sign in sign up
<<
Home , Hotspot (geology), Iceland hotspot, Mantle plume

P ERSPECTIVES

GEOLOGY The apparent controversy can be broken down into two questions. Is there evidence that deep plumes exist? And do all volcanoes not associated with plate bound- Deep Origin of Hotspots— aries require a deep ? The an- swers seem most likely to be “yes” and “no,” respectively. the Mantle Plume Model Several observations support a deep (that

Donald J. DePaolo and Michael Manga is, plume) origin for some hotspots. Mid-

0

9 ocean ridges are able to migrate over 1

0

0 2

0

he workings of the hot interiors of the 1 hotspots without changing the 2

0 2

rocky planets of the solar system are 2 track, which implies that the hotspot most dramatically expressed by the source is deeper than about 200 km. T 30 size and arrangement of their volcanoes. At , the high produc- Depth 20

Most volcanoes on are a result of 10 km tion rate in a small area requires an . At mid-ocean ridges, the 0 upwelling velocity of ~50 cm/year spreading of the ocean floor generates up- 300 (6), about 10 times the average veloc- ward flow of hot mantle rock beneath the ity of plates. The Hawaiian upwelling ridge. This flow generates magma as a re- must therefore be distinct from flow sult of adiabatic decompression (1). At 650 associated with plate motions. zones, plates returning to the The upwelling mantle under depths of the mantle carry down in Hawaii must also be 200 to 300 K hot- hydrous minerals. The water, when re- ter than the surrounding mantle to leased by metamorphism, causes already 1000 achieve the required large melt frac- hot rock material beneath arcs to tions at depths below the 80-km-thick melt (2). (6). Such hot rock materi- But not all volcanoes on Earth are lo- al must come from a thermal bound- cated at mid-ocean ridges or subduction ary layer. The core-mantle boundary zones. “Hotspots”—regions with particu- 1450 is the most likely source, unless there larly high rates of —are not nec- is another interface within the mantle essarily associated with plate boundaries. between compositionally distinct lay- Hawaii, the premier example, is thousands ers. The chemistry and isotopic com- of kilometers from the nearest plate bound- position of many hotspot , espe- ary yet exudes at a higher rate per unit cially the high 3He/4He ratios, indi- 1900 area than at any other place on Earth. The cate that the hotspots sample a part of Hawaiian volcanic anomaly has remained the mantle distinct from that sampled mostly stationary for tens of millions of by mid-ocean ridge (7). years and produced a 6000-km-long chain Numerical simulations of plumes re- of and . This phenome- 2350 produce many of the geophysical ob- non is not explained by plate tectonics. It servations (6), such as the rate of requires a separate mantle process that can magma production and the topogra- account for narrow, long-lived upwellings phy and gravity anomalies produced of unusually hot mantle rock. by plume material as it spreads be-

Shortly after the discovery of plate tec- 30 2800 neath the lithosphere.

1 20 tonics in the late 1960s, Morgan (3) pro- 90 Theoretical and laboratory studies 2 00 2 10 posed that hotspots represent narrow (100 10 of fluids also predict that plumes 22 km diameter) upwelling plumes that origi- 0 0 should form in the deep Earth. nate within the . Since that Mapping deep plumes. Maps of velocity anom- Because the core is much hotter than time, evidence from , fluid dy- alies (red, slow; blue, fast) under the the mantle, heat conducted from the namics, , and has obtained with finite-frequency tomography, which cor- core warms the base of the mantle, supported if not required the existence of rects for the effects of wavefront healing (16). The red forming a thermal boundary layer. As mantle plumes. For many geoscientists, the and yellow areas are interpreted as regions of anom- this layer thickens, it can become mantle plume model is as well established alously high temperature in the mantle. These images gravitationally unstable. The hot as plate tectonics. suggest that the Hawaiian mantle plume can be traced buoyant boundary layer should then Nonetheless, it is reasonable that we with seismological techniques all the way down to the rise, forming a large mushroom- should want to verify the model by direct core-mantle boundary, at a depth of about 2900 km. shaped plume “head” followed by a observation. The only way we can “see” in- narrow plume “tail” (8). The head is to the deep Earth is with . This stamp that most thought it would. thought to form large flood endeavor has so far not produced the rubber Seismological studies of the Yellowstone provinces once it reaches Earth’s surface, hotspot found no clear evidence for a lower whereas the tail provides a conduit through The authors are in the Department of Earth and mantle source (4), while evidence of a deep which hot mantle continues to flow to the Planetary Science, University of California, Berkeley, plume beneath the hotspot remains hotspot (9). CA 94720, USA. D. J. DePaolo is also in the Earth Sciences Division, Lawrence Berkeley National equivocal (5). Does the model need re- The persistence of flow through the tail Laboratory, Berkeley, CA 94720, USA. E-mail: depaolo@ thinking, or are the seismological tools still for 100 million years or more (several

