How Do Supercontinents Assemble?

J Brendan Murphy & R Damian Nance. American Scientist. Volume 92, Issue 4. Jul/Aug 2004.

For a strong dose of humility, consider that not even the land beneath our feet can be taken for granted. For example, geological data indicate with considerable certainty that between 300 and 200 million years ago all of the Earth’s continental land masses were assembled into a supercontinent, which has been named Pangea (meaning “all lands”), surrounded by a superocean known as Panthalassa (meaning “all seas”). Indeed, the evolution of the Earth over the past 200 million years has clearly been dominated by the breakup of Pangea and the resulting formation of new oceans, such as the Atlantic, between the dispersing continental fragments.

For the past 20 years, however, evidence has been amassing that Pangea itself was only the latest in a series of supercontinents that have assembled and dispersed over 3 billion years. Although the mechanisms responsible are controversial, many geoscientists agree that repeated cycles of supercontinent amalgamation and dispersal have not just taken place, but also have had a profound effect on the evolution of the Earth’s crust, atmosphere, climate and life over billions of years.

The amalgamation of Pangea appears to have been preceded by that of Pannotia about 650 to 550 million years ago, and, although its configuration is debated, there is general acceptance of the existence of the supercontinent Rodinia about one billion years ago. Another supercontinent, variously termed Nuna or Columbia, is thought to have amalgamated about 1.8 billion years ago. Two others, Kenorland and Ur, are believed to have assembled 2.5 and 3.0 billion years ago, respectively.

Since the expression “the past is the key to the present” is one of the basic tenets of geology, a strong probability exists that another supercontinent will form in the future. But how would such a supercontinent form, and what would it look like? There are two competing models: One has the continents drift apart and back together again like an accordion; the other proposes that the continents break apart and march all the way around the Earth to reunite on the other side. To determine which is correct, we must first review the basic principles of plate tectonics, the theory that revolutionized our understanding of the Earth by providing a comprehensive explanation of the forces that shape it.

Plate Tectonics

According to the theory of plate tectonics, the Earth has a rigid outer layer, known as the lithosphere, which is generally 100 to 150 kilometers thick and rides atop a hot, plastic layer in the Earth’s mantle called the asthenosphere. Like a cracked eggshell, the lithosphere is broken up into a mosaic of about 20 slab-like fragments, or plates, which move relative to one another at rates that are typically less than 10 centimeters per year. As they move, the plates interact along their boundaries, where they may converge and collide, diverge and separate, or slide past one another. Over millions of years, such interactions have caused mountains to rise where plates collide and continents to break apart where plates diverge.

The continents are embedded in the plates and drift passively with them. Over millions of years, this motion of the continents is sufficient to open and close entire oceans. For example, over the past 180 million years, the divergence between Europe/Africa and North/South America has opened the Atlantic Ocean. The plate boundary along which these continents diverge takes the form of a mid-ocean ridge running the length of the ocean basin. From the crest of this ridge, new ocean floor spreads in both directions as upwelling hot magma from the underlying mantle cools and solidifies, generating new lithosphere between the diverging plates.

On an Earth of constant radius, the creation of new lithosphere in this fashion must be balanced by lithospheric destruction. Over the same period that the Atlantic Ocean has been opening, for example, the convergence of Africa with Europe and of India with Asia has closed an ancient ocean known as Tethys, while the westward motion of the Americas has consumed much of the Pacific.

When continents converge, the intervening oceanic lithosphere re-enters the mantle and is consumed in a process known as subduction. In general, oceanic crust is denser than continental crust, so where continental crust meets oceanic crust at a convergent plate margin, the oceanic lithosphere is preferentially subducted by diving beneath the continental plate. The fact that the vast majority of oceanic lithosphere is no older than 180 million years, compared with lithosphere capped by continental crust up to 4 billion years old, attests to the preferential destruction of oceanic lithosphere. For an ocean such as the Tethys to have closed, more lithosphere must have been consumed by subduction than was created at the Tethys’ mid-ocean ridges.

