocks in the

geological record.

Collisional magmatism, by contrast, is dominated by partial

melting of the pre-existing crust. It is granitic, and although the

volume generated may be small relative to some other tectonic

settings (Fig. 14), it will tend to be preferentially protected

within the enveloping supercontinent and it will have good

preservation potential in the geological record. Denudation of the

orogenic belt formed within the core of the supercontinent as a

result of craton assembly will be a major source of detritus

throughout the supercontinent and along its margins (e.g. the late

Mesoproterozoic ‘Grenville’ and late Neoproterozoic ‘Pan-

African’ age detritus that is recorded in Rodinia and Gondwana;

Cawood et al. 2007). In contrast, the extensional phase is

dominated by mafic magmatism; for example, the flood basalts

associated with the break-up of Gondwana (Storey 1995;

Hawkesworth et al. 1999). This phase is unlikely to result in

large volumes of zircons, and the rocks may in any case be

relatively sensitive to erosion into the oceans.

It is argued that the record of magmatic ages is likely to be

dominated by periods when supercontinents assembled, not

because this is a major phase of crust generation but because it

provides a setting for the selective preservation of crust. The

preservation potential, particularly for crystallization ages of

zircons, is greater for late-stage collisional events as the super-

continents come together, rather than for subduction- and exten-

sion-related magmatism (Fig. 14). Such an interpretation is

consistent with the data for the last 1.5 Ga (Fig. 1), but at 1.9 and

2.7 Ga the peaks of ages of crystallization also match up with

striking peaks of crust generation. The processes of crust forma-

tion dominate the geological record at these times. The events at

1.9 Ga are still much debated, but by 2.7 Ga the processes that

shaped the geological record were strikingly different.

The end of the Archaean (2.9–2.6 Ga) is the time of the

stabilization of the Archaean cratons as preserved at present.

This reflects a particular stage in the cooling of the Earth (e.g.

Vlaar et al. 1994; King 2005; Korenaga 2006), and the

Fig. 13. A schematic cross-section of convergent, collisional and extensional plate boundaries associated with supercontinent cycle showing estimated

amounts (in km3 a�1) of continental addition (numbers in parentheses above Earth surface) and removal (numbers in brackets below surface). Data from

Scholl & von Huene (2007, 2009). The volume of continental crust added through time via juvenile magma addition appears to be compensated by the

return of continental and island arc crust to the mantle. Scholl & von Huene (2007, 2009) estimated that the long-term global average rate of arc magma

additions is 2.8–3.0 km3 a�1 (see also Franz et al. 2006). The total volume of crustal material moved into the mantle at convergent and collisional

boundaries is around 3.2 km3 a�1. This rate is sufficient that if plate tectonics has been operating since around 3.0 Ga (see Cawood et al. 2006) then a

volume equal to the total current volume of continental crust would have been recycled into the mantle (Scholl & von Huene 2007). The implication is

that the net growth of continental crust at the present day is effectively nil, and convergent plate margins are sites of crustal recycling and reworking, as

well as continental addition.

THE CONTINENTAL CRUST 241

Page 14: The Generation and Evolution of the Continental Crust

predictable consequence is that relatively large volumes of rock

are preserved from that time (Fig. 1). One possibility is that the

2.9–2.6 Ga rocks are representative of the rocks and associations

that formed earlier, although for how far back in time is not well

constrained, but is simply trapped and preserved by the accident

of history at this stage in the cooling of the outer part of the

Earth. For rocks formed earlier in the Archaean, the preservation

potential is extremely poor (as a result of increased recycling

rates in a hotter Earth), and we infer that the record is therefore

much less likely to be biased by the tectonic setting in which the

rocks were formed.

Thus it is envisaged that the late Archaean marks the transition

from a period of relatively poor preservation to one in which the

geological record is biased by the tectonic setting in which the

rocks were formed. It follows that for events older that 2.9 Ga

zircons may provide the best available record, because rocks

from different settings then have similar preservation potential,

albeit extremely poor, and the record is less biased by the

controls on preservation that mark post-Archaean magmatic

processes.

An important inference that can be drawn from the compila-

tions of volumes of continental generation and recycling (Fig.

13) is that the current volume of continental crust would have

been recycled back into the mantle over the last 2–3 Ga and thus

this volume of continental crust must have existed prior to this

time.

The igneous record: insights from granites

Igneous rocks are involved in the generation of new continental

crust, and recent isotope studies have focused on (1) the relative

contributions of new and pre-existing crust in the generation of

the granites, and on tracking the geodynamics of crustal evolu-

tion, and (2) unmixing granite sources to pinpoint times of

crustal growth. They are well illustrated using the Tasmanides of

southeastern Australia (Kemp et al. 2006, 2009b), a 700 km wide

segment of an 18 000 km long orogenic system that developed

along the palaeo-Pacific margin of Gondwana from the Neopro-

terozoic to early Mesozoic (Cawood 2005).

Within the Tasmanides, the Lachlan Fold Belt has a prominent

place in the development of ideas about the petrogenesis of

granite. It is where the influential classification scheme of I- and

S-type granites was developed (Chappell & White 1974, 1992),

and it is an area that has been at the centre of debates about the

balance of new and pre-existing crust involved in the generation

of granite, and how those are best determined (Gray 1984; Keay

et al. 1997; Collins 1998, 1999; Chappell et al. 1999). This is

clearly a central issue for discussions of when and how the

continental crust was generated. The I- and S- classification

scheme was developed at a time when granites were thought to

be windows on their source regions, and those source regions

could be inferred to be infracrustal (i.e. broadly igneous) or

supracrustal (typically sedimentary) material. Granites were

regarded as a way of looking into compositional variations deep

in the crust. One difficulty was over the introduction of heat to

induce partial melting. In many models this would come from

the mantle and therefore include contributions from mantle-

derived magmas. Furthermore, the Lachlan Fold Belt granites

define an apparently simple, overlapping array on the �Nd v.

initial 87Sr/86Sr diagram (McCulloch & Chappell 1982), which

has been almost universally used to infer a large-scale mixing

process between primitive magmas from the depleted upper

mantle and evolved crustal end-members (e.g. Gray 1984; Faure

1986). The implication is that single granites were unlikely to

have been derived from single source regions.

The Lachlan Fold Belt has two main components, a mono-

tonous succession of mature (quartz- and clay-rich) Ordovician

turbidites and a large volume of granitic rocks. The turbidites

apparently accumulated on an oceanic substrate and were subject

to episodic deformation, low-grade regional metamorphism and

massive igneous intrusion from c. 450 to 340 Ma (Gray & Foster

2004). Such granite–turbidite belts appear strikingly different

from orogenic tracts developed in collisional belts through a

Wilson cycle of ocean opening and closing, and they form part

of an accretionary orogenic belt (Coney 1992; Cawood et al.

2009). The Lachlan Fold Belt is thought to have occupied a

back-arc setting for much of its evolution, behind a broadly

eastward migrating subduction zone (Collins 2002a,b; Gray &

Foster 2004; Cawood 2005; Foster et al. 2005).

