This transcript was generated automatically and lightly edited for readability.

Oliver Strimpel: This is Geology Bites with Oliver Strimpel. The theory of plate tectonics has been widely accepted for about 50 years. It tells us that the surface of the Earth is composed of about a dozen rigid plates that are in constant motion. There are two basic plate types: oceanic plates, which are continually being created at mid-ocean ridges and destroyed at subduction zones, and continental plates, which are formed from material distilled out of melted mantle rocks. But understanding the processes that drive plate tectonics does not tell us the actual history of plate motions, plate formation and plate destruction that has taken place throughout Earth's history. Dietmar Müller is a professor of geophysics at the University of Sydney. He is interested in combining the wealth of very disparate geological and geophysical data that is available into a self-consistent picture of the Earth's history over deep time. Over the past decade, his group has completed the Herculean task of building just such a model, spanning a full billion years of Earth history. You might want to pause now to watch the 42-second animation of this entire span of time. There's a link on the podcast web page and in this episode's notes. Dietmar Müller, welcome to Geology Bites.

Dietmar Müller: Thanks for having me.

Oliver Strimpel: How did the idea of reconstructing the motions of the Earth's plates over hundreds of millions of years come about?

Dietmar Müller: As you mentioned before, the 50-year-old theory of plate tectonics merely explains how plate tectonics works, and not how plate motions have actually evolved through time, across several cycles of supercontinent assembly, breakup and dispersal. The rise of computing power in the '80s and '90s has made it possible to apply the theory of plate tectonics to the actual evolution of the Earth.

Oliver Strimpel: So one of the challenges of building such a model is that the sources of data are extremely heterogeneous. What kinds of data sources did you actually use?

Dietmar Müller: Yeah, the sorts of data we used differ a lot depending on how far back in time we are going. Most of the preserved ocean floor is younger than 170 million years; that corresponds to the Late Jurassic, when dinosaurs roamed the Earth. Ocean floor is the best constraint for modeling plate motions, because it contains magnetic stripes visible in the Earth's magnetic field. So ocean crust is like a giant tape recorder: while it is being created at mid-ocean ridges, it records reversals in the Earth's magnetic field. We can use magnetic field data to make fairly accurate models of plate motions through time. But reconstructing pre-Jurassic ocean floor is much harder, and even younger ocean floor that's now been subducted and no longer exists is more difficult. The same also goes for the motions of terranes in very complex regions, like the Caribbean or parts of Southeast Asia. For this, we need to compile a diversity of geological and geophysical data to assess the geological affinities of different terranes and continents through time, pretty much like Alfred Wegener did when he developed the theory of continental drift, but using digital databases. These data include structural, petrological, paleontological, stratigraphic and age constraints from regional geology. In addition, we also track the latitude of terranes using paleomagnetic data.

Oliver Strimpel: Gosh, that's a really wide-ranging and disparate set of data. So how did you manage to wrangle it all into a self-consistent model of the Earth?

Dietmar Müller: Yeah. So the problem is that going back in time, less and less of the geological record is preserved. So in order to generate a complete set of tectonic maps through time, we need to interpolate between all these sparse data. But the biggest hurdle that we faced in doing that was not actually the incompleteness of the data itself, but having software that allows the user to take all this data and turn it into a self-consistent model. One of the basic rules of plate tectonic theory dictates that all plate boundaries must be connected. And so the further one goes back in time, the more gaps open up in the plate boundary system. So we have to interpolate these plate boundaries, and we needed software that would allow us to essentially reconstruct a continuously evolving network of tectonic plate boundaries through time. That's why we developed our GPlates software. It's community software that allows us to cookie-cut, as we say, all our geodata. We assign the data to their respective tectonic plates and then rotate them back through time. GPlates is designed to do all of this, and to go much farther than just continental drift. Wegener, of course, was completely unaware of the existence of plate boundaries, and plate tectonic theory clarified how plate boundaries work. But without GPlates, we wouldn't have been able to develop a global model through deep time.

Oliver Strimpel: I can see that it must be an immensely complex and, in some ways, very messy problem. From what I can tell, people often try to reconstruct a regional history, and they don't think, well, this all has to fit on the surface of the Earth. They don't worry about what happens on the other side; that's not their department. So was one of GPlates' main roles to help handle the complexity while still enforcing this global self-consistency of the model?

Dietmar Müller: Yeah, that's exactly right. GPlates is designed to enforce the rules of plate tectonics. For example, we want to ensure that plates do not exceed plate speed limits, and we want to ensure that other rules are not violated. For example, we don't want to have so-called quadruple junctions. GPlates also ensures that the Earth is not expanding or contracting, and it allows us to make sure that the entire shell of the Earth, all the plates as a group, doesn't rotate at an unreasonable speed.

