The growth of megacontinent Gondwana resulted from plate reorganization following the breakup of Rodinia during the late Neoproterozoic–Cambrian interval. Gondwana assembly involved the suturing of different blocks within the East African, Brasiliano, Damara and Kuunga orogens and concomitant destruction of the Mozambique, Brasiliano, and Damaran oceans1,2,3,4,5,6,7,8. The East African Orogen (EAO) is a long-lived (>400 Ma) accretionary-style margin at one side of the Mesoproterozoic–Neoproterozoic Mozambique Ocean, which separated India and Azania (central Madagascar, Somalia, eastern Ethiopia and Arabia) from the rest of Africa6,7,8,9. Azania collided with eastern parts of Africa between 640 and 600 Ma6. This was followed by final Gondwana amalgamation between 550 and 500 Ma1,2,3. The assembly of Gondwana coincides with a period of Earth’s history marked by dramatic changes in atmospheric circulation, oceanic circulation, the rapid motion of continents, nucleation of the inner core, hyperactive reversing magnetic field, a decrease of geothermal gradients along subduction zones, climatic variations, steep increase in O2 and the rise of metazoan life10,11,12,13,14,15,16,17.

From a geodynamic perspective, the closure of the Mozambique Ocean along the EAO can be examined in the context of supercontinent assembly and breakup via introversion, extroversion, or orthoversion18,19, which is related to the evolution of the structure of the lower mantle structure beneath Africa20,21. Despite the numerous controversies and general lack of high-quality palaeomagnetic data during the Late Ediacaran, the paleogeographic evolution leading to the closure of the Mozambique Ocean and the development of the EAO is vital for constraining the evolution of the Neoproterozoic Earth system.

Various models that deal with the progressive assembly of Gondwana have been proposed2,3,4,5,6,22,23. Recently, coeval Neoproterozoic orogenic activity was identified in the Tarim craton24 and the Lhasa Terrane hinting at possible links to orogenic events in eastern Gondwana25. In this paper, we compile available geological, geochronological, and paleomagnetic data to present a new orthoversion model for the breakup of Rodinia, closure of the Mozambique Ocean, and amalgamation of central Gondwana.

The Mozambique Ocean and the East African Orogen

The Mozambique Ocean was one of three prominent ocean basins that existed or formed during the Rodinia break-up1,6,7,26,27. The Mozambique Ocean closed as India and Azania converged with eastern Africa (Kalahari, Congo, Sahara metacraton) during the East African Orogeny1,2,3. The EAO encompasses most of eastern Africa extending from the Arabian-Nubian Shield (ANS, in the north) to Mozambique (in the south). Sandwiched within the EAO lies a broad band of Archean to Paleoproterozoic crust between the Indian Shield and Congo–Tanzanian Cratons that were identified Azania and the Afif–Abas terranes28 (Fig. 1). The Azania block and the Afif–Abas terrane (the Al-Mafid Terrane in Yemen and Abas terrane into Saudi Arabia) were isolated microcontinents within the Mozambique Ocean28. Geochronologic and structural evidence suggests that Azania separated as a ribbon continent from the Congo–Tanzania Craton29,30. The Mozambique Ocean was subdivided into a West and East branch by Azania at ~750 Ma6,31. Closure of the Mozambique Ocean involved island-arc collisions and microcontinent accretion (ca. 1080–650 Ma) followed by continental collision with the Congo craton and Saharan Metacraton between ca. 650 and 620 Ma along the length of the EAO (for review see Fritz et al.32). Younger orogenic events in eastern Gondwana include the ~550–520 Malagasy Orogeny between Azanian-Afif-Abas and western India3 and coeval Kuunga Orogeny which completed the amalgamation of eastern Gondwana2. Based on the published geological and geochronologic results (Fig. 2, Supplementary Data), each of these sub-regions within the EAO is described below.

