* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Supercontinents, mantle dynamics and plate
Survey
Document related concepts
History of geomagnetism wikipedia , lookup
Geomorphology wikipedia , lookup
Age of the Earth wikipedia , lookup
History of geology wikipedia , lookup
Abyssal plain wikipedia , lookup
History of Earth wikipedia , lookup
Oceanic trench wikipedia , lookup
Post-glacial rebound wikipedia , lookup
Future of Earth wikipedia , lookup
Geological history of Earth wikipedia , lookup
Plate tectonics wikipedia , lookup
Large igneous province wikipedia , lookup
Transcript
Earth-Science Reviews 105 (2011) 1–24 Contents lists available at ScienceDirect Earth-Science Reviews j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e a r s c i r ev Supercontinents, mantle dynamics and plate tectonics: A perspective based on conceptual vs. numerical models Masaki Yoshida a,⁎, M. Santosh b a b Institute for Research on Earth Evolution (IFREE), Japan Agency for Marine-Earth Science and Technology (JAMSTEC), 2-15 Natsushima-cho, Yokosuka, Kanagawa 237-0061, Japan Division of Interdisciplinary Science, Faculty of Science, Kochi University, Akebono-cho 2-5-1, Kochi 780-8520, Japan a r t i c l e i n f o Article history: Received 1 June 2010 Accepted 7 December 2010 Available online 14 December 2010 Keywords: supercontinents mantle dynamics plate tectonics Wilson Cycle supercontinent cycle numerical model a b s t r a c t The periodic assembly and dispersal of supercontinents through the history of the Earth had considerable impact on mantle dynamics and surface processes. Here we synthesize some of the conceptual models on supercontinent amalgamation and disruption and combine it with recent information from numerical studies to provide a unified approach in understanding Wilson Cycle and supercontinent cycle. Plate tectonic models predict that superdownwelling along multiple subduction zones might provide an effective mechanism to pull together dispersed continental fragments into a closely packed assembly. The recycled subducted material that accumulates at the mantle transition zone and sinks down into the core–mantle boundary (CMB) provides the potential fuel for the generation of plumes and superplumes which ultimately fragment the supercontinent. Geological evidence related to the disruption of two major supercontinents (Columbia and Gondwana) attest to the involvement of plumes. The re-assembly of dispersed continental fragments after the breakup of a supercontinent occurs through complex processes involving ‘introversion’, ‘extroversion’ or a combination of both, with the closure of the intervening ocean occurring through Pacific-type or Atlantic-type processes. The timescales of the assembly and dispersion of supercontinents have varied through the Earth history, and appear to be closely linked with the processes and duration of superplume genesis. The widely held view that the volume of continental crust has increased over time has been challenged in recent works and current models propose that plate tectonics creates and destroys Earth's continental crust with more crust being destroyed than created. The creation–destruction balance changes over a supercontinent cycle, with a higher crustal growth through magmatic influx during supercontinent break-up as compared to the tectonic erosion and sediment-trapped subduction in convergent margins associated with supercontinent assembly which erodes the continental crust. Ongoing subduction erosion also occurs at the leading edges of dispersing plates, which also contributes to crustal destruction, although this is only a temporary process. The previous numerical studies of mantle convection suggested that there is a significant feedback between mantle convection and continental drift. The process of assembly of supercontinents induces a temperature increase beneath the supercontinent due to the thermal insulating effect. Such thermal insulation leads to a planetary-scale reorganization of mantle flow and results in longestwavelength thermal heterogeneity in the mantle, i.e., degree-one convection in three-dimensional spherical geometry. The formation of degree-one convection seems to be integral to the emergence of periodic supercontinent cycles. The rifting and breakup of supercontinental assemblies may be caused by either tensional stress due to the thermal insulating effect, or large-scale partial melting resulting from the flow reorganization and consequent temperature increase beneath the supercontinent. Supercontinent breakup has also been correlated with the temperature increase due to upwelling plumes originating from the deeper lower mantle or CMB as a return flow of plate subduction occurring at supercontinental margins. The active mantle plumes from the CMB may disrupt the regularity of supercontinent cycles. Two end-member scenarios can be envisaged for the mantle convection cycle. One is that mantle convection with dispersing continental blocks has a short-wavelength structure, or close to degree-two structure as the present Earth, and when a supercontinent forms, mantle convection evolves into degree-one structure. Another is that mantle convection with dispersing continental blocks has a degree-one structure, and when a supercontinent forms, mantle convection evolves into degree-two structure. In the case of the former model, it would take longer time to form a supercontinent, because continental blocks would be trapped by different downwellings thus inhibiting collision. Although most of the numerical studies have assumed the continent/supercontinent ⁎ Corresponding author. Tel.: + 81 46 867 9814; fax: +81 46 867 9315. E-mail address: [email protected] (M. Yoshida). 0012-8252/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.earscirev.2010.12.002 2 M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 to be rigid or nondeformable body mainly because of numerical limitations as well as a simplification of models, a more recent numerical study allows the modeling of mobile, deformable continents, including oceanic plates, and successfully reproduces continental drift similar to the processes and timescales envisaged in Wilson Cycle. © 2010 Elsevier B.V. All rights reserved. Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Supercontinent cycle and Wilson Cycle . . . . . . . . . . . . . . 3. Supercontinent cycle and mantle convection . . . . . . . . . . . 4. Thermal and mechanical interaction between continent and mantle 5. Formation of degree-one mantle structure . . . . . . . . . . . . . 6. Long-wavelength thermal structure in the mantle . . . . . . . . . 7. Timescale of the supercontinent assembly . . . . . . . . . . . . . 8. Supercontinent breakup . . . . . . . . . . . . . . . . . . . . . 9. Stability and longevity of the continent . . . . . . . . . . . . . . 10. Origin and growth of the continental crust . . . . . . . . . . . . 11. A preliminary Wilson Cycle model . . . . . . . . . . . . . . . . 12. Continents and plate tectonics . . . . . . . . . . . . . . . . . . 13. Next supercontinent . . . . . . . . . . . . . . . . . . . . . . . 14. Summary and conclusion . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Introduction The history of growth, evolution and dispersion of supercontinents on the globe through time has received considerable attention in the recent years, particularly with respect to the impact of the assembly and dispersion of continental fragments on mantle dynamics, surface processes and life evolution (for a recent compilation, see Santosh and Zhao, 2009, and papers therein). A recent synthesis of the various conceptual models suggests that supercontinent tectonics in relation to mantle dynamics provides a key to evaluate the history of evolution and destruction of the continental crust, to understand the history of life, and to trace the major surface environmental changes of our planet (Santosh, 2010b). The postulated Neo-Archean continental assemblies (e.g., Rogers and Santosh, 2004; Eriksson et al., 2009) and the increasing evidence for the Paleo-Mesoproterozoic supercontinent Columbia (Meert, 2002; Rogers and Santosh, 2002; Zhao et al., 2002; Rogers and Santosh, 2009), Neoproterozoic Rodinia (Dalziel, 1991; Hoffman, 1991; Z.X. Li et al., 2008) and Late Neoproterozoic– Cambrian Gondwana (Collins and Pisarevsky, 2005; Meert and Lieberman, 2008), among other proposed supercontinents, support the notion that global cycles of continental reorganization have occurred throughout Earth's history (Worsley et al., 1984; Nance et al., 1986). Seismic tomographic images suggest that the Earth's mantle structure is characterized by different modes of flow: (1) Subducting plates mainly beneath the Circum-Pacific region, some of which are stagnated at the 660 km phase boundary (i.e., spinel to perovskite + magnesiowüstite phase transition boundary), whereas others penetrate into the deeper lower mantle (e.g., Fukao, 1992; van der Hilst et al., 1997; Fukao et al., 2001; Zhao, 2004); (2) Large-scale, broad upwelling-plumes beneath the South Africa–South Atlantic and South Pacific regions (e.g., Fukao, 1992; Masters et al., 2000; Mégnin and Romanowicz, 2000; Ritsema and van Heijst, 2000); (3) Smallscale, localized upwelling-plumes originating from the core–mantle boundary (CMB) or 660 km phase boundary, which were detected mainly by the recent highly-resolved tomographic model (e.g., Wolfe et al., 1997; Montelli et al., 2004, 2006; Wolfe et al., 2009). Geochemical evidence and geodynamic models support this global . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 3 3 6 7 9 11 12 13 14 15 16 19 20 21 21 view of mantle structure, although several models with various compositional heterogeneities have also been proposed (see review by Tackley, 2000a, 2007). On the other hand, a continent/supercontinent is isolated from the convecting mantle in terms of the rheology, composition, large radiogenic internal heating production (e.g., Schubert et al., 2001), and the longevity over geologic time (e.g., Carlson et al., 2005). The thermal and mechanical interaction between the continental drift and mantle convection has not been, however, fully resolved. The numerical studies of mantle convection have markedly progressed toward the realization of seismic tomography images of mantle structure and a better understanding of geodynamic mechanisms in accordance with the advancement in numerical modeling techniques as well as the increase of computational power and resource. The mantle convection theory is comprehensively summarized by several papers and textbooks (e.g., McKenzie et al., 1974; Jarvis and McKenzie, 1980; Christensen, 1984; Busse, 1989; Schmeling, 1989; Davies, 1999; Schubert et al., 2001; Turcotte and Schubert, 2002; Ricard, 2007). A review of mantle convection studies and the numerical simulation techniques used are beyond the scope of this paper, and can be found in the several textbooks and papers with broader view and perspective (e.g., Richards and Davies, 1992; Tackley, 2000a; Schubert et al., 2001; Ricard, 2007; Zhong et al., 2007b; Ismail-Zadeh and Tackley, 2010). In particular, numerical studies performed in the 80s–90s mainly by using two-dimensional (2-D) model with continents/supercontinents are carefully reviewed in the textbook by Schubert et al. (2001). The relationship between the supercontinent and mantle convection processes is clarified in a review by Condie (2001) from the viewpoint of geology and geochronology. However, there have been very little attempts so far to link the geophysical numerical models and the geological and tectonic conceptual models to understand the history of plate tectonics, Wilson Cycle and supercontinent cycle. Furthermore, it is important to link these models with the actual surface geological records, and the quantitative geophysical data from various types of geophysical surveys. In the recent years, several numerical models of mantle convection have addressed the assembly and breakup of supercontinents using three-dimensional (3-D) models. This paper M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 aims to synthesize the salient results from these studies in an attempt to provide a unified approach in evaluating the numerical studies and speculative geodynamic models to better constrain the mechanism of formation and disruption of supercontinents. 2. Supercontinent cycle and Wilson Cycle Whereas the term supercontinent cycle (Worsley et al., 1984, 1986; Nance et al., 1988; Unrug, 1992) describes the periodic assembly and dispersal of continental fragments, the term Wilson Cycle (after J. Tuzo Wilson, the pioneer who laid the base for the modern concept of plate tectonics; Dewey and Spall, 1975) has been used to explain the periodic opening and closing of ocean basins (Wilson, 1966). In a simple sense, supercontinent cycle and Wilson Cycle are complimentary because fragmentation of supercontinents opens new ocean basins and amalgamation of supercontinents leads to the closure of ocean basins. However, supercontinent cycles involve more complex processes that occur during the assembly of multiple continental fragments of various ages including ancient cratons together with accreted terranes. This is in contrast to the rather simple components in Wilson Cycle involving the production and destruction of ocean floor, with the oldest oceanic crust on the planet aged more than 200 Ma. Among various distinctions between the two cycles, one important aspect is the time interval, with a shorter time span for the Wilson Cycle as against a much longer time span in which supercontinent cycle operates. Whereas this aspect still remains vague, several complex processes associated with supercontinent cycles have been proposed, involving ‘introversion’ and ‘extroversion’ of continental fragments rifted from a previous supercontinent, and sometimes even a combination of these two processes (Murphy and Nance, 2005; Murphy et al., 2009). Hoffman (1991) proposed the concept of ‘inside-out’ process as a mechanism of assembling crustal fragments after the breakup of an earlier supercontinent. If the supercontinent was rifted on one side to bear a passive margin such as in the case of the Atlantic Ocean, then the opposite side becomes a consuming boundary along the continents, such as in the case of the Pacific margin. With time, the Pacific Ocean would shrink and finally close by the passive collision of two continents to generate a large continental mass. In the case of Atlantic, initially both continental margins are passive, and later turn to active margins, and finally shrink in size and totally disappear. If introversion occurs in this case, the Atlantic would close. On the other hand, during much of the Mesozoic– Cenozoic the Indian Ocean has had two different types of margins simultaneously in operation, active and passive, transporting the northern continental margin of Gondwana by the Atlantic-type process and amalgamating the rifted continents to the southern margin of Asia by the Pacific-type process. Therefore, the Indian Ocean-type process illustrates simultaneous continental break up and continental amalgamation exemplifying ‘inside-in’ mechanism (Murphy and Nance, 2003, 2005). Similar simultaneous rifting and accretion on different margins were also proposed in the case of the Paleoproterozoic supercontinent Columbia by Rogers and Santosh (2002). The Tethyan process started at least by the Triassic to Permian time and operated simultaneously with the Pacific process to the east. Subsequently, double-sided subduction started to define the frontier of the future supercontinent (Maruyama et al., 2007). Thus, the Tethyan region is an example for the ‘inside-in’ reassembly of supercontinents (Hoffman, 1991; Murphy and Nance, 2003, 2005; Murphy et al., 2009) whereas the Pacific region is related to the ‘inside-out’ configuration. Thus, the process of completion of a supercontinent would involve a combination of both introversion and extroversion (Fig. 1a–e). Such a combination operated in the case of Rodinia assembly, and is also implied in the prediction of the amalgamation of the future supercontinent Amasia (see Maruyama et al., 2007; Santosh et al., 2009). In another recent work, Murphy et al. (2009) discussed the two geodynamically distinct tracts of oceanic lithosphere generated 3 during the breakup of supercontinents: a relatively young interior ocean floor that develops between the dispersing continents, and a relatively old exterior ocean floor, which surrounded the supercontinent before breakup. The geologic and Sm/Nd isotopic record synthesized in their study (Fig. 1f) suggests that supercontinents may form by two end-member mechanisms: introversion, in which interior ocean floor is preferentially subducted, and extroversion, in which exterior ocean floor is preferentially subducted. Murphy et al. (2009) speculated that ‘superplumes’ (Larson, 1991; Fukao et al., 1994; Maruyama, 1994; Maruyama et al., 2007), perhaps driven by slab avalanche events (Machetel and Weber, 1991; Honda et al., 1993; Tackley et al., 1993), could occasionally overwhelm top–down geodynamics, imposing a geoid high over a pre-existing geoid low and causing the dispersing continents to reverse their directions to produce an introverted supercontinent. In another different model, Silver and Behn (2008) proposed two modes of ocean closure during supercontinent formation which they termed as P-type (Pacific type) and A-type (Atlantic type), a concept broadly similar to the model proposed in Murphy and Nance (2003, 2005) and Murphy et al. (2009). When a supercontinent surrounded by subduction zones begins rifting, the resulting continental breakup creates an internal ocean. As the breakup continues, the size of the internal ocean increases at the expense of the external ocean. In A-type closure, the subduction begins at a passive margin of the internal ocean and the internal ocean begins to close. In P-type closure, the internal ocean continues to grow and becomes an enlarged ocean basin. When a supercontinent is assembled through A-type closure, the internal ocean closes, shutting down subduction zones in the internal ocean, while subduction and sea-floor spreading in the external ocean continue. In Ptype closure, the external ocean closes, shutting down all subduction. Silver and Behn (2008) suggested that A-type and P-type closures can be distinguished by the age of former oceanic crustal material (such as ophiolites) that is trapped in the suture zone along which the supercontinent assembled. In the case of A-type closure, the age of the oceanic material will be younger than that of breakup of the previous supercontinent. This is because the subduction initiation postdates the previous breakup. On the other hand, in the case of P type closure, the oceanic crustal material can predate the breakup of the previous supercontinent because subduction initiation predates the breakup. Similar to the case of ‘introversion’ and ‘extroversion’ (Murphy and Nance, 2003, 2005), the P-type and A-type closures represent two endmember cases, and the actual mode of closure may likely involve a combination of the two processes. According to Silver and Behn (2008), the supercontinent Pangaea, which was formed by the closing of the Iapetus Ocean, appears to have formed primarily by A-type closure. In contrast, both the supercontinents Pannotia and Rodinia formed primarily by P-type closure. Although the Paleoproterozoic supercontinent Columbia might have formed by A-type closure (Rogers and Santosh, 2002, 2009), the available data are not adequate to draw a firm conclusion. 