One of the major obstacles to our understanding of the growth of continental crust is that of estimating the balance between extraction rate of continental crust from the mantle and its recycling rate back into the mantle. As a first step it is important to learn more about how and when juvenile crust is preserved in orogens. The most abundant petrotectonic assemblage preserved in orogens (both collisional and accretionary) is the continental arc, whereas oceanic terranes (arcs, crust, mélange, Large Igneous Provinces, etc.) comprise <10%; the remainder comprises older, reworked crust. Most of the juvenile crust in orogens is found in continental arc assemblages. Our studies indicate that most juvenile crust preserved in orogens was produced during the ocean-basin closing stage and not during the collision. However, the duration of ocean-basin closing is not a major control on the fraction of juvenile crust preserved in orogens; regardless of the duration of subduction, the fraction of juvenile crust preserved reaches a maximum of ~50%. Hafnium and Nd isotopic data indicate that reworking dominates in external orogens during supercontinent breakup, whereas during supercontinent assembly, external orogens change to retreating modes where greater amounts of juvenile crust are produced. The most remarkable feature of εNd (sedimentary rocks and granitoids) and εHf (detrital zircons) distributions through time is how well they agree with each other. The ratio of positive to negative εNd and εHf does not increase during supercontinent assembly (coincident with zircon age peaks), which suggests that supercontinent assembly is not accompanied by enhanced crustal production. Rather, the zircon age peaks probably result from enhanced preservation of juvenile crust. Valleys between zircon age peaks probably reflect recycling of continental crust into the mantle during supercontinent breakup. Hafnium isotopic data from zircons that have mantle sources, Nd isotopic data from detrital sedimentary rocks and granitoids and whole-rock Re depletion ages of mantle xenoliths collectively suggest that ≥70% of the continental crust was extracted from the mantle between 3500 and 2500 Ma.
Although it is well established that U/Pb ages of zircons define episodic populations (Rino et al., 2004; Iizuka et al., 2005; Condie et al., 2009; Condie and Aster, 2010; Belousova et al., 2010; Hawkesworth et al., 2010), the origin of this distribution and how it is related to the growth and preservation of continental crust is not well understood. Both igneous and detrital zircon populations show that major age peaks or peak clusters (2700, 1900, 1000, 600 and 300 Ma) are global in extent (Condie and Aster, 2010). The common interpretation of these peaks has been that they represent periods of enhanced production of continental crust (Taylor and McLennan, 1995; Condie, 1998; Wang et al., 2009). Geochemical data from oceanic basalts and subduction-related volcanics have been interpreted to support new growth of continental crust during collisional orogenies (Niu and O'Hara, 2009). However, as pointed out by several investigators in recent years, there is a striking correlation of major zircon age peaks with the assembly of supercontinents and this has led to the suggestion that they are not crustal production peaks, but crustal preservation peaks (Fig. 1) (Condie, 1990; Kemp et al., 2006; Hawkesworth et al., 2009; Condie and Aster, 2010). Another feature of zircon age spectra that is not well understood are the age minima between times of supercontinent formation. Do these represent true minima in juvenile crustal production or could they be due to enhanced recycling of crust into the mantle and thus also be related to crustal preservation (Kemp et al., 2006; Roberts, 2012)? The answer to these questions partly depends on our understanding of collisional orogens, where most juvenile crust is preserved. Orogens are of three types, but only two are of importance in the production and preservation of juvenile continental crust: accretionary and collisional orogens (Cawood et al., 2009). As ocean basins close, accretionary orogens are active on one or both continental margins and final closure is marked by a continent-continent collision or a large terrane collision (Fig. 2). The ages of juvenile additions preserved in many Proterozoic orogens date mostly to the ocean-basin closing stage and not to the actual collision (Condie et al., 2011; Condie, 2013). However, crust is preserved in orogens during collisional phases, with the highest degree of preservation accompanying the final collisions (Hawkesworth et al., 2009).