Oliver Strimpel: If triple junctions are allowed, why not quadruple junctions of plates?

Dietmar Müller: Yeah, that's a good question. Quadruple junctions are places where four plates would come together. When plate tectonic theory was initially developed in the late '60s and early '70s, it was shown that they are unstable and merely exist for fleeting moments of geological time, if at all, and therefore we cannot have them as persistent features in our model.

Oliver Strimpel: Does GPlates assemble a purely two-dimensional jigsaw puzzle of a global spherical shell over time?

Dietmar Müller: Yeah. It turns out that observations of sinking slabs from seismic mantle tomography are now becoming integrated into our plate models, and they're used to constrain the location of subduction zones through time, especially for intraoceanic subduction, where little other information can be used to reconstruct their location. Seismic tomography works like an ultrasound system, using seismic waves generated during earthquakes. These waves travel through the mantle and the crust, and through subducted, sinking slabs, which accelerate these waves, and the waves are eventually recorded. When a large set of seismic travel times is used together to generate images of the mantle, we can see subducted slabs. And if we then assume that subducted slabs mostly sink vertically, and if we have a rough idea of their sinking speed, we can infer the surface location of subduction zones through time by looking at the location and depth of these subducted blobs of material in the mantle.

Oliver Strimpel: Hmm, that's interesting. So there really is a very important third-dimensional aspect to what is still essentially a two-dimensional model.

Dietmar Müller: That's exactly right. But it only helps us for the last 200 million years, because earlier subduction is not reflected in mantle structure very well.

Oliver Strimpel: So despite your ability to integrate such a diversity of evidence about the history of the plate motions, are there gaps or inconsistencies in the data, such that as you go back in time, or for a certain interval of time, the history still remains very uncertain?

Dietmar Müller: Yeah. So the model of the continents is pretty well constrained, but of course the continents were separated by ocean basins, and it becomes much harder to reconstruct ocean basins before the Late Jurassic, seeing that the oldest preserved ocean floor is about 170 million years old, in the northwest Pacific. For reconstructing these ancient ocean basins, all we can do is generate scenarios using the rules of plate tectonics and some sparse preserved data on the continents that give us some clue as to what such oceans might have looked like. For example, take the Tethys Ocean, which separated the ancient continent of Gondwana in the south and Eurasia in the north. In this ocean, terranes were rifted off Gondwana, crossed the entire ocean basin and were eventually accreted to Eurasia. These terranes are very handy, as their geological and paleontological data allow us to reconstruct the creation and destruction of ocean basins during the passage of these terranes across the Tethys. So even though all this ocean floor is now gone, the terranes give us some indirect clues as to the existence of these now vanished ocean basins.

Oliver Strimpel: One of the things that was discussed in some of the previous podcasts in the series is that there's a period before Pangaea and after Rodinia when the Earth's magnetic field went pretty much haywire, or maybe just went close to zero, and so the paleomagnetic data are rather sparse, or just not really there. Was that one of the periods that you were still able to model with all the other kinds of data you mentioned earlier, like the petrological, paleontological, stratigraphic and so on?

Dietmar Müller: Yeah, that's right. For those times, the geological and stratigraphic data become especially valuable, because they tell us something about the affinity between different plates and terranes, their juxtaposition. And we can also use paleoclimate indicators, which tell us something about the paleolatitude of individual continents and terranes without paleomagnetic data. So we need to rely more heavily on geological data when we have a sparsity of paleomag data.

Oliver Strimpel: So you and your colleagues finally published your billion-year model in February of 2021. Can you estimate how many separate geological studies you incorporated into this model? And is there a measure of how much effort went into this?

Dietmar Müller: Yeah, I think this is the most massive paper I've ever been involved in. It has nearly 500 references, and many of the papers that we reference in turn summarize other papers. So I'd say it's easily over a thousand geological studies that form the basis of this paper, perhaps conducted over half a century. The whole process is so labor-intensive because the model making involves a lot of decisions when it comes to connecting plate boundaries in the face of sparse data. We need to review a ton of the positions of the plates, and we need to review their history: Where did they come from? Where are they going to go? And the whole evolution needs to be self-consistent and obey the rules of plate tectonics. At the moment, we are not yet able to teach a computer to do all this work by itself. So we actually need a geologist with wide-ranging knowledge, or a team of geologists, to collaborate on this and to pick out things that don't work and say, oh no, this is unreasonable. So it's a long, iterative process. The bulk of this work was led and coordinated by my ex-student Andrew Merdith. And I also want to say that this work would not have been possible without Alan Collins from the University of Adelaide and his group of students. They were actually the experts in pre-Pangaea geological data sets. They've done a lot of fieldwork across the Gondwana continents, including Australia, India, Madagascar and Africa.

