Neoarchean Peridotites in the North China Craton and Implications for the oneset of Plate Tectonics

The compositional and mineralogical distinction between mantle lithosphere underlying old continental cratons and that beneath orogenic belts and ocean basins is well established (Boyd, 1989). The origin of many orogenic and oceanic peridotites as residues of partial melting at pressures less than 3 GPa in the spinel stability field is generally accepted. At such pressures, residues will be Ol-rich. In contrast, the highly depleted character, yet relatively Olpoor (and Opx-rich) mineralogy recognized in xenoliths from many cratons require that such lithosphere is not a simple residue. Hypotheses for the formation of cratonic lithosphere were reviewed by Kelemen et al. (1998) who provided compelling evidence that most of the mantle lithosphere beneath cratons originates by partial melting in the spinel stability field, and that cratonic garnet peridotites with higher equilibration pressures have been tectonically transported to such depths. In this way, ‘on-craton’ lithosphere could be considered oceanic peridotite that has underplated to form Archean continents with the deep lithospheric keels. Nonetheless, of the hypotheses reviewed by Kelemen et al. (1998), none have convincingly tied the petrological origin of cratonic mantle lithosphere to a specific tectonic setting. It has been speculated that Opx-enrichment in cratonic mantle resulting from melt/rock reaction would proceed in a subduction zone, whereas the same feature would occur from high degrees of melting and related cumulate enrichment in the root of a mantle plume. Distinct chemical or mineralogical lines of evidence to link the mantle samples to these tectonic environments is so far lacking. When interpreted in the light of a growing experimental data base for V partitioning (Canil, 1997; 1999) the systematics for V in cratonic and oceanic mantle help to further distinguish the conditions and the tectonic setting during melt depletion to form these two contrasting types of lithosphere. Examination of a data base of 200 spinel- and garnet peridotite analyses of on- and off-craton mantle xenoliths, and orogenic massifs from various settings show that the V abundances in mantle lithosphere lies along two distinct coherent trends (see Figure) when plotted against a depletion index such as wt% MgO. For a given degree of depletion, the V abundances of all cratonic peridotites, whether spinel- or garnet-bearing, are at lower levels than those of oceanic peridotites. The covariation of V with depletion for oceanic peridotites is explained (and modelled) as partial melting in the spinel stability field. Spinel has a DV that is ~ 10 times higher than any other peridotite mineral at a given oxygen fugacity (fO2), and as a residual phase during partial melting at low/moderate pressures, and fO2’s near that of MORB generation (NNO- 2 or -3), will dominate the budget for V. On the other hand, the spectrum of cratonic mantle compositions in the Figure requires either of the following: (1) Oceanic and cratonic peridotites represent residues from two completely different primitive mantle (PUM) sources; the source for cratonic mantle having a lower overall V abundance relative to PUM. (2) Unlike oceanic peridotites, cratonic peridotites are residues of melting at high pressures in the garnet stability field. Because DV Gt is less than DV Sp, less V is retained in the residue for a given degree of depletion. Quite possibly, neither garnet nor spinel are residual in the original melt depletion events related to cratonic peridotites, and both pyroxene and/or olivine, having low DV are responsible for the smaller amount of V in the residues. (3) Cratonic mantle originates by depletion in the Sp stability field at low pressure where Sp is a residual phase, but at much higher fO2 than oceanic peridotites related to MORB generation. The DV for Sp is very fO2 sensitive and varies almost an order of magnitude over the range of terrestrial fO2’s. The lower DV Sp during melting at higher fO2 (near NNO) would result in less V being retained in the residue. Hypothesis (3) would constrain the formation of cratonic peridotites to a region of melt generation in the mantle of relatively high FO2, most likely related to subduction. The distinctly lower Ti/V ratios recognized in many arc-related tholeiites (Shervais, 1982) are a complimentary signature related to this process. The above hypotheses will be quantitatively described and explored using appropriate peridotite melting models and partition coefficients. If hypothesis (3) is accepted then partial melting in the Archean was more oxidized than hitherto realized, some form of subduction was operative, and the mantle lithosphere beneath Archean cratons is intimately related to the subduction process.