3. Supercontinent cycle and mantle convection One of the major challenges in earth sciences is to resolve the thermal and mechanical feedback between mantle convection and continental/supercontinental drift. A widely-accepted theory proposes that a supercontinent thermally insulates the underlying mantle and isolates it from subduction (i.e., ‘thermal blanket effect’), and eventually breaks into pieces that move towards colder regions of mantle downwellings (Anderson, 1982; Gurnis, 1988; Anderson, 1994). Continental aggregation might lead to a re-organization of the convective flow in the mantle and a positive temperature excursion beneath the supercontinent (e.g., Yoshida et al., 1999; Coltice et al., 2009). Fig. 2 illustrates an example of the time evolution of mantle convection with a rigid, undeformable supercontinent that covers 30% of the total surface and 250 km thickness from the surface boundary in 4 M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 Fig. 1. (a–e) Stages of breakup and assembly of supercontinents through ‘introversion’, ‘extroversion’ or a combination of both introversion and extroversion as discussed in the text (after Murphy and Nance, 2005). (f) Schematic representation of the Sm/Nd isotopic evolution of oceanic lithosphere from the interior and exterior oceans (from Murphy et al., 2009). The depleted mantle ages for the interior ocean (TI) are younger than the time of supercontinent breakup (TR). Whereas in the case of exterior ocean (TE) the ages are older. Relative to the breakup of Rodinia supercontinent, the Mozambique Ocean is an exterior ocean; relative to the breakup of Pannotia, the Iapetus and Rheic Oceans are interior oceans. 2-D spherical-shell geometry. Initially, the mantle convections are organized by the ‘old’ supercontinent (‘old’ in Fig. 2), and upwelling plumes dominate beneath the old supercontinent, while downwelling plumes dominate beneath the antipodes. Subsequently, at the elapsed time of 0 Myr, we instantaneously imposed a ‘new’ supercontinent (‘new’ in Fig. 2) on the antipodes of the old supercontinent where downwelling plumes dominate. After the set of supercontinent, the downwellings beneath the supercontinent tend to become weak and are not discernable at ~32 Myr. A new upwelling plume occurs at ~92 Myr, and the mantle temperature beneath the supercontinent tends to increase due to the thermal insulation effect (see Section 4). The large-scale horizontal mantle flow beneath the supercontinent, including the high temperatures, swept the downwelling into the antipodes and induced a planetary-scale flow in the whole mantle for b200 Myr. M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 5 Fig. 2. Time sequence of mantle convection with a rigid, highly viscous supercontinent in two-dimensional spherical-shell geometry with a thickness of 2867 km. The circumference ratio between the inner and outer shell is considered to be the ratio of the area between the CMB and top surface boundary of the Earth. Color contour shows the temperature. Orange lid indicates the position of the supercontinent with a thickness of 250 km, covering 30% of total surface area (note that for clarify, the supercontinent is set out on the outer surface in this illustration). ‘Old’ and ‘new’ indicate an old and new supercontinent: the new supercontinent is instantaneously imposed on the antipodes of the ‘old’ supercontinent at 0 Myr. In this model, realistic Rayleigh number (5.9 × 107) is considered in the mantle which is heated from the bottom and internally heated by the radiogenic elements. The radiogenic heating rate of mantle materials are 8.0 × 10− 12 W/kg, which is close to the value at 2 Ga (Turcotte and Schubert, 2002). Eventually the upwelling plumes tend to dominate beneath the new supercontinent, as the elapsed time passes. Fig. 3 illustrates a cartoon showing the time series of mantle reorganization by the amalgamation of supercontinent. The fragments of the supercontinent would migrate to a downwelling flow of the mantle. The mantle beneath the supercontinent warms up by the thermal blanket effect, and the largescale, hotter flow emerges from beneath the supercontinent to subcontinental regions (i.e., oceanic lithosphere). This large-scale flow in the shallower mantle produces large-scale (planetary-scale) flow via the deeper mantle. Eventually, the bottom thermal boundary layer is perturbed by this return flow and then the upwelling plumes are concentrated beneath the supercontinent. Continental rafts impose their own wavelength on mantle convection by impeding downwelling below them (Gurnis, 1988). Thus, the assembly of supercontinents would force larger scales of convection and drive the underlying mantle towards higher temperature. Some models consider that the extensive subduction associated with the assembly of supercontinents involves the transport of large volumes of oceanic as well as continental materials into the deep mantle, and that some of these recycled materials act as fuel to generate mantle plumes which rise up and eventually disintegrate the supercontinents (e.g., Maruyama et al., 2007). Although models on the assembly and disruption of supercontinents are diverse, most of the studies emphasize the implications of supercontinent tectonics on Fig. 3. A cartoon showing the time series of mantle reorganization by the existence of amalgamation of supercontinent. 6 M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 the rheological properties of the mantle, which in turn control the major geological processes in our planet (Karato, 2010a). Based on the topology of Y-shaped triple junctions in major supercontinental assemblies, Santosh et al. (2009) recognized two distinct categories of subduction zones on the globe: the CircumPacific subduction zone and the Tethyan subduction zone. In the randomly distributed subduction zones on the surface of the earth, the low temperature and dense subducted material sinks down to the bottom of the mantle (e.g., Maruyama et al., 2007; Fukao et al., 2009). In the scenario where the subduction is double-sided, the triangular regions with Y-shaped topology selectively refrigerate the underlying mantle, reducing the temperature and turn down the temperature in these domains as compared to the surrounding regions. The Y-shaped domains also accelerate the refrigeration through larger amounts of subduction and thus promote stronger downwelling as compared to other regions of the mantle. Once this process is initiated, a runaway growth of a region characterized by cold downwelling starts to develop, producing a zone of super-downwelling to pull together the continental material on the surface into a tight assembly. The ‘maximum close packing’ of continental fragments within supercontinents suggested by Rogers and Santosh (2004) can be explained through this process. The continental lithosphere exhibits a general behavior that is almost decoupled from the convecting mantle and the motion of the oceanic plates because the material constituting this lithosphere is less dense than that of the mantle and oceanic lithosphere and is less deformable than that of the mantle. It thus acts as an assemblage of fairly rigid bodies ‘floating’ on the top of the mantle. In numerical studies of mantle convection, this allows the continent to be modeled as a nondeformable, highly viscous lid (HVL) or a rigid cap. A number of previous studies have addressed the effects of supercontinents on the dynamics and structure of the mantle using 2-D models (Gurnis, 1988, 1990; Gurnis and Zhong, 1991; Lowman and Jarvis, 1993; Zhong and Gurnis, 1993; Lowman and Jarvis, 1995, 1996; Nakakuki et al., 1997; Trubitsyn et al., 2006) and 3-D Cartesian/spherical-shell models (Trubitsyn and Rykov, 1995; Lowman and Gable, 1999; Yoshida et al., 1999; Honda et al., 2000; Trubitsyn and Rykov, 2001; Coltice et al., 2007; Zhong et al., 2007a; Trubitsyn et al., 2008; Coltice et al., 2009; Phillips and Coltice, 2010; Yoshida, 2010b). A pioneering work with a numerically modeled mantle convection with a rigid supercontinent (Gurnis, 1988) presents crucial results. His 2-D rectangle model has presented crucial results on the dynamic feedback between the mantle and continent/supercontinent. In his model, a supercontinent and the dispersing continental fragments from it can migrate in accordance with the horizontal velocity of the mantle beneath the supercontinent. The supercontinent that is initially located above an upwelling mantle plume in the welldeveloped convection breaks up, with its fragments migrating to a downwelling mantle plume, where they reassemble. They eventually reassemble at a site of the downwelling mantle plume. When the downwelling plume waning from the bottom of the supercontinent, the colder mantle beneath the supercontinent is replaced by the hotter mantle with a new upwelling plume as an initial thermal structure before the breakup of the old supercontinent. Such a periodic supercontinental cycle and the formation of long-wavelength thermal structures during the cycle have also been observed in 2-D cylindrical models with wide model parameters (Gurnis and Zhong, 1991; Zhong and Gurnis, 1993). The flow reorganization of mantle due to the existence of the supercontinent is carefully examined by Lowman and Jarvis (1993, 1995) using 2-D Cartesian box models with various model parameters such as an aspect ratio of box, the thermal diffusivity and thickness of the continent, and a radiogenic internal heating ratio of mantle. They have concluded that the flow reorganization appears prominently in a model with a wider and thicker supercontinent and smaller thermal diffusivity of lid, higher continental crustal heating, and lower radiogenic internal heating of mantle. Lowman and Jarvis (1996) incorporated rigidly moving continental blocks with a finite thickness in a 2-D Cartesian mantle convection model. They suggested that stresses generated by flow reorganization below the aggregated continents are sufficient to produce continental rifting, and that the subduction of oceanic plates at the margins of a thick continental plate may be the key event in triggering the subsequent continental breakup. 4. Thermal and mechanical interaction between continent and mantle The thermal and mechanical interaction between mantle and continent/supercontinent would significantly affect the evolution of the Earth's mantle. The question as to whether the temperature increase beneath the supercontinent due to the thermal insulating effect of the supercontinent (Anderson, 1982) aids continental rifting and breakup is not clear in geodynamics. The subsequent mantle reorganization that has been hypothesized to occur in response to thermal insulating effect is probably one cause of continental breakup and dispersal (Gurnis, 1988). The concept of thermal blanket envisages the breakup of supercontinents through radiogenic heating (e.g., Gurnis, 1988). Granites contain higher K, U and Th compared to mantle peridotite. Recent studies recognize that substantial volume of arc crust of granitic composition is subducted during the assembly of supercontinents through arc subduction, sediment trapped subduction and tectonic erosion. This tonalite–trondhjemite–granodiorite (TTG) material is dragged down and is thought to accumulate in the mantle transition zone (Senshu et al., 2009). It is possible that the radiogenic elements in the subducted TTG crust heat up the overlying mantle with time to initiate continental rifting and dispersion leading to the opening of oceans. Komabayashi et al. (2009) performed phase assemblage analysis in the system mid-oceanic ridge basalt (MORB)– anorthosite–TTG down to the CMB conditions. Their results show that all these materials can be subducted even up to the CMB leading to the development of a compositional stratification in the D” layer. Numerical modeling studies, however, have not yet considered the higher radiogenic heating rate through subduction of continental crust. Maruyama et al. (2007) proposed that the exothermic reaction during the perovskite and post-perovskite phase transition heats up the cold materials accumulating at the CMB and leads to the formation of superplumes which rise up and disintegrate the supercontinent. Effects of thermal insulation by the supercontinent on the mantle convection are so far investigated by the previous numerical models of mantle convection. They have revealed that horizontal flow due to temperature differences between the supercontinent and the rest of the ‘oceanic’ region produces the large-scale flow in the mantle and large upwelling plumes originating from the CMB beneath the supercontinent (Gurnis, 1988; Gurnis and Zhong, 1991; Zhong and Gurnis, 1993; Nakakuki et al., 1997; Yoshida et al., 1999; Honda et al., 2000; Phillips and Bunge, 2005, 2007; Yoshida, 2010b) (see also Fig. 2). It appears that the temperature increase by the thermal insulation effect sustains for long geological timescales when radiogenic internal heating in the mantle is considered (Yoshida et al., 1999; Honda et al., 2000). The thermal insulating effects and flow reorganization of mantle may cause ‘global mantle warming’ and large-scale partial melting beneath the supercontinent (Coltice et al., 2007, 2009) (Section 8). On the other hand, Lowman and Gable (1999) have suggested that, rather than the thermal insulating effect, the cessation of subduction beneath a supercontinent plays a major role in warming the bottom of the supercontinent. The temperature increases beneath the supercontinent may be caused by upwelling plumes originating from the CMB as a return flow of plate subduction occurring at supercontinental margins (Zhong et al., 2007a). Nakakuki et al. (1997) investigated the effects of a combined system of highly-viscous continental lid and subcontinental regions (i.e., ocean) using a 2-D, long aspect-ratio Cartesian model with M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 multiple phase transitions at 410 km and 660 km and depthdependent properties (i.e., viscosity, thermal expansivity, and thermal conductivity). They showed that beneath the ocean, thermal boundary layer at the CMB is suppressed by the large-scale flow in the highly-viscous lower mantle, whereas beneath the continental lid, a thickened boundary layer developed at the CMB. The large-sized upwelling plumes developed in the lower mantle are concentrated beneath the continental lid. When the position of continental lid and ocean is switched instantaneously during the simulation, the upwelling plumes are concentrated beneath the newly-positioned continental lid due to the mantle flow reorganization. This 2-D numerical model suggests that there is a greater tendency that the supercontinent enhances the longest-wavelength thermal structure in the mantle. During the amalgamation of continental fragments, the subducted oceanic lithosphere of intervening oceans either moves down to the deep mantle or is flattened, becoming stronger as stagnant slabs in the mantle transition zone between ca. 410 and 660 km depth (Fukao et al., 1992; van der Hilst et al., 1997; Fukao et al., 2001; Grand, 2002; Zhao, 2004; C. Li et al., 2008; Fukao et al., 2009; Zhao, 2009). Blobs of these stagnant slabs sink down into the deep mantle and accumulate at the CMB. Zhao (2004) synthesized a P-wave tomographic image for the Western Pacific region, along a transect covering Beijing to Tokyo, where about 1200 km-long stagnant slabs are seen floating in the mantle transition zone. The image shows the presence of a high Pwave velocity anomaly close to the bottom of the mantle and immediately above the CMB which has been interpreted as a ‘slab graveyard’ (Richards and Engebretson, 1992). The recycled oceanic lithosphere at the CMB contributes potential fuel for generating superplumes (Maruyama et al., 2007) which rise from the core–mantle interface to the uppermost mantle, penetrating the mantle transition zone and eventually giving rise to hot spot (e.g., Maruyama et al., 2007). Highly resolved seismic tomography models clarify the configuration and amplitude of the velocity anomaly of two superplumes beneath Southern Africa–Southern Atlantic Ocean and Southern Pacific. For instance, Ritsema et al. (1999) observed that beneath Southern Africa–Southern Atlantic Ocean the low-velocity anomaly extends from the CMB into the upper mantle and has an average velocity that is ~ 1% lower than normal mantle. Meanwhile, Suetsugu et al. (2009) observed that beneath Southern Pacific, the low-velocity anomaly extends 1000 km above CMB and has an average velocity that is ~0.5% lower than normal mantle and smallscale low-velocity anomaly originating at the 1000 km-depth or 660 km phase boundary. Multiple subduction zones promote the rapid amalgamation of continental fragments into supercontinents and also act as major zones of material flux into the deep mantle transporting substantial volume of trench sediments and arc crust through sediment subduction and tectonic erosion (e.g., Santosh et al., 2009; Yamamoto et al., 2009; Santosh, 2010a). Due to buoyancy, the subducted TTG material is stacked in the mid mantle region and may not sink down to deeper levels. 5. Formation of degree-one mantle structure With the advancement in numerical modeling techniques as well as the enhancement in computational power and resource, it is now possible to simulate the highly-resolved 3-D spherical-shell mantle convection model with lateral variation of viscosity and realistic rheology (i.e., temperature- and/or strain-rate dependent rheology) (Zhong et al., 2000; Richards et al., 2001; Yoshida and Kageyama, 2004). Several previous studies have attempted to characterize the effects of supercontinents on the dynamics and structure of the mantle in 3-D spherical-shell geometry (Yoshida et al., 1999; Phillips and Bunge, 2005; Coltice et al., 2007; Phillips and Bunge, 2007; Zhong et al., 2007a; Trubitsyn et al., 2008; Zhang et al., 2009; Yoshida, 2010b). 