Beyond the Plume Myth

 Compiled by John Michael Fischer, 2003

For something that has never been seen, the mantle plume has become a fact in the minds of many.   The public rarely gets to hear from those who oppose the notion.  Here you can see what researchers Don Anderson, H.C. Sheth, Alan Smith, Charles Lewis, and Scott King have to say as quoted from professional journals.  They also describe better ways to interpret the facts. Examination of the plume issue can rapidly descend into the weeds of esoteric minutia in the various sub-fields of geology and geodynamics.  This discussion, instead, will be kept at as general a level as possible.  Those wishing to delve into the details should read the full texts of the references cited and related articles.

Some terms

Cratons: Areas of continents that have not been subject to fold deformation for great lengths of geologic time.  Rifting can fragment a craton into several cratons.

Continental lithosphere is the part of the continental crust and upper mantle that can support long term geologic loads (II p. 127).  “The lithosphere under ancient cratons is cold and thick.  It may be of the order of 150-200 km in thickness, as measured both by seismic techniques and by flexure” (II p. 137).  “One expects little strength, even under cratons, below some 150-200 km.  There is no direct evidence for any load bearing capacity below this depth” (II p. 142).

Seismic tomography: analyses the velocity of waves from earthquakes to determine the viscosity of the inner earth between the source and measuring devices at many points.

In seismic tomographic models, low velocity anomalies (LVA) are considered to be hot, more fluid regions, while high velocity anomalies (HVA) are thought to be cold, more solid regions.  “HVA extend to about 250 km under Archean cratons with only some cratons showing HVA to ~400 km” (II p. 141).   Deep HVA could be transient phenomena such as downwellings generated by lateral changes in thickness and temperature, or by the motion of the continent relative to surrounding mantle (II p. 142).  Until recently, tomographic models have assumed that the mantle is compositionally homogenous.  Now, supercomputer modeling suggests "that compositional heterogeneity is ubiquitous, and is particularly strong beneath 2000-km depth"  (emphasis added).  "Low seismic velocities are often attributed to elevated temperatures.  But the new results suggest that such direct scaling is unwarranted because shear speed and temperature are poorly correlated.  In addition, at the high pressures of the deep mantle there is little thermal expansion, and so the chemical contribution to buoyancy outweighs the thermal effect" (XI p. 818).

“Depleted: deficient in the incompatible or mobile trace elements that are extracted in melts or fluids; low in 87Sr/86Sr, 144Nd/143Nd, 206Pb/204Pb, and so on.” Enriched is the opposite of depleted. (II p. 127)

“Mid-ocean ridge basalts (MORB): most abundant and most depleted magma type. Generally attributed to passive spreading” (II p. 127).   “The asthenosphere is thought to be depleted and the source of MORB” (II p. 113).

Large igneous province (LIP): areas of massive extrusion of magma.  On continents they are Continental Flood Basalts (CFB), and elsewhere are Oceanic Plateau Provinces.

Oceanic Island Basalts (OIB) are from ocean islands and, in plume theory, are thought to be due to hotspots. “CFB and OIB share many chemical characteristics.  Differences may be due to processes and depth, rather than source” (II p. 135).

“Technically, the `asthenosphere' is a weak layer in the upper mantle but it is now assumed by geochemists and petrologists to be the depleted reservoir.  Furthermore, the perceived attributes of the 'depleted asthenosphere' have been transferred to the whole upper mantle.  Thus, the words 'asthenosphere', 'upper mantle', 'convecting mantle' and 'depleted mantle' are all used interchangeably in the current geochemical literature” (III p. 100).

“Plumes: hypothetical entities considered to be strong, active upwellings in contrast to passive upwellings caused by plate divergence.  Plumes are assumed to provide magma to hotspots such as ocean islands.  Plume heads are assumed by some authors to be responsible for surface uplift, breaking of the lithosphere and LIPs” (II p. 128).

Plume theory

“Acceptance of the theory of plate tectonics was accompanied by the rise of the mantle plume/hotspot concept which has come to dominate geodynamics from its use both as an explanation for the origin of intraplate volcanism and as a reference frame for plate motions.  However, even with a large degree of flexibility permitted in plume composition, temperature, size, and depth of origin, adoption of any limited number of hotspots means the plume model cannot account for all occurrences of the type of volcanism it was devised to explain.  While scientific protocol would normally demand that an alternative explanation be sought, there have been few challenges to “plume theory” on account of a series of intricate controls set up by the plume model which makes plumes seem to be an essential feature of the Earth.  The hotspot frame acts not only as a reference but also controls plate tectonics.  Accommodating plumes relegates mantle convection to a weak, sluggish effect such that basal drag appears as a minor, resisting force, with plates having to move themselves by boundary forces and continents having to be rifted by plumes” (IX p. 135).

