Scholarly article on topic 'Refinement of the supercontinent cycle with Hf, Nd and Sr isotopes'

Refinement of the supercontinent cycle with Hf, Nd and Sr isotopes Academic research paper on "Biological sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Geoscience Frontiers
OECD Field of science
Keywords
{"Supercontinent cycle" / "Hf isotopes" / "Nd isotopes" / "Collisional orogens" / "Accretionary orogens"}

Abstract of research paper on Biological sciences, author of scientific article — Kent C. Condie, Richard C. Aster

Abstract The combined use of Hf, Nd and Sr isotopes is more useful in understanding the supercontinent cycle than the use of only Hf isotopic data from detrital zircons. Sr and Nd seawater isotopes, although not as precise as ɛ Nd and ɛ Hf distributions, also record input from ocean ridge systems. Unlike detrital zircons where sources cannot be precisely located because of crustal recycling, both the location and tectonic setting often can be constrained for whole-rock Nd isotopic data. Furthermore, primary zircon sources may not reside on the same continent as derivative detrital zircons due to supercontinent breakup and assembly. Common to all of the isotopic studies are geographic sampling biases reflecting outcrop distributions, river system sampling, or geologists, and these may be responsible for most of the decorrelation observed between isotopic systems. Distributions between 3.5 and 2 Ga based on ɛ Hf median values of four detrital zircon databases as well as our compiled ɛ Nd database are noisy but uniformly distributed in time, whereas data between 2 and 1 Ga data are more tightly clustered with smaller variations. Grouped age peaks suggest that both isotopic systems are sampling similar types of orogens. Only after 1 Ga and before 3.5 Ga do we see wide variations and significant disagreement between databases, which may partially reflect variations in both the number of sample locations and the number of samples per location. External and internal orogens show similar patterns in ɛ Nd and ɛ Hf with age suggesting that both juvenile and reworked crustal components are produced in both types of orogens with similar proportions. However, both types of orogens clearly produce more juvenile isotopic signatures in retreating mode than in advancing mode. Many secular changes in ɛ Hf and ɛ Nd distributions correlate with the supercontinent cycle. Although supercontinent breakup is correlated with short-lived decreasing ɛ Hf and ɛ Nd (≤100 Myr) for most supercontinents, there is no isotopic evidence for the breakup of the Paleoproterozoic supercontinent Nuna. Assembly of supercontinents by extroversion is recorded by decreasing ɛ Nd in granitoids and metasediments and decreasing ɛ Hf in zircons, attesting to the role of crustal reworking in external orogens in advancing mode. As expected, seawater Sr isotopes increase and seawater Nd isotopes decrease during supercontinent assembly by extroversion. Pangea is the only supercontinent that has a clear isotopic record of introversion assembly, during which median ɛ Nd and ɛ Hf rise rapidly for ≤100 Myr. Although expected to increase, radiogenic seawater Sr decreases (and seawater Nd increases) during assembly of Pangea, a feature that may be caused by juvenile input into the oceans from new ocean ridges and external orogens in retreating mode. The fact that a probable onset of plate tectonics around 3 Ga is not recorded in isotopic distributions may be due the existence of widespread felsic crust formed prior to the onset of plate tectonics in a stagnant lid tectonic regime, as supported by Nd and Hf model ages.

Academic research paper on topic "Refinement of the supercontinent cycle with Hf, Nd and Sr isotopes"

Contents lists available at SciVerse ScienceDirect China University of Geosciences (Beijing)

Geoscience Frontiers

journal homepage: www.elsevier.com/locate/gsf

Focus paper

Refinement of the supercontinent cycle with Hf, Nd and Sr isotopes

Kent C. Condie a *, Richard C. Astera b

CrossMark

a Department of Earth and Environmental Science, New Mexico Institute of Mining and Technology, Socorro, NM 87801, USA b Geosciences Department, Colorado State University, Fort Collins, CO 80523, USA

ARTICLE INFO

Article history: Received 29 April 2013 Received in revised form 31 May 2013 Accepted 2 June 2013 Available online 22 June 2013

Keywords:

Supercontinent cycle Hf isotopes Nd isotopes Collisional orogens Accretionary orogens

ABSTRACT

The combined use of Hf, Nd and Sr isotopes is more useful in understanding the supercontinent cycle than the use of only Hf isotopic data from detrital zircons. Sr and Nd seawater isotopes, although not as precise as £Nd and tHf distributions, also record input from ocean ridge systems. Unlike detrital zircons where sources cannot be precisely located because of crustal recycling, both the location and tectonic setting often can be constrained for whole-rock Nd isotopic data. Furthermore, primary zircon sources may not reside on the same continent as derivative detrital zircons due to supercontinent breakup and assembly. Common to all of the isotopic studies are geographic sampling biases reflecting outcrop distributions, river system sampling, or geologists, and these may be responsible for most of the decorrelation observed between isotopic systems. Distributions between 3.5 and 2 Ga based on tHf median values of four detrital zircon databases as well as our compiled tNd database are noisy but uniformly distributed in time, whereas data between 2 and 1 Ga data are more tightly clustered with smaller variations. Grouped age peaks suggest that both isotopic systems are sampling similar types of orogens. Only after 1 Ga and before 3.5 Ga do we see wide variations and significant disagreement between databases, which may partially reflect variations in both the number of sample locations and the number of samples per location.

External and internal orogens show similar patterns in tNd and tHf with age suggesting that both juvenile and reworked crustal components are produced in both types of orogens with similar proportions. However, both types of orogens clearly produce more juvenile isotopic signatures in retreating mode than in advancing mode. Many secular changes in tHf and tNd distributions correlate with the supercontinent cycle. Although supercontinent breakup is correlated with short-lived decreasing tHf and tNd (< 100 Myr) for most supercontinents, there is no isotopic evidence for the breakup of the Paleoproterozoic supercontinent Nuna. Assembly of supercontinents by extroversion is recorded by decreasing tNd in granitoids and metasediments and decreasing tHf in zircons, attesting to the role of crustal reworking in external orogens in advancing mode. As expected, seawater Sr isotopes increase and seawater Nd isotopes decrease during supercontinent assembly by extroversion. Pangea is the only supercontinent that has a clear isotopic record of introversion assembly, during which median tNdand tHf rise rapidly for <100 Myr. Although expected to increase, radiogenic seawater Sr decreases (and seawater Nd increases) during assembly of Pangea, a feature that may be caused by juvenile input into the oceans from new ocean ridges and external orogens in retreating mode. The fact that a probable onset of plate tectonics around 3 Ga is not recorded in isotopic distributions may be due the existence of widespread felsic crust formed prior to the onset of plate tectonics in a stagnant lid tectonic regime, as supported by Nd and Hf model ages.

© 2013, China University of Geosciences (Beijing) and Peking University. Production and hosting by

Elsevier B.V. All rights reserved.

* Corresponding author. Tel.: +1 505 822 8794. 1. Introduction

E-mail address: kcondie@nmt.edu (K.C. Condie).

Many studies in recent years have focused on the correlation and significance of zircon U/Pb age peaks with the supercontinent cycle and the growth of continental crust using Hf isotopes in detrital zircons. However, relatively few have used Nd and Sr isotopes in whole-rock samples to augment the Hf isotopic record (Wang et al., 2009; Belousova et al., 2010; Hawkesworth et al.,

Peer-review under responsibility of China University of Geosciences (Beijing)

1674-9871/$ — see front matter © 2013, China University of Geosciences (Beijing) and Peking University. Production and hosting by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.gsf.2013.06.001

2010; Condie et al., 2011). It has been shown that most juvenile crust preserved in orogens is produced in continental arcs (92%) with only small amounts produced in oceanic arcs and other oceanic tectonic settings (Condie, 2013; Condie and Kroner, 2013). Of the juvenile crust preserved in orogens, most was produced during the subduction stage (ocean basin closing) rather than during the collisional stage. Hence, the probability of preservation must be at least moderate during ocean-basin closing for this juvenile crust to survive long enough to be preserved during the collisional stage (Hawkesworth et al., 2009). What is not well constrained is what types of orogens detrital zircons come from, since most older zircons are recycled multiple times in the geologic record.