Kemp et al. (2007a) reported U–Pb, Hf and O isotope

measurements on magmatic zircons from three granite suites in

the Lachlan Fold Belt. The striking feature of the Hf isotope data

is that zircons from single whole-rock samples exhibited a

spectrum of �Hf values of up to 10 �Hf units. Such variations

within a single sample are reconciled only by the operation of

open-system processes that are capable of modifying the 176Hf/177Hf ratio of the magmas from which the zircons precipitated.

The polarity of Hf isotope changes during zircon growth can be

determined from the isotope changes during the growth of single

crystals, and by pairing the isotope variations with trace element

ratios (e.g. Th/U) that are proxies for the degree of differentia-

tion. In most cases, 176Hf/177Hf decreases with increasing

differentiation (see Fig. 4) as would be induced by addition of an

unradiogenic (continental crust-like) component during crystal-

lization. Significantly, however, the zircons from both I- and S-

Fig. 14. The volumes of magma generated (continuous line), and their

likely preservation potential (dashed lines), may vary in the three stages

associated with the convergence, assembly and breakup of a

supercontinent (after Hawkesworth et al. 2009). The preservation

potential in the first stage is greater at margins where the subduction

zone retreats oceanward to form extensional basins than at margins

where the subduction zone advances toward the continent. Thus, peaks in

the crystallization ages that are preserved (shaded area) reflect the

balance between the magma volumes generated in the three stages and

their preservation potential.

C . J. HAWKESWORTH ET AL .242

Page 15: The Generation and Evolution of the Continental Crust

type granites have Hf isotope ratios that trend back towards

values similar to those in mantle-derived magmas. The in situ

analyses of Hf isotopes in zircons are most consistent with there

being mantle contributions in the generation of both I- and S-

type granites.

Figure 15 summarizes the variations in �Hf in zircons from

granitic rocks with their age of crystallization across the

Tasmanides (Kemp et al. 2009b). The changes from compres-

sional to extensional tectonic regimes are marked by changes in

zircon �Hf (and whole-rock �Nd, not shown). Thus, for example,

crustal reworking after back-arc closure is registered by the

marked decrease in zircon �Hf that follows major crustal thicken-

ing episodes in each Tasmanide terrane (Fig. 15). These inflec-

tions signify the emplacement of S-type rocks generated under

granulite-facies conditions largely from metasedimentary precur-

sors (White & Chappell 1977). Subsequent extension is accom-

panied by increases in �Hf and an accompanying change from

peraluminous S-type to metaluminous I-type compositions, pre-

sumably in response to a waning sedimentary contribution, as

suggested by a steady decrease in zircon �18O (Kemp et al.

2009b). The links between the shape of the isotope–time trends

and the pattern of compressional and extensional events in the

Tasmanides (Fig. 15) highlight the feedback between tectonic

activity and magma source. In detail, Hf isotope ratios of zircons

have been used to estimate the mantle contributions in different

granite bodies generated at different tectonic stages. These range

from 30–40% in the S-types analysed to c. 70% in the I-types,

and up to 90% in the A-types, and clearly these represent the

volumes of new crust generated in the Lachlan Fold Belt

orogenic episode (Kemp et al. 2009b). Unexpectedly, because the

S-type granites are more voluminous than the other granite types,

a large proportion of the new crust appears to have been

generated during S-type granite magmatism.

The second aspect is in using inherited and detrital zircons to

pinpoint periods of crustal growth. The detrital zircons in

greywackes from the Lachlan Fold Belt and the inherited zircons

in the granitic rocks yield similar distributions of crystallization

ages (Kemp et al. 2006; Fig. 16). There is a marked peak at

500–650 Ma and another at 0.9–1.2 Ga that manifest Pan-

African and Grenvillian phases of supercontinent-related orogen-

esis, respectively, and then there is a scattering of ages back to

3.5 Ga. The peaks of zircon crystallization ages are taken to

reflect peaks of magmatic activity prior to the formation of the

Lachlan Fold Belt, and the initial question is the extent to which

these reflect periods of new crustal growth, or crustal reworking

primarily through remelting of the pre-existing crust.

Hf model ages represent the times when the crustal source

rocks for granitic magmas were themselves derived from the

mantle (Fig. 3). Kemp et al. (2006) used O isotopes to identify

zircons that crystallized from magmas that contained a contribu-

tion from sedimentary source rocks (those with �18O . 6.5‰),

Fig. 15. The isotope and tectono-magmatic

evolution of the Tasmanides defined by the

�Hf values of granite-hosted zircons (after

Kemp et al. 2009b). Filled diamonds

represent analyses from mafic units. The

grey shaded time slices correspond to

major contractional episodes, as follows:

HB, Hunter–Bowen; K, Kanimblan;

T, Tabberabberan; Bo, Bowning; Be,

Benambran (early phase); D, Delamerian.

The timing of these events is taken from

Landenberger et al. (1995), Collins &

Hobbs (2001), Collins (2002a), Gray &

Foster (2004), Foden et al. (2006) and

Cawood & Buchan (2007).

Fig. 16. A histogram of ages obtained from inherited and detrital zircons

from c. 430–380 Ma old granites and Ordovician turbidites in the

Lachlan Fold Belt (Kemp et al. 2006). The crystallization ages have

peaks at c. 500 and 1000 Ma, whereas the Hf model ages (see Fig. 3),

here termed crust formation ages, are much older with peaks at c. 1.9

and 3.3 Ga. The peaks of model ages in zircons with �18O , 6.5‰

indicate when new crust in the areas sampled by these zircons was

generated. The model ages of zircons with �18O . 6.5‰ contain a

contribution from supracrustal rocks, and they are therefore more likely

to represent hybrid model ages.

THE CONTINENTAL CRUST 243

Page 16: The Generation and Evolution of the Continental Crust

as these are likely to have hybrid model ages that are unlikely to

represent discrete crust-forming events. Zircons with �18O ,

6.5‰ are thought to have been derived from magmas derived

from igneous source rocks, and hence to have model ages that

are more likely to reflect periods of crust generation. The results

are very striking. Zircons with �18O , 6.5‰ define two marked

peaks of Hf model ages at 1.9 and 3.3 Ga (Fig. 16, using the

depleted mantle model of Griffin et al. 2002), and these are

taken to be periods of generation of significant volumes of new

continental crust in the source region. In contrast, the zircons

with �18O . 6.5‰ yield a range of model ages. Many are

intermediate between 1.9 and 3.3 Ga, they have a broad peak at

2.1–2.2 Ga, and so they presumably represent mixtures of rocks

generated at 1.9 and 3.3 Ga. The oxygen isotope data therefore

allowed Kemp et al. (2006) to conclude that the 2.1–2.2 Ga age

peak represented mixing in the sedimentary environment, rather

than a real crust-forming event. More generally, with this

approach we can now contrast the evolution of the igneous and

sedimentary reservoirs in the continental crust, and compare the

information from zircons with the models for the evolution of

the continental crust based on Nd isotope ratios in shales

(Allegre & Rousseau 1984).