“The plume hypothesis has constantly evolved.  In 1963 Wilson thought intraplate volcanoes were fed from the interiors of convection cells. In 1971 Morgan envisaged plumes as part of the general convection system of the Earth” (VII p. 3).  “In this theory, plumes drive the plates and are, essentially, the main source of buoyancy and mass flow in the mantle” (III p. 100). “The modern thinking about plumes is substantially different; plumes are considered to reflect a secondary mode of convection unrelated to (and little affected by) plate-scale convection” (VII p. 3).  Flood basalts, large igneous provinces and all intraplate magmatism are almost unanimously ascribed to features called mantle plumes in current geodynamics literature (VII p. 1).

“The idealized plume has two components: a plume head, supposedly responsible for very short-lived, massive igneous events, and a narrow plume tail which generates long-lived hotspot tracks.  The source of mantle plumes is a thermal boundary layer deep in the mantle, perhaps the core-mantle boundary, although some have argued for a shallower source” (IV p. 269).

A mantle “plume is 200-300 K hotter than normal mantle, so it is buoyant relative to and less viscous than its surroundings” (VIII p. 440).  It is supposed to rise from “just above the core-mantle boundary due to thermal and gravitational instabilities, and as a plume moves upwards it entrains material from the surrounding lower mantle, forming a large, bulbous, blob-like 'head'.  With continuing ascent and entrainment the head becomes bigger and bigger, and is connected to the source region through a narrow, axial, tube-like conduit, the 'tail', along which primitive plume-source material rises, with little entrainment of the surrounding mantle, at least in principle” (VII p. 3).  “As the giant sphere enters the upper mantle it no longer entrains ambient mantle but simply pushes it aside.  Magmatism resulting from plume heads is therefore a combination of lowermost and lower mantle material but not upper mantle or asthenospheric material” (III p. 112).

“Most explanations of the distribution of seamounts and oceanic islands have invoked hotspot theory, where a chain of volcanoes is formed as the oceanic plate moves relative to a site of active volcanism caused by an upwelling plume from the mantle.  The geometry is similar to a series of burns caused by moving your hand slowly over a candle flame, and the expected result is a regular progression in the age of seamounts and islands at increasing distance from the active hotspot” (VIII p. 439).  “The small scale and perceived fixity of hotspots motivated the hypothesis of deep, narrow plumes originating below moving plates” (III p. 100).  An adjustment to this concept is that the plume head spreads “along the base of the lithosphere where it may further melt causing volcanism away from the main hotspot.   That is, the candle analogy should not be carried too far” (VIII p. 440).

“There are three main hypotheses that have been proposed to explain the relationship between mantle plumes and flood basalts”.  The first is the Campbell-Griffith hypothesis.  The “plume head, which is assumed to have a diameter of 1000 km, rises to form, beneath the lithosphere, an oblate circular disk, with a diameter of 2000-2500 km.  This leads to an uplift of the overlying lithosphere of 0.5-1.0 km, and the development of volcanic activity.  Plume head melting occurs as the consequence of adiabatic decompression when the top of the plume reaches the top of the upper mantle.  The final diameter of the plume head is considered to be approximately 2000 km.  The melt at the surface for a relatively small CFB may be as little as 0.2% of the original volume of the hypothetical plume head,” up to “12% for a large oceanic plateau” (VI p. 116).

“The starting plume assumed by White and McKenzie is much smaller than that taken by Campbell and Griffith and also it is assumed to have been initiated at a shallower depth.  In addition, these authors assume that the plume head has a diameter of 200 km, so that it has a volume of less than 1%” of the other model (VI p. 117).

“An alternative approach is to invoke super-plumes or wet-plumes which are initiated in a 'wet' transition zone between the upper and lower mantle.  Providing the wet mantle material contains about 0.3-0.4% water, the dimensions and rates of movement of mantle material could be increased to a degree that would provide the required volumes of melt at the base of the lithosphere.   However, at this time, the assumed water content in the mantle is somewhat conjectural” (VI p. 118).

“The plume hypothesis, to a large extent, is based on the hypothesis that the 'normal' state of the mantle is isothermal, cold, static, dry, and homogeneous, and that small-scale convection effects are not important” (I p. 3623).  “The system is heated from below” and “has a few hot plumes to remove the heat from the bottom of the system to the top” (III p. 112).  “If the `upper mantle' is defined as depleted, homogeneous and the MORB-source then, of course, it cannot provide enriched magmas (OIB, CFB) or be the source of plumes” (III p. 100).

“Common amendments to plume theories involve large plume heads, easy capture of plumes by ridges, long distance lateral plume flow through shallow channels, large geochemical radius of influence around plumes and tilting of plume conduits by mantle flow” (III p. 110).  “In deep mantle plume hypotheses, hotspots appear at random times and places since the source layer is remote and decoupled from plate tectonics” (III p. 114).

