there are hundreds of thousands of small volcanic seamounts, most unrelated to mantle plumes. This entry will
focus on only hot spots related to mantle plumes. Campbell (2007) and White (2010) provide additional reviews.
Plumes and large igneous provinces
Early laboratory and numerical simulations of plumes (e.g.,
Whitehead and Luther, 1975; Campbell and Griffiths, 1990;
Griffiths and Campbell, 1990; Olsen, 1990) suggested that
plumes necessarily start with a large voluminous head. As
Morgan (1972) noted, a number of oceanic island and seamount chains extrapolate to large igneous provinces, or
LIPs, which include flood basalts or “traps” on continents
and oceanic plateaus in the oceans and result from massive
eruptions of basalt over a relatively short period of time. For
example, the Reunion plume, currently beneath volcanically active Reunion Island, can be connected through Mauritius, the Mascarene Plateau, and the Chagos-Laccadive
Ridge to the Deccan Traps flood basalt province of India,
produced 65 Ma ago by the eruption of more than
500,000 km
3 of basalt in roughly a million-year period.
These observations led to the hypothesis that LIPs are the
products of plume heads and oceanic islands are the product
of plume tails (Richards et al., 1989). This idea became an
essential component of the plume theory. However, relatively few oceanic island chains can be confidently associated with LIPs. Subsequent numerical simulations indicate
that the surface expression of thermochemical plumes, as
opposed to purely thermal plumes, “may be a headless,
age-progressive volcanic chain” (Farnetani and Samuel,
2005). Oceanic plateaus, a variety of LIP, are the subject
of a separate entry and will not be further discussed here.
How many plumes?
Table 1 lists 29 oceanic island chains/hot spots possibly
produced by mantle plumes, but just how may mantle
plumes are presently active is unclear and debated. Morgan suggested there were about 20 mantle plumes. Sleep
(1990) estimated buoyancy fluxes for 37 plumes.
Courtillot et al. (2003) considered 49 “hot spots” and concluded that only 7–10 were products of deep mantle
plumes based on five criteria: (1) a long-lived track, (2) a
LIP at initiation, (3) a topographic anomaly indicative of
a buoyancy flux in excess of 10
3 kg/s, (4) high
3 He/
4
He
and
20
Ne/
22 Ne ratios, and (5) negative shear-wave anomalies indicative of hot mantle. Only those hot spots meeting at least three or more of these criteria were
considered to be of “primary plumes” of likely deep mantle origin. They suggested that many of the remaining hot
spots were “secondary,” perhaps triggered by large low
shear-wave velocity provinces (LSSVPs) located at the
base of the mantle and extend up to 1,000 km above it
beneath Africa and the central equatorial Pacific.
As noted earlier, it is possible that plumes might begin
without large buoyant heads and some may be short lived
or young. In addition, Courtillot’s noble gas isotope criterion assumes mantle plumes always consist of relatively
undegassed material, whereas other chemical evidence
discussed below suggests recycled oceanic crust may be
the dominant component in many cases. Thus, of the
Courtillot’s 5 criteria, the last, seismic evidence of hot
mantle, is perhaps the most important. Subsequent seismic
studies by Montelli et al. (2004, 2006) found slow seismic
wave velocity anomalies under 30 of 33 hot spots that
extended at least to depths greater than the upper–lower
mantle boundary at 660 km. Of these, the velocity anomalies extend to depths of greater than 1,450 km in 23 of
the hot spots examined. In addition, Montelli et al. (2006)
identified three “plumes” in the lowermost mantle that did
not extend into the upper mantle (and are not associated
with surface hot spots).
Ebinger and Sleep (1998) suggested that ponding and
flow of the “Ethiopian” or Afar plume beneath the African
lithosphere could account not only for the Mid-Tertiary
Ethiopian flood basalt province and modern Afar volcanism but also hot spots in central North Africa such as
Tibesti, the Cameroon line volcanism, and the oceanic
island volcanism on the Comoros. Furthermore, some of
the plumes imaged by Montelli et al. appear to merge at
depth, such as the Ascension and St. Helena plumes. Thus,
a single plume in some cases might be responsible for
more than one hot spot.
Hot spots aren’t entirely stationary
Both Wilson (1963) and Morgan (1971) noted the apparently fixed nature of hot spots, and this became integral
to mantle plume theory and was often used to reconstruct
plate motions. Subsequent work showed that hot spots do
move relative to one another (e.g., Molnar and Stock,
1987), although the velocities are much lower than those
of lithospheric plates. In addition, the relative motion
between Pacific hot spots and between Indo-Atlantic hot
spots is much smaller than the relative motion between a
Pacific hot spot and an Atlantic or Indian hot spot (e.g.,
Courtillot et al., 2003). It was perhaps always overly simplistic to expect hot spots to remain stationary; after all,
convection plumes in the atmosphere are not stationary:
thunderstorms move and can be bent by atmospheric
shear. We might expect similar behavior in the mantle.
Indeed, based on a mantle convection model, Steinberger
et al. (2004) were able to explain nearly all the apparent
relative motion between hot spots. The surprising thing
is how little, not how much, hot spots move relative to
one another. The most likely reason is that three-fourths
of a plume’s journey to the surface takes place through
the lower mantle, whose viscosity is probably one to two
orders of magnitude greater than that of the upper mantle
and consequently flow is sluggish.
Petrology and geochemistry of oceanic island
basalts and mantle plumes
Morgan (1971) pointed out that his model was compatible
with the compositional difference between oceanic island
basalts (OIB) and mid-ocean ridge basalts (MORB)
318
HOT SPOTS AND MANTLE PLUMES
focus on only hot spots related to mantle plumes. Campbell (2007) and White (2010) provide additional reviews.