Fig. 1: Distribution of Africa and Eurasian continental crust, ocean basins and plate boundaries at 0 Ma (modified after Merdith et al.7).
figure 1

Light brown polygons are areas of continental lithosphere in the Neoproterozoic that our model used, White polygons are areas of present-day continental lithosphere. ANS Arabian–Nubian Shield; Al Altyn Tagh, Qa Qaidam, Qi Qilian, MB Mozambique belt, NM Northern Madagascar, CM Central Madagascar, SM Southern Madagascar, IAP Iran–Afghanistan–Pakistan, ADMB Aravalli–Delhi mobile belt, M Mewar block, NIB North Indian Block, T Tiekelike, A Aksu, K Kuluketage, MOS Inferred final Mozambique Ocean Suture (after Collins et al.6).

Fig. 2: Late Meoproterozoic-Cambrian magmatic or metamorphic events.
figure 2

Error bars represent the 1-sigma uncertainties. Data source in Supplementary Data. Cam Cambrian, ANS Arabian–Nubian Shield, MB Mozambique belt, NM Northern Madagascar, CM central Madagascar, SM Southern Madagascar, MNIS Malani–Nagar igneous rocks, ADMB Aravalli–Delhi mobile belt. EAO East African Orogen; MO Malagasy Orogen. See text for details, sources, and discussion. Orange, gray, and green shadings are accretion, collision, and rifting/extensional settings, respectively.

The Arabian–Nubian Shield

Meso- to Neoproterozoic age island arcs, including volcanic rocks, gneisses, and sedimentary protoliths, are recognized in the ANS and developed between ca. 1.03–0.93 Ga and ca. 0.87–0.73 Ma31,32,33,34,35,36. In addition, numerous dismembered fore-arc and back-arc ophiolites are found throughout the ANS37,38,39. Blueschist-facies metamorphic assemblages have been reported from slices within western portions of the ANS40. Subduction, volcanic arc formation, and terrane accretion (ca. 830–650 Ma) overlap with migmatization and high-grade metamorphism (T: 700–850 °C and P: 0.7–0.9 GPa) (dated at ~775 Ma and 720–715–Ma34, 41,42). The granitoids in the ANS show early arc-related affinities followed by collision-related calc-alkaline assemblages (~715–700 Ma), subsequent calc-alkaline post-collisional (~640–600 Ma), anorogenic alkaline A-type granitoids and volcanics (~620–550 Ma), and finally by post-collisional extension as exemplified by the intrusion of mafic–felsic dikes (591–545 Ma)42,43,44. Although local variations exist, an age of ~620 Ma is proposed for the cessation of subduction and initiation of the post-collisional extensional regime in the western ANS34,35,44,45.

Neoproterozoic basement of Oman

Basement rock exposure is limited in Oman, which is currently located on the eastern side of the ANS. The volcanic and intrusive activity extended from 850 to 785 Ma and was followed by the emplacement of granodiorite around 750 Ma and culminated with the intrusion of mafic dyke swarms at ~620 Ma46,47,48. Collision-related metamorphic and intrusive activity ceased around 750 Ma resulting in a relatively strong lithosphere and establishing a fundamental tectonic boundary in the central part of the Arabian plate dividing it into two distinct domains6,49 (Fig. 1). Because of the differences in the tectonic history, geochemistry, and geology, Oman is proposed to have occupied a position closer to NW India and Pakistan during the mid-late Tonian49.

The Mozambique belt

The Neoproterozoic MB represents the central and southern regions of the EAO and consists of high-grade granulite- and amphibolite-facies rocks2,50,51. Most published zircon U–Pb age data on metamorphic zircons, metamorphic rims on igneous zircons are well grouped between ~650 and 620 Ma52,53. At least two collisional events occurred within the MB, the first between ~650 and ~620 Ma, particularly in the eastern and western granulite belts, and the second at ~550 Ma, especially along the more southern reaches of the MB including the Tanzania Craton and farther east in Madagascar where it has reached regional ultrahigh-temperature metamorphism (UHTM) conditions2,53,54,55 (Fig. 2).