7 In general, a mantle convection planform with short-wavelength structures and a large number of downwellings may not lead to the assembly of supercontinents because continental blocks would be trapped by different downwellings thus inhibiting collision. When the continental blocks are smaller than the wavelength of mantle flow, it takes a longer time for them to form the supercontinent (Phillips and Bunge, 2007; Zhang et al., 2009) (see also Section 7 for the timescale of supercontinent assembly). Thus, long-wavelength convection is preferred for effective close packing of continental fragments into a supercontinent assembly. Both Rodinia and Pangea are considered to have been largely surrounded by subduction zones (Maruyama et al., 2007) which suggests the presence of major downwellings and upwellings, and leading to kinematic models of supercontinent cycles which require mantle convection of very long-wavelengths at spherical harmonic degree one or degree two (e.g., Monin, 1991; Evans, 2003). The present-day mantle is predominated by degree-two structures that include the seismically fast, cold anomalies around Circum-Pacific and two large seismically slow, hot anomalies beneath Africa and Pacific (Section 4), translated into one cold downwelling zone and two hot upwelling zones (e.g., Maruyama et al., 2007). A mantle convection model with a supercontinent in Earth-like 3D spherical-shell geometry was proposed for the first time by Yoshida et al. (1999). They modeled the supercontinent as an elongated-diskshaped HVL with a thickness of 200 km, covering 30% of the total surface, and imposed it on well-developed, isoviscous mantle convection with the short-wavelength thermal heterogeneity. The mantle is heated externally from the core and internally by radiogenic elements. Their results show that the presence of a supercontinent with the HVL produces large-scale horizontal mantle flow, thereby reorganizing the thermal structure of the mantle interior. Large-scale upwelling plumes arising from the CMB beneath the supercontinent are observed and the thermal structure dominated by the spherical harmonic degree of one (i.e., degree-one convection) develops in the mantle (Yoshida et al., 1999; Yoshida, 2010b). Fig. 4 illustrates a result of 3-D spherical-shell mantle convection model with a rigid, spatially fixed HVL. The mantle is considered realistic convection vigor and a temperature-dependent rheology. The viscosity contrast between the HVL and surrounding mantle is fixed at 102. The result reveals that the very short-wavelength structure is reorganized by a degree-one structure for ~ 500 Myr due to the existence of the supercontinent. Yoshida (2010b) proposed that largescale upwelling plumes produce tensional stresses within the supercontinent with a stress magnitude of the order of 10 MPa, irrespective of model parameters studied here, which is comparable to the strength of the Earth's continent, and thus may be responsible for the subsequent continental rifting and breakup, as previously suggested by the 2-D model of Lowman and Jarvis (1996). It is quite important that the formation of degree-one convection found in Yoshida et al. (1999) may be closely related to the periodicity of supercontinent cycles. Phillips and Bunge (2005) have succeeded in modeling the nondeformable HVL freely floating over the surface by a torque balance method (i.e., calculating Euler poles through force balance and applying repulsive forces to prevent overlap) with a 3-D spherical-shell model and have reproduced the supercontinent cycles in the limited geophysical situations. They demonstrated that degreeone convection that develops after the setup of the disk-shaped rigid cap forms in cases for (1) convection models without a viscosity increase at the 660 km phase boundary and purely basal heating (i.e., no radiogenic internal heating), and (2) convection models with a viscosity increase at the 660 km. The broad, active upwelling plumes due to the highly viscous lower mantle and strong heating from the CMB might trigger the formation of the longest-wavelength thermal structure in the mantle when the supercontinent is imposed. Such upwelling plumes from the CMB have also been envisaged in recent conceptual models (e.g., Maruyama et al., 2007). When degree-one convection is established, the dispersing continental fragments 8 M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 Fig. 4. Time sequence of mantle convection with a rigid, highly viscous supercontinent in three-dimensional spherical-shell geometry with a thickness of 2867 km. The blue and purple isosurfaces of the temperature anomaly δT (i.e., the deviation from horizontally averaged temperature at each depth) indicate − 250 K and + 250 K, and orange indicates the position of elongated disk-shaped supercontinent with a thickness of 250 km, covering 30% of total surface area (note that for clarity, the supercontinent is largely transparent). The white sphere indicates the bottom of mantle (i.e., core–mantle boundary). The supercontinent is instantaneously imposed on the well-developed mantle convection with temperature-dependent rheology. The viscosity contrast between the coldest upper surface boundary (i.e., uppermost part of high viscous lithosphere) and hottest bottom surface boundary (i.e., core–mantle boundary) is 102. The viscosity of supercontinent is 102 times larger than the surrounding ‘oceanic’ lithosphere. The elapsed times are scaled by an Earthlike timescale. The details of numerical methodology and model parameters are found in the work of Yoshida (2010b). migrate towards the antipodal downwelling plumes and reassemble to form a new supercontinent. A subsequent study by Phillips and Bunge (2007) demonstrated that active upwelling plumes caused by the significant basal heat flow from the core disrupt the periodicity of supercontinental cycles. They suggest that the periodic supercontinent cycles are unlikely to occur in the history of Earth. We note here that degree-one convection is observed when supercontinent is not imposed in the mantle convection with temperature-dependent rheology. The viscosity of mantle materials strongly depends on the temperature, pressure, and composition, among other parameters (e.g., Karato, 2008). It has been accepted that mantle convection patterns with temperature-dependent rheology are classified into three regimes from the 2-D convection studies (Solomatov, 1995; Solomatov and Moresi, 1997). Recent advances on computational techniques and resources facilitate the numerical simulation of 3-D spherical mantle convection with strong temperature-dependence of viscosity (e.g., Yoshida and Kageyama, 2004; McNamara and Zhong, 2005a; Roberts and Zhong, 2006; Yoshida and Kageyama, 2006; Zhong et al., 2007a; Yoshida, 2008). Fig. 5a illustrates a regime diagram for mantle convection with purely basal heating (i.e., no internal heating rate) and with a linearized viscosity form (i.e., η ∝ exp(ET), where η is the viscosity and E is the activation energy that controls the degree of viscosity's temperature (T)dependence) (Yoshida and Kageyama, 2006). The degree-one convection has been produced by some numerical models that considered temperature-dependent rheologies. When internal heating is included with moderately temperature-dependent viscosity, degree-one convection occurs (McNamara and Zhong, 2005a). On the other hand, Yoshida and Kageyama (2006) have shown that degree-one convection occurs even without internal heating when temperature-dependence of the viscosity is moderately strong. This degree-one convection pattern belongs to the ‘sluggishlid regime’ (Solomatov, 1993) or ‘transitional regime’ (Solomatov, 1995), in which the moderately highly viscous layer around the top surface boundary slowly moves with less Rayleigh–Taylor instability and eventually sinks into the mantle, and the resultant mantle convection has longest-wavelength thermal structure. Considering the Rayleigh number of the Earth's mantle and the viscosity contrast between the lithosphere and underlying mantle, the Earth's mantle presumably falls into sluggish-lid or ‘stagnant-lid’ regimes (orange box in Fig. 4a). In the stagnant-lid regime, a deformable, immobile lid develops near the surface boundary of mantle convection. In order to let the surface plate-like motion, we need to impose the visco-plastic rheology (see Section 12). Degree-one convection is found in mantle convection with a more realistic viscosity form (i.e., Arrhenius-type form, η ∝ exp(E / T)) and/ or with three end-member heating modes, i.e., purely basal heating (i.e., no internal heating rate), mixed heating (i.e., basal and internal heating) (Yoshida, 2008) (Fig. 5b–c), and high internal heating (McNamara and Zhong, 2005a) modes. Therefore there is a highly possibility that degree-one convection is one of the basic structure of mantle convection without the surface heterogeneities such as continent and plate motions. However, as reported by Yoshida (2008), it seems that the geophysically relevant degree-two convection with sheet-like downwellings (i.e., Circum-Pacific subduction zone) and two upwelling plumes (i.e., superplumes) is not observed in the mantle convection model with real convective vigor and without the surface heterogeneities. On the basis of a previous numerical result that a supercontinent spends most of the time over the cold, downwelling mantle (Gurnis and Zhong, 1991), Zhong et al. (2007a) considered in their 3-D spherical model that the continental fragments migrate towards the downwellings in degree-one mantle convection and settle into the downwellings for a certain period of time. When a disk-shaped HVL is imposed on well-organized degree-one convection, the oceanic material begins to subduct at a margin of the HVL and downwellings beneath the supercontinent eventually vanishes. Subsequently, new upwelling plumes tend to originate at the CMB for a short timescale (~50 Myr) and the degree-two thermal structure with old and new upwellings develops in the mantle (see Zhong et al., 2007a for details). Following their scenario, when the continental fragments fully disperse on the Earth so that the mantle does not 'feel' the existence of continents, the degree-two pattern may go back to degree-one structure. However it has not been resolved yet whether or not mantle convection is dominated by degree-one when plate motions and continental drift continuously occur throughout out the Earth's history. There is a possibility that the present-day degree-2 pattern is realized both by the history of plate motions and large-scale thermochemical upwelling plumes (McNamara and Zhong, 2005b). The thermochemical upwelling plumes or thermochemical piles stratified in the deep lower mantle, reproduced in the mantle convection model (Tackley, 1998; Kellogg et al., 1999; McNamara M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 9 Fig. 5. (a) Regime diagram showing three convection regimes with a linearized viscosity form and with varying Rayleigh number (Rabot) and the viscosity contrast across the shell (γ); the mobile-lid (circles), the sluggish-lid regime (triangles), and the stagnant-lid regimes (squares). Solid, open, and shaded symbols show the results from 3-D spherical-shell models by Yoshida and Kageyama (2006) and Ratcliff et al. (1997), and a 3-D Cartesian box model by Trompert and Hansen (1998b), respectively. The regime boundary (dashed curve) between convection regime and no-convection regime is referred with the reviews by Schubert et al. (2001). For the definitions of three regimes, see text, and also see Yoshida and Kageyama (2006) and papers herein. Dashed line shows the approximate boundaries that separate the three convection regimes. An orange box indicates a parameter range appropriate to the Earth's mantle. (b–c) Snapshots of convection pattern with an Arrhenius-type viscosity form and with weak and moderate viscosity contrasts (γ = 102 and 104, respectively). The convection models with (b) purely basal heating and (c) mixed (i.e., basal and internal) heating modes are shown. The blue and yellow isosurfaces of the temperature anomaly indicate colder and hotter regions. For cases with γ = 104 (degree-one convection) cross sections of mantle temperature are shown. and Zhong, 2005b), are expected to be a part of the superplumes beneath South Atlantic–South Africa and South Pacific regions (Larson, 1991; Maruyama, 1994; Maruyama et al., 2007). The thermochemical piles with abundance of radioactive elements produces ‘superswells’ i.e., the broadly elevated topography at the earth's surface (McNutt, 1998). 6. Long-wavelength thermal structure in the mantle The longer thermal structures in the mantle may shorten the period of supercontinent cycle, because continental blocks would smoothly drift on the Earth's surface without being inhibited by upwelling and downwelling plumes. Theoretical and numerical studies on dynamics of planetary mantle convection demonstrate that the effects of depth- or pressure-dependent properties such as viscosity and thermal expansion coefficient (or thermal expansivity) contribute to the predominance of the long-wavelength thermal structure in the planetary mantle, because they reduce convective vigor effectively (Schubert et al. (2001); Yuen et al. (2007)). Since the mid-1990s, several numerical models of mantle convection within a 3-D geometry have investigated the effects of depth-dependent viscosity on the mantle convection pattern with real convective vigor (Zhang and Yuen, 1995; Bunge and Richards, 1996; Bunge et al., 1996; Tackley, 1996; Zhang and Yuen, 1996; Bunge et al., 1997). On the other hand, high pressure experiments of mantle rocks have recently proposed a substantial decrease in the thermal expansion coefficient with depth. Katsura et al. (2009) suggested on the basis of their experiments that the thermal expansion coefficient of MgSiO3 perovskite generally decreases with increasing pressure and that the Anderson–Grüneisen parameter (Anderson, 1967; Birch, 1968), δT, of MgSiO3 perovskite is 6.5. As for the comparison of this δT with δT derived from the previous high-pressure experiments for various mantle rocks, see, for instance, Katsura et al. (2009) and Schubert et al. (2001). The decrease in the thermal expansivity with pressure is expected to reduce the buoyancy force of mantle convection in the deep mantle. Furthermore, it is expected to offset the adiabatic temperature gradient because the degree of adiabatic heating/cooling is linearly proportional to the thermal expansivity (e.g., Jarvis and McKenzie, 1980). Here we attempt a simulation of the mantle convection models with both the depth-dependent viscosity and thermal expansivity. Fig. 6a and b shows the depth profiles of viscosity and thermal expansivity, respectively, used in the simulation. The parameters used here are V which controls the maximum viscosity contrast and δT which controls the thermal expansivity contrast across the mantle. We considered a viscosity profile with different viscosity contrast 10 M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 Fig. 6. Depth profiles of (a) viscosity and (b) thermal expansivity variations used in the model. (a) Dotted, thin, and thick lines represent V = 0, ln 30, and ln 100, respectively. (b) Dotted, thin, and thick lines represent δT = 0, 3, and 6.5, respectively. (c–k) Plan views of temperature field at middle depth of mantle (1433 km) for different combinations of V and δT (see the values embedded in the figure). (l–n) Cross sections of temperature and velocity fields for three cases that correspond to plan views of (c), (i), and (k), respectively. The cross sections are cut along the great circles drawn by dashed lines marked 'A'–'D', 'E'–'H' and 'I'–'L' in the corresponding plan views. The contour interval in the temperature plots is 250 K. The viscosity and thermal expansivity are fixed at 3 × 1021 Pa s and 3 × 10− 5 K− 1 at the top surface boundary. The mantle is modeled as a fluid with an infinite Prandtl number within a 3-D spherical shell geometry with a thickness of 2867 km, and that it is heated from the bottom and from within. An extended-Boussinesq approximation (Christensen and Yuen, 1985) is applied to the mantle fluid. Free-slip, impermeable, and isothermal conditions are imposed on the top and bottom boundaries. The Rayleigh number and the dissipation number are 1.91 × 107 and 0.67. The dimensionless internal heating production rate per unit mass due to radioactive decay is 9.19, which corresponds to the internal heating rate at the present day (3.50 × 10− 12 W/kg) on the basis of the abundance of the radioactive elements in chondritic meteorite (Turcotte and Schubert, 2002). M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 across the mantle (Fig. 6a). The peak viscosity is assumed to be located at the depth of 2000 km, in accordance with the results obtained by the joint inversion of convection and glacial isostatic adjustment data (Mitrovica and Forte, 2004) and other studies (Ricard and Wuming, 1991; Forte and Mitrovica, 2001). It is assumed that the viscosity below depths of 2000 km, including at the hot thermal boundary layer, decreases because of the temperature dependence of viscosity. In the present study, we investigate the cases with V = 0, ln 30, and ln 100; the range of these values was selected on the basis of post-glacial rebound analyses (e.g., Lambeck and Johnston, 1998; Peltier, 1998) and geoid inversion studies (e.g., King, 1995). Meanwhile, as for the thermal expansivity, we test the cases with both the ‘classical’ value of δT = 3 used in the previous numerical studies (e.g., Anderson, 1987; Zhang and Yuen, 1987; Hansen et al., 1993; Hansen and Yuen, 1994) and a ‘new’ value of δT = 6.5. The resultant thermal expansivity contrasts between the top and bottom surfaces are ~ 0.2 and ~ 0.04, respectively (Fig. 6b). The results of the numerical calculations for the nine cases representing different combinations of V = 0, ln 30, ln 100 and δT = 0, 3, and 6.5 are presented in Fig. 6c–n. For each case, calculations were repeated until the output parameters such as root-mean square velocity in the mantle, volume-averaged temperature in the mantle, and top and bottom heat fluxes reached a statistically steady-state. The case with the constant viscosity and thermal expansivity (i.e., V = 0 and δT = 0) shows the thermal structure with quite short wavelengths and numerous upwelling and downwelling plumes that exist in the mantle after the statistically steady state was reached (Fig. 6c and l). This well-developed convection pattern from the first case was used as the starting scenario for the calculations for the other eight cases. We first focus on the effect of depth-dependent viscosity on the convection pattern. Fig. 6c, d, and e shows the cases with V = 0, ln 30, and ln 100, respectively, and a constant thermal expansivity. Even when the highest viscosity contrast