Hafnium and Nd isotopes together with U/Pb zircon ages can be used to constrain the growth and recycling of continental crust (Wang et al., 2009; Belousova et al., 2010; Hawkesworth et al., 2010; Condie et al., 2011; Condie and Aster, 2013; Condie, 2013). We know that most juvenile crust preserved in orogens is produced in continental arcs and back arcs with only small amounts produced in oceanic arcs and other oceanic tectonic settings (Condie, 2013; Condie and Kroner, 2013). What is not well constrained is the location of orogens from which detrital zircons are derived, as most zircons are recycled multiple times in the geological record. Furthermore, primary zircon sources may not reside on the same continent as derivative detrital zircons due to past supercontinent breakup and assembly. Unlike detrital zircons, both the location and tectonic setting can often be constrained for igneous-rock Nd isotopic data. Common to all isotopic studies are geographic sampling biases, either by geologists or by river systems and this appears to be responsible for some decoupling between the isotopic systems.
In the present study, the main focus is on the questions of when, how and where juvenile continental crust is generated at convergent margins. To evaluate these questions, data from 150 post-Archaean orogens (or orogen segments) are compiled in Supplementary Appendix 1 (Supplementary Appendices 1 and 2 have been deposited with the Principal Editor of Mineralogical Magazine and are available from www.minersoc.org/pages/e_journals/dep_mat_mm.html), with corresponding information on tectonic settings and juvenile crust distribution. The Archaean is not included because the relatively hot orogens in the Archaean may have been quite different in preservation potential from the cooler orogens thereafter (Sizova et al., 2013).
Durations of orogeny
It is useful to examine the duration of both the ocean-basin closing (subduction) and collisional stages, estimates of which are compiled in Supplementary Appendix 1, in order to better understand how and when crust is preserved in orogens. Note that many of the compiled orogens are ‘orogen segments’, which, upon reconstruction of past supercontinents, may continue into other cratons. Both accretionary and collisional orogens are included in the compilation. The age of onset of subduction is equated with the ages of the oldest (post-basement) arc volcanic and plutonic rocks and is assumed to represent a minimum for the onset of ocean-basin closing. Duration of subduction is estimated as the difference between the age of the onset of subduction and the onset of collision, which is a maximum age for the duration of closing of an oceanic basin, as subduction associated with the accretionary stage of an orogen may have started before the actual closing of the ocean basin. The onset of collision is estimated from the oldest syntectonic granitoids and structures accompanying major collisions. The termination of collision is more difficult to estimate because these processes generally have a gradual rather than abrupt ending as colliding plates are annealed to each other and subduction and delamination cease. The termination is estimated from the oldest post-tectonic intrusions and structures, which is a maximum age for the end of collision, because delamination and crustal foundering may continue after collision.
Subduction (ocean-basin closing) duration ranges from 57 to >900 m.y. but most durations are between 50 and 200 m.y. and collision duration ranges from 10–260 m.y. with most falling between 40 and 150 m.y. Orogens fall into two broad groups: those with subduction durations ≤600 m.y. (mean = 115 m.y.) and those with durations of >600 m.y. (mean = 750 m.y.) (Fig. 3). Only seven examples of the >600 m.y. group are recognized: Penokean-Yavapai-Mazatzal, Makkovikian-Labradorian-Grenville, Baltica (Svecofennian-Sveconorwegian), Amazonia (Sunsas and Oaxaquia components), Altaids and Xiong'er. However, the actual number of orogens in each group depends on how orogens are divided up (lumpers vs. splitters). The first three orogens in the list above comprise the Great Proterozoic Accretionary Orogen (GPAO), which may be the longest-lived orogen of all time, in some segments enduring for >800 m.y. (Figs 3 and 4) (Condie, 2013) (Supplementary Appendix 1). The GPAO began as a segmented accretionary orogen during the amalgamation of Nuna and did not terminate in collisions until Rodinia formed at 1200–1000 Ma. It is noteworthy that the subduction duration of most post-Archaean orogens is considerably longer than the lifespan (oldest igneous rock age minus accretion age) of typical terranes in these orogens (50–300 m.y.; Condie, 2007). This means that, on average, arc magmatism in terranes continues for several hundred million years after terrane docking.