Plumes and large igneous provinces
Early laboratory and numerical simulations of plumes (e.g.,
Whitehead and Luther, 1975; Campbell and Griffiths, 1990;
Griffiths and Campbell, 1990; Olsen, 1990) suggested that
plumes necessarily start with a large voluminous head. As
Morgan (1972) noted, a number of oceanic island and seamount chains extrapolate to large igneous provinces, or
LIPs, which include flood basalts or “traps” on continents
and oceanic plateaus in the oceans and result from massive
eruptions of basalt over a relatively short period of time. For
example, the Reunion plume, currently beneath volcanically active Reunion Island, can be connected through Mauritius, the Mascarene Plateau, and the Chagos-Laccadive
Ridge to the Deccan Traps flood basalt province of India,
produced 65 Ma ago by the eruption of more than
500,000 km
3 of basalt in roughly a million-year period.
These observations led to the hypothesis that LIPs are the
products of plume heads and oceanic islands are the product
of plume tails (Richards et al., 1989). This idea became an
essential component of the plume theory. However, relatively few oceanic island chains can be confidently associated with LIPs. Subsequent numerical simulations indicate
that the surface expression of thermochemical plumes, as
opposed to purely thermal plumes, “may be a headless,
age-progressive volcanic chain” (Farnetani and Samuel,
2005). Oceanic plateaus, a variety of LIP, are the subject
of a separate entry and will not be further discussed here.
How many plumes?
Table 1 lists 29 oceanic island chains/hot spots possibly
produced by mantle plumes, but just how may mantle
plumes are presently active is unclear and debated. Morgan suggested there were about 20 mantle plumes. Sleep
(1990) estimated buoyancy fluxes for 37 plumes.
Courtillot et al. (2003) considered 49 “hot spots” and concluded that only 7–10 were products of deep mantle
plumes based on five criteria: (1) a long-lived track, (2) a
LIP at initiation, (3) a topographic anomaly indicative of
a buoyancy flux in excess of 10
3 kg/s, (4) high
3 He/
4
He
and
20
Ne/
22 Ne ratios, and (5) negative shear-wave anomalies indicative of hot mantle. Only those hot spots meeting at least three or more of these criteria were
considered to be of “primary plumes” of likely deep mantle origin. They suggested that many of the remaining hot
spots were “secondary,” perhaps triggered by large low
shear-wave velocity provinces (LSSVPs) located at the
base of the mantle and extend up to 1,000 km above it
beneath Africa and the central equatorial Pacific.
As noted earlier, it is possible that plumes might begin
without large buoyant heads and some may be short lived
or young. In addition, Courtillot’s noble gas isotope criterion assumes mantle plumes always consist of relatively
undegassed material, whereas other chemical evidence
discussed below suggests recycled oceanic crust may be
the dominant component in many cases. Thus, of the
Courtillot’s 5 criteria, the last, seismic evidence of hot
mantle, is perhaps the most important. Subsequent seismic
studies by Montelli et al. (2004, 2006) found slow seismic
wave velocity anomalies under 30 of 33 hot spots that
extended at least to depths greater than the upper–lower
mantle boundary at 660 km. Of these, the velocity anomalies extend to depths of greater than 1,450 km in 23 of
the hot spots examined. In addition, Montelli et al. (2006)
identified three “plumes” in the lowermost mantle that did
not extend into the upper mantle (and are not associated
with surface hot spots).
Ebinger and Sleep (1998) suggested that ponding and
flow of the “Ethiopian” or Afar plume beneath the African
lithosphere could account not only for the Mid-Tertiary
Ethiopian flood basalt province and modern Afar volcanism but also hot spots in central North Africa such as
Tibesti, the Cameroon line volcanism, and the oceanic
island volcanism on the Comoros. Furthermore, some of
the plumes imaged by Montelli et al. appear to merge at
depth, such as the Ascension and St. Helena plumes. Thus,
a single plume in some cases might be responsible for
more than one hot spot.
Hot spots aren’t entirely stationary
Both Wilson (1963) and Morgan (1971) noted the apparently fixed nature of hot spots, and this became integral
to mantle plume theory and was often used to reconstruct
plate motions. Subsequent work showed that hot spots do
move relative to one another (e.g., Molnar and Stock,
1987), although the velocities are much lower than those
of lithospheric plates. In addition, the relative motion
between Pacific hot spots and between Indo-Atlantic hot
spots is much smaller than the relative motion between a
Pacific hot spot and an Atlantic or Indian hot spot (e.g.,
Courtillot et al., 2003). It was perhaps always overly simplistic to expect hot spots to remain stationary; after all,
convection plumes in the atmosphere are not stationary:
thunderstorms move and can be bent by atmospheric
shear. We might expect similar behavior in the mantle.
Indeed, based on a mantle convection model, Steinberger
et al. (2004) were able to explain nearly all the apparent
relative motion between hot spots. The surprising thing
is how little, not how much, hot spots move relative to
one another. The most likely reason is that three-fourths
of a plume’s journey to the surface takes place through
the lower mantle, whose viscosity is probably one to two
orders of magnitude greater than that of the upper mantle
and consequently flow is sluggish.
Petrology and geochemistry of oceanic island
basalts and mantle plumes
Morgan (1971) pointed out that his model was compatible
with the compositional difference between oceanic island
basalts (OIB) and mid-ocean ridge basalts (MORB)
318
HOT SPOTS AND MANTLE PLUMES