is imposed (i.e., V = ln 100), the convection pattern shows a quite short wavelength thermal structure (Fig. 6e). This convection pattern is quite different from those with a viscosity profile with a step-increase of viscosity at 660 km and without the gradual decrease of the viscosity below 2000-km depth (Bunge et al., 1996; Zhong et al., 2000): The convection pattern shifts to larger wavelengths and is re-organized into well-established convection cells with stable large-sized upwelling plumes originating at the CMB and sheet-like plumes downwelling from the top surface. Focusing next on the cases with V = 0 (Fig. 6c, f, and i), we find that the number of large-sized upwelling plumes is reduced with increasing δT. When δT = 6.5 is considered (Fig. 6i and m), the convection pattern drastically changes compared with the case where δT = 3 (Fig. 6f). The convection pattern shifts to larger wavelengths and is re-organized into well-established convection cells with two stable large-sized upwelling plumes and a ‘great-circle’ downwelling. Also, when other values of V (V = ln 30 and ln 100) are considered, there is a tendency for the number of upwelling plumes to decrease with increasing δT. The large-sized upwelling plumes are increased in width by the effect of depth-dependent viscosity. We note that with increasing δT, downwelling plumes tend to effectively ‘subduct’ into the deep lower mantle. This implies that in the real Earth, the largely reduced thermal expansivity with depth facilitates the subduction of plates into the deep lower mantle and resultant large-scale flow circulations in the whole mantle (e.g., van der Hilst et al., 1997) (for instance, ‘M’ in Fig. 6m). Depth-dependent viscosity alone does not produce such a large-scale mantle circulation, even when the largest viscosity contrast is considered (Fig. 6e). As stated in Section 5, the temperature-dependent viscosity also leads to the long-wavelength structure because of reduced thermal instability in the shallow part of the mantle (Fig. 5), while the depthdependent properties lead to a long-wavelength structure because of the reduced thermal instability at the bottom of the mantle. There 11 may be some other factors that define the long-wavelength thermal structure in the mantle. For instance, theoretical and numerical analyses demonstrate that the low-viscosity asthenosphere (channel) wedged between the highly viscous lithosphere and underlying mantle promotes a long wavelength flow in the mantle (Busse et al., 2006; Lenardic et al., 2006; Höink and Lenardic, 2008). Single, spatially-isolated plumes rising from the deep mantle beneath the Hawaii (e.g., Wolfe et al., 2009) and Iceland (e.g., Bijwaard and Spakman, 1999) hotspots have been detected by seismic tomography studies. However, in general, the origin of the Earth's hotspot plumes remains a controversial issue in geodynamics. The large-sized upwelling plumes observed in Fig. 6, which have been quite steady during the geological timescale, may be one of the important keys that could resolve the question regarding the origin of the isolated plumes, regardless of whether convection in the deep mantle is dominated by compositional driving force or not. The largely reduced thermal expansivity in the deep mantle may allow the thermo-chemical ‘megaplume’ (Tackley, 1998) or thermochemical ‘dome’ or ‘piles’ (Tackley, 1998; Davaille, 1999; Kellogg et al., 1999; Tackley, 2000a; Davaille et al., 2002; Jellinek and Manga, 2002; Tackley, 2002; Jellinek and Manga, 2004; McNamara and Zhong, 2005b) to be more stable. The degree-two convection pattern for the case with V = 0 and δT = 6.5 (Fig. 6i and n) is similar to the Earth's convection pattern characterized by two large-scale upwellings like Earth's superplumes beneath the South Pacific and South Atlantic–South Africa and one ‘great-circle’-downwelling in the Circum-Pacific, even when the mantle convection studied here is purely dominated by the thermal buoyancy. It seems that such great-circle-downwelling is also observed in the models with relatively weak temperature-dependent viscosity and relatively low yield stress (van Heck and Tackley, 2008; Yoshida, 2008). The planetary-scale, long downwelling like the Circum-Pacific subduction zone are selfconsistently reproduced by the mantle convection model even with more realistic geophysical conditions (see Section 12). 7. Timescale of the supercontinent assembly Although it is well established that supercontinent cycles operated throughout the Earth history, the timescales of their assembly and dispersal are not well-constrained. A stable configuration is hypothesized for Columbia between 1800 and 1500 Ma (Meert, 2002) which assigns a time gap of 450–500 Myr prior between the breakup of Columbia and the assembly of Rodinia. On the other hand, if we take into consideration the hypothetical supercontinent Pannotia at ca. 600 Ma (Dalziel, 1991), it leaves only 150 Myr between the birth of this supercontinent and the demise of Rodinia and 300 Myr before the formation of the subsequent supercontinent Pangea. Senshu et al. (2009) evaluated the history of supercontinents based on a number of parameters such as the surface configuration based on paleogeographic reconstructions, super downwelling events as suggested by the rapid amalgamation of the majority of continents, and mantle dynamics. These parameters predict that the only supercontinent that was assembled within a short period of time was the Paleoproterozoic supercontinent Columbia. Although Gondwana and Pangea assemblies do not tally with this predicted model in Senshu et al. (2009) and were therefore considered together by these authors, these assemblies have distinct mountain building events and clearly represent two different supercontinental assemblies. In a speculative model, Senshu et al. (2009) suggested that the precise time when a younger true supercontinent might have existed in terms of mantle dynamics is around 340 Ma. According to their model, the lifespan of supercontinents becomes longer in the younger Earth, a phenomenon assigned to a decrease in the rate of heat generation of the subducted TTG materials with time. However, this model requires further 12 M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 evaluation as there is more subduction erosion with time, and this factor outweighs the decreasing heat production. Numerical models with mobile continents/supercontinents can address the timescale of the assembly of supercontinents. Phillips and Bunge (2007) demonstrated that using a 3-D spherical-shell model with isoviscous mantle, three pieces of continental fragments whose size broadly corresponds to 10% of the whole surface of the globe, tend to assemble for an interval of 300–500 Myr, which is comparable to or shorter than the actual supercontinental cycles, 500–700 Myr (e.g., Silver and Behn, 2008). Zhang et al. (2009) demonstrated that using a 3-D spherical-shell model with temperature-dependent rheology and deformable continents, continents assemble to form a supercontinent for ~ 250–650 Myr that depends on the wavelength of mantle structure. This timescale also seems to be comparable to or shorter than the actual supercontinental cycles. Because the timescales obtained from numerical simulation models depend on model parameters that control flow speed (in particular, rheological parameters), it is difficult for us to precisely compare the timescales of the assembly of supercontinents obtained by different numerical models with the actual supercontinent cycles. However, the regularity of the assembly of supercontinents can be resolved by the numerical simulation. One of the outstanding problems in this context is that the effects of the size of supercontinent on the behavior of mantle convection and supercontinent cycles. Phillips and Bunge (2007) have shown that when six continental fragments whose size is 5% of the whole surface of the Earth are considered, a much larger time interval (~ 1500 Myr) is required to form the supercontinent. This implies that the size of supercontinents/continental fragments significantly affects the timescale of supercontinent cycles. There is a possibility that the timescale of supercontinent cycles may be closely related to the timescale of generation of strong upwelling plumes from the deep mantle. Previous numerical simulation models suggest that the timescale of the plume generation and resulting mantle reorganization is much shorter than that of the actual supercontinent cycle. Yoshida et al. (1999) and Honda et al. (2000) have concluded that using an isoviscous mantle model, welldeveloped upwelling plumes originate from the CMB beneath the highly viscous supercontinent over a timescale of 200–400 Myr when real convection vigor is considered in the model. This timescale may be shorter than the actual supercontinent cycles. However, if this timescale is rescaled by the Earth timescale using a typical speed of motion of Earth's plates, the duration may be further shortened by a factor of two or three, that is, ~ 50–100 Myr. On the other hand, Zhong et al. (2007a) concluded that using a temperature- and depthdependent viscosity model, the transition from degree-one convection to degree-two convection is observed with a timescale of ~ 50 Myr. This implies that mantle reorganization may occur within a short time after a supercontinent is assembled. 8. Supercontinent breakup A related topic of ongoing discussion is whether the initiation of continental breakup is caused by active rifting initiated by mantle plumes originating in the deep mantle (e.g., Morgan, 1983; Richards et al., 1989, 1991; Storey, 1995; Dalziel et al., 2000; Condie, 2004), by passive rifting associated with a pre-existing weak zone (e.g., Ruppel, 1995) or by a combination of both these phenomena. Some geological evidence does not support a plume origin for the breakup of Pangea (e.g., McBride, 1991; McHone, 2000). Coltice et al. (2007) tested the hypothesis that the assembly of supercontinents would force a largescale thermal structure and therefore drive the underlying mantle to higher temperatures by using a numerical model. The position of the continents was fixed and an equilibrium temperature field was computed by stacking the temperature fields over several billion years in order to obtain a statistical steady state. Their results show that the sub-continental lithospheric mantle temperature correlates inversely with the number of continents. Thus, with a single supercontinent, the convection planform is dominated by spherical harmonic degree one and the temperatures are 100 K higher than those with dispersed continents. Even when the convective parameters were changed, the large temperature excursion observed in the supercontinent configuration is maintained. The convection modeling in an internally heated mantle by Coltice et al. (2007) led them to conclude that the assembly of continents into supercontinents would naturally lead to ‘global mantle warming’ without the involvement of hot active plumes. However, this model is considered as one of the two end member models to explain the formation of continental flood basalts that occur over a supercontinent which are characterized by wide and diffusive magmatism and a lower rate of magma supply. The other end member model relates to plume-derived continental flood basalts that are characterized by a very brief and high rate of magma supply over a restricted and radiating area followed by continuous hotspot activity. A robust example for plume-related mafic magmatism which ultimately broke-up the globe's first coherent supercontinent (Columbia) has been evaluated by Hou et al. (2008). They synthesized data on the 1.3–1.2 Ga fan-shaped Mackenzie dyke swarm and other similar aged dyke swarms in the Canadian Shield which constitute the sub-swarms of a Late Mesoproterozoic giant radiating dyke swarm. These dyke swarms also correlate with the Late Mesoproterozoic mafic dyke swarms in Australia and East Antarctica which constitute additional sub-swarms of the giant radiating dyke swarm. Hou et al. (2008) proposed a Late Mesoproterozoic mantle plume at the focal area of the giant radiating dyke swarm between North America and the landmass comprising West Australia–East Antarctica (see also, Ernst and Buchan, 2003). They suggested that this mantle plume triggered the continuous extension at ca. 1.3–1.2 Ga and voluminous mafic magma emplacement, which extended into much of the Columbia supercontinent, and led to its final fragmentation. It must be noted that rifts started developing in the Columbia assembly immediately after its amalgamation as indicated by the emplacement of several suits of mafic dykes in various constituent fragments of this Paleoproterozoic supercontinent (Rogers and Santosh, 2009). A protracted event of mafic magmatism during the Paleoproterozoic and early Mesoproterozoic recorded from this continental assembly would probably relate to a global mantle warming process, as envisaged in the model of Coltice et al. (2007). However, the ultimate fragmentation of the supercontinent in the latest Mesoproterozoic probably involved plume-related activity as indicated by the formation of radiating giant dyke swarms and voluminous mafic magmatism. The fragmentation history of Columbia might thus represent a combination of prolonged global warming and incipient rifting culminating in a plume-triggered final fragmentation. We therefore propose that the two end member models of supercontinent break-up (global mantle warming and plume activity) might have operated in conjunction in the case of at least some of the supercontinents in Earth history. A specific case of supercontinent break-up was evaluated by Storey (1995) with regard to the disintegration of the Gondwana supercontinent which started at about 180 Ma ago. The magmatic and tectonic signatures along the proto-Pacific margin of Gondwana record important changes in subduction zone forces during its initial fragmentation. The absence of subduction along the Neotethyan margin of Gondwana, together with the observation that initial rift formed at right angles to the active subduction margin as demonstrated by Storey (1995) suggest a potential active role for mantle plumes in the initial separation of the East and West blocks of the supercontinent. The initial plume-related magmatism is represented by the ca. 182 Ma Karoo province and the final breakup is constrained at ca. 156 Ma. The large extent of magmatic provinces also suggests that the Gondwana supercontinent was underlain by a broad shallow M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 region of higher temperature, rather than a single plume. Although not considered as a mantle plume, this anomaly is broadly similar to the thermal blanket effect (Anderson, 1982, 1994). Storey (1995) concluded that either a plume point source or a regional heat anomaly could have weakened the Gondwana lithosphere leading to the high magma production rates and local rifting leading to the fragmentation of the supercontinent into the East and West Gondwana blocks. In summary, the role of plumes in initiating the disintegration of supercontinents is emphasized in the case of Gondwana also. Similar models involving the role of mantle plumes have also been proposed for the fragmentation of the Neoproterozoic supercontinent Rodinia (Maruyama et al., 2007 and references therein) and Pangea (Isozaki, 2009) with significant implications on surface environment and life history of the planet. 9. Stability and longevity of the continent Based on information from heat flow, geochemistry and the relative delay times of seismic waves in different settings, Jordan (1975) proposed the ‘tectosphere’ (highly depleted relatively low density upper mantle layer) model, in which a zone moves with the motion of the plate lying beneath the old continental shields and is expected to be up to 400 km thick. The depth to which the tectospheric mantle extends is still a hotly debated topic, although most tomographic images of the cratons show high velocity roots 13 extending to at least 200 km depth, and in some cases to depths greater than 300 km (Grand, 2002; Gung et al., 2003; Romanowicz, 2009) (Fig. 7). The tectosphere, also known as continental keel or cratonic root, is thus a rigid, cold and chemically distinct raft that supports the continental crust (Jordan, 1988) and occurs only beneath old cratons. With time, this keel is extensively eroded by younger magmatic and subduction-erosion processes as in the case of the North China Craton (Kusky et al., 2007; Santosh, 2010a). The longevity of the continental root in the geological timescale is closely related to the rheological properties (i.e., viscosity and strainrate of materials) not only of the continental root itself but also of the surrounding mantle. As the viscosity contrast between the continental root and oceanic upper mantle (including low-viscosity asthenosphere) is larger, it takes longer time for the continental root to cause Rayleigh–Taylor instability (at the bottom of the keel) or convective erosion (mainly at the subduction zone) and to be involved into the surrounding mantle. It has been accepted that the continental root is depleted in volatile elements due to a high degree of partial melting in which case it is dehydrated, and has high viscosity (Pollack, 1986; Karato, 2008, 2010b). However, considering the plausible values of the degree of water content and activation volume of dry olivine, the viscosity of continental roots is likely to be no more than the order of 1021 Pa s, even when the combined effects of differences in temperature and water content on the viscosity are considered (Karato, 2010b). The viscosity of continental roots may be two or three orders Fig. 7. Distribution of tectosphere defined by S-wave tomography (Grand, 2002). Note the selective occurrence of high velocity anomaly under cratons older than 2.0 Ga and downward thinning of tectosphere. Maximum depths are about 300 km. The continental crust is only about one tenth of the continental keel. 14 M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 larger than that of the surrounding asthenosphere with orders of 1018–1019 Pa s, on the basis of postglacial rebound analyses (e.g., Bills and May, 1987; Okuno and Nakada, 2001) and tectonic observations (e.g., Gordon, 2000). The resultant viscosity contrast between continental root and the surrounding mantle is, therefore, 102–103, and the assumption of continents as highly viscous bodies made in the numerical models discussed above seems reasonable. In a recent study based on the Kaapval craton of southern Africa, Peslier et al. (2010) concluded that the deep roots of cratons are protected by a layer of unusually dry minerals, and provide sufficient viscosity contrast with underlying asthenosphere. Geochemical studies have revealed that the continental lithospheres (or cratonic root) beneath some of the old cratons have survived for periods more than 3 billion years in spite of later tectonic disturbance (Carlson et al., 2005). However, the numerical model with HVL described in the previous section cannot address the mechanism for stability and longevity of the cratonic lithosphere, because the model supercontinent does not deform by the convection force with time. In contrast, in some numerical studies, a continent is modeled as a highly viscous, compositionally buoyant fluid in a 2D Cartesian model (e.g., Doin et al., 1997; Lenardic and Moresi, 1999; Shapiro et al., 1999; Lenardic et al., 2003). One of their significant observations is that continental roots must be 1000 times as viscous as the surrounding mantle in order to stabilize the roots with compositionally buoyant materials over geological timescales (Doin et al., 1997; Lenardic and Moresi, 1999). This viscosity contrast is in the range of that obtained by the mineral physics described above. Lenardic et al. (2003) presented a 2-D numerical model with three chemically distinct materials: continental crust, continental mantle lithosphere and bulk mantle. The visco-plastic rheology is imposed in the oceanic lithosphere to realize the plate boundary and plate motion. They concluded that a high yield stress for cratonic lithosphere relative to the oceanic lithosphere is found to be an effective and robust means for providing tectonic stability of cratonic crust and the relative longevity of deep cratonic lithosphere. They also suggested that the degree of yield stress variations between cratonic and oceanic lithosphere required for the stability and longevity can be decreased if cratons are bordered by continental lithosphere that has a relatively low yield stress (i.e., mobile belt). This means that the mobile belts protect cratons from being deformed for certain periods of geologic timescale. The longevity of cratonic lithosphere has not been investigated by a 3-D numerical model so far. Yoshida (2010a) demonstrated for the first time that using a 3-D spherical-shell model, the deformable and mobile continental lithosphere is stable only for b1 billion years, that is, the continental materials tend to gradually merge into the oceanic plates and underlying mantle with time, when the viscosity contrast between cratonic lithosphere and surrounding oceanic plates is 102 (see Section 11 for details of this result). However, their previous model (Lenardic et al., 2003; Yoshida, 2010a) did not consider the origin and growth of continental crust. 