Are splash plumes the origin of minor hotspots?

J.H. Davies School of Earth, Ocean and Planetary Sciences, Cardiff University, Main Building, Park Place, Cardiff, CF10 3YE
Wales, UK

H.-P. Bunge Department of Earth and Environmental Sciences, Ludwig Maximilians University Munich, Theresienstrasse 41,
Munich DEU D-8033, Germany
ABSTRACT
It has been claimed that focused hot cylindrical upwelling plumes cause many of the
surface volcanic hotspots on Earth. It has also been argued that they must originate from
thermal boundary layers. In this paper, we present spherical simulations of mantle circulation at close to Earth-like vigor with significant internal heating. These show, in addition to thermal boundary layer plumes, a new class of plumes that are not rooted in
thermal boundary layers. These plumes develop as instabilities from the edge of bowls of
hot mantle, which are produced by cold downwelling material deforming hot sheets of
mantle. The resulting bowl and plume structure can look a bit like the splash of a water
droplet. These splash plumes might provide an explanation for some hotspots that are not
underlain by thermal boundary layersourced plumes and not initiated by large igneous
provinces. We suggest that in Earths mantle, lithospheric instabilities or small pieces of
subducting slab could play the role of the model downwelling material in initiating splash
plumes. Splash plumes would have implications for interpreting ocean-island basalt geochemistry, plume fixity, excess plume temperature, and estimating core heat flux. Improved seismic imaging will ultimately test this hypothesis.
Keywords: mantle, hotspot, plume, convection.
INTRODUCTION
Hotspots were explained by Morgan (1972)
as resulting from hot cylindrical upwelling
plumes rooted in the deep mantle, which it has
been argued must originate from instabilities
in thermal boundary layers (Schubert et al.,
2001). Therefore, it has been expected that
plumes are rooted at the core-mantle boundary
or possibly at the lower-mantleupper-mantle
boundary, if the mantle is layered.
Seismologically, plumes have been very
difficult to image (Nataf, 2000). Many studies
have claimed to trace plumes through parts of
the upper mantle (e.g., Ritsema and Allen,
2003; Wolfe et al., 1997), with fewer studies
extending them to the lower mantle (Bijwaard
and Spakman, 1999; Montelli et al., 2004; Nataf and Vandecar, 1993; Rhodes and Davies,
2001; Wolfe et al., 1997). The recent seismic
tomography of Montelli et al. (2004) imaged
a range of plumes, a small number of which
extended all the way through the mantle and
others which died out at various depths
through the mantle. While it is possible that
this difference could result from varying ray
coverage beneath different hotspots, could
some of them really not be rooted at the coremantle boundary?
It has been argued that plumes originating
from the core-mantle boundary will start with
large heads, which will lead to large igneous
provinces, followed by thin tails, which will
lead to a time-progressive chain of volcanoes
(Duncan and Richards, 1991; Richards et al.,
1989). There are, however, hotspots that do
not seem to start with a large igneous province
and/or have a time-progressive chain of vol-

canic centers. Courtillot et al. (2003) argued

that the mantle has three classes of plumes
that might be able to explain this difference.
Others have argued that hotspots result from
different non-thermal boundary layer plume
sources, such as cracks (Turcotte and Oxburgh, 1973), compositional heterogeneity
(Foulger and Natland, 2003; Schilling et al.,
1980), edge convection (King and Ritsema,
2000), or phase transition instabilities (Cserepes and Yuen, 2000). These are all shallow
sources that would be unable to explain
plumes rooted in the middle of the lower mantle. In this work, we demonstrate, using numerical mantle convection models, that small
cylindrical thermal upwelling need not be
rooted in thermal boundary layers, and that
they can be produced in purely thermal isochemical convection models. These new
plumes might be significant, since they provide a means for explaining some of the small
hotspots that are not associated with large igneous provinces and some of the plumes that
seismically seem to originate away from thermal boundary layers.
MODELING
Here we concentrate on presenting the discovery of the plumes not rooted in thermal
boundary layers through the results of one numerical simulation. The method used here
closely follows the work of Bunge and coworkers (Bunge and Davies, 2001; Bunge et
al., 2002; Davies and Bunge, 2001) and uses
TERRA (Baumgardner, 1985; Bunge et al.,
1997), an established, widely used, benchmarked (Bunge et al., 1997) parallel code. The