Whole-rock Nd isotope data have the advantage of not only having a known source location, but also in most cases, they are able to constrain the tectonic setting of the source. When used in conjunction with Hf isotope zircon data, these isotopes increase the potential for a more complete understanding how the production and preservation of juvenile crust is related to the supercontinent cycle. In this study we analyze and compare a combined Hf isotopic database from detrital zircons with whole-rock Nd isotopic data from detrital sedimentary rocks and granitoids. We also include Sr and Nd isotopic data from marine carbonates to track seawater composition with time. Using these combined databases, we propose refinements to the supercontinent cycle and discuss uncertainties that have not been fully addressed in previous studies. Also included is a discussion of the use of isotopic data to constrain

modes of assembly of supercontinents by extroversion and introversion and complications arising from overlap in breakup and assembly of supercontinents during the last 1000 Ma.

2. Controls of isotopic data

Many recent papers have focused on the types of orogens preserved in the geologic record and how these relate to the supercontinent cycle (Cawood et al., 2009; Collins et al., 2011; Lancaster et al., 2011; Stern, 2011; Roberts, 2012; Condie, 2013). Four end-member scenarios relating orogens to the supercontinent cycle are illustrated in Fig. 1. Each of these scenarios influences the Hf and Nd isotopic compositions of rocks produced and/or preserved in orogens (Hawkesworth et al., 2010; Collins et al., 2011; Condie et al., 2011 ). During supercontinent breakup, external accretionary orogens commonly shift into an advancing phase (a), resulting in more reworking of older crust by such processes as sediment subduction, delamination, and subduction erosion (Scholl and von Huene, 200 , 2009). This scenario is epitomized during in the last 100—200 Ma by the Andean orogen as the South Atlantic opened (Scholl and von Huene, 2007). In contrast, when an ocean basin closes (b), external orogens typically change into a retreating phase and the fraction of juvenile crust production increases (Collins et al., 2011). Internal orogens exhibit both reworked and juvenile crust production depending on such factors as terrane accretion and whether or not subduction zones are present on one or both margins of the closing ocean basin (Collins et al., 2011). During continent-continent

a OPENING OCEAN BASIN c COLLISION

Accretionary Orogens Collisional Orogens

External Advancing (Reworking) External (Mixed)

External Retreating (Juvenile) Internal (Reworking)

Internal Advancing (Reworking)

Internal Retreating (Juvenile)

b CLOSING OCEAN BASIN d SUPERCONTINENT

Figure 1. Map views showing four end-member scenarios for the relationship of internal zones (no dip direction implied).

and external orogens to the supercontinent cycle. Red barbed lines indicate subduction

collisions (c), crust is largely reworked in internal orogens, but as new external orogens are established, juvenile crustal contributions may again become important. During the supercontinent stasis stage (d), new external orogens form and may involve both reworked and juvenile crustal contributions.

In interpreting zircon ages and Hf isotopic data, the effects of zircon alteration must be considered. The d18O in zircons is sensitive to hydrous alteration and to water in the sites of magma generation. The d18O of altered zircons may depart from that of unaltered zircons with mantle isotopic signatures (Valley et al., 2005). An important question for our analysis is whether or not detrital zircons that come from igneous rocks derived from water-rich magma sources (or aqueously altered rocks) that show biases in £Hf. To check this we use published data that include d18O analyses of zircons. As examples, d18O vs £Hf plots of three suites of detrital zircons are shown in Figs. A1-A3 (prefix A refers to Appendix figures). If we assume a range for mantle d18O of 4.5-6.5%„, these plots show that zircons that fall outside of the mantle d18O window (lower or higher) have similar distributions in £Hf from those that plot within the mantle d18O window. In the Russian rivers database 43% of zircons with mantle d18O values have negative £Hf compared to 51% for non-mantle d18O sources (Wang et al., 2011); corresponding numbers for African rivers are 59% and 62% (Iizuka et al., 2012) (Figs. A1 and A2). In the much larger Dhuime et al. (2012) database, 53% with mantle d18O are negative and 65% of non-mantle d18O are negative. Although the proportion of negative epsilon values is somewhat enhanced due to the effects of water, negative d18O do not appear to introduce a significant bias in identifying juvenile and reworked crustal sources in large detrital zircon populations.

Sr and Nd isotopic data from marine carbonates record a balance between input into the oceans from erosion of continents (which largely samples reworked crust) and juvenile input chiefly from hydrothermal processes along ocean ridges (Keto and Jacobson, 1988; Prokoph et al., 2008; Peucker-Ehrenbrink, 2010). During supercontinent breakup, ocean-ridge control may dominate, and thus Sr and Nd isotopic signatures in marine carbonates should reflect more juvenile input into the oceans (lower 87Sr/86Sr and higher 143Nd/144Nd). The magnitude of this effect as well as the geographic distribution may vary due to different resident times of these elements in seawater. In general, however, rivers sampling external orogens may introduce a reworked isotopic signature (higher 87Sr/86Sr and lower 143Nd/144Nd) into the oceans, and thus seawater should reflect a balance between these two inputs. During supercontinent assembly the opposite trends may exist: loss of ocean ridges and increasing numbers of high mountains in collisional orogens should enhance isotopic signals of reworked crust sent to the oceans.

3. Comparison of databases

There are differences between published detrital zircon databases, many of which reflect different geographic sampling biases in the zircon populations. For instance, studies by Wang et al. (2011) and Iizuka et al. (2012) of detrital zircon populations for Siberian and African rivers show considerable variation in both age and Hf (t) (hereafter £Hf) distribution between single rivers. To compare published Hf databases we show median Hf for each of four databases plotted with time averaged over a 50 Myr bin moving 1 Myr per time increment (Fig. 2). Median values from our £Nd(t) database (hereafter £Nd) (Appendix 1) are shown for comparison, along with the distribution of major zircon age peaks (Fig. 2c). Fig. 2b shows some unanticipated similarities between the Hf curves and the Nd curve. Between 3.5 and 2 Ga the epsilon distributions are relatively flat (Iizuka et al., 2012 excepted). All four databases also show a

rather tight grouping between 2 and 1 Ga with only small amplitude variations. Only after 1 Ga and before 3.5 Ga do we note wide discrepancies in databases and in some instances, clear divergence. These variations often reflect time periods with either few data and/or a small number of sample locations. Within specific Hf databases, we note that the Dhuime et al. (2012) data show a large positive £Hf peak around 150 Ma whereas the Voice et al. (2011) data show a negative peak at this time; only GEMOC-Condie data show a large peak at 400 Ma. The Voice et al. (2011) data show a large negative peak around 3000 Ma and the Iizuka et al. (2012) data show large amplitude spikes into negative epsilon space near 2500 Ma. Most of these large deviations may arise from geographic sampling biases. Although zircon age peaks (Fig. 2c) also show considerably variation between the Hf databases, the total Hf database (S £Hf) and the global £Nd database show most of the previously recognized peaks (e.g., Condie and Aster, 2010; Condie et al., 2011).