Oliver Strimpel: I must say, the result is truly mesmerizing. I saw the video, actually, on the front page of the New York Times when it came out, and it shows not only the locations of the plates and their motions, but what kind of edges they have between them: the spreading ridges, the subduction zones, the boundaries where plates just slide past each other. I mean, there's a huge amount of information there. Do you know how many people have watched the video by now?

Dietmar Müller: Yeah, the video had a huge impact. It's been viewed nearly two million times on YouTube. Nothing we've ever done before has had such an enormous impact. I highly recommend watching the video; it provides a view of the dynamic Earth unlike anything you would have seen before. You'll see an ever-changing web of plate boundaries that enclose shifting continents, imaging the growth and destruction of ocean basins. The Earth has always been a story of dramatic change, with multiple cycles of supercontinent assembly and dispersal, and the movie really brings home that point.

Oliver Strimpel: Just seeing this magisterial overview of the planet's tectonic motions through time satisfies our historical curiosity. But this must also have far-reaching implications for geoscience.

Dietmar Müller: Yes. It connects many aspects of geology and climate: the making of mountain belts, volcanic arcs, continents and ocean basins, through greenhouse and icehouse climates. We are hoping that this model will establish a totally new, holistic approach to deep-time Earth system modeling that will ultimately allow us to gain better insights into how tectonic cycles regulate the concentration and availability of elements in the atmosphere and oceans, which are responsible for modulating climate and biological evolution.

Oliver Strimpel: When we talked earlier, you also mentioned that looking at this global history through deep time also throws light on processes going on in the deep Earth. Can you say a bit more about that?

Dietmar Müller: Yeah. The subduction of oceanic plates stirs up the mantle. Subducting plates will sink all the way to the core–mantle boundary, where they pile up and ultimately give rise to mantle plumes, which are cylindrical upwellings of mantle material that cause giant volcanic eruptions when they reach the surface. These so-called large igneous provinces are known to have played a role in driving mass extinctions through time. So there's a clear connection between tectonics, volcanism, and extinctions and radiations of life as well. Of course, the most famous example of tectonics and volcanism driving extinctions is the so-called Siberian Traps, which erupted 250 million years ago and caused an event also known as the Great Dying. It's the Earth's most severe known extinction event, with around 81 percent of all marine species and 70 percent of terrestrial species becoming extinct.

So the models of plate motion and mantle convection over a billion years allow us to study how plate tectonics has driven subduction history, and how subducted slabs have sunk through the mantle and piled up on the core–mantle boundary. We have already made some interesting observations through these models. For example, there are large delays between changes in plate tectonic configurations at the surface and the time it takes for these changes to be seen at the core–mantle boundary, to percolate their way down to these deepest portions of the Earth. It takes anything between 160 and 240 million years for the lower mantle structure to start reflecting plate motion changes at the surface. And so we're learning something about the time offsets in these connections between plate tectonics, mantle convection, the stirring at the core–mantle boundary, plumes coming up, and then giant volcanic eruptions occurring.

Oliver Strimpel: Could you give us a specific example of how the timing of events documented in your reconstructions is actually manifest in the geological record?

Dietmar Müller: Yeah. There was a period of what we called slab superflux into the mantle in the Early Cretaceous, peaking around 130 million years ago. This spike reflects a twofold increase in mid-ocean ridge lengths following the breakup of Pangaea, and a contemporaneous increase in convergence rates, with average speeds exceeding 10 centimeters per year. That's pretty fast as far as plate tectonics goes, as plates normally move at the speed at which fingernails grow, which is more like 2 to 5 centimeters per year. So this slab superflux drove a surge in slab penetration into the lower mantle, and an associated increase in the vigor of mantle convection and mantle return flow, because for everything that goes down, something has to come up. So this mid-Cretaceous event may have triggered the formation of the so-called Darwin Rise, a topographic swell in the Pacific. In a recent paper led by one of my ex-students, Madison East, who is now doing a PhD at Cambridge, we proposed that the immense pushing up of the surface in the Cretaceous Pacific Ocean actually reflects this large-scale mantle return flow in response to the very fast and voluminous plate subduction in the Cretaceous. This upward return flow is naturally concentrated in the interior of plates, far away from subduction zones, where the mantle is dominated by downward flow. So if we look at this globally, we find two global highs of this massive upward flow of the mantle today: one underneath the central South Pacific and one underneath Africa. But in the Cretaceous, these upwellings went into overdrive and pushed the surface up much more than they do today.

Oliver Strimpel: So can your model help us understand the timing of the changes in the Earth's climate over geological time as well?

