Upon reading “Looking for mantle plumes” by Eugene Humphreys and Brandon Schmandt (Physics Today, August 2011, page 34), I had a déjà vu moment. The first part was very familiar. It is based on laboratory fluid-injection experiments of the 1980s, the 1988 Cambridge thin-plate-geotherm model, and geochemical papers of the time.1 (The 1988 model has been abandoned by Cambridge seismologists and is not supported by data or by realistic simulations.) Thermodynamics, self-compression (gravity), secular cooling, radioactivity, and contemporaneous geophysics were ignored, as noted at the time.2 The model favored in the 1980s required a homogeneous, unheated, adiabatic, melt-free upper mantle and a primordial gas-rich lower mantle and relied on simulations that were unscaled for size, pressure, and gravity.1 Plate-tectonic and upper-mantle boundary-layer mechanisms were discounted ab initio. Physics, which fundamentally rules out narrow upwellings in an internally heated Earth-size planet,3 was not a consideration, as is apparent in the figures of the Physics Today article.

Although mantle plume hypotheses are moving targets, the one adopted is particularly out of date.4 The thin-plate geotherm in the article’s figure 2 does not explain the low-velocity zone, anisotropy, absolute wave speeds, and vertical gradients of wave speeds. Geophysical and tectonic models that take into account the neglected physics tell a different story,2,3 as do newer USArray data and plate-tectonic reconstructions5 (four-dimensional tomography). Red blobs in relative tomography apparently are slab gaps and ambient mantle, not hot plumes. The unphysical scaling assumptions (red equals hot upwelling) plus the 1988 homogeneity and thin-plate assumptions1 implied that Yellowstone could not have a plausible tectonic, nonplume, or shallow-mantle explanation as most continental hotspots do.6 

Geophysical modeling produces large-scale subadiabatic structures in the deep mantle that are typical of normal internally heated planetary convection, a mode that the original mantle plume hypothesis was intended to replace. Well-constrained seismic inversions invariably produce a thick (220-km) anisotropic, heterogeneous boundary layer, which violates the thin-plate and ambient-mantle assumptions and explains why some seismic experiments apparently image near-vertical streaks under the array.3 Significantly, sources deep within that layer satisfy thermodynamically constrained petrological data and, as J. Tuzo Wilson pointed out in 1963, are fixed enough to create volcanic chains.

A hypothesis needs to be challenged if it violates physics and thermodynamics. Plume hypotheses are too ill-defined and flexible1,4 to be tested, but scaling relations and assumptions can be. There has not been a recent physically based critique of the hypothesis. However, boundary layer physics; the effects of radioactivity, pressure, and secular cooling on the geotherm and on mantle dynamics; and the effects of anelasticity, anharmonicity, and anisotropy on Earth models and on the interpretation of tomography indicate that apparent sightings of plumes are not based on data or theory. They are instead the result of the unphysical assumptions and nonunique interpretations of data such as relative near-vertical travel times. Studies that address fundamental problems in mantle and planetary physics2,3 usually ignore the plume hypothesis and will not appear in searches for mantle plumes, but they make a strong case against narrowly focused and spontaneous deep upwellings (not imposed by unnatural boundary conditions that violate the second law) being responsible for surface volcanoes.

Implications of such studies3 include the following:

‣ The geotherm of the mantle sampled at ridges is not representative of ambient mantle.

‣ When the conductive gradient shown in the article’s figure 2b is extended to depths required by surface waves, temperatures are hundreds of degrees hotter than shown.

‣ Ambient mantle, from midplate geophysical and petrological data, is 150–200 K hotter than assumed in thin-plate models.1 

‣ The global low-velocity zone contains 1–2% melt, on average.

‣ Thermodynamically consistent subplate geotherms are subadiabatic and 300 K colder at lower mantle depths than assumed; that makes plumes, if they exist, useless for providing excess temperatures.

The effects of compression, secular cooling and anisotropy, and properly scaled simulations eliminate mantle plumes as an observational fact or a viable physical theory.3 The physics-based and surface-wave-based plume alternative is simply this: Midplate volcanoes tap into a thick sheared anisotropic boundary layer that is sufficiently hot, fertile, large, and fixed, at depth, to explain volcanic chains; the layer is disrupted at ridges.3,4 

1.
D.
McKenzie
,
M. J.
Bickle
,
J. Petrol.
29
,
625
(
1988
);
M. A.
Richard
,
R. A.
Duncan
,
V. E.
Courtillot
,
Science
246
,
103
(
1989
), and references therein.
For physically realistic models see
A.
Maggi
 et al,
Geophys. J. Int.
166
,
1384
(
2006
), and
W. B.
Moore
,
J. Geophys. Res.
113
,
B11407
(
2008
).
2.
A. M.
Dziewonski
,
D. L.
Anderson
,
Phys. Earth Planet. Inter.
25
,
297
(
1981
);
H.
Sato
,
I. S.
Sacks
,
T.
Murase
,
J. Geophys. Res.
94
,
5689
(
1989
);
D. L.
Anderson
,
Theory of the Earth
,
Blackwell Scientific
,
Boston
(
1989
); 
D. L.
Anderson
,
T.
Tanimoto
,
Y.-S.
Zhang
,
Science
256
,
1645
(
1992
);
R.
Jeanloz
,
S.
Morris
,
Geophys. Res. Lett.
14
,
335
(
1987
).
4.
For discussions of “modern” mantle plume theories, see http://www.mantleplumes.org/TopPages/ShearDrivenUpwellingTop.html and http://www.mantleplumes.org/LLAMA.html.For discussions of “modern” mantle plume theories, see http://www.mantleplumes.org/TopPages/ShearDrivenUpwellingTop.html and http://www.mantleplumes.org/LLAMA.html.
5.
G. L.
Pavlis
 et al,
Tectonophysics
532–535
,
82
(
2012
);
D. E.
James
 et al,
Earth Planet. Sci. Lett.
311
,
124
(
2011
);
6.
D.
Zhao
 et al,
Chin. Sci. Bull.
49
,
1401
(
2004
);
M. J. R.
Wortel
,
W.
Spakman
,
Proc. K. Ned. Akad. Wet.
95
,
325
(
1992
);
R.
Meyer
,
G. R.
Foulger
,
“The European Cenozoic Volcanic Province is not caused by mantle plumes,”
http://www.mantleplumes.org/Europe.html.
See also
Geological Society of America Special Papers
, vols. 388,
430
, 470, http://specialpapers.gsapubs.org/content/by/year.