Roadmap | The review process | Home
   What the hell?
Insight into Motion of the Hawaiian Hotspot from Paleomagnetism

William W. Sager

Department of Earth and Atmospheric Sciences, University of Houston, Houston, TX 77204-5007


The situation with hotspots – and in particular the best known, the Hawaiian hotspot – reminds me of an old Saturday Night Live skit.  To quote Steve Martin: “What the hell is that!?”  Like his character, I can’t seem to decide just what it is, nor can I turn away.  For 25 years, I thought I knew exactly what it is.  Now recent research is casting the Hawaiian hotspot in a new light.  Perhaps our earlier views on this hotspot were just too simplistic.  Perhaps they were just wrong.  In a paper that is in press in the forthcoming GSA volume “Plates, Plumes and Planetary Processes” I compare the motion of the Pacific plate a) reconstructed from paleomagnetism, and b) from models of plate motion relative to the hotspots.  I find that the two reference frames fit together nicely for the past ~49 Ma, but before that they are drastically different. (Sager, in press)  Those differences may in fact be illuminating and I would like to share a few salient points here.

For many years it appeared that the common explanation for the Hawaiian hotspot – and others by implication – was just right.  The Pacific plate seemed to have about the right amount of northward motion to be explained by a steady drift of the plate over the nearly fixed hotspot, with a kink in the middle where the Hawaiian and Emperor chains join [Ed: See also The Emperor and Hawaiian Volcanic Chains and Speculations on tectonic origin of the Hawaii hotspot].  One of the first indications that there might be a problem with this explanation was from Deep Sea Drilling Project Leg 55 and a study by Masoru Kono (Kono, 1980) showing that lavas from Suiko Seamount, in the middle Emperor chain, were formed at a paleolatitude of ~27° rather than the ~20° expected from the present location of the hotspot.  Surprisingly, this difference created little in the way of controversy about the fixity of the Hawaiian hotspot.  Ocean Drilling Program Leg 197 took this observation to the next level, deriving paleolatitudes for several more Emperor seamounts and showing a progressive decrease in paleolatitude with time and a total of ~13° displacement (Tarduno et al., 2003).  This time, the paleolatitude discrepancy results were seen differently because other study results were also suggesting that the hotspot was not fixed.  Efforts to reconstruct the Hawaiian-Emperor chain geometry from a plate circuit and plate motions relative to the hotspots in other oceans also showed less northward drift of the plate during the Emperor chain period (e.g., Raymond et al., 2000).  Furthermore, models of mantle flow implied that hotspot conduits ought not to stay fixed, but should instead “sway in the mantle wind” (e.g., Steinberger & O’Connell, 2000; Steinberger, 2000).  My impression is that most marine geologists still accept the hotspot explanation for the Hawaiian-Emperor chain, but accept also that it has drifted significantly.  Perhaps this is true, but the results of my work, and comparison with others, stretches this view still further.

In my study, I compiled a paleomagnetic apparent polar wander path (APWP) for the Pacific plate in much greater detail than previously.  I compared this APWP with a synthetic APWP constructed from models of Pacific plate motion relative to the hotspots (Wessel et al., 2006).  If the hotspots are fixed and the Earth’s spin axis stays fixed relative to the mantle, the two reference frames should give comparable results.  For the past ~49 Ma, the two APWP match well and are statistically indistinguishable (Figure 1).  Thus, the hotspot model of plate motion seems to fit well over that time period.  Before that, there are major differences, however.

Figure 1. Pacific apparent polar wander path.  Red stars denote pole positions defining the most likely APWP (Sager, 2006; Beaman et al., submitted), shown by the blue, heavy dashed line.  Poles are surrounded by 95% confidence ellipses and labeled by age in Ma.  Blue star denotes Ontong Java Plateau pole, which is considered anomalous (Sager, 2006).  Green squares show poles determined from magnetic lineation skewness (73, 76, and 81 Ma poles from Petronotis & Gordon (1999), Vasas et al., (1994); 139 and 142 Ma poles from Larson & Sager (1992)).  The Late Cretaceous skewness poles are considered anomalous (Beaman et al., submitted).  Thin dashed lines show the predicted polar wander path from plate/hotspot motion models of Duncan & Clague (1985) (purple with triangles) and Wessel et al. (2006) (blue with dots).  Triangle and dot symbols show predicted pole positions at 5-Ma intervals, labeled every 10 Ma.  Red lines show offset between paleomagnetic and hotspot model predicted poles.  Inset sketch map shows interpreted phases of polar wander.  Plot is an equal area map.  Numbers are pole ages in Ma.  Figure from Sager (in press). Click here or on Figure for enlargement.