“Ascent velocities estimated for plumes have ranged from 10-or-less to 60 to 200 cm per year.  The lower estimates can be regarded as the more reasonable since to avoid plumes 'burning' through the plate and generating large melt fractions, the ascent velocities of the plume material must be of the same magnitude or less than plate velocities.  A delicate balance then has to be achieved between plume, plate, and mantle velocities, because if the latter approach or exceed plume ascent rates, the result would be significant motion between hotspots or the mixing of plumes into the convecting mantle.  Accommodating the plume model therefore constrains mantle convection to be a sluggish effect with velocities around 1 to 5 cm per year.  To move plates by [mantle] drag alone would require mantle flow rates of 10 to 20 cm per year, so if the mantle is required to be nearly static, drag cannot be an active driving force.  Plates must then move themselves by boundary forces.  In conventional plate tectonic models, it is the motion of the plates which induces flow in the asthenosphere” (IX pp. 142-3).

Plate rifting is “attributed to deviatoric stresses arising from the interaction of plume heads with the continental lithosphere, though as the distribution of stresses in a plume should be radial, a multitude of plumes, all fortuitously lined up along the axis of rifting, is required.  The plumes' arrival is generally regarded as the impetus for continental breakup and associated volcanism” (IX p. 144).

“The complex picture of the Earth that emerges in the plume model is compounded by uncertainties in the number of hotspots, the depth of origin of plumes, whether hotspots are fixed, the composition of plumes, the amount of melting in plumes, and the relationship of plumes to mantle convection” (IX pp. 136-7).

What the evidence shows

H.C. Sheth is a specialist in the Deccan flood basalt provinces of India, and as recently as 1997 supported the plume theory.  By 1998 he had made a complete reversal.  “Superficially the mantle plume explanation seems attractive and has had a tremendous appeal.  However, its numerous built-in fallacies, contradictions and failings are unfortunately little discussed in much of the current literature, and it has acquired the status of an unchallengeable dogma and an obvious fact” (VII p. 2).   “Few predictions and requirements of the mantle plume model seem to be fulfilled in the actual geology” (VII p. 20).  “The popular and widespread notion that hotspot tracks are simply the products of one or more plumes beneath moving plates is actually far from reality” (VII p. 22).  “Most LIPs of the world (if not all), despite widespread and popular conceptions, seem inconsistent with plumes.  All the evidence that has been used so far to support the plume model - geochemical, petrological, thermal, tomographic - is equivocal at best, if indeed not contrary.  The plume idea is ad hoc, artificial, unnecessary, inadequate, and in some cases even self-defeating, and should be abandoned” (VII p. 23).   “The plume hypothesis seems to have led Earth scientists along a blind alley” (VII p. 16).  What do he and other researchers know that could lead to this conclusion?

“The complete plume hypothesis is untestable.  The narrow plume tails, 10-200 km in diameter, and extending deep into the mantle are below the resolution of geophysical techniques and cannot be resolved by numerical or theoretical computation.  They give no signal and have no measurable effect.  The geophysical effects (bathymetry, geoid, tomography) of the large flattened plume heads are little different from alternative models of upper mantle structure.  Large scale hot regions in the upper mantle can be generated by a variety of mechanisms” (III p. 114).

“Although no geophysical or computational technique can resolve the narrow plume tails conjectured to link plume heads with the deep mantle, there are other predictions of the theories that can be tested with available tomographic resolution” (III p. 115) and “new geochemical and geophysical tools such as the Sm-Nd, Lu-Hf, Re-Os isotopic systems” (IX p. 136).

What does tomography show at fairly shallow, 230 km depth?  “In the plume models we expect to see large low velocity circles surrounding all hotspots, or their conjectured initiation sites (CFB), or elliptical patches elongated in the plate direction.  We expect to see shallow linear bands of low-velocity material under ridges, updrafts caused by spreading.  However, we see broad areas of both low and high seismic velocity.  It should be pointed out that some of the slowest regions at depths greater than 200 km are not associated with hotspots at all.  They occur along some continental edges and near some subduction zones” (III p. 115).

“At sublithospheric depths there are very large LVA in the Pacific and Indian oceans and in the North and South Atlantics.  The large continental and oceanic flood basalt provinces seem to have formed over these large, presumably hot, regions” (III p. 99).  “Yet there is no evidence of lithosphere damage or the presence of a massive plume head which has been conjectured to be dragged away from the scene by subsequent plate motions.  The hotspots associated with all of these events (except possibly for Siberia, which has left no track) are embedded in slower than average mantle and these presumably hot regions can be traced down to 200 to 300 km.  At these depths, however, the hotspots are not spots but are simply part of an extensive, meandering low-velocity region of the upper mantle.  These LVA are often parallel to ridges and coastlines rather than to spreading directions” (III p. 118).  “There is as yet no evidence from seismology, or any other source, that plumes originate deeper than about 300 km or that the broad low-velocity regions in the upper mantle are connected by narrow tails extending deep into the lower mantle” (III p. 100).   “The large plume head, even after it flattens into a pancake shape beneath the lithosphere, should be readily resolved by seismic tomography but, to date, there is no evidence for either plume heads or damaged lithosphere beneath CFB provinces” (IV p. 270).

