This transcript was generated automatically and lightly edited for readability.

Oliver Strimpel: This is Geology Bites with Oliver Strimpel. In every ocean, chains of volcanic islands poke out and trail off into underwater seamounts that grow steadily older along the chain. The best-known example is Hawaii and the Emperor Seamount chain. It's widely accepted that these are manifestations of hotspots, the surface trace of hot mantle plumes anchored deep within the Earth that burn through the drifting oceanic plates so that fresh magma breaks through and heaps up onto one island after another. The trail of islands becomes a record of the plate's journey, but that rests on the assumption that hotspots stay put so that they can provide a global frame of reference. But are hotspots actually fixed? And what do we mean when we ask that question? Fixed relative to what?

Bernhard Steinberger has spent his career working on how the deep Earth moves and how hotspots move along with it. He is a geophysicist at the German Centre for Geosciences in Potsdam. Bernhard Steinberger, welcome to Geology Bites.

Bernhard Steinberger: Thanks for inviting me. It's a pleasure to talk to you.

Oliver Strimpel: I think there are two related lines of inquiry here. The first is to understand the hotspots themselves. Are they anchored in the deep mantle, perhaps at the core-mantle boundary? Do their motions through the mantle on the way up to the surface cause them to move? And if so, is that a systematic motion or a different, perhaps random motion for each hotspot? And second, can we in fact use them as a global frame of reference against which to measure plate motions? Does that kind of make sense to you?

Bernhard Steinberger: Yeah, yeah, that's exactly the kind of questions I have been addressing during much of my career.

Oliver Strimpel: OK. So we actually have no way of measuring the motion of hotspots directly. Can you describe exactly what it is that we can actually go out and measure? And then we'll talk about how those measurements are related to the quantities we are actually interested in, namely whether and how hotspots move and whether we can use them as a record for plate motions. So first, what observations bear on these questions?

Bernhard Steinberger: Yeah, there are two kinds of observations. One is the paleomagnetic observations: you measure the magnetic field recorded in the rocks by the hotspot, and they can tell you at what latitude relative to the pole the hotspot would have erupted. And then there are the age dates from along the hotspot track. So you measure at different places along the track at what time the rocks were erupted, and from that you can then measure how the hotspot has moved relative to the plate. Now if you have several hotspot tracks on one plate, or you have them on different plates and there are observations that let you determine the motion of the plates relative to each other, then you can also find out whether those hotspots have moved relative to each other. And then this will help you to devise this hotspot reference frame, or moving hotspot reference frame, which I'm sure we'll talk about later in more detail.

Oliver Strimpel: OK, let's take these one at a time, starting with the paleomagnetism. What exactly can we learn when we measure that?

Bernhard Steinberger: OK, what you measure, from the seamounts that are generated along a hotspot track, is in what direction the rocks are magnetized. It's not as easy as it sounds, because there are various overprints to this magnetization. But with some good methods you can extract the primary magnetization, and then you know the direction of the magnetic field where the seamount was originally generated. Now the direction of the magnetic field can tell you about the latitude. Basically, the further north you are, the steeper the magnetic field is. At the equator it's horizontal; at the pole it's vertical. And in this way you then determine the latitude where the hotspot formed.

Now this is the latitude relative to the magnetic pole. And the first assumption that goes into that is that the Earth has had a dipole field like it has now, so that you can really always relate the latitude to the inclination of the magnetic field. Now that has worked for the last few million years. You can determine plate reconstructions that make a lot of sense based on this assumption. So this is an assumption that's mostly accepted: that through most of at least recent Earth history, the Earth has had this magnetic field, and that you can use magnetic inclinations to determine latitudes. Another thing, which we will come back to, is that the Earth as a whole can rotate relative to the spin axis, which is called true polar wander. But I'm sure we'll come back to this as well.

But essentially at this point, we can now determine how the hotspot has moved relative to the spin axis. Because if you have different seamounts along the hotspot track and you have different inclinations for those seamounts, then you can find out that the hotspot has moved in latitude. Now it cannot tell you anything about longitudinal, east-west motion. So it can only tell you how the hotspot has moved north-south, and only relative to the magnetic pole, which is presumably the spin axis, but which might be different from the mantle reference frame which we want to determine from the hotspots.