For the period 80-49 Ma, the paleomagnetic APWP shows no motion, i.e., a polar "standstill".  This result indicates that the Pacific plate had no discernible northward motion.  Given the uncertainty regions for the paleomagnetic poles, a slight amount of northward motion could have occurred, but not much more than about 3-4°.  In contrast, the Emperor chain itself implies N-S motion of the Pacific plate by ~19° over the same period.  Although the paleomagnetic data are far from being free of uncertainties, the magnitude of this difference is so great that I doubt whether one can dismiss the discrepancy as a result of poor data.  There are too many different paleolatitude estimates from different sources and the data agree.  This result is very important because it shows that the change in paleolatitude came about virtually entirely as a result of southward motion of the melting anomaly.  Most models of Pacific plate motion in the Late Cretaceous and early Cenozoic call upon a large amount of northward drift, but the paleomagnetic data argue otherwise (Figure 2).   This finding makes sense because it is thought that the northern Pacific plate did not begin subducting in the Aleutian trench until the Eocene , so the plate would not have experienced a great amount of northward slab pull (Sager, in press).

Figure 2. Paleolatitude (top) and northward drift (bottom) implied by Pacific paleomagnetic data vs. age.  Top: Filled circles show estimated paleolatitudes of Hawaiian-Emperor seamounts determined by distance from paleomagnetic poles to seamount sites (estimated from Wessel et al. (2006) model).  Open triangles show paleolatitudes from DSDP/ODP basalt drill cores (Tarduno et al., 2003). Bottom: Filled circles show northward drift implied by paleomagnetic poles.  Dash-dot line and dots show northward drift of Hawaiian-Emperor Seamounts with time (from Wessel et al., 2006).  Small dashed line and triangles denote northward drift shown by equatorial sediments (Parés & Moore, 2005).  Gray triangles show additional estimates of northward drift from equatorial sediments (Sager & Bleil, 1987).  Heavy dashed line and stars are estimates of seamount latitude from the plate circuit model of Raymond et al. (2000).  Numbers are pole ages. Figure from Sager (in press).

The formation of the Emperor chain becomes all the more interesting when new radiometric ages are considered.  Sharp & Clague (2006) show that the volcanic progression along the chain was not steady, as had been thought previously (Figure 3).  Instead, the motion implied for the southern Emperor chain (17 cm/yr) is about twice the old rate (~9 cm/yr; Clague & Dalrymple, 1989).  If these dates are accurate, they raise the following questions.  Would one expect a hotspot to undergo such drastic changes in drift rate (from 5 cm/yr to 17 cm/yr) and would one expect the drift rate to be so high (17 cm/yr)?  I am not privy to the details of mantle flow models, but my impression from articles on the subject is that reasonable mantle parameters do not give drift rates that are so fast (Steinberger et al., 2004).  

Figure 3. 40Ar/39Ar ages of Hawaiian-Emperor volcanoes vs. distance along the chain from the modern hotspot at Kilauea volcano. Age errors shown include geological uncertainties, as discussed by Sharp & Clague (2006).  Boxed values are volcanic migration rates for respective segments of the Hawaiian-Emperor chain in cm/year. Previously published ages (blue symbols) are from Duncan & Keller (2004). Figure from Sharp & Clague (2006).

The paleomagnetic results also allow a crude estimate of the motion of the hotspot relative to a “fixed” mantle reference frame.  As shown in Figure 4, the motion of the hotspot relative to the mantle, mVh, can be described as the vector sum of the motion of the plate relative to the hotspot, hVp, and the motion of the plate relative to the mantle, mVp.  In Figure 4, I have made crude estimates of these using a planar approximation.  Although the velocity estimates are rough, the actual numbers are less important than the implications.  For the period of the Emperor chain, the chain shows the relative motion of the plate relative to the hotspot (hVp).  The motion of the plate relative to the mantle is ill-constrained, but the paleomagnetic data testify that the northward motion was negligible.  Assuming the spin axis did not move significantly relative to the mantle, the motion of the Pacific plate should have had little or no northward component.  In Figure 4, I show two possibililities:

  1. plate drift relative to the mantle was equivalent to that shown by the Hawaiian chain for the Cenozoic (i.e., the Hawaiian chain shows the motion), or
  2. the motion was entirely E-W.

Either way, the melting anomaly must have had a significant westward component in its motion (about the same as the southward component).  It seems unlikely that the plate was standing still, so the westward component of plate and hotspot motion had to nearly cancel to produce the N-S trending Emperor chain.  This large westward motion for the Hawaiian hotspot is not predicted by mantle flow models (e.g., Steinberger & Antretter, 2006).

Figure 4.  Sketch of motion vectors indicating Hawaiian hotspot drift during the formation of the Emperor Seamounts.  Motion of plate relative to hotspot, hVp (red vector), given by trend of Emperor Seamounts.  Motion of plate relative to mantle (assumed fixed relative to spin axis), mVp (purple vector), is assumed to be same as at present (Hawaiian Chain).  Sum is motion of hotspot relative to the mantle, mVh (yellow vector), which has a large westward component.  Horizontal vector at bottom (magenta) shows Pacific plate motion if the plate had no northward component of velocity.  Dashed line vectors show predicted motion of hotspot relative to mantle if Pacific plate motion had no northward component.  Different dashed lines correspond to different westward velocities.  Background is a shaded relief plot of Hawaiian-Emperor Chain bathymetry.  Figure from Sager (in press).

