The tectonics of introversion and extroversion: redefining interior and exterior oceans in the supercontinent cycle Open Access

The assembly and breakup of supercontinents have controlled the evolution of our planet since at least the Paleoproterozoic (Nance et al. 2014). Efforts to describe the kinematic mechanisms of supercontinent amalgamation have resulted in the establishment of two end-member conditions for their formation – introversion and extroversion (Nance et al. 1988; Hartnady 1991; Hoffman 1991). Introversion describes the formation of a new supercontinent through destruction of ‘interior’ oceans formed during rifting of the preceding supercontinent (Nance et al. 1988; Murphy and Nance 2003, 2013). Extroversion, by comparison, describes the formation of a new supercontinent by the destruction of the ocean ‘exterior’ to the previous supercontinent (Hartnady 1991; Hoffman 1991).

In contrast to these kinematic concepts, Mitchell et al. (2012) tied supercontinent cycles to the geodynamic concept of orthoversion, which relates the amalgamation and breakup of supercontinents to global-scale mantle convection patterns. Orthoversion does not consider supercontinent configuration or the age of oceanic lithosphere but focuses on processes such as coupled supercontinent–mantle harmonics and true polar wander (TPW). During orthoversion, the new supercontinent assembles along a girdle of mantle downwelling which is approximately orthogonal to two antipodal, sub-equatorial zones of mantle upwelling, one of which formed beneath the previous supercontinent.

While accepting that orthoversion is the geodynamic rule guiding the assembly of supercontinents, it is still kinematically important to determine whether that assembly preferentially consumes newly formed ‘interior’ or relatively old ‘exterior’ oceanic lithosphere. That is because orthoversion is inherently removed from kinematics, but kinematics are driven by feedbacks between surface tectonic processes and mantle convection patterns, and these ultimately result in the geological expression of supercontinent configuration. The concepts of introversion and extroversion, as originally envisaged over 30 years ago, describe end-member possibilities for supercontinent formation. However, the use of these original definitions for introversion and extroversion (e.g. Murphy and Nance 2003) have resulted in individual supercontinent formations being ascribed to contradictory mechanisms; for example, Pangaea has been variously argued to have formed by introversion from Pannotia (Murphy and Nance 2003), and by extroversion from Rodinia (Li et al. 2019, 2023). A potential area of exploration to correct such contradictions is through using the surface expression of mantle convection (Conrad et al. 2013) to update the tectonic definitions of the processes involved in supercontinent formation. Despite advances in understanding the coupling between mantle convection patterns and plate motions (e.g. Zhong et al. 2007, 2008; Li and Zhong 2009; Yoshida and Santosh 2011, 2014; Yoshida 2016; Heron 2019; Gün et al. 2021; Langemeyer et al. 2021; Wolf and Evans 2021), the tectonic and kinematic descriptions of these geodynamic scenarios in the context of supercontinent cycles are lacking. Here, we propose new tectonic definitions for interior and exterior oceans that acknowledge the improved understanding of coupled supercontinent and convective mantle dynamics, thus allowing for the consistent identification of supercontinent formation by introversion and/or extroversion.

The fundamental difference between introversion and extroversion is the relative geography of the ocean that is closed with respect to the cratons that comprise the preceding supercontinent (Fig. 1). For introversion, the successor supercontinent forms through the closure of young ‘interior’ oceans that opened between the drifting cratons as the previous supercontinent broke up. That is, the oceans that open during supercontinent breakup are the ones that are preferentially closed during amalgamation of the successor supercontinent (Nance et al. 1988; Murphy and Nance 2003; Fig. 1). According to the Murphy and Nance (2003) introversion definition, the oceanic lithosphere consumed is therefore younger than the age of supercontinent breakup. By contrast, extroversion describes the formation of the next supercontinent by preferential closure of the ‘exterior’ ocean that surrounded the previous supercontinent (Fig. 1). In this scenario, the bulk of the oceanic lithosphere consumed must be older than the age of supercontinent breakup (Murphy and Nance 2003). The distinction between interior and exterior oceans was recognized using the Sm–Nd crust formation ages (depleted mantle model ages) of ophiolites accreted in collisional orogens (Murphy and Nance 2003). In interior oceans, model ages must be younger than the age of rifting of the preceding supercontinent, to demonstrate that the oceanic lithosphere that produced the ophiolites formed as a result of supercontinent breakup and did not exist before its assembly. Conversely, ophiolites that yield model ages older than rifting ages must have formed in exterior oceans, as the oceanic lithosphere from which the ophiolites were produced is required to have existed prior to supercontinent breakup.