Oliver Strimpel: So as you said, you're measuring the latitude with respect to the magnetic pole, and we know that the magnetic pole actually wanders with respect to the spin axis pole. But are you saying that basically over a few millennia that kind of all averages out and it actually is effectively aligned?

Bernhard Steinberger: Yeah, it's not over a few millennia, but maybe over a few hundred thousand years, it averages out. So actually, when you drill through those seamounts to determine the paleolatitude, it's not just that you take one piece out and measure the inclination. You have to have a drill core, and you have to have enough flows that you get enough variation that you can be confident that you can average out those secular variations.

Oliver Strimpel: OK, let's talk about the hotspot tracks and the chain of mid-ocean islands and the seamounts. What is that actually recording?

Bernhard Steinberger: So the assumption is that there are plumes upwelling from deep inside the Earth, and once they come close to the surface, due to the pressure release there is some melting, and these melts will then penetrate through the plate. And in this way you get a continuous chain of seamounts where these melts erupt. And for those seamounts you can then, by radiometric dating, determine the age at which they were erupted along the seamount chain. You get different ages, and this then will tell you where the hotspot was relative to the plate at any given time. And so this relative motion can of course be due to plate motion or due to hotspot motion or due to any combination.

Now going back to Jason Morgan, he found out that you can fit several hotspot tracks on the Pacific with one plate motion. So coming back to that time, it seems that it's mainly due to plate motion, but there may be some small amount of hotspot motion as well. And this is something which I have spent a lot of time on, trying to disentangle these two different motions.

Oliver Strimpel: So what we're getting is the relative motion between the plate and the hotspot. Now you said that's all very well if we're looking at different hotspots in the same plate, but plates move relative to each other. So we also need to measure the relative motion of the plates. How do we do that?

Bernhard Steinberger: The main method to measure relative motion is this stripe pattern on the seafloor.

Oliver Strimpel: So it's again due to the magnetic field.

Bernhard Steinberger: Yeah, because the magnetic field is not only a dipole, but it also reverses. So sometimes the North Pole is where the North Pole is now, and sometimes it's the other way round. And so the seafloor that gets generated at the mid-ocean ridges gets magnetized, and the magnetization is opposite when the magnetic field is opposite. So in this way, once the seafloor spreads away from the ridges, you get this striping pattern and can then use this to determine relative plate motions.

Now there is also another way, the older way, which is to determine it from paleomagnetic data on continents. You can bring the apparent polar wander paths for several continents into agreement. And this way, even for earlier times before there is any seafloor, you can determine the relative motion of continents. But in terms of hotspot tracks, which are mainly in the ocean, those magnetic stripe patterns are really what is important to determine relative plate motion.

Now that works well in the African hemisphere, because all the plates are linked through spreading ridges. It's a bit more complicated if you want to link plate motions in the Pacific realm to the African realm, because there are some plate boundaries which are not all connected through spreading ridges. And in this way, there's a bit more uncertainty about how plates in the Pacific realm have moved relative to the African. But in the African realm, all the relative plate motions are very well determined, and much better than the absolute plate motions from hotspot tracks, because magnetic stripe patterns are really a very good way of determining relative plate motions.

Oliver Strimpel: So we have the paleolatitudes from the paleomagnetism, and we have the relative motion of the plates with respect to the underlying hotspots from the hotspot tracks that we see, and the relative motion of plates with respect to each other from magnetic striping or other measurements. So how do these measurements fit together to address the two lines of inquiry I mentioned, namely the intrinsic hotspot motions themselves and the motion of plates relative to the mantle?

Bernhard Steinberger: Yeah. So we have those hotspot tracks, which determine the relative plate motion, and then we determine with a numerical model how the hotspot would have moved in a convecting mantle. And by bringing this computed hotspot motion into optimal agreement with the observed hotspot tracks, we can determine a so-called moving hotspot reference frame. So that's the sort of reference frame of the lower mantle, which is based on the hotspot tracks but which takes those computed hotspot motions into account. So we have this moving hotspot reference frame, which is meant to represent the lower mantle, and we compute plate motions in this reference frame.