It is debated when plate tectonics first operated on Earth. One of the arguments against the Archean (> 2.5 Ga) operation of plate tectonics is the lack of rock records that can be best explained to be formed at convergent plate boundaries, such as continental lithosphere metamorphosed at ultrahigh-pressures (> 2.7 GPa or 80–100 km). Here we report Archean ultrahigh-pressure peridotites in Eastern Hebei, the North China Craton. Bulk-rock and mineral compositions suggest that these peridotites are likely cumulates or slivers of metasomatized continental lithospheric mantle. Garnet pseudomorphs and pyroxene exsolution textures are preserved in these Archean peridotites, indicating decompression-induced breakdown of the original garnet and pyroxene from high pressures. We reintegrate the original garnet and clinopyroxene based on mass proportions and compositions of decompression-induced breakdown products. The reconstructed garnet and clinopyroxene compositions of these Archean peridotites indicate that they were brought up from mantle depths of 110–130 km. We propose that these ultrahigh-pressure peridotites are tectonic slivers of a collisional complex, possibly subducted to mantle depths and then exhumed to crustal levels during Neoarchean subduction and subsequent arc/continental collision, similar to those from Phanerozoic continental collisional zones. The Archean ultrahigh-pressure peridotites in the North China Craton provide direct evidence for operation of continental collisional plate tectonics since at least 2.5 billion years ago.

In general, Archean rocks exhibit rather ordinary moderate-P-high-T facies series metamorphism; neither blueschists nor any record of deep continental subduction and return are documented. However, the abundance and scale of ultrahigh-temperature (UHT) metamorphic belts from the Neoarchean to the Cambrian imply a significant change in geodynamics during the Neoarchean Era, after which transient sites of high heat flow were available at intervals throughout this period of Earth evolution. Many Neoproterozoic-Cambrian UHT metamorphic belts appear to have developed in settings analogous to modern backarcs that were closed and inverted during crustal aggregation and formation of the Gondwana supercontinent. If backarcs were the general setting for UHT metamorphism, then on a hotter Earth the cyclic formation of supercratons (in the Neoarchean Era) and supercontinents (in the Proterozoic Eon) required the destruction of oceans floored by thinner lithosphere that may have generated hotter backarcs than those associated with the current destruction of the Pacific Ocean on the modern Earth. The inherent weakness of the lithosphere in a hotter thermal regime inevitably localized magmatism and deformation at these sites contemporaneously with UHT metamorphism. Medium-temperature eclogites of crustal derivation and associated highpressure granulites are also first recognized in the Neoarchean, for example within the Belomorian Mobile Belt, and they occur at intervals throughout the Proterozoic, for example in the Orosirian Usagaran Orogen and in the Grenvillian belts of the Proto-Atlantic region, and Paleozoic, for example in the circum-North Atlantic Caledonides and European Variscides. Eclogite-high-pressure granulite (E-HPG) metamorphism is predominantly a Proterozoic-Paleozoic phenomenon—complementary to but sparser than UHT metamorphism to begin with, but extending further into the Paleozoic than does UHT metamorphism—that is inferred to record subduction-to-collision orogenesis. Blueschists appear in the Neoproterozoic Era, becoming common through the Phanerozoic Eon; they record the low thermal gradients associated with modern subduction. Lawsonite-bearing blueschists and eclogites, and ultrahigh-pressure (UHP) metamorphism characterized by coesite or diamond are predominantly Phanerozoic phenomena related to deep subduction within subductionto-collision orogens. In general, UHP metamorphic belts in subduction-to-collision orogens are not associated with a contemporary magmatic arc in the hanging wall; this suggests that deep subduction of continental crust may inhibit the generation of calc-alkaline magmas, a feature that may have enabled preservation during exhumation of the mineralogical evidence for extreme pressures. However, an enigma concerning UHP metamorphism is the first evidence of deep subduction of continental crust in the rock record. At issue is the recycling implied by the geochemistry of Archean and Proterozoic diamonds (since entrained from mantle to upper crust in younger magmatic events), which requires a supracrustal component. Are these diamonds evidence of deep subduction of continental crust and an early record of UHP metamorphism, or was some other mechanism (e.g., delamination of underthrust lithosphere, some form of slab break-off) responsible for taking a supracrustal component deep into the mantle source? The Archean and Proterozoic eons were characterized by higher but decreasing mean mantle temperatures and radioactive heat production (RHP), and a thinner thermal boundary layer (TBL) with a shorter residence time than modern Earth. Modeling the effect of increased RHP on the thermal evolution of crust instantaneously doubled in thickness predicts that metamorphic rocks in Archean collisional orogens should have experienced maximum temperatures several hundreds of degrees Celsius higher than those recorded by metamorphic rocks in modern collisional orogens. However, there is no evidence of this—the Archean record is dominated by rather ordinary P-T conditions and crustal melting at relatively low temperatures probably fluxed by water. I argue that a duality of metamorphic belts—reflecting a duality of thermal environments—is the characteristic metamorphic imprint of plate tectonics in the rock record, and it appears only since the Neoarchean Era. Based on the occurrence of both UHT and E-HPG metamorphic belts since the Archean-to-Proterozoic transition, I suggest this transition records the onset of a "Proterozoic plate tectonics regime," although the total ridge length and the number of plates undoubtedly were larger in the beginning than on modern Earth and the style of tectonics likely involved some differences. The Neoproterozoic transition to the "modern plate tectonics regime" registers a change to subduction of continental crust deeper into the mantle and its (partial) return from depths of up to 300 km, a change perhaps related to whole mantle convection as oceanic lithosphere became thicker with decreased thermal gradients. Although there was overlap in time and space between E-HPG and UHP metamorphism during the Paleozoic, since the Carboniferous UHP metamorphism has dominated, and extreme thermal metamorphism is rare.