eps.berkeley.edu not quite up to the task, or perhaps both? times the number of years required for CREDIT: G. UNIVERSITY NOLET/PRINCETON

920 9 MAY 2003 VOL 300 SCIENCE www.sciencemag.org P ERSPECTIVES plume heads to rise through the mantle) plumes or superswells. They require other achieved with standard techniques is too implies that the plume is much less viscous models that are not yet generally agreed poor for this task. However, preliminary re- than the surrounding mantle (10). In a upon. sults (16) suggest that image resolution can planet with plate tectonics such as Earth, The mantle plume model has implica- be improved using alternative data reduc- cooling of the mantle by the subduction of tions beyond accounting for the spatial dis- tion approaches, and that the deep roots of tectonic plates allows large variations of tribution of volcanism. If plumes come mantle plumes can already be resolved viscosity to exist at the base of the mantle, from the base of the mantle, then the erupt- with available data (see the figure). and hence between the plume and its sur- ed lavas from hotspot volcanoes may carry roundings (11). clues about the workings of the deepest References Deep mantle plumes may not be the mantle and even the core. Plumes provide a 1. D. J. McKenzie, J. Petrol. 25, 713 (1984). cause of all hotspots. Courtillot et al. (12) connection between geochemical and iso- 2. Y. Tatsumi, S. Eggins, Subduction Zone (Blackwell Science, Boston, 1995). argue that the main features predicted by topic reservoirs (inferred from studies of 3. W. J. Morgan, Geol. Soc. Am. Mem. 132,7 (1972). the plume model are clearly evident in on- lavas) and seismological structures imaged 4. E. D. Humphreys, K. G. Dueker, D. L. Schutt, R. B. Smith, ly seven hotspots—including Hawaii and within the mantle. The origin of hotspots is GSA Today 10,1 (2000). Iceland, but not Yellowstone. Criteria used therefore linked with our ability to inte- 5. R. M. Allen et al., J. Geophys. Res. 107, 2325 to recognize plumes include the presence grate geochemical and seismological ob- (2002). 6. N. M. Ribe, U. R. Christensen, Earth Planet. Sci. Lett. of a hotspot track and an associated flood servations with geodynamic models. 171, 517 (1999). basalt province, a large buoyancy flux (the Moreover, plumes potentially provide a 7. P. E. Van Keken, E. H. Hauri, C. J. Ballentine, Annu. Rev. product of the volume flux through the constraint on the heat flux from the core Earth Planet. Sci. 30, 493 (2002). plume and the density difference between (14) and hence insight into the energy 8. R. W. Griffiths, Phys. Earth Planet. Inter. 43, 261 (1986). the plume and its surroundings), a high budget for the core dynamo that generates 9. M.A. Richards, R.A. Duncan,V. Courtillot, Science 246, 3 4 He/ He ratio, and a monotonic age pro- Earth’s magnetic field (15). 103 (1989). gression in the chain of volcanoes. Many natural phenomena were deduced 10. N. H. Sleep, M. A. Richards, B. H. Hager, J. Geophys. Superswells—regions of the lower mantle correctly from indirect effects before in- Res. 93, 7672 (1988). 11. H.-C. Nataf, Tectonophysics 187, 261 (1991). beneath Africa and the Pacific that are struments were developed with sufficient 12. V. Courtillot,A. Davaille, J. Besse, J. Stock, Earth Planet. characterized by low seismic wave velocity sensitivity to verify their existence directly. Sci. Lett. 205, 295 (2003). (13)—are inferred to be broad, hot up- Direct evidence for mantle plumes will re- 13. B. Romanowicz, Y. C. Gung, Science 296, 513 (2002). wellings. Secondary, weaker plumes are quire seismic imaging at resolution suffi- 14. G. F. Davies, Geophys. J. Int. 115, 132 (1993). 15. B. A. Buffett, Geophys. Res. Lett. 29, 1566 (2002). proposed to form in the mantle from these ciently high to detect narrow conduit-like 16. R. Montelli, G. Nolet, G. Masters, F. A. Dahlen, S.-H. large superswells. Still other hotspots are structures in the lower mantle. The present Hung, EGS-AGU-EUG Joint Assembly, Nice, France, 6 not obviously associated with either deep lateral resolution of 1000 km that is to 11 April 2003, abstract EAE03-A03002.