Improved accuracy in palaeogeographical reconstructions combined with advances in geodynamic modelling indicate that application of the current definitions for introversion/extroversion are problematic for two reasons. First, the definitions assume supercontinental rifting and breakup occurred over a narrow time interval. It is widely accepted, however, that rifting and breakup of supercontinents are notoriously protracted and commonly temporally overlap with the collisions that form the successor supercontinent (Li et al. 2008, 2023; Johansson 2014; Nance and Murphy 2019; Murphy et al. 2021). Second, because the definitions of interior and exterior oceans are defined relative to breakup of the previous supercontinent, they require consensus on the definition of a supercontinent, which becomes an issue in the Ediacaran–Cambrian where the supercontinent status of Pannotia or Gondwana is disputed (Cawood et al. 2001; Meert 2012; Nance and Murphy 2019; Evans 2021; Murphy et al. 2021; Nance et al. 2022). The varying interpretations of the status of Pannotia as a supercontinent highlight problems with the current definition of interior and exterior oceans. For example, in defining a superocean cycle complementary to the supercontinent cycle, Li et al. (2023, 2019) proposed the survival of an exterior ocean from c. 2 to 0.3 Ga, and thus, proposed that Rodinia formed by introversion from Nuna/Columbia, whereas Pangaea formed by extroversion from Rodinia (Pannotia was not considered a supercontinent in their model). By contrast, using the relative age of oceanic crust, Murphy and Nance (2003) identified Gondwana (or Pannotia) to have formed by extroversion from Rodinia, that is, by preferential subduction of the Mirovoi Ocean exterior to Rodinia, with Pangaea forming by introversion, that is, by subduction of the oceans (Iapetus, Rheic) that formed due to the breakup of Pannotia. The ambiguity in the definition of interior and exterior oceans has led to Columbia, Rodinia and Gondwana being described as forming by extroversion and introversion by different workers (Hoffman 1991; Murphy and Nance 2003; Rogers and Santosh 2003; Rino et al. 2008; Oriolo et al. 2017; Li et al. 2019, 2023).

Seismic tomography has revealed that the modern structure of the convecting mantle is a spherical harmonic degree-2 planform (Mégnin and Romanowicz 2000), composed of two antipodal zones of upwelling (large low shear-wave velocity provinces – LLSVPs), associated with dynamic topographical highs centred on the equator, and bisected by a downwelling girdle (Fig. 2) that broadly correlates with the modern location of the circum-Pacific subduction ring (Burke 2011; Torsvik et al. 2014). LLSVP development is argued to be geodynamically driven by supercontinent assembly (Zhong et al. 2007; Mitchell et al. 2020b), and the modern Pacific and African LLSVPs likely formed as a result of the formation of Pangaea (Torsvik et al. 2014). Figure 2 depicts Pangaea reconstructed to 300 Ma overlain on a seismic tomography slice of the mantle demonstrating the relationship between deep mantle and surface dynamics in a degree-2 condition. The two LLSVPs are shown in red colours, one beneath Pangaea (African/continental LLSVP, dubbed Tuzo) and one beneath the palaeo-Pacific ocean (Pacific/oceanic LLSVP, dubbed Jason). The two LLSVPs are separated by a relatively fast shear-wave velocity girdle shown in blue colours – the mantle downwelling girdle. The downwelling girdle broadly correlates with Pangaea's external subduction ring which encircles the supercontinent. The location of the continents and subduction zones are after Matthews et al. (2016). All continental material lies inside the subduction ring and is separated by it from an oceanic hemisphere.