10. Origin and growth of the continental crust Continental growth occurs predominantly through the addition of juvenile crust by arc magmas (e.g., Rapp and Watson, 1995; Nair and Chacko, 2008; Windley and Garde, 2009). In the early history of the Earth, the parallel collision of intra-oceanic arcs was an important process to build embryonic continents (e.g., Santosh et al., 2009; Windley and Garde, 2009). Recent studies from Archean terranes in different parts of the world have offered important clues for the process of amalgamation of composite arcs (Komiya et al., 2002, Santosh et al., 2009 and references therein). A modern analogue for the Archean process is the western Pacific domain where 60–70% of island arcs are concentrated in the oceanic domain. Archean TTG magmatic suites represent the oldest coherent pieces of felsic continental crust. In a recent study, Nair and Chacko (2008) presented results from long-duration dehydration melting experiments which suggest melting depths of N48 km. This result is consistent with the early crustal evolution models which propose melting at the base of oceanic crust or oceanic plateaus to explain the origin of early Archean continental crust. Under the high Archean geothermal gradient, the subducted oceanic crust would melt to produce TTG. The model proposed by Nair and Chacko (2008) differs from earlier models of subduction initiation in that the subduction of oceanic lithosphere occurs through the protomantle lithosphere at the base of the newly formed oceanic plateau crust. Their model can thus effectively explain the origin of subduction systems, TTG, TTG-mafic and/or ultramafic magma association, stabilization of continental crust and the broadly coeval formation of cratons and their lithospheric roots. It is now generally understood that plate tectonics not only creates but also destroys Earth's continental crust, both occurring mostly at subduction zones, the former by arc magmatic creation and the latter by subduction removal (Stern, 2008; Stern and Scholl, 2010) (Fig. 8). Since subduction zones are the only major routes through which material is returned to great depths within the Earth, the mass flux through convergent plate margins has received considerable attention in recent studies particularly to evaluate the origin and growth of the continental crust. Clift and Vannucchi (2004) classified convergent plate margins into those showing long-term landward retreat of the trench and those dominated by accretion of sediments from the subducting plate. They proposed that tectonic erosion is favored in regions where convergence rates exceed 6 ± 0.1 cm yr− 1 and where the sedimentary cover is less than 1 km. In regions of slow convergence (b 7.6 cm yr− 1) and/or trench sediment thickness exceed 1 km, preferential accretion occurs. Large volumes of continental crust are subducted at both erosive and accretionary margins. At present, creation and destruction of continental crust are either in balance (ca. 3.2 km3 yr− 1) or more crust is being destroyed than created. The creation–destruction balance changes over a supercontinent cycle, with crustal growth being greatest during supercontinent break-up due to high magmatic flux at new arcs and crustal destruction being greatest during supercontinent amalgamation due to subduction of continental material and increased sediment flux due to orogenic uplift (Stern and Scholl, 2010). Ongoing subduction erosion also occurs at the leading edges of dispersing plates, which also contributes to crustal destruction, although this is only a temporary process. The widely held view that the volume of continental crust has increased over time through plate tectonic activity (e.g., Hurley and Rand, 1969) has thus been challenged, with the possibility that the volume has in fact decreased. The mechanism of crust production and growth should be incorporated in a future numerical model in order to test the concept that plate tectonics creates and destroys continental crust with time, and also to evaluate whether more volume of crust is being destroyed than that is being created as suggested in some recent works discussed above. Such a study can also test whether the geologically suggested episodic emergence of supercontinents is realized in the numerical model. The thermal effect of accumulation of continental crust with compositionally buoyant materials on mantle convection has been previously evaluated in a series of papers (Lenardic and Kaula, 1994, 1995, 1996; Lenardic, 1997; Moresi and Lenardic, 1997; Lenardic, 1998; Moresi and Lenardic, 1999). Convective overturn of the mantle causes chemically light crust to thicken above the mantle downwelling and surface heat flux above the thick crust is lower than that above the mantle (Lenardic and Kaula, 1995). A number of such studies are carefully reviewed in Schubert et al. (2001). Walzer and Hendel (2008) performed simulations of mantle convection with the growth of a continent using a 3-D spherical-shell model with real convective vigor (i.e., Rayleigh number), temperature- and pressure dependent rheologies, and viscoplastic yield stress (i.e., maximum of rupture strength) of the lithosphere to induce the M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 15 Fig. 8. Cartoon illustrating tectonic erosion and sediment trapped subduction (compiled after Clift et al., 2009; Senshu et al., 2009). plate-like motion in the lithosphere. Their modeled continent was not artificially imposed on the convection model in contrast to models discussed in previous sections and temporally evolved with mantle convection by considering the chemical differentiation and redistribution of incompatible elements, U, Th, and K. The distribution of radiogenic elements in the mantle is represented by the tracer particles. In their model, chemical differentiation generates the depleted MORB mantle from the primordial mantle, if the conditions for partial melting are fulfilled, and a large amount of the differentiated primordial mantle is assumed to produce the new continental crust with higher abundances of incompatible elements (see, Walzer and Hendel, 2008, for the details of methodology). With a wide range of parameter space for convective vigor and yield stress of the lithosphere, their result supports evidence that continental crust has grown progressively with time in an episodic manner (Condie, 1998). However, the creation–destruction balance of continental crust is not presented in their work. Despite this, among mantle convection models proposed so far, this work has produced the first realistic distributions of continents. 11. A preliminary Wilson Cycle model A 3-D mantle convection model with mobile and deformable continents and visco-plastic oceanic plates has not been proposed so far, primarily because of the difficulty to accurately solve the advection of a chemically distinct body (i.e., continent/supercontinent) without numerical diffusion (i.e., artifact diffusion). In particular, serious hurdles remained for simulations that attempted tectonic processes occurring over geological timescales. Overcoming this difficulty, Yoshida (2010a) has recently presented a new numerical model of mantle convection with a deformable, mobile continental lithosphere within 3-D regional spherical-shell geometry (Fig. 9). In this preliminary model, a supercontinent is instantaneously imposed on well-developed mantle convection. The supercontinent has a spherical-square-shaped cratonic lithosphere with a uniform thickness of 250 km (dark orange regions in Fig. 9), which is comparable to the tectosphere thickness inferred from seismological analysis and geodynamic evidence (Section 9). It covers ~ 30% of the total surface area of the model domain, and is initially separated into four pieces by the weak (low-viscosity) continental margins (WCMs; light orange regions in Fig. 9). In order to ensure the conservation of the continental materials, a process of advection with approximately zero chemical diffusion is solved by a kind of tracer particle method (e.g., Christensen and Hofmann, 1994). A viscosity increase due to the 660-km phase transition is assumed to be 30 on the basis of the results of postglacial rebound analyses (Mitrovica and Forte, 1997; Lambeck and Johnston, 1998; Peltier, 1998). The maximum viscosity contrast between the continents and the oceanic lithosphere is 102 at the top surface boundary, which is within the range predicted from rheological experiments for dehydrated continental root materials relative to a reference mantle (Hirth and Kohlstedt, 1996; Karato, 2010b; Peslier et al., 2010) (Section 9). The viscosity contrast between the cratonic lithosphere and the WCM is fixed at ~ 101.5. The viscoplastic oceanic lithosphere and the associated subduction of oceanic plates are incorporated in this model. Earth-like continental drift and the characteristic thermal interaction between the mantle and a continent are observed in this preliminary numerical model (Fig. 9). By ca. 100 Myr, the supercontinent begins to break up because of the tensional stress induced by mantle upwelling, deforms with time because of the surrounding convecting force, and gradually moves towards a side boundary of the model domain (note that in this model, the continents reaching the lateral boundary of the model domain appear at the opposite lateral boundary due to the periodic boundary condition). Fig. 10 illustrates an early stage of the composition-field (representing 1 for continental material and 0 for oceanic material) and the corresponding temperature anomaly (i.e., the deviation from horizontally averaged temperature at the same depth) δT at a shallower mantle (depth of 358 km) to observe the thermal insulating effect. The regional upwelling plumes with high buoyancy and maximum temperature anomaly of ≥+300 K originating from the CMB are remnants of the initial thermal condition (indicated by arrows ‘A’ in Fig. 10), while broad hot regions with a temperature anomaly of approximately +100 K (‘B’ in Fig. 10) are due to the thermal insulating effect by the existence of a supercontinent and the radioactive elements in the mantle. The beginning of the continental breakup is likely to be mainly responsible for the tensional force of these broad hot regions. The tensional force generated by upwelling plumes may additionally help the continental rift and breakup. The temperature anomaly of +100 K generated by the thermal insulating effect observed here is consistent with that proposed by Anderson (1982). Fig. 9 shows that by ca. 150 Myr, Continent-A (or Continent-B) is completely separate from Continent-C (or Continent-D), while Continent-A relatively stays with Continent-B. All the continents seem to maintain their positions around the lateral boundary of the convection domain, as observed from the snapshots at 237 Myr. By the time the large-scale upwelling plumes in the center of the convection domain become weak by ca. 237 Myr, pieces of continent tend to go back to the central part of the model domain; for instance, Continent-B and Continent-D are about to collide with each other, as 16 M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 Fig. 9. Time sequence of mantle convection with deformable, mobile continents. The computational domain of mantle convection is confined to three-dimensional regional spherical-shell geometry in spherical polar coordinates and has a thickness of 2867 km and a lateral extent of 90° × 90° in the latitudinal and longitudinal directions. The blue and purple isosurfaces of the temperature anomaly δT (i.e., the deviation from horizontally averaged temperature at each depth) indicate − 250 K and + 250 K, and the orange isosurfaces indicate the position of continents. The white spherical surface indicates the bottom of the mantle (i.e., core–mantle boundary). The supercontinent is composed by the four continental fragments (named ‘A’ to ‘D’), surrounding by the weak (low-viscosity) continental margins (WCM) (light orange), and it instantaneously imposed on the welldeveloped mantle convection with temperature-dependent rheology. The viscosity contrast between the coldest upper surface boundary (i.e., uppermost part of high viscous lithosphere) and hottest bottom surface boundary (i.e., core–mantle boundary) is 104. The elapsed times are scaled by an Earth-like timescale. The details of numerical methodology and model parameters are found in the work of Yoshida (2010a). seen from a snapshot at 322 Myr. And eventually, Continent-B and Continent-D collide with each other by 375 Myr. A series of these behaviors may mimic a Wilson Cycle, self-consistently reproduced by the numerical model. Paleographic reconstruction models show that the continental drift from the Rodinia supercontinent formed at ~ 900 Ma to the Pangea supercontinent formed at ~ 350 Ma is likely to occur on the hemispheric scale of the Earth's surface, not the planetary-scale, through the introversion process (Murphy and Nance, 2003, 2005) or A-type ocean closure (Silver and Behn, 2008) (Section 2). Thus, this mimicked Wilson Cycle observed in the regional spherical-shell geometry may be relevant to the real Wilson Cycle. Continental splitting and thinning, i.e., rift valley formation, that occur in the first stage of the Wilson Cycle are also observed in this model because the model allows a continent to laterally and radially deform with the driving force of mantle convection. A continental collision, the final stage of the Wilson Cycle, is observed after about ~ 375 Myr, which is broadly similar to the timescales for the actual Wilson Cycles as inferred from geological criteria (Section 1). This model presented here would represent an important step towards formulating a more realistic model that could be used to address many outstanding problems about the thermal and mechanical feedbacks between the mantle and continents, the cycle of continental breakup and collision, the duration of a supercontinent, the mechanism of continental drift, and the temporal evolution of the Earth's mantle structure. The continental splitting and thinning (i.e., rift valley formation) that occur in the first stage of the Wilson Cycle are observed in the present model (Figs. 9 and 10), because this model allows a continent to laterally and radially deform with the driving force of mantle convection. However this model used a relatively simple rheology especially for the continental materials, and the result does not answer the question of how supercontinents break apart, because the weak zones are imposed a priori in a model supercontinent (i.e., no yield mechanism in the continent). Nevertheless, among the 3-D models proposed so far, this model has the largest potential to explain the Wilson Cycles documented from geological and geochronological data. 12. Continents and plate tectonics The upwelling and downwelling plumes, plate tectonics, and continental drift are a part of the convecting system in the mantle. It is considered that the flow scale that characterizes the upwelling plumes is independent from the one that is characterized by plate subduction. As stated in Section 1, the Earth's mantle structure has different scale of downwellings (i.e., planetary-scale subduction of plate) and upwellings (i.e., multi-scale upwelling plumes). In particular, if the heat from the core is significantly small compared with the heat production by radiogenic elements of mantle rocks (Davies, 1988; Sleep, 1990), mantle plumes define only the secondary flow compared to the main flow determined by the plate tectonics (Davies, 1988, 1999), and therefore do not impart significant influence on the global-scale plate motion. The formation and rapid M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 17 Fig. 10. Early stage of the time evolution of mantle convection with supercontinents. The composition field (left panels) and the corresponding temperature anomaly δT (right panels) at a shallower mantle (depth of 358 km). The composition field indicates 1 for the continent and 0 for oceanic plates. The surface velocity fields are superimposed on the composition field. Interval of line counters for δT is 100 K. The arrow ‘A’ indicates the hot