GEOLOGY structures have been reported, the results are often not confirmed by more detailed ex- Is “Hotspot” Volcanism a periments. Second, “hotspots” are not fixed relative to one another (3). Hawaii has not remained stationary; it changed direction Consequence of Plate Tectonics? radically at the time of the bend in the Hawaiian-Emperor chain ~50 million years G. R. Foulger and J. H. Natland ago, when the did not change direction. Third, heat-flow measurements he plate tectonic model, first proposed deliver large amounts of melt to the surface. and petrological observations provide little in the mid-1960s, elegantly accounted The volcanic chains associated with many evidence of the high eruptive temperatures Tfor the distribution of most volcanism of these melt anomalies (or “hotspots”) ap- required by deep plumes (4, 5). on Earth’s surface—that which occurs at peared to all trend in the same direction and The plume hypothesis survived largely as plate boundaries. However, it did not appear to age regularly along the chains, suggest- a belief system and had to be extensively to explain areas of unusually profuse vol- ing that the “hotspots” are fixed relative to modified to account for unexpected observa- canism within plates and near mid-ocean one another. Morgan suggested that this in- tions. From Morgan’s initial estimate of ~20 ridges, for example, at Yellowstone, Hawaii, dicated that the plumes are rooted in the plumes in Earth’s mantle, the proposed num- and Iceland. In 1971, Morgan proposed a lower mantle, below the level of vigorous ber peaked at 5200 in 1999, but most lists second, independent mode of to convection associated with plate tectonics. now contain ~50 plumes. Subdivision into account for this type of volcanism: mantle Despite some early skepticism about the deep, intermediate, and shallow “hotspots” plumes (1). More recently, explanations for of the proposed plumes, the has been proposed, and the most recent esti- “anomalous” volcanism based on plate tec- hypothesis was subsequently used to account mate for the number of plumes ascending tonics have come to rival the plume model. for almost all anomalous volcanism. from the core-mantle boundary has dropped According to the plume model, columns However, the original predictions of the hy- below 10 (6) (see the figure). Most scientists of hot material ascending from great depths pothesis—including seismic-wave anomalies actively working on the subject now accept in the lower mantle, relative fixity, and high that not all “hotspots” are underlain by deep G. R. Foulger is in the Department of Geological temperatures—have not been confirmed. mantle plumes. Sciences, University of Durham, Durham DH1 3LE, UK. First, seismologists have generally not What, then, causes those melt anom- She is currently on leave at the U.S. Geological Survey, detected the vertical structures with low alies and enriched geochemical signatures Menlo Park, CA 94025, USA. E-mail: g.r.foulger@ wave speeds predicted to underlie the that do not arise from deep mantle plumes, durham.ac.uk J. H. Natland is at the Rosenstiel School of Marine and Atmospheric Science, University of “hotspots” and extend into the deep mantle, and are there any true deep mantle plumes Miami, Miami, FL 33149, USA. for example, at Yellowstone (2). Where such anywhere within Earth? The answers to

www.sciencemag.org SCIENCE VOL 300 9 MAY 2003 921