The supercontinent cycle and mantle convection are geodynamically coupled, and numerical modelling of mantle convection and supercontinents also yields a degree-2 planform (Zhong et al. 2007, 2008; Li and Zhong 2009; Conrad et al. 2013; Doucet et al. 2019). A major outcome of this modelling is that following supercontinent assembly, a subduction ring forms along the periphery of the new supercontinent (e.g. Zhong et al. 2007). This external subduction ring separates the planet into two domains, one containing the continents and the other dominated by oceanic lithosphere. Orthogonal to the subduction ring and bisected by the downwelling girdle are two antipodal LLSVPs, one beneath the new supercontinent and the other within the oceanic hemisphere (Yoshida and Santosh 2011). Therefore, the subduction ring is the major tectonic expression of coupled mantle and supercontinent dynamics (Conrad et al. 2013). In this contribution, we apply these subduction ring dynamics to re-interpret tectonic definitions of introversion and extroversion.

The development of a long-lived external subduction ring along the periphery of a supercontinent is an integral feature in supercontinent-coupled geodynamic models (Zhong et al. 2007; Li and Zhong 2009) and is typically preserved as a collage of accretionary orogens (Collins 2003; Cawood and Buchan 2007; Collins et al. 2011; Spencer et al. 2019; Martin et al. 2020b; Cawood et al. 2021). The longevity of the subduction ring is evidenced from the geological record of long-lived external accretionary orogenic systems (circum-Pacific >500 Ma, circum-Columbia c. 800 Ma; Cawood 2005; Collins et al. 2011; Spencer et al. 2013) and outlasts supercontinent cycles (∼600 Ma, e.g. Li et al. 2019). Subduction in the circum-Pacific accretionary orogenic system began in the late Neoproterozoic along the margin of Pannotia/Gondwana (Cawood 2005; Cawood et al. 2009, 2021; Goodge 2020), and likely began earlier in NW Gondwana with the Avalonian–Cadomian arcs initiating by c. 760 Ma (Murphy and Nance 1991; Murphy et al. 2019; Cawood et al. 2021). External accretionary orogens along the margin of the circum-Pacific record >500 Ma history for the Terra Australis segment (Collins et al. 2011), at least ∼450 Ma for the proto-Andean segment (Chew et al. 2008; Martin et al. 2020a), and at least a ∼350 Ma history for the Cordilleran segment (Matthews et al. 2016; Spencer et al. 2019, and references therein), pre-dating Pangaea amalgamation (Nance et al. 2010; Matthews et al. 2016) and outlasting its breakup. Unlike Wilson cycle-dominated collisional orogenic systems, circum-Pacific accretionary orogens develop with continents predominantly residing in the upper plate, relative to the oceanic lower plate. Temporary exceptions occur when subduction zones dip away from the craton. These are short-lived systems because they create a narrow oceanic basin and an oceanic arc (upper plate) with which the flanking continent (lower plate) must inevitably collide. Such arc–continent collisions are typically followed by a subduction polarity flip, which places the continent on the upper plate (Collins 2002).

Long-lived external accretionary orogenic systems have resulted in substantial lateral growth of continents in the western Pacific Ocean (Cawood et al. 2009; Kemp et al. 2009; Collins et al. 2019). They produce zircon Lu–Hf (and whole-rock Sm–Nd) signatures that become increasingly radiogenic (juvenile) with time (Kemp et al. 2009; Collins et al. 2011), reflecting the decreased contribution from the ancient craton as the thinned crust migrates outboard. This overall trend may be punctuated by unradiogenic (crustal) Hf and Nd isotopic excursions related to crustal thickening produced by transient contractional events (Collins 2002; Kemp et al. 2009; Nelson and Cottle 2018; Martin et al. 2020b).