The sedimentary record: erosion models andcontinental maturation

Fine-grained continental sediments sample the available crust at

the time of deposition. They are widely used to obtain average

abundances of the upper crust for insoluble elements, and so

there is considerable interest in using Nd and Hf isotopes in

sediments to constrain models for the evolution of the crust.

Allegre & Rousseau (1984) modelled Nd isotopes in shales of

different ages in terms of the growth of the continental crust. On a

plot of the model age of the sample against its sedimentation age,

or crystallization age in the case of zircons, new crust that is then

reworked in subsequent events results in horizontal arrays (Fig.

17). Samples that have younger model ages in younger sediments,

as seen for Nd isotopes in shales, indicate that new crust was

generated in younger events and it has then been sampled by the

younger sediments. As the slope of the data array flattens out, it

indicates that with time less and less new crust has been gener-

ated. Allegre & Rousseau (1984) assumed that on average new

crust was generated every 500 Ma, and they evaluated the links

between the sediments analysed and their source rocks using an

erosion factor ‘K’. K relates the model age of the sediments

analysed to the average model age of their source rocks, given that

some source terrains are more susceptible to erosion than others.

Models were evaluated for different values of K, largely because

K has been difficult to constrain (Allegre & Rousseau 1984;

Goldstein & Jacobsen 1988; Jacobsen 1988; Kramers & Tolstikhin

1997; Tolstikhin & Kramers 2008).

It remains difficult to assess independently the relative con-

tributions of sources of different ages in a sample of sediment.

One way forward is to combine Hf isotopes in zircon with Nd

isotopes in whole-rock samples. The distribution of Hf model

ages in detrital zircons in a sediment offers insight into the

proportions of different source terrains that have contributed to

the bulk sample, and hence on the actual value of K. However,

zircon is a heavy mineral and the Nd budget in sediments is

dominated by the clay fraction (see Vervoort et al. 1999), and it

has to be established how well these records compare.

Dhuime et al. (2009) undertook a Hf and Nd isotope study on

samples of recent sediment from the Frankland River in south-

western Australia, building on the earlier study of Cawood et al.

Fig. 17. (a) Model ages v. crystallization ages of detrital zircons from five recent sediments from the Frankland River in SW Australia (Dhuime et al.

2009). The bold diamonds are the means of the zircon data (small dots) grouped into five main periods of zircon crystallization: 3.4–3.0, 2.8–2.5,

2.4–1.4, 1.35–1.0 and 0.7–0.4 Ga. The average values for zircons of different ages define a similar trend through time to the model Nd ages of shales of

different ages from the Australian continent (dashed curve). The latter was used by Allegre & Rousseau (1984) to model the evolution of the continental

crust using different values of the erosion constant ‘K’ (see text). The inset illustrates that new crust would plot on the 1:1 line, and new crust

subsequently defines a horizontal array retaining the same model age through younger geological events. Displacement down the diagram (i.e. to younger

model ages) requires a contribution from younger crustal material. (b) The relative contribution of zircons from the Yilgarn block in the zircon

populations in recent sediments plotted against distance along the Frankland River (grey envelope curve). These are compared with the proportion of

Archaean source rock available in the catchment for each sediment sample (open squares). The calculated K values for each sediment, and the change in

elevation down the river, are also shown. The inflexion in the river’s profile at about 80 km from the coast is responsible for the increase of K. The high

values of K close to the contact with the Yilgarn are subject to large errors, and are poorly understood. Values of K ¼ 4–6 appear representative of mature

river systems that sample large areas of continental crust.

C . J. HAWKESWORTH ET AL .244

Page 17: The Generation and Evolution of the Continental Crust

(2003). The Frankland River is c. 320 km in length and the

catchment area is c. 4630 km2. It drains two terranes with

different zircon crystallization and model age distributions, the

Archaean Yilgarn craton and the Proterozoic Albany–Fraser

mobile belt. Thus it offers an opportunity to compare the

distribution of the rocks sampled in different sediments and the

proportions of those rocks in the catchment area for each

sediment. The Hf and U–Pb age data on zircons from four recent

sediments are summarized in Figure 17. Although there is a large

range of Hf model ages in any group of zircons with similar U–

Pb ages, it is striking that the trend of the average value of the

Hf model ages through time is broadly similar to that for Nd

model ages in Australian shales (Allegre & Rousseau 1984). This

suggests that the distribution of zircon data may be used to

constrain values of K for different sediment samples.

Dhuime et al. (2009) used the model age and the crystal-

lization age data on detrital zircons to calculate the contribution

from the Yilgarn craton and the Albany–Fraser belt in each

sample, and to compare that with the proportion of those two

units in the catchment again for each sample. Overall there is a

downstream decrease in the proportion of Yilgarn material in the

recent sediments (Fig. 17b, grey curve). Moreover, the K values

increase with distance from the Yilgarn craton, and with the

gradient of the river profile (K varies from c. 4–6 to c. 15–17,

Fig. 17b). Samples with K c. 9–10 and c. 15–17 are from below

the inflection that reflects a weak escarpment at c. 80 km from

the coast associated with Miocene–Pliocene uplift (Cawood et

al. 2003) (Fig. 17b). The steepening of the river’s profile has

resulted in preferential erosion of material from the Albany–

Fraser belt, and hence an increase in the calculated K values. In

turn the ‘stable’ segment of the Frankland River is best sampled

above the escarpment at distances of 100–150 km from the

coast. The implication is that K values of 4–6 are representative

of mature river systems that sample large areas of continental

crust, and models using values of K ¼ 2–3 result in biases

towards younger ages in the global models of crust formation

and evolution through time (Allegre & Rousseau 1984; Goldstein

& Jacobsen 1988; Jacobsen 1988; Kramers & Tolstikhin 1997).

It is encouraging that Nd and Hf isotopes can now be combined

to evaluate the erosion constant K and how it varies with changes

in discharge rates and relief.

Synthesis

New analytical approaches have offered new insights into the

generation and evolution of the continental crust. This remains

the key archive of how processes and conditions have changed

during the evolution of the Earth. The oldest rock is 4 Ga old,

and only 7% of the preserved continental crust is older than 2.5

Ga. Yet most of the crust is generally inferred to have been

generated by that time (see Fig. 11), and the challenge is to

decipher the major events in the generation of the crust from this

small fragment of the geological record. From 4.4 to 4.0 Ga the

only samples are tiny crystals of detrital zircons, and therefore

much of the discussion has focused on the information available

from the zircon crystals themselves and the silicate inclusions

within them. Zircons are the basis for the geological time scale,

in that they yield precise crystallization ages, and they have been

increasingly used to determine crust formation ages using Hf

isotopes. The continental record is marked by peaks in ages of

crystallization, which in turn imply periods of enhanced mag-

matic activity, and by peaks in ages of crust formation, which

might be taken to reflect periods of enhanced crust generation.

