Plate reconstructions and the question of relative hotspot fixity

Joann Stock

California Institute of Technology


We discussed the question of whether the surface traces of the “hotspots” form a single, fixed reference frame. Proponents of hotspot fixity assume that the hotspot sources are all fixed to one another in a “hotspot reference frame” and that the lithospheric plates are moving around on top of this reference frame. If this reference frame exists, then all the hotspots should be fixed to one another, and if we reconstruct the plates correctly, then the past position of all of these hotspots, relative to one another, should not vary as one goes back in time. To test hotspot fixity we need three types of information:

  1. Dated hotspot tracks on one plate, that have been used to derive a set of rotations describing the relative motion of the hotspot reference frame with respect to this plate. For example, for Africa we could use the dated tracks of the Walvis Ridge (from the Tristan da Cunha hotspot) and other hotspots in the South Atlantic Ocean to obtain a set of rotations describing the motion of the Africa plate with respect to the hotspots through time.
  2. A set of rotations describing the relative motion of the hotspots with respect to another plate. For example, for the Pacific we could use the dated tracks of the Hawaiian-Emperor chain and the Louisville Ridge.
  3. A knowledge of the past relative motions of the same two plates, derived from marine magnetic anomalies and fracture zone trends, and expressed as a set of rotations of one plate relative to another plate.

We start by noting that Mueller et al. (1993) used a set of selected hotspots in the Indian and Atlantic oceans and concluded that they were fixed to one another. They provided a set of rotations for these hotspots relative to each of the major plates in the Indo-Atlantic region. A separate set of rotations for the Pacific plate with respect to another set of selected hotspots has been derived by using the geometry and ages from the Hawaiian-Emperor seamounts and the Louisville Ridge (Lonsdale, 1988). The main new test here is that we can show, using the same hotpot traces as used by Mueller et al. (1993) and Lonsdale (1988), that even with the latest information on the plate reconstructions, there does appear to be relative motion between the Indo-Atlantic hotspots and the Pacific hotspots (Raymond et al., 2000).

To test if all of the hotspots constitute a fixed reference frame, we test the following: does the motion of Africa to Pacific as determined from the marine magnetic anomalies equal the motion of Africa to Pacific assuming fixed hotspots? If they agree, then it would appear that the African and Pacific hotspots do form a fixed reference frame. If they do not agree, then either we have an unrecognized problem with the plate reconstructions, or else the assumption of fixed hotspots is wrong.

The actual kind of test we run for this is as follows: we assume that the Hawaiian hotspot belongs to the same reference frame as the African plate hotspots. We then take the modern position of the Hawaiian hotspot and rotate it back through the following series of rotations for 10 Ma: African hotspots back to Africa, then Africa to East Antarctica (using the rotation from magnetic anomalies) then West Antarctica to Pacific (again using the magnetic anomalies). If the Hawaii hotspot is fixed to the African hotspots then the point we reconstruct for 10 Ma should lie on top of the 10 Ma age position of the Hawaiian seamounts. We do this for a series of past times, e.g. 10 Ma, 20 Ma, 33 Ma…and see if we can reproduce the geometry of the Hawaiian-Emperor seamount chain – or not.

It turns out when we do this that the Hawaiian hotspot does not follow the track of the Hawaiian-Emperor seamount chain. This was first done by Molnar & Stock (1987) but since the plate reconstructions are better known now, it was done again more recently (Raymond et al., 2000). Instead, reconstructed positions of the Hawaiian hotspot lie at the wrong place (too close to Hawaii) even for the early Miocene (20 Ma) and they lie south of the Hawaiian seamount chain, not on it. They never take enough of a bend prior to 43 Ma to reproduce the geometry of the Emperor part of the track at all.

So what is wrong with our assumptions? I don’t think the relative plate motions are the problem, because Africa-Antarctica is really well known now, and Pacific-Antarctica is also well known. There was a plate boundary within Antarctica, but we have very good data now to constrain the time and amount of motion across this boundary as far back as 43 Ma. The motion is certainly not enough to reconcile the plate circuit reconstructions with the fixed-hotspot assumption. And the motion stopped at about 28 Ma so for younger times it does not affect the plate circuit calculations at all. We can also actually get from Africa to Pacific a different way back to 42 Ma – by going through the plate circuit Africa to East Antarctica to Australia to Pacific. This latter relative motion, of Australia to Pacific, is obtained directly through the South Tasman Ocean Crust and the Emerald basin (reconstructions of Keller, 2003), completely bypassing any boundary between E and W Antarctica. This latter circuit shows the same result, that the Hawaiian hotspot trace cannot be fixed with respect to the Indo-Atlantic hotspots. The agreement between these two independent methods of calculation of Australia to Pacific motion suggests that we do understand well the limits of motion within Antarctica since 43 Ma, and they are not enough to reconcile the issue of hotspot fixity. So a more likely explanation is that the Pacific hotspots and the Indo-Atlantic hotspots do not form a single, consistent reference frame at all.

What could change these conclusions? If the ages we are using for the Hawaiian and Emperor seamounts are wrong, that could change the story for the young part of the chain. It would definitely change the locations that we expect the Hawaiian hotspot to reconstruct to, when we do the reconstructions treating it as part of the African hotspot reference frame. However, we still have the problem that for the pre-20 Ma part of the chain, the reconstructed points don’t even fall on any of the Hawaiian or Emperor seamounts themselves – they fall consistently too far south of the Hawaiian seamounts and too far west (and south) of the corresponding aged Emperor seamounts.

If the African hotspot tracks get re-dated and revised so that the rotations of Africa to the hotspots change, that could change the story too. But these seem dated quite well, and it is hard to show that they could plausibly change enough to affect the answer.

One other consideration is that we don’t still have the best data for the pre-43 Ma motion between East and West Antarctica. Geological data suggests it has to be small, but small motions on one part of the globe can be equivalent to much larger motions elsewhere on the globe, so further investigations of the plausible limits of this geological deformation within Antarctica would be useful.

Another type of data constraining hotspot relative motion, which was not discussed in our talk but which is important to this issue, is that of paleo-latitude of the hotspot source through time. This is constrained by paleomagnetics, because the inclination of the remnant magnetization can be related to paleo-latitude once the effect of secular variation is averaged out. Recent results from drilling the Emperor seamount chain confirm that there has been variation of paleomagnetic latitude of the Hawaiian-Emperor chain that cannot be easily reconciled with the assumption that the Hawaiian hotspot is a fixed source with respect to the Indo-Atlantic hotspots.


  • Muller, R. D., J-Y. Royer and L. A Lawver, 1993, Revised plate motion relative to the hotspots from combined Atlantic and Indian Ocean hotspot tracks, Geology, 21, 275-278.
  • Keller, W., 2003, Cenozoic plate tectonic reconstructions and plate boundary processes in the southwest Pacific, Ph.D. thesis in revision, California Institute of Technology, Pasadena, CA.
  • Lonsdale, P., 1988, Geography and history of the Louisville hotspot chain in the Southwest Pacific, J. Geophys. Res., 93, 3078-3104.
  • Molnar, P., and J. Stock, 1987, Relative motions of hotspots in the Pacific, Atlantic, and Indian Oceans since Late Cretaceous time, Nature, 327, 587-591.
  • Raymond, C. A., J. M. Stock, and S. C. Cande, 2000, Fast Paleogene motion of the Pacific hotspots from revised global plate circuit constraints, in M. A. Richards, R. G. Gordon, and R. D. van der Hilst, eds., The History and Dynamics of Global Plate Motions, AGU Geophysical Monograph 121, 359-375.