In the Proterozoic, fewer ophiolites are preserved, the palaeomagnetic record is sparse, and continental reconstructions have greater degrees of freedom, making it increasingly difficult to employ the original definitions of extroversion and introversion in earlier supercontinents. However, the record of protracted accretionary orogenesis that identifies the external subduction ring is preserved. Crustal growth on a scale similar to that recorded in circum-Pacific orogens has been attributed to long-lived Paleoproterozoic to Mesoproterozoic external accretionary orogenic systems along the margins of Laurentia, Amazonia, Australia, Kalahari and Baltica (e.g. Karlstrom et al. 2001; Zhao et al. 2002; Cordani and Teixeira 2007; Dickin et al. 2010; Spencer et al. 2013; Roberts and Slagstad 2015; Martin et al. 2020b; Johansson et al. 2022). Here, external accretionary orogens were established at c. 2.0–1.8 Ga, pre-dating final Nuna/Columbia collisions at c. 1.6 Ga (Zhang et al. 2012; Pourteau et al. 2018; Wang et al. 2021), and continued until the terminal collisions of Rodinia assembly between 1.25 and 1.0–0.9 Ga (Karlstrom et al. 2001; Cordani and Teixeira 2007; Roberts and Slagstad 2015; Cawood and Pisarevsky 2017). The onset of subduction coincides with the early stages of assembly of Nuna/Columbia between 2.0–1.8 Ga (e.g. Pisarevsky et al. 2014; Wang et al. 2021) and reflects the establishment of external subduction along the margin of the assembling Nuna/Columbia supercontinent (Li et al. 2019; Martin et al. 2020b). As Nuna/Columbia broke up, roll-back of the external subduction ring towards the downwelling girdle accommodated rifting (e.g. Fig. 3b, c), resulting in increasingly isotopically juvenile crustal compositions within the external accretionary system (Martin et al. 2020b). Prolonged symmetrical subduction ceased when the accretionary margins of cratons along the edge of the exterior ocean collided to form Rodinia by c. 1.0 Ga (e.g. Cawood et al. 2013; Spencer et al. 2013; Martin et al. 2020b). Thus, the fingerprint of the subduction ring in the geological record is long-lived (>500 myr) opposing accretionary orogens.

By contrast, the major orogens of the late Neoproterozoic–Early Cambrian that formed Pannotia/Gondwana (e.g. Pan-African and Brasiliano orogens) and the Phanerozoic orogens that formed Eurasia, are dominated by Wilson-style tectonics. That is, ocean opening, asymmetrical (one-sided) subduction and collisional orogenesis, with one continent on the downgoing plate relative to the other (da Silva Schmitt et al. 2018, and references therein; Spencer et al. 2013). Wilson-style orogens juxtapose ancient cratonic crust against younger mobile belts (Spencer et al. 2013), resulting in the incorporation of ancient cratonic material and more evolved isotopic signatures in magmatic and crustal rocks within the orogens, as demonstrated in the Eurasian collage (Collins et al. 2011).

The formation of a supercontinent leads to the establishment of an external ring of subduction around the supercontinent periphery (Zhong et al. 2007; Li and Zhong 2009; Conrad et al. 2013; Cawood et al. 2016; Spencer et al. 2019). Thus, from both a mantle circulation and tectonic perspective, the exterior ocean should be defined as the oceanic cell that is exterior to the supercontinent and separated from it by the external subduction ring (Figs 2 & 3). By contrast, interior oceans are all those located within the continental hemisphere of the degree-2 structure, separated from the exterior ocean by the external subduction ring (Figs 2 & 3). In this definition, oceans can be defined as interior whether they form as a result of rifting of the supercontinent, such as the Atlantic Ocean, or if their oceanic lithosphere pre-dates rifting, such as the Tethys oceans in Pangaea (Matthews et al. 2016). The latter would be classified as exterior oceans by the Murphy and Nance (2003) scheme.