Madagascar is made up of several domains with Archean to Neoproterozoic rocks. Intrusive igneous rocks formed between 1080 and 900 Ma are restricted to the Ikalamavony domain in southern Madagascar, which was interpreted to represent a magmatic arc and marginal volcano-sedimentary sequence within a continental back-arc tectonic setting outboard of the Antananarivo domain56,57,58,59,60,61. 850–750 Ma granitoids and gabbros are widespread throughout Madagascar57,58,61. Younger, 750–720 Ma volcanic, granitoid, and sedimentary rocks are developed in the Bemarivo domain of northern Madagascar59,62. There are ~670–630 Ma intermediate-felsic volcanic and intrusive rocks developed in the Vohibory and Androyan–Anosyan domains of South Madagascar63,64. Central and South Madagascar were deformed and underwent amphibolite to granulite facies metamorphism between ~650 and 610 Ma57,63,64,65 (Fig. 2). Ediacaran–Early Cambrian (580–510 Ma) post-collisional granitic magmatism, regional metamorphism, and deformation occurred throughout Madagascar (Fig. 2). This younger magmatism overlaps with the latter stages of high-grade metamorphism in northern Madagascar between ca. 550 and 530 Ma (750–850 °C and 6–8 kbar)64,65. In central Madagascar, U-Pb metamorphic ages obtained from zircon rims range from 670 to 540 Ma57. In southern Madagascar, high-temperature and ultra-high temperature (UHTM) metamorphism took place between 580 and 510 Ma based on ages from metamorphic monazite in metapelite, and igneous zircons, and monazite in syn- and post-tectonic granitoids57,63,64,65,66,67 (Fig. 2). The tectonic history of Madagascar is controversial. Collins and Windley28 divide Madagascar along the Betsimaraka suture and include the central and northern regions into Azania. According to the Azania model, central/northern Madagascar collided with the Dharwar craton sometime prior to the final amalgamation of Gondwana. Another model, the so-called SMIWH (South Madagascar–India–Wanni-Highland) model, posits a Paleoproterozoic assembly of these regions around 1.8 Ga from then until the breakup of Gondwana68. More recently, central and southern Madagascar were separated into two distinct domains64. The Vohibory–Graphite–Androyen domains represent the ‘western’ region of Madagascar on one side of the Mozambique Ocean whereas the Anoysen-Ikalamavony domains were positioned on the other side. In that model, the final Madagascar assembly took place at ca. 580–520 Ma, and the Beraketa high-strain zone marks the closure of the Mozambique Ocean.

The Tarim Craton

The Tarim Craton is made up of Archean to Neoproterozoic rocks covered by younger desert deposits that hinder direct outcrop sampling. There are two main areas (Tiekelike and Altyn Tagh–Dunhuang) at the southern margin of the Tarim basin, and the Aksu and Kuluketage areas at the northern margin of the Tarim basin (Fig. 1). In previous models, the Central-South Altyn was considered an integral part of the Tarim Craton; however, the Central-South Altyn is now thought to represent an exotic terrane accreted after the late Neoproterozoic69. The northern margin (N Tarim) is characterized by late Mesoproterozoic–Neoproterozoic tectonothermal events at (Fig. 2; see Ge et al.24 for a review): (1) ca. 1050–900 Ma granitic magmatism and arc volcanic rocks; (2) ca. 830–750 Ma K-rich adakitic granite, 823–800 Ma mafic dykes, 816–787 Ma high-pressure granulites and ca. 830 Ma crustal anatexis; (3) ca. 780–700 Ma basic dykes, volcanic rocks, and granites, ca. 750–730 Ma Aksu blueschists; (4) 680–600 Ma basic dykes, basalt, leucogranites and alkaline/ferroan/A-type granites. These data suggest that N Tarim represents a long-term subduction-accretion arc built along the northern Tarim margin between ca. 1050 and 600 Ma. These ages correspond well with tectonic events in the East African Orogen (Fig. 2). In the Tiekelike and West Kunlun regions along the southwestern margin of the Tarim Craton (SW Tarim), there are ca. 900–880 Ma within-plate bimodal volcanic rocks69, OIB-type basic dykes which formed at 802 ± 9 Ma and the Kudi bimodal igneous complex which crystallized at 783 ± 10 Ma70. Between ca. 0.9–0.7 Ga, SW Tarim evolved as a passive margin with alluvial and shallow-marine deposition and within-plate bimodal volcanism69.