upwelling plumes from the core–mantle boundary; ‘B’ indicates the broad hot regions caused by the thermal insulating effect. expansion of the Pacific and Indian oceanic plates in the Cretaceous took place in the broadly low-velocity regions because they have not been cooled or displaced by cold oceanic lithosphere for more than 200 Myr, and mantle plumes are not required to explain such a plate formation and expansion (Anderson, 1994). In contrast to the nearly independent relation between plates and mantle plumes, the behavior of mantle plumes is closely linked with the continental aggregation and breakup as stated in Section 8. However, the mechanical interaction between continent/supercontinent and moving/subduction plates has been poorly understood in geodynamics. Because continental fragments drift in accordance with plate motions on the Earth, there should be a mechanical coupling between continental drift and plate motions throughout the Earth's history. Although the preliminary study by Yoshida (2010a) discussed in Section 11 has considered the aspects of both mobile continents and oceanic plates in numerical models, it is still one of the major challenges in numerical studies of mantle convection to selfconsistently reproduce moving plates in 3-D space (e.g., Bercovici et al., 2000). This is because the mechanisms for the generation of narrow plate boundary and the associated plate motion are not adequately clear in geodynamics (e.g., Bercovici, 1996, 1998; Bercovici et al., 2000). There is a high possibility that chemical and rheological heterogeneities due to moving plates, as well as drifting continent/supercontinent (Sections 4–5), give rise to the long-wavelength structure in the 3-D spherical mantle. Fig. 11 shows the results of mantle convection with the oceanic plates but without continents/supercontinent. The highly viscous oceanic lithosphere is damaged by imposing a yield stress 18 M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 (i.e., maximum limit of strength of oceanic lithosphere), which was often used for the previous 2-D and 3-D mantle convection models (e.g., Moresi and Solomatov, 1998; Trompert and Hansen, 1998a). When the yield stress is small, the oceanic lithosphere is highly damaged. When it is large, the oceanic lithosphere is not substantially damaged (left panels of Fig. 11). The plate-like behavior is obtained in a moderate yield stress of around τy = 100 MPa (Tackley, 2000b; Richards et al., 2001; van Heck and Tackley, 2008; Walzer and Hendel, 2008; Yoshida, 2008; Foley and M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 Becker, 2009), which is almost determined by the convection stress. When the yield stress is 100–125 MPa, narrow divergence and convergence zones coexist in the lithosphere, and both of them are comparable to the ‘spreading center’ and ‘trench’ of the Earth, respectively. The Y-shaped subduction zone observed in the western Pacific regions of the present-day Earth is also reproduced (see red arrow in the map with τy = 100 MPa). Broadly deformed regions with low-viscosity (for instance, ‘D’ in the map with τy = 100 MPa) may correspond to the ‘diffuse plate boundary’ of the Earth's lithosphere (Gordon and Stein, 1992; Gordon, 1998, 2000; Zatman et al., 2005). The Earth's continental and oceanic plates are not uniformly rigid, in contrast with the theory of plate tectonics that can be attributed to the near rigidity of plates and narrow plate boundaries (Wilson, 1966; McKenzie and Parker, 1967; Le Pichon, 1968; Morgan, 1968) and the previous 3-D models for plate dynamics (Hager and O'Connell, 1979, 1981; Zhong and Davies, 1999; Yoshida et al., 2001). The resultant convection pattern shows a long-wavelength thermal structure organized by subducting ‘oceanic plates,’ when the value of yield stress is moderate, 100–125 MPa (right panels of Fig. 11). The results indicate that the subduction zones surrounding the margins of the past supercontinents since the Palaeozoic era (i.e., Gondwana and Pangaea), as seen in some paleographic reconstructions (e.g., Collins, 2003), may be an inherent characteristic of mantle convection only with thermally induced driving force, that is, without any chemical heterogeneity like the compositionally distinct continent/supercontinent. The interplay between plate motions and continental assembly–dispersal in the evolving mantle is not clear in the numerical study at present. Future simulation studies should focus on mantle convection models which take into account both the continents/supercontinents and plate-tectonic motions. A possible mechanical coupling between supercontinent cycle and plate tectonics throughout the Earth's history is proposed with a conceptual model by Silver and Behn (2008) (Section 2). They hypothesized that drastic reductions or temporary cessations of plate subduction (‘intermittent plate tectonics’) have occurred in situations where a supercontinent forms primarily by external ocean closing and the dispersing continental fragments assemble in the antipode of the Earth. Namely, plate tectonics temporarily stops in the process of continental assembly because all the subduction of the plates diminishes at the continental margins. This external ocean closure primarily occurred to form the supercontinent Rodinia at 900 Ma and Pannotia at 650 Ma (Silver and Behn, 2008). However, in contrast to their conceptual model, there may be a possibility that plate tectonics temporarily stops by a change in the stress magnitude in the oceanic lithosphere (i.e., lithospheric strength) in the process of the mechanical interaction between drifting continent/supercontinent and moving oceanic plates, because plate-like motion may occur in a narrow range of the yield stress as shown in Fig. 11. 13. Next supercontinent The formation of the future supercontinent is of broad interest not only for earth scientists but also for the larger audience who are interested in the future shape of the Earth. The name ‘Amasia’ has 19 been proposed for the next supercontinent (Hoffman, 1992). The North American and Asiatic plates already fused along a zone of uncertain nature through eastern Asia, and it is possible that the next supercontinent will develop as more of the present continents collide around the Asia–North America nucleus (Rogers and Santosh, 2004). Part of this fusion might result from further accretion of Gondwana blocks to southern Asia, perhaps by closure of the Indian Ocean. If South America and Antarctica moved in a pattern that would accrete them to Asia in about 500 Myr, then the next supercontinent would reach its maximum packing at the same time in the cycle as previous supercontinents. It has also been proposed that if modern subduction in the Caribbean and Scotia arcs spreads along the Atlantic seaboard, then convergence and destruction of the Atlantic Ocean would result in a supercontinent termed as ‘Pangea Ultima’ (Scotese, 2000) that would form by introversion (Murphy et al., 2009). Maruyama et al. (2007) proposed a conceptual model which argues that the Western Pacific Triangular Zone (WPTZ) is a potential candidate for the frontier of a future supercontinent (Fig. 12). According to these authors, the presence of a double-sided subduction zone in this region where the Pacific plate subducts under the WPTZ, and the Indo-Australian plate subducts and collides locally against the WPTZ would promote the assembly of continental fragments. This concept was further elaborated recently by Santosh et al. (2009) where they identified two major categories of subduction zones on the globe based on Y-shaped trip junctions, termed as the CircumPacific Tethyan subduction zones (Section 3). The Circum-Pacific subduction zone covers a length of over 20,000 to 30,000 km, and the remaining 10,000 km zone lies within the northern margin of the Indo-Australian plate starting from the junction of the Pacific subduction zone to the East in New Zealand to Fiji region. The zone continues to the west through Indonesia, Himalaya and Middle East to the Mediterranean through Turkey and is finally connected to the Mid Atlantic Ridge. The Tethyan subduction zone broadly covers the zone along the northern margin of India and the Tethyan Ocean. The past 200 million year history along this zone shows a series of continent collision leading to the growth of Eurasia (cf. Metcalfe, 2006, in press), and must have also witnessed the subduction of substantial volume of oceanic lithosphere. Along the eastern margin of the Y-shaped subduction zone running from Kamchatka down to New Zealand, the passive subduction of Pacific oceanic lithosphere is speculated to continue, leading to the consumption of the oceanic lithosphere deep in the mantle off North and South America. The trench will then migrate to the west leading to a reduction in the size of the Pacific Ocean with time. According this model, in another 250 Myr, the Pacific Ocean might vanish from the globe leading to the assembly of Asia with the amalgam of North and South America. Although the fate of Antarctica is not known, presumably this continent will also join the amalgam in course of time through the propagating westward trench along the northern margin of Antarctica. Thus, the assembly of the future supercontinent Amasia will be finally completed through Pacific ocean closure and extroversion. However, this model faces the problem that the Pacific superplume is situated in between. Also, alternate models of the Atlantic Ocean closing first require evaluation. Fig. 11. Snapshots of mantle convection with varying the magnitude of yield stress (τy; maximum rupture strength) of the lithosphere. Note that only the mechanism for plate-like motion is considered, and continents/supercontinents are not imposed in this model. The degree-one component of toroidal velocity field that represents the rigid-body rotation is subtracted from solutions at each simulation time-step. Left panel shows viscosity and velocity fields on the top surface of model domain. Blue and yellow show the higher and lower viscosities, respectively. Right panel shows the corresponding isosurfaces of temperature anomaly. The blue and yellow isosurfaces of the temperature anomaly δT (i.e., the deviation from horizontally averaged temperature at each depth) indicate − 250 K and + 250 K. The white sphere indicates the bottom of mantle (i.e., core–mantle boundary). The highly viscous oceanic lithosphere is realized by the temperature-dependent rheology of mantle materials, and the plate-like motions in the oceanic lithosphere are realized by imposing the moderate yield stresses (τy = 100 and 125 MPa): ‘S’ and ‘T’ indicate divergence and convergence zones in the lithosphere, i.e., correspond to the spreading center and trench in the Earth. Broadly deformed regions with low-viscosity (for instance, ‘D’ in the map with τy = 100 MPa) may correspond to the ‘diffuse plate boundary’ in the Earth (Gordon, 1998). A red arrow in the map with τy = 100 MPa shows the Y-shaped subduction zone. When the yield stress is lower (i.e., τy = 50 MPa), the lithosphere is highly ruptured, while it is higher (i.e., 150 MPa), the lithosphere is not ruptured so much and the plate boundary is incompletely evolved. The details of numerical methodology and model settings are similar to a previous study by Yoshida (2008), but the viscosity contrast between the coldest upper surface boundary (i.e., uppermost part of high viscous lithosphere) and hottest bottom surface boundary (i.e., core–mantle boundary) is taken to be a higher value, 106. The Rayleigh number defined by the reference viscosity (1021 Pa s) is fixed at 5.72 × 106. 20 M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 Fig. 12. Location of the western Pacific triangular zone (WPTZ), considered as the frontier of the future supercontinent (after Maruyama et al., 2007). East Asia is the location of a double-sided subduction zone, where the old Pacific plate subducts from the east, and the Indo-Australia plate from the south. Due to subduction, and hence refrigeration, the upper and lower mantle here are the coldest mantle regions in the world. A schematic cross-section of WPTZ is shown below to illustrate the stagnant slabs, hydrous mantle transition zone, and formation of hydrous plumes at 410 km depth. We will now examine how numerical simulations approach the shape of the future globe in terms of supercontinent configurations. As stated in Section 11, it is difficult for numerical simulations to reproduce the continental drift on the 3-D spherical model because of some limitations in the techniques currently employed. Nevertheless, Trubitsyn et al. (2008) constructed a 3-D numerical model to realize the continental drift from present to 100 Myr later and to predict a configuration of future supercontinent. They consider continents as thin rigid (non-deformable) spherical caps, the motion of which is described by Euler's solid body equation (see Trubitsyn et al., 2008 for the details of numerical methodology and model parameters used). The model continents are arranged in a real distribution of continents of the present Earth. The initial temperature condition of mantle is given by the present mantle flow pattern estimated from a seismic tomography model (that is, it has the Circum-Pacific subduction zone and two superplumes in the Africa–Atlantic and the South Pacific regions). Their simulation reveals that after 100 Myr, all continents will be assembled in the southern hemisphere. Africa, Eurasia, Australia, Antarctica and South America form a future supercontinent around Antarctica, because the Y-shaped subduction zone in the Western Pacific pulls the continents. According to the model, North America will not be part of the future, probably because the tensional force due to both the South Pacific and African superplumes prevents North America from moving to the southern hemisphere and joining the future supercontinent amalgam. 14. Summary and conclusion A synthesis of some of the conceptual models on supercontinent amalgamation and disruption, together with recent information from M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 numerical studies, leads to a better understanding of the Wilson Cycle and supercontinent cycle. The superdownwelling along multiple subduction zones predicted in plate tectonic models might provide an effective mechanism to pull together dispersed continental fragments into a closely packed assembly. Plumes and superplumes which breakup supercontinents are considered to be fueled by the recycled subducted material that initially accumulates at the mantle transition zone and ultimately sinks down into the CMB. The process of reassembly of dispersed continental fragments occurs through complex processes involving ‘introversion’, ‘extroversion’ or a combination of both, with the closure of the intervening ocean occurring through Pacific-type or Atlantic-type processes. The timescales of the assembly and dispersion of supercontinents have varied through the Earth history, and appear to be closely linked with the processes and duration of superplume genesis. It is now widely recognized that plate tectonic processes lead to both creation and destruction of continental crust. The production–destruction balance changes over a supercontinent cycle, with a higher crustal growth through magmatic influx during supercontinent break-up as compared to the tectonic erosion and sediment-trapped subduction in convergent margins associated with supercontinent assembly which erodes the continental crust. The link between deep mantle dynamics and lithospheric plate motions emphasizes the periodic assembly and dispersal of supercontinents. Model relations between the Wilson Cycle and mantle convection have a strong bearing on the ‘superplume’ hypothesis. Despite the controversy surrounding the superplume concept, the geophysical and geochemical tools at our disposal have aided in the formulation of persuasive models on the role of superplumes. The beginning of a consensus is visible in the various studies concerning the origins of ‘hot spots’ and likely ‘superplume’ candidates. The processes responsible for the formation and disruption of past supercontinents and the hypothetical future supercontinent amalgam have important bearing on future biotic evolution and climate change. Often, unusual phenomena such as backarc basin propagation occur associated with some of the subduction systems. There are also conflicting interpretations of ‘proto-continents’ as drifted fragments or terranes extruded from major continental collision. From the geological point of view, certain specific rock types have been considered as the signature of plumes, although similar rock types also occur in alternate tectonic settings. Another case concerns contiguous marginal basins which are considered as the products of collision-induced lateral mantle flow. Although these may reflect ‘plate-driven’ lateral displacement of ductile upper mantle, they do not negate the ‘super’ or ‘regular’ plume models. Whereas the Tethyan collisions (Africa, India, and Australia) clearly share common lithospheric indications, this does not necessarily apply to their respective asthenospheric responses. They are probably similar in kind but need to be compared and contrasted from geological evidence within the erstwhile sutures between constituent Gondwana terranes. The numerical studies reviewed herein suggest that there is a significant feedback not only between mantle convection and continental drift, but also between mantle convection and plate motions. The process of assembly of supercontinents induces a temperature increase due to the thermal insulating effect. Such thermal insulation leads to a planetary-scale reorganization of mantle flow and results in longest-wavelength thermal heterogeneity in the mantle, i.e., degree-one convection in 3-D spherical geometry. The formation of degree-one convection seems to be integral to the emergence of periodic supercontinent cycles. The rifting and breakup of supercontinental assemblies may be caused by either tensional stress due to the thermal insulating effect, or large-scale partial melting due to the flow reorganization and the resultant temperature increase beneath the supercontinent. Supercontinent breakup has also been correlated with the temperature increase due to upwelling plumes originating from the CMB as a return flow of plate subduction 21 occurring at supercontinental margins. The active mantle plumes from the CMB seem to disrupt the regularity of supercontinent cycles. If mantle convection with dispersed continental blocks is wellorganized degree-one structure owing to its temperature-dependent rheology, the continental blocks may move to downwellings and form a supercontinent. Although most of the numerical studies have assumed the continent/supercontinent to be rigid or nondeformable body mainly because of numerical limitations as well as a simplification of models, a more recent numerical study allows the modeling of mobile, deformable continents, including oceanic plates, and successfully reproduces continental drift similar to the processes and timescales envisaged in Wilson Cycle. On the basis of the numerical studies synthesized here, we can draw two possible end-member scenarios of the mantle convection cycle. One is that mantle convection with dispersing continental blocks has a short-wavelength structure (or close to degree-two structure as the present Earth), and when a supercontinent forms, mantle convection evolves into degree-one structure. Another is the degree-one to degree-two model (Section 5). As for the former model (high-degrees to degree-one), it would take longer time to form a supercontinent, because continental blocks would be trapped by different downwellings thus inhibiting collision. The timescale of mantle convection cycle may be, however, significantly affected by the plate motion and subduction that stir mantle convection. In order to better constrain some of the key topics covered in this paper, the future studies based on numerical modeling need to address the following aspects: (1) Thermal and mechanical interaction between continent/supercontinent and plate motion. (2) The mechanism of formation of large-scale upwellings beneath the supercontinent. (3) The mechanism of origin and growth of the continental crust. (4) The mechanism of supercontinent breakup. (5) The episodic growth of continental crust and supercontinent. (6) The growth and destruction of continental crust both related to subduction zone processes. (7) The thermal and mechanical interaction between supercontinent cycle and mantle convection during the Earth's history, as well as the thermal and chemical evolution of convecting mantle. Acknowledgements We thank Editor Prof. M.F.J. Flower and Reviewer Prof. J. Brendan Murphy for their valuable suggestions and helpful comments. M.Y. was supported by a Grant-in-Aid (No. 20740260) from the Ministry of Education, Culture, Sports, Science and Technology, Japan. References Anderson, O.L., 1967. Equation for thermal expansivity in planetary interiors. J. Geophys. Res. 72, 3661–3668. Anderson, D.L., 1982. Hotspots, polar wander, Mesozoic convection and the geoid. Nature 297, 391–393. Anderson, D.L., 1987. A seismic equation of state II. Shear properties and thermodynamics of the lower mantle. Phys. Earth Planet. Inter. 