Figure 3 provides an updated schematic of introversion and extroversion by placing them in a geodynamic context. Figure 3a shows an external subduction ring around the supercontinent (i.e. a surface tectonic expression of a supercontinent) and a mantle downwelling girdle that bisects the two antipodal LLSVPs (i.e. deep mantle expression of a supercontinent). As the downwelling girdle is a dynamic topographical low, there is an overall tendency for the subduction ring as well as the dispersing continents to migrate towards the girdle (Mitchell et al. 2012; Wang et al. 2021; Fig. 3b, c). During introversion, therefore, the continents will coalesce along the mantle downwelling girdle, but remain within the continental cell isolated from the exterior ocean by the external subduction ring (Fig. 3d, e). The oceans that were closed to form the new supercontinent existed within the continental hemisphere and were enclosed within the subduction ring. As the mantle downwelling girdle is orthogonal to the LLSVP that formed beneath the preceding supercontinent, introversion will always satisfy orthoversion (Mitchell et al. 2012; Wang et al. 2021). Introversion can occur either by migration of the cratons sub-equatorially (Fig. 3d) or to sub-polar regions along the subduction girdle (Fig. 3e). In the latter case, large-amplitude TPW would subsequently bring the polar supercontinent to a more equatorial position (Mitchell et al. 2012). Assembly of a supercontinent by introversion is completed by asymmetrical subduction followed by collisional orogenesis that is progressively included within the continental interior through the repeated collision of continental blocks (Murphy and Nance 2003), an example of which is the formation of the Eurasian collage throughout the Phanerozoic (Collins 2003; Collins et al. 2011).

Extroversion must occur by closure of the oceanic cell exterior to the subduction ring (Fig. 3f, g). To close this ocean, continents must drift across the exterior cell and collide with continents on the opposing side. Collisions may occur by cratons from opposing sides of the exterior ocean drifting towards the centre of that ocean where they collide. If so, the new supercontinent will form 180° from the centroid of the previous one (Extroversion I, Fig. 3f). Or, if continents remain stalled at one margin of the exterior ocean, and drift across the ocean from the opposing side, the exterior ocean will close on the mantle downwelling girdle thus satisfying orthoversion (Extroversion II, Fig. 3g). In each instance, the formation of a supercontinent by extroversion results in the collapse of the external subduction ring onto itself (Fig. 3f, g). The common characteristic feature of Extroversion I and II is the collision of long-lived external accretionary orogenic belts with predominately opposing tectonic vergence, which developed along opposing continental margins of the former supercontinent (Silver and Behn 2008; Spencer et al. 2013; Martin et al. 2020b). This feature is recognized in the geological record by the younging of orogenesis and arc magmatism towards the suture on each juxtaposing craton on timescales of >500 myr.

Closure of some interior oceans, orthogonal to the external subduction ring (Fig. 3e, f) occurs during extroversion, while the closure of some captured exterior oceans occurs during introversion. On modern Earth, this would be akin to closure of the Caribbean Sea and eventual collision of South America with North America, or closure of the Bering Strait and collision of North America with Eurasia. As introversion and extroversion are end-member processes, supercontinents likely form by a combination of both, consistent with some geodynamic models (e.g. Yoshida and Santosh 2014). For example, closure of the Carribean Sea, a small captured exterior ocean, involves destruction of a segment of the subduction girdle. However, evidence for the juxtaposition of long-lived (>500 Ma) external accretionary orogenic belts is rare in the geological record. Therefore, we argue that collisional orogenesis and closure of some interior oceans likely occurs prior to an extroversion event, but closure of the exterior ocean does not occur during introversion.

The formation of Pangaea by introversion from Pannotia/Gondwana and Rodinia's extroversion from Nuna/Columbia is presented in a series of plate reconstructions in Figure 4. Our view of supercontinent-coupled mantle dynamics for interpretation of the following section is that, at the time of supercontinent amalgamation, the LLSVPs and mantle downwelling girdle reposition to the degree-2 structure consistent with the reigning supercontinent. These interpretations and complications arising from them are further described and discussed in the mantle convection and further questions section of this paper.

By 530 Ma, Gondwana amalgamation was complete, and the Iapetus Ocean had opened, separating Laurentia and Baltica from Gondwana and thereby breaking up the c. 620–550 Ma supercontinent Pannotia (if indeed, it existed, e.g. Nance and Murphy 2019). Subduction encircled most of Gondwana and bordered the northern margins (reconstruction reference frame) of Laurentia, Siberia and North China (Merdith et al. 2017, 2021). The exterior Panthalassa Ocean occupied the northern hemisphere and was separated from a continental hemisphere to the south by this external subduction ring.