However, an alternative interpretation is that these peaks of ages

are not a true record of such igneous processes, but instead an

artefact of preservation. The geological record is far from

complete, and the fossil record is biased by the nature of the

fossils, sedimentary facies and the nature of the rock record. The

peaks of crystallization ages also mark the times of super-

continent formation, and it is increasingly understood that the

preservation potentials of rocks generated in different tectonic

settings are very different. It may simply be that the development

of a supercontinent offers markedly improved preservation

potential for magmas formed at such times (Hawkesworth et al.

2009).

The nature of the crustal source rocks of the magmas from

which Hadean and early Archaean zircons crystallized can be

inferred from thermal considerations (Kamber et al. 2005), and

the estimated source Lu/Hf ratios. Both approaches indicate that

the early crust was predominantly mafic in composition, but that

there was a felsic component, as sampled by some of the

magmas from which the Jack Hills zircons crystallized (Harrison

et al. 2005, 2008; Blichert-Toft & Albarede 2008). The early

crust is likely to have had a bimodal distribution in silica, and as

such to have been different from geological younger subduction-

related associations. For rocks formed in the Hadean and early

Archaean the preservation potential was extremely poor, presum-

ably because of meteorite bombardment and increased recycling

rates in a hotter Earth. In many cases detrital and inherited

minerals, such as zircon, remain the main geological archive, and

there is little sense that this early geological record is biased by

the tectonic setting in which the rocks were formed. The end of

the Archaean (2.9–2.6 Ga) is the time of the stabilization of the

Archaean cratons as preserved at present. One implication is that

this reflects a particular stage in the cooling of the Earth, and

that is why relatively large volumes of rock are preserved from

that time (Fig. 1). It may be that the 2.9–2.6 Ga rocks are

representative of the rocks and associations that formed earlier,

although for how far back in time is not well constrained, but is

simply trapped and preserved by the accident of history at this

stage in the cooling of the outer part of the Earth.

Many models indicate that large volumes of continental crust

were generated by the end of the Archaean, and that the volumes

of new crust generated decreased markedly since that time (e.g.

Fig. 11). However, these arguments were in part based on the

evolution of the depleted mantle that was attributed to the

generation of the continental crust. Yet if the depleted mantle is

linear in its isotope evolution (e.g. Fig. 8), and the depletion is

attributed to early differentiation events, then it cannot be used to

monitor the development of continental crust. Instead, arguments

in favour of large volumes of continental crust, and the likely

thickness of felsic and mafic crust, before the end of the

Archaean rely on thermal models for the decay in radiogenic

heat production and the progressively cooling Earth (Davies

1999; Kamber et al. 2005). These are in turn consistent with

recent estimates that the rates of crust generation and destruction

along subduction zones are strikingly similar (Clift & Vannuchi

2004; Scholl & von Huene 2007, 2009). The implication is that

the present volume of continental crust was established 2–3 Ga

ago.

The Late Archaean marks a change, with the subsequent

record controlled by the competing forces of preservation and

generation (Fig. 14). Supercontinents were developed intermit-

tently, and palaeomagnetic data indicate that blocks of crust

moved laterally relative to one another (Cawood et al. 2006).

Plate tectonics may have been active before 3 Ga, but it was

dominant since the end of the Archaean. The geological record,

however, is biased by peaks of ages linked to the development of

THE CONTINENTAL CRUST 245

Page 18: The Generation and Evolution of the Continental Crust

supercontinents (Fig. 11), and, as we have argued elsewhere, by

the development of granulite-facies rocks that are in turn difficult

to destroy (Kemp et al. 2009b). One test of the links between the

development of granulite-facies rocks and the stabilization of the

continental crust is through Pb isotopes. Granulites are character-

ized by low U/Pb, and hence with time unradiogenic Pb isotope

ratios (e.g. Rudnick & Goldstein 1990; Zartman 1990). There

has been a lengthy discussion on the need to identify a low U/Pb

reservoir to model the Pb isotope evolution of the crust and

mantle (Rudnick & Goldstein 1990; Zartman 1990; Asmerom &

Jacobsen 1993; Kramers & Tolstikhin 1997; Hofmann 2008).

Some have invoked granulite-facies lower crust, and typical

models suggest � values (238U/204Pb) of 3–4 in the lower crust,

10–11 in the upper crust and 7–8 in the bulk crust. Thus, c.

30% of the continental crust may be granulite-facies rocks that

are relatively difficult to destroy, and so help in the stabilization

of the crust.

One implication of the different preservation potential of rocks

generated in different settings is that different records should be

preserved in sediments that survive from different settings. There

is also increasing evidence that models for the evolution of the

continental crust now need to integrate large-scale, plate-tectonic

cycles that result in the development and destruction of super-

continents, and shorter-lived 25–100 Ma cycles that differ

depending on the extent to which the plate margin is advancing

or retreating. Such cycles have been identified in the western

American Cordilleras, an advancing orogen where the overriding

plate exerts the dominant control (DeCelles et al. 2009). In sharp

contrast, the Tasmanides represent the retreating orogens of the

western Pacific, which are controlled by the lower (subducting)

plate and where new crust is generated, and stabilized, in back-

arc settings (Kemp et al. 2009b). There is now considerable

scope to establish the evolution of single orogens and to use

isotopes in igneous rocks to develop more detailed models of the

geodynamics of crust generation and evolution.

We thank R. Strachan for inviting this review, and his subsequent

patience during the writing of the manuscript, and M. Whitehouse, S.

Daly and A. Prave for their thorough and supportive reviews. C.H.

gratefully acknowledges support from the NERC (NE/E005225/1) and a

Royal Society Wolfson Award, as does C.S. for an NERC Fellowship NE/

D008891/1, and A.K. for an Australian Research Council Fellowship

(DP0773029).

References

Albarede, F. 1998. The growth of continental crust. Tectonophysics, 296, 1–14.

Allegre, C.J. & Rousseau, D. 1984. The growth of the continent through

geological time studied by Nd isotope analysis of shales. Earth and Planetary

Science Letters, 67, 19–34.

Allegre, C.J., Hart, S. R. & Minster, J.-F. 1983. Chemical structure and

evolution of the mantle and continents determined by inversion of Nd and Sr

isotopic data, I. Theoretical methods. Earth and Planetary Science Letters,

66, 177–190.

Amelin, Y., Lee, D.-C., Halliday, A.N. & Pidgeon, R.T. 1999. Nature of the

Earth’s earliest crust from hafnium isotopes in single detrital zircons. Nature,

339, 252–255.

Arculus, R.J. 1999. Origins of the continental crust. Journal and Proceedings of

the Royal Society of New South Wales, 132, 83–110.

Armstrong, R.L. 1981. Comment on ‘Crustal growth and mantle evolution:

inferences from models of element transport and Nd and Sr isotopes’.

Geochimica et Cosmochimica Acta, 45, 1251.

Armstrong, R.L. 1991. The persistent myth of crustal growth. Australian Journal

of Earth Sciences, 38, 613–630.