45 (4), 307–327. Anderson, D.L., 1994. Superplumes or supercontinents? Geology 22, 39–42. Bercovici, D., 1996. Plate generation in a simple model of lithosphere-mantle flow with dynamic self-lubrication. Earth Planet. Sci. Lett. 144, 41–51. Bercovici, D., 1998. Generation of plate tectonics from lithosphere-mantle flow and void-volatile self-lubrication. Earth Planet. Sci. Lett. 154, 139–151. Bercovici, D., Ricard, Y., Richards, M.A., 2000. The relation between mantle dynamics and plate tectonics: a primer. The History and Dynamics of Global Plate Motions: In: Richards, M.A., Gordon, R.G., van der Hilst, R.D. (Eds.), Geophys. Monogr. Ser., New York, pp. 5–46. Bijwaard, H., Spakman, W., 1999. Tomographic evidence for a narrow whole mantle plume below Iceland. Earth Planet. Sci. Lett. 166, 121–126. Bills, B.G., May, G.M., 1987. Lake Bonneville: constraints on lithospheric thickness and upper mantle viscosity from isostatic warping of Bonneville, Provo, and Gilbert Stage shorelines. J. Geophys. Res. 92, 11493–11508. Birch, F., 1968. Thermal expansion at high pressure. J. Geophys. Res. 73, 817–819. Bunge, H.-P., Richards, M., 1996. The origin of large scale structure in mantle convection: effects of plate motions and viscosity stratification. Geophys. Res. Lett. 23 (21), 2987–2990. 22 M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 Bunge, H.-P., Richards, M.A., Baumgardner, J.R., 1996. The effect of depth dependent viscosity on the planform of mantle convection. Nature 379, 436–438. Bunge, H.-P., Richards, M.A., Baumgardner, J.R., 1997. A sensitivity study of 3-D spherical mantle convection at 108 Rayleigh number: effects of depth dependent viscosity, heating mode and an endothermic phase change. J. Geophys. Res. 102, 11991–12007. Busse, F.H., 1989. Fundamentals of thermal convection. In: Peletier, W.R. (Ed.), Mantle Convection: Plate Tectonics and Global Dynamics, The Fluid Mechanics of Astrophysics and Geophysics, vol 4. Gordon and Breach, New York, pp. 23–95. Busse, F.H., Richards, M.A., Lenardic, A., 2006. A simple model of high Prandtl and high Rayleigh number convection bounded by thin low viscosity layers. Geophys. J. Int. 164, 160–167. Carlson, R.W., Pearson, D.G., James, D.E., 2005. Physical, chemical, and chronological characteristics of continental mantle. Rev. Geophys. 43, 1–24 Physical, chemical, and chronological characteristics of continental mantle. Christensen, U., 1984. Convection with pressure and temperature dependent nonNewtonian rheology. Geophys. J. R. Astron. Soc. 77, 343–384. Christensen, U., Hofmann, A.W., 1994. Segregation of subducted oceanic crust in the convecting mantle. J. Geophys. Res. 99, 19867–19884. Christensen, U.R., Yuen, D.A., 1985. Layered convection induced by phase transitions. J. Geophys. Res. 90, 10291–10300. Clift, P., Vannucchi, P., 2004. Controls on tectonic accretion versus erosion in subduction zones: implications for the origin and recycling of the continental crust. Rev. Geophys. 42, RG2001. doi:10.1029/2003RG000127. Clift, P., Schouten, H., Vannucchi, P., 2009. Arc–continent collisions, sediment recycling and the maintenance of the continental crust. Geol. Soc. Lond. Spec. Publ. 318, 75–103. Collins, W.J., 2003. Slab pull, mantle convection, and Pangaean assembly and dispersal. Earth Planet. Sci. Lett. 205, 225–237. Collins, A.S., Pisarevsky, S.A., 2005. Amalgamating eastern Gondwana: the evolution of the Circum-Indian orogens. Earth Sci. Rev. 71, 229–270. Coltice, N., Phillips, B.R., Bertrand, H., Ricard, Y., Rey, P., 2007. Global warming of the mantle at the origin of flood basalts over supercontinents. Geology 35, 391–394. doi:10.1130/G23240A.1. Coltice, N., Bertrand, H., Rey, P., Jourdan, F., Phillips, B.R., Ricard, Y., 2009. Global warming of the mantle beneath continents back to the Archaean. Gondwana Res. 15 (3–4), 254–266. doi:10.1016/j.gr.2008.10.001. Condie, K.C., 1998. Episodic continental growth and supercontinents: a mantle avalanche connection? Earth Planet. Sci. Lett. 163, 97–108. Condie, K.C., 2001. Mantle Plumes and their Record in Earth History. Cambridge Univ. Press, U.K. 306 pp. Condie, K.C., 2004. Supercontinents and superplume events: distinguishing signals in the geologic record. Phys. Earth Planet. Inter. 146, 319–332. Dalziel, I.W.D., 1991. Pacific margins of Laurentia and East Antarctica–Australia as a conjugate rift pair: evidence and implications for an Eocambrian supercontinent. Geology 19, 598–601. Dalziel, I.W.D., Lawver, L.A., Murphy, J.B., 2000. Plumes, orogenesis, and supercontinental fragmentation. Earth Planet. Sci. Lett. 178 (1–2), 1–11. Davaille, A., 1999. Simultaneous generation of hotspots and superswells by convection in a heterogeneous planetary mantle. Nature 402, 576–760. Davaille, A., Girard, F., Le Bars, M., 2002. How to anchor hotspots in a convecting mantle? Earth Planet. Sci. Lett. 203, 621–634. Davies, G.F., 1988. Ocean bathymetry and mantle convection: 1. Large-scale flow and hot spots. J. Geophys. Res. 93, 10467–10480. Davies, G.F., 1999. Dynamic Earth: Plates, Plumes and Mantle Convection. Cambridge University Press, Cambridge. 460 pp. Dewey, J., Spall, H., 1975. Pre-Mesozoic plate tectonics: how far back in Earth history can the Wilson Cycle be extended? Geology 3, 422–424. Doin, M.-P., Fleitout, L., Christensen, U., 1997. Mantle convection and stability of depleted and undepleted continental lithosphere. J. Geophys. Res. 102 (B2), 2771–2787. Eriksson, P.G., Banerjee, S., Nelson, D.R., Rigby, M.J., Catuneanu, O., Sarkar, S., Roberts, R.J., Ruban, D., Mtimkulu, M.N., Raju, P.V.S., 2009. A Kaapval craton debate: nucleus of an early small supercontinent or affected by an enhanced accretion event? Gondwana Res. 15, 354–372. Ernst, R.E., Buchan, K.L., 2003. Recognizing mantle plumes in the geological record. Annu. Rev. Earth Planet. Sci. 31, 469–523. Evans, D.A.D., 2003. True polar wander and supercontinents. Tectonophysics 362, 303–320. Foley, B.J., Becker, T.W., 2009. Generation of plate-like behavior and mantle heterogeneity from a spherical, viscoplastic convection model. Geochem. Geophys. Geosyst. 10, Q08001. doi:10.1029/2009GC002378. Forte, A.M., Mitrovica, J.X., 2001. Deep-mantle high-viscosity flow and thermochemical structure inferred from seismic and geodynamic data. Nature 410, 1049–1056. Fukao, Y., 1992. Seismic tomogram of the Earth's mantle: geodynamic implications. Science 258, 625–630. Fukao, Y., Obayashi, M., Inoue, H., Nenbai, M., 1992. Subducting slabs stagnate in the mantle transition zone. J. Geophys. Res. 97 (B4), 4809–4822. Fukao, Y., Maruyama, S., Obayashi, M., Inoue, H., 1994. Geologic implication of the whole mantle P-wave tomography. J. Geol. Soc. Jpn 100 (1), 4–23. Fukao, Y., Widiyantoro, S., Obayashi, M., 2001. Stagnant slabs in the upper and lower mantle transition region. Rev. Geophys. 39 (3), 291–323. Fukao, Y., Obayashi, M., Nakakuki, T., Deep Slab Project Group, 2009. Stagnant slab: a review. Annu. Rev. Earth Planet. Sci. 37, 19–46. Gordon, R.G., 1998. The plate tectonic approximation: plate nonrigidity, diffuse plate boundaries, and global plate reconstructions. Annu. Rev. Earth Planet. Sci. 26, 615–642. Gordon, R.G., 2000. Diffuse oceanic plate boundaries: strain rates, vertically averaged rheology, and comparisons with narrow plate boundaries and stable plate interiors. The History and Dynamics of Global Plate Motions. : In: Richards, M.A., Gordon, R., van der Hilst, R.D. (Eds.), Geophys. Monograph Series. Am. Geophys. Union, Washington D.C., pp. 143–159. Gordon, R.G., Stein, S., 1992. Global tectonics and space geodesy. Science 256, 333–342. Grand, S.P., 2002. Mantle shear-wave tomography and the fate of subducted slabs. Philos. Trans. R. Soc. Lond. A 360, 2475–2491. Gung, Y., Panning, M., Romanowicz, B., 2003. Global anisotropy and the thickness of continents. Nature 422, 707–711. Gurnis, M., 1988. Large-scale mantle convection and the aggregation and dispersal of supercontinents. Nature 332, 695–699. Gurnis, M., 1990. Plate–mantle coupling and continental flooding. Geophys. Res. Lett. 17 (5), 623–626. Gurnis, M., Zhong, S., 1991. Generation of long wavelength heterogeneity in the mantle by the dynamic interaction between plates and convection. Geophys. Res. Lett. 18 (4), 581–584. Hager, B., O'Connell, R., 1979. Kinematic models of large-scale flow in the Earth's mantle. J. Geophys. Res. 84 (B3), 1031–1048. doi:10.1029/JB084iB03p01031. Hager, B., O'Connell, R., 1981. A simple global model of plate dynamics and mantle convection. J. Geophys. Res. 86 (B6), 4843–4867. doi:10.1029/JB086iB06p04843. Hansen, U., Yuen, D.A., 1994. Effects of depth-dependent thermal expansivity on the interaction of thermal–chemical plumes with a compositional boundary. Phys. Earth Planet. Inter. 86, 205–221. Hansen, U., Yuen, D.A., Kroening, S.E., Larsen, T.B., 1993. Dynamical consequences of depth-dependent thermal expansivity and viscosity on mantle circulations and thermal structure. Phys. Earth Planet. Inter. 77, 205–213. Hirth, G., Kohlstedt, D.L., 1996. Water in the oceanic upper mantle: implications for rheology, melt extraction and the evolution of the lithosphere. Earth Planet. Sci. Lett. 144, 93–108. Hoffman, P.F., 1991. Did the breakout of Laurentia turn Gondwanaland inside-out? Science 252, 1409–1412. Hoffman, P.F., 1992. Rodinia, Gondwanaland, Pangea and Amasia: alternating kinematic scenarios of supercontinental fusion. Eos Trans. AGU Fall Meet. Suppl. 73 (14), 282. Höink, T., Lenardic, A., 2008. Three-dimensional mantle convection simulations with a low-viscosity asthenosphere and the relationship between heat flow and the horizontal length scale of convection. Geophys. Res. Lett. 35, L10304. doi:10.1029/ 2008GL033854. Honda, S., Yuen, D.A., Balachandar, S., Reutelcr, D., 1993. Three dimensional instabilities of mantle convection with multiple phase transitions. Science 259, 1308–1311. Honda, S., Yoshida, M., Ootorii, S., Iwase, Y., 2000. The timescales of plume generation caused by continental aggregation. Earth Planet. Sci. Lett. 176 (1), 31–43. Hou, G.T., Santosh, M., Qian, X., Lister, G.S., Li, J.H., 2008. Tectonic constraints on 1.3 1.2 Ga final breakup of Columbia supercontinent from a giant radiating dyke swarm. Gondwana Res. 14, 561–566. Hurley, P.M., Rand, J.R., 1969. Pre-drift continental nuclei. Science 164, 1229–1242. Ismail-Zadeh, A., Tackley, P., 2010. Computational Methods for Geodynamics. Cambridge Univ. Press. Isozaki, Y., 2009. Illawarra Reversal: the fingerprint of a superplume that triggered Pangean breakup and the end-Guadalupian (Permian) mass extinction. Gondwana Res. 15, 421–432. Jarvis, G.T., McKenzie, D.P., 1980. Convection in a compressible fluid with infinite Prandtl number. J. Fluid Mech. 96 (3), 515–583. Jellinek, A.M., Manga, M., 2002. The influence of a chemical boundary layer on the fixity, spacing, and lifetime of mantle plumes. Nature 418, 760–763. Jellinek, A.M., Manga, M., 2004. Links between long-lived hot spots, mantle plumes, D", and plate tectonics. Rev. Geophys. 42. doi:10.1029/2003RG000144. Jordan, T.H., 1975. The continental tectosphere. Rev. Geophys. 13, 1–12. Jordan, T., 1988. Structure and formation of the continental tectosphere. J. Petrol. Spec. Lithosphere Issue 11–37. Karato, S., 2008. Deformation of Earth Materials: Introduction to the Rheology of the Solid Earth. Cambridge Univ. Press, U. K. Karato, S.-I., 2010a. Rheology of the Earth's mantle: a historical review. Gondwana Res. 18, 17–45. Karato, S., 2010b. Rheology of the deep upper mantle and its implications for the preservation of the continental roots: a review. Tectonophysics 481 (1–4), 82–98. Katsura, T., Yokoshi, S., Kawabe, K., Shatskiy, A., Geeth, M.A., Manthilakea, M., Zhai, S., Fukui, H., Chamathni, H.A., Hegoda, I., Yoshino, T., Yamazaki, D., Matsuzaki, T., Yoneda, A., Ito, E., Sugita, M., Tomioka, N., Hagiya, K., Nozawa, A., Funakoshi, K., 2009. P–V–T relations of MgSiO3 perovskite determined by in situ X-ray diffraction using a large-volume high-pressure apparatus. Geophys. Res. Lett. 36, L01305. doi:10.1029/2008GL035658. Kellogg, L.H., Hager, B.H., van der Hilst, R.D., 1999. Compositional stratification in the deep mantle. Science 283, 1881–1884. King, S.D., 1995. The viscosity structure of the mantle. Rev. Geophys. 33 (S1), 11–18. Komabayashi, T., Maruyama, S., Rino, S., 2009. A speculation on the structure of the D" layer: the growth of anti-crust at the core–mantle boundary through the subduction history of the Earth. Gondwana Res. 15, 342–353. Komiya, T., Hayashi, M., Maruyama, S., Yurimoto, H., 2002. Intermediate P/T type Archean metamorphism of the Isua supracrustal belt: implications for secular change of geothermal gradients at subduction zones and for Archean plate tectonics. Am. J. Sci. 302, 806–826. Kusky, T.M., Windley, B.F., Zhai, M.G., 2007. Tectonic evolution of the North China block: from orogen to craton to orogen. In: Zhai, M.G., Windley, B.F., Kusky, T.M., Meng, Q.R. (Eds.), Mesozoic Sub-continental Lithospheric Thinning Under Eastern Asia Geological Society of London Special Publication, pp. 1–34. M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 Lambeck, K., Johnston, P., 1998. The viscosity of the mantle: evidence from the analysis of glacial rebound phenomena. In: Jackson, I. (Ed.), The Earth's Mantle, Composition, Structure and Evolution. Cambridge Univ. Press, U.K., pp. 461–502. Larson, R.L., 1991. Latest pulse of Earth: evidence for a mid-Cretaceous superplume. Geology 19, 547–550. Le Pichon, X., 1968. Sea-floor spreading and continental drift. J. Geophys. Res. 73, 3661–3697. Lenardic, A., 1997. On the heat flow variation from Archean cratons to Proterozoic mobile belts. J. Geophys. Res. 102 (B1), 709–722. Lenardic, A., 1998. On the partitioning of mantle heat loss below oceans and continents over time and its relationship to the Archaean paradox. Geophys. J. Int. 134, 706–720. Lenardic, A., Kaula, W.M., 1994. Self-lubricated mantle convection: two dimensional models. Geophys. Res. Lett. 21, 1707–1710. Lenardic, A., Kaula, W.M., 1995. Mantle dynamics and the heat flow into the Earth's continents. Nature 378, 709–711. Lenardic, A., Kaula, W.M., 1996. Near-surface thermal/chemical boundary layer convection at infinite Prandtl number: two-dimensional numerical experiments. Geophys. J. Int. 126, 689–711. Lenardic, A., Moresi, L.-N., 1999. Some thoughts on the stability of cratonic lithosphere: effects of buoyancy and viscosity. J. Geophys. Res. 104 (B6), 12747–12758. Lenardic, A., Moresi, L.-N., Muhlhaus, H., 2003. Longevity and stability of cratonic lithosphere: insights from numerical simulations of coupled mantle convection and continental tectonics. J. Geophys. Res. 108 (B6), 2303. doi:10.1029/2002JB001859. Lenardic, A., Richards, M.A., Busse, F.H., 2006. Depth-dependent rheology and the horizontal length-scale of mantle convection. J. Geophys. Res. 111, B07404. doi:10.1029/2005JB003639. Li, C., van der Hilst, R.D., Engdahl, E.R., Burdick, S., 2008. A new global model for P wave speed variations in Earth's mantle. Geochem. Geophys. Geosyst. 9, Q05018. doi:10.1029/2007GC001806. Li, Z.X., Bogdanova, S.V., Collins, A.S., Davidson, A., Waele, B.D., 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 breakup history of Rodinia: a synthesis. Precambrian Res. 160 (1–2), 179–210. doi:10.1016/ j.precamres.2007.04.021. Lowman, J.P., Gable, C.W., 1999. Thermal evolution of the mantle following continental aggregation in 3D convection models. Geophys. Res. Lett. 26 (17), 2649–2652. Lowman, J.P., Jarvis, G.T., 1993. Mantle convection flow reversals due to continental collisions. Geophys. Res. Lett. 20, 2087–2090. Lowman, J.P., Jarvis, G.T., 1995. Mantle convection models of continental collision and breakup incorporating finite thickness plates. Phys. Earth Planet. Inter. 88, 53–68. Lowman, J.P., Jarvis, G.T., 1996. Continental collisions in wide aspect ratio and high Rayleigh number two-dimensional mantle convection models. J. Geophys. Res. 101, 25485–25497. Machetel, P., Weber, P., 1991. Intermittent layered convection in a model mantle with an endothermic phase change at 670 km. Nature 350, 55–57. Maruyama, S., 1994. Plume tectonics. J. Geol. Soc. Jpn 100, 24–49. Maruyama, S., Santosh, M., Zhao, D., 2007. Superplume, supercontinent, and postperovskite: mantle dynamics and anti-plate tectonics on the core–mantle boundary. Gondwana Res. 11, 7–37. Masters, G., Laske, G., Bolton, H., Dziewonski, A.M., 2000. The relative behavior of shear velocity, bulk sound speed, and compressional velocity in the mantle: implications for chemical and thermal structure, in Earth's Deep Interior. Mineral Physics and Tomography from the Atomic to the Global Scale: In: Karato, S.