This transcript was generated automatically and lightly edited for readability.

Oliver Strimpel: This is Geology Bites with Oliver Strimpel. Ever since Alfred Wegener proposed the theory of continental drift in 1912, we've been aware that blocks of the Earth's lithosphere are moving with respect to each other. With the advent of plate tectonics in the 1960s, these moving blocks became identified with the tectonic plates that tile the Earth's surface. We now have accurate measurements of plate motion speeds, which range from about half a centimeter per year to about 10 centimeters per year. But there is still no general consensus as to what makes plates move. Broadly speaking, there are two competing explanations. In the first, the plates ride on top of convection cells of a vigorously convecting mantle. In the second, it is the forces acting on plate boundaries, principally the pull of dense lithospheric slabs subducting into a less dense mantle, that drive the plates.

Douwe van Hinsbergen is a professor of global tectonics and paleogeography at the University of Utrecht. He has reconstructed the history of plate motions in various locations around the world, with the primary goal of using this history to understand the dynamics of the mantle. And his latest research is directed to shedding light on the long-standing question as to what drives tectonic plates. Douwe van Hinsbergen, welcome to Geology Bites.

Douwe van Hinsbergen: It's a pleasure to be here.

Oliver Strimpel: Before we talk about what drives the motion of tectonic plates, I'd like to ask you about your plate motion reconstructions. Are they global in nature, like those of Dietmar Müller's group, which we discussed in an earlier podcast, or rather a regional?

Douwe van Hinsbergen: The reconstructions that I make are mostly based on mountain belts, and they are therefore regional. So we try to reconstruct the plates that we have lost because they have subducted.

Oliver Strimpel: What regions have you reconstructed?

Douwe van Hinsbergen: Almost all regions that disappeared in the last 200 million years. So that is especially the Mediterranean region, the area between Arabia and Eurasia, the Himalaya and Tibet, Southeast Asia, and the circum-Pacific regions.

Oliver Strimpel: How did you go about performing these reconstructions, and what observations did you use?

Douwe van Hinsbergen: What we try to figure out is the paleogeography of plates that existed, and the plate boundaries that existed, in areas that have fully subducted. And the records that we can use for that are mountain belts, particularly accretionary mountain belts. And with accretionary mountain belts, I'm referring to offscraped relics that formed the upper few kilometers of plates that have now subducted. You find them in mountain belts like we have in Japan or in New Zealand, but also the Himalaya or the Alps, and many of the mountain belts of the Mediterranean region are made of those offscraped relics. And in those mountain belts, in those offscraped relics, we see all sorts of rock types. The nature of the rock types, the fossils in the rock types, the ages of the rocks give us a lot of information about what the geographic features were that these rock types formed on. For instance, we can find barrier reef-type platforms, carbonate platforms, and that will probably tell you that you've been on large continental fragments, whereas if it is a volcanic atoll, like we may find around Hawaii, you will get a very different geological record. So the relics that we find in these mountain belts tell us the type of rock, the age of the rock, and the moment that that rock arrived at a trench. And that is the type of data that we use to, let's say, put them back on the now-subducted lithosphere and make reconstructions with them.

Oliver Strimpel: That's fascinating. So just by identifying the rock types in the mountain belts, which are, as you put it, offscraped relics, literally scraped off the top few kilometers of plates when they subducted, you can infer the extent of those subducted plates, and when their various parts reached the trench and subducted.

Douwe van Hinsbergen: So what we can see in a mountain belt is what is still there. And that probably is the biggest problem that we have in geology: geologists can only study what is there, particularly in these mountain belts. These are minimum records. If there was an ocean 5,000 kilometers wide of which no rock ever accreted, we will never be able to tell that that ocean was there. So as a first starting point for these reconstructions, we use reconstructions like Dietmar Müller has made. Dietmar and his group are the world's leading experts in making reconstructions of the oceans that we do have today. Magnetic stripes in the ocean floor and large faults on the ocean floor, the fracture zones, tell us what the formation history and the growth history of these ocean basins have been. If we go back to 150 million years ago, only about half of the surface of the Earth is covered by rocks that we have today, and the other half is lost.

