90 km. Thus, the extent of melting alone is not a good indicator of mantle temperature.
A useful comparison is between melting in ridgecentered plumes and mid-ocean ridges distant from them.
Dick et al. (1984) found that abyssal peridotites in the
vicinity of hot spots had mineral abundances and compositions suggesting they had melted to greater degree than
had peridotites away from hot spots. They noted that the
peridotite mineralogy was correlated regionally with
basalt chemistry and concluded that hot spots were indeed
hotter and hence likely products of mantle plumes.
Michael and Bonatti (1985) reached similar conclusions
from a study of peridotite compositions. Klein and
Langmuir (1987) concluded from a comprehensive
study of MORB chemistry that mantle temperatures
beneath the mid-ocean ridges varied by as much as
250
C. The highest temperatures, and highest extents of
melting, invariably occurred near hot spots such as Iceland. This, of course, makes some sense because Iceland
crust is far thicker (~30 km) than normal oceanic crust
(~6–7 km), requiring much greater amounts of melt for
its creation.
Putirka et al. (2007) used a refined version of the
Roeder and Emslie (1970) olivine geothermometer to calculate melt temperatures of magmas from mid-ocean
ridges and oceanic islands and from them derived mantle
potential temperatures. They calculated mantle potential
temperatures of 1,454 Æ 81
C for mantle beneath
mid-ocean ridges, 1,722
C for mantle beneath the Hawaiian and Samoan chains, and 1,616
C for mantle beneath
Iceland. These translate to excess temperatures of 268
C
and 162
C, respectively, in overall good agreement with
earlier studies. Consistent with this, Putirka et al. (2011)
calculated that the Hawaiian magmas are generated by
up to 20 % melting, compared to 8 % for MORB.
Potential temperatures calculated by Herzberg
et al. (2007) for ambient mantle beneath mid-ocean ridges
were lower: in the range of 1,280–1,400
C, but they nevertheless concluded that plumes are typically 200–300
C
hotter. Herzberg and Asimow (2008) estimated mantle
excess potential temperatures for a variety of other plumes
and found that they scattered between those for Hawaii
(~200
C) and for Iceland (~100
C). These estimates
are also in good agreement with studies that use entirely
different approaches, such as the width of the geochemical
anomaly along ridges (Schilling, 1991) and excess topography and geoid height (Watson and Mckenzie, 1991).
Finally, Courtier et al. (2007) found a correlation between
petrologically determined temperatures in OIB and
MORB and the seismically determined thickness of the
underlying mantle transition zone. Beneath hot spots, the
transition zone is thinner, precisely the expected result if
hot spots result from hot mantle rising through the transition zone. There is, however, not a complete consensus
on this point as Green et al. (2001) and Falloon
et al. (2007), also using the olivine geothermometer, found
no difference in potential temperatures between the
sub-ridge and the sub-Hawaiian mantle. Possible
explanations for this discrepancy are discussed by Putirka
et al. (2007).
The melt production rate is related to the upwelling
rate. The extent of radioactive disequilibrium between
230
Th and its parent,
238
U, should depend on these rates
(e.g., Spiegelman and Elliot, 1993). Consequently, the
degree to which the
230
Th/
238 U activity ratio (activity
ratios will hereafter be designated by parentheses) exceeds
1 should be a measure of the rate of mantle upwelling.
Indeed, Chabaux and Allegre (1994) demonstrated that
(
230
Th/
238 U) ratios in OIB are inversely correlated with
buoyancy flux as estimated by Sleep (1990). Bourdon
et al. (2006) demonstrated that the relationship observed
by Chabaux and Allègre is precisely that of the expected
based on simple assumptions about melt porosity and partition coefficients and the relationship between melt production and upwelling rate. They further showed that
(
230
Th/
238 U) and (
231
Pa/
235
U) ratios vary with distance
from the assumed center of plumes in a manner that is consistent with buoyancy fluxes and excess temperatures in
the range of 100–300
C.
Rejuvenescent volcanism
One aspect of oceanic island volcanism that remains
somewhat puzzling is the so-called “post-erosional” or
“rejuvenescent” volcanism: eruptions that occur after a
long hiatus, sometimes several million years. The Honolulu volcanic series of Oahu, which includes the iconic
Diamond Head cone, is the classic example. This phenomenon was first recognized in Hawaii (e.g., Sterns, 1946;
MacDonald, 1968) and has since been recognized in at
least Samoa, the Society Islands, the Marquesas, Madeira,
the Canaries, the Louisville Seamounts, Mauritius, and,
most recently, Jasper Seamount (Konter et al., 2009). Usually these cases mimic the Hawaiian pattern in which the
post-erosional phase involves eruption of alkaline to
highly alkaline, low-silica lavas enriched in incompatible
elements (all of which indicate low extents of melting)
but with a more depleted isotopic signature than lavas of
the main shield-building phase (e.g., Chen and Frey,
1983). The eruption rates in the post-erosional phase are,
however, far lower than in the shield-building phase.
Garcia et al. (2010) estimated the rejuvenescent Koolau
Series on Kaua’i comprising about 0.1 % of the volcano
volume, while Moore et al. (2011) estimate the rejuvenescent phase on Mauritius comprising about 0.05 % of the
volume, even though these lavas cover most of the
island’s surface. Rejuvenescent volcanism has not been
identified in on- or near-ridge hot spots such as the
Galapagos and Azores.
In Samoa, the rejuvenescent volcanism is so extensive
on the westernmost island of Savai’i that it was initially
thought that age progressed from east to west, opposite
of the prediction of the hot spot model (Hawkins and
Natland, 1975). Natland (1980) argued that Samoan volcanism resulted from lithospheric flexure as the Pacific
plate passes north of the Tonga–Kermadec subduction
HOT SPOTS AND MANTLE PLUMES
323
A useful comparison is between melting in ridgecentered plumes and mid-ocean ridges distant from them.