According to plate-tectonic theory, the process of subduction can be either directly or indirectly responsible for mountain building, or orogeny (from oros, the Greek word for a mountain). First, the subduction process itself generates mountains. Heating of the cold, dense oceanic plate as it descends into the Earth’s interior triggers a wide array of processes above the subduction zone, including the generation of buoyant molten magma. This magma ascends to the surface to fuel volcanoes and swells the crust to create mountains. The Andes are a modem example of mountains formed by such processes, and many of the highest peaks of this range are either volcanically active or were active in the recent past.

The collision of a continent with small continental blocks or oceanic islands (collectively known as terranes) is another form of orogeny. All modern oceans contain islands, such as Japan or the Hawaiian chain, and if the oceanic lithosphere that separates them from the margin of a continent is consumed by subduction, they will be slowly swept toward the continent and must ultimately collide. This collision results in the deformation of rocks and igneous activity that combine to form mountains. Such is the case for the mountains on the west coast of North America. This continental margin has suffered repeated collisions with numerous Pacific islands over the past 200 million years, resulting in the westward growth of the North American plate by some 500 kilometers, from Baja California in the south to Alaska in the north. Collectively, the processes of subduction and terrane collisions form peripheral orogens, so named because they form along the periphery of continents.

If the subducting plate carries continental crust, an eventual collision between continents is inevitable. In this climactic scenario, an entire ocean is consumed, and huge mountains rise as the continents meet head-on. The collision of India with southern Asia to form the Himalayas, and that of northern Africa with southern Europe to form the Alps, are geologically quite modern examples of this process. Because these episodes resulted from the annihilation of continental edges, they are examples of interior orogens. Both were the result of the consumption of the Tethys Ocean, which closed as the Atlantic Ocean opened.

The Supercontinent Cycle

As techniques for dating rocks have become more advanced, geologists have come to realize that orogenic activity is not evenly distributed in time. Instead, there were relatively short intervals of about 100 to 200 million years during which significant numbers of continental collisions took place and other, larger intervals (up to 300 million years long) during which such activity was minimal. This led to a hypothesis known as the supercontinent cycle in which the temporary assembly of all continents into a single landmass, or supercontinent, occurs roughly every 500 million years.

What are the signatures of the steps in this cycle? Periods of breakup and dispersal are recorded first by the injection of basaltic magma into the fractures developed by rifting, to form basaltic dike swarms, and later, as the continents drift apart, by the development of continental margins, like those of the modern Atlantic, and new oceanic lithosphere at mid-ocean ridges. Most geoscientists think that a supercontinent breaks up and disperses because it acts as an insulator that traps mantle heat, much like a cap on one’s head. As a result, the mantle heats up, generating basaltic magma that rises to the surface.

Continental convergence, which ultimately leads to renewed supercontinent amalgamation, is represented sequentially by the onset of subduction and the destruction of intervening oceanic lithosphere, the accretion of terranes to continental margins and, finally, continent-continent collisions. The forces that initiate subduction are hotly debated, although most geoscientists agree that as oceanic plates age, they cool and so become progressively denser, eventually diving into the mantle. As the oceanic plate descends, the force of gravity pulls the remainder of the plate behind it, just as a tablecloth will slide off a table when enough hangs over the edge. As a result, continents embedded in subducting plates are dragged toward the subduction zones, where they ultimately collide.

Following supercontinent breakup, the trailing edges of the dispersing continents become tectonically inert, or passive, and are typified by wide continental shelves, like those that have developed along the margins of the Atlantic since the breakup of Pangea. At the same time, subduction-related orogenic activity continues along the leading edges of the dispersing continents, resulting in a succession of terrane collisions and protracted volcanic activity, like that which has typified the western margin of the Americas since the breakup of Pangea.

What Goes Around Comes Around?