What of true polar wander (TPW), the motion of the spin axis relative to the mantle?  Some authors, myself included, have called upon this mechanism to explain the southward motion of the Hawaiian hotspot (e.g., Gordon & Cape, 1981; Sager & Bliel, 1987).  While there is some support for this explanation in global TPW curves (see Sager, in press), I think it is unsatisfactory because under TPW, the plate and hotspot would move together.  How then can we explain the Hawaiian-Emperor bend?  The simplest explanation would be that the southward motion of the melting anomaly relative to the plate stopped.  If that southward motion includes TPW, then both the TPW and motion and the hotspot motion have to stop simultaneously. 

Although I am not yet brave enough to jump entirely off the hotspot bandwagon (it has been a good ride for 25 years and besides, if there was ever a hotspot, you would think that it would be Hawaii), I have to admit that the behavior of the hotspot seems unhotspotlike.  It apparently zoomed southwest at high speeds, sometimes going fast and sometimes slower, and screeched to a halt at the time of the Hawaiian-Emperor bend.  I suspect something is amiss.  Either we are being misled by some of the tectonic or age data, or we do not understand how the Hawaiian-Emperor chain formed.  Back to Saturday Night Live: What the hell is that?  I don’t know what the hell that thing is!


  • Clague, D. A., and Dalrymple, G. B., 1989, Tectonics, geochronology, and origin of the Hawaiian-Emperor volcanic chain, in Winterer, E. L., et al., eds., The Geology of North America, The eastern Pacific and Hawaii, Decade of North American Geology, vol. N, Geol. Soc. Amer., Boulder, CO, p. 188-217.
  • Gordon, R. G., and Cape, C., 1981, Cenozoic latitudinal shift of the Hawaiian hotspot and its implications for true polar wander: Earth and Planetary Science Letters, v. 55, p. 37-47.
  • Kono, M., 1980, Paleomagnetism of DSDP Leg 55 basalts and implications for the tectonics of the Pacific plate, in Initial Reports of the Deep Sea Drilling Project, v. 55: Washington, D.C., U.S. Government Printing Office, p. 737-752.
  • Raymond, C. A., Stock, J. M., and Cande, S.  C., 2000, Fast Paleogene motion of the Pacific hotspots, in Richards, M. A., et al., eds., The history and dynamics of global plate motions: Washington, D. C., American Geophysical Union Geophysical Mongraph 121, p. 359-375.
  • Sager, W. W., and Bleil, U., 1987, Latitudinal shift of Pacific hotspots during the Late Cretaceous and early Tertiary: Nature, v. 326, 488-490.
  • Sharp, W. D., and Clague, D. A., 2006, 50-Ma initiation of Hawaiian-Emperor bend records major change in Pacific plate motion: Science, v. 313, p. 1281-1284.
  • Steinberger, B., 2000, Plumes in a convecting mantle: Models and observations for individual hotspots: Journal of Geophysical Research, v. 105, p. 11,127-11,152.
  • Steinberger, B., and Antretter, M., 2006, Conduit diameter and buoyant rising speed of mantle plumes: Implications for the motion of hot spots and shape of mantle plumes: Geochemistry, Geophysics, and Geosystems, v. 7, p. 1-25, doi: 10.1029/2006GC001409.
  • Steinberger, B., and O’Connell, R. J., 2000, Effects of mantle flow on hotspot motion, in Richards, M. A., Gordon, R. G., and van der Hilst, R. D., eds., The history and dynamics of global plate motions: Washington, DC, American Geophysical Union, Geophysical Monograph, v. 121, p. 377-398.
  • Steinberger, B., Sutherland, R., and O’Connell, R. J., 2004, Prediction of Emperor-Hawaii seamount locations from a revised model of global plate motion and mantle flow: Nature, v. 430, p. 167-173.
  • Tarduno, J. A., Duncan, R. A., Scholl, D. W., Cottrell, R. D., Steinberger, B., Thordarson, T., Kerr, B. C., Neal, C. R., Frey, F. A., Torii, M., and Carvallo, C., 2003, The Emperor Seamounts: Southward motion of the Hawaiian hotspot plume in the Earth’s mantle: Science, v. 301, p. 1064-1069, doi: 10.1126/science.1086442.
  • Wessel, P., Harada, Y., and Kroenke, L. W., 2006, Toward a self-consistent, high-resolution absolute plate motion model for the Pacific: Geochemistry, Geophysics, Geosystems, v. 7, Q03L12, doi: 10.1029/2005GC00100.
last updated 4th June, 2007