Stable isotope geochemistry of ultrahigh pressure metamorphic rocks from the Dabie–Sulu orogen in China: implications for geodynamics and fluid regime

Developing the plate tectonics from oceanic subduction to continental collision

Metamorphic evolution of ultrahigh-pressure rocks from Chinese southwestern Tianshan and a possible indicator of UHP metamorphism using garnet composition in low-T eclogites

Regional metamorphism occurs in plate boundary zones. Accretionary orogenic systems form at subduction boundaries in the absence of continent collision, whereas collisional orogenic systems form where ocean basins close and subduction steps back and flips (arc collisions), simply steps back and continues with the same polarity (block and terrane collisions) or ultimately ceases (continental collisions). As a result, collisional orogenic systems may be superimposed on accretionary orogenic systems. Metamorphism associated with orogenesis provides a mineral record that may be inverted to yield apparent thermal gradients for different metamorphic belts, which in turn may be used to infer tectonic setting. Potentially, peak mineral assemblages are robust recorders of metamorphic P and T , particularly at high P – T conditions, because prograde dehydration and melting with melt loss produce nominally anhydrous mineral assemblages that are difficult to retrogress or overprint without fluid influx. Currently on Earth, lower thermal gradients are associated with subduction (and early stages of collision) whereas higher thermal gradients are characteristic of back-arcs and orogenic hinterlands. This duality of thermal regimes is the hallmark of asymmetric or one-sided subduction and plate tectonics on modern Earth, and a duality of metamorphic belts will be the characteristic imprint of asymmetric or one-sided subduction in the geological record. Accretionary orogenic systems may exhibit retreating trench–advancing trench cycles, associated with high (>750 °C GPa −1 ) thermal gradient type of metamorphism, or advancing trench–retreating trench cycles, associated with low (<350 °C GPa −1 ) to intermediate (350–750 °C GPa −1 ) thermal gradient types of metamorphism. Whether the subducting boundary advances or retreats determines the mode of evolution. Accretionary orogenic systems may involve accretion of allochthonous and/or para-autochthonous elements to continental margins at subduction boundaries. Paired metamorphic belts, sensu Miyashiro, comprising a low thermal gradient metamorphic belt outboard and a high thermal gradient metamorphic belt inboard, are characteristic and may record orogen-parallel terrane migration and juxtaposition by accretion of contemporary belts of contrasting type. A wider definition of ‘paired’ metamorphism is proposed to incorporate all types of dual metamorphic belts. An additional feature is ridge subduction, which may be reflected in the pattern of high d T /d P metamorphism and associated magmatism. Apparent thermal gradients derived from inversion of age-constrained metamorphic P – T data are used to identify tectonic settings of ancient metamorphism, to evaluate the age distribution of metamorphism in the rock record from the Neoarchaean Era to the Cenozoic Era, and to consider how this relates to the supercontinent cycle and the process of terrane export and accretion. In addition, I speculate about metamorphism and tectonics before the Mesoarchaean Era.