Over the past 30 years, two contrasting models for the global-scale forces that form supercontinents have emerged. These models differ in regard to the type of ocean that closes during supercontinent assembly. When supercontinents rift, fragment and disperse, the world inherits two contrasting types of ocean. Between the dispersing continental blocks, interior oceans are created, which are floored by oceanic lithosphere that is younger than the rifting event of supercontinent breakup. Surrounding the supercontinent, however, is an exterior ocean floor, some of which is older than the rifting event. As the interior oceans widen, the exterior ocean must contract, usually by subducting its oldest lithosphere. As a result, the age contrast between interior and exterior oceans is greatest just after supercontinents break up and diminishes as the resulting continents drift apart.

In one model of supercontinent formation, the interior oceans close. If this model is correct, the next supercontinent will be produced by closure of the Atlantic Ocean and the resulting collision of Europe and Africa with North and South America. In the other model, however, it is the exterior ocean that closes. In this case, the next supercontinent should be produced by closure of the Pacific Ocean. Accordingly, Australia would continue its northward movement into eastern Asia, while North and South America would continue to move west until the Pacific Ocean closed.

As envisaged by Canadian geophysicist J. Tuzo Wilson, the first model proposes that subduction of the oceanic lithosphere generated between the dispersing continental blocks following supercontinent breakup ultimately leads to the assembly and amalgamation of a new supercontinent. In a process often dubbed “accordion tectonics” because of its resemblance to the opening and closing of an accordion, such a supercontinent “turns inside in,” or “introverts.” Thus the interior continental margins of the supercontinent at the dispersal stage become the interior orogenic belts of the next supercontinent. The evolution of the Appalachian-Caledonide-Variscan orogen of North America and Western Europe is a potential example of such introversion. Here, subduction of the oceanic lithosphere created by the breakup of Pannotia some 550 million years ago culminated in the continent-continent collisions associated with the amalgamation of Pangea some 250 million years later.

In contrast, the second model proposes that, following supercontinent breakup, the dispersing continental blocks migrate to the far side of the globe. In this model, the upwelling of mantle heat trapped beneath the supercontinent eventually results in its fragmentation and the dispersal of the resulting continental fragments toward antipodal regions of mantle downwelling and subduction. Hence, the supercontinent in this scenario “turns outside in,” or “extroverts,” such that the exterior continental margins of the supercontinent at the dispersal stage become the interior orogenic belts of the next supercontinent.

In 1991, Paul Hoffman of Harvard University proposed that the breakup of the supercontinent Rodinia some 760 million years ago, “turned Gondwanaland inside out.” Since Gondwanaland is the name given to an ancient amalgamation of the southern continents, this event is a potential example of extroversion. According to Huffman, the Pacific Ocean first formed 760 million years ago when a landmass consisting of Australia and Antarctica rifted away from ancestral western North America. As this landmass, which would later form East Gondwana, migrated away from the site of rifting, the ancient oceanic crust that surrounded Rodinia was preferentially subducted. Subduction continued until East Gondwana collided with an amalgamated Africa and South America (or West Gondwana) some 600 to 550 million years ago to form the supercontinent Pannotia. So, in contrast to introversion, in which the younger interior oceans close to form the next supercontinent, it is the exterior ocean that closes to form the supercontinent during extroversion.

Distinguishing Between the Models

In reality, supercontinent breakup and reassembly by introversion and extroversion occupy the extremes of a spectrum of possibilities. Nevertheless, these two models have first-order differences and produce geodynamically distinct modes of supercontinent amalgamation that should be distinguishable in the geologic record.

The key to determining which model was at play at a given period lies in calculating the age of the oceanic lithosphere that was subducted during supercontinent assembly. In the case of introversion, the subducted lithosphere is younger than that of supercontinent breakup but older than that of the next supercontinent’s amalgamation. In contrast, during extroversion, the first lithosphere to be subducted will have formed before supercontinent breakup.