Dick et al. (1984) found that abyssal peridotites in the
vicinity of hot spots had mineral abundances and compositions suggesting they had melted to greater degree than
had peridotites away from hot spots. They noted that the
peridotite mineralogy was correlated regionally with
basalt chemistry and concluded that hot spots were indeed
hotter and hence likely products of mantle plumes.
Michael and Bonatti (1985) reached similar conclusions
from a study of peridotite compositions. Klein and
Langmuir (1987) concluded from a comprehensive
study of MORB chemistry that mantle temperatures
beneath the mid-ocean ridges varied by as much as
250
C. The highest temperatures, and highest extents of
melting, invariably occurred near hot spots such as Iceland. This, of course, makes some sense because Iceland
crust is far thicker (~30 km) than normal oceanic crust
(~6–7 km), requiring much greater amounts of melt for
its creation.
Putirka et al. (2007) used a refined version of the
Roeder and Emslie (1970) olivine geothermometer to calculate melt temperatures of magmas from mid-ocean
ridges and oceanic islands and from them derived mantle
potential temperatures. They calculated mantle potential
temperatures of 1,454 Æ 81
C for mantle beneath
mid-ocean ridges, 1,722
C for mantle beneath the Hawaiian and Samoan chains, and 1,616
C for mantle beneath
Iceland. These translate to excess temperatures of 268
C
and 162
C, respectively, in overall good agreement with
earlier studies. Consistent with this, Putirka et al. (2011)
calculated that the Hawaiian magmas are generated by
up to 20 % melting, compared to 8 % for MORB.
Potential temperatures calculated by Herzberg
et al. (2007) for ambient mantle beneath mid-ocean ridges
were lower: in the range of 1,280–1,400
C, but they nevertheless concluded that plumes are typically 200–300
C
hotter. Herzberg and Asimow (2008) estimated mantle
excess potential temperatures for a variety of other plumes
and found that they scattered between those for Hawaii
(~200
C) and for Iceland (~100
C). These estimates
are also in good agreement with studies that use entirely
different approaches, such as the width of the geochemical
anomaly along ridges (Schilling, 1991) and excess topography and geoid height (Watson and Mckenzie, 1991).
Finally, Courtier et al. (2007) found a correlation between
petrologically determined temperatures in OIB and
MORB and the seismically determined thickness of the
underlying mantle transition zone. Beneath hot spots, the
transition zone is thinner, precisely the expected result if
hot spots result from hot mantle rising through the transition zone. There is, however, not a complete consensus
on this point as Green et al. (2001) and Falloon
et al. (2007), also using the olivine geothermometer, found
no difference in potential temperatures between the
sub-ridge and the sub-Hawaiian mantle. Possible
explanations for this discrepancy are discussed by Putirka
et al. (2007).
The melt production rate is related to the upwelling
rate. The extent of radioactive disequilibrium between
230
Th and its parent,
238
U, should depend on these rates
(e.g., Spiegelman and Elliot, 1993). Consequently, the
degree to which the
230
Th/
238 U activity ratio (activity
ratios will hereafter be designated by parentheses) exceeds
1 should be a measure of the rate of mantle upwelling.
Indeed, Chabaux and Allegre (1994) demonstrated that
(
230
Th/
238 U) ratios in OIB are inversely correlated with
buoyancy flux as estimated by Sleep (1990). Bourdon
et al. (2006) demonstrated that the relationship observed
by Chabaux and Allègre is precisely that of the expected
based on simple assumptions about melt porosity and partition coefficients and the relationship between melt production and upwelling rate. They further showed that
(
230
Th/
238 U) and (
231
Pa/
235
U) ratios vary with distance
from the assumed center of plumes in a manner that is consistent with buoyancy fluxes and excess temperatures in
the range of 100–300
C.
Rejuvenescent volcanism
One aspect of oceanic island volcanism that remains
somewhat puzzling is the so-called “post-erosional” or
“rejuvenescent” volcanism: eruptions that occur after a
long hiatus, sometimes several million years. The Honolulu volcanic series of Oahu, which includes the iconic
Diamond Head cone, is the classic example. This phenomenon was first recognized in Hawaii (e.g., Sterns, 1946;
MacDonald, 1968) and has since been recognized in at
least Samoa, the Society Islands, the Marquesas, Madeira,
the Canaries, the Louisville Seamounts, Mauritius, and,
most recently, Jasper Seamount (Konter et al., 2009). Usually these cases mimic the Hawaiian pattern in which the
post-erosional phase involves eruption of alkaline to
highly alkaline, low-silica lavas enriched in incompatible
elements (all of which indicate low extents of melting)
but with a more depleted isotopic signature than lavas of
the main shield-building phase (e.g., Chen and Frey,
1983). The eruption rates in the post-erosional phase are,
however, far lower than in the shield-building phase.
Garcia et al. (2010) estimated the rejuvenescent Koolau
Series on Kaua’i comprising about 0.1 % of the volcano
volume, while Moore et al. (2011) estimate the rejuvenescent phase on Mauritius comprising about 0.05 % of the
volume, even though these lavas cover most of the
island’s surface. Rejuvenescent volcanism has not been
identified in on- or near-ridge hot spots such as the
Galapagos and Azores.
In Samoa, the rejuvenescent volcanism is so extensive
on the westernmost island of Savai’i that it was initially
thought that age progressed from east to west, opposite
of the prediction of the hot spot model (Hawkins and
Natland, 1975). Natland (1980) argued that Samoan volcanism resulted from lithospheric flexure as the Pacific
plate passes north of the Tonga–Kermadec subduction
HOT SPOTS AND MANTLE PLUMES
323