So the first thing that we do is estimate, using reconstructions like Dietmar's, how much lithosphere was lost, not only how much ocean lithosphere was lost, but on which plate it was lost. And it tells us how fast plates were moving together and in what direction they were moving together. And then, on those plates that we, let's say, pulled back up from the mantle, we stick the bits and pieces that we find in the mountain belts, systematically, one by one, across the planet. And then we get a picture of, let's say, 90 percent white void, no information, and 10 percent little fragments, bits and pieces. And those little bits and pieces still make for quite an accurate geography, because you can make an educated guess of what is missing in between.

Oliver Strimpel: Is there a way to test these reconstructions with observations of a different nature, to make sure they're consistent?

Douwe van Hinsbergen: There are. Most directly, you can see where those plates might currently reside in the Earth's mantle. We have seismological techniques, the CAT scans of seismic tomography, that allow us to see where we have anomalously high wave speed anomalies in the mantle; that could be subducted lithosphere. So we can predict whether we find subducted lithosphere in places where our reconstruction tells us there is subducted lithosphere. And clearly, if we do that at modern subduction zones, then it works, because that's how we know that it's subducted lithosphere in the first place: we can calibrate tomography with modern subduction zones. But in mountain belts, like in Iran or in the Mediterranean region, we see the relics of subduction zones that formed and died in the geological past, in locations far away from the modern mountain belt. And for those reconstructions, we can use tomography as a test.

Another independent test, to see whether we place a deformed rock unit that we now have in a mountain belt in the right position on the Earth's lithosphere, is through paleomagnetic data. The plates that we have, we can place relative to the Earth's spin axis by using paleomagnetic data from stable continents. And then we can see whether we put the rock units that we have in mountain belts in the right position, at the right latitude, in the geological past.

Oliver Strimpel: That's really interesting. So you actually don't use the seismic tomography or the paleomagnetic data as inputs to your reconstruction; you just use them as tests once you've done it, using your examination of belts and the other logic that you described earlier.

Douwe van Hinsbergen: Exactly. What I try to do is predict, based on the structure of mountain belts and based on the history of the seafloor, what unit must have been where, with the information that we have in mountain belts on fault motions, plus simple geometry. You know, it has to fit. If you take something as complex as the Mediterranean region, there are so many constraints from so many different corners that each individually may have multiple options to interpret, but if you put them all together, there are not that many options where you can put everything. You know, if two regions both received sediments, they cannot have been on top of each other. So by using geometry, we normally end up with a reconstruction that has limited degrees of freedom, and then we predict paleomagnetism. And in some cases, like for instance in Corsica and Sardinia in the Cretaceous, in the Mediterranean region, paleomagnetism told us they had a really different position from what I had reconstructed. It was permitted by what I knew, but there was no way to tell from the get-go. So that is an example where we changed it. But if it is possible to avoid using paleomagnetism as input, then I do, also because paleomagnetism has a much larger uncertainty than we have appreciated in the past.

The mantle I specifically do not use as input, because it is precisely the evolution of the mantle and the dynamics of the mantle that I try to reconstruct. So if I force my reconstructions to fit a pattern in the mantle that I think should belong to it, then I lose my independent control on what happened in the mantle. And if I want to understand how the mantle is moving, convecting or not, and I use the mantle as input for our reconstructions, I inevitably end up in circular reasoning.

Oliver Strimpel: Let's talk about how reconstructions can shed light on what drives plate motions. So can you recap for us what the two competing theories are about what drives tectonic plates?