most significant difference from the earlier

work is the higher spatial resolution (nearly
an order of magnitude more nodes: !80 million nodes), allowing the calculation to have
a lower reference viscosity (asthenosphere viscosity !8 " 1020 Pa s). This leads to a vigor
of convection which is significantly Earthlike. With a free-slip upper-boundary condition, the model produces root mean square
surface velocity of !4 cm/yr and an uppermantle temperature of around 1600 K. While
all the essential details of the model are presented here, further supplementary details of
certain aspects are presented in the GSA Data
Repository item.1 The case had spatially uniform internal heating (approximately presentday chondritic Earth), was undergoing secular
cooling, and had fixed temperature at its upper
surface and the core-mantle boundary, such
that the ratio of heat crossing the core-mantle
boundary to heat crossing the surface was
10% (see footnote one). Compressible convection was assumed using the same Murnaghan equation of state as Bunge et al.
(1997). This led to a coefficient of thermal
expansion monotonically decreasing with
depth from !4 " 10#5 K#1 at the surface to
1.25 " 10#5 K#1 at the core-mantle boundary.
The viscosity profile is the same as presented
in Bunge et al. (2002), i.e., with a 40-fold increase in viscosity from the upper mantle to
the lower mantle, but with all values being
over an order of magnitude lower. The actual
parameter values are listed in Table 1. The
surface of the circulation model is driven by
plate-motion history of the past 119 m.y. in
11 stages. The initial condition for the circulation model was produced by driving the surface with the plate motions of 119 Ma for !1
b.y. (see footnote one). The actual lithosphere
in the final model was comparatively thin (average !65 km) compared to present-day Earth
(!100 km; see footnote one); while the heat
flow was higher by roughly the same proportion, at 157 mWm#2 (cf. 90 mWm#2 for Earth
today). This difference has the effect of reducing the magnitude of the negative buoyancy that collects in the upper thermal boundary layer (lithosphere).
1GSA Data Repository item 2006070, supplementary model details, is available online at
www.geosociety.org/pubs/ft2006.htm, or on request
from editing@geosociety.org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 803019140, USA.

! 2006 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or editing@geosociety.org.
Geology; May 2006; v. 34; no. 5; p. 349352; doi: 10.1130/G22193.1; 1 figure; 2 tables; Data Repository item 2006070.

349

TABLE 1. PARAMETERS FOR REFERENCE SIMULATION

Parameter

Value

Unit

Surface temperature
CMB temperature
Internal heat generation
Asthenosphere viscosity
Thermal conductivity
Coefficient of thermal expansion ($)
Gravitational acceleration
Specific heat

300
2700
6 " 10#12
8 " 1020
5.5
Varies with depth
!10
103

K
K
W Kg#1
Pa s
W m K#1
K#1
m s#2
J kg#1 K#1

Note: Case 500.

We illustrate how these plumes initiate and

develop using a multipart figure showing different views of the thermal anomaly structure
of the final present-day time step of the circulation model. The plumes (Fig. 1A, feature
b) originate from bowls of hot mantle (Fig.
1A, features a and c). These develop from regions unaffected by recent subduction cooling
that are heated up by uniform distributed internal heating and that rise slowly. These regions are then met from above by cold downwelling material (e.g., see above c in Fig. 1A),
which pushes them down, compressing them
into sheets and then bowls (Fig. 1A, a and c).
The hot sheets envelop the underside of the
downwelling material and rise up around the
side (c, d). Hence, there are frequently regions
of cold downwelling surrounded, rimmed, by
a relatively thin region of hot material on radial surfaces. These upwelling sheets can develop plume-like instabilities on their rims
(Fig. 1A, feature B). The driving downwelling
material results in widespread minor linear instabilities, which are more significant at subduction zones; as it descends, it becomes
thicker and more cylindrical. Given the visual
similarity of these plumes to the splash of a
water droplet caught in a high speed camera,
we use the term splash plumes as an abbreviation for these plumes not rooted in a
thermal boundary layer. Note the physics
of splash plumes is different from inertiadominated water splashes.
Figure 1B focuses on the Pacific hemisphere and allows us to view the American
splash plumes from the side, showing clearly
that they are not rooted in the core-mantle
boundary. This figure shows the large sheet
feature (e) beneath North America with two of
its plumes, one rising beneath eastern North
America (f) and another beneath the Amazon
(g), and another sheet being downwarped beneath Asia (c). We also show that these sheets
can be pushed all the way down to the coremantle boundary, as beneath Samoa (h). These
splash plumes (h) would be hard to distinguish
from traditional plumes rooted in a thermal
boundary layer. This figure shows that away
from regions affected by the downwelling material, e.g., middle of Pacific (i), ridge and
plume-type structures, such as those expected
from earlier work (Houseman, 1990), exist. It
would be possible for two or more downwell350