In this investigation we examine median values of a combined Hf isotope database (with duplicate entries removed), together with median Nd from whole-rock sedimentary rocks and granitoids (Fig. 3; Appendix 1). Power spectra density analyses of the detrended zircon ages in both Hf and Nd databases show similar patterns, with a suggestion of a weak periodicity of about 250 Myr (Figs. A4 and A5). The interpretation of average or median epsilon curves with time has varied among various investigators. Although most agree that increases in epsilons with time reflect greater input of juvenile crust at convergent or/and collisional orogens, decreases in epsilon values have been interpreted as reworking of older continental crust or/and destruction of continental crust and recycling into the mantle (Belousova et al., 2010; Collins et al., 2011; Condie et al., 2011; Dhuime et al., 2012; Roberts, 2012). We suggest that decreasing epsilons with time reflects only reworking of older crust, since recycling of continental crust into the mantle should selectively remove rocks with specific zircon ages, and thus should appear on the epsilon-age spectra as "age gaps", and not a change in slope of the epsilon data (Condie, 2013). Fig. 4 shows the correlation coefficient (r) of £Hf with £Nd for the median values with time using a 150 Myr bin moving in 1 Myr increments. For the entire database, only 15% of these windows show r values of >0.7. This indicates that although there is overall correlative agreement in the Hf and Nd curves on timescales of hundreds of millions of years, this is not so for timescales of <150 Myr. The overall poor correlation on the 150 Myr time scale may arise from different sample sites for the populations. Consistently high positive correlations between Hf and £Nd occur between about 2100 and 1200 Ma. During this time interval, most the Nd and many of the Hf data come from sites in the Great Proterozoic Accretionary Orogen and carry similar isotopic signals (Condie, 2013).

To evaluate the variation in the combined Hf database and in the £Nd database, detrended variation of both data sets is shown in Figs. A6 and A7 as one standard deviation of the mean for a 50 Myr bin moving in 1 Myr increments. As expected, the variation in both epsilons increases with time as a result of growth of radiogenic isotopes in both the depleted mantle reservoir and in various continental reservoirs that have separated from depleted mantle over time (Fig. 5). However, beginning at about 1 Ga there is a rather large increase in the variation of the average standard deviation of both epsilons, and especially in Nd. This probably reflects a combination of decreases in both the number of samples per site and in the number of sample sites after 1000 Ma. Some sites have only a few samples, whereas others, such as the Arabian-Nubian shield, have large numbers of samples. A power spectral density analysis of the detrended epsilon variation curves shows no evidence of periodicity in either Hf or Nd with time (Figs. A6 and A7).

Age (Ma)

Figure 2. tHf for detrital zircons and tNd for whole-rock sedimentary rocks and granitoids. (a) tHf for all Hf databases; (b) Median values of tHf and tNd for each database for a 50 Myr bin, moving at 1 Myr increments; (c) U/Pb zircon age spectra for 50 Myr bins. Data sources: Comb, combined GEMOC and Condie-Aster database (Belousova et al., 2010; Condie and Aster, 2010); Dhuime et al. (2012); Voice et al. (2011), Iizuka et al. (2012); Nd and Hf databases given in Appendix 1.

4. Results

Fig. 5 shows the secular distribution of £Nd from sedimentary and granitic rocks from our Nd isotopic database color coded by orogen type (Appendix 1). On the whole, the tNd distributions and the major zircon age peaks are very similar to those from the Hf database we reported earlier (Condie and Aster, 2010; Condie et al., 2011). As with the Hf data, the major age peaks show both reworked and juvenile input into the continental crust with a sparsity of reworked data in the 1600—800 Ma time window. Unlike the Hf data discussed by Collins et al. (2011), we see no evidence of differences in the spread of the Nd data between collisional and accretionary phases of orogens, either in the last 500 Myr or at any other time. Both types of orogens have similar epsilon distribution patterns with accretionary orogens showing a slight preference for juvenile crust and collisional orogens for reworked crust (Fig. 5, inset). In the last 1000 Ma, both types of orogens show greater input from reworked continental crust, with negative epsilons (with values as low as -30) comprising about 60% of both the whole-rock Nd and zircon Hf databases. Before 1000 Ma, both databases show a slight preference for positive epsilons (61% £Nd; 54% £Hf).

The Hf and Nd median values are shown as a function of age in Fig. 3 (part b enlarged for the last 1000 Ma). The Sr and Nd isotopic curves for seawater are also shown, and although not as precise as the Hf and Nd data, they provide further insight into the ratio of juvenile to reworked crust with time (Shields, 2007; Peters and Gaines, 2012). The Sr isotope curve is the recalibrated seawater curve from Shields (2007). Also shown are assembly and breakup times of supercontinents, major zircon age peaks, and the distribution of major orogenic collisions from Appendix 2. Although there are differences between the Hf and Nd distributions as described above, the overall trends in median epsilons are similar prior to 1000 Ma. Compared to the £Hf curve before 2000 Ma, the Nd curve is offset to more juvenile compositions, and after 1000 Ma, the two curves often deviate and the amplitude of variation can be large. After 1000 Ma, about 60% of both £Hf and £Nd are negative, whereas before 1000 Ma only 40—50% are negative. Decoupling of Nd and Hf isotopic systems has been recognized in other studies, and is related to such factors as metasomatism, garnet fractionation, and selective elimination of radiation-damaged and altered zircons during weathering and erosion (Bizimis et al., 2004; Ionov et al., 2005; Yu et al., 2009). The consistently negative Hf signature before 2 Ga compared to Nd

Figure 3. £Hf for detrital zircons, tNd for whole-rock sediments and granitoids, and seawater Sr and Nd isotopic compositions with increasing age. tHf and tNd are median values of our combined Hf database and our Nd database (Appendix 1; Fig. 2b). Trends are displayed as the median value within a 50 Myr time window stepping in Myr increments. Sr and Nd seawater isotopic curves are from Shields (2007) and Keto and Jacobson (1988), respectively. Also shown are the major zircon peak ages (bold vertical arrows) (Condie and Aster, 2010), the supercontinent cycle, and major orogens (complete list given in Appendix 2) (black collisional, green accretionary). (a) Present to 4 Ga, and (b) expanded view of (a) from Present to 1000 Ma with tNd seawater included. "DM" indicates an approximation of the global tHf and tNd depleted mantle curve.

may reflect extensive recycling of zircons. The large amplitude variation prior to 3700 Ma and the negative peaks in Hf at about 3050 and 3400 Ma reflect few sample locations and few data. For instance, detrital zircons from a quartzite in the Limpopo orogen in Southern Africa (Zeh et al., 2008) are entirely responsible for the negative £Hf peak at 3050 Ma. The parallel changes in slope or peaks in both epsilon distributions at 2100,1950, 750, 550 and 150 Ma are striking and reflect widespread global changes in the ratio of juvenile to reworked crust produced and preserved at these times. For the most part, the major zircon age peaks (shown with vertical arrows) are not reflected in the secular epsilon variations. Exceptions are age peaks at 2100 Ma and also possibly at 100 Ma.

Many changes in the £Hf and tNd distributions with time correlate with changes in the supercontinent cycle (Fig. 3). The short-lived decrease in epsilons at 2.7—2.6 Ga, most apparent in Nd,

correlates with assembly of the first supercratons (Aspler and Chiarenzelli, 1998; Bleeker and Ernst, 2006). These isotopic changes may record reworking of older felsic crust as plate tectonics rapidly propagated around the globe in the late Archean (Condie and O'Neill, 2010) and, in particular, collisions leading to assembly of the supercratons may have increased the rate of recycling of older crust. Although the Sr seawater curve is not sensitive enough to show this short-lived change, it's gradual increase reflects an increasing amount of continental crust during the late Archean. The first and perhaps only event recorded by both £Hf and tNd and by a zircon age peak is at 2.1 Ga, where both epsilon curves drop dramatically. This appears to reflect a shift in subduction zones to the periphery of fragmenting Archean supercratons. If so, it would be the onset of new external accretionary orogens in advancing modes that led to more reworking of older crust (Fig. 1a). There are at least 10 external

Figure 4. Correlation coefficient (r) between tHf and £Nd median values shown as function of age for a 150 Myr bin moving a 1 Myr increments.

orogens that developed at this time: Usagaran-Tanzania, Ubendian, West Congo, Luizian, Venturari-Tapajos, Volhyn, NE Greenland, Angara, Wopmay, and Magondi-Kheis (Appendix 2). Reworking in four collisional orogens (Eburnean, Birimian-Transamazonian, Volga-Don, Limpopo), leading into the assembly of the next supercontinent Nuna, also may have contributed to the falling epsilons. The slightly climbing or flat 87Sr/86Sr seawater curve during this time records an approximate balance between juvenile and

reworked crustal signals into the oceans, the juvenile component probably coming from the opening new ocean basins as the Archean supercratons fragment.