Tectonic settings preserved in orogens
It is important to characterize ancient tectonic settings preserved in orogens in order to better understand the distribution and origin of juvenile crust. Together with geochemical features, petrotectonic assemblages are key in identifying ancient tectonic settings. Details of how tectonic settings are identified and how juvenile and reworked crust are estimated, using a combination of Nd isotopes and geological maps, are given in previous papers (Condie, 1993; Condie and Chomiak, 1996; Condie, 2007). Such assemblages either survive subduction during ocean-basin closing or form during collision and they include oceanic and continental tectonic regimes as represented by both supracrustal and plutonic rocks. Greenstones are particularly useful in identifying ancient tectonic settings and both arc and within-plate oceanic greenstones show similar distributions with age (Fig. 5). The large preservational peak in both types of greenstones in the late Archaean probably reflects rapidly propagating plate tectonics at this time (Condie and O'Neill, 2010) as well as formation of the first supercratons (Bleeker and Ernst, 2006). Another preservational peak at 2100–1700 Ma corresponds to the assembly and stasis of the supercontinent Nuna. Unlike Nuna, very few greenstones that accompanied the assembly of Rodinia between 1200 and 1000 Ma are preserved and the reason for this is an important outstanding question. The peak in greenstone ages at ~500 Ma corresponds to the breakup of Pannotia and the peak at 200–100 Ma corresponds to the breakup of Pangea. The latter peak, which is dominated by within-plate oceanic greenstones, also reflects increased preservation of ophiolites and increasing numbers of studies of ophiolites of this age (Fig. 5) (Dilek and Furnes, 2011).
If petrotectonic assemblages are divided into three groups (continental arc, oceanic settings and reworked crust), the most abundant assemblage preserved in orogens is the continental arc ranging from 10–90% by volume (mean = 53%) (Fig. 6), but there is no relationship between abundance of continental arc assemblages and orogen age, at least since the end of the Archaean. In most orogens, remnants of continental arcs comprise 40–80% of supracrustal rocks, but in a few examples, such as the Limpopo, Usagaran-Tanzania, Ubendian, New Quebec and Angara orogens, they comprise ≤20% (Supplementary Appendix 1). Some of these orogens may have involved a large transpressive subduction component, such that arc magmatism was minimal. Only in the Wopmay and Tanami orogens do remnants of continental arcs comprise ≥90% of the preserved rocks. In contrast to continental arcs, oceanic terranes (including arcs, oceanic crust, serpentinites, pelagic sediments, oceanic islands and LIPs (large igneous provences)) comprise a very small proportion of tectonic settings preserved in orogens (usually ≤10%) and in some orogens, oceanic assemblages are not recognized (Fig. 6). Three exceptions stand out: the Trans-Hudson, Arabian-Nubian and Birimian-Transamazonian orogens, all of which contain 30–40% of oceanic terranes (Supplementary Appendix 1). Also, there is no apparent secular change in relative abundances of oceanic terranes with orogen age, except in the last 200 m.y. where the number of preserved ophiolites increases as mentioned above. The small number of fragments of oceanic arcs preserved in orogens is in agreement with the results of Condie and Kroner (2013), who suggest that oceanic arcs are not major components of continental growth.
Almost everything else in orogens comprises reworked crust, which includes some combination of Archaean basement, accreted terranes with older crust and detrital sediments. Reworked components comprise between 10 and 90% of orogens (most between 20 and 50%) with an average of 40%, (Supplementary Appendix 1; Fig. 6). Microcratons, such as the Sask craton in the Trans-Hudson orogen, are difficult to identify without geophysical data, because they are often not exposed at the surface. Thus, microcratons may be more abundant than suggested by the data in Supplementary Appendix 1. Four orogens are unusual in that they comprise ≥80% reworked components (Limpopo, Usagaran-Tanzania, Angara and Ubendian) and all of them fall in the 2100–1800 Ma time window. It is possible that these four orogens involved largely transpressive collisions with minimal amounts of concurrent arc magmatism.