Douwe van Hinsbergen: The two theories are both more than 50 years old. The first one essentially originates from the famous Arthur Holmes, who was the first to realize that radioactive decay in the Earth's mantle must be a heat engine. And he envisaged, rather like a boiling pan of milk, a whirling, convecting mantle trying to lose its heat. And he supposed that, for the continental drift hypothesis of Wegener at the time, which was lacking a mechanism, the horizontal flowage, as he called it, would be a driver of that horizontal motion of blocks, or whatever they were, because they didn't really know then. When plate tectonics started to become the paradigm, Forsyth and Uyeda, in I think '75, had the perfectly logical notion that if you have ocean floor, and that ocean floor cools and becomes denser than the mantle, it will sink. And if it sinks at subduction zones, its negative buoyancy will act as a pull force on the downgoing plate. Apart from that, the aging of oceanic lithosphere also comes with a topographic effect that will give a horizontal force known as ridge push, which is thought to be subordinate to slab pull. But that suggests that it's the density structure of the moving plates, rather than the boiling of the mantle underneath, convection in the mantle, that drives them.

But there's one thing that both ideas have in common, or two things. First of all, they kind of make sense. And the second is, they were both made when we knew nothing of the mantle; we had no direct observations of the mantle below our feet. And that is, I think, the interesting point where we are now, with all the decades of seismological work, geochemical work, and the interpretation of the geological history of the mantle that came with the connection of seismic tomography to records that we see in mountain belts. We're now coming to the point that we can not only see how plates move over the surface, but also how they move through the Earth's mantle, and we start to see a history below our feet. We can start to evaluate to what extent these ideas of pulling slabs and whirling mantles are actually consistent with what we see.

Oliver Strimpel: What are the observations of the mantle that bear on the question of plate drivers?

Douwe van Hinsbergen: The first thing that comes to mind is seismological images that map out where seismic waves in the mantle are fast and where they're slow. One way of explaining fast and slow, and that is the way that is most common, is that regions that are colder than average will have a seismic velocity that is higher than average, and regions that are hotter have slower seismic waves. But seismologists have also explained that compositional differences or water differences may also have such an effect. What we see in the mantle is that there are regions, elongated, large bodies of anomalously fast stuff, whatever it might be, that, when you compare them to plate reconstructions, have volumes that go nicely with a history of lithosphere consumption. So with seismic tomography, we see a current snapshot of where the bodies in the mantle are that might be slabs. And the next step that we took is a new interpretation step, whereby you connect the interpreted geological history of subduction with those bodies in the mantle. If you correlate the shallowest bodies in the mantle with the youngest records of subduction, and you assume that older subduction records tend to correlate to deeper slabs, deeper anomalies, then you get a very consistent pattern, whereby all subduction zones of the last 200 to 250 million years are accounted for in the mantle, and there are no major anomalies left unaccounted for.

On the other hand, the geochemists have been working on the chemical composition of volcanic rocks that apparently come from different reservoirs in the mantle. There's volcanism correlated to regions of subduction, volcanism correlated to regions of extension, at mid-ocean ridges, and there's volcanism that is intraplate. And that is not just intraplate, but is intraplate in very specific regions relative to anomalous structures in the deepest mantle. And those different volcanic provinces have different signatures, suggesting that there are different reservoirs in the mantle, but also that the mantle is much more heterogeneous than we would expect from a mantle that is very quickly mixing and convecting. So that is a constraint.

Then a third constraint, that I think is very spectacular, is that there are places in the mantle where people in the last 20, 25 years have found zircons, minerals that normally form in magmatic rocks. And we find zircons that are way older than the volcanoes, or the crust, the oceanic crust, on which they formed, in places where they should not be, like on the Mid-Atlantic Ridge. These are features that are now so often observed that we start to be able to piece together a picture of the mantle directly from observations that was not available at the time that we came up with the ideas of the drivers of plate tectonics.

Oliver Strimpel: So there's the seismic tomography, which is one constraint; there's the geochemical signature, which we get by analyzing the trace element and the isotopic composition of lavas sourced from the mantle below at mid-ocean ridges, above subduction zones and at intraplate hotspots, which all point to a heterogeneity in the mantle; and there is the presence of crystals of zircon, which we know really could only have formed in the magmatic rock that then crystallized that's coming up from the mantle. So those three things together are new data that tell us about the mantle that we didn't have at the time that these two competing theories were formulated initially.