Both £Hf and £Nd increase rapidly at 1950 Ma and then steadily but slowly increase thereafter to about 1500 Ma. The rapid increases at 1950 Ma could represent closing of at least five oceans leading to collisions that produced the Usagaran-Tanzania, Taltson, Genburgh, Thelon, and Lapland orogens (Appendix 2). Closing of these oceans

Figure 5. Distribution of £Nd(T) in whole-rock samples of sedimentary rocks and granitoids with age. Two growth lines of continental crust are shown (147Sm/143Nd = 0.107). Also shown are major U/Pb zircon age peaks as mean values with bars of one standard deviation (peaks at 150,560,1400,1700,1900,2100,2500,2700 Ma). DM, depleted mantle growth curve. Number of data points: 4089 accretionary orogens, 1070 collisional orogens. Data from Appendix 1.

may have changed external orogens into retreating modes (Fig. 1b) increasing the rate of juvenile crust production, thus increasing the epsilons at this time. The supercontinent Nuna was largely assembled between 1.9 and 1.8 Ga by numerous craton-craton collisions (Zhao et al., 2002), probably by extroversion if the reconstruction of the late Archean supercraton Kenorland by Bleeker and Ernst (2006) is approximately correct. During this time epsilons remain constant or slightly decrease, indicating a near balance between juvenile and reworked crustal production. The external orogens must have remained largely in retreat mode producing enough juvenile crust to partially offset the reworked input from the numerous collisional orogens. An alternative interpretation based on reliable paleomag-netic data is that Nuna was assembled from two or three large fragments between 1650 and 1580 Ma (Pisarevsky et al., 2013). Supportive of this interpretation is a drop in £Nd at this time (Fig. 3a), which could reflect enhanced recycling of continental crust during continental plate collisions. More than any other period of time, between 1.7 and 1.3 Ga (and extending to 1 Ga), the median £Hf and tNd curves are nearly coincident, recording a gradual increase in juvenile crustal contributions until around 1.5 Ga. The slightly falling seawater Sr isotope curve supports this interpretation. During this time juvenile crust produced in external orogens slightly dominated reworked crust produced largely in internal orogens (Fig. 1c,d). There are only three known collisional orogens with ages between 1.8 and 1.5 Ga (Yapungku, Kimban and Volhyn Central Russian). External orogens formed during this time include the very extensive Great Proterozoic Accretionary Orogen (GPAO; Condie, 2013), as well as the Nimrod-Ross, Racklan-Forward, Olarian and Kararan orogens (Appendix 2). Of the available Nd isotopic data, 82% come from the GPAO.

There is no evidence in Hf and Nd distributions nor in the seawater Sr isotope curve for the breakup of Nuna. Between 1.5 and 1.3 Ga, the epsilon curves are flat reflecting a balance between juvenile and reworked inputs into continental crust (Fig. 3a). There are only two collisions, Musgrave and Albany-Fraser near 1.3 Ga, in the probable time window of Nuna fragmentation. As previously suggested (Roberts, 2012), it is probable that the breakup of Nuna did not result in widespread dispersal of cratons, but only jostling around of the continental plates, and thus no strong signal comes through in the isotopic data for the breakup. Supporting this conclusion is the fact that many "craton groups" survived between Nuna and Rodinia, although their relative positions changed (Meert, 2002; Meert and Torsvik, 2003; Vakubchuk, 2010; Evans and Mitchell, 2011). Examples of craton groups are Baltica-Laurentia-Siberia, Australia-Antarctica-S China, West Africa-Amazonia-Rio de la Plata, and Congo-Tanzania. During assembly of Rodinia between 1.25 and 0.9 Ga, collision frequency increases (Fig. 3a), but unlike Nuna, the £Hf and Nd curves gradually fall, but remain tightly grouped (our combined Hf database [Appendix 1] does not show the negative peak a 1 Ga reported by Roberts [2012]). The change in slope of the epsilon curves beginning about 1.3 Ga may reflect a transfer of crustal production to external orogens in advancing mode (Fig. 1a) as Nuna fragmented, thus enhancing the reworked isotopic crustal signatures. The radiogenic Sr curve during the 1.25—0.9 Ga time period is flat (or slightly decreasing), indicating a close balance between juvenile and reworked input into the oceans. This appears to be controlled by a balance between ocean ridges and external orogens with exclusively or largely juvenile isotopic signatures, and internal orogens with largely reworked isotopic signatures. The increasing epsilons between 900 and 800 Ma (Fig. 3b) may reflect juvenile input from external orogens around the newly formed Rodinian supercontinent. The increasing Sr curve at this time, however, clearly does not have the same control, but may reflect erosion from newly formed mountains carrying a largely reworked isotopic signature.

The breakup of Rodinia and growth of the next supercontinent Gondwana occur between about 800 and 550 Ma (Fig. 3b). The peak

in Hf and Nd at 750 Ma may not be global because it is controlled almost entirely by data from the Arabian-Nubian shield (ANS) (Stern and Johnson, 2010), which is an external accretionary orogen, largely in retreat mode at this time. If Rodinia began to breakup before 750 Ma (Meert and Torsvik, 2003; Li et al., 2008), the ANS data may completely mask the isotopic evidence for this. If we ignore the epsilon peak at 750 Ma, which may be of only local significance, the breakup of Rodinia between 800 and 650 Ma shows decreasing epsilon distributions and increasing seawater Sr isotopes. This is consistent with formation of new internal oceans and a shift of crustal production to advancing external orogens, where reworked components exceed juvenile components. Five external orogens were active during this time (ANS, Lhasa, West Africa, Yenesei, Taimyr), whereas only one craton-craton collision is recorded by the Jiangnan orogen in South China (Appendix 2). This scenario is consistent with Rodinia fragmenting by extroversion as previously suggested (Murphy et al., 2009; Roberts, 2012). Unlike the seawater Sr isotope curve, the seawater Nd isotope curve is flat during this time indicating a balance between reworked and juvenile inputs.

Beginning around 650 Ma, £Hf and £Nd decrease rapidly, reflecting the growth of Gondwana-Pannotia (650—550 Ma). Both Sr and Nd seawater curves also steepen at this time, thus all of the isotopic systems show significant input of reworked older crust accompanying the assembly of Gondwana-Pannotia. As with the breakup of Rodinia, the breakup of Pannotia at 560—530 (Dong et al., 2011) (not at 450—400 Ma as suggested by Roberts [2012]) continued the reworking of older crust in advancing external orogens, and thus epsilons continued to shift to more negative values (Fig. 3b). Terra Australis external orogen was established by 580 Ma and continued until at least 400 Ma, mostly in advancing mode until Gondwana was fully formed (Cawood et al., 2009; Murphy et al., 2011). It is likely that collisional orogens also contributed to the reworked isotopic signatures, since most of the "anastomizing" Gondwana orogens in Africa and South America contain large volumes of reworked Precambrian crust. At 550—500 Ma, Hf and Nd both reached minimal values corresponding to the completion of Gondwana. We also see peaks in seawater 87Sr/86Sr and in seawater Nd at about 480 Ma. Opening of the Iapetus Ocean at 560—530 Ma (Schotese, 2009) (with juvenile input to seawater at the new ocean ridges) apparently did not affect the reworked crustal Sr and Nd isotopic signals transferred into seawater. All four isotopic indices consistently record the assembly of Gondwana-Pannotia from 650 to 500 Ma and the peaks near 480—500 Ma reflect the greatest contribution of reworked crust for any supercontinent assembly. Although it appears that Gondwana-Pannotia also formed by extroversion (Schotese, 2009; Murphy et al., 2011), it differed from the assembly of Rodinia in that it formed more rapidly (150 Myr compared to 350 Myr for Rodinia), involved a greater contribution of reworked older crust, and fragmented as two supercontinents (Pannotia and Gondwana).