Juvenile crust preserved in orogens
The proportion of juvenile crust (crust derived from the mantle with a short crustal residence time) preserved in orogens can be estimated from a combination of Hf and Nd isotopic data interpreted together with geological maps of various scales (Condie, 2007, 2013; Condie and Aster, 2010; Belousova et al., 2010; Dhuime et al., 2012). One uncertainty in this approach, however, is that if the zircon ages are biased by selective preservation, this will also affect Hf and Nd model ages. On average, continental arcs contain about 40–50% of juvenile crust, whereas oceanic terranes (including oceanic arcs, crust, islands and plateaus) comprise ~90% juvenile input (Condie, 2013; Condie and Aster, 2013). As shown in Fig. 7, most juvenile crust preserved in orogens occurs in remnants of continental arcs, with only minor amounts in oceanic arcs and other oceanic tectonic settings. The fraction of juvenile crust preserved in orogens ranges from 5–60% (mostly in the range of 20–40% with a mean = 30%) and the corresponding reworked crust from 40–95% (mostly in the range of 60–80% with a mean = 70%). Although there is a scatter in the data, with the exception of six orogens in the 2000–1800 Ma time window, there is a suggestion of a decrease in the fraction of juvenile crust preserved with decreasing age (Supplementary Appendix 1; Fig. 8). The Limpopo orogen contains the smallest amount of juvenile crust (~5%) and 16 other orogens containing ≤20% are mostly associated with the assembly of either Nuna or Gondwana. At the other extreme are 10 orogens that contain ≥50% juvenile input and these are mostly associated with the assembly of Rodinia or Nuna. The Arabian-Nubian orogen at 750 Ma contains the largest amount of juvenile crust at ~60%. The decrease in juvenile crust with time in orogens reflects an increasing volume of reworked crust with time.
Although Hawkesworth et al. (2009, 2010) make a case for the poor to moderate preservational potential of igneous rocks formed at convergent margins and high potential of syncollisional igneous rocks, our results show that most juvenile crust (60–90%) is generated during the ocean-basin closing stage rather than during the actual collision (Fig. 9) (Supplementary Appendix 1). Although the preservational potential is probably highest during the collisional phase, most juvenile crust trapped in collisional orogens has already survived earlier subduction. An example of this, based on Nd isotopic data, is presented by Dickin et al. (2010) for the Grenville orogen in eastern Canada.
One might expect the longer the duration of ocean-basin closing the greater the amount of juvenile crust preserved, however, this is not supported by the data. Although the seven orogens with subduction durations of >600 m.y. all contain ~50% juvenile components, many of the orogens with shorter subduction durations contain similar amounts of juvenile crust (Fig. 10). This indicates that the duration of ocean-basin closing is not a major control on the fraction of juvenile crust preserved in orogens. It is important to note that the seven orogens with subduction durations >600 m.y. all have long-lived accretionary phases. Three out of the seven comprise part of the GPAO, which contains the largest volume of juvenile crust preserved globally between 2 and 1 Ga (Condie, 2013).
Supercontinents and juvenile crust
Several investigations have shown that major U/Pb zircon age peaks correlate with supercontinent assembly and may represent preservational rather than production peaks of continental crust (Fig. 1) (Condie and Aster, 2010, 2013; Hawkesworth et al., 2009, 2010; Kemp et al., 2009; Roberts, 2012). Condie and Aster (2013) have also shown that the combined use of Hf and Nd isotopes is more useful in understanding the supercontinent cycle and its relation to juvenile crust preservation than the use of Hf isotopic data only from detrital zircons. Unlike detrital zircons, where sources cannot be located precisely because of crustal recycling, both the location and tectonic setting can often be constrained for Nd isotopic data from granitoids. Murphy (2003) has suggested that supercontinents form in two tectonic scenarios: extroversion, which involves closing of external ocean basins and introversion, which involves closing of internal ocean basins. Extroversion and introversion have different signals in εNd and εHf distributions in the rock record (Murphy and Nance, 2003; Collins et al., 2011; Roberts, 2012; Condie and Aster, 2013). The extroversion types show rapid drops in ε during assembly, which lasts the order of 100 m.y. (Fig. 11) (Condie and Aster, 2013; latest update of Nd isotopic data given in Supplementary Appendix 2). This decrease in ε values begins during supercontinent breakup as shown by the breakup of the late Archaean supercratons (2150–2050 Ga), Rodinia (750–650 Ma) and Pannotia (550 Ma) and reflects reworking in external orogens as they converge and collide with each other. Breakup of the Archaean supercraton, Kenorland, using the reconstruction of Bleeker and Ernst (2006) also supports an assembly of Nuna by extroversion. However, there is no evidence in Nd or Hf isotope records for the breakup of Nuna (Fig. 11). The only supercontinent that clearly formed by introversion is Pangea (Murphy et al., 2009) and both εNd and εHf show a rapid increase during its assembly around 400 Ma. 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 around 150 Ma is accompanied by a sharp decrease in both εNd and εHf, followed by an abrupt increase. If the correlation of Nd and Hf isotopic data with supercontinent evolution shown in Fig. 11 is correct, it implies that external orogens may have a significant control on detrital zircon populations and Roberts (2012) reached a similar conclusion based solely on Hf isotope data.