Douwe van Hinsbergen: Yeah, I'm sure there are more. I think particularly the seismologists and the mineral physicists have a few other tricks up their sleeve that I have yet to learn more about. But there are now quite a lot of observations that will constrain at least how the Earth is operating now. It doesn't mean that it was always operating that way, but I think we can get a better idea about to what extent the mantle is vigorously convecting or not.

Oliver Strimpel: So how does all this relate to the plate motion reconstructions that we discussed earlier on?

Douwe van Hinsbergen: The mantle is a geological product itself. It's the product of a geological history. So if we find zircons in the mantle in a location where they cannot have formed recently, like on the Mid-Atlantic Ridge, how would you get them there? These are zircons that are 3 billion years old, up all the way until a couple of hundred million years old. How do you get all those zircons together in the first place? Well, you do that in a sandstone. That's how you get crystals together on a geological surface. How do you get sand into a mantle? Well, you need to have a subduction zone for that. Sand doesn't sink through a mid-ocean ridge into the mantle. So the plate reconstruction should predict, for an anomaly that we see in tomography, when that anomaly could have formed and where. But we know where: tomography tells us where that thing is today. And it seems reasonable to assume, so far, that once slabs break off from lithospheric plates, they do not move a lot horizontally anymore, so they just sink vertically. So if we have a slab deep in the mantle somewhere and we want to figure out when the last opportunity has been to form that thing with subduction, we need a plate reconstruction, and we need to know that plate reconstruction relative to the Earth's mantle, which is not straightforward, but there are ways to estimate that.

So there was a paper last year by an Australian group, led by Marianne Richter, who found something quite unexpected. Near the North Pole, on the Gakkel Ridge, an active ridge today, they found a signature of subduction in the geochemistry. And they noted that in the mantle below their feet, at a depth of 1,500 kilometers or so, there is an anomaly that suggests that there has been subduction near the North Pole. And if you then take a plate reconstruction, the youngest subduction zone that was reconstructed at that location is something like 130 million years old, which suggests that that signature survived in the location where it formed, still right above that slab, for over 100 million years. And we made a similar case with rocks with a magmatic signature that we see on New Guinea, but that seem to have formed in the absence of subduction. And we also noticed that we start to find those rocks as soon as our reconstructions tell us that New Guinea started to move over a subduction-related anomaly in the mantle, over a slab in the mantle, which formed because of intraoceanic subduction 30 million years prior. What that might suggest is that those anomalies do not move much horizontally in the mantle over a very long period of time. And if that is true, then it means that the mantle is not moving that much, at least not in the last hundred and thirty million years.

Oliver Strimpel: OK, so let me see if I fully understand the logic here. So when you have a subduction that goes down into the mantle, and because of a certain amount of melting and interaction with the mantle, that will change the chemistry of the mantle to a certain extent, which we can detect later using our modern geochemical techniques. So by using your reconstructions, we can tell the history and the location of past subductions. And what we seem to be finding is that the geochemical imprint of those past subductions has survived through geological time in roughly the same locations where that imprint was originally created by the subduction itself. If I got it right.

Douwe van Hinsbergen: You explain it better than I do. But yes.

Oliver Strimpel: So by putting together the reconstructions with the modern geochemistry and tomography, we are now beginning to piece together a picture that maybe the mantle is really pretty still, maybe not stagnant, but really not moving very much. Is that the conclusion?

Douwe van Hinsbergen: Yeah, that is the conclusion that I'm slowly starting to draw.

Oliver Strimpel: And so the implications are that the mantle is not like a boiling pail of milk with vigorous convection cells. Therefore plate motions would not really be driven, because there's no convecting cell to drive them.