Between 500 and 450 Ma following the breakup of Pannotia, subduction shifted to external orogens and both £Hf and £Nd increased (Fig. 3b). This change probably records continual closure of internal oceans, which enhanced juvenile crustal production in external orogens in retreating mode (Fig. 1b, d). Seawater Sr and £Nd both reached peak values at this time showing a balance between ocean ridge and reworked isotopic signals reaching the oceans. Although the most volumetrically important external orogen was Terra Aus-tralis, the Antler, Ellesmerian, and Verkhoyansk-Kolyma accretionary orogens also contributed juvenile crust during this time.

Pangea assembled between 400 Ma and 250 Ma by introversion accompanying the closing of internal oceans (Murphy and Nance, 2003). Growth of Pangea begins with the Acadian and Scandian collisions at 420—390 Ma continuing until about 250 Ma, with the final collision of Africa and Laurentia during the Alleghanian-Ouachita orogenies (370—230 Ma). The steep climb in Hf and Nd

and in seawater Nd and the fall in seawater Sr at about 400 Ma during the Acadian collisions (420-380 Ma) may reflect significant input of juvenile crust in external orogens, which were now largely in retreating mode. The fact that both £Hf and £Nd are rather flat between 400 and 200 Ma indicates a balance between juvenile and reworked crustal input during this time, which may be a characteristic of supercontinent assembly by introversion. Subduction around the closing internal oceans may have involved a significant reworked component, enough to offset the juvenile crust input in external orogens. The decrease in seawater 87Sr/86Sr and irregular increase in seawater Nd between 480 and 200 Ma is puzzling since these ratios should change in the opposite directions during supercontinent assembly. Perhaps an increasing number of ocean ridges in external oceans and new external orogens in retreat mode introduced enough juvenile component into the oceans to offset the reworked component coming from collisional orogens. If real, the significance of Nd seawater peaks about 400 and 250 Ma is not understood, but could represent incomplete mixing of Nd in the oceans at these times.

After 200 Ma we see significant differences between Hf and Nd curves, probably resulting from relatively few data and geographic sampling biases. This is the time that Pangea fragmented (Murphy and Nance, 2003). Most of the detrital zircons are from Mississippi River sediments and probably come from the Cordilleran external orogen (Wang et al., 2009; Iizuka et al., 2010). Nd data come from widespread geographic locations, all from external orogens (such as Japan, the Cordillera in western North America, the Andes, and South China). The negative spike in £Nd and £Hf and in seawater 87Sr/86Sr around 150 Ma corresponds to the beginning of breakup of Pangea as the Atlantic Ocean opens. The opening of the Atlantic may have caused external orogens along the west coasts of the Americas to change to advancing modes, producing more reworked crust. The minimum in seawater 87Sr/86Sr at this time could reflect new juvenile input into the oceans as the Mid-Atlantic ridge grew. The positive spike in both £Nd and £Hf at about 100 Ma may reflect continued closing of the Tethys and shift of subduction to the Terra Australis external orogen, which by now is largely in retreat mode. Although breakup of Pangea continues until at least 30 Ma, the isotopic signatures after 100 Ma are complicated by collisions leading into the assembly of a new supercontinent Amasia. The "overall fall" in £Nd and £Hf beginning just after 100 Ma reflects the continued breakup of Pangea with a shift of crustal production to external orogens where reworked crustal components dominate. The negative Nd peak at 50 Ma is due entirely to sediments from Taiwan (Chen et al., 1990) and probably is not a global signature. The rapid increase in seawater 87Sr/86Sr in the last 50 Ma is caused chiefly by input from rapidly rising mountains in the Himalayan-Alpine orogen (Peucker-Ehrenbrink, 2010; Peters and Gaines, 2012).

5. Discussion

One of the chief problems in using detrital zircons to identify orogen types is that most or all zircons liberated during weathering are recycled by sedimentary processes. Some studies compare heights of detrital zircon age peaks to areas of craton of known age exposed on a specific continent which is being sampled by a modern river (Iizuka et al., 2005; Wang et al., 2009). However, most detrital zircons are not derived directly from their original sources, but from later sediments that may have been recycled more than once before entering a modern river system. Hence, zircon peak heights are controlled by the distribution and preservation of detrital sedimentary rocks and not by the geographic area of primary sources. Further complicating this issue is the supercontinent cycle. Primary zircon sources may not even reside on the same continent as the detrital zircons sampled by modern rivers, because past supercontinent breakup has dispersed these sources (Meinhold et al., 2013).

We suggest that this is a major problem with the dual orogen model proposed by Collins et al. (2011) based on Hf isotope distributions in detrital zircons. The contrasting patterns in Hf in external and internal orogens (Collins et al. [2011 ; Fig. 2]) depend critically on the locations and tectonic settings of the original sources of the detrital zircons, neither of which is well constrained. Although Hf distribution in detrital zircons reflects a balance between external and internal orogen sources, it also depends on whether external orogens are in retreating or advancing modes (Fig. 1 ). Our data suggest that reversals between retreating and advancing modes in external orogens are not short-term reversals (<50 Myr) as suggested by Collins et al. (2011). Data from the Great Proterozoic (1.9—1.5 Ga) and Andean (<300 Ma) external orogens indicate these reversals can last several hundred million years. During opening of internal oceans, external orogens are dominantly in advancing mode where production of reworked crust with negative epsilons dominates (Fig. 1 ). Detrital zircons from the Andes with ages of200—500 Ma exhibit £Hf values from +5 to -10, with most falling in the range of -2 to -10 (Bahlburg et al., 2009; Dahlquist et al., 2013; Herve et al., 2013) clearly showing the importance of reworked crust during this 300 Myr time window. And finally, the statement by Collins et al. (2011, p. 336) that negative whole-rock Nd events in external orogens are short-lived (<50 Myr) is not supported by our Nd database (Appendix 1). £Nd from continental arcs shows an overwhelming proportion of reworked crust, and both external and internal orogens show similar "spreading patterns" in the last 500 Ma (Fig. 5). Whether or not juvenile or reworked crustal production predominates in external orogens depends critically on if they are in retreating or advancing modes.

Still another problem in using Hf in detrital zircons to identify primary sources is the commonly made assumption that the ages of detrital zircons in modern river sediments are biased by young zircons from high-mountain sources (Iizuka et al., 2010). This assumes that relatively young igneous rocks predominate in high mountains. However, some of the highest mountains in North America (in Wyoming and Colorado) expose Proterozoic or Archean basement, which can feed zircons directly into modern river systems such as the Mississippi. In this respect, whole-rock Nd isotopic data have an advantage in being able to identify source locations, as well as to constrain ancient tectonic settings of these sources.

If there was little if any continental crust before plate tectonics began about 3 Ga (Condie and Kroner, 2008 ; Van Huenen and Moyen, 2012), there should be few data points that plot in negative epsilon space before this time. Yet there are many negative epsilons (both £Nd and Hf), some of which project back to CHUR or depleted mantle (DM) intersections of 4 Ga or more (Fig. 5) ( Condie et al., 2011 ). There are two possible explanations for this observation: (1 ) plate tectonics operated on planet Earth beginning at least 4 Ga, or (2) significant volumes of felsic crust were produced before the onset of plate tectonics, perhaps in a stagnant lid tectonic regime (Piper, 2013). Since the first really strong evidence for widespread plate tectonics appears in the late Archean (Condie and Kroner, 2008), the second explanation needs to be seriously considered. Perhaps during a stagnant lid tectonic regime, which probably dominated before 3 Ga, partial melting of mafic (eclogitic?) roots of a thick crust resulted in widespread, but not voluminous, TTG magma production. And because very little crust older than 3 Ga has survived, this requires extensive recycling into the mantle as a result of later plate tectonics.