It is not hard to find cases of hotspot-type eruptions without hot regions at plume source depth, while other regions have low seismic velocities but no hotspots (IX p. 152).   An important example is Hawaii, “which should have the most readily resolvable conduit as it is situated away from ridge systems and is "supposedly the strongest plume.”  Investigators “searched for low-velocity anomalies in the lower mantle beneath the hotspot, but found no low-velocity anomaly which correlated with the surface expression of volcanism” (IX pp. 152-3).   “There are numerous examples of LIPs that have no associated hotspot track (Keweenawan, Karoo, Siberian Traps, Caribbean, Pacific plateaus, Ethiopia, continental margin basalt sequences) and numerous hotspot tracks that have no associated LIP (Hawaii-Emperor, Line Islands, Cameroon Islands).  The LIP on the North American Atlantic margin does not appear to have a hotspot origin.  It is a rather remarkable fact that few of the numerous hotspot tracks in the Pacific can be related to the large plateaus (presumed plume-head products), either geometrically or geochemically, or to any other LIP” (IV p. 270).

“Unfortunately for the hotspot model, measurements along the axis of the Hawaiian swell suggested an increase in heat flow with distance away from the supposed site of the plume.  The swell may not be dynamically supported, but merely represents a thick section of basalt” or of “thickened peridotite predating Hawaiian volcanism.  The size of the swell does not decline along the Hawaiian chain, and there is no corresponding swell associated with the Emperor chain.  While Hawaii is thought be the strongest currently active plume, its effects are small compared to those expected during the mid-Cretaceous superplume event invoked for the formation of many of the intraplate edifices on the Pacific seafloor.  Uplift associated with a Cretaceous event has been observed, but fails to reach the amounts predicted in the plume model.  The largest CFB province, the Siberian Traps, would be expected to be associated with the greatest lithospheric uplift in continental regions.  But a detailed analysis of the geological record has shown that emplacement of the basalts was not accompanied by any more uplift than could be accounted for by local asthenospheric convection. If predictions of the plume model with regard to supposedly the strongest plume, the most recent superplume event, and the largest continental igneous province all fail, little confidence can be put in explanations of the model for less pronounced intraplate features” (IX p. 140).

In a detailed study, a group of researchers “did not find the expected age progression in the area, a 3 by 3 degree region in the Cook-Austral chain near the Macdonald seamount.  The information conflicts with the standard hotspot theory” (VIII pp. 439-440).   “Radiometric dates from many island samples are younger or older than would be predicted if a single plume currently located at volcanically active Macdonald seamount was responsible for all of the volcanoes.  These southern Austral volcanoes are actually composed of three distinct volcanic chains with a range of ages spanning 34 million years and with inconsistent age progressions” (V p. 479).  There is other evidence.  “Deeply submerged, flat-topped seamounts are interspersed with young volcanoes, suggesting erosion at wave base during a much earlier phase of volcanic activity when the sea floor was shallower.  Furthermore, individual islands within the long chain display totally different isotopic signatures, thought to be characteristic of different mantle source regions, making the hypothesis of a single plume even more unlikely.  Modifications to plume theory that might explain the Cook-Austral chain include proposals such as fortuitous alignment of several distinct plumes or volcanic rejuvenation that repaves older features with a veneer of young rock long after it has passed over the plume.  Our data suggest that such minor adjustments to plume theory are inadequate” (V p. 482).  “Gravity anomalies and seafloor fabric suggest that the volume and location of volcanism in this region is controlled by stress in the lithosphere rather than the locus of narrow plumes rising from the deep Earth” (V p. 479).

Radiometric dating of ocean island basalts has shown that many do not progress in a linear sequence of older to younger.  “To fit the local expression of volcanism in many regions requires the addition of multitudes of plumes (7 for the Line Islands, 3 for the Cook-Austral Islands, 2 for the Kodiak-Bowie chain and so on), all fortuitously lined up along the axis of volcanism.  Other plume models have resorted to a variety of ad hoc plumbing arrangements for melt to reach the surface including swaying or deflection of plumes, channeling of plume material by lithospheric structure, multiple conduits from a single plume head, or have devised special shapes such as tabular plumes.  A further option is plume splitting, which has been invoked for multiple episodes of volcanism on oceanic plateaus.  Multiple episodes of volcanism are also found in continental provinces, but the events are often separated in time by hundreds of millions of years.  Plumes would either have to hit the same spot repeatedly under continents, or be channeled to the appropriate position.  Even with such variation, the plume model still does not fit all examples of intraplate volcanism” (IX p. 146).

“Giant flood basalt fields, such as the Deccan Traps in India, have volumes of 100,000 to 10,000,000 cubic kilometers, often erupted in short time periods of 1-4 million years” (VII P. 2).   “The Deccan traps formed at 65 million years ago, presumably over the Reunion hotspot.  The responsible plume head would underlie most of West and South India.  We therefore expect to see a large area of thin lithosphere and very slow seismic velocities under most of India.  The tomography, however, shows fast velocities under the Indian subcontinent and indicates thick and cold lithosphere, and cold asthenosphere (say 110-200 km depth).  Heat flow and flexural studies are consistent with a thick, cold lithosphere today and during emplacement of the Deccan traps” (III p. 115).  Seismic tomographic evidence rules out a plume head beneath India (or beneath any other flood basalt province)” (VII pp. 11-12).