The Lhasa Terrane

The Lhasa terrane (southern Tibetan Plateau) is located between the Qiangtang and Tethyan Himalayan terranes (Fig. 1). The Lhasa terrane contains a suite of Neoproterozoic meta-sedimentary rocks, meta-diabase/gabbro, and meta-granite, which have undergone amphibolite–HP granulite-facies metamorphism and varying degrees of deformation25. Recent studies indicate that the Lhasa terrane is characterized by ca. 930–902 Ma rift-related magmatic and sedimentary rocks71,72, ca. 830–700 Ma arc-related calc-alkaline and tholeiitic mafic rocks and granitoids (Fig. 2); followed by ca. 690–650 Ma collision-related magmatism and HP granulite-facies peak-metamorphism metamorphism and amphibolite-facies metamorphism at 605–590 and 480 Ma25,73. ~530–470 Ma high-K calc-alkaline and shoshonitic granitoid emplacement along with ~530–495 Ma A-type ultrapotassic rhyolites indicate a post-collision setting or an extensional environment associated with an active margin74.

NW India

The 1.1–0.8 Ga Aravalli–Delhi mobile belt (ADMB) of NW India marks a Late Mesoproterozoic–Neoproterozoic subduction–collision orogen, thought to be associated with the collision of the Marwar Block with the North Indian Block during Rodinia assembly75,76 (Fig. 1). The ca. 1.1–0.9 Ga metamorphic event is widespread in metapelites, ortho- and paragneisses, which are characterized by a metamorphic transition from HP granulite to HP amphibolite facies along clockwise P–T paths77. Widespread Meso–Neoproterozoic aged volcano-plutonic rocks constitute major parts of the southern segment of the Delhi Fold Belt (SDFB). Peak metamorphic conditions in the SDFB were followed by isothermal decompression78. The intrusion of the ‘Erinpura granites’ is coeval with the timing of shear activity and retrograde metamorphism of granulite exhumation in the DFB (830–820 Ma)79,80. The youngest Neoproterozoic igneous activity in the region (770–700 Ma) constitutes the Malani Igneous Province81 (NW India) and coeval plutonic rocks in the Nagar Parkar region82 (Pakistan) where ages overlap with Malani but also evidence for magmatic activity as young as ~650 Ma. The Marwar block is extensively intruded by the Erinpura and related granites, and in several places has been covered by younger volcano-sedimentary sequences belonging to the Sindreth, Punagarh, and Marwar groups. The Malani igneous rocks include voluminous felsic lavas and tuffs (occasionally bimodal at the base), granite emplacement, and felsic and mafic dyke intrusions75,81. Because felsic volcano-plutonic rocks are distributed over a large area >100,000 km2 in NW India, SE Pakistan, Seychelles, and Mauritia this late-Neoproterozoic magmatic episode has been regarded as a silicic LIP83. Some have argued that the Malani–Nagar–Parkar LIP formed in an extensional setting81,82,83. The debate regarding the tectonic setting for Malani magmatism is far from settled as support for an active Andean-type margin has been argued based on similar-aged magmatic activity in Seychelles, Madagascar, and Mauritia80,84,85,86,87.

Results and discussion

Ca. 900 Ma Southern Tarim connection with Congo, West Africa, and Lhasa in the southern hemisphere

Traditionally, Tarim is positioned either between Australia and Laurentia88,89 or close to north India or west Australia in Rodinia90,91. However, these locations are difficult to reconcile with the contrasting tectonic settings in north and south Tarim. Geologic evidence from northern Tarim indicates an active continental margin24 and supports a peripheral location of Tarim in Rodinia. However, the development of a rift margin in southwestern Tarim around 900 Ma92, contradicts both the “missing-link“89 and peripheral reconstruction models of Tarim in Rodinia. In those reconstructions, Tarim is placed in the northern hemisphere88,89,90,91. The northern hemispheric choice is presumably based on Tarim’s early Paleozoic affinity with eastern Gondwana. Despite this common assumption, there is no a priori reason to rule out a southern hemisphere position for Tarim between 900 and 800 Ma as discussed below (Supplementary Table 1, Fig. 3). Here we compile high-quality Neoproterzoic paleomagnetic poles from Tarim (Supplementary Table 1, reliability score (R) ≥ 593), and find that a southern hemisphere placement of Tarim before the Ediacaran is possible. We demonstrate that by reversing the 900–720 Ma palaeopoles from Tarim (all poles are listed in Supplementary Table 1), we can generate a simpler APWP through the Neoproterozoic (Fig. 3a). By inverting these poles, we reduce the arc distance between 720 and 635 Ma poles dramatically (from 129.7° to 50.3°) and decrease the APW rates from 17 to 6.5 cm yr−1, We, therefore, propose a palaeomagnetically based reconstruction with Tarim located in southern hemisphere until the Ediacaran (Figs. 3b and 4). Geological evidences are also offered to support our conjecture.