Now there is an independent way of determining plate motions, based on paleomagnetic data from each continent. Each continent has a so-called apparent polar wander path, which represents the way that the pole has moved relative to each continent. But in reality it's the continents that have moved relative to the pole. So you can determine each plate motion also in this so-called paleomagnetic reference frame. And by comparing the two reference frames, what you get is the so-called true polar wander. So that's the motion of the spin axis in this mean mantle, or moving hotspot, reference frame which we determined before, and in this way we determine this so-called true polar wander.

Now, as I said, there are also independent data from the hotspot tracks: from the Hawaii-Emperor Seamounts, you can also determine the paleolatitudes. And then we can test whether these paleolatitudes from the Hawaii-Emperor Seamounts really agree with the motion of the spin axis in the hotspot reference frame. There's still some disagreement, and at the moment it's still an open question. But in principle, these are the lines of inquiry we are following.

Oliver Strimpel: So in order to get at the true motions of the plates with respect to the mantle, we have to be able to subtract out true polar wander, as you said.

Bernhard Steinberger: Yeah, if those motions are based on paleomagnetic data, yes.

Oliver Strimpel: And true polar wander is the entire solid Earth, the lithosphere and the mantle, reorienting itself so its axis of maximum moment of inertia stays lined up with the spin axis as things are moving around inside, such as dense slabs subducting or the lower-density plumes rising.

Bernhard Steinberger: Yeah, exactly. But it's kind of a theoretical concept, because again it depends on what reference frame you fix your true polar wander to. And we tend to use this lower mantle reference frame that takes hotspot motions into account. This has moved relative to the spin axis. Because we think that in the lower mantle the motions are slower than the plate motions, it kind of makes sense to take this mean motion of the lower mantle relative to the spin axis as the so-called true polar wander.

Oliver Strimpel: Now we can get to the questions we really wanted to answer. First of all, and I think you've already mentioned this, are hotspots fixed relative to one another?

Bernhard Steinberger: No, they're not absolutely fixed, but because they're anchored in the lower mantle, they move relative to each other more slowly in general than plate motion. So that's the reason why this idea of the fixed hotspot reference frame, and hotspot tracks as a reference frame for plate motion, which was proposed by Jason Morgan in the '70s, came up in the first place. But already back then, work by Peter Molnar and others showed that the Hawaii hotspot probably has moved relative to the other hotspots. And that's something which we can also address with numerical modeling, because the lower mantle is of course convecting. So you would never expect that things are absolutely fixed. But you can have high enough viscosity in the lower mantle to have hotspot motions that are sufficiently slow to be compatible with the observations.

Oliver Strimpel: OK. So we have the hotspots moving because of these convection patterns. Is that systematic? For example, are the Atlantic hotspots moving systematically with respect to the Pacific ones, or is there even a lot of relative motion within each group?

Bernhard Steinberger: Yeah, I would say that it is mostly relative motion within each group, because it really depends where the hotspot is situated relative to a convection cell. Now there has been this concept of relative motion between Atlantic and Pacific hotspots, but that's partly because the Pacific moves northwest. There is this large-scale convection cell, and the Pacific hotspots tend to have a component opposite to plate motion, because there is this return flow in the lower mantle. They also move relative to each other, but they tend to have this component opposite to plate motion. So there is a bit of relative motion of groups of hotspots, but this is not a concept which I would carry too far. I would say that it's really individual hotspot motions, depending on where in the mantle convection cell a hotspot is located.

Oliver Strimpel: So are those individual motions essentially random? I mean, as you might expect, a little bit of random motion in a convecting plume.

Bernhard Steinberger: They can maybe be seen as random motion, but it's really determined by how the mantle underneath moves. So basically the idea is that plumes, not all hotspots, but probably many of them, come from the deep mantle, from the core-mantle boundary. So there is a motion of the source along the core-mantle boundary, and then they move through the convecting mantle, and both the plume head and the conduit that follows the plume head get further deflected as the plumes rise through the mantle. And then the position where this deflected and tilted conduit reaches the surface over time, that's where the hotspot is. And that then determines the hotspot motion. So in principle, I think it can be computed how a hotspot moves, but you would need to know the mantle convection pattern, and there is still quite a bit of uncertainty. So maybe, if you look at them, you can model them as a bit of random motion. But in the end, it's really all determined by the flow pattern in the mantle beneath.