The Reunion 'hotspot' is thought to be the manifestation of an underlying Reunion 'plume', producing the Laccadive-Reunion track and Deccan volcanism.  However, “the data indicate that the Deccan eruptions may have had no genetic connection with the Reunion hotspot; nor may be the Reunion hotspot underlain by a deep mantle plume.  The 'tail' part of the hotspot track may be a southerly propagating fracture tapping a shallow, enriched mantle of the Indian Ocean, or a huge shear zone, and reflects the stress field of the Indian Oceanic lithosphere, not its displacement history” (VII pp. 21-2).  “The grand Deccan volcanic episode is not the product of any hypothetical mantle plume but the result of protracted processes of lithospheric rifting and early low-volume alkaline magmatism, ending in a catastrophic culmination due to lithospheric splitting” (VII p. 23).

Regarding geochemistry, “every major radiogenic isotope endmember (BE, DM, HIMU, EM1, EM2) has now been equated with a plume.  Such diversity has been embraced rather than questioned.  But even with the flexibility of any lithosphere or mantle component being allowed to characterize a plume-source, the model still fails to account for evidence for hydrous minerals in the source of most OIB” (IX p. 147).

“All large continental igneous provinces and most high-temperature magmas (picrites, komatiites) are found on the margins of cratonic lithosphere.  The standard plume model of flood basalt formation offers no explanation for this observation” (IV p. 269).  “The interaction of the hot, buoyant plume head with the lithosphere should also generate uplift and stretching prior to, or concurrent with volcanism.  In many large igneous provinces, the evidence for uplift, heating, or stretching is lacking.  In fact, some CFBs occur on top of deep sedimentary sequences and in depressions. Cratonic lithosphere adjacent to CFB provinces shows no evidence for heating or thinning” (IV p. 270).

“It bears emphasizing that there is no chemical or physical characteristic of hotspots that demands a lower mantle or core-mantle boundary origin, or even an origin in a thermal boundary layer.  We still have no convincing evidence that hotspots are the result of active upwellings or that they are due to lower thermal boundary instabilities.  If hotspots are the tops of active upwellings, they could represent hot upper mantle convection cells or an overturning upper mantle convection cell.  They do not need to be the tops of deep mantle thermals” (III p. 120).

“As far as the actual surface geology is concerned, several simplistic schemes and all superficial textbooks show pictures of hotspot tracks due to a plume beneath a moving plate, but never do they show the numerous extensive fracture zones that traverse the entire plate in question.”  Recent bathymetric maps “show an extensive, pervasive, fracture zone fabric in the ocean basins.  Many ocean island chains - with or without any systematic age progression - are along fracture zones, and flood basalt provinces are at the orthogonal intersections of the fracture zones.   Most, if not all, CFBs likewise show a definite lithospheric control on their locations and are associated with/erupted from major lithospheric breaks or rifts at craton-noncraton boundaries.  Most CFBs are built on long-subsiding sedimentary basins, indicating crustal extension” (VII p. 3).  “Hotspot tracks may not reflect plate motion above stationary deep mantle upwellings but the stress state of the lithosphere, as originally proposed decades ago.  The locations of large igneous provinces, hotspots and hotspot tracks are strongly controlled by lithospheric architecture and history: they are not placed randomly on the globe, indifferent to the surface geology; neither are large igneous provinces sudden chance events within the vastness of geological time (which plume theories typically allege)” (VII p. 1).

“The initiation and evolution of hotspot features appear, on average, not to be midplate phenomena but strongly favor plate boundaries, lithospheric discontinuities and pre-existing weak areas.  In fact, the correlation of LIPs to cratons and lithospheric discontinuities is much stronger than the correlation of CFBs and hotspot tracks” (IV p. 271).  “Flood basalts are found on the edges of thick, high-velocity lithosphere that correlates with Archean cratons (Archons) and, when backtracked to their eruption sites, occur over large low-velocity regions of the asthenosphere.  In general, these large mantle domains have been insulated by a supercontinent or have not been cooled by subduction since the breakup of Pangea” (IV p. 270).  Anderson et al. suggest that the continents around Africa “have been moving from hot mantle toward cold mantle.  If so, most of the continents have reached the end of their journey since, to proceed further, they must intrude into hot Pacific mantle.  North America, in particular, seems to be boxed in by hot mantle” (III p. 111).

“By assuming that plumes are responsible for continental rifting, a temperature of 200 to 300 degrees C greater than the surrounding asthenosphere can be estimated for the core of a plume.”  These are equivalent to temperatures estimated to produce high-temperature melts such as komatiites.  “To account for the virtual absence of komatiite compared to the number of supposed hotspots,” rapid plate movement is required.  "Otherwise, using a candle-stick analogy, the plume would 'burn' through the plate and generate” lots of melt products.  While that “may work for fast moving plates such as the Pacific plate, it does not explain the melt products of plumes impacting on slow moving plates. At the other extreme are eclogite-rich plumes with temperatures only 100 degrees C higher than the surrounding mantle.  But as temperature decreases, viscosity increases, slowing the ascent rate of the plume, such that it is unclear how such plumes could avoid entrainment by mantle flow and ever reach the shallow mantle” (IX p. 140).