Although the combined Nd and Hf databases increase our insight into the supercontinent cycle, they also present several new and important questions. It appears that external orogens often control both εNd and εHf distributions when considered on a global scale (Hawkesworth et al., 2010; Collins et al., 2011; Condie, 2013) and if this is correct, we must ask why internal orogens (accretionary or collisional) do not contribute significantly to isotopic signatures in the preserved crust? Another intriguing question is why Nuna appears to be the only supercontinent with a long-lived (≥800 m.y.) accretionary orogen (the GPAO) leading into the assembly of the next supercontinent? These questions are also related to what determines whether a supercontinent forms by extroversion or introversion and why the introversion case is so rare?
Recycling of continental crust into the mantle
If the major U/Pb zircon age peaks result from enhanced preservation of continental crust during the assembly of supercontinents, do the valleys between the peaks represent (1) relatively few zircon-producing magmatic events during these times, (2) sampling biases by geologists and rivers, (3) the background extraction and recycling rates of continental crust from the mantle as suggested by Condie et al. (2011), or (4) enhanced recycling of continental crust into the mantle? When both Hf and Nd isotopic data are considered, 1 and 2 seem very unlikely (Condie, 2013; Condie and Aster, 2013). Numerous accretionary orogens were active during the minima between zircon age peaks and most or all of these produced zircon-bearing granitoids. Also, as discussed in previous papers, it is unlikely that geologists and rivers have similar sampling biases such that both show age minima at the same times (Condie et al., 2009; Condie and Aster, 2010). It is difficult with our existing data to distinguish between possibilities 3 and 4 and thus both possibilities need to be tested against future data.
If the zircon peaks were truly crustal generation peaks, the ratio of positive to negative εNd and εHf should increase at the same times and if recycling does not discriminate between juvenile and reworked crust, the ratio of juvenile to reworked crust should be about the same in age peaks and valleys. The most remarkable feature of the ε curves is how well they agree with each other through time (Fig. 11). At times of supercontinent assembly, median εNd and εHf do not show peaks in positive ε, but are close to zero, thus not favouring new crustal production at these times. Both ε values are also close to zero for supercontinent breakups. The somewhat high positive ε values at 700 and 1400 Ma reflect juvenile crust generation in long-lived accretionary orogens (Arabian-Nubian and Central Asian Orogens at 800–600 Ma and the GPAO at 1900–1300 Ma) and not fragmentation of the supercontinents. The large difference between εNd and εHf at 100 Ma is puzzling and it may be that the continued fragmentation of Pangea and the beginning of the assembly of a new supercontinent (Amasia) have caused decoupling in these isotopic systems. In summary, the εNd and εHf distributions do not favour enhanced production of juvenile continental crust either during or between supercontinent assemblies.