Supercontinents forming by extroversion and introversion have rather different signals in Nd and Hf distributions (Murphy and Nance, 2003; Collins et al., 2011; Roberts, 2012). The extroversion types show rapid drops in epsilon during their assembly, which lasts the order 100 Myr (this includes a possible new supercontinent Amasia) (Fig. 3). This decrease in epsilons begins during supercontinent breakup as shown by the breakup of the late

Archean supercratons (2.1 Ga), Rodinia (800-650 Ma), and Pan-notia (550-500 Ma). The extroversion breakup-assembly pattern clearly reflects extensive reworking in external orogens as they converge and collide with each other. Although some investigators have suggested that Nuna formed by introversion (Roberts, 2012), there is a decrease in tNd and £Hf between 1.9 and 1.8 Ga (Fig. 3a), when most of the Nunian collisions occurred, consistent with an extroversion origin. The reconstruction of the late Archean supercontinent Kenorland proposed by Bleeker and Ernst (2006) also supports an assembly of Nuna by extroversion. The only supercontinent that clearly formed by introversion is Pangea (Murphy et al., 2009) and both epsilon curves show a rapid increase during its assembly (Fig. 3b). This is consistent with external orogens shifting to retreating modes (as internal oceans close), where a greater proportion of juvenile crust is added to the continents. The breakup of Pangea is accompanied by a sharp decrease in both epsilon curves, followed by an abrupt increase. This is more difficult to interpret because collisions in the Tethys Ocean (leading to the onset of assembly of Amasia) overlapped with continuing breakup of Pangea. If this interpretation of £Nd and £Hf during supercontinent evolution is correct, it implies that external orogens may have a significant control on detrital zircon populations. Roberts (2012) reached a similar conclusion based solely on Hf isotope data.

6. Conclusions and future outlooks

From our analysis of Hf, Nd and Sr isotopic data in ancient minerals and rocks, it is clear that these isotopes reflect balances between juvenile and reworked crustal inputs, but that these balances are not always the same for the three isotopic systems. Sr and Nd isotopic records in marine carbonates are sensitive to changes in juvenile and reworked inputs into the oceans, and unlike the crustal Nd and Hf isotopic records, they also record juvenile input from ocean ridges. It is not usually possible from detrital zircon data alone to identify the primary source or the tectonic setting from which the zircons were ultimately derived because of zircon recycling and supercontinent breakup. This is where Nd isotopes from whole-rock samples become important: not only do we know the location of the samples (in most instances), but often we can constrain the tectonic setting of the rocks. Common to all of the isotopic studies are geographic sampling biases, either by geologists or by river systems, and this appears to be responsible for most of the decoupling observed between isotopic systems. In the time period from 2 to 1 Ga, £Nd and £Hf show remarkable agreement and the median values exhibit relatively small amplitude changes. This suggests that both isotopic systems are dominantly sampling the same or similar types of orogens. Prior to 2 Ga, Nd is consistently more juvenile than Hf, probably as a result of sampling of orogens with different ratios of juvenile to reworked crust, and perhaps also due to recycling of older zircons. There is no isotopic signal for the possible onset of plate tectonics around 3 Ga, a feature that may result from widespread, but not voluminous felsic crust formed prior to the onset of plate tectonics.

Our study suggests that the combined application of Hf, Nd and Sr isotopes is more productive in understanding the supercontinent cycle than the use of only Hf isotopic data from detrital zircons. Results from all three isotopic systems are consistent with an extroversion assembly of all supercontinents except Pangea. Supercontinent breakup is characterized by a short-lived (<100 Myr) decrease in epsilon values, which continue to decrease as the next supercontinent forms by extroversion. However, there is no evidence in Nd, Hf or Sr isotope records for the breakup of Nuna. Supercontinent assembly by introversion is characterized by a short-lived increase in both Nd and Hf. Because of overlapping breakup and assembly phases in the last 1000 Ma, the interpretation of the isotopic data is

complex and often ambiguous. Although many papers have appeared in the last few years on the significance of zircon age peaks and the supercontinent cycle, we still have many outstanding questions regarding the nature of these peaks and how they are related to both production and preservation of continental crust. If the interpretations set forth in this and other recent papers are on the right track (Hawkesworth et al., 2010; Collins et al., 2011; Condie, 2013), it looks like external orogens often control both £Nd and £Hf distributions when considered on a global scale. Why is this the case, and are there times when internal orogens (accretionary or collsional) control isotopic distributions in the crust? If so, how do we recognize them and of what significance are they to the supercontinent cycle? Also, reliable paleomagnetic data now suggests that Nuna did not assemble until 1650-1580 Ma. If so, what is the significance of all the collisional orogens between 1900 and 1800 Ma (Appendix 2)? Another intriguing question is why Nuna appears to be the only supercontinent with a long-lived (the order of 1000 Myr) accretionary orogen (the Great Proterozoic Accretionary Orogen) leading into assembly of the next supercontinent? The answer to this question may also be relevant to the question of why we do not see isotopic evidence for the breakup of Nuna. These questions are also related to what determines whether a supercontinent forms by extroversion or introversion, and why is the introversion case so rare? And finally, we have known for many years that the supercontinent cycle seems to be speeding up with time (Condie, 2002; Korenaga, 2006), with breakup and assembly phases significantly overlapping after 1 Ga. Could this be due to recycling of seawater into the mantle, lowering mantle viscosity as suggested by Korenaga (2006, 2011), and causing the rate of the convection and plate speed to increase with time?

Appendix A. Supplementary data

Supplementary data related to this article can be found online at http://dx.doi.org/10.1016/j.gsf.2013.06.001.

Appendix figures

-40 -30 -20 -10 0 10 20

Figure A1. Distribution of tHf (£Hf(t)) as a function of d18O in detrital zircons from Russian rivers (Wang et al., 2011). Shown is the range of d18O for mantle sources (4.5—6.5&, Valley et al., 2005).

Figure A2. Distribution of tHf (fHf(t)) as a function of d O in detrital zircons from African rivers (Iizuka et al., 2012). Shown is the range of d O for mantle sources (4.5-6.5&, Valley et al., 2005).

Figure A3. Distribution of tHf (tf)) as a function of in detrital zircons from Dhuime et al. (2012). Shown is the range of d O for mantle sources (4.5-6.5&, Valley et al., 2005).

Figure A4. Power spectral density of demeaned and detrended zircon ages for the complete Hf isotopic database (Fig. 2c). Dashed contours indicate estimates 1 a confidence limits. Power spectral estimation was performed using the adaptively weighted prolate spheroidal multitaper method of Thomson (1982), implemented using the MATLAB pmtm function with a time bandwidth product of 4.

Figure A5. Power spectral density of demeaned and detrended zircon ages for the Nd isotopic database (Fig. 2c) calculated as in Fig. A4.

Figure A6. Detrended tHf variation (1a variation within 50 Myr moving windows with a step of 1 Myr) and associated power spectrum calculated as in Fig. A4.

Figure A7. Detrended tNd variation (1a variation within 50 Myr moving windows with a step of 1 Myr) and associated power spectrum calculated as in Fig. A4.

References

Aspler, L.B., Chiarenzelli, J.R., 1998. Protracted breakup of Kenorland, a Neoarchean supercontinent? Geochronologic, tectonostratigraphic and sedimentologic evidence from the Paleoproterozoic. Sedimentary Geology 120, 75—104.