ing bodies to descend to the core-mantle

boundary near each other and push the hot
material together, forming a large upwelling or
superswell.
Figure 1C focuses on the Americas. It
shows the East Pacific Rise (j) as very shallow, as would be expected for a passive structure. We also see the eastern North America
(f), Amazon (g), and Hudson (k) plumes all
rising from a relatively complex interaction
between a hot sheet and cold downwelling
material, and another plume rising from a different sheet beneath southeastern Brazil (l).
The figure also shows a cross section of the
small scale of the instabilities descending
from the surface (m), and again shows the way
that they combine to move the passive hot material down (n).
The purpose of this report is to draw attention to this new mantle convection phenomenon (it is unclear whether it is a new fluid
mechanic phenomenon). While we do not yet
have the computational resources to undertake
a comprehensive search of parameter space,
we have undertaken many additional simulations to better characterize the conditions for
the formation of the splash plumes. The
changes in the parameters away from the reference case for the additional simulations are
listed in Table 2. The result of these additional
simulations is that splash plumes still exist,
even when the ratio of bottom heating to internal heating is much greater, and when an
endothermic phase change is included at 660
km depth, but not when the coefficient of thermal expansion is made constant with depth.
The additional runs also show that splash
plumes are weaker in cases where the lithosphere is more viscous and hence more stable.
In the two cases where the splash plumes are
weaker or not present, the downwelling bodies
are larger in magnitude (either because of increased negative buoyancy due to a higher coefficient of thermal expansion or increased
lithosphere viscosity), and hence they are
more capable of reaching the core-mantle
boundary. In such cases, where downwelling
bodies reach the core-mantle boundary, it is
very difficult to tell splash plumes generated
by such downwelling material from thermal
boundary layer plumes. We saw an example
of this beneath Samoa in our reference case,
in Figure 1B (feature h).

Figure 1. Present-day thermal anomaly

structure of mantle circulation model; i.e.,
layer average is removed. Scale bar shows
thermal perturbations (in Kelvin). Solid
lines at surface represent coastlines, while
dashed lines are tectonic boundaries. A:
Focused on Indian Ocean, showing a radial surface just above core, a cross section, and a hot (yellow) isosurface (!300
K), restricted to deeper than 1910 km to
better visualize deep structure. Figure
shows four plume-like features coming off
bowl feature toward Africa (B). B: Focused
on Pacific Ocean, with hot isosurface extending to surface. Figure shows splash
plumes f and g clearly not rooted at the
core-mantle boundary. C: Focused on
Americas. As in B, except isosurface is
!250 K and not restricted, and radial surface is at 1656 km depth. This figure is discussed further in text.

GEOLOGY, May 2006

TABLE 2. CHANGED PARAMETER VALUES FROM CASE 500 (TABLE 1) FOR ADDITIONAL CASES
Case

Difference from reference caseCase 500

Splash plumes?

501
502

Rate of internal heating % 3 " 10#12 W/Kg, TCMB % 3500 K

Phase changesEquivalent Clapeyron Slopes 410 km % 1.65 MPa/K, 660 km
% #2.25 MPa/K; assuming olivine component makes up 65% of mantle
$ % 4 " 10#5 K#1 Constant with depth, Boussinesq
More viscous lithosphere100 times more viscous