Arndt, N.T. & Goldstein, S.L. 1989. An open boundary between lower

continental crust and mantle: its role in crust formation and crustal cycling.

Tectonophysics, 161, 201–212.

Asmerom, Y. & Jacobsen, S.B. 1993. The Pb isotopic evolution of the Earth:

inferences from river water suspended loads. Earth and Planetary Science

Letters, 115, 245–256.

Bennett, V.C. 2003. Compositional evolution of the mantle. In: Carlson, R.W.

(ed.) Treatise on Geochemistry, Vol. 2: The Mantle and Core. Elsevier,

Amsterdam, 493–519.

Bierlein, F.P., Groves, D.I. & Cawood, P.A. 2009. Metallogeny of accretionary

orogens—the connection between lithospheric processes and metal endow-

ment. Ore Geology Reviews, 36, 282–292.

Blichert-Toft, J. & Albarede, F. 2008. Hafnium isotopes in Jack Hills zircons

and the formation of the Hadean crust. Earth and Planetary Science Letters,

265, 686–702.

Bouvier, A., Vervoort, J.D. & Patchett, P.J. 2008. The Lu–Hf and Sm–Nd

isotopic composition of CHUR: Constraints from unequilibrated chondrites

and implications for the bulk composition of terrestrial planets. Earth and

Planetary Science Letters, 273, 48–57.

Bowring, S.A. & Williams, I.S. 1999. Priscoan (4.00–4.03 Ga) orthogneisses

from northwestern Canada. Contributions to Mineralogy and Petrology, 134,

3–16.

Boyet, M. & Carlson, R.W. 2006. A new geochemical model for the Earth’s

mantle inferred from 146Sm–142Nd systematics. Earth and Planetary Science

Letters, 250, 254–268.

Brown, M. 2007. Metamorphic conditions in orogenic belts: A record of secular

change. International Geology Review, 49, 193–234.

Brown, M. & Rushmer, T. 2006. Evolution and Differentiation of the Continental

Crust. Cambridge University Press, Cambridge.

Campbell, I.H. & Allen, C.M. 2008. Formation of supercontinents linked to

increases in atmospheric oxygen. Nature Geoscience, 1, 554–558.

Carlson, R.W. & Boyet, M. 2008. Composition of the Earth’s interior: the

importance of early events. Philosophical Transactions of the Royal Society,

Series A, 366, 4077–4103.

Cavosie, A.J., Valley, J.W., Wilde, S.A. & Edinburgh Ion Microprobe Facility

2005. Magmatic �18O in 4400–3900 Ma detrital zircons: A record of the

alteration and recycling of crust in the Early Archean. Earth and Planetary

Science Letters, 235, 663–681.

Cawood, P.A. 2005. Terra Australis Orogen: Rodinia breakup and development of

the Pacific and Iapetus margins of Gondwana during the Neoproterozoic and

Paleozoic. Earth-Science Reviews, 69, 249–279.

Cawood, P.A. & Buchan, C. 2007. Linking accretionary orogenesis with

supercontinent assembly. Earth-Science Reviews, 82, 217–256.

Cawood, P.A., Nemchin, A.A., Freeman, M. & Sircombe, K. 2003. Linking

source and sedimentary basin: Detrital zircon record of sediment flux along a

modern river system and implications for provenance studies. Earth and

Planetary Science Letters, 210, 259–268.

Cawood, P.A., Kroner, A. & Pisarevsky, S. 2006. Precambrian plate tectonics:

Criteria and evidence. GSA Today, 16, 4–11.

Cawood, P.A., Nemchin, A.A., Strachan, R., Prave, T. & Krabbendam, M.

2007. Sedimentary basin and detrital zircon record along East Laurentia and

Baltica during assembly and breakup of Rodinia. Journal of the Geological

Society, London, 164, 257–275.

Cawood, P.A., Kroner, A., Collins, W.J., Kusky, T.M., Mooney, W.D. &

Windley, B.F. 2009. Accretionary orogens through Earth history. In:

Cawood, P.A. & Kroner, A. (eds) Earth Accretionary Systems in Space and

Time. Geological Society, London, Special Publications, 318, 1–36.

Chappell, B.W. & White, A.J.R. 1974. Two contrasting granite types. Pacific

Geology, 8, 173–174.

Chappell, B.W. & White, A.J.R. 1992. I-Type and S-Type Granites in the Lachlan

Fold Belt. Transactions of the Royal Society of Edinburgh: Earth Sciences,

83, 1–26.

Chappell, B.W., White, A.J.R., Williams, I.S., Wyborn, D., Hergt, J.M. &

Woodhead, J.D. 1999. Discussion—Evaluation of petrogenetic models for

Lachlan Fold Belt granitoids: implications for crustal architecture and

tectonic models. Australian Journal of Earth Sciences, 46, 827–831.

Clemens, J.D., Yearron, L.M. & Stevens, G. 2006. Barberton (South Africa)

TTG magmas: Geochemical and experimental constraints on source-rock

petrology, pressure of formation and tectonic setting. Precambrian Research,

151, 53–78.

Clift, P. & Vannucchi, P. 2004. Controls on tectonic accretion versus erosion in

subduction zones: Implications for the origin and recycling of the continental

crust. Reviews of Geophysics, 42, 1–31.

Clift, P., Schouten, H. & Vannucchi, P. 2009. Arc–continent collisions,

sediment recycling and the maintenance of the continental crust. In:

Cawood, P.A. & Kroner, A. (eds) Accretionary Orogens in Space and Time.

Geological Society, London, Special Publications, 318, 75–103.

Cloos, M. 1993. Lithospheric buoyancy and collisional analysis: subduction of

oceanic plateaus, continental margins, island arcs, spreading ridges, and

seamounts. Geological Society of America Bulletin, 105, 715–737.

Collerson, K.D. & Kamber, B.S. 1999. Evolution of the continents and the

C. J. HAWKESWORTH ET AL .246

Page 19: The Generation and Evolution of the Continental Crust

atmosphere inferred from Th–U–Nb systematics of the depleted mantle.

Science, 283, 1519–1522.

Collins, W.J. 1998. Evaluation of petrogenetic models for Lachlan Fold Belt

granitoids: implications for crustal architecture and tectonic models. Austra-

lian Journal of Earth Sciences, 45, 483–500.

Collins, W.J. 1999. Reply—Evaluation of petrogenetic models for Lachlan Fold

Belt granitoids: implications for crustal architecture and tectonic models.

Australian Journal of Earth Sciences, 46, 831–836.

Collins, W.J. 2002a. Hot orogens, tectonic switching and the creation of

continental crust. Geology, 30, 535–538.

Collins, W.J. 2002b. The nature of extensional accretionary orogens. Tectonics, 21,

1–12.

Collins, W.J. & Hobbs, B.E. 2001. What caused the Early Silurian change from

mafic to silicic (S-type) magmatism in the eastern Lachlan Fold Belt?

Australian Journal of Earth Sciences, 48, 25–41.

Condie, K.C. 1993. Chemical composition and evolution of the upper continental

crust: Contrasting results from surface samples and shales. Chemical Geology,

104, 1–37.