From 520–400 Ma, a convergent plate margin developed along the eastern margin of Gondwana from North China and northeastern Australia to Antarctica and Amazonia, enclosing the interior Paleotethys Ocean within the subduction ring. Plate reorganization by c. 400 Ma was associated with a series of collisional orogenies that led to the subsequent closure of the interior Iapetus and Rheic oceans to form Pangaea (Fig. 4; Merdith et al. 2021). Subduction along the northern margin of Laurentia from at least c. 410 Ma (Matthews et al. 2016, and references therein) and along the western margin of Laurentia from 360 Ma (Matthews et al. 2016; Merdith et al. 2021, and references therein), is simultaneous with convergence and closure of the Rheic Ocean (Wu et al. 2020). This scenario is consistent with all continents lying within the continental hemisphere of the degree-2 mantle structure (Le Pichon et al. 2019, 2021), indicating that Pangaea formed by introversion from Pannotia, and with the continuous, albeit cryptic, history of accretionary orogenesis along the western margin of Laurentia (Spencer et al. 2019).

It has been proposed that the tectonic processes leading to the assembly of Rodinia were different from those of other supercontinents (Silver and Behn 2008; Spencer et al. 2013; Martin et al. 2020b). On either side of the Rodinian suture (Fig. 4), the age of the continental crust decreases towards the suture, reflecting the juxtaposition of opposing and long-lived (c. 800 Ma) external accretionary orogenic belts (Cawood et al. 2013; Spencer et al. 2013; Martin et al. 2020b) formed by prolonged symmetrical subduction of an exterior ocean. Creation of accretionary orogenic systems that are active for over c. 800 Ma is symptomatic of those cratons being positioned along the periphery of a former supercontinent and adjacent to the external subduction ring (Collins 2003; Collins et al. 2011; Spencer et al. 2019).

Subduction initiation within these accretionary orogens at c. 2.0–1.8 Ga defines the external subduction ring that formed along the margin of continents of the developing supercontinent Nuna/Columbia (Fig. 4) (Cawood et al. 2016; Li et al. 2019, 2023). Subduction of a Paleo-Mesoproterozoic exterior ocean as evidenced by prolonged external accretionary orogenesis on the margins of Laurentia, Baltica, Amazonia, Kalahari, Mawson and Australia (Betts et al. 2011; Aitken et al. 2016; Martin et al. 2020b; Johansson et al. 2022) predates Columbia amalgamation and postdates its breakup (external accretionary orogenesis 2.0–0.9 Ga, Nuna/Columbia tenure 1.6–1.3 Ga; Kirscher et al. 2020; Mitchell et al. 2020a). These relationships suggest that the exterior ocean did not close during Nuna/Columbia amalgamation. Therefore, Nuna/Columbia formed within the continental hemispheric cell (assuming degree-2 mantle configuration existed; Mitchell et al. 2012; Li et al. 2019) and the external subduction ring was persevered until Rodinia assembly, at which time it was destroyed (Fig. 4; Silver and Behn 2008; Martin et al. 2020b) by closure of the ocean exterior to Nuna/Columbia. Therefore, the juxtaposition of opposing and long-lived external accretionary orogenic belts is a strong indication that Rodinia formed by extroversion from Nuna/Columbia (Figs 2f & 4). This juxtaposition is characterized by subdued isotopic excursions (zircon Lu–Hf, seawater Sr and whole-rock Pb) recorded during Rodinia assembly, as a result of collision of young margins dominated by juvenile crust (Spencer et al. 2013; Martin et al. 2020b). These signatures, combined with reduced mantle flux recorded by He isotopes and Nb/Th ratios (Silver and Behn 2008), provide a geochemical marker for extroversion. The TPW axis of Rodinia occurs 90° from the centroid of Nuna/Columbia (Mitchell et al. 2012), suggesting Rodinia formed kinematically by Extroversion II (and orthoversion).