Yes
Yes

503
505

The critical ingredients to generate model

mid-mantle splash plumes appear to be (1) lateral variations in temperature, which could result from regions having different subduction
or cooling histories (these variations are probably favored by internal heating and vigorous
convection); and (2) downwelling bodies that
interact with the hot regions but that are sufficiently weak such that they are unable to
reach the core-mantle boundary. These may be
common characteristics of Earth, at least in
parts of the mantle. The fact that splash
plumes have not been prominent in earlier
work (e.g., Bunge et al., 2002) probably results from the fact that the downwelling material in the cases studied was stronger, partly
as a result of the stronger thicker lithosphere
(!200 km thick) and the higher viscosity of
the asthenosphere (8 " 1021 Pa s), which limited episodes of weak downwelling.
DISCUSSION
Some differences that we might expect between the simulation and Earth are that the
upper boundary in reality would be more viscous and thicker than in the reference simulation, and therefore Earth might form fewer
but stronger downwelling bodies. On Earth,
subducting slabs are the dominant downwelling structures, and so one might expect hot
internal heat to be pushed back down beneath
subduction zones. We might therefore speculate that a process related to splash plumes
might explain, for example, the Korean and
Chinese (Wee, 2002), Massif Central and Eifel
(Wedepohl and Baumann, 1999), and Vietnam
and Tibet magmatism (Hoang and Flower,
1998; Wang et al., 2001) behind their respective subduction zones. We note, however, that
they could also partly reflect backarc extension on continents, and possibly the effects of
slab breakoff (Davies and von Blanckenburg,
1995). The downwelling bodies also change
more quickly with depth from linear to a more
cylindrical planform in our simulation than is
seen in seismic tomography. For computational reasons, we have not incorporated
temperature-dependent viscosity in these simulations. This would tend to make downwelling bodies stiffer and allow the slabs to retain
their linear form deeper into the mantle; and
also it would also make it easier for thin
plumes to form. Another significant factor in
deciding how important mid-mantle splash
plumes will actually be is how far down the
downwelling material can push the hot sheets
before they become unstable and form
GEOLOGY, May 2006

No
Weaker

plumes. If the sheets can be pushed all the

way to the core-mantle boundary, then they
will look similar to traditional plumes rooted
in the hot bottom thermal boundary layer. In
fact, this process might be more important
than boundary-layer instabilities, even for
plumes rooted in the core-mantle boundary.
While this work identifies the exciting new
process of splash plumes, we note that there
is much more work to be done in the future
to understand the controls of various parameters, including heating and rheology. Mindful
of the limitations, we speculate that another
setting where downwelling material might
produce splash plumes includes lowerlithosphere instabilities beneath slow-moving
continental plates. Possible candidates that
might be worthy of further investigation include Tibesti, Hoggar, and Darfur beneath Africa (Dupuy et al., 1993; Franz et al., 1994;
Gourgaud and Vincent, 2004), Newer, in Victoria, Australia (McBride et al., 2001), and
Comores off Madagascar in the Indian Ocean
(Spath et al., 1996). Another possibility might
be an instability triggered by a continental
collisionan example to pursue in this case
might be the volcanics around Baikal, Siberia
(Johnson et al., 2005), where a downwelling
instability might have been initiated around
3540 Ma beneath Baikal by the India-Asian
collision, which might also have initiated the
local rifting. A further possible setting for
generating splash plumes is where subduction
history has been brief, or has resulted in a subducting slab separating into pieces. Finally,
lithospheric instabilities might be possible beneath oceanic lithosphere (Jaupart and Parsons, 1985), and any resulting splash plumes
could be the cause of short-lived chains of islands, such as those between Rurutu Islands
and Raivavae Islands in the Cook-Austral
chain in the central Pacific (Bonneville et al.,
2002).
One of the few direct ways proposed for
estimating the core heat flux is by evaluating
the topographic swell of hotspot tracks and
then attributing the cause of excess buoyancy
to thermal plumes originating at the coremantle boundary (Davies, 1988; Sleep, 1990).
While, as described in the introduction, the
current imaging of mantle plumes is improving, it still does not allow all plume structures
to be traced to their source. If any of the
plumes used to estimate the core heat flux are
splash plumes, then we would be overestimating the heat coming from the core.