Fig. 3: New Rodinia reconstruction based on paleomagnetic and geological data.
figure 3

a Neoproterozoic paleomagnetic poles of Tarim, all are listed in Supplementary Table 1. Blue colors are traditionally interpreted as north poles used to place Tarim in the northern hemisphere7; green colors with dashed circles are reversed poles, assuming that Tarim was located in the southern hemisphere. O1 and O3 are early and late Ordovician poles from Tarim. b Reconstruction of Rodinia at 900 Ma. Brown-shadowed continents are parts of Rodinia. CS, Congo–São Francisco, Tar Tarim, WA West Africa, Lh Lhasa, Amz Amazonia, Ind India, Yan Yangtze, Cat Cathysia. Brown square with a circle is the 900 Ma pole of Tarim. Euler rotation parameters for India, Yangtze, Lhasa, Tarim, and Congo–São Francisco are listed in Supplementary Table 2, others are after Merdith et al.7.

Fig. 4: Paleogeographic reconstruction focused on breakup of Rodinia and evolution of Mozambique Ocean and Gondwana.
figure 4

a 800 Ma, b 750 Ma, c 635 Ma. Aff Afif-Abas, Az Azania, ANS Arabian–Nubian Shield, S. China South China. Orange and brown curves are fixed true polar wander trajectories (great-circles) between 810–790 and 615–570 Ma for South China (orange) and West Africa (brown).

It is generally accepted that Congo-São Francisco, West Africa, and the Lhasa block were assembled in the southern hemisphere during the Neoproterozoic era71. Several rift-related provinces developed during the 900-880 Ma interval on many of these blocks indicating large-scale breakup. In SW Tarim, ca. 900-870 Ma bimodal volcanic and ca. 900–850 Ma A-type granites were used to argue for an intraplate rift setting91, 92,94. The Bahia–Gangila large igneous province (LIP) in Congo-São Francisco suggests two stages of (950–910 and 890–870 Ma) lithospheric stretching and rifting95. Previous studies based on these ca. 900 Ma rift-related LIPs, and Mesoproterozoic magmatism hint at a possible link between South Tarim, North China, and Congo–São Francisco92. Although all of these continents record 900–860 Ma magmatism, Tarim has no record of rift-related magmatism between 950 and 910 Ma in contrast to prominent magmatism in both Congo-São Francisco and North China. Therefore, we question the proposed links between Tarim and North China or São Francisco. However, on the (present-day) eastern side of Congo, bimodal volcanic and sedimentary rocks in the Zambezi and Lufilian belts in Zambia and the Sankuru-Mbuji-Mayi-Lomami-Lovoy failed-rift basin indicate rifting in southeast Congo between ~890 and 800 Ma96. This history is broadly similar to events in SW Tarim. We propose a link between SW Tarim and the Congo Zambezi belt during the 1.0–0.9 Ga interval. Dyke swarms in the Anti-Atlas region of West Africa (Iguerda-Taïfast LIP, ~885-860 Ma), Hoggar and Reguibat shield (~850–800 Ma) are consistent with a rift setting in West Africa97. Mafic dyke swarms also are reported from Ghana at 915 and 860 Ma98. Geochronological studies on mafic rocks in northern Lhasa indicate a late-stage rift setting at ca. 900 Ma71. We place the Lhasa terrane close to Ghana and near southern India between 1000 and 900 Ma. The mafic rift volcanism may represent a LIP beneath eastern Congo, west Africa, Lhasa, and Tarim. Our placement of Lhasa, India, South China, and Tarim (LIST) in the southern hemisphere at 900 Ma changes the size of the Mozambique Ocean and creates a separate Mawson Ocean between LIST and Australia-East Antarctica (Fig. 3b). We also position subduction zones adjacent to Azania and along the western margin of India which are not present in other full-plate models at 900 Ma (see Fig. 15 in Merdith et al.7).