Unfortunately, the subduction of oceanic lithosphere that culminates in continental collision generally destroys much of the primary evidence that would permit us to distinguish between the two models. But in most orogens, there are local environments in which small fragments of the subducting oceanic lithosphere become detached and stick to the continents, a process known as abduction. In contrast to the fate of most of the oceanic lithosphere, these fragments are preserved as part of the geologic record. Furthermore, volcanic island complexes (known as island arcs) that originally formed above subduction zones but are now within the subducting oceanic lithosphere may be decapitated at subduction zones, where they are scraped off the subducting sea floor and attached to the continental margin. As a result, these may also be preserved as part of the geologic record. In the simplest scenario, the crystallization ages of these accreted complexes would reveal whether they are older or younger than the age of supercontinent breakup, and so reveal whether they formed in an interior or exterior ocean. Because the age contrast between interior and exterior oceanic lithosphere is greatest just after supercontinent breakup and diminishes as the continents drift apart, it is the ages of the earliest accreted complexes that are most diagnostic of the ocean in which they formed.

In both models, however, ongoing subduction within the closing ocean may generate new volcanic island arcs with crystallization ages that will postdate supercontinent breakup regardless of the type of ocean (interior or exterior) that is being subducted. Likewise, some volcanic arcs are generated by subduction at the continental margin and so have crystallization ages that postdate the onset of subduction. In neither case can the crystallization ages of such terranes be used to distinguish between the two models. However, for those volcanic terranes whose evolution is dominated by recycling of oceanic lithosphere, there is a powerful isotopic method that can trace this evolution back through time to reveal the age of the source oceanic lithosphere. This age, which dates the crystallization not of the terrane, but of the oceanic lithosphere that was recycled to produce it, can then be used to distinguish between the two models.

Isotopic Analysis

Isotopes are atoms of the same element that differ slightly in mass. All atoms of an element have the same number of protons in their nucleus, but the number of neutrons in the nucleus-and, with them, the mass of the atom-can vary. Each such variation is an isotope of the element, and many elements have several isotopes. Some isotopes are radioactive-that is, they are unstable and spontaneously decay to a more stable form, giving off energy in the process. The time it takes for half of the unstable “parent element” to decay to the stable “daughter element” is known as the half-life of the radioactive element. This half-life is constant and can be precisely measured in the laboratory. The parent-daughter ratio can also be measured, which, when combined with the known half-life of the parent, can be used to calculate the length of time over which decay has occurred. For a volcanic rock, this is the crystallization age of the sample.

In addition to their use in dating rocks and minerals, certain isotopes can be used as tracers to yield information about the original source material from which the volcanic rock was derived. The pioneering work of several geoscientists, most notably Don DePaolo of the University of California, Berkeley, has shown that the decay of the parent element samarium (Sm) to the daughter element neodymium (Nd) provides one of the best tracers for tectonic processes.

Although the evolution of Sm-Nd isotopes with time is complex in detail (see the sidebar for an introduction), it is analogous to the history of the early European immigrants to North America. Some aspects of the heritage of their home countries, such as their surnames, have been handed down from generation to generation. As a result, genealogists are able to trace that heritage back through time and so deduce the date of their departure from that country. In a similar manner, if we independently know the crystallization age of a rock sample, we can trace the evolution of Sm-Nd isotopes in that sample back through time and so deduce the date that the rock’s chemical ancestry departed from what is called depleted mantle-depleted because light rare earth elements, of which Sm and Nd are two examples, congregate in the liquid during melting, leaving the remaining mantle deficient in these elements. Samarium and neodymium have very similar chemistries, so their ratio is largely unaffected by crustal processes and is thus handed down from “generation to generation” during crustal recycling events. As we shall see, it is the timing of a rock’s ancestral departure from the depleted mantle-its depleted mantle age, or TDM-relative to the timing of the breakup of a supercontinent that reveals how the next supercontinent formed. If a large number of rock samples are gathered, these samples together define an “envelope” on a graph of neodymium isotopic evolution that can clearly distinguish between introversion and extroversion.