“A variety of geophysical data indicates that temperature variations of the asthenosphere depart from the mean by + 200 degrees C, not the + 20 degrees C adopted by plume theoreticians.  The 'normal' variation, caused by plate tectonic processes (subduction, cooling, continental insulation, small-scale convection) encompasses the temperature excesses that have been attributed to hot jets and thermal plumes.  There is very little support for the cold isothermal asthenosphere hypothesis, a corollary of the plume hypothesis.  If normal mantle temperatures are 1400 + 200 degrees C, there is no thermal requirement for plumes.  Small scale convection [gives] rise to concentrated, shallow, plume-like upwellings.  Cracks and dikes may control the dimensions of volcanic features.  Thus, there is also no geometric requirement for the deep mantle plume hypothesis” (I p. 3623).

“The narrow, fixed, tubular conduits of plume theories require special circumstances to form and be maintained, even in the laboratory.  Such artificial laboratory conditions are, for example, a non-convecting tank, no phase changes, constant material properties, no plate motions, no internal heating of the plume (all heating of the plume fluid is external), artificial injection of superheated fluid at the base of another fluid, and so on” (VII p. 3).

“The proofs of the plume model given in the tomographic studies amount to choosing the most suitable anomaly to fit the preconceived idea.  When data is evaluated only in terms of the plume model, the number of hotspots can be arbitrarily adjusted, proofs can be made from fragmentary evidence, and lack of suitable correlations cannot be interpreted against the plume model, none of the tomographic studies can be considered to have been a valid test of whether plumes exist or not” (IX p. 153).  “What should not be forgotten is that when a concept lacks direct observation, it is inevitable that a certain amount of circularity in argument based on the assumption that the phenomenon exists will be used to define its features.  Such logic does not constitute a scientific proof.  With regard to what is now known as 'plume theory', it should be noted that constraints put on temperatures, depth of origin, size, composition and degree of melting of plumes have all been made under the assumption that plumes exist, and consequently do not constitute a test of, nor confirm the validity of, the plume model.  That the plume model grew to the complexity it currently exhibits with an almost total lack of criticism should make it remarkable among scientific endeavours of this century.  Every option, except the possibility that it was fundamentally wrong all along, seems to have been explored” (IX p. 173).

Replacing plume theory

“There are essentially two models which have dealt with a shallow origin for the sources of intraplate volcanism on a global scale.  [One is the] concept of an enriched 'perisphere' layer residing between the lithosphere and MORB-source.  The perisphere includes the upper part of the asthenosphere and the thermal boundary layer [at the base] of the continental mantle.  The layer undergoes continuous enrichment from subduction processes, but is essentially static and hence encounters difficulty in generating long-lived volcanism as along the Hawaiian chain” (IX p. 157).  “Melts from enriched mantle are most evident at new or slowly rifting regions, infant subduction zones, new backarc basins, slab windows, and mid-plate environments away from spreading induced upwelling.  Enriched mantle is therefore probably shallow.  [The perisphere] is physically isolated from the depleted mantle not by its strength but by its weakness and buoyancy.  It has the chemical characteristics often attributed to continental lithosphere (or plume heads)” (II p. 125).  “The perisphere/asthenosphere is probably laterally and vertically inhomogeneous” (III p. 119).

“If we do away with the widespread notion that the upper mantle is depleted and the lower mantle is enriched, and accept, based on various compelling grounds, that the enriched layer must be a shallow-level layer and the MORB source deeper, no plumes are necessary.  [Advocates] have repeatedly argued that it is incorrect to conceive of the asthenosphere as 'depleted mantle', or MORB source, the shallow mantle as 'homogeneous' and 'barren', and the deep mantle (the supposed source of plumes) as 'enriched'.  Rather, the shallow mantle may be `enriched' mantle and be distributed all around the Earth (the perisphere).  Such a perisphere layer would be a natural outcome of upward migration of volatiles and fluids, carrying incompatible elements which would concentrate in the upper layers (200-400 km depth) and give rise to enriched chemical signatures that are being ascribed to plumes” (VII p. 17).  “The perisphere removes the need for delamination, remobilization, long-distance transport of continental lithosphere, and plumes or plume heads, to explain enriched basalts or so-called hotspot magmas far from continents” (II p. 143).

“Hot areas of the upper mantle may be due to the absence of cooling rather than the importation of plume heads from great depth in the mantle” (III p. 108).  “A moving plate, overriding a hot region of the mantle, and being put into tension, will behave, in many respects, as if it were being impacted from below by a giant plume head” (III p. 120).  “Hot cells are an alternative to plume heads” (III p. 120).  “Rifting causes massive magmatism if the break occurs over hot cells.  CFB may result from the upwellings of already hot, even partially molten, mantle” (III p.99).