-I., Forte, A.M., Liebermann, R.C., Masters, G., Stixrude, L. (Eds.), Geophys. Monogr. Ser. AGU, Washington, D.C. , pp. 63–87. McBride, J.H., 1991. Constraints on the structure and tectonic development of the early Mesozoic South Georgia rift, southeastern United States; seismic reflection data processing and interpretation. Tectonics 10, 1065–1083. McHone, J.G., 2000. Non-plume magmatism and tectonics during the opening of the central Atlantic Ocean. Tectonophysics 316, 287–296. McKenzie, D.P., Parker, R.L., 1967. The North Pacific: an example of tectonics on a sphere. Nature 216, 1276–1280. McKenzie, D.P., Roberts, J.M., Weiss, N.O., 1974. Convection in the earth's mantle: towards a numerical simulation. J. Fluid Mech. 62, 465–538. McNamara, A.K., Zhong, S., 2005a. Degree-one mantle convection: dependence on internal heating and temperature-dependent rheology. Geophys. Res. Lett. 32, L01301. doi:10.1029/2004GL021082. McNamara, A.K., Zhong, S., 2005b. Thermochemical structures beneath Africa and the Pacific ocean. Nature 437, 1136–1139. McNutt, M.K., 1998. Superswells. Rev. Geophys. 36, 211–244. Meert, J.G., 2002. Paleomagnetic evidence for a Paleo-Mesoproterozoic supercontinent Columbia. Gondwana Res. 5, 207–215. Meert, J.G., Lieberman, B.S., 2008. The Neoproterozoic assembly of Gondwana and its relationship to the Ediacaran–Cambrian radiation. Gondwana Res. 14, 5–21. Mégnin, C., Romanowicz, B., 2000. The 3D shear velocity structure of the mantle from the inversion of body, surface and higher mode waveforms. Geophys. J. Int. 143, 709–728. Metcalfe, I., 2006. Palaeozoic and Mesozoic tectonic evolution and palaeogeography of East Asian crustal fragments: the Korean Peninsula in context. Gondwana Res. 9, 24–46. Metcalfe, I., 2011. Tectonic framework and Phanerozoic evolution of Sundaland. Gondwana Res. 9 (1), 3–21. Mitrovica, J.X., Forte, A.M., 1997. Radial profile of mantle viscosity: results from the joint inversion of convection and postglacial rebound observables. J. Geophys. Res. 102, 2751–2769. 23 Mitrovica, J.X., Forte, A.M., 2004. A new inference of mantle viscosity based upon a joint inversion of convection and glacial isostatic adjustment data. Earth Planet. Sci. Lett. 225, 177–189. Monin, A.S., 1991. Planetary evolution and global tectonics. Tectonophysics 199, 149–164. Montelli, R., Nolet, G., Dahlen, F.A., Masters, G., Engdahl, E.R., Hung, S.-H., 2004. Finitefrequency tomography reveals a variety of plumes in the mantle. Science 303, 338–343. Montelli, R., Nolet, G., Dahlen, F.A., Masters, G., 2006. A catalogue of deep mantle plumes: new results from finite-frequency tomography. Geochem. Geophys. Geosyst. 7, Q11007. doi:10.1029/2006GC001248. Moresi, L.N., Lenardic, A., 1997. Three dimensional numerical simulations of crustal deformation and subcontinental mantle convection. Earth Planet. Sci. Lett. 150, 233–243. Moresi, L.-N., Lenardic, A., 1999. Three-dimensional mantle convection with continental crust: first-generation numerical simulations. Earth Interact 3, 002. Moresi, L., Solomatov, V.S., 1998. Mantle convection with a brittle lithosphere: thoughts on the global tectonics styles of the Earth and Venus. Geophys. J. Int. 133, 669–682. Morgan, W.J., 1968. Rises, trenches, great faults and crustal blocks. J. Geophys. Res. 73, 1959–1982. Morgan, W.J., 1983. Hotspot tracks and the early rifting of the Atlantic. Tectonophysics 94, 123–139. doi:10.1016/0040-1951(83)90013-6. Murphy, J.B., Nance, R.D., 2003. Do supercontinents introvert or extrovert? Sm–Nd isotope evidence. Geology 31, 873–876. Murphy, J.B., Nance, R.D., 2005. Do supercontinents turn inside-in or inside-out? Int. Geol. Rev. 47, 591–619. Murphy, J.B., Nance, R.D., Cawood, P.A., 2009. Contrasting modes of supercontinent formation and the conundrum of Pangea. Gondwana Res. 15, 408–420. Nair, R., Chacko, T., 2008. Role of oceanic plateaus in the initiation of subduction and origin of continental crust. Geology 36, 583–586. Nakakuki, T., Yuen, D.A., Honda, S., 1997. The interaction of plumes with the transition zone under continents and oceans. Earth Planet. Sci. Lett. 146 (3–4), 379–391. Nance, R.D., Worsley, T.R., Moody, J.B., 1986. Post-Archean biogeochemical cycles and longterm episodicity in tectonic processes. Geology 14, 514–518. doi:10.1130/0091-7613. Nance, R.D., Worsley, T.R., Moodey, J.B., 1988. The supercontinent cycle. Sci. Am. 1988, 72–79 (July). Okuno, J., Nakada, M., 2001. Effects of water load on geophysical signals due to glacial rebound and implications for mantle viscosity. Earth Planet. Space 53 (12), 1121–1135. Peltier, W.R., 1998. Postglacial variations in the level of the sea: implications for climate dynamics and solid-Earth geophysics. Rev. Geophys. 36 (4), 603–689. Peslier, A.H., Woodland, A.B., Bell, D.R., Lazarov, M., 2010. Olivine water contents in the continental lithosphere and the longevity of cratons. Nature 467, 78–81. doi:10.1038/nature09317. Phillips, B.R., Bunge, H.-P., 2005. Heterogeneity and time dependence in 3D spherical mantle convection models with continental drift. Earth Planet. Sci. Lett. 233, 121–135. Phillips, B.R., Bunge, H.-P., 2007. Supercontinent cycles disrupted by strong mantle plumes. Geology 35 (9), 847–850. doi:10.1130/G23686A.1. Phillips, B.R., Coltice, N., 2010. Temperature beneath continents as a function of continental cover and convective wavelength. J. Geophys. Res. 115, B04408. doi:10.1029/2009JB006600. Pollack, H.N., 1986. Cratonization and thermal evolution of the mantle. Earth Planet. Sci. Lett. 80, 175–182. Rapp, R.P., Watson, B., 1995. Dehydration melting of metabasalt at 8–32 kbar: implications for continental growth and crust–mantle recycling. J. Petrol. 36, 891–931. Ratcliff, J.T., Tackley, P.J., Schubert, G., Zebib, A., 1997. Transitions in thermal convection with strongly variable viscosity. Phys. Earth Planet. Inter. 102, 201–212. Ricard, Y., 2007. Physics of mantle convection. In: Bercovici, D., Schubert, G. (Eds.), Treatise on Geophysics, 7. Elsevier, pp. 31–87. Ricard, Y., Wuming, B., 1991. Inferring the mantle viscosity and its three dimensional structure from geoid, topography and plate velocities. Geophys. J. Int. 105, 561–571. Richards, M.A., Davies, G.F., 1992. Mantle convection. J. Geol. 100, 151–206. Richards, M.A., Engebretson, D.C., 1992. Large-scale mantle convection and the history of subduction. Nature 355, 437–440. Richards, M.A., Duncan, R.A., Courtillot, V., 1989. Flood basalts and hotspot tracks, plume heads and tails. Science 246, 103–107. Richards, M.A., Jones, D.L., Duncan, R.A., Depaolo, D.J., 1991. A mantle plume initiation model for the Wrangellia flood basalt and other oceanic plateaus. Nature 254, 263–267. Richards, M.A., Yang, W.-S., Baumgardner, J.R., Bunge, H.-P., 2001. Role of a lowviscosity zone in stabilizing plate tectonics: implications for comparative terrestrial planetology. Geochem. Geophys. Geosyst. 2 2000GC000115. Ritsema, J., van Heijst, H.J., 2000. Seismic imaging of structural heterogeneity in Earth's mantle: evidence for large-scale mantle flow. Sci. Prog. 83, 243–259. Ritsema, J., van Heijst, H.J., Woodhouse, J.H., 1999. Complex shear wave velocity structure imaged beneath Africa and Iceland. Science 286, 1925–1928. Roberts, J.H., Zhong, S., 2006. Degree-1 convection in the Martian mantle and the origin of the hemispheric dichotomy. J. Geophys. Res. 111, E06013. doi:10.1029/2005JE002668. Rogers, J.J.W., Santosh, M., 2002. Configuration of Columbia, a Mesoproterozoic supercontinent. Gondwana Res. 5, 5–22. Rogers, J.J.W., Santosh, M., 2004. Continents and Supercontinents. Oxford Univ. Press, New York. Rogers, J.J.W., Santosh, M., 2009. Tectonics and surface effects of the supercontinent Columbia. Gondwana Res. 15, 373–380. Romanowicz, B., 2009. The thickness of tectonic plates. Science 324, 474–476. 24 M. Yoshida, M. Santosh / Earth-Science Reviews 105 (2011) 1–24 Ruppel, C., 1995. Extensional processes in continental lithosphere. J. Geophys. Res. 100, 24187–24215. Santosh, M., 2010a. Assembling North China Craton within the Columbia supercontinent: the role of double-sided subduction. Precambrian Res. 178, 149–167. Santosh, M., 2010b. A synopsis of recent conceptual models on supercontinent tectonics in relation to mantle dynamics, life evolution and surface environment. J. Geodyn. 50 (3–4), 116–133. doi:10.1016/j.jog.2010.04.002. Santosh, M., Zhao, G., 2009. Supercontinent dynamics. Gondwana Res. 15, 225–227. Santosh, M., Maruyama, S., Yamamoto, S., 2009. The making and breaking of supercontinents: some speculations based on superplumes, superdownwelling and the role of tectosphere. Gondwana Res. 15, 324–341. Schmeling, H., 1989. Compressible convection with constant and variable viscosity: the effect on slab formation, geoid and topography. J. Geophys. Res. 94 (B9), 12463–12481. Schubert, G., Turcotte, D.L., Olson, P., 2001. Mantle Convection in the Earth and Planets. Cambridge Univ. Press. 956 pp. Scotese, C.R., 2000. PALEOMAP Project. http://www.scotese.com. Senshu, H., Maruyama, S., Rino, S., Santosh, M., 2009. Role of tonalite–trondhjemite– granite (TTG) crust subduction on the mechanism of supercontinent breakup. Gondwana Res. 15, 433–442. Shapiro, S.S., Hager, B.H., Jordan, T.H., 1999. Stability and dynamics of the continental tectosphere. Lithos 48 (1), 115–133. Silver, P.G., Behn, M.D., 2008. Intermittent plate tectonics? Science 319, 85–88. Sleep, N.H., 1990. Hotspots and mantle plumes: some phenomenology. J. Geophys. Res. 95, 6715–6736. Solomatov, V.S., 1993. Parameterization of temperature- and stress-dependent viscosity convection and the thermal evolution of Venus. Flow and Creep in the Solar System: Observations, Modeling and Theory. : In: Stone, D.B., Runcorn, S.K. (Eds.), NATO ASI Ser. Ser. E. Kluwer, Netherlands, pp. 131–145. Solomatov, V.S., 1995. Scaling of temperature- and stress-dependent viscosity convection. Phys. Fluids 7, 266–274. Solomatov, V.S., Moresi, L.-N., 1997. Three regimes of mantle convection with nonNewtonian viscosity and stagnant lid convection on the terrestrial planets. Geophys. Res. Lett. 24 (15), 1907–1910. doi:10.1029/97GL01682. Stern, R.J., 2008. Neoproterozoic crustal growth: the solid Earth system during a critical episode of the Earth history. Gondwana Res. 14, 33–50. Stern, R.J., Scholl, D.W., 2010. Yin and yang of continental crust creation and destruction by plate tectonic processes. Int. Geol. Rev. 52, 1–31. Storey, B.C., 1995. The role of mantle plumes in continental breakup: case histories from Gondwanaland. Nature 377, 301–308. Suetsugu, D., Isse, T., Tanaka, S., Obayashi, M., Shiobara, H., Sugioka, H., Kanazawa, T., Fukao, Y., Barruol, G., Reymond, D., 2009. South Pacific mantle plumes imaged by seismic observation on islands and seafloor. Geochem. Geophys. Geosyst. 10, Q11014. doi:10.1029/2009GC002533. Tackley, P.J., 1996. On the ability of phase transitions and viscosity layering to induce long wavelength heterogeneity in the mantle. Geophys. Res. Lett. 23, 1985–1988. Tackley, P.J., 1998. Three-dimensional simulations of mantle convection with a thermochemical CMB boundary layer: D"? The Core–Mantle Boundary Region: In: Gurnis, M., Wysession, M.E., Knittle, E., Buffett, B.A. (Eds.), Geodynamics Series. American Geophysical Union, N.Y. , pp. 231–253. Tackley, P.J., 2000a. Mantle convection and plate tectonics: towards an integrated physical and chemical theory. Science 288, 2002–2007. Tackley, P.J., 2000b. Self-consistent generation of tectonic plates in time-dependent, threedimensional mantle convection simulations. Geochem. Geophys. Geosyst. 1 (8). doi:10.1029/2000GC000036. Tackley, P.J., 2002. Strong heterogeneity caused by deep mantle layering. Geochem. Geophys. Geosyst. 3 (1024). doi:10.1029/2001GC000167. Tackley, P.J., 2007. Mantle geochemical geodynamics. In: Bercovici, D., Schubert, G. (Eds.), Treatise on Geophysics, 7. Elsevier, pp. 437–505. Tackley, P.J., Stevenson, D.J., Glatzmaier, G.A., Schubert, G., 1993. Effects of an endothermic phase transition at 670 km depth in a spherical model of convection in the Earth's mantle. Nature 361 (6414), 699–704. Trompert, R.A., Hansen, U., 1998a. Mantle convection simulations with rheologies that generates plate-like behaviour. Nature 395, 686–689. Trompert, R.A., Hansen, U., 1998b. On the Rayleigh number dependence of convection with a strongly temperature-dependent viscosity. Phys. Fluids 10, 351–360. Trubitsyn, V.P., Rykov, V.V., 1995. A 3-D numerical model of the Wilson cycle. J. Geodyn. 20 (1), 63–75. Trubitsyn, V.P., Rykov, V.V., 2001. A numerical evolutionary model of interacting continents floating on a spherical. Earth Russ. J. Earth Sci. 3 (2), 83–95. Trubitsyn, V., Kaban, M., Mooney, W., Reigber, C., Schwintzer, P., 2006. Simulation of active tectonic processes for a convecting mantle with moving continents. Geophys. J. Int. 164 (3), 611–623. doi:10.1111/j.1365-246X.2006.02832.x. Trubitsyn, V., Kaban, M.K., Rothacher, M., 2008. Mechanical and thermal effects of floating continents on the global mantle convection. Phys. Earth Planet. Inter. 171 (1–4), 313–322. doi:10.1016/j.pepi.2008.03.011. Turcotte, D.L., Schubert, G., 2002. Geodynamics. Cambridge Univ. Press, U.K. 456 pp. Unrug, R., 1992. The supercontinent cycle and Gondwanaland assembly: component cratons and the timing of suturing events. J. Geodyn. 16, 215–240. van der Hilst, R.D., Widiyantoro, S., Engdahl, E.R., 1997. Evidence of deep mantle circulation from global tomography. Nature 386, 578–584. van Heck, H.J., Tackley, P.J., 2008. Planforms of self-consistently generated plates in 3D spherical geometry. Geophys. Res. Lett. 35, L19312. doi:10.1029/2008GL035190. Walzer, U., Hendel, R., 2008. Mantle convection and evolution with growing continents. J. Geophys. Res. 113, B09405. doi:10.1029/2007JB005459. Wilson, J.T., 1966. Did the Atlantic close and then re-open? Nature 211, 676–681. Windley, B.F., Garde, A.A., 2009. Arc-generated blocks with crustal sections in the North Atlantic craton of West Greenland: crustal growth in the Archean with modern analogues. Earth Sci. Rev. 93, 1–30. Wolfe, C.J., Bjarnason, I.T., VanDecar, J.C., Solomon, S.C., 1997. Seismic structure of the Iceland mantle plume. Nature 385, 245–247. doi:10.1038/385245a0. Wolfe, C.J., Solomon, S.C., Laske, G., Collins, J.A., Detrick, R.S., Orcutt, J.A., Bercovici, D., Hauri, E.H., 2009. Mantle shear-wave velocity structure beneath the Hawaiian hot spot. Science 326, 1388–1390. doi:10.1126/science.1180165. Worsley, T.R., Nance, R.D., Moody, J.B., 1984. Global tectonics and eustasy for the past 2 billion years. Mar. Geol. 58, 373–400. Worsley, T.R., Nance, R.D., Moody, J.B., 1986. Tectonic cycles and the history of the Earth's biogeochemical and paleoceanographic record. Paleoceanography 1, 233–263. Yamamoto, S., Senshu, H., Rino, S., Omori, S., Maruyama, S., Zhao, D., 2009. Granite subduction: arc subduction, tectonic erosion and sediment subduction. Gondwana Res. 15, 433–442. Yoshida, M., 2008. Mantle convection with longest-wavelength thermal heterogeneity in a 3-D spherical model: degree one or two? Geophys. Res. Lett. 35, L23302. doi:10.1029/2008GL036059. Yoshida, M., 2010a. Preliminary three-dimensional model of mantle convection with deformable, mobile continental lithosphere. Earth Planet. Sci. Lett. 295 (1–2), 205–218. doi:10.1016/j.epsl.2010.04.001. Yoshida, M., 2010b. Temporal evolution of stress state in a supercontinent during mantle reorganization. Geophys. J. Int. 180 (1), 1–22. doi:10.1111/j.1365246X.2009.04399.x. Yoshida, M., Kageyama, A., 2004. Application of the Yin-Yang grid to a thermal convection of a Boussinesq fluid with infinite Prandtl number in a threedimensional spherical shell. Geophys. Res. Lett. 31 (12), L12609. doi:10.1029/ 2004GL019970. Yoshida, M., Kageyama, A., 2006. Low-degree mantle convection with strongly temperature- and depth-dependent viscosity in a three-dimensional spherical shell. J. Geophys. Res. 111 (B3), B03412. doi:10.1029/2005JB003905. Yoshida, M., Iwase, Y., Honda, S., 1999. Generation of plumes under a localized high viscosity lid on 3-D spherical shell convection. Geophys. Res. Lett. 26 (7), 947–950. Yoshida, M., Honda, S., Kido, M., Iwase, Y., 2001. Numerical simulation for the prediction of the plate motions: effects of lateral viscosity variations in the lithosphere. Earth Planet. Space 53 (7), 709–721. Yuen, D.A., Monnereau, M., Hansen, U., Kameyama, M., Matyska, C., 2007. Dynamics of superplumes in the lower mantle. In: Yuen, D.A., Maruyama, S., Karato, S., Windley, B.F. (Eds.), Superplumes: Beyond Plate Tectonics. Springer, Berlin, pp. 239–267. Zatman, S., Gordon, R.G., Mutnuri, K., 2005. Dynamics of diffuse oceanic plate boundaries: insensitivity to rheology. Geophys. J. Int. 162 (1), 239–248. Zhang, S., Yuen, D.A., 1987. Deformation of the core–mantle boundary induced by spherical shell, compressible convection. Geophys. Res. Lett. 14, 899–902. Zhang, S., Yuen, D.A., 1995. The influences of lower mantle viscosity stratification on 3-D spherical-shell mantle convection. Earth Planet. Sci. Lett. 132, 157–166. Zhang, S., Yuen, D.A., 1996. Various influences on plumes and dynamics in time dependent compressible mantle convections in 3-D spherical shells. Phys. Earth Planet. Inter. 94, 241–267. Zhang, N., Zhong, S., McNarmara, A.K., 2009. Supercontinent formation from stochastic collision and mantle convection models. Gondwana Res. 15, 267–275. doi:10.1016/ j.gr.2008.10.002. Zhao, D., 2004. Global tomographic images of mantle plumes and subducting slabs: insight into deep earth dynamics. Phys. Earth Planet. Inter. 146, 3–34. Zhao, D., 2009. Multiscale seismic tomography and mantle dynamics. Gondwana Res. 15, 297–323. 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 Sci. Rev. 59, 125–162. Zhong, S., Davies, G.F., 1999. Effects of plate and slab viscosities on the geoid. Earth Planet. Sci. Lett. 170, 487–496. Zhong, S.J., Gurnis, M., 1993. Dynamic feedback between an non-subducting raft and thermal convection. J. Geophys. Res. 98, 12219–12232. Zhong, S., Zuber, M.T., Moresi, L.N., Gurnis, M., 2000. Role of temperature-dependent viscosity and surface plates in spherical shell models of mantle convection. J. Geophys. Res. 105, 11063–11082. Zhong, S., Zhang, N., Li, Z.-X., Roberts, J.H., 2007a. Supercontinent cycles, true polar wander, and very long-wavelength mantle convection. Earth Planet. Sci. Lett. 261, 551–564. Zhong, S.J., Yuen, D.A., Moresi, L.N., 2007b. Numerical methods in mantle convection. In: Bercovici, D., Schubert, G. (Eds.), Treatise on Geophysics, 7. Elsevier, pp. 227–252.