The longevity (>800 Ma) of the external accretionary orogens along the periphery of the previous supercontinent which subsequently collided to form the core of the successor supercontinent, is unique to Rodinia amalgamation (Silver and Behn 2008; Dickin et al. 2010; Martin et al. 2020b). Thus, Rodinia is the only supercontinent to have formed by extroversion. The apparent destruction of the external subduction ring that became located in the core of Rodinia raises difficulties in defining the mechanism for the assembly of Pannotia/Gondwana. While many Neoproterozoic reconstructions depict an external subduction ring encircling Rodinia (e.g. Li et al. 2008; Cawood et al. 2016; Merdith et al. 2017), few regions, except for the Arabian–Nubian Shield–East African Orogens (Fritz et al. 2013), record long-lived accretionary orogenesis throughout the Neoproterozoic. Instead, plate reconstructions for the Neoproterozoic depict geometries of subduction systems that become increasingly organized throughout the Neoproterozoic, ultimately encircling Gondwana by c. 550 Ma (Cawood et al. 2016, 2021; Merdith et al. 2017). The assembly of Gondwana along its interior orogens is characterized predominately by single-sided (asymmetric) subduction, which results in extensive reworking of old continental crust in the collision zone resulting in pronounced isotopic fingerprints (Spencer et al. 2013). Given the well characterized nature of most Gondwanan orogens, and, in the absence of a geological record for the closure of an exterior ocean in the Gondwanan collisions (that is, the juxtaposition of opposed external accretionary orogenic systems, e.g. Figs 3f, g & 4; Spencer et al. 2013; Martin et al. 2020b), we propose that Gondwana/Pannotia formed by introversion from Rodinia. Importantly, using our tectonic definition for introversion and extroversion, if the Gondwana/Pannotia cycle is omitted, the mechanism for the assembly of Pangaea remains unchanged, that is, Pangaea formed by introversion from Rodinia.

The subsequent closure (post-300 Ma) of the interior Paleo-Tethys, neo-Tethys and Tethys oceans, through multiple collisional orogens within the continental hemisphere, broke up Pangaea and caused amalgamation of the Eurasian collage, and the first steps of formation of the next supercontinent (Hoffman 1999; Mitchell et al. 2021; Wang et al. 2021). Based on the above tectonic definitions of introversion and extroversion, the next supercontinent may form in three possible configurations presented in Figure 5. First, the formation of Pangaea Proxima (Davies et al. 2018) by Introversion I through closure of the Atlantic Ocean (Fig. 5a) and preservation of the subduction ring. Second, the formation of Amasia (Fig. 5b) by Introversion II through closure of the Arctic Ocean (Mitchell et al. 2012), a scenario that would also preserve the subduction ring. Both future supercontinents that form by introversion should satisfy orthoversion, therefore, the centroid of Pangaea Ultima may lie 30–45° further west (Fig. 5). Third, the formation of Novopangaea (Nield 2007; Davies et al. 2018; Livermore 2018) by Extroversion I or II (Fig. 5c, d) through closure of the Pacific Ocean and destruction of the external subduction ring, in the latter case satisfying orthoversion.

Isotope geochemistry of ocean island basalts derived from deep mantle plumes has identified that the Pacific and African mantle hemispheres have been isolated for at least 700 Ma (Doucet et al. 2020). This highlights the dominance of the degree-2 mantle structure and the longevity of the external subduction ring, but also raises questions about the evolution of the degree-2 structure through time. While some proponents argue for fixed location of LLSVPs (e.g. Torsvik et al. 2006, 2010), supercontinent–mantle coupled models, along with principles for the mechanisms that drive TPW, suggest that LLSVPs are mobile and evolve in response to global tectonics (Wolf and Evans 2021). Wolf and Evans (2021) propose a mechanism to physically divide the source of an LLSVP in the lower mantle though impingement of subducting slabs from the subduction girdle providing an explanation for variation in the expression of LLSVPs and possibly their mobility. The tectonic descriptions presented here suggest that LLSVP mobility enables supercontinent assembly by extroversion (Fig. 4f, g). Assembly of a supercontinent along the mantle downwelling girdle (Fig. 3d, e) by introversion would also likely result in interactions between down-going slabs and the continental LLSVP of the previous supercontinent, which may in turn promote LLSVP migration (e.g. Wolf and Evans 2021).