“Splitting of cratonic lithosphere (rather than thinning) may allow adiabatic ascent from great depth (>150 km) and, therefore, extensive melting” (III p. 116).  “Large-volume igneous provinces form where the transition from thick to thin lithosphere is abrupt, setting up strong lateral temperature gradients which induce small-scale convection and rapid movement of material through the melting zone” (IV pp. 275-6).  “Thick lithosphere (usually Archean) adjacent to thinner lithosphere may control the locations of flood basalt provinces.  The boundary between thick and thin lithosphere focuses both the strain in the lithosphere and the upwelling convection.  In addition, the condition actually induces a small-scale form of convection that is not present in simple convection and plume models” (IV p. 269). 

To produce OIBs, “the perisphere must be removed by melting before the underlying depleted reservoir would begin to be tapped, and this observation is beautifully consistent with factual evidence from ocean island, innumerable continental rifts of the world, and new ocean ridges which, before they erupt MORB-like melts, erupt lavas of numerous enriched varieties” (VII pp. 18-19).

“The alternative to the perisphere model is that the sources of intraplate volcanism are derived solely from continental mantle.  This can be envisaged as a cold buoyant layer (lithosphere) overlying a thermal layer which is transitional to the asthenosphere.  The advantage of the continental mantle is that it is a reservoir where isotopic signatures may age, and where signatures produced under different tectonic regimes throughout Earth history may be retained.  [The question is - how is it moved] throughout the asthenosphere[?]  The perisphere model has its source components in place, so does not encounter the same problem.  The perisphere model is also compatible with plate motions in the hotspot reference frame.  But if the model is intended to replace plumes, there is no requirement to retain the hotspot frame.  When the effects of this reference frame are removed, there not only arises a mechanism for the lateral distribution of eroded continental mantle throughout the asthenosphere, but also mechanisms for tapping such material as a result of Earth rotation” (IX pp. 158-9).

“The starting point for the construction of any counter-model has to be an acceptance of the evidence for amphibole and phlogopite in the source of OIB.”  The presence of these “minerals indicates that intraplate volcanism results predominantly from compositional, not thermal anomalies.  The low melting point of such minerals would make them susceptible to shear melting to generate tracks, while entrainment into ridge upwelling or local asthenospheric convection cells would lead to rapid generation of large quantities of melt, thereby explaining the generation of LIPs” (IX p. 156).

“In opening ocean basins, the convection cells induced will be transverse rolls with the axis of the cells parallel to that of the rift.” (IX p. 161).  “The convection cells expand as rifting progresses, cycling continental mantle from the suture toward the spreading axis, thereby producing volcanic tracks which may be mistaken as plume tail effects, while erosion of continental mantle from sutures parallel to the rift axis may produce oceanic rises” (IX p. 162).

“Before the hotspot frame was proposed, the Antarctic plate was used as a reference because it is nearly surrounded by ridges and so can be regarded as stationary.  Initial modeling of plate interactions using this plate as a fixed reference indicated a net westward drift of plates at an average rate of 5 cm per year. Unfortunately, many of the early explanations for a net westward plate drift focused on causes external to the Earth such as a tidal lag.  Such mechanisms were subsequently shown to be inadequate, but also the observation of westward plate movement [was shown to be] incompatible with plate motions in the hotspot frame.  The possibility that it was the plume model that was wrong was never considered” (IX p. 159).

“An internal explanation for a westward plate drift can be found when the Earth is envisaged as a series of shells of differing viscosity rotating about a central axis.  Plate lag results from the asthenosphere acting as a zone of decoupling between the lithosphere and the deeper, convecting mantle.  The result is a differential rotation characterized by what appears to be a net westward lag of plates.  The effect will not be purely west-east as undulations in the flow lines are expected as the position of the rotation axis is expected to move in response to plate movements changing the mass distribution at the Earth's surface (polar wander).  As the effect is also a plate lag relative to the deep mantle, it can alternatively be envisaged as eastward mantle flow.  A net westward drift of plates, albeit reduced to a global average of 1.7 cm per year, still exists in the hotspot frame.  The average polar wander rate plus the rotation differential in the hotspot frame equals the average 5 cm per year plate lag determined from the rotation axis reference.  If hotspots were truly fixed and independent of mantle flow, there should be no net westward plate lag.  That requirements for it still exist, and that the magnitude is reduced in the hotspot frame, suggests that (1) westward plate lag is a real effect and not an artifact of using the Antarctic plate as a reference and (2) hotspots are actually a series of shallow melting anomalies which are moving relative to both the lithosphere and mesosphere” (IX p. 160).

“In the plume model, the motions of the various shells of the Earth appear chaotic, with the inner and outer core moving in opposite directions about the rotation axis, the mantle rotating about an equatorial axis, and the lithospheric plates moving in no predominate direction over the surface.  In the differential rotation model, the motion of the shells is more regular with the lithosphere and convection pattern in the outer core showing a net westward drift, while the mantle and inner core move eastwards relative to the rotation axis.  It is reasonable to expect changes in lithospheric plate configuration to also affect the balance of angular momentum.  As the total angular momentum of the Earth must remain constant in the absence of any changes in external torques, balancing angular momentum changes in the lithosphere by changes in core convection patterns would provide a link between magnetic reversals and plate tectonics” (IX p. 160).