Neoproterozoic long-lived Andean-type subduction zone and terranes in the EAO

Northern Tarim is characterized by Andean-style subduction-accretion from 900 to 650 Ma which closely parallels the tectonic environment in the ANS, MB, central and southern Madagascar, but is strikingly different from northern Australia and western Laurentia in this period99.

If our ca. 900 Ma southern hemisphere paleogeographic reconstruction for LIST is correct (Fig. 3), then there is a marginal subduction setting in the eastern Mozambique Ocean around ~900 Ma with Azania and Tarim close to the eastern African margin (Fig. 3b). The suprasubduction system, constituted by Tarim–Azania–Afif–Abas–Lhasa (TAL) terranes/blocks, is characterized by volcanic arcs, intrusive rocks and terrane accretion (ca. 830–650 Ma) including migmatization and high-grade metamorphism (Fig. 2). Therefore, we suggest that the TAL terranes constitute the northern margin of the West Mozambique Ocean, and the Saharan, Congo-Tanzania cratons formed the southern margin of the West Mozambique Ocean (Fig. 4a, b).

Peak metamorphism, migmatization, and associated syn-collisional and syn-metamorphic orogenic granites within the TAL developed between 660 and 620 Ma. In our model, TAL collides with the eastern margin of the Congo craton and Saharan metacraton between 660–620 Ma with a general north–south progression. This orogenic belt extends from the ANS, Northern Tarim, Lhasa into Madagascar and the MB (Figs. 2 and 4c). The West Mozambique ocean is closed by continent–continent collision causing granulite-facies metamorphism in Lhasa, southern Madagascar, and the MB (Fig. 2). The metamorphic grade generally decreases from south to north, with lower-grade metamorphism in ANS and Tarim.

Between 590 and 520 Ma, high-grade metamorphism in southern India, Sri Lanka, Madagascar, and Mozambique was interpreted to represent the final collision between India with Azania during the Kuunga or Malagasy Orogeny and closure of the East Mozambique Ocean2,3. Continued compression and shortening between India with Azania and amalgamated Africa might induce the northward-directed escape of Tarim and Lhasa from the EAO via lateral escape tectonics1,100,101.

Evolution of the Mozambique Ocean and growth of mega continent Gondwana

Geological records from the Neoproterozoic ANS suggest a marked change in the orientation of subduction zones in the Mozambique Ocean at ca. 720 Ma, from N–S to E–W directed6. Hence, we divide the evolution of the West Mozambique Ocean into two phases: spreading and closure before and after 720 Ma. Based on our interpretation of the paleomagnetic data from Tarim89,90, we posit that Tarim and Lhasa separated from Rodinia before 850 Ma and the TAL blocks moved toward the equator opening the West Mozambique Ocean (Fig. 4a). The East Mozambique ocean developed as India-South China broke away from Rodinia at ca. 790 Ma102. The lengthy N–S striking subduction system in the East Mozambique ocean appears to be a relic of the circum-subduction system established during the tenure of Rodinia. Following breakup around 720 Ma, the Tonian to Cryogenian magmatic arcs and associated metamorphic basement of the TAL moved equatorward establishing a new E–W striking subduction girdle along the equator (Fig. 4b).

Breakup of Rodinia along the (present-day) western and Arctic margins of Laurentia (~720 Ma) established a new downwelling locus along the relic circum-subduction girdle (Fig. 4b). Subsequent motion of the continents was directed towards this new downwelling (Fig. 4c). The closure of the West Mozambique Ocean was related to subduction reorganization around 720 Ma from a predominately N–S strike to an E–W strike along with spreading in the East Mozambique Ocean6. The growth of the East Mozambique Ocean resulted from the southerly motion of the TAL blocks. Paleomagnetic data from South China indicate little latitudinal motion during the 780–635 Ma interval (Fig. 5), which would also apply to India if the two blocks remained in contact102. Final closure of the East Mozambique Ocean results from the post-635 Ma southward movement of South China–India (Fig. 4b, c).