The structure of the mantle between supercontinent cycles is complex. According to Wang et al. (2021), supercontinent assembly may be a two-step process with the initial nucleation of a megacontinent (Nuna before Columbia; Ukmondia before Rodinia; Gondwana before Pangaea; Eurasia before Amasia) at a degree-1 locus along the degree-2 girdle, prior to assembly of the supercontinent. The degree-2 structure persists today despite Pangaea's breakup and the assembly of Eurasia. Thus, it is not clear from the present mantle structure at what stage in the supercontinent cycle the degree-2 mantle structure reorganizes in response to formation of the subsequent supercontinent. Some models predict the dominance of a degree-1 condition (one hemispheric downwelling and one upwelling, Zhong et al. 2007; Zhang et al. 2010) or a global state of small convections (Li and Zhong 2009) as an interim step between supercontinents. The c. 700 Ma geochemical isolation of the African and Pacific mantle hemispheres argues against the transient degree-1 condition, as loss of the subduction ring that physically divides the hemispheres should allow for some chemical mixing. Alternatively, the mantle structure may be a balance between dominance of the degree-1 and degree-2 conditions as described by Wang et al. (2021).

An important implication of the chemical differences between the two current LLSVPs (Doucet et al. 2020) is that they are driven by different processes and so are unlikely to have the same dynamics. For example, the Pacific LLSVP has 30% less volume and a lower topography than its African counterpart (Doucet et al. 2019). In this case, the geometry of the degree-2 planform may be also variable, such that the downwelling girdle would migrate towards the hemisphere with the weaker LLSVP, enabling extroversion (e.g. Fig. 4c). Figure 2f and g depicts a subdued ‘exterior ocean’ LLSVP, as we speculate the LLSVP could be diminished or deflected through sustained interaction with a slab curtain of the external subduction ring (e.g. Wolf and Evans 2021), a process that may enable extroversion. Future work could focus on further geodynamic modelling of these scenarios that may explain the complex relationship between supercontinents and mantle convection on a supercontinent cycle timescale (c. 600 myr, Mitchell et al. 2019).

To understand supercontinent cyclicity and its impact on Earth, we must decipher the processes involved and, in particular, the dynamic relationships between palaeogeography, tectonics and mantle convection. We propose tectonic definitions for the end-member mechanisms of supercontinent formation, relative to the spherical harmonic degree-2 planform structure of the mantle, which is fundamentally linked to the supercontinent cycle. The subduction ring that forms along the periphery of supercontinents divides Earth into two domains, with one dominated by continents plus smaller oceans and the other a largely oceanic domain. The subduction ring separates interior oceans within the continental hemisphere, from the exterior ocean. The subduction ring is represented geologically by long-lived (>500 Ma) accretionary orogens, and can be identified in the geological record in two geological time intervals: in the Phanerozoic as the circum-Pacific subduction system; and, in the Mesoproterozoic as the c. 800 Ma accretionary orogens of Laurentia, Baltica, Amazonia and Australia that developed along the periphery of Columbia as it assembled. By these definitions, the destruction of the subduction ring would be required for extroversion to occur, leading to the collision of external accretionary orogens. Rodinia is the only example of a supercontinent that formed in this way. The remaining supercontinents formed by collisional orogenesis and therefore by introversion and orthoversion.

We thank the Australian Research Council (grant FL160100168) for continued support. JBM acknowledges a Hadyn Williams Fellowship at Curtin University and JBM and PJH acknowledges the support of NSERC Canada. The authors thank Ross Mitchell and Gabriel Gutiérrez-Alonso for careful and insightful reviews and Shoufa Lin for editorial handling. This is a contribution to IGCP 648 Supercontinent Cycles and Global Geodynamics.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper

ELM: conceptualization (lead), writing – original draft (lead), writing – review & editing (lead); PAC: conceptualization (supporting), writing – review & editing (supporting); JBM: conceptualization (supporting), writing – review & editing (supporting); RDN: conceptualization (supporting), writing – review & editing (supporting); PJH: conceptualization (supporting), writing – review & editing (supporting).

This paper was supported by the Australian Research Council, award FL160100168 to PAC, Funding ID ID0ETYAE4796.

No new data were used in the publication. Data used in this publication (Gplates reconstruction files) are available in Merdith et al. (2021); Pehrsson et al. (2016) and Pisarevsky et al. (2014).