Oceanic lithosphere consumed during introversion was derived from the interior ocean that formed after the previous supercontinent broke up, whereas oceanic lithosphere consumed during extroversion was derived from the exterior ocean, some of which formed before the breakup of the previous supercontinent. By combining these observations with our knowledge of the typical Sm/Nd ratio of crustal rocks, we can define an “envelope” of expected isotopic values for the two types of ocean. This approach predicts two distinct envelopes for the interior and exterior oceans (see sidebar). The parameter eNd is the difference in the ratio of two isotopes of neodymium in a sample from that of the bulk Earth at the time the rock crystallized. Isotopic analyses of fragments of the lithosphere of these oceans, or of volcanic terranes formed by recycling them, should therefore enable us to distinguish between the two end-member models of supercontinent amalgamation.

Amalgamation of Pangea

The principal collisional orogenies associated with the amalgamation of Pangea are the Appalachians of eastern North America, the Caledonian Belt of the North Atlantic borderlands, the Variscan Belt of southern Europe and the Ural Mountains of Russia. The breakup of the previous supercontinent, Pannotia, began about 550 million years ago, allowing us to define two envelopes, one each for the exterior and interior oceans. By tracing the ancestry of oceanic rock samples from these mountain belts back to the time they left the depleted mantle, we can produce growth lines for the samples that will lie within one of these envelopes and so clearly distinguish between introversion and extroversion as the mechanism of Pangea’s assembly.

If oceanic terranes accreted during the amalgamation of Pangea were derived from the interior ocean, they must have been derived from the mantle less than 550 million years ago. Although neodymium isotopic data are not yet available for the Urals, a number of oceanic terranes exist within the other orogens that contain rocks with growth lines indicating that they were indeed derived from an interior ocean. Their initial εNd values plot very close to the depleted mantle curve at the time of their crystallization, implying that the magmas were extracted from the depleted mantle reservoir at or about the time of their formation. In each case, these values are consistent with an origin in an interior ocean.

For example, the Appalachians were formed by the closure of an ocean known as Iapetus and contain terranes associated with both the birth and subsequent destruction of this ocean. In Quebec, 600-million-year-old rocks from such terranes have very high initial εNd values (+6.9 to +10.0) that are close to the depleted mantle curve and are interpreted to reflect derivation from the depleted mantle. The age of these rocks coincides with the breakup of the previous supercontinent, Pannotia, and the terranes are thought to reflect the early stages of breakup. Similarly, in Newfoundland, oceanic terranes that are about 480 million years old were obducted onto ancestral North America as the Iapetus Ocean closed and have initial εNd values (+5.6 to +7.7) that are typical of depleted mantle 480 million years ago. Hence, these terranes were derived from the mantle at a time that postdates Pannotia breakup, and so can be considered vestiges of an interior ocean.

Oceanic terranes associated with both the opening and closing of the Iapetus Ocean consequently have compositions that lie at or very close to the TDM at their respective times of emplacement, such that their model ages closely match their ages of crystallization, which do not exceed the age of Pannotia’s rifting. These data, together with the lack of oceanic complexes with model ages older than the rifting event, indicate that the oceanic lithospheric source of these suites was generated in an interior ocean.

The Variscan Belt of western Europe was formed between about 320 million and 285 million years ago by the closure of an ocean known as the Rheic. Terranes associated with both the rifting and subsequent subduction of this ocean are preserved in Britain, France and Spain. In the Massif Central of central France, for example, basalts that formed during the initial rifting of this ocean about 480 million years ago have initial εNd values (around +6.8), similar to that of the contemporary depleted mantle. Similarly, 360- to 350-million-year-old subduction-related basaltic rocks in central France and southwestern Spain have initial εNd values close to depleted mantle values (+6.1 to +8.0 and +7.9 to +9.2, respectively). In southern Britain, an obducted fragment of Rheic ocean floor preserved on the Lizard Peninsula also has εNd values (ranging from +9.0 to +11.8) typical of the depleted mantle and a post-breakup crystallization age of 390 million years.