Bahlburg, H., Vervoort, J.D., Frame, S.A., Du, Bock, B., Augustsson, C., Reimann, C., 2009. Timing of crust formation and recycling in accretionary orogens: insights learned from the western margin of South America. Earth-Science Reviews 97,215—241.

Belousova, E.A., Kostitsyn, Y.A., Griffin, W.L., Begg, G.C., O'Reilly, S.Y., Pearson, N.J., 2010. The growth of the continental crust: constraints from zircon Hf-isotope data. Lithos 119, 457—466.

Bizimis, M., Sen, G., Salters, V.J.M., 2004. Hf—Nd isotope decoupling in the oceanic lithosphere: constraints from spinel peridotites from Oahu, Hawaii. Earth and Planetary Science Letters 217, 43—58.

Bleeker, W., Ernst, R.E., 2006. Short-lived mantle generated magmatic events and their dyke swarms: the key unlocking earth's paleogeographic record back to 2.6 Ga. In: Hanski, E., Mertanen, S., Rämö, T., Vuollo, J. (Eds.), Dyke Swarms — Time Markers of Crustal Evolution. A.A. Balkema, Rotterdam, pp. 3—26.

Cawood, P.A., Kröner, A., Collins, W.J., Kusky, T.M., Mooney, W.D., Windley, B.F., 2009. Accretionary orogens through Earth history. Geological Society of London, Special Publication 318,1—36.

Chen, C.-H., Jahn, B.-M., Lee, T., Chen, C.-H., Cornichet, J., 1990. Sm/Nd isotopic geochemistry of sediments from Taiwan and implications for the tectonic evolution of southeast China. Chemical Geology 88, 317—332.

Collins, W.J., Belousova, E.A., Kemp, A.I.S., Murphy, J.B., 2011. Two contrasting Phanerozoic orogenic systems revealed by hafnium isotope data. Nature Geo-science 4, 333—337.

Condie, K.C., 2002. The supercontinent cycle: are there two patterns of cyclicity? Journal of African Earth Sciences 35, 179—183.

Condie, K.C., 2013. Preservation and recycling of crust during accretionary and collisional phases of Proterozoic orogens: a bumpy road from Nuna to Rodinia. Geosciences 2013 (3), 240—26 .

Condie, K.C., Kroner, A., 2008. When did plate tectonics begin? Evidence from the geologic record. Geological Society of America Special Paper 440, 281 —294.

Condie, K.C., Kroner, A., 2013. The building blocks of continental crust: evidence for a major change in the tectonic setting of continental growth at the end of the Archean. Gondwana Research 23, 394—402.

Condie, K.C., Aster, R.C., 2010. Episodic zircon age spectra of orogenic granitoids: the supercontinent connection and continental growth. Precambrian Research 180, 227—236. http://dx.doi.org/10.1016/j.precamres.2010.03.008.

Condie, K.C., O'Neill, C., 2010. The Archean-Proterozoic boundary: 500 My of tectonic transition in Earth history. American Journal of Science 310, 775—790.

Condie, K.C., Bickford, M.E., Aster, R.C., Belousova, E., Scholl, D.W., 2011. Episodic zircon ages, Hf isotopic composition, and the preservation rate of continental crust. Geological Society of America Bulletin 123, 951—957.

Dahlquist, J.A., Pankhurst, R.J., Gaschnig, R.M., Rapela, C.W., Casquet, C., Alasino, P.H., Galindo, C., Baldo, E.G., 2013. Hf and Nd isotopes in Early Ordovician to Early Carboniferous granites as monitors of crustal growth in the Proto-Andean margin of Gondwana. Gondwana Research 23,1617—1630.

Dhuime, B., Hawkesworth, C.J., Cawood, P.A., Storey, C.D., 2012. A change in the geodynamics of continental growth 3 billion years ago. Science 335,1334—1336. http://dx.doi.org/10.1126/science.1216066.

Dong, X., Zhang, Z., Santosh, M., Wang, W., Yu, F., Liu, F., 2011. Late Neoproterozoic thermal events in the northern Lhasa terrane, south Tibet: zircon chronology and tectonic implications. Journal of Geodynamics 52, 389—405.

Evans, D.A.D., Mitchell, R.N., 2011. Assembly and breakup of the core of Paleo-Mesoproterozoic supercontinent Nuna. Geology 39, 443—446.

Hawkesworth, C., Cawood, P., Kemp, T., Storey, C., Dhuime, B., 2009. A matter of preservation. Science 323, 49—50.

Hawkesworth, C., Dhuime, B., Pietranik, A., Cawood, P., Kemp, T., Storey, C., 2010. The generation and evolution of the continental crust. Journal of the Geological Society of London 167, 229—248. http://dx.doi.org/10.1144/0016-76492009-072.

Herve, F., Calderon, M., Fanning, C.M., Pankhurst, R.J., Godoy, E., 2013. Provenance variations in the late Paleozoic accretionary complex of central Chile as indicated by detrital zircons. Gondwana Research 23,1122—1135.

Iizuka, T., Hirata, T., Komiya, T., Rino, S., Katayama, I., Motoki, I., Maruyama, S., 2005. U-Pb and Lu-Hf isotope systematics of zircons form the Mississippi River sand: implications for reworking and growth of continental crust. Geology 33, 485—488.

Iizuka, T., Komiya, T., Rino, S., Maruyama, S., Hirata, T., 2010. Detrital zircon evidence for Hf isotopic evolution of granitoid crust and continental growth. Geochimica et Cosmochimica Acta 74, 2450—2472.

Iizuka, T., Campbell, I.H., Allen, C.M., Gill, J.B., Maruyama, S., Makoka, F., 2012. Evolution of the African continental crust as recorded by U-Pb, Lu-Hf and O isotopes in detrital zircons from modern rivers. Geochimica et Cosmochimica Acta. http://dx.doi.org/10.1016Zj.gca.2012.12.028.

Ionov, D.A., Blichert-Toft, J., Weis, D., 2005. Hf isotope compositions and HREE variations in off-craton garnet and spinel peridotite xenoliths from central Asia. Geochimica et Cosmochimica Acta 69, 2399—2418.

Keto, L.S., Jacobson, S.B., 1988. Nd isotopic variations of Phanerozoic palaeoceans. Earth and Planetary Science Letters 90, 395—410.

Korenaga, J., 2006. Archean geodynamics and the thermal evolution of Earth. American Geophysical Union Monograph 164, 7—32.

Korenaga, J., 2011. Thermal evolution with a hydrating mantle and the initiation of plate tectonics in the early Earth. Journal of Geophysical Research 116, B12403.

Lancaster, P.J., Storey, C.D., Hawkesworth, C.J., Dhuime, B., 2011. Understanding the roles of crustal growth and preservation in the detrital zircon record. Earth and Planetary Science Letters 305, 405-412.

Li, Z.X., Bogdanova, S.V., Collins, A.S., Davidson, A., De Waele, B., Ernst, R.E., Fitzsimons, I.C.W., Fuck, R.A., Gladkochub, D.P., Jacobs, J., Karlstrom, K.E., Lu, S., Natapov, L.M., Pease, V., Pisarevsky, S.A., Thrane, K., Vernikovsky, V., 2008. Assembly, configuration, and break-up history of Rodinia: a synthesis. Precambrian Research 160,179-210. http://dx.doi.org/10.1016/j.precamres.2007.04.021.

Meert, J.G., 2002. Paleomagnetic evidence for a Paleo-Mesoproterozoic supercontinent Columbia. Gondwana Research 5, 201-215.

Meert, J.G., Torsvik, T.H., 2003. The making and unmaking of a supercontinent: rodinia revisited. Tectonophysics 375, 261-288.

Meinhold, G., Morton, A.C., Avigad, D., 2013. New insights into peri-Gondwana paleogeography and Gondwana super-fan system from detrital zircon U-Pb ages. Gondwana Research 23, 661 -665.