“Had alternatives been given the same consideration as the plume model was given when it was first introduced, a less complex picture of the Earth could potentially have been developed in the 1980s from the following concepts:

     1.  Volatile-rich sources (`wetspots'): minerals such as amphibole and phologopite in the sources of intraplate volcanism [remove] the need for large thermal anomalies in the generation of intraplate melts. 

     2.  Eastward mantle flow: Differential rotation of lithosphere and mantle resulting from the transmission of stress through the asthenosphere exerts a drag on the base of continental plates which leads to continental rifting, and provides a mechanism for the lateral introduction and tapping of source material for intraplate volcanism into the oceanic asthenosphere.

     3.  Stress fields: Drag and plate boundary forces acting in opposing directions on oceanic plates set up a counterflow in the asthenosphere and cause melting anomalies generated by shearing to appear stationary.

     4.  Marble cake mantle: Re-mixing of subducted crust into the depleted mantle MORB-source” (IX p. 173).


Without plumes, plate tectonics loses a number of capabilities.  The most important is the ability to rift continents.  The uplift, thinning, and dragging apart of the lithosphere as the plume head strikes will be hard for plate tectonics to replace.  Rifting continents joins the initiation of subduction zones far from continents, mountain building (see article at this website), and the sudden redirection of the Pacific plate as the primary mechanical problems of the theory.  This is due to the strength of the lithosphere plus the need to direct sufficient force in the right directions.  Initiation of subduction zones anywhere has been a difficult issue for plate tectonics.  The theoretical scenario in vogue is that an approximately 10 km high pile of sediment accumulates along the edge of a continent, and appropriate tensile and compressive stresses break the lithosphere.  One wonders what becomes of the vast sediment load, and where the next such event is developing today.

The synthesis of the shock dynamics theory with the differential rotation theory of Smith and Lewis (X pp. 97-116) presents an extraordinary alternative solution.  Shock dynamics explains rifting of the protocontinent and the placement of all continental crust; the formation, shape, and slant of trenches; the opening of mid-ocean ridges; plus the building and location of mountains.  Differential rotation adds explanations for the generation and location of intraplate volcanism as well as its geochemistry, and for lithospheric creep and stress.  The failure of plume theory is an opportunity to discover a new understanding of the Earth.



I  Anderson, Don L. November 15, 2000. The Thermal State of the Upper Mantle; No Role for Mantle Plumes. Geophysical Research Letters, Vol. 27, No. 22, pp. 3623-3626.

II  Anderson, Don L. February 1995. Lithosphere, Asthenospher, and Perisphere. Reviews of Geophysics, Vol. 33, No. 1, pp. 125-149.

III  Anderson, Don L., Yu-Shen Zhang, Toshiro Tanimoto. 1992. Plume heads, continental lithosphere, flood basalts and tomography. In: Magmatism and the Causes of Continental Break-up, Geological Society Special Publication No. 68, eds. Storey, B. C., T. Alabaster, R. J. Pankhurst, pp. 99-124. 

IV  King, Scott D., Don L. Anderson. 1995. An alternative mechanism of flood basalt formation. Earth and Planetary Science Letters, Vol. 136, pp. 269-279.

V  McNutt, M. K., D. W. Caress, J. Reynolds, K. A. Jordahl, R. A. Duncan. 2 October 1997. Failure of plume theory to explain midplate volcanism in the southern Austral islands. Nature, Vol. 389, pp. 479-482.

VI  Price, Neville J. 2001. Major Impacts and Plate Tectonics - A model for the Phanerozoic evolution of the Earth's lithosphere. Routledge, London.

VII  Sheth, H. C. 1999. Flood basalts and large igneous provinces from deep mantle plumes: fact, fiction, and fallacy. Tectonophysics, Vol. 311, pp. 1-29.

VIII  Sleep, Norman. 2 October, 1997. The puzzle of the South Pacific. Nature, Vol. 389, pp. 439-440. 

IX  Smith, Alan D., Charles Lewis. 1999. The planet beyond the plume hypothesis. Earth-Science Reviews, Vol. 48, pp. 135-182.

X  Smith, Alan D., Charles Lewis. 1999. Differential rotation of lithosphere and mantle and the driving forces of plate tectonics.  Geodynamics, Vol. 28, pp. 97-116.

XI  van der Hilst, Robert D. 29 October 2004. Changing Views on Earth's Deep Mantle. Science, Vol. 306, pp. 817-818.  and
Trampert, Jeannot, Frederic Deschamps, Joseph Resovsky, Dave Yuen. 29 October 2004. Probabilistic Tomography Maps Chemical Heterogeneities Throughout the Lower Mantle. Science, Vol. 306, pp. 853-856.