Douwe van Hinsbergen: Yeah. That as such may not be a shocking conclusion, because people have argued that plate motions themselves can be explained by slabs going down. But even models that have slabs driving the plates have those same slabs also driving convection, so that the mantle itself is vigorously convecting, from a geodynamic point of view, is still the paradigm. The question has only been: is that convection driving it, or is that convection passive? And what these observations seem to suggest (and I think we're only at the start of this whole thing) is that not only are the plates driving themselves, at least in the last few hundred million years, but they're not doing much to the mantle underneath. The mantle underneath appears to be a more or less sluggish, passive body of rock that melts if you decompress it or you add some water, and that will certainly move and swirl where you move your paddle through the water, where your slabs are moving through the mantle. They certainly are stirring the mantle. If a slab is retreating, which we can demonstrate happens, it moves mantle around. So it's not a solid body of rock. It can move if somebody pushes it, but on the large scale it seems not to be doing too much.

Oliver Strimpel: So in other words, the convection driven by the radioactive heat is not really driving these plate motions, as Arthur Holmes might have suggested originally when he proposed the theory, but what motions there are would be responsive to the dynamics of the plates themselves and the forces acting on the edges, whether it's push or pull.

Douwe van Hinsbergen: Yes, and those motions are local, and on the large scale minor, one or two orders of magnitude slower than the plates are moving. I don't say that I understand the dynamic consequences of that statement. I do kinematics. I try to piece out how stuff moved, not so much why it moved. Sure, I do that with colleagues, but in my research I'm either blessed or compromised by a very poor knowledge of mathematics and physics. So my limitations tell me that I can figure out when stuff moved and where it moved, but not why. So from that point of view, I came to this conclusion. And I think that, on the other hand, this can be a beautiful constraint on trying to figure out what is driving things, because whatever is driving it has to fulfill these observations, or perhaps we misinterpret these observations. If this is correct, it gives a rather unique constraint on the workings of the planet that we didn't realize before.

Oliver Strimpel: Yeah, it's fascinating. I mean, obviously this is a question that's been around for many, many decades, and now appears this kind of new line of evidence that might bear on the question of what drives plate tectonics. So if the mantle is, let's say, relatively stagnant, and the driver is the forces acting on plate boundaries, what does that tell us about individual plate motions that we see? Can it help us understand why some plates are moving fast and some plates are moving slowly?

Douwe van Hinsbergen: It must. But all our experiments, numerical experiments or analog experiments of plate motion, say: if I have a plate, and there's something buoyant on that plate, like a continent, and the continent hits the trench, then I should see a change in my plate motion. But I'm not sure whether we do. There are many places. Australia has collided twice with intraoceanic subduction zones, and slabs attached to it, and they've broken off, and nothing happened in its plate motion history; it just happily kept going. In India we see the same thing. India is still moving northward, and the last phases of slab breakoff have barely any effect on plate motion. In the Mediterranean region, there's been continental subduction for 100 million years, on and off. We don't really see it in the plate motions. So even though slab pull must be the driving factor of plate tectonics, as we think, the loss of slab does not really have the predicted result. So I'm not sure whether we understand.

And to make life even more difficult, there is a paper that just came out by Chuck DeMets and Sergei Merkouriev, who are well known for high-resolution reconstructions of plate motions. We know that India went super fast 50 or 60 million years ago, up to 150 millimeters a year or even more, and then it slowed down, and everybody has always been saying that it had something to do with collision. Well, what they showed is that actually, in the time that it went super fast, India was speeding up and slowing down by 50 to 100 percent of its speed on timescales of 2 to 3 million years, which we really cannot explain. And these are crazy high-resolution records. So, as usual, I think that the more we learn, the more we figure out that we do not understand what's going on. Again, if you look at where these concepts came from, they make perfect sense, but in the light of everything we know now, there are a couple of observations that show that they're either incorrect or incomplete, and we can speculate a little bit about what might be driving it. I think it's an exciting time.

Oliver Strimpel: Douwe van Hinsbergen, thank you very much.

Douwe van Hinsbergen: You're welcome.

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