Oliver Strimpel: Let's come back to the most famous example of a hotspot. That's the Hawaii-Emperor chain. There's a famous bend in that chain. And given what we've talked about now, can we disentangle whether that bend was caused by a sudden change in direction of the Pacific plate motion or was it the hotspot itself that was moving?

Bernhard Steinberger: Yeah, there have been these two different lines of thought, and it hasn't really been totally resolved. And the reason is because we cannot really find a good explanation. Why would the Pacific plate motion suddenly change direction, but also, why would the hotspot suddenly change direction? So I usually tend to think that it's a combination. For the hotspot motion, you would have to have initially a southwestward hotspot motion that then stopped; that would give you the bend. There possibly was an intra-oceanic subduction zone towards the northeast, and that drove a convection cell with the lower mantle flowing towards the southwest, and the plume, before the Emperor bend, moving towards the southwest. So there is some indication in some recent numerical models that show that. So that can give you part of the story.

But then there is also probably, some time ago, before 50 million years ago, to the northwest of the Pacific plate, still another plate, the Izanagi plate. But then this was totally subducted. And then after that the Pacific plate itself got subducted westward. So there was this slab pull: the pull of the slab that's attached to the plate pulled the Pacific plate westward. But as I said, individually you would not expect either of them to give a sudden change, and I think it must be some combination: through some coincidence, both things working together give this sharp bend. And I would say it's mainly the hotspot motion that has made the chain longer, but there has also been a change in Pacific plate motion.

There are also other hotspot tracks in the South Pacific which also show a little bit of a bend at this time, but not as clearly. And that's why I think there must have been some change in plate motion, but maybe not as pronounced as the bend would indicate. And for the Hawaii plume, the hotspot motion was in addition. And this combination gives a much sharper bend for the Hawaii hotspot track, whereas for other Pacific hotspot tracks this bend is not nearly as pronounced.

Oliver Strimpel: Well, that's fascinating. So something that looks like it would probably have a single cause, namely a sharp bend at a particular time, you actually have to invoke two separate motions in combination to really understand it. So I think you said that we can model the motion of the hotspots by trying to understand what the convective patterns in the mantle are. And so even though there is probably some individual residual motion of each hotspot, if they are moving in a systematic way as you suggest, can we still use them to build a frame of reference that we can then apply to understanding the plate motions?

Bernhard Steinberger: Yeah, yeah. We can determine the motion of hotspots relative to each other, and in combination with modeling we can find the best-fitting reference frame to determine the plate motion over them.

Oliver Strimpel: So in fact, my initial question about whether or not the hotspots do provide a reference frame, the answer is if you account for the appropriate motions, yes.

Bernhard Steinberger: Yes, right. You don't get it only from measurements, because then you still only get relative motions, so you don't get anything absolute. So in order to fix it to an absolute lower mantle, I think some kind of modeling needs to be involved.

Oliver Strimpel: OK, let's pull back from the individual hotspots. Can we tell whether the entire lithosphere, all the plates together, is on average moving relative to the mantle underneath?

Bernhard Steinberger: Yeah. If you take any hotspot reference frame, both fixed and in particular moving hotspot reference frames, you find that the whole lithosphere on average is moving. So it's about 1 centimeter, with a pole in the southern Indian Ocean, and the maximum motion is about 1 centimeter per year. Now the thing is that this mean motion of the lithosphere is in just the same direction as the Pacific plate. And I would think the main reason for this mean motion is because the Pacific plate is very big and it's being pulled by a slab, and perhaps, because it's moving fast, that reduces the viscosity further underneath, so it biases the global plate motion. So in this way you get this mean motion of the lithosphere.

Oliver Strimpel: But this westward drift of about 1 centimeter a year, then, that's real. And what do we think might be causing that?