We might expect plumes resulting from the

same splash to be relatively fixed, but we
would observe greater relative motion for
plumes related to different sheets. This might
also apply to plumes with core-mantle boundary sources, if they are controlled by downwelling phenomena like the splash plumes
pushed to the core-mantle boundary here; we
might then expect only slow relative motion
as the plumes of a splash are pushed slowly
apart and affected by the background mantle
flow as they rise (Steinberger et al., 2004).
This might, therefore, explain why we find
clusters of relatively fixed plumes, but with
more mobility between different clusters, e.g.,
Pacific versus Atlantic/Indian Ocean plumes
(Molnar and Stock, 1987; Steinberger et al.,
2004). Splash plumes would also have geochemical implications, since they would introduce the interior away from the thermal
boundary layer as another source reservoir for
hotspots.
A problem for a simple version of the traditional thermal boundary layer plume model
is the low excess temperature frequently evaluated for hotspots from petrologic data compared to the much higher values predicted by
simple dynamic plume models (Bunge, 2005;
Farnetani, 1997). While this can be addressed
by considering non-isochemical systems (Farnetani, 1997), or the influence of internal heating and non-adiabaticity (Bunge, 2005), some
cases might be caused by splash plumes. Since
the temperature increase in the source region
of a splash plume is only 250 K over 2 b.y.
(assuming no thermal loss by conduction,
chondritic rate of heating of !5 " 10#12 W/
kg, and a specific heat capacity of !1.2 KJ/
mol), we would generally not expect very
large excess temperatures. Equally, there are
interpretations of plumes with buoyancy fluxes so low that if they originated from the coremantle boundary, they would not, from simple
modeling, be expected to reach the surface
(Albers and Christensen, 1996). Splash
plumes, since they originate closer to the surface and with lower excess temperatures,
might allow a simple way to reconcile these
problems for certain hotspots.
While, as mentioned earlier, tomography results sometimes do show plumes starting in
mid-mantle (Montelli et al., 2004), we are not
aware of any definite tomographic imaging of
splash plumes. The difficulty in imaging
splash plumes should not be surprising, since
the horizontal bowls are hard to image, while
the downwelling and upwelling bodies are
likely to be thinner than in our simulations
due to temperature-dependent viscosity. They
would therefore currently be difficult to resolve seismically and hence will tend to average to a weak signal. The plume signature
will also be reduced by wavefront healing
(Gudmundsson, 1996), where the fast parts of
the wavefronts that avoided the slow region
351

diffract to join together and remove the signature of the slow features along the travel
path. This could be especially true with splash
plumes, since they are near fast waveguides of
subducting slabs or cold downwelling bodies.
Speculatively, one might combine regional
and local seismic studies to argue for imaging
of splash plumes, though, for example, the
very slow velocities imaged at depth beneath
Europe might in part reflect a large bowl
(Goes et al., 1999; Hoernle et al., 1995), while
the plumes imaged beneath Eifel and Massif
Central might then be a splash plume originating from that bowl (Granet et al., 1995;
Keyser et al., 2002). Ultimately, the hypothesis that splash plumes exist in the mantle will
be testable by improved seismic imaging.
CONCLUSION
In numerical mantle circulation simulations,
we have discovered upwelling plumes that are
not rooted in a thermal boundary layerwe
call them splash plumes. They have the potential to explain hotspots not initiated by
large igneous provinces, with moderate excess
temperatures, with tails that do not reach the
core-mantle boundary, and with short life
spans.
ACKNOWLEDGMENTS
I. Thomas (Cardiff) and A. Heath (Liverpool) developed the visualization software used for Figure 1.
Higher Education Funding Council for Wales (Strategic
Research Infrastructure Fund) and Higher Education
Funding Council for England (Joint Research Equipment Initiative) funded the computing clusters Helix
(Cardiff) and Networked Earth Sciences Supercomputing Cluster (Liverpool). We acknowledge insightful
comments from J. Ritsema, A. Kerr, and anonymous
reviewers.

REFERENCES CITED
Albers, M., and Christensen, U.R., 1996, The excess
temperature of plumes rising from the core-mantle boundary: Geophysical Research Letters,
v. 23, p. 35673570.
Baumgardner, J.R., 1985, Three dimensional treatment
of convective flow in the Earths mantle: Journal
of Statistical Physics, v. 39, p. 501511.
Bijwaard, H., and Spakman, W., 1999, Tomographic evidence for a narrow whole mantle plume below
Iceland: Earth and Planetary Science Letters,
v. 166, p. 121126.
Bonneville, A., Le Suave, R., Audin, L., Clouard, V.,
Dosso, L., Gillot, P.Y., Janney, P.,
Jordahl, K., and Maamaatuaiahutapu, K., 2002,
Arago Seamount: The missing hotspot found in
the
Austral
Islands: Geology, v. 30,
p. 10231026.
Bunge, H.P., 2005, Low plume excess temperature and
high core heat flux inferred from nonadiabatic geotherms in internally heated mantle
circulation models: Physics of the Earth and Planetary Interiors, v. 153, p. 310.
Bunge, H.P., and Davies, J.H., 2001, Tomographic images of a mantle circulation model: Geophysical
Research Letters, v. 28, p. 7780.
Bunge, H.P., Richards, M.A., and Baumgardner, J.R.,
1997,
A
sensitivity
study
of
threedimensional spherical mantle convection at 10(8)
Rayleigh number: Effects of depth-dependent viscosity, heating mode, and an endothermic phase
change: Journal of Geophysical Research, Solid
Earth, v. 102, p. 11,99112,007.
Bunge, H.P., Richards, M.A., and Baumgardner, J.R.,