Where plates converge, one-sided subduction generates two contrasting thermal environments in the subduction zone (low dT/dP) and in the arc and subduction zone backarc or orogenic hinterland (high dT/dP). This duality of thermal regimes is the hallmark of modern plate tectonics, which is imprinted in the ancient rock record as penecontemporaneous metamorphic belts of two contrasting types, one characterized by higher-pressure–lower-temperature metamorphism and the other characterized by higher-temperature–lower-pressure metamorphism. Granulite facies ultrahigh-temperature metamorphism (G-UHTM) is documented in the rock record predominantly from the Neoarchean to the Cambrian, although it may be inferred at depth in some younger Phanerozoic orogenic systems. Medium-temperature eclogite–high-pressure granulite metamorphism (E-HPGM) also is firecognized in the Neoarchean, although well-characterized examples are rare in the Neoarchean-to-Paleoproterozoic transition, and occurs at intervals throughout the Proterozoic and Paleozoic rock record. The fi rst appearance of E-HPGM belts in the rock record registers a change in geodynamics that generated sites of lower heat fl ow than previously seen, inferred to be associated with subduction-to-collision orogenesis. The appearance of coeval G-UHTM belts in the rock record registers contemporary sites of high heat fl ow, inferred to be similar to modern arcs, abd backarcs, or orogenic hinterlands, where more extreme temperatures were imposed on crustal rocks than previously recorded. Blueschists fi rst became evident in the Neoproterozoic rock record, and lawsonite blueschists, low-temperature eclogites (high-pressure metamorphism, HPM), and ultrahigh-pressure metamorphism (UHPM) characterized by coesite or diamond are predominantly Phanerozoic phenomena. HPM-UHPM registers low to intermediate apparent thermal gradients typically associated with modern subduction zones and the eduction of deeply subducted lithosphere, including the eduction of continental crust subducted during the early stage of the collision process in subduction-to-collision orogenesis. During the Phanerozoic, most UHPM belts have developed by closure of relatively short-lived ocean basins that opened due to rearrangement of the continental lithosphere within a continent-dominated hemisphere as Eurasia was formed from Rodinian orphans and joined with Gondwana in Pangea,

Understanding continental subduction: A work in progress

Multistage growth of garnet fingerprints the behavior and property of metamorphic fluids in a Paleotethyan oceanic subduction zone

A classical example of Barrovian type metamorphic sequence is observed in pelitic schists, gneisses and eclogites from Balakot in the southwest to Babusar Pass in the northeast, Kaghan – Naran Valley, Pakistan Himalaya. This sequence comprises the first appearance of chlorite followed by biotite, garnet, staurolite, kyanite and sillimanite. Based on mineral chemistry, pressure – temperature conditions of garnet, staurolite, kyanite zones and eclogites in kyanite zone were estimated as 6.2 – 6.9 kbar and 420 – 478 ° C; 7.1 – 8.3 kbar and 558 – 605 ° C; 12.8 – 14.6 kbar and 658 – 700 ° C; and 27 – 32 kbar and 727 – 799 ° C respectively (Figure 1). Based on detailed field survey in the study area and petrography, the previously called basement and cover sequence of Higher Himalayan sequence from the structural bottom to top is recently classified into three tectonic units referred herein as unit I, II and III (this study). Unit I mainly comprises the basement sequence and has a tectonic contact with the Main Central Thrust to the southwest. This unit mainly consists of pelitic schists and gneisses. Unit II representing ultrahigh-pressure (UHP) metamorphism, is sandwiched in between units I and III. It constitutes the lower cover having pelitic schists; gneisses; felsic gneisses, calcareous gneisses/marbles and eclogites, while unit III is uppermost part of the cover sequence and has a tectonic juxtaposition with low grade Tethyan metasediments locally and Main Mantle Thrust in particular to the northeast. It is also comprised of low grade pelitic gneisses. Presence of coesite relics in clinopyroxene from eclogites and as inclusions in zircon in gneisses from Higher Himalayan crystalline rocks (unit II) gives evidence of deep continental subduction. Geothermal interpolations from petrological data and presence of coesite proves the hypothesis that deep continental subduction occurred when Indian plate collided with Asian plate sandwiching Kohistan Arc approximately at 53 Ma with the closure of Tethys. At the collision boundary marked by Main Mantle Thrust, continental rocks along with oceanic crust subducted beneath Kohistan Arc reaching about 100 ± 10 km depth. Tectonic setup and relative P – T conditions (Figure 2a, b) interprets that the grade of metamorphism in Higher Himalayan sequence increased towards north close to the subduction front. Ultrahigh-pressure metamorphism took place in unit II when it reached to a considerable depth sufficient for the development of coesite. At this event the felsic/pelitic rocks metamorphosed to UHP gneisses and basaltic sills and flows metamorphosed to eclogites. The UHP rocks underwent medium-pressure Barrovian metamorphism during their exhumation stage. SHRIMP data for zircon core and rims from the felsic gneisses of unit II close to eclogite body yields the protolith age as of 253-170 Ma and UHP metamorphic age as 46.2 ± 0.7 Ma (Kaneko et al. 2003). Petrologic and P – T data indicate that these rocks exhumed to earth surface from depths of up to about 90~110 km evidenced by coesite retrogression to quartz and omphacite to amphibole in eclogites.