Dietmar Müller: Yes, indeed it can. The disposition of continents and mountains affects ocean and atmospheric circulation. And I want to go back again to the time when the supercontinent Rodinia existed. About 750 million years ago, in a period called the Neoproterozoic, a very unusual configuration of the tectonic plates occurred, causing one of the most striking episodes of the planet's geologic and atmospheric history, with the Earth's surface completely covered by ice, hence the term snowball Earth. This process started when the supercontinent Rodinia formed, about 1.1 billion years ago. But when it started to break into smaller continental plates, things started to change. Rodinia was located near the equator, and the resulting continents also remained in the tropics. Rodinia itself, while it was still assembled, was probably extremely arid, because large amounts of moisture couldn't reach its interior. But when it broke up, the smaller continents had a warm and humid climate. And so this configuration accelerated the weathering of rocks at the surface, and the process of silicate rock weathering removes large quantities of CO2 from the atmosphere.

How does this actually work? Carbon dioxide is dissolved in rivers as part of the erosion of rocks, and this carbon dioxide is transported to the ocean, where pelagic organisms like corals use it to make their skeletons. When they die and fall to the seabed, the associated carbon gets stored away for a long time. After the supercontinent Rodinia broke up, atmospheric CO2 levels sank below 200 ppm; that's less than half our current levels. With this little CO2, ice sheets began to grow, reflecting more and more sunlight back to space and cooling the climate even further. And of course, the weathering process also slowed right down, with the continents being covered by ice. Eventually, the entire Earth was covered by ice. Weathering was brought to a halt, and this great carbon storage mechanism in the deep sea no longer worked, because it was interrupted at its source. So, ultimately, volcanism, as a by-product of tectonics, is assumed to have pumped enough CO2 into the atmosphere to help melt the ice again. But how exactly these feedbacks work between tectonics, volcanism, weathering and deep-sea sedimentation is still a very active area of research. It's particularly relevant for understanding the future pathways of the Earth.

Oliver Strimpel: Does the global timing of the disposition of the continents that we can get from your model also have implications for the evolution and distribution of life?

Dietmar Müller: Absolutely. The processes that govern radiations and extinctions of life are known to be closely connected to plate tectonics. Recently, a global fossil database was connected to paleogeographic reconstructions by Shanan Peters and his collaborators at the University of Wisconsin in the USA, and they found that a significant component of the trajectory of marine animal diversity over the past 440 million years can actually be attributed to the assembly and disassembly of the supercontinent Pangaea, which, of course, changed the configurations of ocean basins and sea level, and modulated the continental area covered by shallow seas. Mountains are another connection between plate tectonics and biodiversity, as originally observed by Alexander von Humboldt. With about 25% of all land area, mountain regions are home to more than 85% of the world's species of amphibians, birds and mammals. But such a large area of mountains is a very recent phenomenon. The Earth didn't always look the way it looks now. This unusual state of the Earth is restricted to the last 50 million years, when the progressive dispersal of the continents after the breakup of Pangaea resulted in continents running into each other and causing mountain building. In the previous 150 million years, the Earth was quite flat and had relatively high sea level, and so in those days, biodiversity was mainly concentrated in shallow seas, not in mountainous regions. So one can easily see how plate tectonic cycles play a huge role in modulating biodiversity.

Oliver Strimpel: Does the model enable us to keep the clock running into the future and predict what's going to happen next?

Dietmar Müller: Yes, definitely. The big unknown is whether the Atlantic Ocean will continue to open or whether a subduction zone will be initiated along one of its margins, and the result will be dramatically different. Either the Americas will run into eastern Eurasia sometime in the future and the entire Pacific Ocean will have vanished, or the Americas will join up with Africa again, forming a Pangaea-like continent. And I don't know which one it's going to be.

Oliver Strimpel: My last question: if you had unlimited research funds, how would you use them?

Dietmar Müller: I would build a 3D model of the world's geology, something like a global stratigraphic database, which doesn't currently exist. It would allow us to understand the deep structure of the continents, the history of intrusions, the connections between uplift, erosion and sedimentation, and therefore the connection between deep Earth dynamics, mantle convection, and plate and surface processes through time. Erosion is a killer, because, just like with subduction, the eroded material gets removed, but this eroded material contains a lot of information about uplift and surface processes through time. Now, lucky for us, much of the material that gets eroded in elevated regions of continents is still there, but it's deeply buried in sedimentary basins. So if we knew the deep structure of global sedimentary basins in better detail, we would be in a much better position to build a virtual Earth, including making models of the carbon cycle through time. Much of these data actually exist, but they have been collected by exploration companies. If I could coordinate industry to share at least some of these resources for research and make them available to the community, it would be an enormous boon to a better understanding of the Earth's evolution through geological time. So we wouldn't even need unlimited research funds to make that happen.

Oliver Strimpel: Dietmar Müller, thank you very much.

Dietmar Müller: My pleasure.

Oliver Strimpel: For more about Geology Bites, as well as pictures and diagrams that illustrate this podcast, you can go to geologybites.com.