Condie, K.C. 1998. Episodic continental growth and supercontinents: a mantle

avalanche connection? Earth and Planetary Science Letters, 163, 97–108.

Condie, K.C. 2000. Episodic continental growth models: afterthoughts and

extensions. Tectonophysics, 322, 153–162.

Condie, K.C. 2004. Supercontinents and superplume events: distinguishing signals

in the geologic record. Physics of the Earth and Planetary Interiors, 146,

319–332.

Condie, K.C. 2005. Earth as an Evolving Planetary System. Elsevier, Amsterdam.

Coney, P.J. 1992. The Lachlan belt of eastern Australia and circum-Pacific tectonic

evolution. Tectonophysics, 214, 1–25.

Dalziel, I.W.D. 1992. On the organization of American plates in the Neoproter-

ozoic and the breakout of Laurentia. GSA Today, 2, 238–241.

Davies, G. 1999. Dynamic Earth: Plates, Plumes and Mantle Convection.

Cambridge University Press, New York.

DeCelles, P.G., Ducea, M.N., Kapp, P. & Zandt, G. 2009. Cyclicity in

Cordilleran orogenic systems. Nature Geoscience, 2, 251–257.

Dhuime, B., Hawkesworth, C.J., Storey, C. & Cawood, P.A. 2009. Linking the

continental growth record of igneous and sedimentary rocks. Geochimica et

Cosmochimica Acta, 73, A287.

Eiler, J.M. 2001. Oxygen isotope variations of basaltic lavas and upper mantle

rocks. In: Valley, J.W. & Cole, D.R. (eds) Stable Isotope Geochemistry.

Mineralogical Society of America, Reviews in Mineralogy and Geochemistry,

43, 319–364.

Ellam, R.M. & Hawkesworth, C.J. 1988. Is average continental crust generated

at subduction zones? Geology, 16, 314–317.

Faure, G. 1986. Principles of isotope Geology, 2nd edn. Wiley, New York.

Foden, J., Elburg, M.A., Dougherty-Page, J. & Burtt, A. 2006. The timing

and duration of the Delamerian orogeny: Correlation with the Ross Orogen

and implications for Gondwana assembly. Journal of Geology, 114, 189–210.

Foley, S.F., Buhre, S. & Jacob, D.E. 2003. Evolution of the Archaean crust by

delamination and shallow subduction. Nature, 421, 249–252.

Foster, D.A., Gray, D.R. & Spaggiari, C. 2005. Timing of subduction and

exhumation along the Cambrian East Gondwana margin, and the formation of

Paleozoic backarc basins. Geological Society of America Bulletin, 117, 105–

116.

Franz, G., Lucassen, F., Kramer, W., et al. 2006. Crustal evolution at the

central Andean continental margin: a geochemical record of crustal growth,

recycling and destruction. In: Oncken, O., Chong, G., Franz, G., et al.

(eds) The Andes: Active Subduction Orogeny. Springer, Berlin, 45–64.

Fyfe, W.S. 1978. The evolution of the earth’s crust: Modern plate tectonics to

ancient hot spot tectonics? Chemical Geology, 23, 89–114.

Garrels, R.M. & Perry, E.A. 1974. Cycling of carbon, sulfur, and oxygen through

geologic time. In: Goldberg, E.D. (ed.) The Sea, Vol. 5. Wiley, New York,

303–336.

Goldstein, S.J. & Jacobsen, S.B. 1988. Nd and Sr isotopic systematics of river

water suspended material: implications for crustal evolution. Earth and

Planetary Science Letters, 87, 249–265.

Gray, C.M. 1984. An isotopic mixing model for the origin of granitic rocks in

southeastern Australia. Earth and Planetary Science Letters, 70, 47–60.

Gray, D.R. & Foster, D.A. 2004. Tectonic evolution of the Lachlan Orogen,

southeast Australia: historical review, data synthesis and modern perspectives.

Australian Journal of Earth Sciences, 51, 773–818.

Griffin, W.L., Wang, X., Jackson, S.E., Pearson, N.J., O’Reilly, S.Y., Xu, X.

& Zhou, X. 2002. Zircon chemistry and magma mixing, SE China: In-situ

analysis of Hf isotopes, Tonglu and Pingtan igneous complexes. Lithos, 61,

237–269.

Grigne, C. & Labrosse, S. 2001. Effects of continents on Earth cooling: thermal

blanketing and depletion in radioactive elements. Geophysical Research

Letters, 28, 2707–2710.

Gurnis, M. & Davies, G.F. 1985. Simple parametric models of crustal growth.

Journal of Geodynamics, 3, 105–135.

Gurnis, M. & Davies, G.F. 1986. Apparent episodic crustal growth arising from a

smoothly evolving mantle. Geology, 14, 396–399.

Harrison, T.M., Blichert-Toft, J., Muller, W., Albarede, F., Holden, P. &

Mojzsis, S.J. 2005. Heterogeneous Hadean Hafnium: Evidence of Continental

Crust at 4.4 to 4.5 Ga. Science, 310, 1947–1950.

Harrison, T.M., Schmitt, A.K., McCulloch, M.T. & Lovera, O.M. 2008. Early

(>4.5 Ga) formation of terrestrial crust; Lu–Hf, �18O, and Ti thermometry

results for Hadean zircons. Earth and Planetary Science Letters, 268, 476–

486.

Hawkesworth, C.J. & Kemp, A.I.S. 2006a. The differentiation and rates of

generation of the continental crust. Chemical Geology, 226, 134–143.

Hawkesworth, C.J. & Kemp, A.I.S. 2006b. Evolution of the continental crust.

Nature, 443, 811–817.

Hawkesworth, C.J., Kelley, S., Turner, S., le Roex, A. & Storey, B.C. 1999.

Mantle processes during Gondwana break-up and dispersal. Journal of

African Earth Sciences, 28, 239–261.

Hawkesworth, C.J., Cawood, P.A., Kemp, A.I.S., Storey, C. & Dhuime, B.

2009. A matter of preservation. Science, 323, 49–50.

Hofmann, A.W. 2008. The enduring lead paradox. Nature Geoscience, 1, 812–

813.

Hurley, P.M. & Rand, P.J. 1969. Pre-drift continental nuclei. Science, 164, 1229–

1242.

Jacobsen, S.B. 1988. Isotopic constraints on crustal growth and recycling. Earth

and Planetary Science Letters, 90, 315–329.

Jacobsen, S.B. & Wasserburg, G.J. 1979. The mean age of crustal and mantle

reservoirs. Journal of Geophysical Research, 84, 7411–7427.

Kamber, B.S., Whitehouse, M.J., Bolhar, R. & Moorbath, S. 2005. Volcanic

resurfacing and the early terrestrial crust: Zircon U–Pb and REE constraints

from the Isua Greenstone Belt, southern West Greenland. Earth and

Planetary Science Letters, 240, 276–290.

Kay, R.W. & Kay, S.M. 1991. Creation and destruction of lower continental crust.