Plate tectonics shapes our dynamic planet through the creation and destruction of lithosphere. This work focuses on increasing our understanding of the processes at convergent and divergent boundaries through geologic and geophysical observations at modern plate boundaries. Recent work had shown that the subducting slab in central Mexico is most likely the flattest on Earth, yet there was no consensus about what caused it to originate. The first chapter of this thesis sets out to systematically test all previously proposed mechanisms for slab flattening on the Mexican case. What we have discovered is that there is only one model for which we can find no contradictory evidence. The lack of applicability of the standard mechanisms used to explain flat subduction in the Mexican example led us to question their applications globally. The second chapter expands the search for a cause of flat subduction, in both space and time. We focus on the historical record of flat slabs in South America and look for a correlation between the shallowing and steepening of slab segments with relation to the inferred thickness of the subducting oceanic crust. Using plate reconstructions and the assumption that a crustal anomaly formed on a spreading ridge will produce two conjugate features, we recreate the history of subduction along the South American margin and find that there is no correlation between the subduction of a bathymetric highs and shallow subduction. These studies have proven that a subducting crustal anomaly is neither a sufficient or necessary condition of flat slab subduction. The final chapter in this thesis looks at the divergent plate boundary in the Gulf of California. Through geologic reconnaissance mapping and an intensive paleomagnetic sampling campaign, we try to constrain the location and orientation of a widespread volcanic marker unit, the Tuff of San Felipe. Although the resolution of the applied magnetic susceptibility technique proved inadequate to contain the direction of the pyroclastic flow with high precision, we have been able to detect the tectonic rotation of coherent blocks as well as rotation within blocks.

Mid-ocean ridge peridotites record significantly greater variability in major and trace elements, isotopic compositions, and thermodynamic properties such as oxygen fugacity (fO2) than do their basaltic counterparts. This variability may derive from modern ridge processes related to melting and melt-rock interaction or from long-lived source heterogeneity related to recycled material or ancient melting events. In this study, we investigate variations in spinel geochemistry as well as silicate major and trace element chemistry and oxygen fugacity of a suite of peridotites from a single segment of the Southwest Indian Ridge (SWIR). We present new petrographic analysis and trace element data for samples with previously-published fO2 results and combine this with new data for a suite of SWIR gabbro-veined peridotites. We find that SWIR residual lherzolites record low spinel Cr# (Cr# = 100*Cr/(Cr+Al) < 30) and represent low to moderate degrees of melting (∼5-8%) beneath the ridge axis, with no change in oxygen fugacity during melting. In contrast, a subset of SWIR peridotites with high spinel Cr# (Cr#>30) record both melt extraction as well as melt-rock interaction. In these samples, spinel Cr# has been substantially elevated by reaction of spinel to form plagioclase during melt addition, complicating the use of spinel Cr# in mid-ocean ridge peridotites as a proxy for degree of melt extraction alone. While spinel Cr# remains a robust proxy for melt extraction within residual, non-melt-influenced samples, mid-ocean ridge peridotites must first be evaluated to ensure that modification by melt-rock reaction has not occurred. Although addition of MORB melt to a peridotite residue modifies spinel Cr#, this melt addition does not result in significant changes to the fO2 recorded by the peridotite. Residual SWIR lherzolites record fO2 of 0.66±0.39 relative to the quartz-fayalite-magnetite buffer (QFM), statistically indistinguishable from melt-influenced and veined SWIR samples (QFM+1.13±0.61). In contrast to other tectonic settings, such as subduction zones, ocean islands, and continental cratons—locations where peridotite is oxidized by petrogenetically unrelated, presumably high-fO2 melts/fluids—ridge peridotites interact with MORB, which has little to no oxidizing power over its own mantle residues. Thus, modern processes beneath the ridge modify peridotite major and trace elements, but do not generate variability in oxygen fugacity.