Accretionary orogens go through retreating and advancing phases and during retreating phases new continental crust is generated in forearc regions and preferentially preserved in back arc regions (Fig. 12a) (Kemp et al., 2009; Cawood et al., 2009). In contrast, advancing orogens may lose significant volumes of crust by recycling into the mantle at subduction zones by a combination of subduction erosion, sediment subduction and delamination (Cawood et al., 2009; Scholl and von Huene, 2007, 2009; Clift et al., 2009; Stern, 2011). Studies of both the Japanese and Andean orogens show that large volumes of juvenile and reworked crust have been recycled into the mantle during the Phanerozoic (Kay et al., 2005; Scholl and von Huene, 2007, 2009; Clift et al., 2009). Enhanced subduction erosion in the Andes in the last 250 m.y. may reflect the opening of the Atlantic basin, during which the Andean arc changed from a retreating to an advancing arc (Fig. 12b). Scholl and von Huene (2007, 2009) have estimated that continental crust around the Pacific basin is being recycled into the mantle at a rate of 3.2 km3/yr, approximately equal to the rate at which it is being extracted from the mantle. At this rate, the entire width of an accretionary orogen can be destroyed in a few hundred million years and if this rate is typical of the last 3 g.y., a volume equal to the current continental crust would have been recycled into the mantle during this time interval.
If recycling can explain age gaps in single orogens, could it also explain the large minima in ages between the global zircon age peaks (Fig. 1)? At least in part, two of these age minima correlate with supercontinent breakup. During the 1500–1250 Ma minimum, Nuna may have partially fragmented leading into the assembly of Rodinia (Pisarevsky et al., 2014). Consistent with this possibility is the fact that crustal remnants of this age occur in the GPAO as well as in other external accretionary orogens such as Namaqua-Natal, Kibaran, Irumides and Albany-Fraser (Supplementary Appendix 1). During the 900–700 Ma minimum, Rodinia fragmented and minor pieces of crust of this age are widespread in external accretionary orogens such as the Irumides, Kibaran, Jiangnan, Arabian-Nubian, Central Asian, Zambezi and East African Orogens. Although the 400 Ma mimimum correlates with the beginning of the assembly of Pangea, crustal remnants of this age formed chiefly in external accretionary orogens, although they are now preserved in collisional orogens (examples include the Taconic-Caledonian, CAOB, Qinling-Dabi-Sulu, Antler, Variscan, Pampean and Acadian Orogens). Neodymium and Hf model ages from these orogens indicate that continental crust must have formed during the age minima, but that it was not preserved. This observation supports a recycling explanation for the large valleys between the major zircon age peaks (Fig. 1). The large, well established crustal age gap at 2400–2200 Ma (Condie et al., 2009) correlates with stasis of the late Archaean supercratons, which fragmented at 2150–2050 Ma. Although this age gap was originally suggested to record a slow-down in global magmatism and plate tectonics (Condie et al., 2009), it more likely reflects recycling of continental crust into the mantle. Supporting this interpretation are widespread Hf and Nd model ages as well as detrital zircon ages from South America and south China ~2300 Ma (Sun et al., 2009; Berman et al., 2010; Teixeira et al., 2014).
Growth rate of continental crust through time
The best estimate of the net growth rate of continental crust uses a combination of model ages of Nd in sediments and granitoids and of Hf in zircons. To avoid inclusion of significant volumes of reworked continental crust, it is important to consider Hf isotope data in zircons that are filtered with O isotopes so as to include only those with mantle isotopic signatures (δ18O = 4.5–6.5‰ (Kemp et al., 2006; Iizuka et al., 2012; Wang et al., 2011; Dhuime et al., 2012). A comparison of δ18O filtered and unfiltered data shows a similar distribution of juvenile crust with broad model age peaks at 3500–3000 and 2000–1200 Ma, which is in striking contrast to the sharp peaks of the zircon age spectra (Figs 1, 13a,b). Note that the 2000δ1200 Ma peak is much smaller using the δ18O filtered data. The Nd model ages of sediments and granitoids show much the same pattern as the Hf isotope data (Fig. 13c). The Archaean peak is somewhat younger, centred ~3000–2700 Ma and the Proterozoic peak somewhat older at 2300–1400 Ma. There also is a suggestion of renewed crustal growth around 900–600 Ma in the Nd data and 600–400 Ma in the Hf data (both filtered and unfiltered).