352

2002, Mantle-circulation models with sequential

data assimilation: Inferring present-day mantle
structure from plate-motion histories: Philosophical Transactions of the Royal Society of London,
ser. A, v. 360, p. 25452567.
Courtillot, V., Davaille, A., Besse, J., and Stock, J.,
2003, Three distinct types of hotspots in the
Earths mantle: Earth and Planetary Science Letters, v. 205, p. 295308.
Cserepes, L., and Yuen, D.A., 2000, On the possibility
of a second kind of mantle plume: Earth and Planetary Science Letters, v. 183, p. 6171.
Davies, G.F., 1988, Ocean bathymetry and mantle convection, 1. Large-scale flow and hotspots: Journal
of Geophysical Research, Solid Earth and Planets,
v. 93, p. 10,46710,480.
Davies, J.H., and Bunge, H.P., 2001, Seismically fast
geodynamic mantle models: Geophysical Research Letters, v. 28, p. 7376.
Davies, J.H., and von Blanckenburg, F., 1995, Slab
breakoffA model of lithosphere detachment
and its test in the magmatism and deformation of
collisional orogens: Earth and Planetary Science
Letters, v. 129, p. 85102.
Duncan, R.A., and Richards, M.A., 1991, Hotspots,
mantle plumes, flood basalts, and true polar wander: Reviews of Geophysics, v. 29, p. 3150.
Dupuy, C., Dostal, J., and Chikhaoui, M., 1993, Traceelement and isotopic geochemistry of Cenozoic
alkali basaltic lavas from Atakor (central Sahara):
Geochemical Journal, v. 27, p. 131145.
Farnetani, C.G., 1997, Excess temperature of mantle
plumes: The role of chemical stratification across
D: Geophysical Research Letters, v. 24,
p. 15831586.
Foulger, G.R., and Natland, J.H., 2003, Is hotspot
volcanism a consequence of plate tectonics?: Science, v. 300, p. 921922.
Franz, G., Pudlo, D., Urlacher, G., Haussmann, U.,
Boven, A., and Wemmer, K., 1994, The Darfur
Dome, western SudanThe product of a subcontinental mantle plume: Geologische Rundschau,
v. 83, p. 614623.
Goes, S., Spakman, W., and Bijwaard, H., 1999, A lower mantle source for central European volcanism:
Science, v. 286, p. 19281931.
Gourgaud, A., and Vincent, P.M., 2004, Petrology of
two continental alkaline intraplate series at Emi
Koussi volcano, Tibesti, Chad: Journal of Volcanology and Geothermal Research, v. 129,
p. 261290.
Granet, M., Stoll, G., Dorel, J., Achauer, U., Poupinet,
G., and Fuchs, K., 1995, Massif-Central
(France)New constraints on the geodynamical
evolution from teleseismic tomography: Geophysical Journal International, v. 121, p. 3348.
Gudmundsson, O., 1996, On the effect of diffraction on
traveltime measurement: Geophysical Journal International, v. 124, p. 304314.
Hoang, N., and Flower, M., 1998, Petrogenesis of Cenozoic basalts from Vietnam: Implication for origins of a diffuse igneous province: Journal of
Petrology, v. 39, p. 369395.
Hoernle, K., Zhang, Y., and Graham, D., 1995, Seismic
and geochemical evidence for large-scale mantle
upwelling beneath the eastern Atlantic and western and central Europe: Nature, v. 374, p. 3439.
Houseman, G., 1990, The thermal structure of mantle
plumes: Axisymmetric or triple-junction?: Geophysical Journal International, v. 102, p. 1524.
Jaupart, C., and Parsons, B., 1985, Convective instabilities in a variable viscosity fluid cooled from
above: Physics of the Earth and Planetary Interiors, v. 39, p. 1432.
Johnson, J., Gibson, S., Thompson, R., and Nowell, G.,
2005, Volcanism in the Vitim volcanic field, Siberia: Geochemical evidence for a mantle plume
beneath the Baikal rift zone: Journal of Petrology,
v. 46, p. 13091344.
Keyser, M., Ritter, J.R.R., and Jordan, M., 2002, 3D
shear-wave velocity structure of the Eifel plume,
Germany: Earth and Planetary Science Letters,
v. 203, p. 5982.