Continental deep subduction after the closure of large oceanic basins is commonly ascribed to the gravitational pull of the subducting oceanic slab. However, it is not clear how continental lithosphere adjacent to small oceanic basins was subducted to mantle depths. The Sesia Zone in the Western Alps provides an excellent target for exploration of subduction dynamics in such a tectonic setting. Here we report the first finding of coesite in a jadeite-bearing orthogneiss from the Sesia Zone, providing the first evidence for deep subduction of the continental crust to mantle depths for ultrahigh-pressure (UHP) metamorphism in this zone. Three coesite inclusions were identified by laser Raman spectroscopy in two garnet grains. Based on zircon U-Pb dating and trace element analysis, the UHP metamorphic age was constrained to be 76.0±1.0 Ma. The phase equilibrium modeling yields peak metamorphic pressures of 2.8-3.3GPa, demonstrating the continental deep subduction to mantle depths of >80km. The subducted continental crust was a rifted hyperextended continental margin, which was converted to the passive continental margin during seafloor spreading and then deeply subducted during the oblique convergence between the Adria microplate and Eurasian plate in the Late Cretaceous. Because the slab pull could only play a limited role in closing small oceanic basins for continental collision, the distal push of either continental breakup or seafloor spreading is suggested as the major driving force for the deep subduction of continental crust in the Western Alps. Therefore, deep subduction of the continental crust bordering small oceanic basins would have been induced by the far-field stress of compression, whereas that bordering large oceanic basins was spontaneous due to the oceanic slab pull. This provides a new insight into the geodynamic mechanism of continental deep subduction.

Subduction zones and their associated slabs are the main drivers of plate tectonics and mantle flow, but how these zones initiate remains enigmatic. In the Scotia Sea region, subduction started in the Late Cretaceous/Early Cenozoic in a pristine ocean basin setting devoid of other subduction/collision zones. How this subduction zone initiated remains intensely debated, as exemplified by the variability of published plate tectonic reconstructions. Despite such variability, several works argue for a subduction initiation mechanism, in which a South America-Antarctica relative plate motion change, in combination with a particular plate boundary geometry in the western Weddell Sea, caused convergence across a transform plate boundary segment that subsequently evolved into a subduction zone. Here we discuss this kinematic model of subduction initiation, and, following geometric and kinematic arguments, highlight several unsolved issues that call for alternative explanations. Furthermore, we present new tectonic reconstructions of the Scotia region involving a simpler middle-Late Cretaceous plate boundary configuration, which avoid the geometric and kinematic problems of earlier reconstructions and that call for a new mechanism of subduction initiation. We refer to this mechanism as Subduction Invasion Polarity Switch (SIPS), which involves a long-lived and wide subduction zone (South American-Antarctic subduction zone) with lower mantle slab penetration, which imposes major horizontal trench-normal compressive deviatoric stresses on the overriding plate. This plate consists of a narrow continental lithospheric (land) bridge at the trench (Cretaceous-Early Cenozoic Antarctica-South America land bridge) with oceanic lithosphere behind it (Weddell Sea-Atlantic Ocean). The stresses cause shortening and thrusting at the continent-ocean boundary in the backarc region of the overriding plate, forcing oceanic lithosphere under continental lithosphere, starting the subduction initiation process, and eventually leading to a new, self-sustaining, subduction zone (Scotia subduction zone) with an opposite polarity compared to the long-lived subduction zone. The model thus involves invasion of a new subduction zone into a pristine ocean basin (Atlantic Ocean), with the primary driver being a long-lived subduction zone in another ocean basin. To test the physical viability of the SIPS model, we have conducted numerical geodynamic simulations of buoyancy-driven subduction. Numerical results demonstrate that the SIPS model is viable, with compressive stresses in the overriding plate resulting from strong trenchward basal drag induced by subduction-driven whole-mantle poloidal return flow and compression at the subduction zone plate boundary. Subduction initiation starts in the overriding plate after ∼100 Myr of long-lived subduction, eventually resulting in the formation of a new, opposite-dipping, subduction zone. This new subduction zone develops at the continent-ocean boundary for models without and with a pre-imposed weak zone. We further propose that the SIPS model might explain subduction initiation elsewhere, including the New Caledonia subduction zone in the Southwest Pacific, the Lesser Antilles-Puerto Rico subduction zone in the Caribbean region, and the subduction zones that consumed the Rocas Verdes and Arperos backarc basins in South America and Central America, respectively. We further postulate that active backarc shortening in the Japan Sea, with eastward under-thrusting of Japan Sea oceanic lithosphere below the Japan arc, represents an early stage of SIPS.