Fig. 5: Neoproterozoic–Early Paleozoic palaeomagnetic poles of South China (yellow) and Tarim (lime), and the Ediacaran poles from West Africa (brown).
figure 5

Orange and brown curves are fixed true polar wander trajectories (great circles) between 810–790 Ma for South China (orange) and 615–570 Ma for West Africa (brown). The TPW great circles are used to position the loci of subduction and to establish the relative paleo-longitude for South China, Tarim, and West Africa. All poles are listed in Supplementary Table 1.

Our reconstructions also differ from recent full-plate models7 in that we use TPW to constrain the relative paleolongitude of the isolated blocks. Figure 5 illustrates the paleomagnetic data from South China and West Africa between 810–790 and 615–570 Ma with proposed TPW great-circle trajectories (brown and orange lines). Both South China and West Africa APWPs are close to, or overlap, the true polar wander paths. Our interpretation is that South China and West Africa were located near the downwelling girdle that surrounded the Rodinia supercontinent between 810–790 and 615–570 Ma, respectively19. In our reconstruction, the older (>750 Ma) subduction zone in the Mozambique Ocean is a relic subduction system related to the formation of the Rodinia. Therefore, they are subparallel with the true polar wander path (Fig. 4a). After 750–720 Ma, this relic subduction system ceased, and a new subduction system developed, which eventually led to the closure of Mozambique Ocean (Fig. 4b, c). The evolution of these subduction systems is consistent with the orthoversion supercontinent model19, 103 because they lie along a girdle that is ~90° away from the center of Rodinia.

The “introversion” and “extroversion” models of supercontinent evolution predict that succeeding supercontinents will form either by closure of an interior ocean (such as the classic “Wilson-cycle” for the Atlantic Ocean)18 or by the closure of an exterior ocean (such as the Pacific Ocean)104. In the general case, neither the introversion nor extroversion models provide an adequate explanation for the closure of young oceans (e.g. Rheic, Iapetus, and Tethys oceans during Gondwana and Pangea formations). In this specific case, the Mozambique Ocean is analogous to the younger oceans cited above. Its closure fate was sealed, as it just overlapped with the subduction girdle as suggested by the “orthorversion” model (Fig. 4b, c). The new tectonic reconstruction could be turned into a full plate reconstruction and used as a boundary condition of mantle flow models in future work.


Available Neoproterozoic geological and paleomagnetic data are compatible with Tarim in the southern hemisphere during the 900–650 Ma interval (Figs. 2 and 3). The Tarim, Lhasa, Arabia, Azania, and Afif–Abas terranes constitute an enlarged EAO along the northern margin of the West Mozambique Ocean. The West Mozambique Ocean grew in size until ~720 Ma. A post-720 Ma subduction reorganization developed almost orthogonal to the centroid of the former Rodinia supercontinent and resulted in the closure of the relatively young West Mozambique Ocean. This geometry is consistent with the orthoversion model of supercontinent evolution from Rodinia to Gondwana.


The metamorphic and magmatic rocks produced during mountain building is crucial to constrain geodynamic processes. We compiled zircon U-Pb ages between 1.1 and 0.52 Ga from Neoproterozoic metamorphic and magmatic rocks from Tarim, the Arabian–Nubian Shield, Oman, Mozambique belt, NW India, Madagascar, and the Lhasa terrane. The vast majority of ages in this compilation are from felsic to mafic intrusive or extrusive rocks, or their metamorphic equivalents. The dataset is available in Supplementary Data.

This manuscript is based on a tectonic reconstruction of the last billion years that was developed on the open-access software GPlates ( The continent’s shape and rotation files are from Merdith et al.7 with the exception of the rotation parameters of the Tarim, Yangtze, Cathysia, India, Lhasa, and Congo-São Francisco, and Afif-Abas regions. These parameters are listed in Supplementary Table 2. The position of Tarim is constrained by using the palaeomagnetic data to place it in the southern hemisphere prior to 635 Ma.