King, S.D., and Ritsema, J., 2000, African hot spot volcanism: Small-scale convection in the upper mantle beneath cratons: Science, v. 290,
p. 11371140.
McBride, J.S., Lambert, D.D., Nicholls, I.A., and Price,
R.C., 2001, Osmium isotopic evidence for crustmantle interaction in the genesis of continental
intraplate basalts from the Newer Volcanics Province, southeastern Australia: Journal of Petrology,
v. 42, p. 11971218.
Molnar, P., and Stock, J., 1987, Relative motions of hotspots in the Pacific, Atlantic and Indian Oceans
since Late Cretaceous time: Nature, v. 327,
p. 587591.
Montelli, R., Nolet, G., Dahlen, F.A., Masters, G., Engdahl, E.R., and Hung, S.H., 2004, Finite-frequency tomography reveals a variety of plumes in the
mantle: Science, v. 303, p. 338343.
Morgan, W.J., 1972, Plate motions and deep mantle
convection, in Shagam R., et al., eds., Studies in
earth and space sciences: Geological Society of
America Memoir 132, p. 722.
Nataf, H.C., 2000, Seismic imaging of mantle plumes:
Annual Review of Earth and Planetary Sciences,
v. 28, p. 391417.
Nataf, H.C., and Vandecar, J., 1993, Seismological detection of a mantle plume: Nature, v. 364,
p. 115120.
Rhodes, M., and Davies, J.H., 2001, Tomographic imaging of multiple mantle plumes in the uppermost
lower mantle: Geophysical Journal International,
v. 147, p. 8892.
Richards, M.A., Duncan, R.A., and Courtillot, V.E.,
1989, Flood basalts and hot-spot tracksPlume
heads and tails: Science, v. 246, p. 103107.
Ritsema, J., and Allen, R.M., 2003, The elusive mantle
plume: Earth and Planetary Science Letters,
v. 207, p. 112.
Schilling, J.-G., Bergeron, M.B., and Evans, R., 1980,
Halogens in the mantle beneath the North Atlantic: Philosophical Transactions of the Royal Society of London, ser. A, v. 287, p. 147178.
Schubert, G., Turcotte, D.L., and Olson, P., 2001, Mantle convection in the Earth and planets: Cambridge, UK, Cambridge University Press, 940 p.
Sleep, N.H., 1990, Hotspots and mantle plumesSome
phenomenology: Journal of Geophysical Research, Solid Earth and Planets, v. 95,
p. 67156736.
Spath, A., LeRoex, A.P., and Duncan, R.A., 1996, The
geochemistry of lavas from the Comores archipelago, western Indian Ocean: Petrogenesis and
mantle source region characteristics: Journal of
Petrology, v. 37, p. 961991.
Steinberger, B., Sutherland, R., and OConnell, R.J.,
2004, Prediction of Emperor-Hawaii seamount locations from a revised model of global plate motion and mantle flow: Nature, v. 430, p. 167173.
Turcotte, D.L., and Oxburgh, E.R., 1973, Mid-plate tectonics: Nature, v. 244, p. 337339.
Wang, J.H., Yin, A., Harrison, T.M., Grove, M., Zhang,
Y.Q., and Xie, G.H., 2001, A tectonic model for
Cenozoic igneous activities in the eastern IndoAsian collision zone: Earth and Planetary Science
Letters, v. 188, p. 123133.
Wedepohl, K.H., and Baumann, A., 1999, Central European Cenozoic plume volcanism with OIB
characteristics and indications of a lower mantle
source: Contributions to Mineralogy and Petrology, v. 136, p. 225239.
Wee, S.M., 2002, Geochemistry and isotopic
systematics of Cenozoic alkaline volcanic
rocks in Korea and NE China: Neues Jahrbuch
fur Mineralogie Abhandlungen, v. 177,
p. 213240.
Wolfe, C.J., Bjarnason, I.T., VanDecar, J.C., and Solomon, S.C., 1997, Seismic structure of the Iceland
mantle plume: Nature, v. 385, p. 245247.
Manuscript received 22 August 2005
Revised manuscript received 5 December 2005
Manuscript accepted 14 December 2005
Printed in USA

GEOLOGY, May 2006