Other than the late Archaean peak in Nd model ages, there is no correlation of model age peaks with zircon age peaks (Fig. 1), which supports the interpretation of the zircon peaks as preservation peaks rather than production peaks of continental crust. However, both Nd and Hf model ages may also be biased because an unknown number of model ages reflect mixing rather than isotopic growth. Another way to test the preservation model is using Re/Os isotopic ages of mantle xenoliths (Carlson et al., 2005; Pearson et al., 2007). Rhenium depletion ages of xenoliths may record major melting events in the mantle and if so, there is strong evidence of a major event around 2800–2500 Ma (Fig. 13d). The broad peak in Re depletion ages in the last 1000 m.y. probably reflects a sampling bias, as many isotopic studies have concentrated on mantle xenoliths in volcanics underlain by Phanerozoic crust. If most weight is given to the filtered Hf isotopic data in zircons, the whole-rock Nd isotopic data from granitoids and sediments and the whole-rock Re depletion ages of mantle xenoliths, it would appear that most of the continental crust (≥70%) was extracted from the mantle between 3500 and 2500 Ma (Belousova et al., 2010; Condie and Aster, 2010; Collins et al., 2011; Dhuime et al., 2012).
Most orogens have durations of ≤600 m.y. whereas a few, with long-lived accretionary phases, can last for up to 800 m.y. or more. The deformation in orogens generally ends with a continent–continent collision, although some segments of accretionary orogens can end with terrane collisions. Most of the long-lived accretionary orogens are part of the gigantic GPAO, which extended along the southern margins of Laurentia and Baltica (present day coordinates) from 2.0 to 1.2 Ga.
The most abundant tectonic setting preserved in orogens is the continental arc, at least since the
Juvenile continental crust is generated and preserved in accretionary orogens during ocean-basin closing, chiefly in external orogens, but in some cases also in internal orogens. The ultimate preservation of juvenile crust occurs during continental collisions, or in some instances in accretionary orogens, during collision and accretion of large terranes.
Supercontinents that form by extroversion show rapid drops in εNd and εHf during their assembly, which lasts the order 100 m.y. This decrease in ε begins during supercontinent breakup and reflects extensive reworking in external orogens as they converge and collide with each other. The only supercontinent that clearly formed by introversion is Pangea and both εNd and εHf show a rapid increase during its assembly. This is consistent with a greater proportion of juvenile crust production in external orogens, which shift into retreating modes as internal oceans close.
The most remarkable feature of εNd (sedimentary rocks and granitoids) and εHf (detrital zircons) distributions through time is how well they agree with each other. The ratio of positive to negative εNd and εHf does not increase during supercontinent assembly (co-incident with zircon age peaks), which suggests that supercontinent assembly is not accompanied by enhanced crustal production, assuming recycling does not discriminate between juvenile and reworked crust. The proportions of positive to negative εNd and εHf also do not correlate with supercontinent breakup.
Neodymium model ages from detrital sedimentary rocks and granitoids, Hf model ages from mantle-sourced zircons and whole-rock Re depletion ages of mantle xenoliths collectively suggest that ≥70% of the continental crust was extracted from the mantle between 3500 and 2500 Ma.
Although some age minima can be related to the breakup of supercontinents, others may correlate with supercontinent assembly or stasis. Minima between U/Pb zircon age peaks at 1500–1250 Ma and 900–700 Ma may reflect recycling of continental crust into the mantle and correspond to either partial or complete breakup of supercontinents. In contrast, minima at 2400–2200 and 400 Ma correspond to stasis or growth of supercontinents, respectively. No peaks or minima occur during overlap of supercontinent assembly and breakup between 750 and 500 Ma and ≤150 Ma. A major question that must be addressed is that of what mantle processes control the balance of extraction and recycling of continental crust out of and into the mantle and why should one or the other be more important?
This paper is based on the Hallimond lecture given at the meeting How to Build Strong Continents, September 2–4, 2013 in Portsmouth, UK. The author acknowledges the Mineralogical Society of the UK for the invitation to give this lecture.