Bernhard Steinberger: Yeah. So as I already said, my preferred explanation is that it's because the Pacific plate can move relatively fast, and this then causes it when you average globally, because it's covering a third of the globe. Now there has also been this other theory that it's due to the Moon causing tidal drag. But for that to work, I think you need much lower viscosity in the asthenosphere, like 10 to the 14 or 10 to the 15 pascal-seconds. Whereas for this slab pull, it's enough if the asthenosphere, the layer underlying the plates, beneath the Pacific is down to about 10 to the 19 pascal-seconds. Now those numbers maybe don't mean much to normal people, but it's basically 5 orders of magnitude higher viscosity still permitted. Typical average asthenosphere viscosities are about 10 to the 20, and underneath the Pacific it might be lower than that still. So that's why the Pacific can move faster.

Oliver Strimpel: I'm wondering how well we have all these motions pinned down today. It seems as if the motion of the plates relative to one another is very well known. What about the tie to the mantle and an absolute frame?

Bernhard Steinberger: Yeah, this is exactly where the hotspot tracks and hotspot motions come in. So that's much less well known. It really depends on modeling the motion of plumes. And these models of motion depend on the mantle viscosity structure, and the mantle viscosity structure is still fairly poorly known. Whereas the relative motions, you can determine directly from observations. Of course, the motions of plates relative to hotspots you can also determine from observations. But how do the hotspot motions relate to the deep mantle? That's something for which you don't have direct observations.

Oliver Strimpel: Is there a way in which you and others can work to shrink these uncertainties and improve the link between the surface and the deep mantle?

Bernhard Steinberger: Yeah, we need to collect observations that our model predictions can be matched to. Models also predict what the plume conduits would look like at the present day, so if there are some observations around plumes to determine what the plume conduit looks like right now, then this can be compared to model predictions. There have been seismic arrays around hotspots like Hawaii and La Réunion, and also the Canary Islands, and now we're planning another experiment around Iceland. What you see there is much more complicated.

Improving models of mantle viscosity will also help, because mantle viscosity is what determines our flow computations and the computations of how much hotspots have moved. And I've already said that the fact that hotspots have moved slowly already tells you that the lower mantle viscosity has to be fairly high. But there is still quite some uncertainty. For example, I've just recently been on a cruise in the Pacific where we tried to measure where the material that was emplaced underneath the Emperor chain is located now. And depending on how much it has been sheared off from the Emperor chain, it will tell you how well the mantle is coupled to the plate, and hence how low the viscosity is directly underneath the plate: whether the mantle moves with the plate, if there is no low-viscosity zone, or, if there is a low-viscosity zone, whether it's sheared off. So that's a new method to determine mantle viscosity.

Oliver Strimpel: When you say that, the more we look, the more complicated the picture becomes. I'm reminded of the podcast episode I did with Gillian Foulger, who questions the very existence of deep mantle plumes. I guess you're not quite in that camp.

Bernhard Steinberger: Yeah, none of the seismic images has really given those narrow plumes which you imagine. So there are some models that seem to show fat plumes coming up, which could be because plumes entrain a lot of heavy material and they have difficulty coming up, and that's why they're fat coming up through the mantle. And also, in some of the models, they don't seem to come straight through, but they get deflected in the mid-mantle. So it could be that the plumes come up from the deep mantle, then somehow they hit some barrier, or they have too much heavy material, and then somewhere else some small plumes come up. And then, as I already said, there are those ocean-bottom seismometer experiments which show something which looks like a whole plume tree structure, which could be that either the plume comes up, then spreads out, then individual plumes come up, or it could also be that separate plumes came up and then the bases merged and the tops are still separate. So there are many different complications.

So as you say, I think the more you look, the more complicated things get. So that's why my hope is that this simple concept of plume motions, hotspot motions and the hotspot reference frame based on that still has some truth to it. Now, I don't think, like Gillian Foulger, that you can do without plumes at all. But you cannot say every plume comes from the deep mantle. And the relation between coming up from the deep mantle and the surface observations might be more complicated than those simple conduits. I mean, you have heat coming out of the core, and every geodynamic model gives you plumes. So I'm quite confident that plumes exist. But what they exactly look like, there's still a lot to find out.

Oliver Strimpel: Bernhard Steinberger, thank you very much.

Bernhard Steinberger: Thank you.

Oliver Strimpel: To see pictures and illustrations that support this podcast, go to geologybites.com, where you'll also find a subject matter index of all the episodes. There you can also give me feedback, which I welcome, as well as sign up to get my emails about new episodes.