As with the Iapetus, the correspondence of εNd data with contemporary depleted mantle values, and thus the close match between the depleted mantle model and crystallization ages for Rheic Ocean samples, suggest that the oceanic lithospheric sources of these suites were generated after the breakup of Pannotia, and, hence, were formed within an interior ocean. Taken together, we conclude that Pangea was formed by introversion.

Amalgamation of Pannotia

But is introversion the only way in which supercontinents form? To test this possibility we now turn to the data from collisional orogens associated with the amalgamation of Pannotia, about 600 million years ago. These orogens include the Borborema Belt in Brazil, and the Trans-Saharan and Mozambique belts of North and East Africa, all of which preserve terranes derived from the oceans that closed to form Pannotia. Because the previous supercontinent, Rodinia, broke up about 760 million years ago, oceanic terranes derived from the interior ocean should be younger than 760 million years old, whereas many of those derived from the exterior ocean should have ages between 760 million and 1.1 billion years, the approximate time span of Rodinia.

In the Tocantins Province of central Brazil, juvenile island arc terranes with crystallization ages of 950 million to 850 million and 760 million to 600 million years have initial εNd values of +0.2 to +6.9, and TDM between 1.2 billion and 900 million years. In the Trans-Saharan orogenic belt of southwestern Algeria and southern Morocco, island arc terranes with high initial εNd values (+1.0 to +5.0) yield model ages of 1,200 to 950 million years.

The Mozambique Belt also contains a variety of oceanic arc terranes. At the northern end of the belt in the Sudan, for example, 800-million-year-old basaltic rocks with high initial εNd values (+3.6 to +5.2) have depleted mantle model ages of 900 to 800 million years. To the south, 740-million-year-old rocks with initial εNd values of +2.9 to +3.4 yield TDM ages of 980 million to 960 million years. Hence, each of these orogens contain accreted oceanic terranes with TDM model ages that exceed the 760-million-year breakup age of Rodinia. This suggests that the oceanic lithospheric sources of these suites were generated prior to the breakup of Rodinia; thus the ocean whose closure the orogen records was part of the exterior ocean that surrounded this supercontinent. In contrast to the formation of Pangea, therefore, we conclude that Pannotia was formed by extroversion.


The recognition that supercontinents have repeatedly assembled and dispersed over the past 3 billion years argues strongly that some form of plate tectonics has dominated the Earth’s geology over this time period. It has also led to the suggestion of a supercontinent cycle in which interaction between convection currents in the Earth’s mantle and the overlying lithosphere has caused repeated supercontinent amalgamation and dispersal. However, the current analysis shows that supercontinents can form by fundamentally different mechanisms. It appears that after the breakup of Rodinia, the dispersed continents may well have migrated to regions of mantle downwelling, represented by subduction zones, such that the positions of these regions of downwelling profoundly influenced the formation and location of Pannotia.

The mechanisms responsible for the formation of Pangea, however, are more difficult to deduce. Because the continents did converge, we know that the oceanic lithosphere in the closing interior oceans was consumed by subduction at a faster rate than it was created. On a global scale, the destruction of oceanic lithosphere must be balanced by creation of new lithosphere. Thus the closure of the interior oceans must be accompanied by net lithospheric creation in the exterior ocean. But was it the creation of new lithosphere in the exterior ocean that drove the continents together, or did rapid subduction in the interior ocean pull the continents together, allowing hot mantle to well up in the exterior ocean? It is evident that the mechanisms responsible for the formation of Pangea are proving to be as elusive today as they were when the existence of Pangea was first proposed by Alfred Wegener, the father of continental drift, nearly 100 years ago.