Murphy, J.B., Nance, R.D., 2003. Do supercontinents introvert or extrovert?: Sm-Nd isotopic evidence. Geology 31, 873-876.

Murphy, J.B., Nance, R.D., Gutiérrez-Alonso, G., Keppie, J.D., 2009. Supercontinent reconstruction from recognition of leading continental edges. Geology 37,595-598.

Murphy, J.B., van Staal, C.R., Collins, W.J., 2011. A comparison of the evolution of arc complexes in Paleozoic interior and peripheral orogens: speculations on geo-dynamic correlations. Gondwana Research 19, 812-827.

Peters, S.E., Gaines, R.R., 2012. Formation of the 'Great Unconformity' as a trigger for the Cambrian explosion. Nature 484, 363-366.

Peucker-Ehrenbrink, B., 2010. Continental bedrock and riverine fluxes of strontium and neodymium isotopes to the oceans. Geochemistry Geophysics Geosystems 11. http://dx.doi.org/10.1029/2009GC002869.

Piper, J.D.A., 2013. A planetary perspective on Earth evolution: lid tectonics before plate tectonics. Tectonophysics 589, 44-56.

Pisarevsky, S.A., Elming, S.-A., Pesonen, L.J., Li, Z.-X., 2013. Mesoproterozoic paleo-geography: supercontinent and beyond. Precambrian Research (in press).

Prokoph, A., Shields, G.A., Veizer, J., 2008. Compilation and time-series analysis of a marine carbonate 518O, S13C, 87Sr/86Sr and S34S database through Earth history. Earth-Science Reviews 87, 113-133. http://dx.doi.org/10.1016/j.earscirev. 2007.12.003.

Roberts, N.M.W., 2012. Increased loss of continental crust during supercontinent amalgamation. Gondwana Research 21, 994-1000. http://dx.doi.org/10.1016/ j.gr.2011 .08.00 .

Scholl, D.W., von Huene, R., 2007. Crustal recycling at modern subduction zones applied to the past e issues of growth and preservation of continental basement crust, mantle geochemistry, and supercontinent reconstruction. Geological Society of America Memoir 200, 9-32.

Schotese, C.R., 2009. Late Proterozoic plate tectonics and paleogeography: a tale of two supercontinents, Rodinia and Pannotia. Geological Society of London, Special Publication 326, 67-83.

Shields, G.A., 2007. A normalised seawater strontium isotope curve: possible implications for Neoproterozoic-Cambrian weathering rates and the further oxygenation of the Earth. eEarth 2, 35-42.

Scholl, D.W., von Huene, R., 2009. Implications of estimated magmatic additions and recycling losses at the subduction zones of accretionary and collisional orogens. Geological Society of London, Special Publication 318,105-125.

Stern, C.R., 2011. Subduction erosion: rates, mechanisms, and its role in arc mag-matism and the evolution of the continental crust and mantle. Gondwana Research 20, 284-308.

Stern, R.J., Johnson, P., 2010. Continental lithosphere of the Arabian plate: a geologic, petrologic and geophysical synthesis. Earth-Science Reviews 101, 29-67.

Thomson, D.J., 1982. Spectrum estimation and harmonic analysis. Institute of Electrical and Electronics Engineers Proceedings 70,1055-1096.

Valley, J.W., Lackey, J.S., Cavosie, A.J., Clechenko, C.C., Spicuzza, M.J., Basei, M.A.S., Bindeman, I.N., Ferreira, V.P., Sial, A.N., King, E.M., Peck, W.H., Sinha, A.K., Wei, C.S., 2005. 4.4 billion years of crustal maturation: oxygen isotope ratios of magmatic zircon. Contributions to Mineralogy and Petrology 150, 561-580.

Vakubchuk, A., 2010. Restoring the supercontinent Columbia and tracing its fragments after its breakup: a new configuration and a Super-Horde hypothesis. Journal of Geodynamics 50, 166-175.

Van Huenen, J., Moyen, J.-F., 2012. Archean subduction: fact or fiction? Annual Reviews of Earth and Planetary Sciences 40, 195-219.

Voice, P.J., Kowalewski, M., Eriksson, K.A., 2011. Quantifying the timing and rate of crustal evolution: global compilation of radiometrically dated detrital zircon grains. The Journal of Geology 119,109-126. http://dx.doi.org/10.1086/658295.

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.

Wang, C.Y., Campbell, I.H., Stepanov, A.S., Allen, C.M., Burtsev, N., 2011. Growth rate of the preserved continental crust: II. Constraints fromHfand O isotopes in detrital zircons from Greater Russian Rivers. Geochimica et Cosmochimica Acta 75,1308-1345.

Yu, S.Y., Xu, Y.G., Huang, X.L., Ma, J.L., Ge, W.C., Zhang, H.H., Qin, X.F., 2009. Hf and Nd isotopic decoupling in continental mantle lithosphere beneath Northeast China: effects of pervasive mantle metasomatism. Journal of Asian Earth Sciences 35, 554-570.

Zhao, G., Cawood, P.A., Wilde, S.A., Sun, M., 2002. Review of global 2.1-1.8 Ga collisional orogens and accreted cratons: a pre-Rodinia supercontinent? Earth-Science Reviews 59,125-162.

Zeh, A., Gerdes, A., Klemd, R., Barton Jr., J.M., 2008. U—Pb and Lu—Hf isotope record of detrital zircon grains from the Limpopo Belt—evidence for crustal recycling at the Hadean to early-Archean transition. Geochimica et Cosmochimica Acta 72, 5304—5329.

Kent C. Condie is professor of geochemistry at New Mexico Institute of Mining and Technology, Socorro, NM where he has taught since 1970. Degrees: BS Geology (1959) and MA mineralogy (1960), University of Utah; PhD, University of California, San Diego, geochemistry (1965). Prior to that time he was at Washington University in St. Louis, MO (1964—1970). Textbook, Plate Tectonics and Crustal Evolution, which is widely used in upper division and graduate courses in the Earth Sciences, was first published in 1976 and has gone through four later editions. In addition Condie has written a beginning historical geology textbook with coauthor Robert Sloan, Origin and Evolution of Earth (Prentice-Hall, 1998), an advanced textbook, Mantle Plumes and Their Record in Earth History (Cambridge University Press, 2001), and a research treatise, Archean Greenstone Belts (Elsevier, 1981). His most recent book written as an upper division/graduate textbook, is Earth as an Evolving Planetary System (Elsevier, 2005; second edition 2011). He also has edited two books, Proterozoic Crustal Evolution (Elsevier, 1992) and Archean Crustal Evolution (Elsevier, 1994). His CD ROM, Plate Tectonics and How the Earth Works is widely used in upper division Earth Science courses in the United States and Europe. Condie's research, primarily dealing with the origin and evolution of continents and the early history of the Earth, has over the years been sponsored chiefly by the U. S. National Science Foundation. He is author or co-author of over 800 articles published scientific journals.

Richard C. Aster is professor of geophysics at New Mexico Institute of Mining and Technology, where he has taught since 1991. Degrees: BS in Electrical Engineering and additional major in Physics at University of Wisconsin (1983), MS, Geophysics, University of Wisconsin-Madison (1986), Ph.D. Earth Sciences, University of California, San Diego (1991). Aster has long-standing research interests in earthquake, volcano, Antarctic and structural/imaging seismology, with over 60 authored/co-authored publications, supported by the U.S. National Science Foundation, U.S. Department of Energy, U.S. Geological Survey, and other sponsors. He recently served (2010—2011) as president of the Seismological Society of America. He is the author (with B. Borchers and C. Thurber) of a widely used textbook in geophysical and mathematical modeling, Parameter Estimation and Inverse Problems, which has recently been published in a second edition (Aster et al., Elsevier, 2012).