Geologische Rundschau, 80, 259–278.

Keay, S., Collins, W.J. & McCulloch, M.T. 1997. A three component Sr–Nd

isotopic mixing model for granitoid genesis, Lachlan Fold Belt, eastern

Australia. Geology, 25, 307–310.

Kemp, A.I.S. & Hawkesworth, C.J. 2003. Granitic perspectives on the generation

and secular evolution of the continental crust. In: Rudnick, R.L. (ed.)

Treatise on Geochemistry, Vol. 3, The Crust. Elsevier, Amsterdam, 349–410.

Kemp, A.I.S., Hawkesworth, C.J., Paterson, B.A. & Kinny, P.D. 2006. Episodic

growth of the Gondwana supercontinent from hafnium and oxygen isotope

ratios. Nature, 439, 580–583.

Kemp, A.I.S., Hawkesworth, C.J., Foster, G.L., et al. 2007a. Magmatic and

crustal differentiation history of granitic rocks from hafnium–oxygen isotopes

in zircon. Science, 315, 980–983.

Kemp, A.I.S., Shimura, T. & Hawkesworth, C.J. 2007b. Linking granulites,

silicic magmatism, and crustal growth in arcs: Ion microprobe (zircon) U–Pb

ages from the Hidaka metamorphic belt, Japan. Geology, 35, 807–810.

Kemp, A.I.S., Foster, G.L., Schersten, A., Whitehouse, M.J., Darling, J. &

Storey, C. 2009a. Concurrent Pb–Hf isotope analysis of zircon by laser

ablation multi-collector ICP-MS, with implications for the crustal evolution

of Greenland and the Himalayas. Chemical Geology, 261, 244–260.

Kemp, A.I.S., Hawkesworth, C.J., Collins, W.J. , Gray, C.M., Blevin, P.L. &

Edinburgh Ion Microprobe Facility 2009b. Nd, Hf and O isotope evidence for

rapid continental growth during accretionary orogenesis in the Tasmanides,

eastern Australia. Earth and Planetary Science Letters, 284, 455–466.

Kesler, S.E. & Wilkinson, B.H. 2006. The role of exhumation in the temporal

distribution of ore deposits. Economic Geology, 101, 919–922.

King, E.W., Valley, J.W., Davis, D.W. & Edwards, G.R. 1998. Oxygen isotope

ratios of Archaean plutonic zircons from granite–greenstone belts of the

Superior Province: indicator of magmatic source. Precambrian Research, 92,

47–67.

King, S.D. 2005. Archean cratons and mantle dynamics. Earth and Planetary

Science Letters, 234, 1–14.

Korenaga, J. 2006. Archean geodynamics and the thermal evolution of Earth. In:

Benn, K., Mareschal, J.-C. & Condie, K. (eds) Archean Geodynamics and

Environments. Geophysical Monograph, American Geophysical Union, 164,

doi:10.1029/164GM03.

Kramers, J.D. & Tolstikhin, I.N. 1997. Two terrestrial lead isotope paradoxes,

forward transport modelling, core formation and the history of the continental

crust. Chemical Geology, 139, 75–110.

Kuno, H. 1968. Origin of andesite and its bearing on the island arc structure.

Bulletin of Volcanology, 32, 141–176.

Landenberger, B., Farrell, T.R., Offler, R., Collins, W.J. & Whitford, D.J.

1995. Tectonic implications of Rb–Sr biotite ages for the Hillgrove Plutonic

Suite, New England fold belt, N.S.W., Australia. Precambrian Research, 71,

251–263.

Lowe, D.R. & Tice, M.M. 2004. Geologic evidence for Archean atmospheric and

THE CONTINENTAL CRUST 247

Page 20: The Generation and Evolution of the Continental Crust

climatic evolution: Fluctuating levels of CO2, CH4, and O2, with an

overriding tectonic control. Geology, 32, 493–496.

McCulloch, M.T. & Bennett, V.C. 1994. Progressive growth of the Earth’s

continental crust and depleted mantle: geochemical constraints. Geochimica

et Cosmochimica Acta, 58, 4717–4738.

McCulloch, M.T. & Chappell, B.W. 1982. Nd isotopic characteristics of S-type

and I-type granites. Earth and Planetary Science Letters, 58, 51–64.

McCulloch, M.T. & Wasserburg, G.J. 1978. Sm–Nd and Rb–Sr chronology of

continental crust formation. Science, 200, 1003–1011.

Nagler, T.F. & Kramers, J.D. 1998. Nd isotopic evolution of the upper mantle

during the Precambrian: models, data and the uncertainty of both. Precam-

brian Research, 91, 233–252.

Nance, W.B. & Taylor, S.R. 1976. Rare earth element patterns and crustal

evolution—II. Archean sedimentary rocks from Kalgoorlie, Australia. Geo-

chimica et Cosmochimica Acta, 41, 225–231.

Nelson, D.R. 1998. Granite–greenstone crust formation on the Archaean Earth: a

consequence of two superimposed processes. Earth and Planetary Science

Letters, 158, 109–119.

O’Nions, R.K., Evensen, N.M. & Hamilton, P.J. 1980. Differentiation and

evolution of the mantle. Philosophical Transactions of the Royal Society of

London, Series A, 297, 479–493.

O’Nions, R.K., Hamilton, P.J. & Hooker, P.J. 1983. A Nd isotope investigation

of sediment related to crustal development in the British Isles. Earth and

Planetary Science Letters, 63, 229–240.

Peck, W.H., Valley, J.W. & Graham, C.M. 2003. Slow diffusion rates of O

isotopes in igneous zircons from metamorphic rocks. American Mineralogist,

88, 1003–1014.

Pietranik, A.B., Hawkesworth, C.J., Storey, C.D., Kemp, A.I.S., Sircombe,

K.N., Whitehouse, M.J. & Bleeker, W. 2008. Episodic, mafic crust

formation from 4.5 to 2.8 Ga: New evidence from detrital zircons, Slave

craton, Canada. Geology, 36, 875–878.

Pietranik, A.B., Hawkesworth, C.J., Storey, C. & Kemp, T. 2009. Depleted

mantle evolution and how it is recorded in zircon. Geochimica et Cosmochi-

mica Acta, 73, A1028.

Plank, T. 2005. Constraints from thorium/lanthanum on sediment recycling at

subduction zones and the evolution of continents. Journal of Petrology, 46,

921–944.

Plank, T. & Langmuir, C.H. 1998. The chemical composition of subducting

sediment and its consequences for the crust and mantle. Chemical Geology,

145, 325–394.

Reymer, A. & Schubert, G. 1984. Phanerozoic addition rates to the continental

crust and crustal growth. Tectonics, 3, 63–77.

Rino, S., Kon, Y., Sato, W., Maruyama, S., Santosh, M. & Zhao, D. 2008.

The Grenvillian and Pan-African orogens: World’s largest orogenies through

geologic time, and their implications on the origin of superplume. Gondwana

Research, 14, 51–72.

Rudnick, R.L. 1995. Making continental crust. Nature, 378, 573–578.

Rudnick, R.L. & Gao, S. 2003. Composition of the continental crust. In:

Rudnick, R.L. (ed.) Treatise on Geochemistry, Vol. 3, The Crust. Elsevier,

Amsterdam, 1–64.

Rudnick, R.L. & Goldstein, S.L. 1990. The Pb isotopic compositions of lower

crustal xenoliths and the evolution of lower crustal Pb. Earth and Planetary

Science Letters, 98, 192–207.

Scholl, D.W. & von Huene, R. 2007. Crustal recycling at modern subduction

zones applied to the past—issues of growth and preservation of continental

basement crust, mantle geochemistry, and supercontinent reconstruction. In:

Hatcher, R.D., Carlson, M.P., McBride, J.H. & Catalan, J.R.M. (eds) 4-

D Framework of Continental Crust. Geological Society of America, Memoirs,

200, 9–32.

Scholl, D.W. & von Huene, R. 2009. Implications of estimated magmatic

additions and recycling losses at the subduction zones of accretionary (non-

collisional) and collisional (suturing) orogens. In: Cawood, P.A. & Kroner,

A. (eds) Earth Accretionary Systems in Space and Time. Geological Society,

London, Special Publications, 318, 105–125.

Shirey, S.B., Kamber, B.S., Whitehouse, M.J., Mueller, P.A. & Basu, A.R.

2008. A review of the isotopic and trace element evidence for mantle and

crustal processes in the Hadean and Archean: Implications for the onset of

plate tectonic subduction. In: Condie, K.C. & Pease, V. (eds) When Did

Plate Tectonics Begin on Planet Earth? Geological Society of America,

Special Papers, 440, 1–29.

Smith, A.B. 2007. Intrinsic versus extrinsic biases in the fossil record: contrasting

the fossil record of echinoids in the Triassic and early Jurassic using sampling

data, phylogenetic analysis, and molecular clocks. Paleobiology, 33, 310–323.

Stein, M. & Hofmann, A.W. 1994. Mantle plumes and episodic crustal growth.

Nature, 372, 63–68.

Storey, B.C. 1995. The role of mantle plumes in continental break-up: case

histories from Gondwanaland. Nature, 377, 301–308.

Sun, S.-S. & McDonough, W.F. 1989. Chemical and isotopic systematics of

oceanic basalts: implications for mantle composition and processes. In:

Saunders, A.D. & Norry, M.J. (eds) Magmatism in the Ocean Basins.

Geological Society, London, Special Publications, 42, 313–345.

Taylor, S.R. 1967. The origin and growth of continents. Tectonophysics, 4,

17–34.

Taylor, S.R. & McLennan, S.M. 1985. The Continental Crust: Its Composition

and Evolution. Blackwell, Oxford.

Taylor, S.R. & McLennan, S.M. 1991. Sedimentary rocks and crustal evolution:

Tectonic setting and secular trends. Journal of Geology, 99, 1–21.

Taylor, S.R. & McLennan, S.M. 1995. The geochemical evolution of the

continental crust. Reviews of Geophysics, 33, 241–265.

Tolstikhin, I. & Kramers, J.D. 2008. The Evolution of Matter. Cambridge

University Press, New York.

Tolstikhin, I.N., Kramers, J.D. & Hofmann, A.W. 2006. A chemical Earth

model with whole mantle convection: The importance of a core–mantle

boundary layer (D’’) and its early formation. Chemical Geology, 226,

79–99.

Valley, J.W. 2003. Oxygen isotopes in zircon. In: Hanchar, J.M. & Hoskin,

P.W.O. (eds) Zircon. Mineralogical Society of America, Reviews in Miner-

alogy and Geochemistry, 53, 343–380.

Vervoort, J.D. & Blichert-Toft, J. 1999. Evolution of the depleted mantle: Hf

isotope evidence from juvenile rocks through time. Geochimica et Cosmochi-

mica Acta, 63, 533–566.

Vervoort, J.D., Patchett, P.J., Gehrels, G.E. & Nutman, A.P. 1996.

Constraints on early Earth differentiation from hafnium and neodymium

isotopes. Nature, 379, 624–627.

Vervoort, J.D., Patchett, P.J., Blichert-Toft, J. & Albarede, F. 1999.

Relationships between Lu–Hf and Sm–Nd isotopic systems in the global

sedimentary system. Earth and Planetary Science Letters, 168, 79–99.

Vlaar, N.J., van Keken, P.E. & van den Berg, A.P. 1994. Cooling of the Earth

in the Archaean: Consequences of pressure-release melting in a hotter mantle.

Earth and Planetary Science Letters, 121, 1–18.

Walter, M.J. 2003. Melt extraction and compositional variability in mantle

lithosphere. In: Carlson, R.W. (ed.) Treatise on Geochemistry, Vol. 2, The

Mantle and Core. Elsevier, Amsterdam, 363–394.

Walzer, U. & Hendel, R. 1997. Time-dependent thermal convection, mantle

differentiation and continental-crust growth. Geophysical Journal Interna-

tional, 130, 303–325.

Wang, C.Y., Campbell, I.H., Allen, C.M., Williams, I.S. & Eggins, S.M. 2009.

Rate of growth of the preserved North American continental crust: Evidence

from Hf and O isotopes in Mississippi detrital zircons. Geochimica et

Cosmochimica Acta, 73, 712–728.

White, A.J.R. & Chappell, B.W. 1977. Ultrametamorphism and granitoid genesis.

Tectonophysics, 43, 7–22.

Whitehouse, M.J., Kamber, B.S. & Moorbath, S. 1999. Age significance of U–

Th–Pb zircon data from early Archaean rocks of west Greenland—a

reassessment based on combined ion-microprobe and imaging studies.

Chemical Geology, 160, 201–224.

Wilde, S.A., Valley, J.W., Peck, W.H. & Graham, C.M. 2001. Evidence from

detrital zircons for the existence of continental crust and oceans on the Earth

4.4 Gyr ago. Nature, 409, 175–178.

Willner, A.P., Gerdes, A. & Massonne, H.-J. 2008. History of crustal

growth and recycling at the Pacific convergent margin of South America at

latitudes 298–368 S revealed by a U–Pb and Lu–Hf isotope study of detrital

zircon from late Paleozoic accretionary systems. Chemical Geology, 253,

114–129.

Zandt, G., Gilbert, H., Owens, T.J., Ducea, M., Saleeby, J. & Jones, C.H.

2004. Active foundering of a continental arc root beneath the southern Sierra

Nevada in California. Nature, 431, 41–46.

Zartman, R. 1990. Granulites and the lead paradox. Nature, 345, 204–205.

Zhang, Y. & Zindler, A. 1993. Distribution and evolution of carbon and nitrogen

in Earth. Earth and Planetary Science Letters, 117, 331–345.

Received 5 June 2009; revised typescript accepted 12 November 2009.

Scientific editing by Rob Strachan.

C. J. HAWKESWORTH ET AL .248


Recommended